 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K IProVAccﬁPres/CIRC/Dateoue Indd

 

SYNTHESIS AND CHARACTERIZATION OF NOVEL CHALCOPHOSPHATE
MATERIALS

Matthew A. Gave

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Chemistry

2008

ABSTRACT

SYNTHESIS AND CHARACTERIZATION OF NOVEL CHALCOPHOSPHATE
MATERIALS

By
Matthew A. Gave
Metal chalcophosphate compounds have received signiﬁcant attention due to
the presence of some industrially relevant physical properties such as spontaneous
ferroelectric ordering,1'3 reversible phase changes,4 nonlinear optical properties,5 and
photoconductivity.6 These compounds contain a chalcogenide (Q = S, Se, or Te) that

is directly bonded to oxidized phosphorous to form various [PnyP‘ anions. The

identity and relative concentration of the metal ion has a dominant effect on the
structure and properties of the resultant compound, and is not simply a requirement
for charge balance.

In this work, several novel chalcophosphate materials are presented, and some
insights are provided into the nature of the molten reactions that form them. The

novel compounds have the general formula MM’ [Pny] where M = Cu, Ag,2 '1'],7 K3,

or Cs,9 M’ = Bi or Hg, and Q = S, or Se. The synthetic conditions required to
produce each compound are detailed and so are the structural features of the
crystallographic reﬁnement. These crystalline compounds are also characterized by
other spectroscopic techniques including Uv/vis and Raman, thermal analysis, and
electronic structure calculations.

The reactions to produce these compounds were characterized by high
temperature nuclear magnetic resonance, and from these data, the presence of a

signiﬁcant quantity of free radicals in the melts and syntheses of chalcophosphate-

containing materials is demonstrated. Additionally, the melt was found to have
relatively low viscosity, and there was evidence that the species that exist in the
molten liquid have small (several A) radii and may be related to the species that are

observed in the room temperature crystalline compounds.

1. Bourdon, X.; Maisonneuve, V.; Cajipe, V. B.; Payen, C.; Fischer, J. E., J. Alloys
Compd. 1999, 283, 122-127.

2. Gave, M.; Bilc, D.; Mahanti, S.; Breshears, 1.; Kanatzidis, M., Inorg. Chem. 2005,
5293-5303.

3. Maisonneuve, V.; Cajipe, V. B.; Simon, A.; VonDerMuhll, R.; Ravez, J., Phys.
Rev. B. 1997, 56, 10860-10868.

4. Chung, 1.; Do, 1.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G., Inorg. Chem.
2004, 43, 2762-2764.

5. Bridenbaugh, P., Mater. Res. Bull. 1973, 8, 1055-1060.

6. Galdamez, A.; Manriquez, V.; Kasaneva, J.; Avila, R. E., Mater. Res. Bull. 2003,
38, 1063-1072.

7. Gave, M.; Malliakas, C. D.; Weliky, D. P.; Kanatzidis, M. G., Inorg. Chem. 2007,
46, 3632-3644.

8. Gave, M.; Weliky, D. P.; Kanatzidis, M. G., Inorg. Chem. 2007, Accepted.

9. Gave, M.; Canlas, C.; Iyer, R. G.; Kanatzidis, M. G.; Weliky, D. P., J. Solid State
Chem. 2007, 180, 2877-2884.

ACKNOWLEDGEMENTS

The efforts of a number of people have been invaluable to complete the work
contained herein. In particular, I was taught many of the standard techniques to form and
to study various materials by Joseph B. Wachter, James R. Salvador and R. G. Iyer. I am
also grateful to Christos Malliakas and Hanna Lind for many useful discussions and
assistance with various aspects of crystallography. Without the dedication of Kermit L.
Johnson, it would have been impossible to collect the NMR spectra. The expertise of
John L. McCracken in the collection and interpretation of electron paramagnetic
resonance spectra is also appreciatively acknowledged. Carly Wunderlich and Matthew
Field are also acknowledged for their dedication and hard work in the lab. Last, but
certainly not least, are the efforts of my advisors Mercouri G. Kanatzidis and David P.

Weliky. It was an honor to have the chance to learn from both.

iv

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES .......................................................................................................... xii
LIST OF TERMS AND ABBREVIATIONS .................................................................. xix
CHAPTER 1

INTRODUCTION ............................................................................................................... 2
REFERENCES .................................................................................................................... 7
CHAPTER 2

ANTIFERROELECT RIC PHASE TRANSITION S DUE TO COOPERATIVE Cu+ AND

Bi3+ ION MOTION.

INTRODUCTION ............................................................................................................. 1 1
EXPERIMENTAL SECTION ........................................................................................... 12
RESULTS AND DISCUSSION ........................................................................................ l9
CONCLUDING REMARKS ............................................................................................. 42
REFERENCES .................................................................................................................. 44
CHAPTER 3

WIDE COMPOSITIONAL AND STRUCTURAL DIVERSITY IN THE SYSTEM
Tl/Bi/P/Q (Q = S, Se) AND OBSERVATION OF VICINAL P-Tl J-COUPLING IN THE
SOLID STATE.

INTRODUCTION ............................................................................................................. 48
EXPERIMENTAL SECTION ........................................................................................... 50
RESULTS AND DISCUSSION ........................................................................................ 61
CONCLUDING REMARKS ............................................................................................. 91
REFERENCES .................................................................................................................. 93
CHAPTER 4

INIPROVED RESOLUTION AND DECECTION OF 3lP-Tl J-COUPLINGS AT 21 T
IN 31F MAGIC ANGLE SPINNING NMR SPECTRA of INORGANIC COMPOUNDS
CONTAINING 'I‘l/Bi/P/S.

INTRODUCTION ............................................................................................................. 98
EXPERIMENTAL SECTION ......................................................................................... 100
RESULTS AND DISCUSSION ...................................................................................... 101
CONCLUDING REMARKS ........................................................................................... 105
REFERENCES ................................................................................................................ 106

CHAPTER 5
NEW POTASSIUM BISMUTH THIOPHOSPHATES INCLUDING THE
MODULATED K151525084),

INTRODUCTION ........................................................................................................... 108
EXPERIMENTAL SECTION ......................................................................................... 109
RESULTS AND DISCUSSION ...................................................................................... 118
CONCLUDING REMARKS ........................................................................................... 141
REFERENCES ................................................................................................................ 143
CHAPTER 6

MIXED METAL ARGYRODITES: PHASE TRANSITIONS BY CHEMICAL
COMPOSITION.

INTRODUCTION ........................................................................................................... 148
EXPERIMENTAL SECTION ......................................................................................... 152
RESULTS AND DISCUSSION ...................................................................................... 159
CONCLUDING REMARKS ........................................................................................... 167
REFERENCES ................................................................................................................ 168
CHAPTER 7

Cs4PzSe10: A NEW COMPOUND DISCOVERED WITH THE APPLICATION OF
SOLID STATE AND HIGH TEMPERATURE NMR.

INTRODUCTION ........................................................................................................... 170
EXPERIMENTAL SECTION ......................................................................................... 17 1
RESULTS AND DISCUSSION ...................................................................................... 178
CONCLUDING REMARKS ........................................................................................... 188
REFERENCES ................................................................................................................ 190
CHAPTER 8

NUCLEAR MAGNETIC RESONANCE AND ELECTRON PARAMAGNETIC
RESONANCE EVIDENCE FOR FREE RADICALS IN CHALCOGENIDE MELTS
AND SYNTI-IESES.

 

 

INTRODUCTION ........................................................................................................... 194
EXPERIMENTAL SECTION ......................................................................................... 196
RESULTS AND DISCUSSION ...................................................................................... 199
CONCLUDING REMARKS ........................................................................................... 21 1
REFERENCES ................................................................................................................ 212
CHAPTER 9

ON THE SPECIATION OF CHALCOPHOSPHATE-CONTAIN IN G MELTS.
INTRODUCTION ...... .. . .................................... 216
EXPERIMENTAL SECTION ......................................................................................... 217
RESULTS AND DISCUSSION ...................................................................................... 223
CONCLUDING REMARKS ........................................................................................... 238
REFERENCES ................................................................................................................ 240

vi

CHAPTER 10
THE SOLID SOLUTIONS Cu3PSe4_xSx AND Fe2P23e6,xSx.

 

 

 

INTRODUCTION . ....................................................................... 243
EXPERIMENTAL SECTION ......................................................................................... 245
RESULTS AND DISCUSSION ...................................................................................... 251
CONCLUDING REMARKS ........................................................................................... 263
REFERENCES ................................................................................................................ 264
CHAPTER 11

CONCLUSIONS AND FUTURE WORK ...................................................................... 266
REFERENCES" .......................................................................... 273
APPENDICES

APPENDD( A .................................................................................................................. 276
APPENDIX B .................................................................................................................. 279
APPENDIX C .................................................................................................................. 282
APPENDIX D .................................................................................................................. 285

vii

LIST OF TABLES

Table 2-1. Crystallographic data, experimental and reﬁnement details for the CuBiPZSe6,
AgB1P2866, and AgB1P286 ............ . I6

 

Table 2-2. Fractional atomic coordinates and U(eq) values for CuBiPZSe6 at 298 K with
standard deviations in parenthesis ..................................................................................... 16

Table 2-3. Fractional atomic coordinates and U(eq) values for CuBiPZSe6 at 173 K with
standard deviations in parenthesis ..................................................................................... 17

Table 2-4. Fractional atomic coordinates and U(eq) values for CuBiPZSe6 at 97 K with
standard deviations in parenthesis ..................................................................................... 17

Table 2—5. Fractional atomic coordinates and U(eq) values for AgBiPzS6 at 298 K with
standard deviations in parenthesis ..................................................................................... 17

Table 3-1. Crystallographic data, experimental and reﬁnement details for the Tl/Bi/P/Se
and Tl/Bi/P/S compounds ..................................................................... 56

 

Table 3-2. Fractional atomic coordinates and U(eq) values for a-TlBiPZSe6 with standard
deviations in parenthesis .................................................................................................... 56

Table 3—3. Fractional atomic coordinates and U(eq) values for B-TlBiPZSe6 with standard
deviations in parenthesis ............................................................. .. 57

 

Table 3—4. Fractional atomic coordinates and U(eq) values for TlBiP286 with standard
deviations in parenthesis ........................................... . 57

 

Table 3-5. Fractional atomic coordinates and U(eq) values for 'I‘13Bi3(PS4)4 with standard
deviations in parenthesis ........................................................ . .................... 57

 

Table 3-6. Fractional atomic coordinates and U(eq) values for T 1BiPZS7 with standard
deviations in parenthesis ........... - - ..... . . - -- . 59

 

Table 3-7. Selected bond lengths and angles for a'TIBIPZSC6 with bond lengths (A) and
angles (deg) with standard uncertainties in parenthesis .................. 66

 

Table 3-8. Selected bond lengths and angles for B- 'TlBiPZSe6 with bond lengths (A) and
angles (deg) with standard uncertainties in parenthesis ..................................................... 69

Table 3-9. Selected bond lengths and angles for TlBiP286 with bond lengths (A) and
angles (deg) with standard uncertainties in parenthesis ................... 72

 

viii

Table 3-10. Selected bond lengths and angles for Tl3Bi3(PS4)4 with bond lengths (A)
and angles (deg) with standard uncertainties in parenthesis .............................................. 77

Table 3-11. Selected bond lengths and angles for a-TlBiPZS7 with bond lengths (A) and
angles (deg) with standard uncertainties in parenthesis ..................................................... 82

Table 3-12. 31P chemical shift (CS), chemical shift anisotropy (CSA), and T] values for
the Tl/Bi/P/Se and TllBi/P/S compounds .......................................................................... 87

Table 4-1. Chemical shifts, linewidths, and integrated intensities of the ﬁtted Lorentzian
components of the spectra of T14Bi2(PS4)2(PZS6), Tl3Bi(PS4)2, and T1BiP2$7 ............. 102

Table 4-2. Average chemical shifts and splittings of the spectral clusters of
TI4B12(PS4)2(P286), TI3BI(PS4)2, and TIBIP237 ........................... .. 102

 

Table 5- 1. Crystallographic Data, Experimental and Reﬁnement Details for the K/Bi/P/S
compounds ................... 1 13

 

Table 5-2. Fractional atomic coordinates and U(eq) values for K3Bi3(PS4)4 with standard
deviations in parenthesis ............................................................... . ..... 113

 

Table 5-3. Fractional atomic coordinates and U(eq) values for the average structure of
K1.5Bi2_5(PS4)3 with standard deviations in parenthesis ................................................. 114

Table 5-4. Fractional atomic coordinates and U(eq) values for KgBi(PS4)4 with Standard
deviations in parenthesis .................................................................................................. 114

Table 5-5. Selected bond lengths and angles for K3Bi3(PS4)4 with bond lengths (A) and
angles (deg), with Standard uncertainties in parenthesis .................................................. 123

Table 5-6. Selected bond lengths for K1.5Bi2'5(PS4)3 with bond lengths (A), with
standard uncertainties in parenthesis ............................................................................... 126

Table 5-7. Modulation parameters for K1_5Bi2.5(PS4)3 ................................................... 130

Table 5-8. Selected bond lengths and angles for KgBI(PS4)4 with bond lengths (A) and
angles (deg), with standard uncertainties in parenthesis .................................................. 133

Table 5-9. 31P chemical shift (CS), chemical shift anisotropy (CSA), and T1 values....139
Table 6-1. Crystallographic data and reﬁnement details for Cu3HgPSe6 ....................... 158

Table 6-2. Fractional atomic coordinates and U(eq) values for Cqu3PSe6 with standard
uncertainties in parenthesis ......................................... .. .. .................... 158

 

ix

Table 6-3. Anisotropic thermal displacement parameters for Cqu3PSe6 with standard
uncertainties in parenthesis .............................................................................................. 159

Table 6-4. Unit cell parameters determined from the indexed powder X-ray diffraction
patterns ....................................................................................... 160

 

Table 6-5. Selected bond lengths and angles for Cqu3PSe6 with bond lengths (A) and
angles (°), with standard uncertainties in parenthesis ...................................................... 164

Table 7-1. Crystallographic data for CS4P23610 .............................................................. 175

Table 7-2. Fractional atomic corrdinates for Cs4PZSe10 with standard deviations in
parenthesis -- .- - . ............................................ .. ...175

 

 

Table 7-3. Selected bond lengths A and angles (°) for Cs4P2Se10 with standard
uncertainties in parenthesis .............................................................................................. 175

Table 8-1. NMR parameters in high temperature melts and room temperature solids.
Samples in quotation marks represent nominal compositions, and those without quotation
marks were pure crystalline compounds at room temperature. The 22 °C data listed
above and below the high temperature data were collected before and after the high
temperature data, respectively ......................................................................................... 201

Table 9-1. The 31P chemical shifts, longitudinal relaxation times (T1), and line widths of

various chalcophosphate-containing materials as well as the ﬁnal product determined
from powder X-ray diffraction ......................................................................................... 225

Table 9-2. The 31P chemical shift and peak width of the “KPl_5Se6_3” sample over the
623 — 823 K temperature range ........................................................................................ 228

Table 9-3. The 31P chemical shift and peak width of the “CS4PZSe16” sample over the
623 - 873 K temperature range ........................................................................................ 229

Table 9-4. Summary of the ﬁtting parameters for each of the samples studied and the
corrected chemical shift and peak width extrapolated to 298 K determined using those
parameters ..................................................................................... 231

 

Table 9-5. The 31P chemical shift and peak width of the “PS5” sample over the 623 - 873
K temperature range ............................................................................... 233

 

Table 9-6. The enthalpy of radical formation The 31P chemical shift was extrapolated to
0 K. .............. ........................................ 237

 

 

Table 10-1. Literature crystallographic parameters for the parent compounds of the solid
solutions ...................................................................................... .. 244

Table 10-2. Summary of the details of the crystallographic reﬁnement of Cu3PSeS3 and
FezPZSe/ZS‘; . - - ............................................... 247

 

 

Table 10-3. Fractional atomic coordinates for Cu3PSeS3 with standard uncertainties in
parenthesis - - - .......................................................................... 248

 

Table 10-4. Fractional atomic coordinates for FezPZSeZS4 with standard uncertainties in
parenthesis .................................................................................... . ....... 248

Table 10-5. Selected bond lengths (A) and angles (deg) for Cu3PSeZS4. Standard
uncertainties are displayed in parenthesis ................................ 249

Table 10-6. Selected bond lengths (A) and angles (deg) for FezPZSeS3. Standard
uncertainties are displayed in parenthesis ................. . - 250

Table 10-7. The cell parameters of each of the Cu3PSe4_xSx compounds as indexed from
the powder X-ray diffraction data ........................................................... 252

 

Table 10-8. The cell parameters of each of the FeZPZSe6.xSx compounds as indexed from
the powder X-ray diffraction data ........... - - . ............................................... 256

 

Table 10-9. The energy in wavenumber (cm'l) of the strongest spectral features in the
Raman spectra of the Cu3,PSe4.xSx compounds .............................................................. 258

Table 10-10. The 31P NMR spectral features observed for the Cu3PSe4,xSx compounds,
in ppm. The peaks corresponding to the Cu7PS6 impurity have been omitted for clarity.

The peaks common to different stoichiometry compounds have been grouped
horizontally ............................. .. ..... 259

 

 

 

Table A-1. Other synthetic attempts with the reactants , maximum temperature, cooling
time, lab notebook page number, and the major products as identiﬁed by X-ray
diffraction. In cases where the powder diffraction peaks had no matches, the product is
listed as “Unknown.” When the powder pattern matched a known diffraction pattern but
with slightly shifted peaks, the product 18 given in italics ...................... . .278

xi

LIST OF FIGURES

Figure 2-1. The optical absorption spectra of (A) CuBiPZSe6, (B) AgBiPZSe6 and (C)
AgBisté. The high residual absorbance below the band gap in (A) is likely the result of
a BiZSe3 impurity. The residual absorbance in (B) is the result of AgBiSez, and the
residual absorbance in (C) is likely Bi2S3 ......................................................................... 21

Figure 2-2. The layered structures of (A) CuBiPZSe6 and (B) AgBiPZSe6 (room

temperature) as viewed down the [110] direction. Small hollow spheres are Cu(Ag)
atoms, large hollow spheres are Bi atoms, solid black spheres are P, and solid grey
spheres are Se ..................................................................................................................... 25

Figure 2-3. View of a single CuBiPZSe6 layer down the c-axis ([001] direction) showing

the ordered arrangement of the Bi3+ and Cu+ ions. The same arrangement exists in
AgBiPZSe6 ......................................................................................................................... 26

Figure 2-4. The immediate coordination environment of (A) Cu+ and (B) Bi3+ in
CUBIPZSC6 at room temperature ........................................................................................ 26

Figure 2-5. The immediate coordination environments of (A) Ag+ and (B) Bi3+ atoms in
AgBiPZSe6 at 298K; (C) Cu and (D) Bi atoms in CuBiPZSe6 at 97K ............................... 28

Figure 2-6. The (A) 298K (B) 173K, and (C) 97K structures of CuBiPZSe6 viewed down

the [110] direction. The direction of the displacement of the copper atoms is indicated
with a dashed arrow, and the direction of the bismuth atoms is indicated with a solid
arrow. The atoms in the layers marked with a horizontal line have not yet ordered ........ 30

Figure 2-7. Structure of AgBiPZS6 viewed down the a-axis showing the lamellae .......... 34

Figure 2-8. View of a single layer of AgBiPZS6. The P-P distance ranges from 2.212(4)A
to 2.225(5)A, and the P-S distances range from 1.993(3) A to 2.044(3) A ....................... 34

Figure 2-9. The immediate coordination environments of (A) Ag“ and (B) Bi3+ in
AgBiP286 ........................................................................................................................... 35

Figure 2-10. Band structure of CUBIPZSC6: (A) antiferroelectric phase and (B)
paraelecu'ic phase ............................................................................................................... 37

Figure 2-11. Total and partial atomic DOS: (A), (C), (E), (G), (I) antiferroelectric phase
and (B), (D), (F), (H), (J) paraelectric phase ..................................................................... 38

xii

Figure 3-1. (A) The intensity weighted orthorhombic reciprocal lattice of Tl3Bi(PS4)3

down b* indicating the satellite reﬂections that could not be indexed with integral (hkl).
(B) The view down the b* axis of the orthorhombic and monoclinic cells that were used
to attempt to model '1‘13Bi(PS4)3 ................................ . .......... 52

Figure 3-2. Schematic of the optimized reaction conditions required to produce the
known T1/Bi/P/S compounds as the dominant crystalline product. The stoichiometric
ratio, soak time, and soak temperature are indicated ......................................................... 62

Figure 3-3. a—TlBiPZSe6 [1] down the [010] direction showing the lamellae separated by a

layer of T1“ ions with the double-chain moiety common to [I] and [II] emphasized by the
rectangular dotted outline. Small black spheres are P, large black spheres are T1, large
white spheres are Bi, and shaded spheres are Se ............................................................... 64

Figure 3—4. (A) The coordination of Tl+ in a-TlBiP28e6 [I] and (B) the coordination of
Bi3+ showing tris-chelation by one [P2Se6]4' and monodentate bonding by each of the
other three [PzSe6]4' anions ............................................................................................... 65

Figure 3-5. Assembly of the chain common to both [I] and [H] viewed down the (001)
direction ............................................................................................................................. 67

Figure 3-6. B-TlBiPZSe6 [11] down the (010) axis showing emphasizing the weak inter-
chain interaction with a dashed line, and double-chain moiety common to [I] and [II]
emphasized by rectangular outline ..................................................................................... 68

Figure 3-7. (A) The coordination of T1+ in B-TlBiPZSe6 and (B) the coordination of Bi3+
showing the weak inter-chain Bi-Se interaction with a dashed line .................................. 68

Figure 3-8. The structure of TlBiP2S6 [III] viewed down the (100) axis showing the
lamellar stacking motif ........ .. .. ...... 71

 

Figure 3-9. (A) the Tl+ coordination environment, and (B) the coordination environment
of Bi3+ in TlBiPZSe6 [III] .................................................................................................. 72

Figure 3-10. (A) Tl3Bi3(PS4)4 [IV] viewed down the a-axis showing the stacking of the
T13Bi3(PS4)4 along the b-axis with a slab emphasized by rectangular outline, (B) an
individual slab of Tl3Bi3(PS4)4 as viewed down the c-axis showing the empty channels
for the Tl+ ions and (C) the coordination environments of Bi(l) and Bi(2) ...................... 74

Figure 3-11. The coordination environments of the crystallographically unique Tl+ ions
in TI3BI3(PS4)4 [IV] ........................................................................................................... 76

xiii

 

Figure 3-12. The structure of TlBiPZS7 [V] viewed down the (001) axis showing the
corrugated lamellar packing ......................................................... . 80

Figure 3-13. (A) The T1“ coordination environment. Tl atom is green. (B) the
coordination environment of Bi3+ in TlBiP2S7 [V] ........................................................... 81

Figure 3-14. (A) The [Bi(PS4)2]3' chain that runs down the b-axis and (B) perpendicular

to the b-axis showing the Bi3+ positional disorder. Bi atoms in white, phosphorous and
sulfur atoms in black and shaded, respectively .................................................................. 83

Figure 3-15. The optical absorption spectra showing energy band gaps of (A) a-
TlBiP2$e6 [I]: 1.23 eV, B-TlBiPZSe6 [II]: 1.27 eV, and TlBiP2S6 [111]: 1.81 eV; and (B)
'I'l3Bi3(PS4)4 [IV]: 1.88 eV, TlBiP2S7 [V]: 1.87 eV, T13Bi(PS4)4 [VI]: 1.87 eV, and
“4312(P286)(PS4)2 [VII]. 1.98 CV .................................................................................... 84

Figure 3-16. 31P NMR spectra of (A) a-TlBiPZSe6 [I], (B) B-TlBiPZSe6 [H], (C)
TlBista [1111, (D) T13Bi3(PS4)4 [IV]. (E) TlBist7 [V]. (F) TlgBia>sn2 [VI]. (G)
T14Bi2(PS4)2(P2S6), and (H) KBiP287. Each spectrum was an average of four scans with

a delay between pulses of 5000 S. Each spectrum was processed with < 100 Hz line
broadening. Peaks and their corresponding spinning sidebands are grouped with like
symbols ................................................... .. .86

 

 

Figure 4-1. (A) The structure of Tl4Bi2(PS4)2(P2S6) viewed down the b-axis. Large black
spheres are Tl, large hollow spheres are Bi, small black spheres are P, and small shaded
spheres are S. A [P286]4' anion is highlighted in a dashed box. Two Tl+ ions have been
removed so that the highlighted [P2S6]4- unit is more clearly displayed. Two of the Ss in

this unit are not visible because they are directly behind displayed Ss. (B) The structure
of T'lBiP2S7 viewed down the c-axis. Large black spheres are Tl, large hollow spheres

are Bi, small black spheres are P, and small shaded spheres are S. A single unit cell is
shown for each structure .................................................................................................... 98

Figure 4-2. 3113 NMR spectra Of TI4B12(PS4)2(P236), TI3BI(PS4)2, and TIBist‘y Obtained
at 9.4 T and 21 T. The experimental spectra are represented by solid lines and the best-ﬁt
Lorentzian functions used to deconvolve the spectra are represented by dashed lines. The
sums of the Lorentzians are also shown as dashed lines and ﬁtted well to the
corresponding experimental spectra. ............................................................................... 100

Figure 5-1. (A) K3Bi3(PS4)4 viewed down the [001] axis showing the large channels that
are occupied by K+ ions. (B) An undulating Bi3S12 chain that forms the walls of the
tunnels .............................................................................................................................. 120

Figure 5-2. The coordination environments of (A) K(1), (B) K(2), and (C) K(3) ions in
K3Bi3(PS4)4 . ................................................................ 121

 

xiv

Figure 5-3. Void space representation of K3Bi3(PS4)4 showing the relative sizes and

contours of each of the channels viewed (A) perpendicular to [001] and (B) parallel to
[001] ..................................................................................................... 122

 

Figure 5-4. (A) The 3D approxirnant structure of K1.5Bi2.5(PS4)3 as viewed down [100]
showing the assembly of the chains centered at 1/2 the [010] and [001] axes, and (B) the
assembly of “Bi2(PS4)4” chains, emphasized in a rectangular outline. (B) An individual
“Bi2(PS4)4” chain as viewed down the [001] axis ........................................................... 125

Figure 5-5. The coordination environment of (A) Bi(l), (B) K(2), and (C) the K(3)/Bi(3)
mixed site in K153125934», .......................................................................................... 126

Figure 5-6. Percent occupancies of the complementary modulated sites K(3)/Bi(3) and
K(4)/Bi(4) in K1.5Bi2_5(PS4)3 with the ordinate representing the fractional distance along

the modulation axis .......................................................................................................... 128

Figure 5-7. View of K1.5Bi2.5(PS4)3 down the b axis, showing how the K and Bi atoms

interchange differently along the c axis for the K3/Bi3 and K4/Bi4 pairs. The subcell is
displayed in black ....................................................... ....................... 129

 

Figure 5-8. (A) View down [010] of K9Bi(PS4)4 and (B) a view of an individual
molecule of [Bi(PS4)]9' .................................................................................................... l3 1

Figure 5-9. The coordination environment of (A) K(1), (B) K(2), (C) K(3), (D) K(4), and
(E) K(S) ions in K9Bi(PS4)4 ............................................................................................ 132

Figure 5-10. The Raman spectra of (A) K3Bi3(PS4)4, (B) K1.5Bi2.5(PS4)3 and (C)
KgBi(PS4)4. The K3Bi3(PS4)4 sample used to collect the spectrum also contained several
crystalline impurities (see text) ........................................................................................ 137

Figure 5-11. 31P NMR spectra of (A) K1.SBi2_5(PS4)3 and (B) K9Bi(PS4)4 with isotropic
peaks and their spinning sidebands grouped under like symbols .................................... 141

Figure 6-1. The room temperature structure of Cu6PSSBr in the F—43m spacegroup

Showing (A) the disordered sites containing Cu+ ions, (B) The anionic framework of
corner shared [PS4]3' tetrahedra, and (C) the icosahedral coordination of the mixed S/Br

Site. Hollow spheres are Cu, light grey spheres are S, and the central atom in (C) is the
mixed S/Br site ................................................................................................................. 149

XV

Figure 6-2. The structure of Cd4GeS6 down the b-axis showing the monoclinic cell.
Hollow spheres are Cd, shaded spheres are S, and black spheres are P .......................... 150

Figure 6-3. Scanning electron micrographs of the products formed in the reaction that
produced Cqu3PSe6 with views of the ~2 um surface detail observed on the faces of the

crystals ............................................................................................................................. 154

Figure 6-4. Powder X-ray diffraction of the argyrodite materials. The Ag7PSe6 sample

has been previously described. The quotation marks indicate the nominal stoichiometry
of the product because a stoichiometric ratio that was reﬁned from a crystallographic
solution was unavailable .................................................................................................. 160

Figure 6-5. The coordination environments of (A) the Cu(l) in a bent coordination
geometry and (B) Hg(1) in a distorted tetrahedral geometry in the structure of
Cqu3PSe6 ............................................................................................................................

The positions of the Hg(1) and Cu(l) atoms were reﬁned to be 0.78(2) A apart ............ 163

Figure 6-6. The coordination environments of (A) the trigonal planar Cu(2)/Hg(2) site,
(B) the trigonal planar Cu(3)/Hg(3) site (C) the distorted tetrahedral Cu(4)/Hg(4) site in
the compound Cqu3PSe6 ............................................................................................... 165

Figure 6-7. (A) The differential thermal analysis and (B) optical absorption spectrum of
Cqu3PSe6 ...................................................................................................................... 166

Figure 7-1. Ambient temperature 31P magic angle spinning NMR spectra of the ﬁnal
products of: (A) a “CS4P28e16” reaction stoichiometry indicating Cs4P2Se10 as the
dominant 31P-containing species at -52.8 ppm; (B) a “CS4P2Se3” reaction stoichiometry
indicating CS4P28e9 as the dominant 31P-containing species at -39.9 ppm; and (C) a
“CS4P2Se12” reaction stoichiometry that produced Cs4P2Se9 and CS4P28610 in a ~3:5
ratio. Panel D displays the static 31P NMR spectrum of a 350 0C melt of the products in
C. The peak with a chemical shift of -22.0 ppm in panels B and C was not assigned and
the peak at -99.8 ppm in panel B was not assigned. For spectra A-C, the spinning
frequency was 14 kHz and spinning sidebands are grouped with the corresponding
isotropic peak using like symbols with * E CS4P28610, + E- Cs4P2Se9, and o E unassigned
peak with -22.0 ppm isotropic shift ................................................................................. 173

Figure 7-2. Structure of CS4P2Se10 (A) down the a-axis showing the [P2Se10]4‘ chains

and (B) along the b-axis with the weak interaction between the [P2Se10]4' chains
emphasized with a dashed line ......................................................................................... 179

xvi

Figure 7-3. Differential thermal analysis (DTA) of a mixture with “CS4P2Se16”

stoichiometry that was quenched rapidly from the melt. The displayed plot is for the ﬁrst
heating/cooling cycle of this mixture. The interpretation of this plot relied on the powder
X-ray diffraction pattern obtained following several heating/cooling cycles. This pattern
indicated that CS4P28610 and elemental Se were the major crystalline products. In the
displayed plot, the peak centered at ~160 °C is assigned to the glass to crystalline
transition of CS4P25610 and Se, the peak centered at at ~315 °C is assigned to the melting
of this mixture, and the peak centered at 277 °C is assigned to the subsequent
recrystallization. The exothermic peak centered at ~160 °C was not observed for the
second and subsequent heating/cooling cycles likely because the cooling rate was slow
enough that crystalline rather than glassy material was formed. Thermal events due to
pure Se were not observed perhaps because of insufﬁcient Se or because of
comelting/cocrystallization of Se with CS4P23610 .......................................................... 182

Figure 7-4. The room temperature 31P NMR spectrum of a mixture with “Cs4PZSe16”
stoichiometry that was cooled over 24 h to produce crystalline CS4P2Se10 and CS4PZSe9 .
The MAS frequency was 7 kHz. The intensities of peaks corresponding to CS4PZSe10
were used for the determination of CSA principal values of this compound .................. 185

Figure 8-1. NMR spectra of a melt mixture with the nominal ratio “PS5” showing the
increase in chemical shift and peak width as a function of temperature. Spectra at 450,

500, 550, and 600 0C were acquired with acquisition delays of 2 s and represent the sums
of 16 scans ........................................................................................................................ 200

Figure 8-2. (A) Peak 31P chemical Shifts and peak widths as a function of temperature in
a melt with the nominal composition “P85”. Chemical shifts and peak widths were

determined by ﬁtting the spectral peak to a 50%/SO% Gaussian and Lorentzian line
function. The chemical Shifts and peak widths are represented by open and closed
squares, respectively. (B) 31P longitudinal NMR relaxation rates for “PS5” as a function

of temperature .................................................................................................................. 202

Figure 8-3. High temperature 31P NMR spectra of (A) KzCu2P4Se10 at 550 °C and (B) a
sample with the nominal composition “KPSe7” at 350 °C. The spectra were acquired

with acquisition delays of 10 ms and 200 ms, respectively and represent the sums of 50
scans . .- -- .................................................................................... 203

 

Figure 8-4. EPR spectra at room temperature of rapidly quenched “PS5” samples. Prior

to quenching, the sample temperature in the furnace was (A) 1000 0C or (B) 550 °C.
Acquisition parameters were: (A) 10 G modulation amplitude, 2000 G sweep width and
(B) 0.1 G modulation amplitude, 50 G sweep width ....................................................... 208

xvii

Figure 8-5. Room temperature EPR of (A) “SnS5” quenched from 550 °C (B) “PSe5”
quenched from 550 °C and (C) K1.5Bi2.5(PS4)3 quenched from 1000 °C. Acquisition

paramaters were: (A) 1 G modulation amplitude, 40 G sweep width; (B) 1 G modulation
amplitude, 500 G sweep width; and (C) 10 G modulation amplitude, 3000 G sweep
width. Each spectrum was a single acquisition ............................................................... 208

Figure 9-1. Fit of the experimental 31P NMR chemical shift and peak width for the
sample “KPl_5Se6.3”. The ﬁt values of yo, A and I were 16.9, 0.26, and 150.7 for the

chemical shift, and 1238, 8.89, and 143.4 for the peak width. The respective R2 values
for these ﬁts were 99.953% and 99.625%. The chemical shift and peak width at T = 0 K
were 18.7 ppm and ~1300 Hz .......................................................................................... 227

Figure 9-2. Pit of the experimental 31P NMR chemical Shift and peak width for the
sample “Cs4P28e16”. The ﬁt values of yo, A and t were -45.2, 0.08, and 127.4 for the

chemical shift, and 907, 0.014, and 70.7 for the peak width. The respective R2 values for
these ﬁts were 99.971% and 99.432%. The chemical shift and peak width at T = 298 K
were -44.4 ppm and ~900 Hz ........................................................................................... 229

Figure 9-3. Plot of the data of the sample with the stoichiometry “KP1_5Se6.3” with the

ordinate as the reciprocal of temperature and the abscissa as the natural logarithm of the
change in chemical shift. The ﬁt had a slope of -3446 and a y-intercept of 8.228 which
indicated an enthalpy change of 30.2 kJ/mol. The R value was 99.743% ....................... 237

Figure 10-1. The X-ray diffraction patterns of the Cu3PSe4_xSx compounds. Each peak
could be indexed to the parent structure of either Cu3PSe4 or Cu3PS4 and the cell
parameters were subsequently reﬁned ............................................................................. 252

Figure 10-2. The cell parameters a (squares), b (circles), and c (upwards pointing
triangles) of each of the Cu3PSe4_xSx compounds, as well as the cell volume (downwards

pointing triangles). The uncertainties were smaller than the indicators used ................. 253

Figure 10-3. The powder diffraction patterns of the Fe2P2Se6_,(Sx compounds. The two
strongest peaks are off scale to Show the smaller peaks at higher angle 20 .................... 255

Figure 10-4. Raman spectra of the Cu3PSe4,xSx compounds. The spectra were sums of
64 scans ................................................................................................... . ........................ 257

Figure 10-5. The 31P NMR spectrum of the Cu3PSe4_xSx compounds. The spectra were

sums of 4 individual scans with a pulse delay pulse delay of 1000 s and the spinning
frequency was 12 kHz, and spinning sidebands were not observed ................................ 260

xviii

Figure 11-2. High temperature 31P NMR spectrum of ngPzSeG at 525 °C. The melting

point of the crystalline compound is 510 °C. The spectral features are presently
unassigned and this is the ﬁrst known example where speciation was observed in a
molten sample. ................................................................................................................. 269

xix

LIST OF TERMS AND ABBREVIATIONS

Ferroelectric — Ferroelectricity results when displacement of charged ions results in an
overall dipole moment. This situation arises when ions cooperatively move in the same
direction. This can occur spontaneously, from the inter-ionic forces, or due to the
application of a coercive electric ﬁeld.

Antiferroelectric — Antiferroelectricity results when the displacement of multiple charged
ions results in a canceling of the dipole moment that would be produced if only one of the
mobile ions were present.

Ferrielectric - Like ferroelectricity, ferrielectlicity is a dipole moment that results from
motion of ions. With ferrielectricity, there are multiple ions that move in such a way that
the dipole moment is partially, but not fully, canceled out.

Spontaneous polarization - Spontaneous polarization is when ferro- or ferri— electricity
occurs in the absence of a coercive electric ﬁeld.

Photoelectric response — When light impinges on material, some electrons are released
from the surface of that material. This effect is the photoelectric response.

SHG - Second harmonic generation is a non-linear optical process that results in the
wavelength of light being halved. Two photons with a given wavelength are effectively
combined into one photon with half of the initial wavelength.

Flux — A ﬂux is an inorganic solvent and is typically solid at room temperature. The
purpose of using a ﬂux is to increase the mobility of the reactants by providing a
relatively non-viscous liquid in which the reactants can diffuse. The composition of the
ﬂux can be tailored to alter the chemical properties of the melt, i.e. Lewis acidity or
basicity

NMR - Nuclear magnetic resonance.

EPR - Electron paramagnetic resonance.

DTA — Differential thermal analysis.

SOJT — Like the ﬁrst order Jahn Teller distortions, second order J ahn Teller distortions
involve a reduction in orbital degeneracy. The distortion in both cases occurs because
degenerate orbitals are unevenly occupied. For ﬁrst order distortions, the degeneracy

occurs between orbitals on the same atom. In the second order case, the orbitals are
located on two different, but proximate, atoms.

‘XX

CHAPTER 1

INTRODUCTION

The earliest example of high temperature synthesis of inorganic materials is ﬁ'om
the Bronze Age, perhaps 4000 BC. By trial and error, copper was mixed with other
metals such as tin and arsenic and heated over a ﬂame to form new alloys with superior
physical properties. High temperature or solid state chemistry continues to contribute to
the development of new inorganic materials including semiconductors, high performance
refractory ceramics, and superconductors. Despite the long history of solid state
chemistry, discovery of new compounds and materials is still accomplished primarily
with exploratory synthesis; i.e. trial and error variation of the elemental stoichiometry and
heating proﬁle.

Patterns of reactivity in solid state reactions have been derived empirically
following the high temperature synthesis by considering the results of a reaction
retroactively. Though this may be perceived as a limitation to the chemistry that the
chemistry can only be understood by trial-and-error, it is also a tremendous asset because
exploratory synthesis remains crucial to discovering compounds with unimagined
structures and physical properties.

Much as they have been for the past 6 millennia, many solid state materials are
produced through high temperature syntheses. These syntheses may or may not be done
at a high enough temperature to melt any of the reactants. True “solid state” reactions
that do not involve melting the reactants have pejoratively described as “heat-and-beat”
or “Shake-and-bake” methods because this type of synthesis often involves prolonged
times at high temperature, followed by grinding in a mortar and pestle, and repeating this
procedure until a homogenous ﬁnal product is formed. In binary solid state reactions, the

product compound is formed at the interface between the two solids. In order for the

reaction to go to completion, interfaces between the reactants must be replenished until
no starting material remains.l For this reason, the repeated cycle of heating and grinding
is necessary.

Though this repeat cycle is somewhat limiting because the reactions can take a
signiﬁcant amount of time to go to completion, it has the beneﬁt of providing convenient
points throughout the reaction where reaction progress can be monitored. For example,

the kinetics of the reaction between NiO and A1203 to form NiA1204 are known to obey

elementary reaction kinetics.1 This determination was feasible because it was possible to
identify and quantitate the reactants and the products at various points in the reaction
progress.

In syntheses where the reactants are in the molten state, the pathway of the
reaction is potentially much different, and the identiﬁcation of the species present is
inherently much more challenging due to the difﬁculty of applying spectroscopic or
diffraction-based techniques at high temperature. There is typically no experimental data
about the species that exist at high temperature and this lack of data is a contributing
factor to the poor understanding of the reaction chemistry whenever melting is involved
in the reaction. A few noteworthy reports have been published using high temperature X-
ray diffraction}5 Raman6'9 electron paramagnetic resonance (EPR),“"14 and nuclear
magnetic resonance (NMR)15‘20 and these experiments provided some information about
the identities of the species present during the course of the reactions at high temperature.

Relative to melt chemistry, reactions in the solution phase may be more
predictable because the molecular reactants tend to remain relatively intact in the course

of the reaction. The primary goal of the solution phase synthesis is to link these building

blocks together and to make other modiﬁcations to the molecular structure or other
modiﬁcations to the functional groups. For example, a common way to synthesize target
peptides is to sequentially link together the requisite amino acids.21 Because many solid
state reactions either use elemental starting materials or provide sufﬁcient energy to break
apart molecular precursors, there is often little predictability of the high temperature

chemistry. For example, crystalline P4S10 is a molecular super tetrahedron composed of

four comer shared PS4 tetrahedra and has been used as a reactant in a number of

syntheses,22‘24 but this supertetrahedral structural moiety has not been observed in any
solid state product compounds. Presumably, this moiety cannot exist in a high
temperature melt although this hypothesis has not been conﬁrmed experimentally.
Recently, so called ﬂux methodologies have been developed, and offer more
control over solid state syntheses that are molten. Fluxes are solvents in which reactions
may proceed. These tend to be liquids at high temperature but solids near room
temperature. Some recent examples are alkali metal chalcogenide ﬂuxes?-5 alkali metal
polychalcophosphate ﬂuxes,26 and alkali metal halide ﬂuxes”,28 The composition of
the flux can be varied to tailor the properties of the ﬂux to produce a desired compound.
For example, in reactions utilizing the alkali metal polychalcophosphate ﬂux method, the
amount of excess elemental P relative to the metal and chalcogen can be changed to

affect the stabilization of more oxidized P-containing species (i.e [PQ4]3' with a P5+) or a

less oxidized P-containing species (i.e. [P2Q6]4‘ with two P“). With this increased

relative amount of P, the oxidizing chalcogen is putatively used up before all of the P
atoms can be oxidized to the 5+ oxidation state, and instead have only been oxidized to

the 4+ oxidation state. Having such control over the reaction allows for certain aspects of

the structure of the resultant product to be targeted, and therefore the trial-and-error
methodology is somewhat enhanced.

The structures of these compounds contain an assortment of [Pny 2' (Q = S, Se,
Te) anions, and some ability to choose the type of anion present in the ﬁnal compound
has been demonstrated by varying the reaction stoichiometry, or by related ﬂux
methodologies.15a26,29 It was proposed that the products formed in these reactions may
be affected by the equilibrium between [Pny 2' anions, and therefore this particular type
of solid state synthesis may have some resemblance to solution chemistry because of the
presence of a small number of discreet species that exist at high temperature.

Chalcophosphate materials are excellent candidates for the elucidation of
chemical reactivity and reaction mechanisms due to the following: (1) it has been
proposed that the ﬂux stoichiometry affects the dynamic equilibrium between various
chalcophosphate moieties, and this equilibrium is potentially measurable or its eﬁ‘ects
observed, 26 (2) many of the reactions result in highly crystalline products that can be
characterized crystallographically and, (3) there is an abundance of 31P in the melts and
resultant compounds and so 31P nuclear magnetic resonance can be used to provide
structural information at room temperature and up to ~600 °C.-”0‘32

Various chalcophosphate compounds are interesting not only because of the
possibility of providing some information about the underlying reaction mechanisms of
solid state or melt chemistry but also because of potentially useﬁll physical properties
such as ferroelectric ordering,33'38 reversible glass-to-crystalline phase changes,15-39'42

and magnetic phenomena."'3‘45

In this work, several new compounds and their physical properties are presented
and insights into the underlying reaction mechanisms are proposed. In Chapter 2, the

new compounds CuBiPZSe6, AgBiPZSe6,and AgBiP2S6 are presented, and the Cu/Bi
antiferroelectric sublattice ordering is discussed.35 In Chapter 3, The new compounds
T13Bi3(PS4)4, TlBiP2S7, TlBiPZS6, T13Bi(PS4)2, ot-TlBiPZSe6, and B—TlBiPZSe6 are
presented, and the reaction stoichiometries required to produce each as the dominant
crystalline compound are given.46 In Chapter 4, ultra high ﬁeld (21 T) NMR
measurements of some of the reported Tl/Bi/P/ S compounds are displayed.47 In Chapter
5, the new compounds K3Bi3(PS4)4, K1.5Bi2.5(PS4)3 and KgBi(PS4)4 are presented, and
the modulated crystalline structure of K1.5Bi2.5(PS4)3 is detailed.48 In Chapter 6,
materials related to the mineral argyrodite are presented, and the composition dependant
phase change that occurs is compared to the existing Argyrodite materials where the

phase change occurs as a ﬁrnction of temperature. In Chapter 7 the new compound
CS4P28610 is presented, and this is the ﬁrst new compound that was discovered by in situ
high temperature NMR syntheses.49 In Chapter 8, evidence for the presence of
signiﬁcant concentrations of ﬁ'ee radicals in chalcogenide-containing melts is given, as
well as discussions into the potential of these radicals to play a signiﬁcant role in
synthetic reactions.50 In Chapter 9, the high temperature NMR data are compared in
reference to the available room temperature 31P NMR data, and some conclusions about

speciation in the melt are drawn. In Chapter 10, solid solutions of Cu3PSe4_xSx and

FeZPZSe6_xSx are presented, and the distribution of the chalcogen atoms is discussed.

REFERENCES

1.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

West, A. R. Solid State Chemistry and its Applications; Wiley: New York, New York,
1984; pp 6-7.

. Okamoto, Y.; Madden, P. A. J. Phys. Chem. Solids 2005, 66, 448-451.

Polonka, J.; Xu, M.; Li, Q.; Goldman, A. 1.; Finnemore, D. K Appl. Phys. Lett. 1991,
59, 3640-3642.

Sugiyama, K; Nomura, K.; Waseda, Y.; Andonov, P.; Kimura, S.; Shigematsu, K. Z.
Natwforsch. A 1990, 45, 1324-1327.

Waseda, Y.; Shinoda, K.; Sugiyama, K_; Takeda, S.; Terashirna, K.; Toguri, J. M. Jpn
J Appl Phys 1 1995, 34, 4124-4128.

Finkman, B.; Defonzo, A. P.; Tauc, J. B. Am. Phys. Soc. 1974, 19, 212-212.

Scott, J. F.; Remeika, J. P. Phys. Rev. B. 1970, 1, 4182-4185.

. Yano, T.; Shibata, S.; Maehara, T. J. Am. Ceram. Soc. 2006, 89, 89-95.

Zhang, X.; Wan, S. M.; Zhang, Q. L.; Yin, S. T.; You, J. L.; Chen, H.; Wang, A. H.
Acta. Phys. Sin-Ch. Ed. 2007, 56, 1152-1155.

Sampath, P. I. J. Chem. Phys. 1966, 45, 3519-3525.

Gardner, D. M.; Fraenkel, G. K J. Am. Chem. Soc. 1956, 78, 3279-3288.
Gardner, D. M.; Fraenkel, G. K. J. Am. Chem. Soc. 1954, 76, 5891-5892.
Koningsberger, D.; Van Wolput, J .; Rieter, P. Chem. Phys. Lett. 1971, 8, 145-147.
Warren, W. W.; Dupree, R. Phys. Rev. B. 1980, 22, 2257-2275.

Chung, I.; Do, J.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2004,
43, 2762-2764.

Maxwell, R.; Eckert, H. J. Phys. Chem. 1995, 99, 4768-4778.
Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1993, 115, 4747-4753.
Maxwell, R; Eckert, H. J. Am. Chem. Soc. 1994, 116, 682-689.

Maxwell, R; Erickson, H.; Eckert, H. Z. Naturforsch. A 1995, 50, 395-404.

20. Schmedt, A. J.; Meise-Gresch, K.; Eckert, H.; Holand, W.; Rheinberger, V. Glass.
Sci. Technol. 2000, 73, 90-97.

21. Frangsmyr, T.; Malmstrbm, B. G. Nobel Lectures, 1981-1990; World Scientiﬁc:
Singapore ; River Edge, NJ, 1992.

22. Aitken, J. A.; Canlas, C.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2001, 40,
6496-6498.

23. McCarthy, T. J.; Kanatzidis, M. G. Chem. Mot. 1993, 5, 1061-1063.
24. Loken, S.; Tremel, W. Eur. J. Inorg. Chem. 1998, 283-289.

25. Kanatzidis, M. G. Chem. Mot. 1990, 2, 353-363.

26. Kanatzidis, M. G. Curr. Opin. Solid. St. M. 1997, 2, 139-149.

27. Gencer, F.; Abel], J. S. J. Cryst. Growth 1991, 112, 337-342.

28. Abell, J. S.; Gencer, F. Physica C 1991, 185, 441-442.

29. Chung, 1.; Karst, A. L.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2006, 45,
2785-2787.

30. Canlas, C. G. Ph.D. Thesis, Michigan State University, East Lansing, MI, 2004.
31. Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.

32. Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

33. Bourdon, X.; Maisonneuve, V.; Cajipe, V. B.; Payen, C.; Fischer, J. E. J. Alloys
Compd. 1999, 283, 122-127.

34. Carpentier, C. D.; Nitsche, R. Mater. Res. Bull. 1974, 9, 1097-1100.

35. Gave, M.; Bilc, D.; Mahanti, S.; Breshears, J.; Kanatzidis, M. Inorg. Chem. 2005,
5293-5303.

36. Maisonneuve, V.; Cajipe, V. B.; Simon, A.; VonDerMuhll, R.; Ravez, J. Phys. Rev.
B. 1997, 56, 10860-10868.

37. Simon, A.; Ravez, J.; Maisonneuve, V.; Payen, C.; Cajipe, V. B. Chem. Mat. 1994, 6,
1575-1580.

38. Vysochanskii, Y. Ferroelectr. 1998, 218, 629-636.

39. Chrissaﬁs, K.; Kyratsi, T.; Paraskevopoulos, K. M.; Kanatzidis, M. G. Chem. Mat.
2004, 16, 1932-1937.

40. Gidon, S.; Lemonnier, 0.; Rolland, B.; Bichet, 0.; Dressler, C.; Samson, Y. App].
Phys. Lett. 2004, 85, 6392-6394.

41. Ohta, T. J. Optoelectron. Adv. M 2001, 3, 609-626.

42. Wachter, J. B.; Chrissaﬁs, K.; Petkov, V.; Malliakas, C. D.; Bilc, D.; Kyratsi, T.;
Paraskevopoulos, K. M.; Mahanti, S. D.; Torbrugge, T.; Eckert, H.; Kanatzidis, M. G.
J. Solid State Chem. 2007, 180, 420-431.

43. Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. Commun. 1998, 1, 55-57.

44. Odile, J. P.; Steger, J. J.; Wold, A. Inorg. Chem. 1975, 14, 2400-2402.

45. Taylor, B.; Steger, J.; Wold, A; Kostiner, E. Inorg. Chem. 1974, 13, 2719-2721.

46. Gave, M.; Malliakas, C. D.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46,
3632-3644.

47. Gave, M.; Johnson, K; Kanatzidis, M. G.; Weliky, D. P. Solid State Nucl. Magn.
Reson. 2007, Accepted. DOI 10.1016/j.ssnmr.2007.11.002.

48. Gave, M.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46, 11063-11074.

49. Gave, M.; Canlas, C.; Iyer, R. G.; Kanatzidis, M. G.; Weliky, D. P. J. Solid State
Chem. 2007, 180, 2877-2884.

50. Gave, M.; Kanatzidis, M. G.; Weliky, D. P. 2007, In preparation.

CHAPTER 2

ON THE LAMELLAR COMPOUNDS CuBiPZSe6, AgBiPzSe6 AND AgBiPZS6.

ANTIFERROELECTRIC PHASE TRANSITIONS DUE TO COOPERATIVE Cu"’
AND Bi3+ ION MOTION.

10

INTRODUCTION

Alkali metal selenophosphate compounds with group 15 elements such as
AMPZSeG (A = K, Rb, Cs, M = Sb, Bi) exhibit closely related structures with a covalently
bonded anionic [MPZSe6]' framework and alkali metals as counterions},2 There have
been several discussions in the literature regarding the inﬂuence of cation size on the
structure of the anion which highlight the general tendency that for a given stoichiometry,
larger counterions favor lower dimensionality in the the anions.2'4 Alkali metal
interactions with the 1/00[BiP2Se6]1' ﬁ‘amework are predominantly electrostatic in nature.

In the case of KSbP2Se6 three different interconvertible forms (a-, ,B- and amorphous)

were identiﬁed.1 The conversion from one form to another is connected to the ability of
K+ ions to move inside the structure. Given the same ionic charge of coinage metals (i.e.
Cu, Ag) with those of the alkali ions, it is often useful to consider the former as pseudo
alkali ions. Therefore, in general it becomes interesting to explore the relationship of
alkali metal compounds to analogous compounds with Cu and Ag. By substituting the
less electropositive Ag or Cu for an alkali metal, Signiﬁcant changes in the crystal and
electronic structure are expected not only due to the counterion effect4 but also due to the
stronger Ag-Q and Cu-Q interactions compared to the electrostatic A-Q interactions (Q =
chalcogen element). This is clearly manifested in marked reductions in size of the energy

band gap as observed for example in A0.5M1.75GeQ4 (A = Ag, Cu, Na; M = Pb, Eu; Q =

S, Se).5 Exploring coinage metal/alkali metal interactions could allow the manipulation

of structures and the possible timing of the optical and electronic properties.

11

In this context, we targeted for study CuBiPZSe6, AgBiPZSe6, and AgBisté. The
ﬁrst two phases have been described brieﬂy and were claimed to possess
photooonductive properties, however, structural characterization has not been reported.6
Furthermore, Cu and Ag in these types of compounds can become mobile and cause the

materials to undergo unusual phase transitions. This is observed in the related layered

compounds of CuInP2S6,7'11 CuInPZSe6,12'14 AgAlPZSe6,13 CuCrPZSe6,12,13 and
CuVP2S6.8-15 These compounds Show interesting physical properties such as ferri- and

ferroelectricity with high spontaneous polarization,14t16t17 as well as strong second

harmonic generation,10 and photoelectric response.6-18~19

Here we describe CuBiPZSe6, AgBiPzSe6, and AgBiP2S6 which represent the
coinage metal analogs of the ABiP2Q6 group. The stronger and more covalent
interactions of Cu+ and Ag+ ions with the framework [BiP2Se6]1' coupled with their
small Size, results in a distinctly different lamellar motif from those of the alkaline metal
counterparts. Interestingly, CuBiPZSe6 exhibits temperature induced anti-ferroelectric
phase transitions associated with motion of Cu+ ions in the structure whereas AgBiPZSe6

is already at this anti-ferroelectric state at room temperature.

EXPERIMENTAL SECTION
Reagents

Chemicals were used as obtained unless otherwise noted: bismuth chunks
(99.999%, Tellurex Inc., Traverse City, MI.), selenium shot (99.999%, Tellurex Inc.,

Traverse City, MI), sulﬁrr (sublimed ﬂowers, Alfa Aesar, Ward Hill, MA), phosphorous

12

(amorphous red, MCB Reagents, Gibbstown, NJ), electrolytic copper dust (Fisher
Scientiﬁc, Pittsburgh, PA), and Silver powder, prepared as below. Selenium and bismuth
were ground in an agate mortar and pestle to ~100 mesh.

Synthesis

Preparation of Silver Powder: Silver powder was prepared by cutting a silver
coin (31g) into centimeter Sized pieces, which were subsequently dissolved in 200mL
concentrated nitric acid by stirring overnight. The solution was then neutralized by
addition of ammonium hydroxide. Finely divided silver particles were obtained after
addition of 10g of sodium borohydride to the solution which were isolated by ﬁltration,
washed with water, and dried in a 100°C oven.

CuBiPzSe6: A mixture of Cu (0.1580 g. 2.5 mmol), Bi (0.5179 g. 2.5 mmol), P
(0.1549 g. 5.0 mmol), and Se (1.1816 g. 15.0 mmol) was loaded into a fused silica tube
and ﬂame-sealed at a reduced pressure of <10'4 mbar. The mixture was heated to 700°C
over 12h and held for 12h, followed by cooling to 50°C at a rate of 15°C/h. An
inhomogeneous grey metallic ingot (approximately 80% product, 20% Bi28e3 / Cu3PSe4)
was isolated that ﬁactured into graphite like layers composed of air-stable black plates.
Microprobe analysis of single crystals gave an average composition of

CuLoBilooPnSeig. Attempts to prepare the related compound CuBiP2S6 compound

were unsuccessful.

AgBiPzSe6: A mixture of Ag (0.2532 g., 2.35 mmol), Bi (0.4918 g., 2.35 mmol),
P (0.1466 g. 4.7 mmol), and Se (1.1205 g. 14.2 mmol) was loaded into a fused silica tube

and ﬂame-sealed at a reduced pressure of <10'4 mbar, and was heated with the same

13

proﬁle as above. An inhomogenous grey metallic ingot (approximately 80% product,

20% AgBiSez) was isolated that fractured into graphite like layers composed of air-stable

black plates. Microprobe analysis of single crystals gave an average composition of
Agl.lBil.0P2.OseS.9-

AgBiP2S6: A mixture of Ag (0.1803 g., 1.7 mmol), Bi (0.3463 g., 1.7 mmol), P
(0.1565 g. 5.1 mmol), and S (0.3203 g. 10.0 mmol) was loaded into a fused silica tube
and ﬂame-sealed at a reduced pressure of <10'4 mbar. The mixture was heated to 850°C
over 12h and kept there for 24h, followed by cooling to 50°C at a rate of 15°C/h. A
homogenous grey metallic-like ingot was obtained, but was coated with amorphous red
phosphorous powder. Upon fracturing, the ingot cleaved in a graphite-like manner to
form black plates. Microprobe analysis gave an average composition of
Agl.oBil.0P2.IS6.0- The phase was pure by powder X-ray diffraction, although it is likely
that some amorphous elemental phosphorous was present. A similar synthesis with Cu

instead of Ag did not result in the compound CuBiPZS6.

Powder X-ray Diffraction

All samples were assessed for phase purity using powder X-ray difﬁ'action.
Powder patterns were obtained using an Inel CPS 120 powder X-ray difﬁactometer with
monochromatized Cu Ka radiation (2 = 1.540598 A) operating at 40 kV and 20 mA
equipped with a position-sensitive detector with a 26 range of 0 - 120° and calibrated

with a LaB6 standard.

14

Single Crystal X-ray Diffraction

Intensity data for Single crystals of CuBiPZSe6 and AgBiP2Se6 were collected on
a Bruker SMART platform CCD diffractometer using Mo Ka radiation operating at 40
kV and 40 mA. The crystals were manually isolated ﬁom the reaction ampoule.
Individual ﬁ'ames were collected with a 10 8 exposure time and a 03° omega rotation.
The SMART software was used for data collection, and SAINT software was used for
data extraction and reduction. An analytical absorption correction to the data was
performed and direct methods were used to solve and reﬁne the structures with the
SHELXTL20 software package.

Intensity data for AgBiP2S6 was collected on a Stoe IPDS II diffractometer with

Mo Ka radiation operating at 50 kV and 40 mA on with a 34 cm image plate. The crystal
used for data collection was manually isolated from the reaction ampoule. Individual
ﬁ'ames were collected with a 60 3 exposure time and a 05° omega rotation. The X-
SHAPE and X-RED21 software packages were used for data extraction and reduction and
to apply an analytical absorption correction. The SHELXTL software package was used
to solve and reﬁne the structure. The parameters for data collection and the details of the
structural reﬁnement can be found in Table 2-1 and the fractional atomic coordinates for

each compound are shown in In all cases the atoms were reﬁned to full occupancy.

15

Table 2-1. Crystallographic data, experimental and reﬁnement Details for the CuBiPZSe6, AgBiPZSe6, and

 

 

AgBiPzS6.
CuBiP2366 0131132566 CuBiPzSe6 AgBiP2866 AgBiP236
Twp?m’°’ 298 173 97 298 298
Crystal system Trigonal Rhombohedral Rhombohedral Rhombohedral Triclinic
Space group P-31c R-3 R-3 R-3 P-l
A, A (Mo Ka) 0.071073 0.071073 0.071073 0.071073 0.071073
3, A 6.5410(9) 6.5591(10) 6.5532(16) 6.6524(13) 6.3833(13)
b, A 6.5410(9) 6.5591(10) 6.5532(16) 6.6524(13) 7.1439(14)
c, A l3.263(3) 79.385(18) 39.762(l3) 39.615(15) 9.5366(19)
0., ° 90 90 90 90 91 .89(3)
[3, ° 90 90 90 90 91.45(3)
y, ° 120 120 120 120 94.05(3)
Z 2 12 6 6 2
digerrﬁtiilns, 0.18x0.l8x O.l6x0.15x O.l6x0.15x 0.24x0.22x 0.3lx0.18x
mm 0.04 0.02 0.02 0.07 0.09
Dcalc g/em3 5.462 5.445 5.445 5.595 4.377
[1, mm'1 42.513 42.381 42.383 41.116 24.26
Rim. % 5.8 6.5 5.8 5.8 3.2
total
reﬂections/ 1678/220 8151/1626 2947/799 3000/498 4155/2136
independant
may?“ 5.5/13.9 9.1/21.2 5.0/11.7 4.5/9.9 4.2/9.2

 

a R = 20w - chl)/E|Fol- R... = [2W(lFol - IFCI)2/2WlFolzll/2

Table 2-2. Fractional atomic coordinates and U(eq) values for CuBiPZSe6 at 298 K with standard

deviations in parenthesis.

 

 

atom y 2 Uma Symmetry
Bi 1.3333 0.6667 0.75 15(1) 2d
Se 1.3322(3) 0.3031(3) 0.6180(1) 20(1) 121
Cu 1 0.75 186(8) 2a
P 1.6667 0.3333 0.6652(6) 10(2) 4f

 

16

a U(eq) = (XiZjUijai*aj*ai'aj)/3 x 1000

Table 2-3. Fractional atomic coordinates and U(eq) values for CuBiPZSe6 at 173 K with standard

deviations in parenthesis.

 

 

atom x y z Ueqa Symmetry
Bi(l) 0.6667 0.3333 0.0412(1) 17(1) 6c
Bi(2) 0.3333 -0.3333 0.1222(1) 15(1) 6c
Se(l) 0.7200(4) 0.0410(4) 0.1031(1) 19(1) 18f
Se(2) 1.0312(4) 0.7015(4) 0.0196(1) 22(1) 18f
Se(3) 0.3503(4) 0.0099(4) 0.1473(1) 16(1) 18f
Se(4) 1.0271(4) 0.3363(4) 0.0638(1) 20(1) 18f
P(l) 1.3333 0.6667 0.0558(1) 15(2) 6c
19(2) 0.6667 0.3333 0.1385(1) 11(2) 6c
P(3) 0.6667 0.3333 0.1 104(1) 13(2) 60
P(4) 1.3333 0.6667 0.0275(1) 14(2) 60
Cu( 1) 2 1 0.1369(2) 91(4) 6c
Cu(2) 0 0 0.0430(3) 178(9) 6c

 

a U(eq) = (XiZjUijai*aj‘ai'aj)/3 x 1000

Table 2-4. Fractional atomic coordinates and U(eq) values for CuBiPZSe6 at 97 K with standard deviations

in parenthesis.

 

 

atom x y z Ueqa Symmetry
Bi 0 0 0.0907(1) 10(1) 6c
Se(l) 0.3761(2) 0.3877(2) 0.1273(1) 12(1) 18f
Se(2) 0.3389(2) 1.0230(2) 0.0386(1) 9(1) 18f
Cu(l) 0.3333 0.6667 0.0535(1) 13(1) 60
Cu(2) 0.3333 0.6667 0.0761(13) 47(9) 6c
P(l) 0.6667 0.3333 0.1128(1) 9(1) 6c
P(2) 0.6667 1.3333 0.0561(1) 7(1) 6c

 

a U(eq) = (XinUijai*aj*ai'aj)/3 x 1000

Table 2-5. Fractional atomic coordinates and U(eq) values for AgBiPZS6 at 298 K with standard

 

 

deviations in parenthesis.
atom x y z Ueqa Symmetry
Bi 0 0 0.0880(1) 23(1) 60
Se(l) 0.3749(2) 0.3761(2) 0.1278(1) 26(1) 18f
Se(2) 0.3431(2) 0.0328(2) 0.0379(1) 24(1) 18f
P(l) 0.6667 0.3333 0.0555(2) 18(1) 6c
P(2) 0.6667 0.3333 0.1123(2) 18(1) 6c
Ag(l) 0.3333 .0.3333 0.0723(1) 64(1) 6c

 

a U(eq) = (EiZjUijai*aj*ai—aj)/3 x 1000

Electron Microscopy

A JEOL J SM-35C scanning electron microscope equipped with a Tracor Northern

energy dispersive spectroscopy detector was used for quantitative microprobe analysis.

Data were collected using an accelerating voltage of 25 kV and a collection time of 60 s.

17

Diﬂ'erential Thermal Analysis

Differential thermal analyses were performed with a Shimadzu DTA-50 thermal
analyzer. A portion of the ingot obtained from the reaction weighing ~ 25mg was ﬁnely
ground and sealed in a quartz ampoule under reduced pressure. An equivalent mass of
alumina was sealed in an identical ampoule to serve as a reference. The samples were
heated to 800 °C at a rate of 10 °C/min, cooled to 150 °C at a rate of 10 °C/min, reheated
to 800 °C at a rate of 10 °C/min, and then cooled to room temperature at a rate of 10

°C/min.

Solid State UV/V is Spectroscopy

Optical band gaps were determined using Kubelka-Munk theoryzz'24 on data
collected by diffuse reﬂectance UVN is spectroscopy on ﬁnely ground sample portions of
the ingot at room temperature. A background was collected before each scan using

B3804. A spectrum was collected for the region of 200-2500 nm with a Shimadzu UV-

3101 PC double-beam, double-monochromator spectrophotometer.

Method of electronic structure calculations

Electronic structure calculations were performed using the self-consistent ﬁrll-
potential linearized augmented plane wave method (LAPW)25 within density functional
theory (DFT),26’27 using the generalized gradient approximation (GGA) of Perdew,
Burke and Ernzerhof for the exchange and correlation potential.28 The values of the

atomic radii were taken to be: 1.8 a.u. for P atoms, 2.2 a.u. for Se and Cu atoms, and 2.6

18

a.u. for Bi atoms, where a.u. is the atomic unit (0.529 A). Convergence of the self-
consistent iterations was performed for 20 k points inside the irreducible Brillouin zone
to within 0.0001 Ry with a cutoff of -6.0 Ry between the valence and the core states.
Scalar relativistic corrections were included and a spin-orbit interaction was incorporated

using a second variational procedure.29 The calculations were performed using WIENZK

program3O

RESULTS AND DISCUSSION
Syntheses

CuBiPZSe6 and AgBiPZSe6 formed as black crystal plates that were manually

isolated from the reaction ampoule. It was not possible to obtain pure compounds as the

former always contained BiZSe3 and Cu3PSe4 and the latter AgBiSez as minor phases.
Presumably, some phosphorous in the reaction to produce AgBiPZSe6 was either

vaporized or remained attached to the walls of the reaction vessel, thus allowing for the
mixture of the target phase as well as a phosphorous-poor ternary phase. The bulk
samples were grey metallic in appearance and broke into graphite-like layers upon
cracking. In our hands, a procedure given for these compounds in an earlier report could
not be repeated.6

Neither the use of P2Se5 nor RbCl/LiCl ﬂux were successful in producing pure
phases of the desired compounds. In each attempt, as much as a 20% Bi28e3/Cu3PSe4 or
AgBiSe2 impurity was found by powder X-ray diffraction. Attempts to prepare

CuBiPZSe6 and AgBiPZSe6 by quenching stoichiometric melts to room temperature or in

19

ice water were not successﬁrl in producing the desired product as a single phase and
resulted only in known binary/ternary compounds. The persistent impurities are likely
the result of their extreme stability.

CuBiPZSe6 and AgBiPzSe6 are semiconductors with a room temperature energy
gap Eg of 1.2 and 1.4 eV respectively, Figure 2-1A,B. BiZSe3 present in CuBiPZSe6 has
a band gap of 0.35 eV and is likely the cause of the high residual absorbance below the
band gap of CUBIP2866. The other impurity phase, Cu3PSe4, has a band gap of 1.3 eV
and, therefore, the absorption edge is nearly collinear with the absorption edge of
CuBiPZSe6. AgBiSe2 is an impurity in AgBiPZSe6 and has a band gap of 0.5 eV which
can be seen at lower energy than the absorption edge of the AgBiPZSe6. The suitable
energy gap and high crystal symmetry could result in stable photoexcited carriers with
high mobility. Preliminary reports suggest that these materials can generate photoexcited
carriers.6 Therefore these systems could be attractive for investigations as possible solar

energy materials. The band gaps of the related KBiPZSe6 phases are somewhat larger

than those reported here at 1.6 eV for the a- and 1.3 eV for the ,6- phase.31

20

 

(A)

AMwmmmo

 

 

 

 

 

 

 

 

 

 

 

12eV
05 1 15 2 25 3 35
eV
(B)
o
i
0
Vi
n
<
14eV
05 1 15 2 25 3 35
eV
(C)
O
8
g
1Jev
05 1 15 2 25 3 35
eV

Figure 2-1 The optical absorption spectra of (A) CuBiPZSe6, (B) AgBiPZSe6 and (C) AgBiP2S6. The high
residual absorbance below the band gap in (A) is likely the result of a BiZSe3 impurity. The residual
absorbance in (B) is the result of AgBiSeZ, and the residual absorbance in (C) is likely BiZS3.

DTA analysis of sample portions of the ingot containing CuBiPZSe6 and

AgBiPZSe6 melt incongruently at 475 °C and 480 °C, respectively. A broad endothermic

21

region in both cases suggested that the title phases form a eutectic mixture with the
corresponding impurity phases. No deviation was observed ﬁ'om the baseline at the
melting points of the impurities. Powder X-ray difﬁ’action of the sample portion after
two heating cycles indicated that the title compounds were obtained in approximately the
same yield and were accompanied by the same impurities.

The synthesis of AgBiP2S6 could only be accomplished in the presence of an
additional equivalent of phosphorous. In each attempt with the stoichiometric ratio of the
elements, phosphorous was oxidized to the 5+ state and produced ternary compounds
containing [PS4]3' units (e.g. Ag7PS6, Ag3PS4). Apparently, the extra equivalent of
phosphorous helps to achieve the P4+ containing [P2S6]4' anions from which AgBiP2S6
could form. The material was found to melt at 590 °C by DTA. A parallel attempt to
produce CuBiPZS6 using extra phosphorous in the synthesis was not successful, instead
producing the very stable phases Cu3PS4 and BiPS4. A band gap of 1.7 eV was

determined spectroscopically for AgBiP2S6, Figure 2-1C.

Structure Description

Not surprisingly, the phases reported here do not adopt the same structure as their
alkali containing analogs KBIPZSC6 and RbBIPZSCé which feature pseudo one-
dirnensional structures. Because of the smaller size of Cu+ and Ag+ ions, the [BiPZSe6]1'

anion favors a two-dimensional structure that interacts strongly with the coinage metals

and “traps” them inside it. This structure is similar to other On and Ag selenophosphates

containing group 13 metals (e.g. Al3+, In3+ etc).

22

CuBiPZSe6 and AgBiPzSe6: These compounds have the lamellar FezPZSeé
structure type. This structure is a modiﬁcation of the CdIz structure where 1/3, of the
metal sites are replaced with a [PZSe6]4' anion with the P-P bond axis along the layer
normal. Each layer has the thickness of one [P2Se6]4‘ unit, and the structures contain

well deﬁned van der Waals gaps. Because of differences in the stacking sequence (i.e.

positions on the slabs relative to each other) the two compounds are not isomorphous

and, at room temperature, adopt hexagonal and rhombohedral space groups respectively.
The structures are closely related to those of CuCrPZSe6 and CuInPZSe6 with the

same [P28e6]4‘ anion lattice bridged by the metal cations. This Fe2P2S66 Slab motif is

generally adopted by compounds with transition M2+ metal ions, or a combination of
MJ’fM3+ ions (e. g. Cu/Al, Ag/In, Cu/Cr etc.) which have no valence electrons (i.e. have a
dnsop0 conﬁguration). When transition M2+ metals are replaced with main group
elements, which formally possess ns2 conﬁguration (e.g. Pb2+, Sn2+), the lamellar
M2P2Se6 structure type changes to a three-dimensional one”,33 Therefore it is
interesting that in CuBiP28e6 and AgBiPZSe6 where an intermediate situation exists i.e.
half the metal sites are Bi3+ (652) the lamellar FeZPZSe6 motif is preserved.

The [P2Se6]4' anions in the room temperature structure of CuBiPZSe6 and
AgBiPZSe6 are bridged by an ordered arrangement of alternating coinage metal and
bismuth metal centers, Figure 2-2. Each anion has three coinage metal and three bismuth

ions as its nearest metal neighbors, Figure 2-3. The Cu+ ions in CuBiPZSe6 have an

octahedral coordination environment of Se atoms, Figure 2-4A. This is unusually high for

23

Cu since it is well known that in chalcogenides, Cu prefers low coordination sites such as
tetrahedral, trigonal planar or even linear. In fact, by inspecting the results of the crystal
structure reﬁnement, there is strong evidence that the apparent high coordination
environment of Cu is misleading. As suggested by the very high anisotropic thernml
parameter of Cu+ ions, U11 = U22 = 175(11), and U33 = 209(18), it is likely that this is the
result of dynamic averaging due to atomic motion in an oversized cavity. This suggests
that the actual position of Cu is a disordered arrangement of various off-center positions.

By moving off center, the Cu+ ions can achieve a lower coordination environment.

24

..I
a...
0
mo

«4..

 

 

 

 

 

 

25

Figure 2-2 The layered structures of (A) CuBiPZSe6 and (B) AgBiPZSe6 (room temperature) as viewed

down the [110] direction. Small hollow spheres are Cu(Ag) atoms, large hollow spheres are Bi atoms,

solid black spheres are P, and solid grey spheres are Se.

 

Figure 2-3 View of a single CuBiPZSe6 layer down the c-axis ([001] direction) showing the ordered

arrangement of the Bi3+ and Cu+ ions. The same arrangement exists in AgBiPZSe6.

 

Figure 2-4 The immediate coordination environment of (A) Cu+ and (B) Bi3+ in CuBiPZSe6 at room
temperature.

26

The Bi site is nearly octahedral with Bi-Se bond lengths of 2.950(2)A, and Se-Bi-
Se bond angles ranging ﬁom 88.37(6)° to 91.79(7)°. The so-called 6s2 lone pair on Bi is
thus not stereochemically expressed, Figure 2-4. The expression of the lone pair would

be manifested as a distortion of the octahedral Bi-Se coordination. The P-Se distance is
2.187(3) A, and the P-P distance is 2.251(16) A. In contrast to the Cu analog, Ag+ ion
disorder is not evident in AgBiP28e6 where these ions are already situated in an off-
center position inside a Se-based octahedron.

The positions of Ag" ions in AgBiPZSe6 are distorted ﬁom an octahedral

environment by an elongation of three of the Ag-Se bonds, thus shifting the Ag+ ions

along the c-axis away from the center of the layer by 0.4 A, Figure 2-5A. All Ag+ ions

are ordered at room temperature, in contrast to the disorder observed in CUBIPZSC6,

described above. The Bi site is a Slightly distorted octahedron with Bi-Se bond lengths of
2.9487(15) A to 2.9550(15) A and Se-Bi-Se angles ranging from 79.70(5)° to 95.35(4)°,
Figure 2-5B The Bi3+ ions are shifted only 0.2A off the center of the octahedron in the
opposite direction along the layer normal of the Ag+ ions, Figure 2-5B. The 6s7- lone pair
of Bi seems to be stereochemically expressed, as evidenced by the slight distortion of the
Bi coordination environment from the ideal octahedron. The Ag-Se bond lengths are
2.762(2)A, with presumably weak bonding to the other side of the Se cage 3.032(3)A
away. The P-Se distances range from 2.186(2) A to 2.193(2) A, and the P-P distance is

2.251(9) A.

27

 

Figure 2-5. The immediate coordination environments of (A) Ag+ and (B) Bi3+ atoms in AgBiPZSe6 at
298K; (C) Cu and (D) Bi atoms in CuBiPZSe6 at 97K

Temperature Induced C11+ Sublattice Ordering

The oversized cavity of the Cu+ ions in CuBiPZSe6 and the possible dynamic

disorder in these cavities prompted us to investigate the possibility of ordering at lower

temperatures. For CuBiPZSe6 we reﬁned the structure at three different temperatures (97,

173 and 298 K) and indeed discovered ordering in the form of two well deﬁned anti-

ferroelectric transitions as the Cu+ ions move and settle off-center of the octahedral sites,

28

Figure 2-6.

The anti-ferroelectric order of Ag+ ions observed in AgBiPZSe6 at room
temperature is achieved in CuBiPZSe6 as the sample is cooled to < 97 K On its way to
the fully ordered state the system passes through a series of intermediate structures with
some slabs being fully ordered and others disordered. The low temperature structure

(97K) has six layers per unit cell, Figure 2-6C. Adjacent layers are shifted with respect to

each other by 1/3 along the a- or b-axes.

29

 

 

 

 

A
IIII O
V

—>

4

 

1
1
1

l
l

l.

l, 1

 

_’ 4.—

‘IIII II...
W 'r

.-

 

 

 

 

 

 

 

 

 

 

iii £13 . . —
ti‘ ’ 1* - - l
' : j j : 35311581158: 4
1? gig: gm _

 

 

 

 

 

 

Figure 2-6. The (A) 298K (B) 173K, and (C) 97K structures of CuBiPZSe6 viewed down the [110]

direction. The direction of the displacement of the copper atoms is indicated with a deﬁned arrow, and the
direction of the bismuth atoms is indicated with a solid arrow. The atoms in the layers marked with a
horizontal line have not yet ordered.

As the temperature is lowered from 298 K to 173 K a new unit cell with a six fold
superstructure along the c-axis was observed. The reﬁnement of the structure at 173 K

suggested an intermediate structure in the phase transition as seen in Figure 2-6B. The

30

Cu+ ions in Six of the twelve “CuBiPzSe6” layers move in a direction parallel to the c-
axis and towards the triangular face of the octahedral Se-cage. This motion is such that

Cu+ ions in adjacent slabs move in opposite directions and towards the same van der
Waals gap. This results in an anti-ferroelectric arrangement. In the remaining layers, the
Cu+ ions are still disordered within the selenium cage and exhibit a very high thermal
displacement parameter.

At 97 K, the transition of the Cu+ ions to a ﬁrlly ordered state is nearly complete

with about 85% of the Cu+ ions are in a well deﬁned off-center position: 1.16 A away
from the center of the octahedron along the 3-fold axis, Figure 2-5C. The remaining 15%
of Cu+ ions are still disordered within the layer as before, but have been omitted for
clarity, Figure 2-6C. Presumably at a lower temperature, all of the Cu+ ions would be
ordered.

The Cu+ ion motion creates a dipole moment inside the CuBiPZSe6 slab and a
new trigonal pyramidal coordination environment with Cu-Se distances of 2.3917(17) A,
Figure 2-5C. The long Cu---Se distances to the opposite side are 3.544(1) A. Within a
single slab, all of the Cu+ ions move the same direction, thus generating a dipole along
the direction perpendicular to that layer. At the same time, however, the Bi3+ ions within
the same layer displace in a similar manner but in the opposite direction of the Cu+ ions,
Figure 2-5D. This motion is possible presumably from the stereochemical expression of
the 682 electrons on Bi. The displacement of Bi3+ ions is 0.31 A ﬁom the octahedral
center and it creates an almost equal but opposite dipole moment. It is interesting that

these ions which bear thrice the charge of the Cu+ ions move only about one third the

31

. distance. This gives rise to a formal antiferrolectric arrangement inside the slab and
results in an intra-slab dipole moment cancellation.

Other examples where temperature dependent Cu atom ordering has been
reported, include CuInP2S6 (315K),7'11,17 CuCrPZSe6 (40K),l3 CuInPZSe6
(240K)8’13,14’17 and CuVPZSeé (20103.15 In CuInPZSe6, the motion of Cu+ and In3+
ions is such that a net dipole moment remains and a ferroelectric phenomenon is
produced. The Cu+ ions in CuInPZSe6 move 1.18 A Off center in the same direction
along the slab-normal, whereas the In3+ ions do not move appreciably. This causes a net

dipole moment in the slab which then induces the corresponding moments in adjacent

Slabs to align ferroelectrically.
At 1.16 A the magnitude of the displacement of the Cu+ ions at room temperature in
CuBiPZSe6 is comparable to that in CuInPZSe6. This large Cu+ ionic displacement in
CuBiPZSeG and in CuBiPZSe6 may be attributed to the degree of covalency between the
Cu and Se atoms. The ordering at low temperatures is likely the result of a second order
Jahn-Teller coupling involving the copper d-orbitals, and the selenium orbitals (vide
infra).34'36

A conversion to a disordered arrangement of coinage metal was not evident in
either Ag analog. It is possible, however, that above room-temperature, a disordered
paraelectric phase exists. The reasons for the differences in behavior observed between
CuBiPZSe6 (antiferroelectric) and CuInPZSe6 (ferroelectric) could be traced to the
existence of the stereochemically active lone pair in Bi3+. The stereochemical expression

of this excess electron density along a particular direction from the Bi3+ atom is the

32

driving force for shifting it off the octahedron center, a counteraction that cancels the Cu-
induced dipole moment. The inability of In3+ to behave similarly obviates the need for a
comparable shift in its octahedral cage, thereby allowing the Cu-induced dipole moments

to induce ferroelectric behavior in CuInP2Se6. This implies that in isostructural

CuMPZSe6 systems (where M is trivalent metal) we could expect ferroelectric ordering

only when M lacks a n52 lone pair and presents an exact size ﬁt in the octahedral cage.

Structure of AgBiPZS6

This compound, though layered, exhibits a different structure type, Figure 2-7.
Half of the [P2S6]4‘ units lie with their P-P axis normal to the layer as in the structures
described above whereas the other half are rotated so their P-P bond axis is nearly

parallel to the layer, Figure 2-8. The [P286]4‘ anion normal to the layer coordinates to

four Ag+ and two Bi3+ ions. The anions lying parallel coordinate to four Bi3+ and two
Ag+ ions. This is a contrast to the selenium-containing analog that has only a

perpendicular arrangement of [P2Se6]4‘ anions thus forming octahedral chalcogenide

cages, which in AgBiP2S6 are absent. Ag+ ions are instead coordinated by a highly
distorted tetrahedron of S atoms within the same layer, and have only weak Ag-S
interactions across the van der Waals gap, Figure 2-9A. The Bi3+ ions are coordinated by
six S atoms with bond distances ranging from 2.752(2) to 3.106(2) A, Figure 2-9B. This
is a highly distorted coordination environment for Bi and clearly there is one side of the
coordination sphere that is devoid of S atoms. This suggests the inert lone pair (652) of

Bi is strongly stereo-chemically expressed. Of the Bi environments presented herein, this

33

coordination geometry has the greatest degree of distortion. The S atoms spanning the
putative lone pair (S3, S5, S6) have large S-Bi-S bond angles of 129.57(6)° to 152.06(6)°.

The S-Bi-S angles that do not span the lone pair have smaller bond angles: 72.86(7)° to

88.77(6).°

 

-“.~2IB."‘.-.
P3 "is? ' A? :4

Figure 2-8. View of a single layer of AgBiP2S6. The P-P distance ranges from 2.212(4)A to 2.225(5)A,
and the P-S distances range ﬁ'om 1.993(3) A to 2.044(3) A.

34

(A)

 

Figure 2-9. The immediate coordination environments of (A) Ag+ and (B) Bi3+ in AgBisté.

The inability of AgBiP2S6 to be isostructural to its Se analog could have its

origins in the large size of Bi for the available S-based octahedral cage as well as its
increased tendency to stereochemically expresses its lone pair of electrons. Being
smaller in Size, the S atoms cannot fully satisfy the coordination shell of Bi resulting an a
highly distorted geometry. Differences in the observed bonding environment of Bi3+
ions in the sulﬁde and selenide containing structures are common as for example in Bi2S3

(distorted octahedral) vis a vis BIZSe3 (perfect octahedral). The large size of Bi3+ ions

must play a dominant role to destabilize the FezPZSe6 structure type given that the

analogous compounds with smaller In3+, AgInPZS6,13 is in fact structurally closely

related to AgBiPZSe6 and not AgBiP2S6.

Electronic structure calculations

In order to understand the differences between the antiferroelectric (the structure

35

at 97K) and paraelectric (the structure at 298K) phases of CuBiPZSe6 we performed

electronic structure calculations for both forms. The band structure results Show that both
phases are semiconductors, Figure 2-10A,B and Figure 2-11A,B. The paraelectric phase
is a direct band gap semiconductor with an energy gap of ~0.66eV at the I‘ point and also
has an indirect gap of the same size ~0.66eV with the valence band (VB) maximum along
the H-A direction of the Brillouin zone, Figure 2-10B. The antiferroelectric phase is an
indirect gap semiconductor with a much greater band-gap value of ~1 . leV, Figure 2-10A.
The substantial increase in the gap is due to the restructuring of the top VB states which
are shifted down in energy. There is, however, a modiﬁcation of all the VB states in
going ﬁom the paraelectric to antiferroelectric phase, Figure 2-10B,A. The total energy
comparison shows that the antiferroelectric phase lies lower in energy by ~0.13eV/f.u.

than the paraelectric phase (f.u.= CuBiPZSe6 ). The VB maximum occurs along the I‘-L

direction in the Brillouin zone of the antiferroelectric phase. This phase transition can be

classiﬁed as a second order Jahn-Teller distortion (see below).

36

 

 

 

 

 

 

 

‘AliQE>/‘“ /
.A

,"°j”\>/§//\f

e , E.

Eon: I
:/'

.2.eU a 7 z = r

 

 

l.

 

 

 

 

 

 

 

 

 

 

l
(B) j
1.0:
=3 ; 7? J
$0.01- , , _._ - Er:
L5 .
. A
- \/
-1.o.
Q
-7n A F“ A
“ M K r». - H

 

A

Figure 2-10. Band structure of CuBiPZSe6 : (A) antiferroelectric phase and (B) paraelectric phase.

37

 

Total DOS (B) Total DOS

MM 51 MM

(A)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(C) C” J (D) W
A ._ In T .7 A
. (B) Sep ’ (F) 389 .
Al MW .1 MW
(G) Bi? (H) RIP 1
l
l
A M M A L M
E (I) Cus (J) 1 Gus
-10 -5 0 5 10 -10 -5 0 5; ““‘1‘0
Energy (eV) Energy (eV)

Figure 2-11. Total and partial atomic DOS: (A), (C), (E), (G), (I) antiferroelectric phase and (B), (D), (F),
(H), (J) paraelectric phase.

38

In order to understand the atomic nature of the energy states we can inspect the
partial DOS. We ﬁnd that the VB states between (-4.0eV, 0eV) have strong Cu (1 and Se
p hybridized character, Figure 2-11C,D,E, and F and to a smaller degree Bi p and Se p
hybridized character, Figure 2-11G,H. This suggests that there are strong covalent
interactions between the Cu-Se and Bi-Se atoms, consistent with the chalcophilic nature
of both metals. The paraelectric to antiferroelectric transition affects all the states (Cu (1,
Se p, and Bi p). Namely, the conduction band states in the range (1.0eV, 5.0eV) consist
mostly of Bi p character with small contributions ﬁ'om P p character, Figure 2-11G,H. In
the (-3.0eV, 0eV) energy interval the Bi p states hybridize more extensively with the Se p
states in the antiferroelectric phase than in the paraelectric phase.

The DOS plots suggest that the observed electronic optical absorption observed in

the visible spectrum of CUBIPZSCé (and by extension for AgBiPZSe6) are mainly due to

excitations between ﬁlled Se based p-orbitals and vacant Bi-based p-orbital states. The
semiconducting energy gap is deﬁned ﬁ'om the band maximum occurring along the T-L
direction to the bottom of the unoccupied band at the I‘ point, leftmost arrow in Figure 2-
10A. There is however another closely lying direct energy gap at the T point which is
only slightly larger at ~1.3 eV, right arrow in Figure 2-10A. It is possible that the
relatively sharp rise in optical absorption observed experimentally at ~1.45 eV (Figure 2-
1A) may involve both direct and indirect transitions given these similarly spaced energy
gaps. In fact, because the direct energy gap is expected to exhibit a greater absorption
probability than the Slightly smaller indirect gap, we may expect decent
photoconductivity in these materials and this conclusion is consistent with the previously

reported results.6

39

Since experimentally we observe a phase transition from the para- to
antiferroelectric state in the 298-97K temperature interval, we wanted to check the
relative stability of the antiferroelectric and hypothetical ferrielectric phases at 0K The
ferrielectlic phase which is not observed is deﬁned by a model in which all Cu+ ions in
all layers move in the same direction along the c-axis so that the same Cu-Se bond
distances are achieved as in the antiferroelectric phase. Similarly, all Bi3+ ions in every
layer are displaced in the opposite direction to the Cu+ ions. We ﬁnd that the total energy
of the ferrielectric phase is higher by ~0.19 eV/f.u. This is consistent with the
experimental results showing that the antiferroelectric phase is the ground state.
Therefore, chemical bonding energetics dominates the Cu ion motion in CuBiP28e6.

It is interesting to explore the driving force behind the off-center structural
instabilities of Cu(Ag) and Bi3+ ions in these compounds. It is known that symmetry-
breaking instabilities can occur as a result of a second-order Jahn-Teller dIO-s coupling
between the conduction and valence band states.37 Calculations carried out by Burdett
et. al.34 and Wei et. a].38 for instabilities in the tetrahedral and octahedral symmetries
support this idea. In the tetrahedral and octahedral symmetries the d-s mixing is not
allowed. The mixing is allowed only if distortions ﬁ'om these symmetries occur and is
accompanied by a lowering of energy. In contrast, in CuBiPZSe6 which have trigonal
symmetry the d-s mixing is allowed even in the paraelectric (undistorted) phase, Figure
2-111. The band calculation indicated only a small increase in the d-S mixing in the range
(-l.5 eV,-0.5 eV) for the antiferroelectric phase Figure 2-111. More importantly, as can be
seen ﬁom the DOS, the states responsible for the lowering in the energy of the

antiferroelectric phase involve mostly d-p mixing which is symmetry-allowed.

40

Consequently, a second-order Jahn-Teller eﬁect associated with the d-p mixing is
the driving force for the off-center displacement of Cu+(and Ag+) ions in these
compounds, which create 3 short Cu-Se2 bonds and 3 long Cu-Sel bonds. The
stabilization of the Sel p orbitals is then achieved by shorter Sel-Bi bonds due to the
concomitant Bi displacement in the opposite direction to Cu(Ag). The displacement of
the B1 atom in the opposite direction which tends to cancel the intra- layer dipole moment
is most likely assisted by the stereochemical expression of its 6S2 lone pair. The so-
called lone pair could then occupy the space opposite to the direction of motion.

Interestingly, in CuInPZSe6, which develops a ferroelectric phase, the lack of a lone pair

in In3+ signiﬁcantly reduces the ion’s ability to shift in the opposite direction which fails
to cancel the incipient dipole moment created by the Cu+ ions. The In3+ ion moves by
0.2 A which is not adequate to cancel the much larger movement of 1.2 A of the Cu+ ions
(i.e. a 6 fold difference). This compares with a 0.3 A shift for Bi3+ against a ~4-fold
opposite shift (1.16 A) for Cu+.

Although the above considerations explain well the driving force for intra-layer
ordering, they do not address the issue of why an inter-layer antiferroelectric ordering of
the dipole moments associated with the Cu+(Ag+) and Bi3+ displacements are observed.
It is clear that due to the displacement of the Cu+(Ag+) and Bi3+ ions in opposite
directions, the net dipole moment inside a Single CUBIPZSC6 slab is considerably reduced,
but perhaps not minimized. There is likely some residual dipole within a layer since the

motion of the Cu+ and Bi3+ ions does not differ by exactly a factor of three as the

difference in formal charge would dictate. This greatly reduces the dipole-dipole coupling

41

between the layers which in general favors an inter-layer ferroelectric coupling.

In order to explore the origin of the antiferroelectric interlayer ordering (as seen in
experiment), we have carried out crystal orbital Hamilton population (COHP)39 analysis
of the Se-Se interlayer bonds. COHP partitions the band structure states into bonding,
nonbonding and antibonding character, whereas the integrated COHP (ICOHP) gives the
energy contribution of a bond to the total energy. We ﬁnd that in the para and
ferroelectric phases, ICOHP is positive decreasing the total energy (indicating anti-
bonding character), whereas in the antiferroelectric phase ICOHP is small and negative
lowering the total energy. This suggests that the stability of the Se- --Se interlayer van der
Waals contacts may be responsible for the observed antiferroelectric ordering in these

compounds.

CONCLUDING REMARKS
The coinage metal containing phases reported here are distinctly two-dimensional.
Although the dimensionality is invariant upon switching ﬁom copper to silver, substantial

differences exist between CuBiPZSe6 and AgBiPZSe6 with regard to the behavior of the
two metals in their respective compounds. The coinage metal interaction with the anionic
[BiPZSe6]1' ﬁamework is very strong relative to the corresponding alkali metal ABiPZSe6
analogs and, in fact, dominates the properties of the compounds. This includes a dramatic
reduction in the energy gap and a tendency for structural distortion. The [BiPZSe6]1'
framework adopts the lamellar Fe2P23e6 structure type motif which is apparently stable

enough to force the coinage metals, which tend to prefer lower coordination, into

octahedral sites of Se atoms. In turn, the coordination preference of the coinage metal is

42

satisﬁed by a motion towards one of the faces of the Se-based octahedron. The apparent
tendency of the [BiPZSe6]1' framework to adopt the lamellar FezPZSeé structure type
suggests that smaller alkali metals such as Na+ and Li+ may produce isostructural
compounds to CuBiPZSe6 and AgBiPZSe6. A local dynamic hopping mechanism of Cu+
ions is the basis for the antiferroelectric, second order Jahn-Teller type instability that sets
in with reduced thermal energy. A ferroelectric state in QlBiPZSe6 fails to develop
because of counteracting opposite shifts in the B13+ ion sublattice. Based on this work
we conclude that in the lamellar system Culv1P2Se6 (M=trivalent metal) several systems

could exhibit ferroelectric or antiferroelectric phenomena depending on the ability of M
to undergo local motion. It would be interesting to explore whether motional ﬁ'ustration

sets in due to competing interactions in mixed metal phases e. g. CuIn1_xBixP2Se6.

43

REFERENCES

1.

2.

Breshears, J. D.; Kanatzidis, M. G. J. Am. Chem. Soc. 2000, 122, 7839-7840.

McCarthy, T. Ph.D. Thesis, Michigan State University, East Lansing, MI, 1994.

. Chondroudis, K; Hanko, J. A.; Kanatzidis, M. G. Inorg. Chem. 1997, 36, 2623-2632.

Kanatzidis, M. G. Phosphorous Sulfur 1994, 93, 159-172.

. Iyer, R. G.; Aitken, J. A.; Kanatzidis, M. G. Solid State Sci. 2004, 6, 451-459.

. Galdamez, A.; Manriquez, V.; Kasaneva, J .; Avila, R. E. Mater. Res. Bull. 2003, 38,

1063-1072.

. Maisonneuve, V.; Cajipe, V. B.; Simon, A.; VonDerMuhll, R.; Ravez, J. Phys. Rev. B.

1997, 56, 10860-10868.

. Maisonneuve, V.; Evain, M.; Payen, C.; Cajipe, V. B.; Molinie, P. J. Alloys Compd.

1995, 218, 157-164.

. Maisonneuve, V.; Reau, J. M.; Dong, M.; Cajipe, V. B.; Payen, C.; Ravez, J.

Ferroelectr. 1997, 196, 577-580.

10. Misuryaev, T. V.; Murzina, T. V.; Aktsipetrov, O. A.; Sherstyuk, N. B.; Cajipe, V.

11.

12.

13.

14.

15.

l6.

17.

B.; Bourdon, X. Solid State Commun. 2000, 115, 605-608.

Simon, A.; Ravez, J.; Maisonneuve, V.; Payen, C.; Cajipe, V. B. Chem. Mat. 1994, 6,
1575-1580.

Bourdon, X.; Maisonneuve, V.; Cajipe, V. B.; Payen, C.; Fischer, J. E. J. Alloys
Compd. 1999, 283, 122-127.

Pfeiff, R; Kniep, R J. Alloys Compd. 1992, 186, 111-133.

Vysochanskii, Y. M.; Molnar, A. A.; Gurzan, M. 1.; Cajipe, V. B.; Bourdon, X. Solid
State Commun. 2000, 115, 13-17.

Burr, G.; Durand, B.; Evain, M.; Brec, R. J. Solid State Chem. 1993, 103, 514-518.

Kroupa, J.; Tyagur, Y. 1.; Grabar, A. A.; Vysochanskii, Y. M. Ferroelectr. 1999,
223, 421-428.

Vysochanskii, Y. Ferroelectr. 1998, 218, 629-636.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Odoulov, S. G.; Shumelyuk, A. N.; Brost, G. A.; Magde, K. M. Appl. Phys. Lett.
1996, 69, 3665-3667.

Shumelyuk, A.; Hryhorashchuk, A.; Odoulov, S. Phys. Rev. A 2005, 72, 6.

SMART, SAINT, SHELXT L V-5 and SADABS, Bruker Analytical X-ray Instruments,
Inc.: Madison, 1998.

X-AREA, X—SHAPE, and X-RED, STOE & Cie GmbH: Darmstadt, 2004.

Wendlandt, W. W.; Hecht, H. G. Reﬂectance spectroscopy; Interscience Publishers:
New York, NY, 1966.

Kortiim, G. Reﬂectance spectroscopy. Principles, methods, applications; Springer:
Berlin, Germany, 1969.

Tandon, S. P.; Gupta, J. P. Phys. Status Solidi 1970, 38, 363-&.

Singh, D. J .; Nordstrbm, L. Planewaves, pseudopotentials, and the LAPW method;
2nd ed.; Springer: New York, NY, 2006.

Hohenberg, P.; Kohn, W. Phys. Rev. B. 1964, 136, B864-B871.

Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, A1133-A1138.

Perdew, J. P.; Burke, K; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865-3868.
Koelling, D. D.; Harmon, B. N. J. Phys. C. Solid State 1977, 10, 3107-3114.

Blaha, P.; Schwarz, K; Madsen, G.; Kvasnicka, D.; Luitz, J. WIENZK, An
Augmented Plane Wave + Local Orbitals Program for Calculating Crystal
Properties, Karlheinz Schwarz Technical University: Wien: Vienna, 2001.
McCarthy, T. J.; Kanatzidis, M. G. J. Chem. Soc. Chem. Commun. 1994, 1089-1090.
Becker, R.; Brockner, W.; Schafer, H. Z. Naturforsch. A 1984, 39, 357-361.

Israel, R.; de Gelder, R; Smits, J. M. M.; Beurskens, P. T.; Eijt, S. W. H.; Rasing, T.;
van Kempen, H.; Maior, M. M.; Motrija, S. F. Z. Kristal]. 1998, 213, 34-41.

Burdett, J. K; Eisenstein, O. Inorg. Chem. 1992, 31, 1758-1762.
Pearson, R G. P. Natl. Acad. Sci. USA. 1975, 72, 2104-2106.

Pearson, R. G. Theochem-Journa] of Molecular Structure 1983, 12, 25-34.

45

37. Burdett, J. K Chemical bonding in solids; Oxford University Press: New York, NY,
1995; pp 246.

38. Wei, S. H.; Zhang, S. B.; Zunger, A. Phys. Rev. Lett. 1993,70, 1639-1642.

39. Dronskowski, R.; Blochl, P. E. J. Phys. Chem. 1993, 97, 8617-8624.

46

CHAPTER 3
WIDE COMPOSITIONAL AND STRUCTURAL DIVERSITY IN THE SYSTEM

TIIBIIP/Q (Q = S, Se) AND OBSERVATION OF VICINAL P-Tl J-COUPLIN G IN
THE SOLID STATE.

47

INTRODUCTION

Bismuth chalcogenide compounds have a great deal of structural diversity due in
part to the ﬂexibility of the coordination environment of Bi and the so-called inert lone
pair effectl'7 In addition to the common octahedral coordination of Bi, examples of
trigonal pyramidal,8 square pyramidal,9 and capped trigonal prismatic10 coordination
have been observed“-13 The bismuth chalcogenide building blocks can form various
modules which are subsequently assembled into more complex crystalline arrangements.
Examples of such assemblies can be seen in various homologous series,14’15 as well as in
the large class of minerals known as sulfosalts.16‘19 The chalcophosphate anions are a
special class of chalcogenides with a broad reactivity and structural characteristics with

general formula of [PnyF’ (Q=S, Se). They can be stabilized as alkali metal salts”-21

or a coordinating ligands to a variety of metal centers.22‘29 Bismuth can bind to various
chalcophosphate ligands, as for example in the compounds KBiP2S7,30 KBiP28631 and
K3Bi(PS4)232 which contain the [P2S7]4", [P2S6]4' and [PS4]3' anions, respectively. Each
anion can be selectively formed by slightly varying the reaction conditions, and a variety
of phases can therefore be rationally synthesized.33 In addition, chalcophosphate
compounds generally are semiconductors with medium to wide energy gaps and can
exhibit fascinating physical properties including pronounced photoconductivity, and
photorefractive properties“,35 pyroelectric and piezoelectric coefﬁcients, and interesting
electro-optic and nonlinear-properties.36 Nevertheless, they are relatively little
investigated compounds.

Thalliurn is often thought of as a pseudo-alkali metal due to its monovalency and

similar ionic radius to potassium, however, with a lower electronegativity, the

48

substitution of the former is expected to yield products that are less ionic in nature, and
therefore likely to feature stronger covalent bonding to the anionic framework. The
effect of increased covalency results in a marked reduction in the size of the forbidden
energy gap37'39 and therefore, the study of this interaction could allow for the tuning of
optical and electronic properties.

Because of the rich structural diversity of chalcophosphate phases and the
possibility of optical and electronic tunability, we targeted the system Tl/Bi/P/Q (Q = 8,
Se) for investigation. Here, we report six new thallium bismuth chalcophosphate
compounds which have 1-, 2- or 3- dimensional structures and span a range of optical
band gaps. The range of bonding modes of the chalcophosphate [Pny]l' ligands
combined with the coordinative ﬂexibility of both the T1“ and Bi3+ cations allows for
rich structural chemistry in this system.

In previous Studies, 31P solid state NMR has been shown to provide deﬁnitive
Structural information. Chemical shifts of selenophosphate compounds are strongly
dependent on the type of anion present,40 and 31P-31P dipolar coupling and 31P chemical
shift anisotropy are dependent on the type of chalcophosphate ion.41 Although NMR
spectroscopy provides information about the types of ions present in a compound, it is
known that crystallographically inequivalent P atoms often have different chemical shifts
even when only a single type of chalcophosphate ion is present. The present work
provides evidence that different chemical shifts in compounds with the same anion do not

arise from distortions within the [Pny]Z' coordination geometry, but instead are
correlated with differences in [Pny]7-' ligand-to-metal ion coordination geometry. To

our knowledge, this work also provides the ﬁrst example of splittings due to P-Tl J-

49

coupling in the solid state and the ﬁrst example of splitting from two-bond P-Tl J-

coupling.

EXPERIMENTAL SECTION
Reagents

Chemicals were used as obtained unless otherwise noted: bismuth chunks
(Tellurex Inc., Traverse City, MI, 99.999%), selenium shot (Tellurex Inc., Traverse City,
MI, 99.999%), phosphorous (MCB Reagents, Gibbstown, NJ, amorphous red), thallium
metal (ABCR, Karlsruhe, Germany, 99.999%), sublirned sulfur ﬂowers (CCI, Vernon,
CA). Selenium and bismuth were ground in an agate mortar and pestle to ~ 100 mesh.

Phosphorous was freeze dried with benzene.

Synthesis

a-TlBiPzSe6 [I I: A mixture of T1 (0.431g, 2.1 mmol), Bi (0.4408g, 2.1 mmol), P
(0.1309g, 4.2 mmol), Se (0.9989g, 12.6 mrnol) was loaded into a fused silica ampoule
and ﬂame sealed under a reduced atmosphere of ~ 10‘4 mbar. After shaking to increase
homogeneity, the tube was loaded into a protective ceramic sheath and placed in a
furnace. The mixture was heated to 700 °C over 24 h, held there for 24 h, and then
cooled back to room temperature over a period of 96 h. A homogenous grey metallic
crystalline ingot was isolated that fractured like graphite to give air-stable black plates.
Energy dispersive Spectroscopic (EDS) microprobe analysis of single crystals gave an

average composition of T11_2Bi1.0P1.9Se6.1.

50

,B-TlBiPZSe6 [II]: A sample portion of [I] was loaded into a fused silica ampoule

and ﬂame sealed under a reduced atmosphere of ~10'4 mbar. The ampoule was then
heated to 400 °C over 4 h, and held there for 22 d, and then qunched in air. A
homogenous grey metallic crystalline ingot was isolated that fractured like graphite to

give air-stable black plates. EDS Microprobe analysis gave an average composition of

TlloBil.0P2.0365.6-

TlBiPzS6 [III]: A mixture of T1 (0.195g, 1.0 mmol), Bi (0.1990g, 1.0 mmol), P

(0.1768g, 5.7mmol), and S (13.4 mmol) was loaded into a fused silica ampoule and ﬂame
sealed under a reduced atmosphere of ~10'4 mbar. After shaking to increase
homogeneity, the tube was loaded into a protective ceramic Sheath and placed in a
furnace. The mixture was heated to 700 °C over 12 h, held there for 24 h, and then
cooled to room temperature over 12 h. All ingot with a silvery surface was obtained that
could be crushed to air stable red plates. EDS Microprobe analysis gave an average
composition of T11.0Bi1_0P2.2S6.4.

773Bi3(PS4)4 [IV]: A mixture of T1 (0.321g, 1.6 mmol), Bi (0.3288g, 1.6 mmol),
P (0.0649g, 2.1 mmol), and S (0.2861 g, 8.9 mmol) was loaded into a fused silica ampoule
and treated as described above. The mixture was reacted with the same heating proﬁle as
in [I]. A deep red crystalline ingot was isolated was composed of air-stable red plates.
Microprobe analysis gave an average composition of TILIBiLOPIBSio-

TlBiPzS7 [V]: A mixture of T1 (0.2921 g, 1.4mmol), Bi (0.2980g, 1.4 mmol), P

(0.0883g, 2.9mmol) and S (0.3201 g, 9.7mmol) was loaded into a fused silica ampoule
and treated as described above. The mixture was heated to 550 °C over 3 h, held there

for 2 h, and then cooled to room temperature over 12 h. A silvery ingot was isolated was

51

composed of air-stable red plates. Microprobe analysis gave an average composition of
Tl1.013410P 1.936.9-

Tl3Bi(PS4)2 [VI]: A mixture of T1 (1.076g, 5.3mmol), Bi (0.3670g, 1.8mmol), P
(0.1083g, 3.5mmol), and S (0.4504g, 14.0mmol) was loaded into a fused silica ampoule
and and treated as described above. The mixture was then allowed to react with the same
heating proﬁle as in [I]. A silvery ingot was isolated that fractured into bright red plates.
Microprobe analysis of several crystals gave an average composition of
T13.lBi1.0P1.838.4-

Single crystals of Tl3Bi(PS4)2 with high enough quality to obtain a satisfactory
reﬁnement could not be obtained. Many crystals were screened, and all showed evidence
of twinning in the form of a rotational smearing of higher angle spots perpendicular to
(010), Figure 3-1. Presumably, each crystal that was mounted was a stack of plates with

a slight rotation from one plate to the next.

 

Figure 3-1 (A) The intensity weighted orthorhombic reciprocal lattice of Tl3Bi(PS4)3 down b"I indicating

the satellite reﬂections that could not be indexed with integral (hkl). (B) The view down the b" axis of the
orthorhombic and monoclinic cells that were used to attempt to model Tl3Bi(PS4)3.

52

It was possible to index the data with a monoclinic cell with a = 25.314(5) A, b =
6.660(1) A, c = 17.950(1) A, p = 110.67(3)°, z = 8, v = 2831.6(13) A3. A solution in

the P21/c space group was found that shows the same zig-zag [Bi(PS4)2]n3“' chains as in

K3Bi(PS4)4.32 The model did not converge as was evidenced by a relatively high
reliability factor (~12%), negative thermal parameters, and high residual peaks of
electron density. Furthermore, the model suggested a positional disorder of the Bi3+ ions

down the chain axis that resulted in a 50% occupation of the Bi3+ ions in each of two

different sites. The space group P21 was also tested to verify that the 50%/50% disorder

was not a result of an artiﬁcial symmetry restriction. A signiﬁcant improvement in
reliability statistics was not obtained.

A view of an intensity weighted reciprocal lattice suggested that the monoclinic
cell might not have been the most appropriate choice, as it was possible to index a
smaller orthorhombic cell that was close to the cell of the potassium analog32 with a
superstructure or modulation vector to describe the remaining reﬂections. An
orthorhombic unit cell with a = 8.973(2) A, b = 6.653(1) A, and c = 23.676(5) A was
used to describe the main reﬂections and the remaining reﬂections could be indexed
using a modulation vector of (1/2, 0, 1/2). This cell becomes the initial cell when
doubled in the a- and c- crystallographic directions and transformed to the monoclinic
setting. The reﬁnement of the structure in the Space group ana converged to a ﬁnal R
value of about 8.8% but a suitable super-space group could not be found that modeled the

satellite reﬂections. The K-analog has P212121 symmetry, a space group which is a

minimal non-isomorphic subgroup of ana. Although a convergent model was obtained

using P212121, it did not result in a signiﬁcant decrease in reliability statistics and again,

53

it was not possible to ﬁnd a suitable corresponding super spacegroup. The cell obtained
by doubling the a- and c- axes removes the need for the modulation vector and is the
orthorhombic translation of the above monoclinic cell, but no convergent model using
this doubled cell was found.

774Bi2(PS4)2(PzS6) [VII]: Though the structure of [VII] was previously

reported,39 the authors made no attempt to optimize the synthesis. Crystals of [VII] were
obtained from a mixture of T1 (0.874, 4.3mmol), Bi (0.3670g, 2.1mmol), P (0.1990g,
6.4mmol), and S (0.4795g, 15.0mmol) was loaded into a fused silica ampoule and ﬂame-
sealed under a reduced pressure of ~ 10'4 mbar. After shaking to increase homogeneity,
the tube was loaded into a protective ceramic sheath and placed in a furnace. The
mixture was then allowed to react with the same heating proﬁle as in [I]. A homogenous
looking silvery ingot was isolated that fractured into bright red plates. Presumably a
minor amorphous P/S containing by product is present but could not be observed by X-

ray difﬁ'action. EDS Microprobe analysis of several crystals gave an average composition

0f Tl2.313il.01’2.8311.1-

Powder X-ray Diffraction

All samples were assessed for phase purity using powder X-ray diffraction.
Powder patterns were obtained using an Inel CPS 120 powder X-ray diffractometer with
monochromatized Cu Ka radiation (it = 1.540598 A) operating at 40 kV and 20 mA
equipped with a position-sensitive detector with a 26 range of 0 - 120° and calibrated

With L336“

54

Single Crystal X-ray Diffraction

Intensity data for single crystals of [I] — [VI] were collected on a Bruker SMART
platform CCD diffractometer using Mo Ka radiation operating at 40 kV and 40 mA. A
full sphere of data was collected and individual frames were acquired with a 10 s
exposure time and a 03° omega rotation. The SMART software was used for data
collection, and SAINT software was used for data extraction and reduction. An
analytical absorption correction to the data was performed and direct methods were used
to solve and reﬁne the structures with the SHELXTL42 software package.

Intensity data for several crystals of [V1] were also collected on a Stoe IPDS II
diffractometer with Mo K01 radiation operating at 50 kV and 40 mA on with a 34 cm
image plate. Individual frames were collected with a 60 3 exposure time and a 05°
omega rotation. The X-SHAPE and X-RED43 software packages were used for data
extraction and reduction and to apply an analytical absorption correction. The
SHELXTL and JANA200044 software packages were used to solve and reﬁne the
structure. The parameters for data collection and the details of the structural reﬁnement
are given in Table 3-1. Atomic coordinates for [I]-[V] are displayed in Tables 3-2

through 3-6.

55

Table 3-1. Crystallographic data, experimental and reﬁnement details for the Tl/Bi/P/Se and Tl/Bi/P/S

 

 

compounds.
[I] [I] [III] [IV] [V]
a-TlBiP28e6 B-TlBiP2Se6 TlBiP2S6 Tl3Bi3(PS4)4 T1BiP287
Crystal system Monoclinic Monoclinic Monoclinic Monoclinic Monoclinic
Space group P21/C P21/C P21 P21/C P21
.1, A (Mo Ka) 0.071073 0.071073 0.071073 0.071073 0.071073
a, A 12.539(3) 12.250(2) 6.5967(12) 20.8189(7) 9.5028(17)
b, A 7.4990(18) 7.5518(15) 7 .3749(14) 13.2281(3) 12.290(2)
c, A 12.248(3) 22.834(5) 9.7911(18) 22.20290) 9.0771(16)
(1, ° 90 90 90 90 90
[3, ° “3.731(4) 97 .65(3) 91.270(3) 1 17.896(2) 90.561(3)
7, ° 90 90 90 90 90
Z 4 8 2 8 4
Crystal 0.26 x 0.10 x 0.10 x 0.08 x 0.10 x 0.04 0.20 x 0.08 x 0.20 x 0.08 x
dimensions, mm 0.03 0.03 x 0.03 0.06 0.02
Dealc g/cm3 5.979 6.022 4.656 4.614 4.384
[1, mm'1 52.879 53.261 36.921 38.764 33.374
Rint’ % 5.3 7 3.1 6.8 4.4
”131 ”mm“ 6233/2495 16039/5255 3068/1367 44454/12396 8406/1522
Independent
ﬁnal R/Rw, %a 4.0/9.8 4.1/7.1 2.3/5.1 4.6/9.9 1.9/4.0

 

a R = 201:0: - men/1:11:01. Rw = [2w(lFol - IFCI)2/£wlFol2]1/2

Table 3-2. Fractional atomic coordinates and U(eq) values for a—TlBiPZSe6 with standard deviations in

parenthesis.

 

 

atom x y z U69:
Bi 0.1206(1) 0.1500(1) 0.3404(1) 14(1)
Tl 0.4226(1) 0.5757(1) 0.6549(1) 20(1)
86(1) 0.4128(1) 0.7343(1) 0.4013(1) 13(1)
Se(2) 0.3113(1) -0.0086(1) 0.5730(1) 14(1)
Se(3) 0.1683(1) 0.5097(1) 0.4384(1) 12(1)
Se(4) 0.3379(1) 0.1983(1) 0.3184(1) 13(1)
Se(S) 0.0568(1) 0.3126(1) 0.1124(1) 12( 1)
86(6) 0.1144(1) -0.l790(1) 0.1911(1) 12(1)
P( 1) 0.2382(3) 0.6313(3) 0.3154(3) 10(1)
P(2) 0.2395(3) 0.0902(3) 0.6948(3) 10(1)

 

a U(eq) = (ZiZjUiJ-ai*aj *ai-aj)/3 x 1000

56

Table 3-3. Fractional atomic coordinates and U(eq) values for B-TlBiPZSe6 with standard deviations in

parenthesis.

Table 3-4. Fractional atomic coordinates and U(eq) values for TlBiPZS6

parenthesis.

 

810 m

X

Y

Z

U a

 

eq
Bi(l) 0.2013(1) 0.6539(1) 0.0796(1) 12(1)
Bi(2) 0.0618(1) 0.1509(1) 0.1714(1) 13(1)
11(3) 0.3379(1) 0.0827(1) 0.0784(1) 15(1)
Tl(4) 0.5826(1) 0.4076(1) 0.1744(1) 15(1)
Se(l) 0.0233(1) 0.8216(1) 0.0961(1) 11(1)
Se(2) 0.0517(1) 0.5042(1) 0.2177(1) 10(1)
Se(3) 0.7851(1) 0.7170(1) 0.2023(1) 11(1)
Se(4) 0.0013(1) 0.3103(1) 0.0575(1) 11(1)
Se(S) 0.1821(1) 0.2003(1) 0.1582(1) 11(1)
Se(6) 0.2782(1) 0.4943(2) 0.0345(1) 12(1)
Se(7) 0.4333(1) 0.6952(1) 0.0899(1) 11(1)
Se(8) 0.4579(1) 0.4922(2) 0.2891(1) 13(1)
Se(9) 0.2560(1) 0.8117(1) 0.1926(1) 11(1)
Se(10) 0.1992(1) 1.0057(1) 0.0323(1) 11(1)
Se(l 1) 0.4689(1) 0.2285(1) 0.0499(1) 11(1)
Se(12) 0.2768(1) 0.3163(1) 0.1541(1) 11(1)
P(l) 0.1448(3) 0.4039(4) 0.0967(2) 10(1)
P(2) 0.9177(3) 0.6268(4) 0.1559(2) 10(1)
P(3) 0.3359(3) 0.1245(4) 0.0927(2) 10(1)
P(4) 0.5979(3) 0.4038(4) 0.3468(2) 10(1)

 

a U(eq) = (Xiszijai*aj*ai‘aj)/3 x 1000

with standard deviations in

 

 

atom x y z Uecﬁ
T1 0.1955(1) 0.1512(1) 0.0002(1) 19(1)
Bi -0.7058(1) 0.4030(1) -0.5193(1) 13(1)
P( 1) -0.3002(4) 0.1370(5) -0. 1971(3) 10(1)
P(2) -0. 1918(5) 0.3652(4) -0.3242(3) 8(1)
8(1) -0.4433(4) 0.5538(4) -0.6965(3) 12(1)
S(2) 0.3277(5) -0.2817(4) 0.0071(3) 14(1)
5(3) -0.8419(5) 0.7509(4) -0.4855(3) 11(1)
3(4) 0.4177(4) 0.0532(4) 0.3228(3) 10(1)
3(5) 0.0843 (5) 0.4454(4) 0.2305(3) 13(1)
8(6) -0.9275(5) 0.4532(4) -0.2443(3) 1 1(1)

 

a U(eq) = (ZiEjUijai*aj*ai-aj)l3 X 1000

Table 3-5. Fractional atomic coordinates and U(eq) values for Tl3Bi3(PS4)4 with standard deviations in

parenthesis.

 

 

atom x y z cha
Bi(l) 0.6636(1) 0.2389(1) 0.3434(1) 28(1)
Bi(2) 0.8190(1) 0.7366(1) 0.6581(1) 26(1)
31(3) 0.8705(1) 0.3959(1) 0.0602(1) 25(1)
31(4) 0.7078(1) 0.1521(1) 0.0621(1) 27(1)
Bi(5) 0.6485(1) 0.6512(1) 0.9339(1) 28(1)

57

Table 3-5 continued

 

Bi(6)
Tl(l)
Tl(2)
Tl(3)
Tl(4)
Tl(5)
Tl(6)
3( 1)
5(2)
5(3)
5(4)
5(5)
3(6)
3(7)
3(8)
5(9)
S(10)
S(l 1)
S( 12)
S( l 3)
S(14)
S(15)
S(16)
S(17)
S(18)
S(l9)
S(20)
S(21)
S(22)
S(23)
S(24)
S(25)
S(26)
S(27)
S(28)
S(29)
S(30)
S(31)
S(32)
P( l)
P(2)
P(3)
P(4)
P(5)
P(6)
P(7)
P(8)

0.8062(1)
0.521 1(1)
0.41 13(1)
0.6264(1)
0.1379(1)
0.8937(1)
0.9909(1)
0.8109(3)
0.9366(3)
0.9140(3)
0.7853(2)
0.9160(3)
0.8550(3)
0.7405(2)
0.8604(2)
0.5846(3)
0.7015(3)
0.6031(3)
0.7349(2)
0.5954(3)
0.6471(3)
0.7693(3)
0.6508(3)
0.9023(2)
0.9639(3)
0.0107(2)
0.0823(3)
0.6924(3)
0.8024(3)
0.6243(3)
0.7240(3)
0.6951(3)
0.7826(3)
0.7668(3)
0.8727(3)
0.6049(3)
0.5044(3)
0.5750(3)
0.5174(3)
0.8626(2)
0.8443(2)
0.6544(3)
0.6643(3)
0.0149(2)
0.7123(3)
0.7776(3)
0.5054(2)

0.6101(1)
0.0410(1)
0.2394(1)
0.4588(1)
0.4876(1)
0.1771(2)
0.0377(1)
0.6054(3)
0.5496(3)
0.7088(3)
0.6790(3)
0.4535(3)
0.2831(3)
0.0954(3)
0.3179(3)
0.4968(4)
0.061 1(3)
0.2556(3)
0.1367(3)
0.5056(4)
0.6384(3)
0.9366(3)
0.7528(3)
0.5593(3)
0.7893(3)
0.3344(3)
0.5952(4)
0.0809(3)
0.1535(4)
0.1959(3)
0.3225(3)
0.8978(4)
0.9244(3)
0.8016(3)
0.6225(4)
0.8652(4)
0.8181(5)
0.0906(3)
0.3256(3)
0.6560(3)
0.3480(3)
0.1094(3)
0.6107(3)
0.3472(3)
0.1866(3)
0.8498(3)
0.8183(3)

0.4352(1)
0.1510(1)
0.3327(1)
0. 143 1(1)
0. 1505(1)
0.8316(1)
0. 1551(1)
0.0628(2)
0.0182(3)
0.5467(3)
0.91 15(2)
0.4887(2)
0.9434(2)
0.9312(2)
0.5828(2)
0.9842(3)
0.4283(2)
0.9471(2)
0.5836(2)
0.5105(3)
0.41 17(2)
0.0670(2)
0.0467(2)
0.7246(2)
0.7367(2)
0.1208(2)
0.8008(2)
0.2713(2)
0.2075(2)
0.1201(2)
0.2644(2)
0.7879(2)
0.6941(2)
0.2470(2)
0.3600(2)
0.2899(2)
0.3638(2)
0.7950(3)
0.7674(2)
0.0107(2)
0.4861(2)
0.4860(2)
0.5090(2)
0.2143(2)
0.2166(2)
0.7718(2)
0.2723(2)

26(1)
54(1)
82(1)
61(1)
55(1)
154(1)
53(1)
28(1)
35(1)
43(1)
24(1)
31(1)
39(1)
29(1)
24(1)
40(1)
31(1)
30(1)
24(1)
38(1)
25(1)
34(1)
31(1)
30(1)
41(1)
28(1)
35(1)
29(1)
38(1)
29(1)
38(1)
36(1)
37(1)
32(1)
45(1)
35(1)
43(1)
43(1)
33(1)
22(1)
22(1)
23(1)
23(1)
25(1)
25(1)
25(1)
27(1)

 

a U(eq) = (Ziszijai*aj *ai-aj)/3 x 1000

58

Table 3-6. Fractional atomic coordinates and U(eq) values for TlBiPZS7 with standard deviations in

 

 

parenthesis.
atom x y z Ueqa
Bi(l) 0.3411(1) 0.3622(1) 0.1199(1) 22(1)
Tl(2) 0.8813(1) 0.1896(1) 0.3771(1) 36(1)
P(l) 0.7782(2) 0.6028(1) 0.5192(2) 14(1)
P(2) 0.6808(2) 0.4630(1) 0.2006(2) 14(1)
S( 1) 0.1160(2) 0.2059(1) 0.1128(2) 21(1)
S(2) 0.1211(2) 0.4454(2) 0.2938(2) 21(1)
S(3) 0.4084(2) 0.1718(1) -0.0697(2) 19(1)
8(4) 0.7526(2) 0.4475(1) 0.4212(2) 19(1)
8(5) 0.5061(2) 0.5563(1) 0.1868(2) 20(1)
S(6) 0.1640(2) 0.4712(2) -0.07 55(2) 20(1)
8(7) 0.6514(2) 0.3058(1) 0.1486(2) 19(1)

 

a U(eq) = (XiZjUijai*aj *ai-aj)/3 x 1000

Energy Dispersive Spectroscopy

A JEOL JSM-35C scanning electron microscope equipped with a Tracor
Northern energy dispersive spectroscopy detector was used for quantitative microprobe
analysis. Data were collected using an accelerating voltage of 25 kV and a collection

time of 60s and results were averaged over several sample areas.

Differential Thermal Analysis
Differential thermal analyses were performed with a Shimadzu DTA-50 thermal
analyzer. A ground sample weighing ~25mg was sealed in a quartz ampoule under

reduced pressure. An equivalent mass of alumina was sealed in an identical ampoule to
serve as a reference. The samples were heated to 850 °C at a rate of 10 °C/min, cooled to

150 °C at a rate of 10 °C/min, reheated to 850 °C at a rate of 10 °C/min, and then cooled

to room temperature at a rate of 10 °C/min.

59

UV/V is Spectroscopy

Optical band gaps were determined using Kubelka-Munk theory45'47 on data
collected by diffuse reﬂectance UVN is spectroscopy on ﬁnely ground samples at room
temperature. A background was collected before each scan using BaSO4. A spectrum
was collected for the region of 200-2500 nm with a Shimadzu UV-3101 PC double-beam,

double-monochromator spectrophotometer.

NMR Spectroscopy

Room temperature solid-state NMR spectra were collected on a 9.4 T
spectrometer (Varian Inﬁnity Plus) using a double resonance magic angle spinning
(MAS) probe. Samples were spun at frequencies ranging from 5 kHz to 15 kHz in
zirconia rotors of 4 mm outer diameter and with a ~50 11L sample volume. Bloch decay
spectra were taken with a 4 [.18 90° pulse (calibrated using 85% H3PO4) and with
relaxation delays ranging from 10 to 5000 3. Each spectrum was processed with S 100
Hz line broadening and up to a 10th order polynomial baseline correction. The spectra
were referenced to 85% H3PO4 at 0 ppm.

Quantitative spectra were collected by ensuring that the pulse delay between
scans was at least 2-5 times T1. All of the compounds except [I] and [HI] had T1
relaxation times in the range of 1500-1750 sec, which is typical for the chalcophosphate
compounds previously studied."'8

Room temperature solid state NMR spectra of [VI] and [VII] were also collected

on a Bruker DSX300 NMR spectrometer using a double resonance magic angle spinning

60

(MAS) probe with zirconia rotors of 4 mm outer diameter. Bloch decay spectra were

collected with a 4.5 us 90° pulse and a relaxation delay of 2500 s.

RESULTS AND DISCUSSION
Syntheses
It was possible to form a—TlBiPZSe6 [I] by direct combination of the elements.

The compound B-TlBiP2Se6 [11], however, was not formed directly from the melt but was

instead formed after annealing the til-phase for an extended time (~22 days) below the
melting point. Presumably, the a-compound is the kinetically stable one, whereas the [3-
phase is thermodynamically stable. The rotation of the chain (vide infra) that accounts
for the difference between the two compounds presumably has a high energy of
activation, and a short (i.e. several hours to several days) anneal time is insufﬁcient to

affect such a conversion. Though the a-MBiPZSe6 (M = K, Tl) phases are isomorphous,
it is interesting to note that B-TlBiPZSe6 and B-KBiPZSe649 are different structure types,
with the latter being formed after annealing a glass formed by rapid quenching. As in the
case of the K analog, it was possible to form a glass by quenching a stoichiometric melt
of Tl/Bi/2P/6Se but after annealing the glass, the a-phase was again isolated. A T1"-
containing compound that is isomorphous to B-KBiPZSe649 was not found.

Compounds [III] - [IV] have a 1:1 ratio of T1 to Bi. By slightly varying the
reaction conditions, it was possible to prepare each as the dominant crystalline phase.
Based on the experimental information we have gathered thus far, we can produce a

schematic reactivity table that places these compounds in a greater context as shown in

Figure 3-2.

61

T|4Bi2(PS4)2(P286)
ll

 

 

 

 

 

 

TIBiPZS, 7' E T1|33i3(PS.)4
as: a “V1
Ratio of
Tl/Bi/P/S
G A. 12:06 0 9 2a,; :0: 71
a e “ .
BiPS, g; g. C TIBIP286
f [Ill]
TI3Bi(PS4)2

[VI]

Figure 3-2. Schematic of the optimized reaction conditions required to produce the known Tl/Bi/P/S
compounds as the dominant crystalline product. The stoichiometric ratio, soak time, and soak temperature
are indicated.

It was initially a challenge to reproduce TlBiPZS7 as Tl3Bi3(PS4)4 often formed.
'I‘lBiP2S7 [IV] was originally discovered reaction conditions identical to those used for 01-
TlBiPZSe6 [I] but with S instead of Se. The intended TlBiP2S6 stoichiometry was not
produced by this reaction, but rather Tl3Bi3(PS4)4 [IV] was produced as the dominant

crystalline product. Although its presence could not be conﬁrmed with powder X-ray
diffraction, '1‘13Bi3(PS4)4 must have been accompanied by a glassy phase composed of
the two extra equivalents of both P and S. Upon further investigation, it was determined
that the additional P is not necessary to produce compound [IV] in pure form, but that
some additional S is essential. The optimized reaction conditions for producing

compound [IV] were determined to be 3Tl/3Bi/4P/17S, that is, an extra equivalent of S.

62

It is possible that the extra equivalent of S in the form of S3” chains reduced the acidity

of the melt wherein compound [IV] could be formed. The metal to chalcogenide ratio is
extremely important in the melt, and changes of as little as 0.2 stoichiometric equivalents
were observed to have the marked effect of producing a completely different crystalline
compound.50 A stoichiometric combination of 3T1/3Bi/4P/ 168 did not produce

compound [TV] but gave a product that contained a signiﬁcant amount of TlBiP2S6 [111]
along with BiPS4 as a minor phase. In terms of a 1:1 leBi ratio, this results in a

deﬁciency of both P and S relative to the stoichiometry of the dominant crystalline
compound, suggesting that the S was consumed by oxidizing the metals before P could
be fully oxidized to the 5+ state. To produce compound [III] as the dominant crystalline
product, it was therefore necessary to employ a ﬂux with limited acidity to stabilize the

[P2S6]4' unit. A reaction with the stoichiometric ratio of Tl/Bi/6P/14S was subsequently

found to produce compound [III] as the sole crystalline product. Attempts to remove the
glassy impurities by washing with ethylenediamine, methanol, and ether resulted in a
product that was amorphous by powder X-ray diffraction suggesting that compound [III]
decomposes under such conditions.

TlBiP2S7 [V] was formed from a stoichiometric mixture of the elements, but with

a heating proﬁle that was both lower in temperature and shorter in time than any of the
other reactions. When slowly cooled, a mixture of Tl/Bi/2P/7S produced mainly

compound [IV] but also some BiPS4. Presumably, compound [V] is more kinetically

stable and a more rapid cooling rate is necessary to avoid the more thermodynamically

stable compounds.

63

It was also possible to prepare TI4B12(PS4)2(P2S6)39 [VII] in pure form by adding
an additional equivalent of P (i.e. 2Tl/Bi/3P/7S). The overoxidation of P to 5+ was
prevented by the extra phosphorous, and it was possible to stabilize both the P4+ and P5+

species in the compound.51

Structure Description
a-TlBiPZSe6 [1] and ,B-TlBiPZSe6 [II]: a-TlBiP2S66 [I] is isomorphous with o.-
KMP2Se652i53 (M = Sb, Bi), Figure 3-3. The structure is built up of [MPﬁed‘ layers

that are separated by Tl+ ions with a mean Tl-Se distance of 3.34 A.

 

 

 

 

 

Figure 3-3. a-TlBiP28e6 [1] down the [010] direction showing the lamellae separated by a layer of Tl+ ions
with the double-chain moiety common to [I] and [11] emphasized by the rectangular dotted outline. Small
black spheres are P, large black spheres are Tl, large white spheres are Bi, and shaded spheres are Se.

Comparatively, the K+ analog has K-Se bond distances averaging 3.36 A. Each T1+

cation is 7-coordinate with a distorted monocapped octahedral geometry, Figure 3-4A,

and each bismuth is 6-coordinate with four different [PS4]3' units, Figure 3-4B. One of
the [PZSe6]4‘ anions is tris-chelated to Bi3+ (Se(3), Se(4), Se(5)) with Bi-Se bonds
ranging from 2.8506(14) A to 3.1253(14) A in length. Neighboring BiS6 are connected

to form a chain along down (010) by bridging Bi-Se(6) bonds at a distance of 3.1253(14)
A.

 

Figure 3-4. (A) The coordination of T1+ in a-TlBiP28e6 [I] and (13) the coordination of 1313+ showing tris-
chelation by one [P28e6l4‘ and monodentate bonding by each of the other three [P28e614' anions.

Finally, a double chain is assembled through the Bi-Se(2) bond where Se(2) is on a
[PzSe6]4' ion in a neighboring chain along the (001) axis. Neighboring double chains are
assembled by the same Bi-Se(2) bond. The layer assembled from the double chains is
separated ﬁ'om an identical layer by T11" ions. The P-P bond distance in the [PZSe6]4'
anion is 2.228(4) A, and the P-Se bond distances range from 2.155(3) A to 2.222(3) A.
There is one Se atom on the [P2Se6]4' anion that is terminal and interacts only with Tl+

ions. Table 3- 7 contains selected bond lengths and angles for compound [I].

65

Table 3-7. Selected bond lengths and angles for a-TlBiPZSe6 with bond lengths (A) and angles (9) with

standard uncertainties in parenthesis.

 

Bi—Se5 2.8506(14) SeS—Bi—Se4 7590(4)
Bi—Se4 2.8676(14) Se5—Bi—Se3 8642(4)
Bi—Se3 2.9148(12) Se4—Bi—Se3 8288(3)
Bi—Se6 3.0531(13) Se5—Bi—Se6 8018(4)
Bi—Se6i 3.0919(13) Se4—Bi—Se6 8089(3)
Bi—Se2 3.1253(14) Se3—Bi—Se6 161.03(4)
Ses—Bi—Se6i 79.78(3)
Tl—Se3 3.2615(14) Se4—l3i—Se6i 146.07(3)
Tl—Sel 3.2814(15) Se3—Bi—Se6i 7225(3)
Tl—Se4V 3.3277(14) Se6—Bi——Se6i ll7.86(3)
Tl—Se4iv 3.3476(14) Se5—Bi—Se2 l49.99(4)
Tl—SeliV 3.3590(13) Se4—Bi—Se2 7412(4)
Tl—Selvi 3.3843(14) Se3—Bi—Se2 9125(3)
Tl—Sezvii 3.3974(14) Se6—Bi—Se2 9368(3)
Tl—Sezviii 3.6999(15) Se6i—Bi—Se2 127.80(4)
Sel—Pl 2.155(3) Sel—Pl—Se6vii 113.6202)
Se2—P2 2.160(3) Sel—Pl—Se3 114.65(l4)
Se3—Pl 2.220(3) Seovii—Pl—Ses 111.9403)
Se4—l>2ii 2.197(3) Se2—l>2—Se4v ll6.69(l4)
Ses—Pzii 2.222(3) Se2—P2—Se5" 116.07(l4)
Se6—1>1iii 2.203(3) Se4V—P2—Se5" 105.46(12)
l>l—l>2ii 2.228(4) Se2—P2—P1V 108.5703)
Sel-—Pl—l>2ii 108.3005)
Se6vii—Pl—P2ii 102.37(l5)
Se3—l>l—-l>2ii 104.7103)
Se4V—P2—P1V 103.4705)
SeSV—PZ—Pl" 105.3204)

 

(1)-x, 0.5+y, 0.5-z; (ii) x, 0.5-y, -O.5+z; (iii) x, -1+y, 2; (iv) l-x, l-y, l-z; (v) x, 0.5-y, 0.5+z;
(vi) x, 1.5-y, 0.5+z; (vii) x, 1+y, 2; (viii) l-x, 0.5+y, 1.5-2; (ix) x, 1.5-y, -0.5+z; (x) l-x, -0.5+y, 1.5-z;
(xi) -x, -0.5+y, 0.5-z.

The compound B-TlBiP2Se6 [II] crystallizes in the monoclinic spacegroup P21/c.
The structure is built up from the same chains described as those in a-KBiPZSe6 and
a-TlBiPZSe6. The difference between the 01- and B-TlBiP28e6 compound is that in the
latter, the layer is broken apart by a rotation about the (010) axis of [BiPzSe6]44‘

double chains such that l-dimensionality is preserved, Figure 3-5.

66

 

Figure 3-5. Assembly of the chain common to both [I] and [11] viewed down the (001) direction.

There are two such chains per unit cell with one centered at the origin, and the other
centered at 1/2 along the c-axis. Each chain is separated along the (100) direction by Tl+
ions, Figure 3-6. Tl(l) forms a distorted bicapped octahedron. Seven of the Tl-Se
distances are relatively short with an average distance of 3.40A. The remaining
interaction between T1( 1) and Se(8) is weak with a bond length of 3.5813( 17) A. Tl(2) is
also a distorted bicapped octahedron, again with seven short and one longer Tl-Se
interaction. The average length of the shorter bonds is 3.34 A, and the longer Tl(2)-

Se(12) interaction is 3.776105) A, Figure 3-7A.

67

 

e
'

4 ﬂ.
eﬁyg as

X'—

O
(A J-~ A .‘.

- -,V/%£‘V ~ -PV/'
' "I
I’ a ‘

 

 

 

Figure 3-6. ﬂ-TlBiPZSeé [II] down the (010) axis showing emphasizing the weak inter-chain interaction
with a dashed line, and double-chain moiety common to [I] and [11] emphasized by rectangular outline.

 

Figure 3-7 (A) The coordination of Tl+ in B-TlBiP2Se6 and (B) the coordination of Bi3+ showing the weak
inter-chain Bi-Se interaction with a dashed line.

Each Bi3+ center is tris-chelated by Se atoms with an average Bi-Se bond length
of 2.88 A, Figure 3-7B. Two chains are then connected together to form a double-chain
with a Bi(1)-Se(6) bond length of 3.1259(16) A, and the chains interact weakly with

neighboring double chains with Bi(2)-Se(2) lengths of 3.2419(16) A and Bi(2)-Se(3)

68

3.2627(18) A, Figure 3-7B. Table 3-8 contains selected bond lengths and angles for
compound [II].

Table 3-8. Selected bond lengths and angles for B- TlBiPZSe6 with bond lengths (A) and angles (9) with
standard uncertainties in parenthesis.

Bil—Se7 2.837404) Se7—Bil—Se9 76.47(5)
Bil—Se9 2.840606) Se7—Bi1—Se10 83.560)
Bil—SelO 2.867103) Se9—Bil—Se10 8670(4)
Bil—Sel 3.097404) Se7—Bil—Se1 147.240)
Bil—Se6 3.1259(16) Se9—Bil—Sel 7983(4)
Bil—Se12 3.135204) SelO—Bil—Sel 7273(4)
Bi2—Se4 2.871706) Se7—Bi1—Se6 7268(4)
Bi2—Se2 2.8786(13) Se9—Bil—Se6 149.08(4)
Bi2—Se5 2.985804) SelO—Bil—Se6 91.840)
Bi2—Se12 2.988504) Sel—Bil—Se6 129.040)
Biz—Selii 3.122603) Se7—Bil—Se12 79.900)
1312—Se2iii 3.2419(16) Se9—Bil—Se12 80.290)
Bi2—Se3i" 3.2627(18) SelO—Bil—SelZ l60.93(4)
Sel—Bil—Se12 117.98(4)
Tll—SeS 3.2565(15) Se6—Bil—Se12 9223(4)
Tll—SelOi 3.280207) Se4—Bi2—Se2 85.740)
Tll—Se7" 3.306503) Se4—Bi2—Se5 73.890)
Tll—Sel lvi 3.3356(17) Se2—Bi2—Se5 8019(4)
Tll—SellVii 3.337704) Se4—Bi2—Se12 80.260)
Tll—Se6vi 3.376404) Sez—Biz—Selz 75.120)
Tll—Se3" 3.379207) Se5—Bi2—Se12 l4530(3)
Tll—Se8iii 3.5813(17) Se4—13i2—Se1ii 78.940)
T12—Se8 3.265807) Se2—l3i2—Se1ii 155900)
T12-—-Se1 1 3.284607) Se5—13i2—Selii 77.760)
T12—Se7 3.292805) Selz—Biz—Selii 119.710)
T12—Se8iv 3.3008(14) Se4—13i2———.Se2iii 139.900)
T12——Se5viii 3.3434(14) st:2—1312—-Sc2iii 8919(3)
T12—Se3 3.405604) Ses—Biz—Se2iii 66.050)
T12—Se9iv 3.473308) Se12—13i2—Se2iii 136.31(4)
T12—Se12 3.776105) Selii—Biz—Seziii 90.660)
T12—Se6ix 3.887608) Se4—Biz—Se3iV 141.720)
Se2—Bi2—Se3iv 6525(4)
Sel—P2vii 2.194(3) Sc5—13i2—Se3iv 121.300)
Se2—P2vii 2.219(4) Se12—l312—Se3iv 68.890)
Se3—P2 2.165(4) Selii—13i2—Se3iv 135.98(3)
Se4-—Pl 2.220(3) Se2ilii—1312—Sc3iv 67.510)
SeS—Pl 2.172(4)
Se6—P1i 2.159(4) Se6i—P1—Se5 118.1505)
Se7—l>4xii 2.205(4) Se6i—Pl—Se4 115.3308)
Se8—P4 2.127(4) Se5—Pl—Se4 106.6103)

69

Table 3-8 continued

 

Se9—1>4xii 2.220(3) Se6i—Pl—P2vii 107.3804)
Solo—P3x 2.213(4) SeS—Pl—P2Vii 103.1407)
Sell—P3 2.156(3) Se4—Pl—P2Vii 104.7605)
Se12—P3 2.204(3) Se3—P2—Se1Viii 1151303)
Se3—P2—Se2viii 111.0807)
Pl—szii 2.230(4) Selviii—Pz—Sezviii 113.5204)
P4—P3xii 2.246(4) Se3—P2—Plviii 108.3506)
Selviii—Pz—Plviii 104.1107)
Sczviii—Pz—Plviii 103.6103)
Sel 1—P3—Se12 112.3803)
Sel 1—-P3—Se10ii 114.8408)
Se12—P3—Se10ii 112.18(14)
Sell—l>3—I>4iv 108.4405)
Se12—1>3—l>4iv 102.6206)
Seloii-—l>3—I>4iv 1052904)
Se8—P4—Se7iv 115.78(15)
Se8—P4—Se9i" 118.3108)
Se7iV—1>4—Se9iv 1051603)
Se8—1>4—l>3xii 110.0804)
Se7i"—-l>4-—l>3xii 101.9307)
Se9iV—P4—P3xii 103.7305)

 

(i) -x, l-y, -z; (ii) x, -1+y, z; (iii) -x, -0.5+y, 0.5-z; (iv) l-x, -0.5+y, 0.5-z; (v) -l+x, -1+y, 2;
(vi) -x, -y, -z; (vii) -1+x, y, 2; (viii) 1+x, y, 2; (ix) l-x, l-y, -z; (x) x, 1+y, z; (xi) -x, 0.5+y, 0.5-z;
(xii) l-x, 0.5+y, 0.5-z; (xiii) 1+x, 1+y, z.

TlBiP2S6 [III]: TlBiPZS6 [III] crystallizes in the monoclinic spacegroup P21. The
structure is closely related to that of K(RE)P28e6 (RE = Y, Gd)54 and isomorphous to B-
KMP2Se6 (M = Sb, Bi)49 and KBiP2S6.31 The structure is composed of [Bi(PZS6)]1'

layers separated by Tl+ cations, Figure 3-8.

70

 

Figure 3-8. The structure of TlBiP2S6 [11]] viewed down the (100) axis showing the lamellar stacking

motif.

The Tl+ ions are spherically coordinated by nine S atoms with an average Tl—S
interaction of 3.35 A, Figure 3-9A. The Bi3+ atoms are tris-chelated by a [P2S6]4' anion
with Bi-Se distances ranging from 2.717(3) A to 2.805(3) A. A second [P286]4' anion is

chelated to the Bi3+ cation with Bi-S bond distances of 2.741(3) A and 3.114(3) A. Two
other [PZS614' anions are chelated to the Bi3+ cation with Bi-S distances of 3.125(3),

Figure 3-8B. The P-P distance is 2.221(4) A.

71

(A) (B)

 

Figure 3-9. (A) the Tl+ coordination environment, and (B) the coordination environment of Bi3+ in
TlBiPZSe6 [111].

Table 3-9 contains selected bond lengths and angles for compound [III]. The
difference between the Se-containing compounds (i.e. [I] and [11]), and compound [III] is
that in the former, the Bi3+ ions undulate in a direction perpendicular to the chain
between two different positions down the length of the chain, so that when the chains

assemble they form a slab that has the thickness of two [PZSe6]4' capped Bi3+ ions. In

compound [III], the Bi3+ ions do not undulate, and therefore the resulting slab is only one

Bi3+ coordination environment thick.

Table 3-9. Selected bond lengths and angles for TlBiPZS6 with bond lengths (A) and angles (9) with
standard uncertainties in parenthesis.

 

Tl—szi 3.186(3) Sl—Bi—S3 84.7800)
Tl—Ss 3.225(3) Sl—Bi—SSVii 7288(9)
Tl—S5ii 3.261(3) S3—Bi—35vii 8106(9)
Tl—S2 3.310(3) S1—13i—S4v 8134(9)
T1—S6iii 3.340(3) S3—Bi—S4" 7690(9)
Tl—S6iv 3.356(3) ssvii—l3i—S4v l47.35(9)
Tl—S2v 3.486(3) S1—Bi—S6 145.26(9)
T1—s1vi 3.499(3) S3—Bi—S6 6785(8)

72

Table 3-9 continued

 

Tl—S4 3.527(3) S5vii—l3i—S6 120.66(8)
S4V—Bi—86 7217(8)
Bi—Sl 2.717(3) Sl—Bi—S4Vii 8108(9)
Bi—S3 2.741(3) S3—Bi—S4vii 157.26(8)
131—85vii 2.805(3) S5Vii—Bi—S4vii 7782(8)
Bi—S4V 2.895(3) S4V—Bi—S4Vii 118.04(6)
Bi—S6 3.114(3) so—Bi—S4vii 131.l9(8)
13i—S4vii 3.125(3)
s2V—l>l—Ssii 1 16.2609)
P1—S2V 1.967(4) S2V—P1—Slx l 18.5(2)
Pl—ssii 2.038(4) ssii—Pl—s1x 106.4(2)
Pl—Slx 2.060(4) S6iV—P2—S4V 1 14.0908)
1>2—S6iv 2.003(4) S6iV—P2—s3x 112.0(2)
P2—S4V 2.035(4) s4V—P2—s3x 112.2109)
I>2—s3x 2.062(4)
SZV—Pl—P2 109.6(2)
P1—P2 2.221(4) ssii—Pl—Pz 101.5908)
SIX—Pl—Pz 102.3508)
S6iV—P2—Pl 108.3707)
S4V—P2—Pl 1055209)
S3x—P2—Pl 103.8006)

 

(i) l-x, 0.5+y, -2; (ii) -x, -0.5+y, -2; (iii) -1-x, -0.5+y, -z; (iv) 1+x, y, 2;
(v) -x, 0.5+y, -z; (vi) -x, -0.5+y, -l-z; (vii) -1+x, y, -1+z; (viii) -1+x, y, 2;
(ix) -1-x, 0.5+y, -1-z; (x) -1-x, -0.5+y, -1-z; (xi) -x, 0.5+y, -1-z; (xii) l-x, -0.5+y, -z;
(xiii) x, -1+y, z; (xiv) -1-x, 0.5+y, -z; (xv) 1+x, y, 1+z.

Tl3Bi3(PS4)4 [IV]: T13Bi3(PS4)4 [1V] crystallizes in the monoclinic space group
P21/c. Crystal twinning in a merohedral fashion was a major difﬁculty in solving and

reﬁning the structure. Data sets on several crystals were collected but could not be
successfully reﬁned as large residual electron densities remained. Careful examination of
the reciprocal lattice revealed a second twin domain rotated 180° along the a-axis. When
the corresponding twin law of

a' 100 a
b'=0-10-b (1)

73

was applied, the R-value dropped from 28% to 5%. This led to small residuals in the
electron density map and with a twin fraction of 43.2(1)%.

The 3-dimensional structure of the compound can be seen as an assembly of
[Bi3(PS4)4]3' slabs Stacked along the b-axis, Figure 3-10A. The Slabs are interconnected

with Bi-S bonds and form channels in each direction which are occupied by T1 atoms.

 

 

Figure 3-10. (A) T13Bi3(PS4)4 [IV] viewed down the a-axis showing the stacking of the T13Bi3(PS4)4
along the b-axis with a slab emphasized by rectangular outline, (B) an individual slab of TlgBi3(PS4)4 as

viewed down the c-axis showing the empty channels for the Tl+ ions and (C) the coordination
environments of Bi( 1) and Bi(2).

74

The T13Bi3(PS4)4 slabs consist of trigonal prismatic and square pyramidal Bi3+
atoms and [PS4]3‘ anions, Figure 3-10C. Each trigonal prismatic Bi3+ atom (Bi(l) and
Bi(2)) is surrounded by four [PS4]3' anions forming a cross-shaped Bi-(PS4)4 unit. These

units are connected with the trigonal prismatic Bi3+ atoms (Bi(3), Bi(4), Bi(5) and Bi(6)),
to form the slab. Bi(l) and Bi(2) are coordinated by six S atoms to form a distorted
trigonal prism Figure 3-10D. The geometry is distorted because of the stereochemically

active 6s2 lone pair of Bi3+. Two [PS4]3' tetrahedra share one edge with the rectangular
planes of the trigonal prism and the other two [PS4]3' anions share a comer with the two

remaining vertices of the prism. All four [PS4]3' anions share one edge with the square
pyramidal Biss units Bi(3), Bi(4), Bi(5) and Bi(6). The remaining S atoms of the two
comer shared [PS4]3' units on the pyramid are attached on two Bi85 units of a different
slab.

The Bi-S distances in the trigonal prism vary from 2.709(5) A to 2.922(4) A for
81(1) and 2.791(5) A to 2.883(5) A for Bi(2). The four equatorial bonds in the BiS5 units
range from 2.768(4) A to 2.996(5) A and the shorter axial distance from 2.627(5) A to

2.706(4) A. The P-S distances in all [PS4]3' units range from 2.007(6) A to 2.064(6) A.

The coordination number of Tl+ ions varies from six up to eight S atoms, Figure
3-10. In particular, Tl(l) is surrounded by eight S atoms forming a distorted
dodecahedron with Tl-S distances from 3.159(5) A to 3.698(5) A, Figure 3-9A. The Tl(l)
dodecahedron shares a comer through the bridging S(29) atom with the six coordinate
Tl(2) center. The distances in this octahedron vary from 3.071(5) A to 3.585(5) A, Figure

3-10B. A corner of the distorted T1(2)S6 octahedron is shared with the seven-coordinate

75

Tl(3) which has Tl-S distances from 3.084(5) A to 3.507(5) A, Fig 11C. Tl(4), Tl(5) and
Tl(6) are located in a different channel from the other Tl atoms. T 1(4) is surrounded by
eight 8 atoms with distances from 3.153(6) A to 3.772(6) A, Fig 11D. Tl(5) and Tl(6) are
six coordinated with one corner shared S(20) atom and Tl-S distances in the range of
3.085(4) A to 3.717(5) A, Figure 3-9E, and 3-9F, respectively. Selected bond lengths and

angles for compound [IV] are given in Table 3-10.

(A)

 

Figure 3-11. The coordination environments of the crystallographically unique Tl+ ions in Tl3Bi3(PS4)4
[TV]-

76

Bil—s31i
Bil—823
Bil—S321
Bil—821
1311—07i
Bil—810
Bil—SI 1i
Biz—S27i
1312—318
B12426
1312—517
13i2—S15i
I312—s1iv
Bi3—Sl9viii
Bi3—820"
Bi3—s2
Bi3—S6
1313—88ii
Bi3—Sl
1313—s12ii
Bi4—s24ii
1314—51 1

1314—312ii
Bi4—s22
1314—513ii
Bi4—S8ii
Bi4——SlS
Bis—s3oii
13i5—S14ii
Bis—$16
Bis—S29ii
13i5—S9xi
I3i5—S10xii
13i6—Sz8xiv
Bi6—S3
13i6—S4i
Bi6—S5
13i6—325i
13i6—s7i
Bi6—Sl4

T11-S32i
Tll—sz4ii

2.792(6)
2.816(6)
2.831(6)
2.855(5)
2.875(5)
2.887(5)
3.101(6)
2.711(6)
2.776(6)
2.817(5)
2.875(5)
2.911(5)
2.918(5)
2.703(5)
2.767(5)
2.845(5)
2.877(5)
2.905(5)
3.050(5)
3.130(5)
2.667(5)
2.820(5)

2.850(5)
2.894(5)

2.943(6)

3.007(5)
3.104(5)

2.687(6)
2.819(5)
2.820(5)
2.891(6)
2.921(6)
3.035(5)
2.634(6)
2.764(5)
2.829(5)
2.892(5)
3.000(6)
3.023(5)
3.123(6)

3.150(6)
3.265(6)

77

S3li—Bi1—sz3
s31i—Bil—s32i
s23—Bil—s32i
S3li—Bil—S21
823—Bil—821
s32i—Bil—Szl
s31i——13il—S7i
S23—13i1—s7i
S32i—13i1—S7i
321-—13i1—s7i
s3li—Bil—SIO
S23—Bil—SlO
s32i—Bil—Slo
SZl—Bil—SIO
s7i—Bi1—SIO
S3li—Bil—Sl 1i
823—1311—311i
S32i—13i1—s11i
SZl—Bil—Slli
S7i—13i1—Slli
SlO—Bil—Sl li
S27i—1312—Sl8
827i—1312—s26
Sl8—Bi2—SZ6
s27i—1312—s17
Sl8—Bi2—Sl7
SZ6—Bi2—Sl7
S27i——1312—315i
818—812—315i
S26—1312—515i
Sl7—13i2—515i
sz7i—1312—SIiv
S18—13i2—Sliv
$26—13i2—Sliv
Sl7—13i2—SIiv
S15i—l3i2—Sliv

S19Viii—Bi3—s20V

S19viii—l3i3—s2
SZOV—Bi3—82
s19viii—Bi3—S6
SZOV—Bi3—S6
S2—Bi3—S6

S19viii——l3i3—S8ii

Table 3-10. Selected bond lengths and angles for Tl3Bi3(PS4)4 with bond lengths (A) and angles (9) with
standard uncertainties in parenthesis.

81.2(2)
71.59(16)
114.29(17)
l30.47(17)
70.65(15)
83.77(17)
72.6305)
81.47(l6)
137.63(17)
137.76(17)
147.2(2)
l3l.56(18)
91.8806)
72.84(14)
106.57(14)
80.84(19)
155.09(15)
75.83(15)
l34.22(l4)
76.77(15)
67.4605)
106.1308)
72.87(15)
87.6(2)
79.90(17)

70.37(16)
138.48(16)

103.00(17)
124.3909)
146.6008)
69.55(l4)
139.15(16)
87.1706)
69.23(14)
140.20(l6)
100.2404)
73.9206)
78.8106)
111.1809)
85.9108)
149.2007)
86.6807)
76.7405)

Table 3-10 continued

Til—s31xiii
Til—$13iii
Til—s21
Tll—Sl4iii
T11—s29xi
Tll—S30iii
T12—s29iii
T12—325xiii
T12—s32i
T12—827xv
T12—S16xv
T12—S1 1i
T12-—T13iii
T12—T11xv
Tl3—823
T13—S30iii
T13—S9
Tl3—S26xvi
T13—suii
T13—325i
T13—S24ii
Tl4—Sl9
Tl4—SZ7XV
Tl4—Sl7"
T14—s2ix
T14—szlxv
T14—S4xiii
T14—s22xv
T14——S6ix
T14—szxvii
T15—s23i
T15_S6XV111
T15—s22i
T15—s20xix
T15—T16xx
Tl5—s3xxi
T15—S5i
T15—1>6i
Tls—Sz8xxii
Tie—ssii
n6—s18vi
T16—SI7ii
'I'l6—828x

3.273(7)
3.291(6)
3.348(6)
3.364(6)
3.603(6)
3.699(7)
3.071 (6)
3.132(6)
3.284(6)
3.380(7)
3.482(5)
3.587(6)
3.883(2)
4.192(2)
3.081(6)
3.239(7)
3.257(6)
3.287(7)
3.350(5)
3.41 1(5)
3.508(6)
3.153(6)
3.305(6)
3.313(6)
3.355(6)
3.363(6)
3.372(5)
3.551(6)
3.720(5)
3.918(6)
3.130(8)
3.254(6)
3.363(7)
3.384(6)
3.646(2)
3.667(6)
3.717(7)
3.872(6)
4.072(8)
3.089(5)
3.128(6)
3.178(6)
3.209(6)

s20V—l3i3—S8ii
S2—13i3——S8ii
S6—l3i3—S8ii
S19viii—Bi3—SI
S20V—Bi3—Sl
S2—Bi3—S1
S6—Bi3—Sl
S8ii—Bi3—Sl
S19Viii—l3i3—312ii
S20V—1313—SI2ii
S2——Bi3—512ii
S6—Bi3—812ii
ssii—1313—512ii
s1—13i3—S12ii
S30ii—1315—SI4ii
S3oii—Bi5—Sl6
Sl4ii—Bi5—Sl6
S30ii—13i5—329ii
Sl4ii—13i5—829ii
Sl6—BiS—S29ii
S30ii—1315—S9xi
s14ii—1315—S9xi
S16—I3is—S9xi
sz9ii—I3is—S9xi
S30ii—I3is—s10xii
314ii—Bis—310xii
816—Bis—swxii
s29ii—I3i5—s10xii
S9Xi—Bi5—SIOXii
s24ii—Bi4—Sl 1
sz4ii—l3i4—s12ii
S1 1—1314—s121'li
sz4ii—Bi4—522
S1 1—1314—522
Slzii—Bi4—S22
s24ii—1314—s13ii
Sl 1—Bi4—S 1 3ii
Sl2ii—l3i4—Sl3ii
s22—13i4—Sl3ii
S24ii—I3i4—S8ii
s1 1—Bi4—88ﬁ
S12ii—Bi4—S8ii
322—13i4—S8ii

82.2507)
147.3905)
70.4204)
128.1105)
82.6705)
67.4905)
128.0305)
145.1004)
136.2804)
72.2805)
139.7204)
110.6506)
71.9703)
73.4903)
80.5309)
89.6508)
71.2605)
71.7006)
84.8506)
151.9706)
75.0908)
141.6606)
79.3807)
114.0709)
116.2309)
151.4306)
128.2404)
79.5205)
66.9105)
85.8006)
80.10(16)
71.5304)
73.3906)
147.3707)
80.1806)
76.1007)
81.6006)
145.1506)
116.0309)
140.93(14)
112.9305)
74.5904)
73.3305)

Table 3-10 continued

 

T16—S5ii 3.280(6) s13ii—13i4—S8ii 138.0705)
Tl6—s20xxiii 3.382(5) S24ii—Bi4—SI5 122.3207)
T16—S3ii 3.927(6) SI 1—Bi4—SlS 127.4405)
512ii—Bi4—S15 148.2105)
Sl—Pl 2.019(7) s22—Bi4—SI5 8517(16)
S2—P1 2.033(7) Sl3ii—Bi4—SIS 66.5505)
S3—l>1iv 2.044(7) S8ii—Bi4—s15 74.2204)
S4-—l>1xi 2.056(6) S2sxiV—Bi6—S3 950(2)
SS—P2 2.024(7) SZ8XiV—Bi6—S4i 84.2309)
86—P2ii 2.044(7) s3—Bi6——S4i 72.0105)
S7—l>2ii 2.056(8) s28xiV—Bi6—ss 7578(17)
SS—P2 2.057(7) S3—Bi6—SS 79.4606)
S9—l>3ii 2.001(7) S4i—Bi6—S5 143.3905)
S10—P3 2.046(8) 528xiV—Bi6—S25i 70.9008)
S11—1>3ii 2.051(7) S3—13i6—s25i 151.6507)
S12—P3 2.061(7) S4i—l3i6—S25i 81.9405)
Sl3—P4 2.020(8) S5—1316—525i 118.4806)
Sl4—P4 2.039(7) SZ8XiV—Bi6—S7i 113.8(2)
SIS—P4ii 2.044(8) s3—I3i6—S7i 128.0306)
Sl6—1>4ii 2.065(7) S4i—l3i6—S7i 148.4205)
Sl7—P5V 2.037(7) S5—13i6—S7i 68.1404)
S18—1>5v 2.046(7) s25L—l3i6—S7i 80.2805)
Sl9—P5 2.045(7) szsxiV—Bi6—S14 136.2106)
S20—P5m 2.042(8) S3—Bi6—S 14 1 1633(17)
821—P6 2.012(7) S4i—Bi6—SI4 78.0304)
$22—P6 2.016(8) SS—Bi6—S14 136.5704)
S23—P6 2.047(7) S25i—Bi6—Sl4 67.1904)
S24—l>6i 2.080(7) s7i—Bi6—s14 71.0804)
S25—P7 1.998(8)
326—P7xxviii 2.030(7)
S27—1>7xxii 2.059(7)
S28—l>7xxii 2.074(8)
S29—P8 2.015(8)
S30—P8 2.042(8)
S31—1>8ix 2.047(7)
s32—1>8ix 2.061(7)

 

(i) x, -1.5-y, 0.5+z; (ii) x, -l.5-y, -0.5+z; (iii) l-x, -0.5+y, 0.5-z; (iv) x, -O.5-y, 0.5+z; (v) l-x, -l-y, l-z; (vi)
2-x, -1-y, l-z; (vii) 2-x, 0.5+y, 1.5-z; (viii) 1+x, y, 2; (ix) l-x, -1-y, -z; (x) 2-x, 0.5+y, 0.5-z; (xi) x, -l+y, z;
(xii) x, -2.5-y, -0.5+z; (xiii) l-x, -2-y, -z; (xiv) x, 1+y, z; (xv) l-x, 0.5+y, 0.5-z; (xvi) x, -0.5-y, -0.5+z;
(xvii) -l+x, y, z; (xviii) x, y, 1+2; (xix) l-x, -0.5+y, 1.5-z; (xx) 2-x, -2-y, l-z;
(xxi) 2-x, -0.5+y, 1.5-z; (xxii) x, -2.5-y, 0.5+z; (xxiii) 1+x, -l.5-y, -0.5+z; (xxiv) x, y, -l+z; (xxv) -x, -1-y,
l-z; (xxvi) -l+x, -1.5-y, 0.5+z; (xxvii) l-x, 0.5+y, 1.5—z; (xxviii) x, 1+y, 1+2; (xxix) 2-x, -0.5+y, 0.5-z;
(xxx) -1+x, -1.5-y, -0.5+z; (xxxi) x, -l+y, -1+z; (xxxii) 2-x, -2-y, -z

79

TlBiP2S7 [V]: TlBiPZS-I crystallizes in the monoclinic spacegroup P21/c and is
isomorphous to KBiP287.32 The structure is composed of corrugated [BiP2S7]1' layers,
Figure 3-12, separated by 8-coordinate T'l+ with an average Tl-S distance of 3.34 A,
Figure 3-13A. Each Bi3+ ion is bis-chelated to three [P237]4‘ anions (average Bi-Se bond
lengths of 2.92 A) and chelated by another [P2S7]4' anion (Bi-S bond length of 2.778(2)

A, Figure 3-12B. Table 3-11 contains selected bond lengths and angles for compound
[Vl-

 

 

9 i, . \ i, . i,
c u 491- u «we
O O O O
v.\ I. I.\ I. \
i {a} . i {A}. .00
Be as, Be ~ (e
O i, \ 0’ ‘9 i, \ 0’ ‘O l,
.. a; , 9.0

 

Figure 3-12. The structure of TlBiP2S7 [V] viewed down the (001) axis showing the corrugated lamellar
packing.

80

 

 

Figure 3-13. (A) The Tl+ coordination environment. Tl atom is green. (B) the coordination environment of
Bi3+ in TlBiPZS7 [V].

81

Table 3-11. Selected bond lengths and angles for a—TlBiPzS7 with bond lengths (A) and angles (9) with
standard uncertainties in parenthesis.

 

Bi—S6 2.7780(18) S6—Bi—S2 74.63(5)
Bi—82 2.8238(18) S6—Bi—Sl 8205(5)
Bi—Sl 2.8752(19) SZ—Bi—Sl 7253(5)
131—83i 2.9124(18) S6—Bi—s3i 144.34(5)
Bi—SS 2.9156(18) 82-——Bi—s3i 7059(5)
Bi—S3 2.9778(18) Sl—Bi—S3i 94.72(5)
Bi—S7 3.0376(19) S6—Bi—SS 9342(5)
sz—Bi—ss 8924(5)
Tl—S6Vi 3.2640(19) Sl—Bi—SS 161.77(5)
Tl—SlVﬁ 3.2973(18) ssi—Bi—ss 78.65(5)
T1—s7i 3.3099(18) S6—Bi—S3 98.15(5)
'I'l—S7 3.3202(19) SZ—Bi—S3 140.53(5)
Tl—slviii 3.3330(19) Sl—Bi—S3 6803(5)
T1—36viii 3.3589( 19) s3i—Bi—s3 1 13.61(4)
T1—s2vi 3.3784(19) SS—Bi—S3 130.19(5)
Tl—S4 3.4224(18) S6—Bi—S7 138.27(5)
n—szvii 3.960(2) SZ—Bi—S7 139.31(5)
Sl—Bi—S7 124.82(5)
Pl—Slii 2.019(2) s3i—Bi—s7 7141(5)
Pl—s3ii 2.022(3) SS—Bi—S7 6947(5)
P1—s2ix 2.028(2) S3—Bi—S7 6992(5)
Pl—S4 2.118(2)
P2—S7 2.009(2) Slii—Pl—s3ii 108.31(11)
P2—SS 2.019(3) Slii—m—szix 116.52(12)
P2—S6Xii 2.037(2) s3ii—P1—82ix 109.89(10)
P2—S4 2.118(2) Slii—Pl—S4 111.82(10)
s3ii—P1—S4 112.05(11)
szix—P1—84 9804(10)
S7—P2—SS 114.80(11)
s7—P2—S6xii 110.5201)
ss—P2—86xii 109.84(11)
S7—P2—S4 100.26(10)
SS—P2—84 111.5(1)
S6xii—P2—S4 109.53(11)
0.5+z; (ii) l-x, 0.5+y, 0.5-z; (iii) -l+x, 0.5-y, -O.5+z;
-O.5+z; (vi) l-x, -O.5+y, 0.5-z; (vii) 1+x, 2; (viii)

(ix) l-x, l-y, l-z; (x) 2-x, l-y, l-z; (xi) 2-x, 0.5+y, 0.5-z; (xii) l-x, l-y, -z.

Tl3Bi(PS4)2 [W]: This compound is isotypic to K3Bi(PS4)2. The structure

solution revealed the same straight [Bi(PS4)2]3' chains as in the K-analog, Fig 3-14A.

82

Furthermore, comparison of the powder X-ray diffraction patterns with simulated ones
from the known K+ containing phase as well as the solutions discussed here suggested
that the compounds were nearly identical. The chains, however, exhibited position

disorder in the Bi3+ atoms, which were separated from each other along the chain axis by

a distance of approximawa 0.85 A, Figure 3-14B.

 

Figure 3—14. (A) The [Bi(PS4)2]3' chain that runs down the b-axis and (B) perpendicular to the b-axis

showing the Bi3+ positional disorder. Bi atoms in white, phosphorous and sulfur atoms in black and
shaded, respectively.

Optical Properties

Diffuse reﬂectance measurements suggest that the optical band gaps of the title
compounds are smaller than those of the K+ analogs. The spectrum of a—TlBiPZSeé [1]
indicated that the band gap was about 1.23 eV, Figure 3-15A, whereas a-KBiPZSe6 was
reported to have a band gap of 1.61 eV.52,53 The average Tl-Se distance in the former
was 3.38 A whereas the average K-Se distance in the latter was 3.36 A. The similarities
in bond lengths is expected given that T1" and K+ have similar ionic radii (1.5 A for 11+

vs. 1.4 A for W)” The reduction in band gap in the 11+ analogs is therefore likely due

to the increased covalency of the interaction between the metal and Se.

83

(A) r

[H— — [ll] -—[|"l

Absorbance (arb. units)

 

U I I

0.5 1 1 .5 2 2.5
Energy (eV)

(B)

Absorbance (arb. units)

 

 

 

Enemy (eV)

Figure 3-15. The optical absorption spectra showing energy band gaps of (A) a-TlBiP28e6 [I]: 1.23 eV, [3-
TlBiPZSe6 [II]: 1.27 eV, and Tuaipzsé [111]: 1.81 eV; and (B) Tl3Bi3(PS4)4 [IV]: 1.88 eV, TlBiP287 [V]:
1.87 eV, T13Bi(PS4)4 [v1]: 1.87 eV, and T14Bi2(P2S6)(PS4)2 [VII]: 1.98 eV.

In most main group metal chalcogenides with semiconducting band gaps, the top
of the valence band is dominated by chalcogen based ﬁlled p-orbitals, whereas the
bottom of the conduction band is mainly based on empty p-orbitals of the main group
metal.37 The substitution of K for T1 in KIM/Q structures enhances the Tl-Q bonding

through increased covalency which further broadens the bands near the energy gap

causing it to shrink.39

84

It is not surprising that, given the similarity between the [BiPZSe6]1' frameworks
in a—TlBiP2866 [I] and B-TlBiPZSe6 [11] and the similar Tl-Se bond lengths that the
optical band gaps are also similar with showing an absorption edge at 1.27 eV. The band
gap of TlBiP236 [III] is slightly larger at 1.81 eV as is expected upon the substitution of
Se for S in structurally related compounds (i.e. containing a [P2Q6]4' anion), Figure 3-
15A.

TlBiPZS7 [V] was determined to have an optical band gap of 1.87 eV, Figure 3-
15B. The isomorphic analog of compound [V], KBiPZS7, was reported at 2.25 eV.5
T13Bi3(PS4)4 [IV], T13Bi(PS4)2 [VI] and T14Bi2(PS4)2(PzS6) [VII] have optical band

gaps of 1.88 eV, 1.98 eV and 1.97 eV.

Solid State NMR

Spectra are displayed in Figure 3-16, and Table 3-12 presents chemical shifts
(CSs), line widths, integrated intensities, chemical shift anisotropies (CSAs), and
longitudinal relaxation times (T18). The chemical shifts and chemical shift anisotropies of
each of the reported thio- and selenophosphate materials are within the ranges expected

for related materials.48

85

 

 

 

 

200 100 onn 400 -200'

 

 

200 100 0 -1bo -2'00 260 160 0'-1bo 200
m p m

(G)

 

 

200 160 0 -100 -2bo 260 160 ('1 400 200
ppm ppm
Figure 3-16. 31? NMR spectra of (A) a—TlBiPzSe6 [I], (B) B—TlBiPZSe5 [II], (C) TlBiP286 [111], (D)
Tl3Bi3(PS4)4 [IV], (E) TlBiP287 [V], (F) T13Bi(PS4)2 [VI], (G) Tl4Bi2(PS4)2(PZS6), and (H) KBiPZS7.
Each spectrum was an average of four scans with a delay between pulses of 5000 s. Each spectrum was
processed with S 100 Hz line broadening. Peaks and their corresponding spinning sidebands are grouped
with like symbols.

86

Table 3-12. 31F chemical shift (CS), chemical shift anisotropy (CSA), and T1 values of the Tl/Bi/P/Se and
TllBi/P/S compounds.

 

 

Compound csa Line Widthsa Integrated b CSAc 71d
(ppm) (Hz) Intensmes (ppm) (s)
a-TlBiP2366 [I] 40.0, 12.4 1060. 1640 100. 90 n.d. n.d.
B-TlBiste6 [II] 48.9, 24.7, 9.1 1000. 1390. 1650 53. 100. 68 n.d. 1500(150)
T133131Ps4)4 [IV] 793, 75.8, 71.0 730. 500.570 100. 30.91 69 1500(100)
T1B1P287 [V] 85.2, 75.4 320. 860 67. 100 n.d. 1750(150)
T133084); 1V1] 79.9 1600 100 n.d. 1500(100)
gﬁzms‘mmzw 98.2, 79.3 410’ 800 91’ 100 137,110. 1300(300)

 

n.d. = not determined, n.o. = not observable.

a Uncertainties in CS are approximately 1 0.5 ppm, and uncertainties in line widths are approximately :1: 10
Hz. These parameters were determined by a mathematical deconvolution of the peak with equal
contributions from Gaussian and Lorentzian line shapes.

b Intensities were normalized with the most intense peak set to 100. For compounds with splittings due to
J-coupling, the integrated intensities of the constituent peaks were summed.

c Principal CSA values were determined using a Herzfeld-Berger analysis computer program (I-[BA 1.5,
Eichele, K., Wasylishen, R.E., Dalhousie University and Universtat Tiibingen, 2006). The reported CSA
= 811 — 833, i.e. the approximate overall width of the static CS powder pattern. Maximum CSA

uncertainties are :1: 20 ppm.
d Uncertainties are indicated in parenthesis.

Phosphorous-Thallium J-coupling: The spectrum of TlBiP257 [V] had two sets of
peaks centered at 85.2 ppm and 72.4 ppm with splittings of 481 Hz and 783 Hz,
respectively, Figure 3-16B. One possible cause of these splittings is two-bond P-Tl J-
couplings and this hypothesis was tested using the isomorphic analog KBiP287.5 The 31P
NMR spectrum of the analog contained two singlet peaks centered at 85.2 ppm and 75.4
ppm, Figure 3-16H, indicating that the splitting observed in the spectrum of [V] was due
to P-Tl coupling. The 31F atoms in the [P287]4' ion are crystallographically inequivalent
and it is interesting that the P-Tl J-coupling is observed while the two-bond P-P J-
coupling is not apparent. It was not possible to rationalize the spitting patterns based on

numbers of equivalent and inequivalent T1. 2031‘] (30% abundant) and 205T1 (70%

abundant) would resonate at 228.89 and 231.14 MHz in a 9.4 T magnet, respectively, and

87

distinct couplings to the different spin 1/2 T1 isotopes were not observed. In liquid state
NMR studies of organophosphorous thallium compounds, distinct 31P-203'I‘1 and 31P-
205'1‘1 couplings were also not observed.56

There was additional evidence for two-bond P-Tl J-coupling in the spectra of

other title compounds. The 9.4 T NMR spectrum of T14Bi2(PS4)2(P2S6) [VII] contained

two peaks centered at 98.2 ppm with a 490 Hz splitting and three peaks centered at 79.3
ppm with neighboring peak splittings of 1390 Hz and 1180 Hz. At 7 T, the respective
splittings were 510 Hz, 1390 Hz, and 1180 Hz. The constancy of the splitting
magnitudes with respect to magnetic ﬁeld is a marker of J-coupling. Similarly, at 9.4 T
the NMR spectrum of 113Bi(PS4)2 [VI] appeared to be a multiplet centered at 79.9 ppm
with neighboring peak separations of ~1400 Hz, Figure 3-16F. At 7 T the same splitting
magnitudes were observed.

The spectral line widths of the Tl-containing chalcophosphate materials are also
relatively broad compared to the chalcophosphate materials that do not contain 11.4057
For example, KBiPZS-l had a line width of 337 Hz, whereas some of the Tl-containing
compounds had line widths greater than 1600 Hz. The increased line widths are likely
due in part to unresolved P-Tl coupling.

To the best of our knowledge, this is the ﬁrst example of P-Tl coupling observed
in the solid state, although one bond P-Tl coupling in the liquid state has been observed.
For example, coupling constants of 3203 and 3144 Hz were detected in spectra of
Ph2P(Me/2Tl)2PPh2.56 To our knowledge, our work is also the ﬁrst example of two-bond
P-Tl J-coupling. Through-space J-coupling is also possible when there is signiﬁcant

overlap of the van der Waals radii of the coupled nuclei.58 For the title compounds there

88

was not a good correlation between the magnitude of J and the distance between P and T1
which suggests that two-bond coupling is dominant. Comparison of these experimental
J-couplings with J-couplings derived from electronic structure calculations may be one
way to now validate the calculation method.

Splittings due to P-P J-coupling have been observed in compounds with directly
bonded crystallographically inequivalent P-atoms.40 Such splittings were a possibility in
compounds [I] and [II], but they were not observed, likely because of the broad spectral

line widths.

Peak Assignments: T14Bi2(PS4)2(P286)39 [VII] provided an interesting example
of a structure with both [P28614‘ and [PS4]3' anions. The chemical shift ranges of these
two anions are known to be overlapped and differences in the CSA magnitudes for the
two anions were used for chemical assignment. The NMR spectrum of [VII] showed two
sets of peaks: a doublet centered at 98.2 ppm, and a triplet centered at 79.3 ppm, Figure
3-16G. The crystallographic structural solution contained two inequivalent P sites: one in
the [PS4]3' anion, and one in the [P2Se6]4' anion. Spinning sidebands were observed
only for the doublet, and so the P atom corresponding to this site must have a larger
magnitude CSA than the triplet. The magnitude of the CSA of [PS4]3' has previously
been shown to be signiﬁcantly lower than that of [P256]4' anions,4O so the triplet and
doublet were therefore assigned to the [PS4]3' and [P286]4‘ units, respectively.

Chemical shift assignment was also made for compounds with
crystallographically inequivalent 31P atoms that were in the same type of [PnyF' anion.

The spectrum of B'TlBiPZSC6 [II], showed three peaks with a ratio of integrated signal

89

intensity of about 1:2:1 with chemical shifts of 48.9, 24.7, and 9.1 ppm, Figure 3-16B.
There were four crystallographically inequivalent P sites with some variation in the
numbers of Bi3+ and Tl+ ions close to the [P28e6]4‘ anions. These variations were
correlated with the chemical shifts using the following scheme. For each P site, spheres
of 4 A diameter were considered around each of the bonded Se atoms and the total
numbers of Bi3+ and 'I‘l+ ions were counted within these spheres. The 4 A diameter was
a reasonable cutoff for deﬁning proximate metal ions as the Se-metal ion distances

formed two groups: one with 2.8 A < dSe-metal < 4.0 A and one with dSe-metal > 4.5 A.

Within these spheres P(l) and P(3) had three Bi3+ and three Tl+ ions, P(2) had four Bi3+
and two 'I‘l+ ions, and P(4) had two Bi3+ and four Tl+ ions. The central NMR spectral
peak can be attributed to P(l) and P(3) which have the same number of vicinal B13+ and
T1" ions and the other distinct peaks can be attributed to P(2) and P(4). It is not possible
using this method to determine the speciﬁc chemical shift that corresponds to each site.
Thismodel only groups 31P sites by similarities in coordination and does not predict the
magnitude or sign of the chemical shift differences.

Tl3Bi3(PS4)4 [IV] is another example of this approach to assignment. The NMR
spectrum showed three peaks with the approximate intensity ratio of 4:1:3 and shifts of
79.3 ppm, 75.8 ppm, and 71.0 ppm, Figure 3-16D. A mathematical deconvolution of the
spectrum was only successful when the feature furthest downﬁeld was modeled by two
closely spaced peaks.

There are eight inequivalent P sites and an assignment was made by grouping
them into three distinct coordination environments using a scheme similar to that for

compound [11]. However, only proximate Bi3+ ions were considered for compound [IV]

90

as the S-Tl distances had a more continuous distribution than was found for compound

[II]. The P(l), P(3) and P(4) centered [PS4]3' anions had four proximate Bi3+ions, the
P(2) centered [PS4]3' ion had four proximate Bi3+ ions and the four remaining [PS4]3'
anions had two proximate Bi3+ ions. It is reasonable that the [PS4]3' anions with similar

coordination environments have the same chemical shifts and the ratio of the intensities
can be explained on this basis.

The numbers of crystallographically inequivalent P atoms and the numbers of
distinct shifts were equal for compounds [I] and [V], Figure 3-16A, B, respectively. Each
peak in the NMR spectrum corresponds to one of these crystallographically inequivalent
sites. Chemical shifts were not calculated and therefore speciﬁc peaks cannot at present
be attributed to particular crystallographic sites. Impurity peaks and the lack of an
accurate crystal structure precluded assignment in compounds [III] and [VI], respectively,

Figure 3-16C, F.

CONCLUDING REMARKS

A broad family of compounds can be stabilized in the Tl/Bi/P/Q (Q=S,Se) system
and contains examples of 1-, 2-, and 3-dimensiona1 structural motifs. This high degree of
structural diversity likely has its origins in the ﬂexibility of the Bi3+ and 11+ coordination
environments, and also in the ﬂexibility of the Tl-Q interaction with the covalent Bi/P/Q
framework. The reaction chemistry of the T1/Bi/P/S system proved to be more diverse
and subtle than we originally expected, and small changes in stoichiometry resulted in
different reaction paths and compounds. This explains why we often experience

problems with reproducibility with some of the compounds we report here (e.g.

91

T13Bi3(PS4)4 vs. TlBiPZS—I). One reason for this rich chemistry seems to be the ability of
the thiophosphate anions to convert from one to another with small changes in P/S ratio.
All three species [PS4]3', [P286]4' and [P287]4' were observed as well as combinations
thereof such as in T14Bi2(PS4)2(P286).39

The corresponding selenium system was less diverse and [PSe4]3' and [PZSe7]4'
species were not observed. In selenophosphate chemistry, the most stable species under
moderate Lewis basic conditions is the P4+ anion [P28e6]4'.22 By slightly varying the
stoichiometry of the reaction conditions, it is often possible to produce a given phase in
pure form. Because of the increased ionicity, it may also be possible to tune the optical
properties of these materials by making solid solutions with potassium. Such a study is
also likely to yield new compounds. Similar richness and diversity might be observed in
other 'I'l/M/P/Q systems perhaps with M = Sb, Au, In, Pd, Zn, Cd, Hg, and the
Lanthanides.

31P solid state NMR spectroscopy of the reported compounds demonstrates
similar chemical shifts and chemical shift anisotropies to related materials.40 It was

possible to assign the peaks of inequivalent 31P atoms from the same [PnyF' anion type
by considering differences in the metal ion coordination to the [PnyF' unit and by

considering CSA magnitudes. More speciﬁc assignment of peaks would require detailed
calculations. Such calculations would be challenging because of the presence of
numerous heavy atoms in the structures. Splittings due to P-Tl J-couplings were
observed, and to the best of our knowledge, this is the ﬁrst example of two bond P—Tl J-

coupling and the ﬁrst observation of P-Tl J-coupling in the solid state.

92

REFERENCES

1.

2.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

Kanatzidis, M. G. Chem. Mot. 1990, 2, 353-363.

Kanatzidis, M. G.; Park, Y. J. Am. Chem. Soc. 1989, 111, 3767-3769.

. Kanatzidis, M. G.; Park, Y. Chem. Mat. 1990, 2, 99-101.
. Park, Y. B.; Kanatzidis, M. G. Angew. Chem. Int. Ed. 1990, 29, 914-915.
. McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 1257-1267.

. Chondroudis, K.; Kanatzidis, M. G. J. Solid State Chem. 1998, 138, 321-328.

Iordanidis, L.; Bilc, D.; Mahanti, S. D.; Kanatzidis, M. G. J. Am. Chem. Soc. 2003,
125, 13741-13752.

McCarthy, T. J .; Ngeyi, S. P.; Liao, J. H.; Degroot, D. C.; Hogan, T .; Kannewurf, C.
R.; Kanatzidis, M. G. Chem. Mot. 1993, 5, 331-340.

Iordanidis, L.; Brazis, P. W.; Kyratsi, T.; Ireland, J.; Lane, M.; Kannewurf, C. R.;
Chen, W.; Dyck, J. S.; Uher, C.; Ghelani, N. A.; Hogan, T.; Kanatzidis, M. G. Chem.
Mat. 2001, 13, 622-633.

Iordanidis, L.; Schindler, J. L.; Kannewurf, C. R.; Kanatzidis, M. G. J. Solid State
Chem. 1999, 143, 151-162.

Kohatsu, I.; Wuensch, B. J. Acta. Crystallogr. B 1976, 32, 2401-2409.

Cook, R.; Schafer, H. Rev. Chim. Miner. 1982, 19, 19-27.

Wiegers, G. A.; Meetsma, A.; Vansmaalen, S.; Haange, R. J .; Wulff, J .; Zeinstra, T.;
Deboer, J. L.; Kuypers, S.; Vantendeloo, G.; Vanlanduyt, J .; Amelinckx, S.;
Meerschaut, A.; Rabu, P.; Rouxel, J. Solid State Commun. 1989, 70, 409-413.
Mrotzek, A.; Kanatzidis, M. G. Acc. Chem. Res 2003, 36, Ill-119.

Kanatzidis, M. G. Acc. Chem. Res 2005, 38, 359-368.

Makovicky, E. Fortschr. Miner. 1981, 59, 137-190.

Makovicky, E. Z. Kristall. 1985, 173, 1-23.

Makovicky, E. F ortschr. Miner. 1985, 63, 45-89.

Makovicky, E. Neues Jahrb. Mineral. -Abh. 1989, 160, 269-297.

93

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Chung, 1.; Do, J .; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2004,
43, 2762-2764.

Knaust, J. M.; Dorhout, P. K. J. Chem. Crystallogr. 2006, 36, 217-223.
Kanatzidis, M. G. Curr. Opin. Solid. St. M. 1997, 2, 139-149.

Coste, S.; Hanko, J .; Bujoli-Doeuff, M.; Louarn, G.; Evain, M.; Brec, R.; Alonso, B.;
Jobic, S.; Kanatzidis, M. G. J. Solid State Chem. 2003, 175, 133-145.

Manriquez, V.; Galdamez, A.; Ruiz-Leon, D. Mater. Res. Bull. 2006, 41, 1337-1344.

Belkyal, 1.; El Azhari, M.; Wu, Y. D.; Bensch, W.; Hesse, K. F.; Depmeier, W. Solid
State Sci. 2006, 8, 59-63.

Gieck, C.; Tremel, W. Chem. -Eur. J. 2002, 8, 2980-2987.

Coste, S.; Kopnin, B.; Evain, M.; Jobic, S.; Brec, R.; Chondroudis, K.; Kanatzidis,
M. G. Solid State Sci. 2002, 4, 709-716.

Hess, R. E; Gordon, P. L.; Tait, C. D.; Abney, K. D.; Dorhout, P. K. J. Am. Chem.
Soc. 2002, 124, 1327-1333.

Coste, S.; Kopnin, B.; Evain, M.; Jobic, S.; Payen, C.; Brec, R. J. Solid State Chem.
2001, 162, 195-203.

McCarthy, T. J .; Kanatzidis, M. G. Chem. Mot. 1993, 5, 1061-1063.

Manriquez, V.; Galdamez, A.; Leon, D. R.; Garland, M. T.; Jimenez, M. Z. Krist.
New Cryst. St. 2003, 218, 151-152.

McCarthy, T.; Kanatzidis, M. G. J. Alloys Compd. 1996, 236, 70-85.

Chung, L; Karst, A. L.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2006, 45,
2785-2787.

Odoulov, S. G.; Shumelyuk, A. N.; Brost, G. A.; Magde, K. M. Appl. Phys. Lett.
1996, 69, 3665-3667.

Shumelyuk, A.; Hryhorashchuk, A.; Odoulov, S. Phys. Rev. A 2005, 72, 6.

Kroupa, J.; Tyagur, Y. 1.; Grabar, A. A.; Vysochanskii, Y. M. Ferroelectr. 1999,
223, 421-428.

94

37.

38.

39.

40.

41.

42.

43.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

Wachter, J. B.; Chrissaﬁs, K.; Petkov, V.; Malliakas, c. D.; Bilc, D.; Kyratsi, T.;
Paraskevopoulos, K. M.; Mahanti, S. D.; Torbrugge, T.; Eckert, H.; Kanatzidis, M.
G. J. Solid State Chem. 2007, 180, 420-431.

Iyer, R. G.; Aitken, J. A.; Kanatzidis, M. G. Solid State Sci. 2004, 6, 451-459.
McGuire, M. A.; Reynolds, T. K.; DiSalvo, F. J. Chem. Mot. 2005, 17, 2875-2884.
Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.
Gunne, J.; Eckert, H. Chem-Eur. J. 1998, 4, 1762-1767.

SMART, SAINT, SHELXT L V-5 and SADABS, Bruker Analytical X-ray Instruments,
Inc.: Madison, 1998.

X-AREA, X-SHAPE, and X-RED, STOE & Cie GmbH: Darmstadt, 2004.

Petricek, V.; Dusek, M.; Palatinus, L. Jana2000. The Crystallographic Computing
System, Praha, Czech Republic, 2000.

Wendlandt, W. W.; Hecht, H. G. Reﬂectance spectroscopy; Interscience Publishers:
New York, NY, 1966.

Kortiim, G. Reﬂectance spectroscopy. Principles, methods, applications; Springer:
Berlin, Germany, 1969.

Tandon, S. P.; Gupta, J. P. Phys. Status Solidi 1970, 38, 363-367.

Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

Breshears, J. D.; Kanatzidis, M. G. J. Am. Chem. Soc. 2000, 122, 7839-7840.
Kanatzidis, M. G.; Sutorik, A. C. Progr. Inorg. Chem. 1995, 43, 151-265.

Gave, M.; Bilc, D.; Mahanti, S.; Breshears, J .; Kanatzidis, M. Inorg. Chem. 2005,
5293-5303.

McCarthy, T. Ph.D. Thesis, Michigan State University, East Lansing, MI, 1994.
McCarthy, T. J .; Kanatzidis, M. G. J. Chem. Soc. Chem. Commun. 1994, 1089-1090.
Chen, J. H.; Dorhout, P. K.; Ostenson, J. E. Inorg. Chem. 1996, 35, 5627-5633.

Greenwood, N. N .; Earnshaw, A. Chemistry of the Elements; 2nd ed.; Butterworth-
Heinemann: Boston, MA, 1997; pp 1176, 1205.

95

56. Walther, B.; Bauer, S. J. Organomet. Chem. 1977, 142, 177-184.
57. Regelsky, G. Ph.D. Thesis, University of Munster, Munster, Germany, 2000.

58. Bryce, D. L.; Wasylishen, R. E. J. Mol. Struct. 2002, 602, 463-472.

96

CHAPTER 4
IMPROVED RESOLUTION AND DETECTION OF 31P-Tl J-COUPLINGS AT 21

T IN 31P MAGIC ANGLE SPINNING NMR SPECTRA OF INORGANIC
COMPOUNDS CONTAINING Tl/Bi/P/S.

97

INTRODUCTION

The ﬁrst examples of phosphorous-thallium J-couplings in the solid state were
recently reported for crystalline compounds containing Tl, Bi, P, and Q (Q = S or Se).
The detection of these couplings was based on comparisons of 31P magic angle spinning
NMR spectra at 9.4 T and 7 T spectra and also on comparisons between 31P NMR
spectra of T1- and non Tl-containing isomorphic analogues.l Figure 4-1 displays the
structures of T14Bi‘-)(PS4)2(P2S6)2 and TlBiPZS7 which were two of the compounds in

this class that exhibited these couplings.

 

 

 

 

 

 

Figure 4-1. (A) The structure of Tl4Bi2(PS4)2(PzS6) viewed down the b-axis. Large black spheres are Tl,
large hollow spheres are Bi, small black spheres are P, and small shaded spheres are S. A [P286]4' anion is.

highlighted in a dashed box. Two r1+ ions have been removed so that the highlighted [st614- unit is

more clearly displayed. Two of the Ss in this unit are not visible because they are directly behind displayed
Ss. (B) The structtn'e of TlBiPZS—I viewed down the c-axis. Large black spheres are T1, large hollow

spheres are Bi, small black spheres are P, and small shaded spheres are S. A single unit cell is shown for
each structm'e.

The Tl4Bi2(PS4)2(PZS6) compound contained [P286]4‘ and [PS4]3— units linked
together in layers by coordinating Bi3+ ions. These layers were separated by layers of Tl+
ions. The TlBiPZS-l compound contained [P287]4‘ units linked together in two-

dimensional layers by coordinating Bi3+ ions with Tl+ ions between the layers. These

structures demonstrated that the detected 500-1400 Hz splittings were due to two-bond P-

98

Tl couplings and such couplings were the ﬁrst example of non-one-bond P-Tl couplings
in either the liquid- or the solid-state.

Although there are seven known compounds containing Tl, Bi, P, and Q and eight
known compounds containing K, Bi, P, and Q,3'9 only B-TlBiPZSeg/B-KBiPZSeG and
TlBiPZS7/KBiPZS7 are structurally isomorphic. Although the crystal structure is not yet
fully reﬁned for Tl3Bi(PS4)2,9 this compound is likely structurally similar to

K3Bi(PS4)2.4 It is somewhat surprising that there are only a few examples of structural

isomorphism because Tl+ and K+ have similar ionic radii (i.e. 0.150 A and 0.138 A,
respectively).10 Electronic structure calculations might provide insight into why the
simple replacement K+ -> Tl+ is infrequently observed in the absence of a signiﬁcant
geometric structural modiﬁcation. Splittings due to P-Tl J-couplings may be useﬁrl

experimental parameters to validate these calculations.

99

T14Bi2(PS.t)2(P286) T13Bi(PS4)2 TlBiPZS7

9.4 T

 

1'10 100 90 8'0 70 60140120100 80 60 40 20100 90 80 7o 60

Whirl.

110 100 90 ' 70 60140120100 80 60 40 20100 90 ' 80

 

Figm'e 4—2. 31P NMR spectra of Tl4Biz(PS4)2(PZS6), Tl3Bi(PS4)2, and TlBiP287 obtained at 9.4 T and
21 T. The experimental spectra are represented by solid lines and the best-ﬁt Lorentzian ftmctions used to
deconvolve the spectra are represented by dashed lines. The sums of the Lorentzians are also shown as
dashed lines and ﬁtted well to the corresponding experimental spectra.

Ultra-high ﬁeld (21 T) NMR spectra of three thallium bismuth thiophosphates are
reported, along with a comparison to the previously obtained 9.4 T spectra. Besides
providing additional conﬁrmation of the P-Tl J-couplings, the 21 T spectra oﬂ‘ered a
signiﬁcant enhancement of chemical shift resolution over the spectra collected at lower

ﬁeld.

EXPERIMENTAL SECTION

The compounds T14312(PS4)2(P286), T13Bi(PS4)2, and TlBiPZS7 were produced
as described previously.1 The ultra high ﬁeld NMR spectra were collected on a Bruker
AVANCE 21 T spectrometer operating at 362 Mhz. The 9.8 T NMR spectra were

collected on a Varian Inﬁnity Plus spectrometer operating at 162 Mhz. Each spectrum

100

was obtained at ambient temperature with magic angle spinning and was the sum of four
Bloch decay scans. Pulse width calibration and chemical shift referencing were done with

85% H3PO4 (= 0 ppm). Each spectrum was processed identically with 50 Hz line

broadening and baseline correction.

RESULTS AND DISCUSSION

Figure 4-2 displays the 9.4 and 21 T 31P magic angle spinning spectra of three
compounds: Tl4Bi2(PS4)2(PZS6), Tl3Bi(PS4)2, and TlBiPZS-l. Each of the spectra could
be ﬁtted well to a sum of Lorentizian functions. In Figure 4- 2, the experimental spectra
are displayed as solid lines, the individual Lorentzian functions are displayed as dotted
lines, and the sums of the Lorentzian ﬁrnctions are also displayed as dotted lines. The
spectra of T14Bi2(PS4)2(PZS6) and TlBiPZS-l were separated into two clusters of
Lorentzian peaks and the spectrum of Tl3Bi(PS4)2 was a single cluster of Lorentzian
peaks. Table 4-1 lists the peak shiﬂs, linewidths, and integrated intensities of the ﬁtted
Lorentzians and Table 4- 2 lists the average shift of each cluster of Lorentzian peaks as
well as the splittings between adjacent Lorentzian peaks in the same cluster. For each
compound, the consistency between the average shifts and the peak splittings at different
ﬁelds conﬁrmed that the splittings were likely due to 31P-Tl J-couplings. Tl has two
abundant isotopes with similar gyromagnetic ratios and splittings due to individual Tl
isotopes were not resolved. The upﬁeld clusters of the TlBiPzS7 spectra were not
included in Table 4-2 because it was not possible to make an assignment with a

consistent average shift and peak splitting at the two ﬁelds. This difficulty may have been

101

due to overlapping peaks ﬁom unidentiﬁed impurities in the sample. Minor impurities

were also apparent as additional peaks in the powder diffraction pattern.

Table 4-1. Chemical shifts, linewidths, and integrated intensities of the ﬁtted Lorentzian components of the
spectra. a

 

 

 

 

9.4 r 21 '1
Compound Peak Linewidth Integrated Peak Linewidth Integrated
(ppm) (Hz) b intensity c (ppm) (Hz) b intensity c
99.8 350 51 98.8 520 42
96.6 420 71 97.4 540 44
T14Bi2(PS4)(P2$6) 87.6 570 31 83 910 40
79.8 1040 100 79.3 1200 100
71.5 760 38 75.7 700 27
99.2 1600 31 88.7 750 11
90 1490 54 84.7 1630 59
T13Bi(PS4)2 80 1890 100 79.8 2060 100
69.4 1410 52 75.3 1300 42
60.3 1710 35 71.6 1610 27
87 360 93 85.9 430 82
84.1 290 66 84.6 390 67
r11311>257 77 510 47 77 650 100
74.4 530 74 71.5 640 67
70.3 460 100 70 430 24

 

a Fitting of the Lorentzian functions was done with the OriginPro 7.5 program
b The linewidth is the ﬁtting parameter ﬁ'om a Lorentzian peak function.
c The Lorentzian component with maximum integrated intensity was assigned a value of 100.

Table 4- 2. Averzﬁe chemical shiﬁs and glittings of the spectral clusters.

 

 

 

 

9.4T 21 T
Cluster Shift Splitting Shift Splitting
“mm“ identity (ppm) (Hz) (ppm) 012)
133303;“ 98.2 520 98.1 510
T14Biz(PS4)(PzS6) Upﬁ 1d
6
triplet 79.6 1270,1350 79.3 1350,1310
T13Bi(PS4)2 Quintet 79.8 ~1580 80.0 ~1560
D°wnﬁeld 85.6 470 85.3 470
. doublet
"1113113237 U ﬁ 1d
p c
72.4 670 74.3 2000
doubleta

 

a This doublet was the two components of the cluster with the greatest integrated intensity.

102

There were two crystallographically inequivalent P in T14Bi2(PS4)2(PZS6) and
one P was associated with the [PS4]3‘ unit and one P was associated with the [P286]?

unit. The downﬁeld “doublet” exhibited much larger chemical shift anisotropy than did
the upﬁeld “triplet” and the doublet and triplet were respectively assigned to the [P286]4‘
and [PS4]3‘ units, respectively},11 The 31P in the [P286]? and [PS4]3“ units had
respectively ﬁve and six Tl separated by two bonds and the reason for the simple
experimental splitting patterns was therefore not obvious. The ratio of relative integrated
intensities of the double and triplet was 0.72 at 9.4 T and 0.51 at 21 T. The 9.4 T result
was closer to the 1.0 ratio expected from the crystal structure. The lower ratio observed at
21 T may be a result of a shorter recycle delay and the consequent effect of differential
longitudinal relaxation for the doublet and triplet.

The spectra of T13Bi(PS4)2 could be ﬁtted as a quintet at both ﬁelds. The structure
of this compound has not been fully reﬁned. The compound K3Bi(PS4)2 is isotypic and

has a well-refmed structure1 so using the reﬁned structural model for comparative
purposes is reasonable. There are two crystallographically inequivalent P atoms in
K3Bi(PS4)2 and one 31P has 10 K" separated by two bonds while the other 31P has 6 K+
ions separated by two bonds. The T13Bi(PS4)2 quintet cannot be not trivially understood
by the replacement of K+ ions with Tl+ ions in this structure.

The spectra of TlBiPZS7 contained two clusters of peaks. At both ﬁelds, the

downﬁeld cluster was ﬁtted as a doublet and the upﬁeld cluster was ﬁtted as a triplet.
There was some variation in the appearance and the ﬁtting of the upﬁeld regions at the

two different ﬁelds. These variations were likely due to: (1) as-yet unidentiﬁed impurities

103

in the samples; and (2) the shorter recycle delay used at 21 T and the consequent possible
effects of differential longitudinal relaxation. There were two crystallo graphically

inequivalent P in TlBiPZS7 and the most reasonable assignment was the downﬁeld

doublet as one 31P and some part of the upﬁeld triplet as the other 31F.

Table 4-1 shows that the linewidths in Hz of the individual Lorentzian
components were not strongly dependent on ﬁeld strength. The splittings due to J-
couplings in Hz were also independent of ﬁeld and there was therefore similar
appearance of individual clusters at the two ﬁelds. However, in ppm units, the total
linewidth of a cluster was approximately inversely proportional to ﬁeld, and this led to
greatly improved resolution between clusters at 21 T, i.e. better chemical shift resolution.
This enhanced resolution was particularly noteworthy for the compound

T14Bi2(PS4)2(PZS6) for which the doublet and triplet were completely resolved at 21 T.

Good spectral deconvolution with Lorentzian ﬁrnctions suggested that the line
widths were dominated by the transverse relaxation rate. Such dominance may hold for

crystalline compounds that have small chemical shielding inhomogeneity at each site. In
addition, T2 may not vary strongly as a function of applied ﬁeld strength if T 2 << T 1.12
This inequality was likely valid for these compounds. It has also been observed that the
31F longitudinal relaxation rates of metal selenophosphates were linearly correlated with

the squares of dipolar couplings to the 31P.13 These couplings are independent of
magnetic ﬁeld strength and could also contribute to the transverse relaxation rates of

metal thiopho sphates.

104

CONCLUDING REMARKS

Motivations for ultra-high ﬁeld magic angle spinning spectroscopy of solids have
included improved resolution in spectra of biological macromolecules due to attenuation
of broadening ﬁ'om 13C-13C J-couplings,l4 reduction of second-order quadrupolar
broadening,15 and improved sensitivity. 16 The present study provides an example of
greatly improved chemical shift resolution at 21 T for spin 1/2 nuclei in crystalline
inorganic solids. This improvement correlated with ﬁeld-independent line widths and
splittings due to 31P-Tl J-couplings. These splittings may be useful parameters for

understanding the electronic structure of these compounds.

105

REFERENCES

1.

Gave, M.; Malliakas, C. D.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46,
3632-3644.

McGuire, M. A.; Reynolds, T. K; DiSalvo, F. J. Chem. Mat. 2005, 17, 2875-2884.

. McCarthy, T. Ph.D. Thesis, Michigan State University, East Lansing, MI, 1994.

. McCarthy, T.; Kanatzidis, M. G. J. Alloys Compd. 1996, 236, 70-85.

. McCarthy, T. J.; Kanatzidis, M. G. Chem. Mat. 1993, 5, 1061-1063.

. McCarthy, T. J.; Kanatzidis, M. G. J. Chem. Soc. Chem. Commun. 1994, 1089-1090.
. Breshears, J. D.; Kanatzidis, M. G. J. Am. Chem. Soc. 2000, 122, 7839-7840.

. Manriquez, V.; Galdamez, A.; Leon, D. R.; Garland, M. T.; Jimenez, M. Z. Krist. New

Cryst. St. 2003, 218, 151-152.

. Gave, M.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, Accepted. DOI

10.1021/ic7012113

10. Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements; 2nd ed.; Butterworth-

ll.

12.

13.

14.

15.

16.

Heinemann: Boston, MA, 1997; pp 75, 222.
Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.

Harris, R K. Nuclear Magnetic Resonance Spectroscopy; Longman Scientiﬁc and
Technical: New York, NY, 1983; p 87.

Canlas, C. G.; Muthukumaran, R B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

Straus, S. K.; Bremi, T.; Ernst, R. R Chem. Phys. Lett. 1996, 262, 709-715.

Hoatson, G. L.; Zhou, D. H. H.; Fayon, F.; Massiot, D.; Vold, R. L. Phys. Rev. B.
2002, 66, 224103-1.

Marulanda, D.; Tasayco, M. L.; Cataldi, M.; Arriaran, V.; Polenova, T. J. Phys.
Chem. B 2005, 109, 18135-18145.

106

CHAPTER 5

NEW POTASSIUM BISMUTH TI-IIOPHOSPHATES INCLUDING THE
MODULATED K1.51312.5(Ps4)3.

107

INTRODUCTION

Chalcophosphate compounds generally are semiconductors with medium to wide
energy gaps and can exhibit fascinating physical properties including pronounced
photoconductivity,l photorefractive gain},3 ferroelectricityﬂ‘6 electro-optic7 and second
harmonic generation8 properties. The chalcophosphate anions are a special class of
polyanions with broad reactivity and structural characteristics with general formula of
[PnyF' (Q=S, Se). They can be stabilized as alkali metal salts9a10 or coordinating
ligands to a variety of metal centers.”18 The chemistry of these materials has been
substantially developed in the past decade thanks to the application of the ﬂux synthesis
technique which served as a medium for reaction chemistry and tool for discovery.
Bismuth chalcogenide compounds have a great deal of structural diversity due in part to
the so-called inert lone pair effectl9'25 and the ﬂexibility in the coordination
environment including trigonal pyramidal,26 square pyramidal,27 and capped trigonal
prismatic”,29 Bismuth can bind to various chalcophosphate ligands, as for example in
the compounds KBiP2S726p30, KBiP28631 and K3Bi(PS4)2,3O which contain the [P287]4',
[P286]4' and [PS4]3‘ anions, respectively. Each anion can be selectively formed by
slightly varying the reaction conditions, and a variety of phases can therefore be
rationally synthesized.32

The K/Bi/P/S chemistry was investigated further to search for K+ analogs of the
recently reported Tl/Bi/P/S compounds.33 Due to their monovalency and similar ionic
radii, K+ and Tl” are often considered to have similar chemical reactivity and form
structurally related compounds. Additionally, related A/M/P/Q (A= K, Rb, Cs, M = Bi,
Pb, Q = S, Se) compounds exhibit a reduction in structural dimensionality as the

proportion of alkali metal is increased-"1934'36 Low-dimensional materials are of interest

108

because they may have accessible networks suitable for ion exchange reactions which

can either produce metastable isomorphic analogs, or following structural
rearrangement, new compounds.37 We describe here the compounds K3Bi3(PS4)4 [I],
K1.5Bi2.5(PS4)3 [II] and KgBl(PS4)4 [III]. The ﬁrst two feature unique extended Bi/P/S
arrangements with [H] exhibiting an incommensurate superstructure arising from a
modulation of the occupation of the Bi and T1 atoms. The third compound [III] is a

molecular salt with an eight coordinate mononuclear [Bi(PS4)4]9‘ complex.

EXPERIMENTAL SECTION
Reagents

Chemicals were used as obtained unless otherwise noted: bismuth chunks
(Tellurex Inc., Traverse City, MI, 99.999%), phosphorous (MCB Reagents, Gibbstown,
NJ, amorphous red), sublimed sulfur ﬂowers (CCI, Vernon, CA), potassium metal,
(Aldrich Chemical Co., Inc., St. Louis, MO) Bismuth was ground in an agate mortar and

pestle to ca. 100 mesh. Phosphorous was freeze-dried and K28 was prepared by a

modiﬁed literature preparation.23.38

Synthesis

Manipulations of Bi, P, and S were carried out in ambient conditions and then
transferred into a Vacuum Atmospheres Dri-Lab glovebox ﬁlled with N2 where the
appropriate amount of K28 was added.

K3Bi3(PS4)4 [I]: A mixture of K28 (0.174g, 1.6 mmol), Bi (0.3941 1.9 mmol), P
(0.0781g, 2.5 mmol), and S (0.3533g, 11.0 mmol) was loaded into a ﬁrsed silica

ampoule and ﬂame sealed under a reduced atmosphere of ~10'4 mbar. After shaking to
109

increase homogeneity, the tube was inserted into a protective ceramic sheath and placed
in a furnace. The mixture was heated to 600 °C over 12h, held there for 12h, and then
cooled back to room temperature over a period of 6h. An inhomogenous red crystalline

ingot was isolated that was a mixture of K3Bi3(PS4)4 [1] (~70%), chsBi2_5(PS4)3 [H]
(~20%), K3Bi(PS4) (5%) and Bi283 (~5%). Microprobe analysis averaged over several

sample areas of the single crystal used for the diffraction experiments gave an average

COIDPOSlthIl Ole.OBil.OP1.4SS.8'
K 1_ 5Bi2. 5(PS4) 3 [II]: An ampoule containing a mixture of KZS (0.039g, 0.35

mmol), Bi (0.2461 g, 1.2 mmol), P (0.0435g, 1.4 mmol), and S (0.1709g, 5.4 mmol) was
allowed to react under the same conditions as were used to produce compound [I]. A
homogenous red crystalline ingot was isolated that was composed of air-stable red plates

of KIO5Bi2.5(PS4)3 in quantitative yield. Microprobe analysis averaged over several

sample areas of the single crystal used for diffraction experiments gave an average
composition of K1.5Bi2.4P3.2S1 1.9.

K9Bi(PS4)4 [111]: An ampoule containing a mixture of KZS (0.414g, 3.8 mmol),
Bi (0.1749g, 0.84 mmol), P (0.1040, 3.4 mmol), and S (0.3078g, 9.6 mmol) that was
prepared in the same manner as [I] was heated to 600 °C over 18h, held there for 2h, and
then cooled back to room temperature over a period of 8h. A homogenous red
crystalline ingot was isolated that was composed of very moisture-sensitive red plates
(~90%) and greenish gelatinous decomposition products (~10%). The relative amount
of the decomposition products increased quickly and under a light microscope, it was
possible to observe regions of the sample decomposing. Microprobe analysis averaged
over several sample areas of the single crystal used for diffraction experiments gave an

average composition of K9.0Bi0.8P4.3SI6_7.

110

Ion Exchange

Because of the large thermal parameters and channel diameters, various sample
portions containing [1] were subjected to conditions wherein ion exchange might occur.
For solid state topotactic ion exchange}9 a ~0.05 mmol sample portion of compound [I]
was ground in an agate mortar and pestle with ~1.9 mmol A1 (A = Rb, Cs) and placed
into a 1 cm die. The sample was then pressed into a pellet with ~4500 kg of force. The
pellets were loaded into glass sample vials and placed into an 80°C oven for 7d. The
products were then washed twice with deionized water, twice with acetone, and once
with ether to isolate a black powder.

Solution phase ion exchange was attempted by combining a ~0.09 mmol sample
portion of compound [I], ~3.3 mmol AC1 (A = Rb, Cs) and ~10 mL deionized water in a
glass vial that was stirred overnight. The resulting black powder was washed twice with
deionized water, twice with acetone, and once with ether.

Microprobe analysis of the products ﬁ'om both the solid state and solution phase
ion-exchange experiments suggested that some exchange had occurred although the
stoichiometric ratio of the elements changed from crystal to crystal. The salt counterion
(C1' or I') was not found in signiﬁcant amounts suggesting that the Rb or Cs was in fact
the compound rather than left over from incomplete washing. Powder X-ray difﬁ’action
of the powders after solution-based ion exchange matched well with the calculated
patterns for [I] while those ﬁom the solid state ion exchange showed a similar pattern of

peaks but with a slight shift towards higher 20 angles.

11]

Powder X-ray Diffraction
All samples were assessed for phase purity using powder X-ray diffraction.
Powder patterns were obtained using an INEL CPS 120 powder X-ray diffractometer

with monochromatized Cu Ka radiation (2 = 1.540598 A) operating at 40 kV and 20

mA equipped with a position-sensitive detector with a 26 range of 0 - 120° and

calibrated with a LaB6 standard.

Single Crystal X-ray Diffraction

Intensity data for single crystals of [I] and [III] were collected on a Bruker
SMART platform CCD diffractometer using Mo K01 radiation operating at 40 kV and 40
mA. A ﬁrll sphere of data was collected and individual frames were acquired with a 10 8
exposure time and a 0.3" omega rotation. The SMART software was used for data
collection, and SAINT software was used for data extraction and reduction. An
analytical absorption correction to the data was performed and direct methods were used
to solve and reﬁne the structures with the SHELXTL4O software package. The details of
data collection and reﬁnement are summarized in Table 5-1. Fractional atomic

coordinates for [I] through [111] are displayed in Tables 5-2 through 5-4.

112

Table 5-1. Crystallographic Data, Experimental and Reﬁnement Details for the Reported Phases.

 

 

K3Bi3(PS4)4[11 K1.513i2.5(1’34)3 [11] K93i(PS4)4[IHl
Temperature, K 298 298 298
Crystal system Tetragonal Monoclinic Orthorhombic
Space group P4/ncc P2/c(a‘/2y)00 P21212
2, A (Mo Ka) 0.071073 0.071073 0.071073
a, A 21.0116(8) 21.793903) 18.306(4)
b, A 21.0116(8) 8.805(8) 8.926(2)
c, A 13.345400) 10.0301(6) 9.710(2)
11, ° 90 90 90
0, ° 90 90.013(5) 90
7, ° 90 90 90
q n/a 0.5a*+0.l6b*+0.25c* n/a
2 8 4 2
c 1 0.40 x 0.03 x 0.20 x 0.15 x
dimmgjzas, mm 0.03 0.10=8 x 0.08 x 0.04 0.08
Dcalc g/cm3 3.114 3.583 2.507
11, mm-l 19.648 24.187 7.992
Rim, % 14.2 7.98 10.6
mialreﬂwmg 611 12/3575 1 1921/6107 16260/3640
independent
6.6/13.3
ﬁnal R/Rw, o/,a 6.7/17.7 5.2/12.7 (main) 5.8/8.9
13.0/24.6 (sat)
. 4.29 1.45
Read"? elecgm (0.80 A from 5.27 (0.45 A from
dens“, A Bi(l» Bi(2))

 

a R = £(IFoI - IFCIVEIFOI' Rw = 12w(IFoI -|Fo|)2/ZWIFo|2]1/2

Table 5-2. Fractional atomic coordinates and U(eq) values for K3Bi3(PS4)4 [I] with standard deviations in

 

 

parenthesis.
atom x y z Ueqa
Bi(l) 0.4904(1) 0.1003(1) 0.3629(1) 26(1)
Bi(2) 0.3987(1) -O. 1013(1) 0.2500 26(1)
K( 1) 0.4569(3) 0.2426(2) 0.0814(4) 45(1)
K(2) 0.2500 -0.2500 0.7500 710(80)
K(3) 0.2500 0.2500 0.5340(1 10) 600(80)
P(l) 0.4260(2) 0.0615(2) 0.1261(3) 19(1)
P(2) 0.5636(2) 0.2467(2) 0.3319(3) 24(1)
S( 1) 0.3777(2) 0.0583(3) 0.2579(3) 36(1)
S(2) 0.5127(2) 0.1040(2) 0.1568(3) 18(1)
S(3) 0.3887(2) 0.1 195(2) 0.5183(3) 27(1)
8(4) 0.4419(3) -0.0291(2) 0.0762(3) 32(1)
8(5) 0.4673(2) 0.2260(2) 0.3371(4) 28(1)
S(6) 0.2728(2) -0.0926(2) 0. 1900(3) 28(1)
8(7) 0.3197(2) -0.852(3) 0.4236(3) 35(1)
S(8) 0.6082(3) 0.1663(2) 0.3767(4) 38(1)

 

a U(eq) = (XiZjUijai*aj*ai'aj)/3 x 1000

113

Table 5-3. Fractional atomic coordinates and U(eq) values for the average structure of K1 .5Biz.5(PS4)3

[II] with standard deviations in parenthesis.

 

atom x y z Ueqa Occupancy

Bil 0.3901 l4( 14) 0.11713(4) 0.85565(3) 0.02332(9)

BiZ 0.890097(14) 0.38299(4) -0.35563(3) 0.02747(9)

Bi3 0.544230 3) 0.5 0.75 0.0225(3) 0.481(3)
K3 0.544226 0.5 0.75 0.0225(3) 0.519(3)
Bi4 -0.04215(9) 0 0.75 0.0238(3) 0.507(3)
K4 -0.042152 0 0.75 0.0238(3) 0.41 1(3)
Kl 0.249920 1) 0.2501(3) 0.4944(2) 0.0376(6)

Pla 0.249950) 0.2520505) 0.9066003) 0.0212(3)

Plb 0.435530) 0.2359805) 0.5409104) 0.0212(3)

Plc 0.93547(7) 0.2639105) -0.04082(14) 0.0212(3)

S l 3 0.2712101) 0.0661(3) 0.7905(2) 0.0217(3)

82a 0.2291503) 0.4333(3) 0.7892(2) 0.0320(4)

S38 0.324240 1) 0.2963(3) 0.0230(2) 0.0249(3)

S4a 0.175760 1) 0.2008(3) 0.0228(2) 0.0268(3)

Slb 0.3852401) 0.3602(3) 0.6767(2) 0.0217(3)

S2b 0.3954802) 0.2456(3) 0.3579(2) 0.0320(4)

S3b 0.439180 1) 0.0186(3) 0.6091(2) 0.0249(4)

S4b 0.521950 1) 0.3251(3) 0.5291(2) 0.0268(4)

Slc 0.8852501) 0.1396(3) -O. 1766(2) 0.0217(3)

sze 0.8954302) 0.2540(3) 0.1422(2) 0.0320(4)

S3c 0.939000 1) 0.4813(3) -0. 1089(2) 0.0249(4)

S40 0.021940 1) 0.1751(3) -0.0290(2) 0.0268(4)

 

Table 5-4. Fractional atomic coordinates and U(eq) values for K9Bi(PS4)4 [III] with standard deviations

in parenthesis.

a U(eq) = (ZiEjUijai*aj*ai-aj)/3 x 1000

 

 

atom x y z Ueqa
Bi 0.0000 0.0000 0.7262(1) 39(1)
K(l) 0.2500(1) 0.4982(3) 1.2486(2) 37(1)
K(2) 0.2308(1) 0.182(3) 0.2405(2) 42(1)
K(3) -0.0766(1) -0.3551(3) 0.9985(3) 34(1)
K(4) 0.0000 0.0000 0.2338(3) 31(1)
K(S) 0.0877(2) 0.3721(3) 0.4841(3) 38(1)
8(1) 0.2456(2) -0.2327(3) 0.9996(5) 29(1)
S(2) 0.1041(1) -0.0053(4) 0.9481(2) 30(1)
S(3) 0.0873(2) 0.3070(3) 1.1525(3) 30(1)
S(4) 0.1013(2) 0.3462(4) 0.8159(3) 46(1)
8(5) 0.1084(1) 0.0209(3) 0.5175(3) 27(1)
S(6) 0.2294(2) 0.2609(3) 0.5246(3) 39(1)
S(7) 0.0827(2) 0.2962(3) 0.3320(3) 31(1)
S(8) 0.0603(2) 0.2893(4) 0.6672(3) 50(1)
P( 1) 0.1359(1) -0.2264(3) 0.9796(3) 18( 1)
P(2) 0.1231(1) 0.2086(3) 0.5085(3) 19(1)

 

a U(eq) = (ZiZjUijai*aj*ai~aj)/3 x 1000

114

The crystal structure of compound [1] indicated that the K“ ion sites within the
channels (vide infra) had large thermal parameters, particularly along the channel axis.
An attempt was made to model this behavior as a series of sequential cation sites with
partial site occupancy, but a signiﬁcant decrease in reliability statistics was not obtained.
Intensity data for a single crystal of [II] was collected on a Stoe IPDS II difﬁ'actometer
with Mo Ka radiation operating at 50 kV and 40 mA on with a 34 cm image plate.
Individual frames were collected with a 05° omega rotation using the X-AREA
software.41 The X-SHAPE and X-RED software packages were used for data extraction
and reduction and to apply an analytical absorption correction. The SHELXTL and
J AN A200042 software packages were used to solve and reﬁne the structure.

The single crystal XRD measurement of [II] was ﬁrst measured with an exposure
time of 2 minutes and showed reﬂections belonging to an orthorhombic cell. Analysis
of the systematic absences indicated the space group Pnna. Direct methods performed
with the SHELX program package gave a partial solution and the structure was then
reﬁned to yield a satisfying description in terms of interatomic distances and residual
electron density values, but with high thermal parameters and residual electron density.
The crystal was then remeasured with a 6 min exposure time. The intention was to check
for extra reﬂections in the diffraction pattern that would reveal additional order in the
structure in the form of a structural modulation. Modulated structures have a periodic
deviation of atomic positions, site occupancies, or thermal displacement parameters.
Therefore in the structure, translational symmetry of the unit cell is lost and is described
in the crystallographic solution as a perturbation to the unit cell. The modulation is
manifested in the diffraction pattern because main reﬂections that correspond to unit cell

periodicity are observed along with weaker satellite reﬂections that correspond to the

115

period of the modulation. These satellite reﬂections are found in the difﬁaction pattern
along the modulation vector q in the reciprocal lattice.43p44 The reciprocal lattice
showed weak ﬁrst order satellite reﬂections, incommensurate with the orthorhombic cell

that could be indexed with the one-dimensional vector q = [1/4, 1/2, 0.16].

This modulation vector can not be combined with orthorhombic symmetry,45 so
the crystal class was lowered and the 3D space group changed to the highest isomorphic
subgroup P2/n. This space group contains two (3+1)-dimensional superspace groups,
P2/c(a‘/2y)00 and P2/c(a‘/2y)0s, of which only the ﬁrst gave a useﬁrl result after
subsequent reﬁnement. Atomic positions were transferred from the Pnna model,
yielding 22 positions, and the main reﬂections quickly reﬁned to low residual values

(Robs/Ra“=0.034/0.068) for the average structure. Then, modulation waves were applied

to the structure to account for satellite reﬂections. Occupational modulation waves were
applied to the mixed K/Bi sites and positional modulation waves were applied to all

atoms in the structure.

The [PS4]3' tetrahedra were modeled as being internally rigid to decrease the

number of crystallographic parameters in the reﬁnement. One [PS4]3' anion was chosen

as the molecular model and the atomic parameters were ﬁrst reﬁned individually. The
molecular positions and parameters common for the entire molecule were then reﬁned in
a rigid body approximation. The ﬁnal overall reliability statistics converged to Rwa =
6.6/ 13.3% (R/Rw = 5.2/ 12.7% for main reﬂections and R/Rw = 13.0/24.6% for satellite
reﬂections) for 11921 reﬂections (I/o(I)>3) and 158 parameters.

In an alternative attempt, it was possible to divide the q vector into a rational
part, q" = [0, V2, 0], and an irrational part, qi = [0.25, 0, 0.16]. The rational part could
then be excluded by applying a centering condition. This resulted in the super space

116

group X2/c(a0 y)00, where X corresponded to the (3+1)-dimensional centering [0, V2, 0,
1/2]. However, this required that the b-axis was doubled and so the overall gain in such a
description for an already large unit cell was limited and was therefore not considered.

Only 35% of the satellite reﬂections met the observation criteria of I>3o and the absence
of higher orders made the modulation difﬁcult to model precisely. The ﬁnal residual
values are about 13% for the satellites but the low overall R value (6.61%) reﬂects the
minor effect of the modulation on the structure. The exchange of Bi for K in two
different sites can apparently be made smoothly, with no drastic effects on the rest of the

structure and thus the weak intensity of the satellite reﬂections.

Electron Microscopy

A JEOL JSM-35C scanning electron microscope equipped with a Tracor
Northern energy dispersive spectroscopy detector was used for quantitative microprobe
analysis. Data were collected using an accelerating voltage of 25 kV and a collection

time of 60 s and results were averaged over several sample areas.

Differential Thermal Analysis
Differential thermal analyses were performed with a Shimadzu DTA-50 thermal
analyzer. A ground sample weighing ~25 mg was sealed in a quartz ampoule under

reduced pressure. An equivalent mass of alumina was sealed in an identical ampoule to
serve as a reference. The samples were heated to 850 °C at a rate of 10 °C/min, cooled
to 150 °C at a rate of 10 °C/min, reheated to 850 °C at a rate of 10 °C/min, and then

cooled to room temperature at a rate of 10 °C/min.

117

Raman Spectroscopy

Raman (750-100 cm‘l) spectra were collected using a BIO-RAD FT
spectrometer equipped with a Spectra-Physics Topaz T10-106c 1.064 nm YAG laser and
a Ge detector. Samples were ground into a ﬁne powder and loaded into fused silica

tubes and 64 scans were averaged.

Solid State NMR

Room-temperature 31P solid-state NMR spectra were collected on a 9.4 T
spectrometer (V arian Inﬁnity Plus) using a double resonance magic angle spinning
(MAS) probe. Samples were spun at ﬁequencies ranging ﬁom 5 kHz to 15 kHz in 4 mm
outer diameter zirconia rotors and a ~50 uL sample volume. Bloch decay spectra were
taken with a 4 us 90° pulse (calibrated and referenced to 85% H3PO4 at 0 ppm) and
relaxation delays ranging ﬁ'om 10 to 5000 s. Each spectrum was processed with 10 Hz

line broadening and up to a 10th order polynomial baseline correction.

The longitudinal relaxation time (T 1) of each compound was estimated by

ﬁtting to the equation:
S(t)=so(1—e"”1) (1)
where ‘l' is the experimental delay time, S(t) is the experimental signal intensity, and S0 is

a ﬁtting parameter representing the signal intensity at r = 00 .46

RESULTS AND DISCUSSION
Synthetic Reactions

The compounds K1.5Bi2.5(PS4)3 [H] and K9Bi(PS4)4 [HI] were produced by

direct stoichiometric reaction of KZS, Bi, P, and S, however, we were not able to prepare

118

K3Bi3(PS4)4 [I] as the dominant compound without additional K28x in the reaction.
Presumably, the K2Sx acted as a ﬂux in which [I] could grow. If KXSy was not added,

[11] was produced in high yields. When the value of x in KZSX was decreased,

K3Bi(PS4)230 was produced along with a signiﬁcant amount of Bi2S3. The synthesis of
KgBi(PS4)4 required even more basic KZSx compositions. In syntheses of compound [I]
we always observed compound [H] and K3Bi(PS4)2 present in detectable quantities.

Compounds [1] and [II] were stable in ambient conditions, but [III] was highly
moisture sensitive and decomposed rapidly when exposed to air. This reactivity can be
attributed to the presence of the highly charged [Bi(PS4)4]9' molecular anion in the
structure.47

In an attempt to prepare pure K3Bi3(PS4)4, a melt of the same composition was

quenched in ice water. The intention was to kinetically trap a solid with the correct
stoichiometry which could then be annealed at a given temperature to produce

crystalline K3Bi3(PS4)4. The resultant solid was an amorphous glass as conﬁrmed with

powder X-ray difﬁ'action. Differential thermal analysis of the glass indicated the onset
of crystallization at about 320 °C. The sample was then careﬁrlly annealed at 320 °C but

the resulting product was found to be K1.5Bi2.5(PS4)3 with no evidence for compound
[I]. It is likely that compound [I] is stabilized by the presence of a KZSx ﬂux and is not

just kinetically stable.

119

Structure Description
K33i3(PS4)4 [I]!

This compound has a complex 3-dimensiona1 [Bi3(PS4)4]3' anionic ﬁamework
with inﬁnite parallel tunnels running through it, Figure 5-1A. The tunnels are parallel to
the c-axis and are ﬁlled with charge balancing K+ ions. There are two types of tunnels in
the structure. Type-I tunnels are located at (V4, V4) in the ab-plane and have K2 atoms in
their center. Type-II tunnels are located at (V4, 3/4) and accommodate the K1 and K3

atoms. Each tunnel type is surrounded by four other tunnels of the other type.

 

 

 

 

(A) ”- . . (B)
‘ I. " K1 7
C C C C
C C .. C C
C C C C
C C C C

 

L. 4‘

Figure 5-1. (A) K3Bi3(PS4)4 viewed down the [001] axis showing the large channels that are occupied by
K+ ions. (B) An undulating Bi3S12 chain that forms the walls of the tunnels.

The K(l) atoms are located at the tunnel edge interacting with S atoms with an
average K-S bond length of 3.33 A, Figure 5-2A. The K(3) central axes are located at
the tunnel center with K-S distances of ~4 A, Figure 5-2C. The K(3) atoms are located
in the same tunnel as the K(l) atoms but instead of being placed on the edge, they are
situated along the walls of the tunnel. The K(2) site is distorted square planar with K(2)-

S interactions of 3.436(5) A, Figure 5-2B. The K2 and K3 ion sites had large thermal
120

parameters and the possibility of this being due to long range order was considered.
Zone photos obtained with X-ray difﬁaction using relatively long exposure times did not
reveal any evidence for weak satellite reﬂections and therefore signiﬁcant contributions
ﬁ'om a modulation are unlikely. Therefore the large thermal parameters are probably due
to static positional disorder of the K+ atoms in the oversized ttmnels. The greatest
displacement in each case was along the (001) axis which suggests the possibility of

high ion mobility within the ttmnels.

 

Figure 5-2. The coordination environments of (A) K(1), (B) K(2), and (C) K(3) ions in K3Bi3(PS4)4.

The size and shape of the two types of tunnels were approximated by plotting the
isosurface that lies between the ﬁamework atoms, Figure 5-3.48 The smaller tunnel
(type-I) has a maximum diameter of ~3.1 A, and the larger one (type-II) has a maximum
diameter of ~4.4 A. The two types of tunnels do not seem to be connected by holes of

signiﬁcant girth.

121

.af'
“‘7 ::=-- ~<
“(21-c

- —r
- \,.

“Car;

.-
‘—
4 ‘

L
Cf;

 

Figure 5-3. Void space representation of K3Bi3(PS4)4 showing the relative sizes and contours of each of
the channels viewed (A) perpendicular to [001] and (B) parallel to [001].

The [Bi3(PS4)4]3' framework itself is built ﬁom undulating chains of Bi-S
polyhedra. If the P-atoms are ignored, the structure is made of pairs of parallel Bi3S12
chains, Figure 5-1B. In each chain, the distorted Bi polyhedra share corners or edges.
Bi(l) is pseudo 8-coordinate and is chelated by three [PS4]3‘ anions. Seven of the Bi—S
distances are relatively short with an average bond length of 2.91 A. The remaining
interaction is between Bi(l) and S(4) distance at 3.451(6) A and is likely elongated due
to the stereochemically active so-called 6s2 lone pair of electrons on Bi. The
stereochemical activity presumably arises from an interaction between the Bi 6s valence
electrons and the chalcogen valence p-orbitals that results in an asymmetric distribution
of the former.49 Two Bi( 1) atoms are bridged by S atoms with Bi(l)-S(2) distances of
2.791(4) and 3.026(4) A. Bi(l) is also connected to Bi(2) by a bridging S atom. Bi(2) is
pseudo 8-coordinate and is chelated by three [PS4]3' anions and forms the inner wall of
the tunnel. Six of the Bi-S distances are relatively short with an average bond length of
2.85 A. The putative lone pair of electrons on Bi is suggested by the elongated Bi(2)-

S(1) of distance of 3.384(6) A. The action of the 4-fold axis on the chain of

122

[Bi3(PS4)4]3' forms into the tunnel. Selected bond lengths and angles are displayed in

Table 5-5.

Table 5-5. Selected bond lengths and angles for K3Bi3(PS4)4 with bond lengths (A) and angles (°), with
standard uncertainties in parenthesis

Bil—S5
Bil—82
Bil—S8
Bil—S1
Bil—S3
Bil—82i
Bil—S4ii
Bi2—S6
Biz—S6i
Biz—S7
Bi2—S71
BiZ—S4
Biz—S4i

1(1—86vii
Kl—SZ
Kl—s3viii
1(1—85ix
1(1——s3ix
Kl—SS
1:1—S7viii
1(1—81vii
K2—S6"
1(2—86xi
K2—S6i"
K2—S6Xii
KZ—S 8x111
itz—ssxiv
1(2—88ii
KZ—SSXV
K3—S3
K3—S3V
K3—S3Vii
1(3—83xvi
K3_Kl XVII
1(3—141iii
K3—K1ii
K3—xlxi
K3—S5V

2.707(5)
2.791(4)
2.844(5)
2.888(5)
3.006(5)
3.026(4)
3.091(5)
2.770(4)
2.770(4)
2.870(4)
2.870(4)
2.916(5)
2.916(5)

3.259(7)
3.296(6)
3.302(7)
3.334(7)
3.340(7)
3.437(7)
3.663(8)
3.703(7)
3.437(5)
3.437(5)
3.437(5)
3.437(5)
3.850(6)
3.850(6)
3.850(6)
3.850(6)
4.006(9)
4.006(9)
4.006(9)
4.006(9)
440(2)

440(2)

440(2)

440(2)

529(7)

S5—Bil—S2
S5—Bil—S8
sz—Bil—ss
SS—Bil—Sl
sz—Bil—s1
S8—Bil—Sl
S5—Bil—S3
S2—Bil—S3
S8—Bil—S3
Sl—Bil—S3
ss—Bil—szi
sz—Bil—szi
ss—Bil—szi
Sl—Bil—szi
s3—Bil—szi
ss—Bil—S4ii
s2—Bil—S4ii
ss—Bil—S4ii
81—B11—84ii
s3—Bil—84ii
szi—Bil—S4ii
S6—Bi2—S6i
S6—Bi2—S7
86i—B12—s7
S6—B12—87i
86i—B12—-s7i
s7—B12—s7i
S6—B12—S4
S6i—B12—S4
s7—Biz—S4
S7i—B12—S4
S6—B12—S4i
S6i—B12——S4i
s7—Biz——S4i
S7i—B12-—S4i
S4—B12——84i

S6vii—K1—sz
S6‘m—K1—S3vm
S2—Kl—S3V‘“

123

82.9903)
71.8605)
84.50(l4)
95.1306)
70.6402)
153.3805)
80.1503)
142.7302)
120.5805)
7804(13)
136.3703)
78.40(12)
67.3803)
114.7604)

l34.64(ll)
116.3605)

150.6403)
81.4404)
125.0604)
65.7802)
7242(12)
102.1(2)
70.9503)
85.2505)
85.2505)
70.9503)
142.0(2)
91.8705)
134.9503)
139.5205)
67.8103)
134.9403)
91.87(15)
67.8103)
139.5205)
107.9709)

7407(15)
134.5709)
60.8503)

83Viii—x1—81vii
S5ix—x1—81vii
S3 ix—1<1—81Vii
ss—xl—SI"ii
s7Viii—K1—81vii
S6x—K2—S6Xi
S6x—K2—S6i"
S6Xi—K2—S6i"
S6x—ltz—86xii
S6Xi—K2—S6’di
S61V_K2_S6Xll
S6x—K2—S8Xiii
S6xi—K2—S8Xiii
S6iv_K2_ngiii
S6xii_K2_38xiii
86x——1<2—-S8xiv
S6X1_K2_88X1V
S6iV—K2—S8’dv
S6Xii—K2—S8Xiv
ngiiLKz_ngiv
S6X—K2—S8ii
soxi—Kz—ssii
S6iV—K2—ssii
S6Xii—K2—S8ii
S8Xiii—K2—S8ii
SSXiV—KZ—S8ii
S6x—K2—S8x"
S6xi—K2—S8XV
S6iV—1t2—88xv
S6Xii—K2—S8XV
S8Xiii—K2—S8XV
SBXiV—KZ—SSXV
ssii—K2—88XV
s3—K3—s3V
s3—K3—s3vii
s3V-—K3-—s3vii
s3—K3—s3xvi
s3V-—1<3—s3xvi
s3vii--1(:3—s3xvi
s3——l<3—ssv

139.5(2)
ll8.42(l6)
57.2103)
56.77(l3)
134.7209)
153.0505)
77.6205)
108.8405)
108.84(15)
77.6205)
153.0405)
64.2901)
103.2401)
141.5501)
54.6601)
103.2401)
64.2901)
54.66(1l)
141.5501)
127.8906)
141.5501)
54.6601)
64.2901)
103.2401)
154.0905)
59.1006)
54.6601)
141.5501)
103.2401)
64.2901)
59.1006)
154.0905)
127.8906)
89.8(2)
89.8(2)
174(4)
174(4)
89.8(2)
89.8(2)
56.8(8)

Table 5-5 continued

 

Pl—s3Viii

 

2.020(6) 86Vii—K1—S5ix 109.8109) S3V—K3—SSV 44.1(7)
Pl—Sl 2.031(6) 82—1t1—85ix 116.7409) s3Vii—x3—ssv 132(2)
P1—S4 2.044(6) s3viii—K1—ssix 86.9806) s3XVli—K3—ssv 119.708)
Pl—SZ 2.071(6) S6Vii—1<1-—s3ix 108.5808) levii—K3—S5V 157(3)
1>2—s7vii 2.015(6) sz—Kl—s3ix 174.7(2) Kliii—K3—85V 101.3(9)
P2—S8 2.022(7) s3viii—K1—s3ix 116.8(2) Klii—KB—SS" 38.9(5)
1>2—S6Vii 2.063(6) SSiLKl—S3i" 67.02(l4) KlXi—KB—SSV 87.0(7)
P2—SS 2.071(7) S6Vﬁ—K1—ss 61.5204)
SZ—Kl—SS 65.47(13) s3viii—P1—81 113.4(3)
s3viii—K1—ss 101.5107) s3viii—P1—S4 109.2(3)
ssiLxl—ss 170.7(2) Sl—Pl—S4 109.4(3)
S3ix—Kl—SS 1 1 1.460 8) s3viii—P1—sz 109.5(3)
86Vii—x1—S7viii 71.44(l6) Sl—Pl—SZ 106.4(3)
sz—K1—s7viii 66.2704) S4—Pl—SZ 108.8(3)
s3Viii—1t1—S7viii 84.9807) S7vii—P2—ss 110.6(3)
S5ix—K1—s7viii 57.3404) S7Vii—P2—S6vii 106.8(3)
S3ix—Kl—S7Vﬁi 118.7508) ss—1>2—S6Vii 110.9(3)
S5—K1—S7Viii 1 19.2008) S7vii—P2—ss 1 1 1.1(3)
S6Vii—Kl—S1Vii 69.1305) S8—P2—SS 105.5(3)
SZ—Kl—SlVii 120.950 7) S6vii—P2—ss 1 12.0(3)
(i) 0.5+y, -0.5+x, 0.5-z; (ii) 0.5-y, 0.5-x, 0.5+z; (iii) x, 0.5-y, 0.5+z; (iv) 0.5+y, -x, l-z;
(v) y, 0.5-x, 2; (vi) l-x, -0.5+y, 0.5-z; (vii) 0.5-y, (viii) 0.5-y, 0.5-x, -0.5+z;
(ix) x, 0.5-y, -0.5+z; (x) x, -0.5-y, 0.5+z; (xi) 0.5-x, y, 0.5+z; (xii) -y, -0.5+x, l-z;

(xiii) -0.5+x, -0.5+y,
(xvii) y, x, 0.5+z; (xviii) l-x, -y, -z.

l-z; (xiv) l-x, -y,

l-z; (xv) y, -1+x, 0.5+z; (xvi) 0.5-x, 0.5-y, z;

K1.5Bi2.5(PS4)3 [11]:

The structure of this compound is incommensurately modulated, and has a 3D
framework structure, Figure 5- 4A. K1.5Bi2.5(PS4)3 [II] adopts the monoclinic super
spacegroup P2/c(aV2y) with the modulation vector having a*, b* and c* components.
The related compound K3Ce2(PS4)3.50 was also found to be modulated and has the
super spacegroup C2/c( a0 y).

In the orthorhombic model before the modulation was introduced, there was one
Bi3+ site (Bi(l)), one K+ site (K(2)), and one site that could only be described as being

mixed between K+ and Bi3+ (K(3)/Bi(3)) with a 50% occupancy, Figure 5-5. The

124

framework is built of interconnected “B12(PS4)4” double chains, Figure 5-4B. Bi(2)
atoms are bridged by S(l 1) along the a-axis in a zigzag pattern to form a single chain.
Two chains are connected by a P(2) centered [PS4]3' ion to form the double chains.
Neighboring double chains are then interconnected by bridging P(l) centered [PS4]3‘

ions and by K(3) ions that undulate down the a-axis. The K(3)/Bi(3) and K(4)/Bi(4)
mixed sites are located in the channels between the chains. Selected bond lengths and

angles are given in Table 5-6.

s

(A) I‘A". A"
3.4! "ls-ﬂ "
>1:'

*9. 39

. -. 1'

    

'

k.
"

 

 

Figure 5-4. (A) The 3D approxirnant structure of K1.5B12.5(PS4)3 as viewed down [100] showing the
assembly of the chains centered at V2 the [010] and [001] axes, and (B) the assembly of “Biz(PS4)4”
chains, emphasized in a rectangular outline. (B) An individual “Biz(PS4)4” chain as viewed down the
[001] axis.

In the case of compounds [I], [III], KBiPZS7,26t30 KBiPZS631 and K3Bi(PS4)230
no cation site disorder was observed, and therefore the possibility of a structural
modulation rather than random disorder was considered. X-ray diffraction data with a

longer exposure time was subsequently collected to look for satellite reﬂections

indicating long range order. The data revealed ﬁrst order reﬂections that could be

125

indexed with the modulation vector q = 0.25a*+0.5b*+0.16c*.

Higher order satellite

reﬂections were not observed.

 

Figln'e 5-5. The coordination environment of (A) Bi(l), (B) K(2), and (C) the K(3)/81(3) mixed site in
K153250503-

Table 5-6. Selected bond lengths for K1.5Bi2.5(PS4)3 with bond lengths (A), with standard uncertainties
in parenthesis.

 

average min max
Bil-Sla 2.712(5) 2.704(6) 2.719(6) Sla-Pla-S2a 110.2(2)
Bil-Slb 2.797(5) 2.740(5) 2.852(5) Sla-Pla-S3a 107.3(2)
Bil-S3a 2.715(5) 2.689(5) 2.740(5) Sla-Pla-S48 108.9(2)
Bil-S2b 3.193(5) 3.020(5) 3.371(5) S2a-P1 a-S3a 1 l 1.1(2)
Bil-S3b 2.829(5) 2.796(5) 2.866(5) S2a-Pla-S4a 109.1(2)
Bil-S3b 3.006(5) 2.970(5) 3.046(5) S3a-Pla-S48 l 10.2(2)
Bil-S4b 2.895(5) 2.876(5) 2.919(5)
BiZ-S2a 2.720(6) 2.717(5) 2.724(5) K4-Slc 3.063(5) 3.012(5) 3.107(5)
Bi2-S48 2.729(5) 2.724(5) 2.735(5) K4-Slc 3.064(5) 3.011(5) 3.108(5)
BiZ-Slc 2.797(5) 2.763(5) 2.833(5) K4-S2c 3.141(5) 3.106(5) 3.169(5)
BiZ-S2c 3.200(5) 3.155(5) 3.241(5) K4-S2c 3.140(5) 3.106(5) 3.170(5)
Bi2-S3c 2.831(5) 2.821(5) 2.839(5) K4-S4c 2.967(5) 2.952(5) 2.981(5)
BiZ-S3c 3.004(5) 2.992(5) 3.015(5) K4-S4c 3.072(5) 3.056(5) 3.087(5)
Bi2-S4c 2.892(5) 2.878(5) 2.907(5) K4-S4c 2.968(5) 2.951(5) 2.982(5)
K4-S4c 3.073(5) 3.056(5) 3.087(5)
Bi3-Slb 3.069(5) 3.043(5) 3.099(5)
Bi3-Slb 3.073(5) 3.043(5) 3.098(5) Kl-Sla 3.419(7) 3.382(7) 3.450(7)
Bi3-S2b 3.132(5) 3.036(5) 3.223(5) Kl-Sla 3.487(7) 3.465(7) 3.506(7)
Bi3-S2b 3.126(5) 3.036(5) 3.223(5) Kl-SZa 3.400(7) 3.377(7) 3.421(7)
Bi3-S4b 2.974(5) 2.899(5) 3.055(5) Kl-SZa 3.499(7) 3.452(7) 3.542(7)
Bi3-S4b 3.063(5) 3.007(5) 3.124(5) Kl-Slb 3.604(8) 3.597(8) 3.612(8)
Bi3-S4b 2.980(5) 2.899(5) 3.056(5) Kl-S2b 3.452(8) 3.385(8) 3.531(8)
Bi3-S4b 3.067(5) 3.006(5) 3. 124(5) Kl-S 1c 3.600(8) 3.554(8) 3.649(8)
Kl-S2c 3.451(9) 3.428(9) 3.476(9)
K3-Slb 3.069(5) 3.043(5) 3.099(5)
K3-S1b 3.073(5) 3.043(5) 3.098(5) Pla-Sla 2.062(5) 2.062(5) 2.062(5)
K3-S2b 3.132(5) 3.036(5) 3.223(5) Pla-S2a 2.035(5) 2.035(5) 2.035(5)
K3-S2b 3. 126(5) 3.036(5) 3.223(5) P1 a-S3a 2.034(6) 2.034(6) 2.03 5(5)
K3-S4b 2.974(5) 2.899(5) 3.055(5) Pla-S4a 2.045(6) 2.044(6) 2.045(5)

126

Table 5-6 continued

 

K3-S4b 3.063(5) 3.007(5) 3.124(5)
K3-S4b 2.980(5) 2.899(5) 3.056(5)
K3-S4b 3.067(5) 3.006(5) 3.124(5)
Bi4-Slc 3.063(5) 3.012(5) 3.107(5)
Bi4-Slc 3.064(5) 3.011(5) 3.108(5)
Bi4-S2c 3.141(5) 3.106(5) 3.169(5)
Bi4-S2c 3.140(5) 3.106(5) 3.170(5)
Bi4-S4c 2.967(5) 2.952(5) 2.981(5)
Bi4-S4c 3.072(5) 3.056(5) 3.087(5)
Bi4-S4c 2.968(5) 2.951(5) 2.982(5)
Bi4-S4c 3.073(5) 3.056(5) 3.087(5)

 

Upon transformation to P2/c(01sz)00, the mixed K/Bi position from the
orthorhombic model was split into two positions (K(3)/Bi(3)) and K(4)/Bi(4)), and all
thermal displacement parameters were reasonable. The application of an occupational
modulation indicated that the modulation between K(4) and Bi(4) had a small amplitude
whereas the modulation between K(3) and 3(3) had a larger amplitude, Figure 5-6. A
summary of the applied modulation waves is displayed in Table 5-7. After taking into

account this occupational modulation, the stoichiometry reﬁned to K1.47Bi2.49P3S12, in

close agreement with the microprobe analysis. A positional modulation was then

applied to all atoms in the structure as well as to the rigid [PS4]3' units. The largest

positional deviations (~ 0.1 A) were observed in Bi(l) and the [PS4]3' tetrahedron

closest to it. The modulation over several unit cells is displayed in Figure 5-7.

127

 

_‘L
(DO
000

O

 

% Occupancy
N 3 O)

0

0.0 0.2 f 0.4 0.6 0.8 1.0
Fractional distance

 

 

 

Figure 5-6. Percent occupancies of the complementary modulated sites K(3)/Bi(3) and K(4)/Bi(4) in
K1.5B12.5(PS4)3 with the ordinate representing the fractional distance along the modulation axis.

128

 

Figure 5-7. View of K1.5Bi2_5(PS4)3 down the b axis, showing how the K and Bi atoms interchange
differently along the c axis for the K3/Bi3 and K4/Bi4 pairs. The subcell is displayed in black lines.

129

Table 5-7. Modulation parameters for K1 5312. 5(PS4)3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Occupation waves x -0.0009(2) -0.0008(2)
Ucos Usin Sla y -0.0007(4) -0.0009(3)
B13 -0.05 12(7) 0.366(5) 2 0.0025(4) 0.005 1(3)
K3 0.0512(7) -0.366(5) x 0.0003(2) 0.0010(2)
Bi4 -0.062(4) 0.0083(6) S2a y 0.0001(4) 0.0004(3)
K4 0.062(4) -0.0083(6) 2 0.0030(4) 0.0090(3)
Displacement waves x -0.00019(18) -0.00188(17)
Ucos Usin S3a y -0.0022(5) -0.0072(5)
x -0.0011 1(5) -0.00083(5) 2 0.0028(4) 0.0110(4)
Bil y -0.01021 (9) -0.00568( 10) x -0.0005 1(18) 00023407)
2 0.00604(9) 0.00927(8) S4a y 0.0008(5) -0.0028(5)
x 0.00004(5) 0.00002(5) 2 0.0026(4) 0.0030(4)
Bi2 y 0.0001301) 0001 15(1 1) x -0.0015(2) -0.00105(19)
z -0.00096(10) -0.00114(10) Slb y -0.0030(5) -0.0028(4)
x 0.001070 1) 0.000150(15) 2 0.0043(4) 0.0095(3)
Bi3 y 0.00108(4) -0.0077(3) x 00015309) 00026008)
2 0.0016009) 0.00022(3) S2b y -0.0039(5) -0.0072(4)
x 0.0010701) 0.000150(15) 2 0.0045(4) 0.0105(3)
K3 y 0.00108(4) -0.0077(3) x -0.00183(l8) -0.00083(18)
2 0.0016009) 0.00022(3) S3b y -0.0035(4) -0.0032(4)
x 0.00008 1(16) 0.0006102) 2 0.0050(4) 0.0119(4)
Bi4 y -0.0025(2) 0.00033(3) x 00014808) 00018008)
2 —0.00025(3) -0.0019 1(19) S4b y -0.0043(5) -0.0034(4)
x 0.00008106) 0.0006102) 2 0.0045(4) 0.0063(4)
K4 y -0.0025(2) 0.00033(3) x -0.00089(l9) -0.0008(2)
z -0.00025(3) -0.00191(19) Slc y 0.0028(5) 0.0030(5)
x -0.0009(3) -0.0001(3) 2 -0.0009(4) -0.0016(4)
Kl y 0.0003(7) 0.0060(7) x -0.00073(19) -0.00082(l9)
2 0.0035(6) 0.0030(6) S2c y 0.0033(5) 0.0021(5)
x 00003405) 00010204) 2 -0.0011(4) -0.0014(4)
Pla y -0.0005(3) -0.0026(3) x 00004308) -0.00042( 18)
2 0.0027(2) 0.0071(2) S3c y 0.0025(4) 0.0024(4)
x -0.00158(l3) -0.00158(l2) 2 -0.0014(4) -0.0008(4)
Plb y -0.0037(3) -0.0042(3) x 00008408) 00008508)
2 0.0045(3) 0.0096(2) S4c y 0.0019(4) 0.0016(4)
x -0.00072(l3) -0.00073(l3) 2 -0.0010(4) -0.0013(4)
Plc y 0.0026(3) 0.0023(3)
2 -0.0011(2) -0.0013(2)
K9Bi(PS4)4 [III]:

Compound [III] has discreet molecules of [Bi(PS4)4]9' and is structurally related

to RbgCC(PS4)447 and K9Ce(PS4)4,51 Figure 5-8A. The [Bi(PS4)4]9' molecules are

arranged in layers parallel to the bc-plane with the K+ atoms in between. The molecular

complex [Bi(PS4)4]9' anion features two chelating [PS4]3' ligands and two monodentate

130

[PS4]3‘ anions, Figure 5- 8B. The average Bi-S distance is 2.86 A. The next-nearest S
atoms on the monodentate [PS4]3' are 3.708(4) A away from Bi. K+ is in a distorted

bicapped octahedral site with K—S distances averaging 3.38 A, Figure 5-9. The P-S

distances average 2.04 A. Selected bond lengths and angles are displayed in Table 5-8.

 

Figm'e 5-8. (A) View down [010] of KgBi(PS4)4 and (B) a view of an individual molecule of [Bi(PS4)]9'.

131

 

Figure 5-9. The coordination environment of (A) K(1), (B) K(2), (C) K(3), (D) K(4), and (E) K(5) ions in
K930804-

132

Table 5-8. Selected bond lengths and angles for K9Bi(PS4)4 with bond lengths (A) and angles (°), with

standard uncertainties in parenthesis.

Bi—S5i
Bi—SS
Bi—S8
Bi—ssi
Bi—SZ
Bi—szi

Kl—Slii
1(1—86ii
1(1——s2ii
Kl—Sl
Kl—SSii
Kl—S6iﬁ
Kl—S3
1(1——S7iii
1(1—1’1ii
xl—Pzii
K2—S1Vi
KZ—SlVii
K2__S6VIII
x2—84vii
K2—S6
K2—S5
1(2—82vi
K2—P2
1(2—Plvi
K2—S7
K3—S4V
1(3—87x
1(3—88i
K3—S2i
1(3—81xi
K3—S3V
x3—s3
K3—S4
K4—S7
1(4—87i
1(4—83vi
1(4—83xii
K4—szvi
1(4—82xii
K4—SS
1(4—85i
K5—S5

2.842(3)
2.842(3)
2.866(4)
2.866(4)
2.877(3)
2.877(3)

3.194(5)
3.215(4)
3.284(3)
3.386(5)
3.453(3)
3.457(4)
3.558(4)
3.661(4)
3.664(3)
3.790(3)
3.250(5)
3.251(5)
3.317(4)
3.348(4)
3.508(4)
3.518(3)
3.673(3)
3.680(3)
3.769(3)
3.781(4)
3.234(4)
3.283(4)
3.284(4)
3.292(4)
3.347(4)
3.372(4)
3.380(4)
3.710(4)
3.192(3)
3.192(3)
3.268(3)
3.268(3)
3.366(3)
3.366(3)
3.400(3)
3.400(3)
3.174(4)

SSI—Bi—SS
ssi—Bi—ss
ss—Bi—ss
ssi—Bi—ssi
ss—Bi—ssi
S8—Bi—88i
ssi—Bi—sz
S5—Bi—S2
S8—Bi—S2
ssi—Bi—sz
S5i—Bi—szi
ss—Bi—-szi
ss—Bi——szi
ssi—Bi—szi
s2—Bi—szi
Slii—Kl—S6ii
Slii—xl—szii
S6ii—1tl—82ii
Slii—Kl—Sl
S6ii—Kl—Sl
szii—K1—81
Slii—Kl—ssii
S6ii—Kl—ssii
szii—Kl—S5ii
Sl—xl—ssii
Slii—Kl—S6iii
S6ii—xl—S6iii
szii—Kl—S6iii
Sl—Kl—S6iii
S5ii—K1—S6iii
Slii—Kl—s3
S6ii—K1—s3
szi'L—Kl—s3
Sl—Kl—S3
ssii—xl—s3
S6iii—K1—s3
Slii—Kl—S7iii
S6ii—1tl—s7iii
szii—ltl—87iii
s1—1<1—s7iii
ssii—Kl—S7iii
S6111—1t1—s7m
s3—1(.1—s7iii
SlVi—KZ—SlVii

89.050 1)
9386(8)
6938(8)
6938(8)
9386(8)
156.9502)
175.95(8)
9404(7)
8479(9)

1 1295(9)
9404(7)
175.95(8)
1 1295(9)
8479(9)
83.0(1)
170.200 1)
6185(8)
108.4700)
8544(5)
8934(1 1)
6751(9)

1 1605(1 1)
6025(8)
76.740)
121.9100)
100.370 1)
8594(5)
121830 1)
170520 1)
62.300)
9790(9)
8614(9)
12253(9)
57.l3(8)
146.0400)
1 1424(9)
8191(9)
107.89(9)
143060 1)
119.28(9)
1 1694(8)
5493(7)
6631(8)
8679(5)

133

ssi—K3—s3V
szi—K3—S3V
SIXi—KB—SBV
S4V—K3—S3
s7X—K3—s3
ssi—K3—83
szi—K3—s3
Slxi—K3—s3
S3V—K3—S3
S4V—K3—S4
S7X—K3—S4
S8i—K3—s4
S2i—K3—S4
Slxi—K3—S4
s3V—K3—S4
S3—K3—S4
S7—K4—s7i
87—1<:4—-83vi
S7i—K4—S3vi
s7—K4—s3xii
s7i—K4—s3xii
s3vi—K4—s3xii
s7—K4—szvi
S7i—K4—szvi
s3vi—K4—szvi
s3xii—x4—szvi
S7—K4—szxii
S7i—K4—82xii
s3Vi—K4—szxii
S3Xii—K4—SZXii
szvi—K4—szxii
S7—K4—S5
S7i—K4—S5
s3vi—x4—85
s3xii—K4—S5
82Vi—x4—85
szxii—K4—ss
S7—K4—S5i
S7i—K4—85i
s3Vi—x4—ssi
S3xi'L—K4—ssi
szvi—K4—ssi
szxii—K4—ssi
ss—x4—ssi

126.8500)
158.6001)
9869(9)
118.1301)
6473(8)
109.2501)
9443(8)
146.1802)
8821(9)
83.0(1)

1 1986(10)
5648(9)
9241(9)
148.3402)
106.37(9)
5520(8)
145.2303)
122.32(8)
6703(7)
6703(7)
122.32(8)
1520503)
8939(8)
120.21(9)
6083(8)
9512(9)
120.21(9)
8939(8)
9512(9)
6083(8)
6898(10)
61.75(7)
8939(9)
8221(7)
121.80(8)
109.67(6)
177.32(8)
8938(9)
61.740)
121.80(8)
8221(7)
177.32(8)
109.67(6)
71.7800)

Table 5-8 continued

 

 

K5—S4 3.239(4) Sl"i—1<2——S6viii 8996(11) S5—K5—S4 7956(10)
1(5—83vi 3.272(4) Slvii—K2—86viii 159.6301) ss—x5—s3vi 8570(9)
1(5——s7xiii 3.310(4) SlVi—K2—84Vﬁ 93.1400) S4—K5—S3Vi 165.0502)
1(5—S8i 3.324(4) Slvii—Kz—S4vii 6090(8) ss—-1<5—87xiii 158.7901)
K5—S7i 3.518(4) 86viii—K2——S4vii 9927(10) 84—K5—S7xiii 120.6501)
K5—S8Xiii 3.542(4) Slvi—K2—S6 172.8803) s3Vi—K5—S7xiii 7373(8)
K5—S6Viii 3.552(4) Slvii—K2—86 9821(11) 85—K5——ssi 7979(10)
K5—sziii 3.805(4) S6viii—K2—S6 8360(5) S4—K5—S8i 6097(10)
S4Vii—K2—S6 8494(9) s3Vi—K5—S8i 1 18.980 1)
Pl—Sl 2.020(4) Slvi—Kz—ss 122.3501) 87xiii—K5—ssi 1145101)
Pl—S2 2.080(4) Slvii—Kz—ss 134.4901) ss—K5—87i 87.60(9)
P1—S3 2.032(4) S6vii'L—x2—ss 6298(8) S4—K5—S7i 118.1501)
Pl—S4 2.018(4) S4Vii—K2—SS 138.2(1) s3Vi—K5—S7i 6327(8)
P2—85 2.068(4) S6—K2—S5 5701(7) S7Xiii—K5—S7i 8773(9)
P2—S6 2.007(4) Slvi—Kz—szvi 5711(9) S8i—K5—S7i 5723(8)
P2—s7 2.023(4) 81Vii—K2—szvi 6441(9) ss—K5—ssxiii 143.9400)
P2—S8 2.054(4) S6Viii—K2—S2Vi 129.1001) S4——1<5—88xiii 6471(9)
84Vii—K2—szvi 118.29(8) s3vi—K5—ssxiii 130.1800)
S6—K2—S2Vi 129.700 1) s7xiii—K5—ssxiii 57.10(8)
S5—x2—szvi 100.55(8) ssi—K5—ssxiii 78.8(1)
s lvi—Kz—s7 132.950 1) S7i—K5_ssxiii 104.4400)
SIVii—KZ—S7 7933(9) ss—x5—86viii 6390(8)
S6viii—K2—S7 116.76(9) S4—K5—S6viii 8586(10)
S4Vii—K2—S7 1 17.3500) s3Vi—K5—S6viii 85.3800)
S6—K2—S7 5347(8) S7xiii—K5—S6viii 108.3400)
S5—K2—S7 55.160) ssi—Ks—S6viii 134.9501)
szvi—xz—s7 7648(8) S7i—K5—S6viii 1395600)
P2—K2—S7 3143(6) ssxiii—K5—86viii 1 1553(10)
plvi—K2—87 101.97(8) S4—Pl—S1 111.9(2)
S4V—K3—s7x 131.9400) S4—Pl—S3 109.0406)
S4V—K3—ssi 6785(9) s l—Pl—S3 1 1024(19)
S7X—K3—SSi 160.190 1) S4—Pl—S2 107.4206)
84V—K3—82i 134.6401) Sl—Pl—S2 108.6006)
S7x—x3—szi 8914(9) S3—Pl—S2 109.5605)
ssi—x3—82i 7214(9) S6—P2—S7 109.3407)
S4V—K3—Slxi 93.4601) S6—P2—S8 113.7608)
s7x—K3—81xi 85.6502) S7—P2—S8 107.1607)
ssi—-K3—s1xi 9298(12) S6—P2—SS 110.7006)
82i—1<3—s1xi 6788(8) S7—P2—S5 1 1 175(16)
S4V—K3—S3V 5984(8) S8—P2—S5 104.0307)
S7x—K3—S3V 7278(8)
(i) -x, -y, 2; (ii) 0.5-x, -0.5+y, 2-2; (iii) x, -l+y, 1+2; (iv) x, y, 1+1;
(v) -x, -l-y, 2; (vi) x, y, -1+z; (vii) 0.5-x, 0.5+y, l-z; (viii) 0.5-x, -0.5+y, 1-z;
(ix) x, 1+y, -1+z; (x) -x, -y, 1+2; (xi) -0.5+x, -0.5-y, 2-z; (xii) -x, -y, -1+z;

(xiii) x, -l+y, z; (xiv) 0.5-x, 0.5+y, 2-z; (xv) 0.5+x, -0.5-y, 2-z; (xvi) x, 1+y, z.

134

Structural Dimensionality

To place the stoichiometry and the dimensionality of the compounds in a greater
context, it is useful to consider each of the known [PS4]3' containing compounds as a
member of the homologous series (K3PS4)n(BiPS4)m, Scheme 1.31434'36 In this
formalism, the compound BiPS4 is a member of the series with an n:m ratio of 0. BiPS4

has a dense three-dimensional ﬁamework. As the nzm ratio increases the dimensionality

of the ﬁamework is reduced. Thus, the Bi/P/S ﬁ'amework in K1.5Bi2.5(PS4)3 [II] at an

n:m ratio of 0.20 is also three-dimensional but the framework architecture is relatively
open with tunnels accommodating K ions. When the n:m ratio is increased to 0.33, an

even more open three-dimensional ﬁamework compound K3Bi3(PS4)3 [I] is stabilized.
A ftu'ther increase of the n:m ratio to 1 gives the one-dimensional chains of K3Bi(PS4)2,
and the largest nzm ratio of 3 gives the molecular K9Bi(PS4)4 [III]. As the nzm ratio is
increased, the three-dimensional covalent BiPS4 framework is dismantled by dilution
with ionic K3PS4 that lowers dimensionality of the resultant compound. Dimensional
reduction of a covalent framework can be a practical formalism in studying structural
evolution in solids and has been discussed earlier as a means to controlling the structures
and properties of materials.52,53 For example, in the incorporation of KZS into HgS
amounts to the introduction of 82" atoms in the diamond lattice of HgS and the
generation of anionic low-dimensional [ngSy]"' frameworks. In essence, the dense
packed lattice of the parent compound is “diluted” with 32- atoms which progressively
dismantle it. The K+ acts as a “non-interfering” counterion. A chemical homology
created in this way is of the type (K28)m(HgS),,. 54 Other such examples in

chalcogenides include (A2Q),,,(CdQ),,55 and (A2Q),,,(Bi2Q3),,26t56 systems.

135

 

.1.
BiPS4 5 A3PS4 ; §K1.5Biz.5(PS4)3

 

 

 

 

lA PS .

3 3 ‘ 2 91631808.).

A P

3 S‘ - K319410802

“3'39“ = KgBi(PS4)4
Schemel

Raman Spectroscopy

Raman spectra were collected for each of the title compounds, Figure 5-10. In

each case, a peak at ~410 cm'1 was observed, which is attributed to the [PS4]3'

symmetrical stretch-‘37 Also, each spectrum showed a band of overlapping peaks ﬁ'om

~150 cm'1 to ~300 cm'1 that showed similarities to the Raman spectrum of Bi2S3

suggesting that those peaks can be attributed to various Bi—S vibrational modes. A small

amount highly crystalline material Bi2S3 may dominate the Raman spectrum even

though it is present in small quantities.

Compound [II] also showed a third intense region of features with peaks centered
at 519, 538, and 562 cm'l, Figure 5-10B. Despite the presence of some of compound
[II] in the sample used to collect a Raman spectrum of compound [I], this region of

features was not evident for the NMR spectrum of K3Bi3(PS4)4, Figure 5-10A. At

present, the exact assignment of these modes that appear to be unique to compound [II]
is not complete, but it is possible that the metal-sulfur bonds would give rise to features
in this region. Perhaps the distortion of the site to accommodate either K or Bi gives rise
to the proper coordination geometry to yield Raman active modes. This region would be

much weaker in the compounds that do not experience modulation, as supported by the

136

Raman spectra of compounds [I] and [HI], Figure 5-10. Given that the intensity ratios of
these peaks are dependent on the composition of the sample, it is unlikely that the higher

energy peaks are overtones of those at lower energy.

238

(A)

263
189

('0
P
V

Intensity (arb. units)

 

 

700 600 500 400 300 200 100
Wavenumber (cm'1)

O

r

(B) 3 8

538
235

189
166

 

Intensity (arb. units)

 

 

700 600 500 400 300 200 100
Wavenumber (cm'1)

a
(C) 8.

a e.

. (O h

s N 22

E a

g V

C

.9

E

 

 

700 600 500 400 300 200 100
Wavenumber (cm'l)

Figure 5-10. The Raman spectra Of (A) K3Bi3(PS4)4, (B) K1_5BiZ.5(PS4)3 and (C) K931033404. The
K3Bi3(PS4)4 sample used to collect the spectrum also contained several crystalline impurities (see text).

137

Solid State NMR

NMR spectroscopy has been a useful tool in studying modulated compounds and
provides structural information that is complementary with the crystallographic data.
The NMR line shapes and spin-lattice relaxation times of modulated compounds are
known to be strongly dependent on the nature of the modulation vector q.58 The spectral
line-widths of modulated compounds tend to be broader than those of related
compounds without modulation because the loss of translational periodicity gives rise to
an essentially inﬁnite number of different chemical environments. Therefore, the
shielding parameter has a wide range of values, and the spectral line is broadened.
Because of enhanced phonon scattering processes, spin-lattice relaxation in
incommensurately modulated compounds is more efficient and longitudinal relaxation
times tend to be shorter in such compounds.58

A summary of the chemical shifts, peak widths, intensities, and T1 relaxation

times for compounds [II] and [III] are found in Table 5-9 and the spectra of [II] and [III]
are displayed in Figure 5-11. The NMR spectra of K1_5Bi2.5(PS4)3 [11] indicated two
isotropic peaks centered at 88.4 ppm and 71.8 ppm with peak widths of 1150 and 890

Hz, respectively, and a longitudinal relaxation time T1 of 600000) s, Figure 5-11A. It is

expected that compounds with signiﬁcant contributions from incommensurate structural
modulations have NMR line widths that are broadened and longitudinal relaxation times
that are shortened relative to phases that are not modulated.58 The NMR data on
compound [11] indicated only a slight increase in line width but no signiﬁcant decrease
in the longitudinal relaxation time relative to the isoelemental compound [HI]. This is
consistent with the crystallographic evidence that suggests the effect of the modulation

on compound [II] is relatively small.

138

Table 5-9. 31F chemical shift (CS), chemical shift anisotropy (CSA), and T) values.

 

 

a ‘ t C
Compound CS Peak Widths In egratedb T1
(ppm) (Hz)a Intensrtles (S)
K1.513i25(PS4)3[II] 88.4, 71.8 1150, 890 45.5, 100.0 (22030)),
K913i(PS4)4 [III] 79.7 640 100 640020)

 

a Uncertainties in CS are approximately :h 0.5 ppm, and uncertainties in line widths are approximately 1
10 Hz. The chemical shift was determined by a mathematical deconvolution of the peaks after ﬁtting with
a Lorentzian lineshape, and peak widths were determined by calculating the full width at half maximum
(FWHM).

b Intensities were normalized with the most intense peak set to 100.
c Uncertainties are indicated in parenthesis.

For compound [II], the upﬁeld peak at 71.8 ppm had roughly double the
integrated intensity of the downﬁeld peak at 88.4 ppm. The crystallographic analysis
suggested three unique P atoms, and the NMR intensity data could be explained by an
overlap of the peaks arising from two of the P atoms. P(lb) and P(lc) are located within
the double chain and have six metal ion neighbors two bonds away. P(la) bridges
neighboring double chains and has ﬁve metal ion neighbors two bonds away. The
upﬁeld peak can be attributed to P(lb) and P(lc) which are related by symmetry, and the
downﬁeld peak can be attributed to P(la).

The NMR spectrum of K9Bi(PS4)4 [III] contained two isotropic peaks centered

at 86.1 ppm and 79.7 ppm, with 2:5 ratio of signal intensity, line widths of 300 and 640
Hz, and longitudinal relaxation times of l350:t125 s and 6402020 s, respectively, Figure
5-1 13. There were two crystallographically inequivalent P sites in [111] that could have
been assigned to the two peaks observed in the NMR. However, deviations from a 1:1
intensity ratio and different line widths and relaxation times for the two peaks do not
support this assignment. A more plausible explanation is that the 64.3 ppm line is due to

an impurity that may have been formed by the partial decomposition of compound [III]

139

by moisture. The narrower line width is consistent with a new mobile species formed
by a reaction of [H1] with atmospheric water. Signals from the two crystallographically
inequivalent P sites in compound [HI] were not resolved and were instead overlapped.

The presence of impurity phases in batches of compound K3Bi3(PS4)4 [I] was
evident in the X-ray powder difﬁaction diagrams. These phases are thiophosphate

species as indicated by the large number of peaks in the 31P NMR spectrum. These

peaks were not assigned.

140

 

 

 

 

(A)
w W
200 100 ppm 0 -100
(B) *

 

 

 

* '1:
1'
200 100 ppm 0 -100

Figure 5.11. 31? NMR spectra of (A) K1_5B12_5(PS4)3 and (B) KgBi(PS4)4 with isotropic peaks and
their spinning sidebands grouped under like symbols.
CONCLUDING REMARKS

It is evident ﬁom this work that the group 15 alkali chalcophosphate class of
compounds contains a rich and diverse collection of species with wide variability in
bonding motifs.19'25 The present report adds three new compounds to this class,

including an example of a phase with an incommensurate structural modulation.

141

Decreased framework dimensionality is favored in compounds with a larger proportion

of K3PS4. It is interesting to note that there are only several isomorphic compounds in

common between the K/Bi/P/S and the previously reported Tl/Bi/P/S classes, namely

MBiP2S7, and M3Bi(PS4)2 (M = K, Tl). None of the new compounds described here
have a known T1+ analog, and even though the K3Bi3(PS4)4 and T13Bi3(PS4)4 have the

same stoichiometry, the structures are quite diﬂ‘erent.33 Attempts to form mixed K/Tl
solid solutions containing K/Tl/Bi/P/S may produce novel pseudo-quaternary materials
with different structures. Finally, ﬂexibility in the Bi coordination in this system is
notable, as well as the chemical conditions required to stabilize each compound. The
ﬂux conditions previously studied-30 did not identify the title compounds. Thus, the
diversity in the group 15 alkali metal chalcophosphates is even greater than was

previously considered, and now even includes an example of a modulated compound.

142

REFERENCES

1. Galdamez, A.; Manriquez, V.; Kasaneva, J .; Avila, R. E. Mater. Res. Bull. 2003, 38,
1063-1072.

2. Odoulov, S. G.; Shumelyuk, A. N.; Brost, G. A.; Magde, K. M. Appl. Phys. Lett.
1996, 69, 3665-3667.

3. Shumelyuk, A.; Hryhorashchuk, A.; Odoulov, S. Phys. Rev. A 2005, 72, 6.

4. Bourdon, X.; Maisonneuve, V.; Cajipe, V. B.; Payen, C.; Fischer, J. E. J. Alloys
Compd. 1999, 283, 122-127.

5. Carpentier, C. D.; Nitsche, R. Mater. Res. Bull. 1974, 9, 1097-1100.

6. Rogach, E. D.; Sviridov, E. V.; Arnautova, E. A.; Savchenko, E. A.; Protsenko, N. P.
Zh. T ekh. Fiz. 1991, 61, 201-204.

7. Kroupa, J.; Tyagur, Y. I.; Grabar, A. A.; Vysochanskii, Y. M. Ferroelectr. 1999, 223,
421-428.

8. Misuryaev, T. V.; Murzina, T. V.; Aktsipetrov, O. A.; Sherstyuk, N. E.; Cajipe, V.
B.; Bourdon, X. Solid State Commun. 2000, 115, 605-608.

9. Knaust, J. M.; Dorhout, P. K J. Chem. Ctystallogr. 2006, 36, 217-223.

10. Chung, 1.; Do, J.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem.
2004, 43, 2762-2764.

11. Kanatzidis, M. G. Curr. Opin. Solid. St. M. 1997, 2, 139-149.

12. Coste, S.; Hanko, J.; Bujoli-Doeuff, M.; Louarn, G.; Evain, M.; Brec, R.; Alonso,
B.; Jobic, S.; Kanatzidis, M. G. J. Solid State Chem. 2003, 175, 133-145.

13. Manriquez, V.; Galdamez, A.; Ruiz-Leon, D. Mater. Res. Bull. 2006, 41, 1337-
1344.

14. Belkyal, 1.; El Azhari, M.; Wu, Y. D.; Bensch, W.; Hesse, K. F.; Depmeier, W.
Solid State Sci. 2006, 8, 59-63.

15. Gieck, C.; Tremel, W. Chem-Eur. J. 2002, 8, 2980-2987.

16. Coste, S.; Kopnin, B.; Evain, M.; Jobic, S.; Brec, R.; Chondroudis, K; Kanatzidis,
M. G. Solid State Sci. 2002, 4, 709-716.

17. Hess, R. F.; Gordon, P. L.; Tait, C. D.; Abney, K D.; Dorhout, P. K J. Am. Chem.
Soc. 2002, 124, 1327-1333.

143

18. Coste, S.; Kopnin, E.; Evain, M.; Jobic, S.; Payen, C.; Brec, R. J. Solid State Chem.
2001, 162, 195-203.

19. Kanatzidis, M. G. Chem. Mot. 1990, 2, 353-363.

20. Kanatzidis, M. G.; Park, Y. J. Am. Chem. Soc. 1989, 111, 3767-3769.

21. Kanatzidis, M. G.; Park, Y. Chem. Mat. 1990, 2, 99-101.

22. Park, Y. B.; Kanatzidis, M. G. Angew. Chem. Int. Ed. 1990, 29, 914-915.

23. McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 1257-1267.

24. Chondroudis, K; Kanatzidis, M. G. J. Solid State Chem. 1998, 138, 321-328.

25. Iordanidis, L.; Bilc, D.; Mahanti, S. D.; Kanatzidis, M. G. Journal of the American
Chemical Society 2003, 125, 13741-13752.

26. McCarthy, T. J.; Ngeyi, S. P.; Liao, J. H.; Degroot, D. C.; Hogan, T.; Kannewurf, C.
R; Kanatzidis, M. G. Chem. Mat. 1993, 5, 331-340.

27. Iordanidis, L.; Brazis, P. W.; Kyratsi, T.; Ireland, J.; Lane, M.; Kannewurf C. R;
Chen, W.; Dyck, J. S.; Uher, C.; Ghelani, N. A.; Hogan, T.; Kanatzidis, M. G. Chem.
Mat. 2001, 13, 622-633.

28. Iordanidis, L.; Schindler, J. L.; Kannewurf, C. R; Kanatzidis, M. G. J. Solid State
Chem. 1999, 143, 151-162.

29. Kohatsu, 1.; Wuensch, B. J. Acta. Crystallogr. B 1976, 32, 2401-2409.
30. McCarthy, T.; Kanatzidis, M. G. J. Alloys Compd. 1996, 236, 70-85.

31. Manriquez, V.; Galdamez, A.; Leon, D. R.; Garland, M. T.; Jimenez, M. Z. Krist.
New Cryst. St. 2003, 218, 151-152.

32. Chung, 1.; Karst, A. L.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2006, 45,
2785-2787.

33. Gave, M.; Malliakas, C. D.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007,
46, 3632-3644.

34. Zimmermann, H.; Carpentier, C. D.; Nitsche, R. Acta. Crystallogr. B 1975, 31,
2003-2006.

35. Chondroudis, K; McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1996, 35, 840-
844.

144

36. Chondroudis, K; Kanatzidis, M. G.; Sayettat, J.; Jobic, S.; Brec, R. Inorg. Chem.
1997, 36, 5859-5868.

37. Sayettat, J.; Bull, L. M.; Jobic, S.; Gabriel, J. C. P.; Forumigue, M.; Batail, P.; Brec,
R; Inglebert, R. L.; Sourisseau, C. J. Mater. Chem. 1999, 9, 143-153.

38. Fehér, F. Handbuch der Priparativen Anorganischen Chemie; Ferdinand Enke
Verlag: Stuttgart, Germany, 1954; Vol. 1, pp 280-281.

39. Chondroudis, K; Kanatzidis, M. G. J. Solid State Chem. 1998, 136, 328-332.

40. SMART, SAINT, SHEIXT L V-5 and SADABS, Bruker Analytical X-ray Instruments,
Inc.: Madison, 1998.

41. X-AREA, X-SHAPE, and X-RED, STOE & Cie GmbH: Darmstadt, 2004.

42. Petricek, V.; Dusek, M.; Palatinus, L. Jana2000. The Crystallographic Computing
System, Praha, Czech Republic, 2000.

43. Evain, M.; Boucher, F .; Gourdon, O.; Petricek, V.; Dusek, M.; Bezdicka, P. Chem.
Mat. 1998, 10, 3068-3076.

44. Gourdon, O.; Hanko, J.; Boucher, F.; Petricek, V.; Whangbo, M. H.; Kanatzidis, M.
G.; Evain, M. Inorg. Chem. 2000, 39, 1398-1409.

45. International Tables of Crystallography Vol. C; Kluwer Academic Publishers:
Dordrecht, Netherlands, 2004.

46. Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

47. Chondroudis, K; Kanatzidis, M. G. Inorg. Chem. Commun. 1998, 1, 55-57.
48. Nagy, T. F.; Mahanti, S. D.; Dye, J. L. Zeolites 1997, 19, 57-64.

49. Stoltzfus, M. W.; Woodward, P. M.; Seshadri, R.; Klepeis, J. H.; Bursten, B. Inorg.
Chem. 2007, 46, 3839-3850.

50. Gauthier, G.; Evain, M.; Jobic, S.; Brec, R. Solid State Sci. 2002, 4, 1361-1366.

51. Gauthier, G.; Jobic, S.; Danaire, V.; Brec, R.; Evain, M. Acta Crystallogr. C 2000,
56, 117.

52. Tulsky, E. G.; Long, J. R. Chem. Mat. 2001, 13, 1149-1166.

53. Long, J. R.; McCarty, L. S.; Holrn, R. H. J. Am. Chem. Soc. 1996, 118, 4603-4616.

145

54. Axtell, E. A.; Park, Y.; Chondroudis, K; Kanatzidis, M. G. J. Am. Chem. Soc. 1998,
120, 124-136.

55. Axtell, E. A.; Liao, J. H.; Pikramenou, Z.; Kanatzidis, M. G. Chem. -Eur. J. 1996, 2,
656-666.

56. Kanatzidis, M. G.; McCarthy, T. J.; Tanzer, T. A.; Chen, L. H.; Iordanidis, L.;

Hogan, T.; Kannewurf, C. R.; Uher, C.; Chen, B. X. Chem. Mat. 1996, 8, 1465-
1474.

57. Evenson, C. R.; Dorhout, P. K. Inorg. Chem. 2001, 40, 2884-2891.

58. Blinc, R. Phys. Rep. 1981, 79, 331-398.

146

CHAPTER 6

MIXED METAL ARGYRODITES: PHASE TRANSITIONS BY CHEMICAL
COMPOSITION.

147

INTRODUCTION

The class of compounds known as the argyrodites are tetrahedrally close-packed
structures with the formula Ami}12,n_x)/mB“+X2'6_xY'x (A = Cu, Ag, Cd, Hg; B = Ga, Si,
Ge, Sn, P, As; X = S, Se; Y = Cl, Br, I and 0 S x S 1).1 The mineral argyrodite,
AggGeSG, is a naturally occurring representative of this class of compounds, and for
historical reasons, the name also refers to the entire class. Most examples of argyrodites
have a crystalline structure at higher temperatures with the spacegroup F -43m and
signiﬁcantly disordered cation sites. At reduced temperatures, various lower symmetry
structures have been reported wherein the cations become more ordered

All of the argyrodite structures can be considered as modiﬁcations to the F—43m

structure of CugPSSBr.l In this structure, the Cu+ ions are distributed randomly over a
number of sites with local tetrahedral symmetry, Figure 6-1A. The F-43m symmetry is
therefore somewhat misleading and only results because of the positional averaging that
is inherent to X-ray crystallography. An anionic framework is formed by corner shared
[PS4]3' tetrahedra, Figure 6-lB. The remaining S2' and Br' ions are randomly mixed in
an icosahedral site surrounded by [PS4]3' tetrahedra, Figure 6-1C.

The parent Cu6PS5Br structure is modiﬁed in a number of different ways to form
each of the other members of the argyrodite class of compounds. For example, the
structure of y-Ag7PSe6 was determined at 473 K,2 and also contains [PSe4]3' tetrahedra,
but only half as many per unit cell as Cu6PS5Br. The other half of the sites are occupied
by Se2' ions. The metal cations, as in the case of Cu6PS5Br are disordered over a number

of tetrahedral sites, but there is no atom in the icosahedral site. At lower temperature

148

(293 K), F-43m structure is modiﬁed, and the resultant compound B-Ag7PSe6 is formed.
This compound has the P213 space group, and differs ﬁom the F-43m structure in the
arrangement of the Ag+ ions. Rather than being disordered over a large number of
tetrahedral sites, the Ag+ ions preferentially occupy a smaller number of sites with
slightly distorted linear, triangular, and tetrahedral metal-selenium coordination}4 The
decrease in symmetry as the temperature is lowered is perhaps misleading and is actually
the result of the averaging of atomic positions that occurs on the relatively long

timescales that diffraction data are collected.

 

Figure 6-1. The room temperature structure of Cu6PSSBr in the F—43m spacegroup showing (A) the
disordered sites containing Cu+ ions, (B) The anionic framework of corner shared [PS4]3‘ tetrahedra, and

(C) the icosahedral coordination of the mixed S/Br site. Hollow spheres are Cu, light grey spheres are S,
and the central atom in (C) is the mixed S/Br site.

149

Another modiﬁcation to the parent F -43m structure occurs in the compound
Cd4GeS6, Figure 6-2.5 This compound differs principally due to lowering of symmetry
ﬁom cubic to monoclinic. This change in symmetry accommodates a distortion in the
packing of the GeS4 tetrahedra which occupy half of the tetrahedral sites, as is the case
with the P213 structure, Figure 6-2. The packing of the 82' ions in the other half of the

sites is also distorted due to the monoclinic symmetry. Another signiﬁcant difference is
that the Cdz' sites are ordered, rather than the disorder or partial disorder that is present in

the cubic argyrodite modiﬁcations.

 

 

 

 

 

 

Figure 6-2. The structure of Cd4GeS6 down the b-axis showing the monoclinic cell. Hollow spheres are
Cd, shaded spheres are S, and black spheres are P.

The elucidation of argyrodite structures are notoriously difﬁcult because of
signiﬁcant twinning that occurs during the phase transition. For example, the structure of

CugGeS6 was solved using a crystallographic model containing a l2-fold twin.6 For the

compound Ag7TaS6, a cubic F-43m structure was observed at 1123 K, and a monoclinic

150

structure was observed at 873 K These structures were solved using the same crystal for
data collection but the high temperature structure was reﬁned as a single crystal, and the
lower temperature structure was reﬁned as a 24-fold twin.7 The twinning might
reasonably occur during the synthesis of these materials since the transition temperatures
are near the temperatures used for the reaction. Signiﬁcant twinning is therefore a
ﬁequent obstacle in the solution of argyrodite structures.

The novel argyrodite compounds presented herein were initially serendipitously
discovered in attempts to synthesize analogs to the previously reported compounds
CuBiPZSe6 and AgBiP2Se68 except with Hg in place of B1. The motivation for this
substitution was to look for novel optical or structural properties that would result from
having vacancies in the ﬁamework as a result of the substitution for Bi3+/Hg2+ i.e. that
1/3 of these metal sites would be vacant in the latter. Isotypic compounds with site
vacancies were not produced, and instead compounds with argyrodite structures were
formed.

This class of compounds has potential commercial relevance. Argyrodite

compounds like Ag7PSe6 have been shown to have signiﬁcant ionic conductivity2 and

therefore may be useful in production of selectively permeable coatings, and in next-
generation metal ion batteries. Additionally, the robustness of the argyrodie ﬁamework
allows for a wide range of chemical compositions with closely related structures, and
therefore tuning physical properties by adjusting the identities and stoichiometric ratios
of the constituent elements may be possible.

The new argyrodite compounds with the possible stoichiometries of AgSHgPSe6,

Ag3Hg2PSe6, Ag3Hg2PSe6, AgHg3PSe6, and Cqu3PSe6 are presented here as well as

151

the as-yet unidentiﬁed novel crystalline compound produced in a reaction targeting

H83.51’366-

EXPERIMENTAL SECTION
Reagents

Chemicals were used as obtained unless otherwise noted: mercury (triple distilled,
Bethlehem Apparatus Co. Inc., Hellertown, PA), selenium shot (99.999%, Tellurex, Inc,
Traverse City, MI), phosphorus (amorphous red, MCB Reagents, Glbbstown, NJ), copper
(electrolytic dust, Fisher Scientiﬁc, Pittsburgh, PA) and silver powder, prepared as below.

Selenium was ground in an agate mortar and pestle to ~100 mesh.

Synthesis

Preparation of Silver Powder: Silver powder was prepared by cutting a silver
coin (31g) into centimeter sized pieces, which were subsequently dissolved in 200 mL of
concentrated nitric acid by stirring overnight. The solution was then neutralized by
addition of ammonium hydroxide. Finely divided silver particles were obtained after
addition of ~10g of sodium borohydride to the solution which were isolated by ﬁltration,
washed with water, and dried in a 100 °C oven.

Ag7,ﬁgx/2PSe6.' Mixtures of Ag, Bi, P, and Se were combined in the appropriate
stoichiometric ratios to form the nominal compositions of Ag7PSe6 AgngPSe6,

Ag3Hg2PSe6, AgHg3PSe6, and Hg3.5PSe6 with a total reactant mass of ~2g. The

resultant mixtures were loaded into fused silica tubes and ﬂame sealed at a reduced

pressure of ~10'4 mbar. The tubes were shaken to increase homogeneity, and then loaded

152

into a protective ceramic sheath. The mixtures were heated to 800 °C over 24h, held
there for 24h, and then cooled back to room temperature over 48h. Shiny black
polyhedral crystals were obtained, some as large as 0.5 mm, Figure 6-3A.

Cqu3PSe6: A mixture of Cu (0.0816 g, 1.28 mmol), Hg (0.2923 g, 1.45 mmol),

P (0.0756 g, 2.44 mmol), and Se (0.5925 g, 7.50 mmol) was loaded into a fused silica
ampoule and prepared for the furnace identically as the Ag7_ngx/2PSe6 samples. The
mixture was then heated to 700 °C over 24 h, held there for 24 h and then cooled over 96
h. Shiny black polyhedral crystals were obtained and EDS analysis on the crystal used
for X-ray difﬁaction had the stoichiometry Culo3Hg2.8P1.0Se4.3. A scanning electron
micrograph of some of the crystals isolated from the reaction is displayed in Figure 6-

3A. One interesting but presently unexplained feature was the ~111m surface detail,

Figure 6-3B.

153

 

Figure 6—3. Scanning electron micrographs of the crystals formed in the reaction that produced Cqu3PSe6

with views of the ~2 um surface detail observed on one of the of the crystals.

154

Powder X-ray Diffraction

Powder X-ray difﬁaction was performed with a Rigaku-Denki RW400F2
(Rotaﬂex) diffractometer equipped with a Cu K01 rotating anode operating at 45 kV and
100 mA with a 2°/min scan rate. The resultant powder patterns were indexed using the
Jade 6.0 software package.9 The indexing was performed by overlaying the powder
pattern of the compound that was known to be an isomorphic analog, then reﬁning the

cell parameters to graft the overlayed peaks onto the experimental powder pattern.

Electron Microscopy

Scanning electron micrographs and energy dispersive spectroscopy (EDS) were
collected using a JEOL ISM-35C electron microscope equipped with a Tracor Northern
energy dispersive spectroscopy detector. EDS data were collected using an accelerating

voltage of 25 kV and an accumulation time of 60 s.

155

Differential Thermal Analysis

Differential thermal analyses were performed with a Shimadzu DTA-50 thermal
analyzer. A portion of the ingot obtained ﬁom the synthetic reaction weighing ~25 mg
was ﬁnely ground and sealed in a fused silica ampoule under a reduced pressure of 10'4
mbar. An equivalent mass of alumina was sealed in an identical ampoule to serve as a
reference. The samples were heated to 900 °C at a rate of 10 °C/min, cooled to 150 °C at
a rate of 10 °C/min. The cycle was then repeated, except then cooled to room

temperature.

Single Crystal X-ray Diffraction of Cqu3PSe6

A single crystal of Cqu3PSe6 with no obvious twinning was selected and X-ray
difﬁaction data was subsequently collected on a Bruker SMART platform CCD
difﬁ'actometer using Mo K01 radiation and operating at 40 kV and 40 mA. Individual
frames were collected with a 10 sec exposure time and a 03° omega rotation. The
SMART software was used for data collection and the SAINT software was used for data
extraction and reduction. An analytical absorption correction to the data was performed
using the SADABS software and direct methods were used to solve and reﬁne the
structure with the SHELXTL software package. 10

The details of the reﬁnement are displayed in Table 6-1. The unit cell of the
crystal was monoclinic with a = 12.515(3) A, b = 7.2328(14) A, c = 12.446(3) A and [3 =
109.36(3)°, and the structure was reﬁned with the Cc space group. The structure closely

resembles Cd4Ge86 with an isomorphic replacement with Cu and Hg for Cd, P for Ge,

and Se for S.

156

The reﬁnement suggested that the atoms Cu(l) and Hg0) had complementary
occupancy but were not located exactly on the same site. Each of the other Cu/Hg mixed
sites, however, were reasonably constrained to occupy the same crystallographic site. In
each case, the site occupancy reﬁned to ~75% Hg and 25% Cu.

One potentially signiﬁcant problem was with the relatively high residual electron
density of 7.185 e'/A3 that was located 1 A away ﬁ'om the Cu(l) atom, Table 6-2. This
residual electron density may be attributable to a third site in addition to the Cu(l) and
Hg(1) atoms that were already accounted for, but this possibility was not explored. The
isotropic thermal displacement parameter U(eq) for several atoms, namely P(l), Se(l),
Se(2), Se(3) and Se(4) were unreasonably small Another potential problem was
observed in the anisotropic reﬁnement of the thermal parameters wherein several of the

tensors were zero, Table 6-3. This may be evidence for an unaccounted twin ﬁ'agment.

157

Table 6-1. Crystallographic data and reﬁnement details for CU3HgPSe6.

 

 

 

 

Empirical formula 01 H83 P Se6
Formula weight 1170.04
Temperature 293(2) K
Wavelength 0.71073 A
Crystal system Monoclinic
Space group Cc
Unit cell dimensions a = 12.515(3) A 01= 90°
b = 7.2328(14) A B= 109.36(3)°
c = 12.446(3) A y = 90°
Volume 1062.9(4) A3
Z 4
Density (calculated) 7.31 l g/cm3
Absorption coeﬂicient 65.799 mm'1
F(000) 1952
Theta range for data collection 3.30 to 29.31°
-17<=h<=l6, -9<=k<=9,
Index ranges -l7<=l<=16
Reﬂections collected 6003
Independent reﬂections 2564 [R(int) = 0-0706]
Completeness to theta = 29.3 1° 98.6 %
Full-matrix least-squares on
Reﬁnement method F 2
Data/restraints/ parameters 2564 / 2/ 105
Goodness-of-ﬁt on F2 1.049
Final R indices [I>o(1)] R1 = 0.0693, wR2 = 0.1552
R indices (all data) R1 = 0.0883, wR2 = 0.1622
Extinction coefﬁcient 0.0004300)
ngm diff. peak and hole 7.185 and -5.l36 e.A'3
Table 6-2. Fractional atomic coordinates and U(eq) values for Cqu3PSe6 with standard uncertainties in
parenthesis.
atom x y z U(eq)a Occupancy
Hg(l) 0.0163(2) 0.0828(3) 0.1250(2) 17(1) 75%
0.10) 0015307) 0.0350(30) 0.162506) 17(1) 25%
Hg(2)|Cu(2) 0.0354(1) 0.06471(3) 0.0665(1) 22(1) 75%125%
Hg(3)|Cu(3) 0.2777(1) 0.1886(3) 0.2582(2) 24(1) 75%I25%
Hg(4)|Cu(4) 0.3635(3) 0.9430(4) 0.6306(2) 49(1) 75%I25%
P(l) -0.2046(6) 0.2623(11) -0.1304(6) 3(1) 100%
Se(l) 0.2261(2) 0.9924(5) 0.4260(2) 4(1) 100%
Se(2) -0.0213(2) 0.2442(4) -0.0805(2) 4(1) 100%
Se(3) 0.2218(2) 0.2639(4) 0.1836(2) 3(1) 100%
Se(4) -0.2588(2) 0.0144(4) -0.0537(2) 3(1) 100%
Se(5) -0.0894(3) 0.2738(6) 0.2364(3) 20(1) 100%
Se(6) 0.0250(4) -0.2665(6) 0.1228(4) 28(1) 100%

 

a U(eq) = (ZinUijai*aj*ai'aj)/3 x 1000

158

Table 6-3. Anisotropic thermal displacement parameters for Cqu3PSe6 with standard uncertainties in
parenthesis.

 

 

atom U11 U22 U33 U23 U13 U12
Hg(l) 18(1) 11(1) 15(1) -4(1) -6(1) 11(1)
Cu(l) 18(1) 11(1) 15(1) .40) -6(1) 11(1)
Hg(2) 9(1) 50(1) 6(1) 1(1) 2(1) -6(1)
Cu(2) 9(1) 50(1) 6(1) 1(1) 2(1) -6(1)
Hg(3) 7(1) 19(1) 43(1) 24(1) 2(1) 2(1)
Cu(3) 7(1) 19(1) 43(1) 24(1) 2(1) 2(1)

Hg(4) 78(2) 37(2) 39(1) -1(1) 30(1) -4(1)
Cu(4) 78(2) 37(2) 39(1) -10) 30(1) 4(1)
P(l) 7(3) 2(4) 2(3) 0(2) 5(2) 1(3)
86(1) 5(1) 2(2) 5(1) 0(1) 2(1) -30)
86(2) 0(1) 4(1) 6(1) 1(1) 0(1) 0(1)
Se(3) 4(1) 5(2) 0(1) -20) -10) -10)
Se(4) 5(1) 3(2) 0(1) 2(1) 1(1) 0(1)

 

Se(S) 13(1) 30(2) 17(2) -8(2) 5(1) -3(2)
Se(6) 29(2) 23(2) 31(2) -7(2) 7(2) .20)
RESULTS AND DISCUSSION

Powder X-ray Diffraction

Ag7PSC6
The reported unit cell parameter for the solved crystal structure of B-Ag7PSe6 is

10.772(2) A1 1 which is outside of the error of the value for the indexed powder pattern of
the present work of 10.783(2), Table 6-4. Indexing of all peaks was successﬁrl assuming
the P213 space group that was determined previously. Presumably, this error is caused
by a miscalibration of the diffractometer and the actual uncertainties in unit cell
parameters are at least 0.007 A. There was no evidence for other crystalline phases in the

powder difﬁaction pattern, Figure 6-4.

159

u I It I A“ l‘ I “Hg3,5PSe6”

 

 

 

 

| A 1 A A “AgHggPSe6”
L;

4.1....2 A .2
, ﬂ _ _.__:.‘._L:' AM AV?“ 1 “A E3Hg2PSE6”
J Agni # “Ag-511813836”
10 20 30 40 50 60
Angle (20)

Figure 6-4. Powder X-ray diffraction of the Argyrodite materials. The Ag7PSe6 sample has been

previously described. The quotation marks indicate the nominal stoichiometry of the product because a
stoichiometric ratio that was reﬁned ﬁom a crystallographic solution was unavailable.

Table 6-4. Unit cell parameters determined from the indexed powder X-ray diffraction patterns.

 

 

S ace

Compound a (A) b (A) c (A) a (deg) 0 (deg) )7 (deg) Ggoup

AE7PSC6 10.783(2) 10.783(2) 10.783(2) 90 90 90 P213

A85H8P366 10.67402) 10.67402) 10.67402) 90 90 90 F-43m

A83H82PSC6 10.54107) 10.54107) 10.54107) 90 90 90 F-43m
A8H83PSe6 12.73102) 7.35400) 12.7320) 90 109.45(5) 90 Ce
H83.5PS°6 12.643(30) 7.232(20) 1259307) 90 1101100) 90 Cc

 

“Ag5HgPSe6 ” and “Ag3Hg2PSe6 ”

The sample with the nominal stoichiometry “Ag5HgPSe6” displayed a powder

pattern that was closely related to y-Ag7PSe6. The reported cell parameter of y-Ag7PSe6

is 10.838(3) A,2 and the cell parameter of “Ag5HgPSe6” was determined to be

10.674( 12). In the former, presumably the change in cell parameter is the result of a

positive thermal expansion as the crystallographic data were collected at elevated

160

temperature to affect the phase transition. In the Hg-containing compound, however, the
change in cell parameter is likely the result of the addition of Hg to the structure. The
effective ionic radii of the Ag“ and Hg2+ are 0.115 and 0.102 A, respectively,12 which
consistent with the decrease in cell parameter with increased substitution of the latter.
The powder X-ray diffraction pattern of the sample with the nominal

stoichiometry “Ag3Hg2PSe6” was also successfully indexed assuming a cubic unit cell
and an F-43m space group. The unit cell was smaller than both the “Ag5HgPSe6” and
Ag7PSe6 cells at 10.54107) A, which is consistent with ﬁrrther substitution of the

smaller Hg2+ ions, Table 6-4.

“AgHg3PSe6 ”
The powder pattern for “AgHg3PSe6” did was not consistent with the cubic unit

cells of the other argyrodite materials, but instead was indexed using a monoclinic cell
with a = 12.731(22) A, b = 7.354(10) A, c = 12.732(9) A, 0 = 109.45(5)°, and using the
Cc space group, Table 6-4. The known argyrodite compound Cd4GeS6 also adopts the
Cc space group with a similarly sized unit cell, namely a = 12.326(3) A, b = 7.088(2) A, c
= 12.365(3) A and [3 110.08(2)°. This suggests that perhaps there is an isomorphic
replacement of S for Se, P for Ge, and the Ag and Hg for Cd to produce the novel

argyrodite material.

‘6Hg3.5PSe699

The X-ray powder pattern for the sample with the nominal stoichiometry of

“Hg3_5PSe6” did not have a powder pattern that matched any known structure. Though

161

the reaction stoichiometry might reasonably have produced some ngPZSe6, difﬁ’action
peaks corresponding to this crystalline compound were not observed. One possible
interpretation of the powder pattern is that a number of the peaks correspond to a
structure that is similar to the Cc symmetry argyrodites but with a smaller unit cell, and
the remaining peaks correspond to a second phase. This interpretation yielded a unit cell
with a = l2.643(30) A, b = 7.232(20) A, c = 12.593(27) A and B = 110.11(20)°. No
successful index of the remaining peaks was obtained, perhaps because the peaks ﬁ'om

the second phase overlap with those of the indexed phase.

Cqu3PSe6 Structural Information by Single Crystal X—ray Diﬂraction
In Cd4GeS6, all of the Cd sites are tetrahedral but in the structurally related
Cqu3PSe6 compound where the Cd atoms are replaced by both Cu and Hg, there are

some signiﬁcant deviations from the parent structure. The structural reﬁnement provided
some insight about this deviation.

The single crystal X-ray reﬁnement of Cqu3PSe6 contained a Cu(l) and Hg(l)
mixed site where the Cu+ and Hg2+ ions were not overlapped but the occupancies of the
sites appeared to be complementary. The positions of the atoms were reﬁned to be
0.78(2) A apart, Table 6-5. This observation may provide some evidence for the Cu+ and

Hg2+ ions to preferentially adopt different coordination environments, and support the

observation that Cu-Se and Hg-Se bond lengths tend to be different. Previously, Cu-Se

distances of 2.392 A were reported in the compound CuBiPZSeg,8 and Hg—Se distances of

2.634 A were reported in the compound HgSe. 13 The average bond lengths of the Cu(l)-

Se atoms in Cqu3PSe6 in this bent coordination environment were 2.310 A, Table 6-5.

162

The Hg(l) ion is coordinated in a tetrahedral geometry with an average Hg(l)-Se bond
length of 2.650 A, Figure 6-5, Table 6-5. None of the other mixed metal sites displayed

any evidence for separation of the crystallographic sites.

 

Figure 6-5. The coordination environments of (A) the Cu( 1) in a bent coordination geometry and (B)
Hg(l) in a distorted tetrahedral geometry in the structure of Cqu3PSe6. The positions of the Hg(l) and
Cu(l) atoms were reﬁned to be 0.78(2) A apart.

The mixed Cu(2)/Hg(2) site was trigonal planar with an average metal-Se bond
length of 2.562 A, the Cu(3)/Hg(3) site was also trigonal planar with an average metal-Se
bond length of 2.583 A, and the average metal-se bond length was 2.493 A, Figure 6-6,
Table 6-5. Each of these average bond lengths is intermediate to the typical Cu-Se and
Hg-Se bond lengths, but only slightly shorter than the Hg-Se bond lengths. This

correlates with the greater Hg content (i.e. each site contains 75% Hg, and only 25% Cu).

163

Table 6-5. Selected bond lengths and angles for Cqu3PSe6 with bond lengths (A) and angles (°), with
standard uncertainties in parenthesis.

 

Hgl—Cul 079(2) Se6—Hgl—Se5 124.4708)
Hgl—Se6 2.530(5) Se6—Hgl—Se2 1 1452(15)
Hgl—SeS 2.608(4) Se5—Hg1—Se2 108.3003)
Hgl—Se2 2.708(4) Se6—Hgl—Se3 115.9604)
Hgl—Se3 2.759(4) Se5—Hg1—Se3 101.5802)
Cul—Se5 230(2) Se2—Hgl—Se3 8427(1 1)
Cul—Se6 233(2) Se5—Cul—Se6 159.4(12)
Hg2|Cu2—Se6i 2.482(5) Se6i—Hg2|Cuz—Sesii 131.4206)
Hg2|cttz—Se5ii 2.499(4) Se6i—1ig2|Cu2—8e4iii 1 1248(13)
Hg2|Cuz—Se4iii 2.704(3) Sesii—nglcttz—Se4iii 1 14.3602)
Hg2|Cu2—Se2 2.991(4) Sc6i—Hg2|Cu2—Se2 102.3904)
Hg3|Cu3—8c6iv 2.507(5) Sesii—Hg2|Cu2—Se2 96.1203)
Hg3|Cu3—Se5 2.535(4) Se4iii—Hg2ICu2—Se2 8199(10)
Hg3|Cu3—Sc4v 2.707(3)
Hg3|Cu3—8e1vi 3.021(4) Se6iV—Hg3ICu3—Se5 129.8107)
Hg4|(:u4—Sesvii 2.401(5) 8e6iV—Hg3ICu3—Se4v 1 15.8504)
Hg4|Cu4—Se6viii 2.419(6) Se5—Hg3lCu3—Se4v 113.9502)
11g4|Cu4—Se3ix 2.568(4) Se6iV—Hg3|Cu3—Selvi 9903(13)
Hg4|Cu4—Sel 2.582(4) Se5—11g3ICu3—Se1vi 9545(12)
Se4V—Hg3 |Cu3—Sc1vi 79.610 1)
P1—Se2 2.172(7)
Pl—Selx 2.190(8) SeSVii—Hg4|Cu4—Se6Viii 114.6(2)
Pl—Se3xi 2.198(7) Sesvii—Hg4lctt4—Se3ix 114.8805)
Pl—Se4 2.239(8) Se6"iii—Hg4|ctt4—Se3ix 11 1.1207)
Sesvii—Hg41Cu4—Se1 112.6506)
Se6viii—Hg4lctr4—Sel 1090006)
Se3ix—Hg4ICu4—Sel 9248(15)
Se2—Pl—Selx 1 168(4)
Se2—Pl—Se3xi 109.3(3)
Selx—Pl—Se3xi 109.6(3)
Se2—P1—Se4 105.7(3)
Selx—Pl—Se4 107.7(3)
8e3xi—P1—Se4 107.4(3)

 

(i) x, 1+y, 2; (ii) x, l-y, -0.5+z; (iii) 0.5+x, 0.5+y, 2; (iv) -0.5+x, 0.5+y, 2; (v) x, -y, 0.5+z; (vi) -0.5+x, -
0.5+y, 2; (vii) 0.5+x, 1.5-y, 0.5+z; (viii) 0.5+x, 0.5-y, 0.5+z; (ix) x, l-y, 0.5+z; (x) -0.5+x, 1.5-y, -0.5+z;
(xi) -0.5+x, 0.5-y, -0.5+z; (xii) x, -l+y, z; (xiii) 0.5+x, -0.5+y, z; (xiv) x, -y, -0.5+z.

In the isomorphic Cd4GeS6 compound, all of the Cd sites are tetrahedrally
coordinated by S atoms, but in Cqu3PSe6, three of the sites have a coordination number

of less than four. This may indicate that some longer range ordering of the Cu and Hg

164

atoms was present but was missed by the reﬁnement. One possible explanation is that the
~2.6 A metal-Se distance is dominated by the large Hg:Cu ratio. Since the Cu-Se bond
length is typically shorter (i.e. ~2.4 A) the Cu atom moves off center in the tetrahedral
cage and adopts a coordination environment with a lower coordination number. For this
model to be correct, presumably each metal site would have to be a mixture of two

crystallographic sites that do not spatially overlap.

 

Figure 6-6. The coordination environments of (A) the trigonal planar Cu(2)/Hg(2) site, (B) the trigonal
planar Cu(3)/Hg(3) site (C) the distorted tetrahedral Cu(4)/Hg(4) site in the compound Cqu3PSe6.

It is likely that a similar study aimed at producing the other members of the class

of compounds Cu7_ngx/2PSe6 would have the opposite trend in size with an increase in
Hg content that was observed for the Ag7_ngx/2PSe6 compounds. A typical Ag-Se bond
length is ~2.8 A as was observed in the compound AgBiPZSe6,8 which is longer than the

typical Hg-Se bond lengths of ~2.6 A.l3 Conversely, the Cu-Se bond lengths are

approximately 2.4 A, so an increase in Cu content should produce a smaller unit cell.

165

Differential Thermal Analysis and Optical Absorption of Cqu3PSe6
The differential thermal analysis of Cqu3PSe6 displayed an endothermic feature

indicative of the onset of melting upon heating at 740 °C. On the ﬁrst cooling cycle a
sharp exothermic event was observed at 645 °C but on the second cycle, an exothermic
event was observed at 675 °C. Presumably, these features correspond to crystallization of
the melt. Interestingly, the line shape represented a physical situation that is impossible,
inset of Figure 6-7A. One interpretation is that crystallization is sufﬁciently exothermic
that it raises the ambient temperature of the ﬁrrnace. This provides evidence for the ready
formation of argyrodite materials since crystallization presumably allows a
thermodynamically very stable product to be formed, and therefore competing crystalline
compounds can be avoided. Powder X-ray difﬁaction patterns collected before thermal
treatment, after a single heating cycle, and after subsequent heating cycles did not have
signiﬁcant differences, and the shift in the crystallization temperature has not been

successﬁrlly rationalized.

 

 

 

 

 

 

 

 

 

 

(A) 120- . (B)
loo-i 4 ﬂ
80- l 8
£60. A J] I] g
t5 40 600 ' 640 ' 680 ' .8
a l j <2:
20- 3;
0.1
-204
0 20 40 60 80 100 0.5 1.0 1.5 2.0 2.5
Temperature (°C) eV

Figure 6-7. (A) The differential thermal analysis and (B) optical absorption spectrum of Cqu3PSe6.

166

The optical absorption spectrum of Cqu3PSe6 contained a strong absorption

edge at 0.9 eV, which correlates with the black color of the polyhedral crystals, Figure 6-
7B. There was also a minor feature that could be an absorption edge at ~1.3 eV which

could presumably be assigned to an as yet unidentiﬁed crystalline phase.

CONCLUDING REMARKS
The new argyrodite materials presented herein are in interesting case where the
cationic site has been populated with multiple metals. There is some preliminary

evidence ﬁom the single crystal reﬁnement of Cqu3PSe6 that some site preference for

the monovalent cation (i.e. Cu+) versus the trivalent cation (i.e. Hg3+) is present, but
problems with the crystallographic reﬁnement make this observation questionable. The

reﬁnement of each of the cationic positions in the unit cell of the Ag7,ngx/2PSe6

compounds was unsuccessﬁrl using the available powder diffraction data, and no
satisfying reﬁnement could be obtained perhaps due to limitations in the accuracy of the
diﬁractometer. One additional possibility that could not be explored with the presently
available data is that atoms with different sizes and formal charges do not randomly
substitute but instead demonstrate longer range order.14 In the ﬁrture, studies using
synchrotron radiation or neutron diffraction may be more suited to such a powder
reﬁnement, and the behavior of the cation sites may be ﬁnther elucidated. More intense
sources such as these may make the elucidation of longer range order possible.
Additionally, some of the known argyrodite materials have temperature-dependant phase
transitions, and variable-temperature difﬁ’action experiments may indicate that the novel

materials presented here have similar phase transitions.

167

REFERENCES
l. Kuhs, W. F.; Nitsche, R; Scheunemann, K. Mater. Res. Bull. 1979, 14, 241-248.

2. Evain, M.; Gaudin, B.; Boucher, F.; Petricek, V.; Taulelle, F. Acta. Crystallogr. B
1998, 54, 376-383.

3. Gaudin, B.; Boucher, F.; Petricek, V.; Taullele, F .; Evain, M. Acta. Crystallogr. B
2000, 56, 402-408.

4. Gaudin, E.; Petricek, V.; Boucher, F.; Taulelle, F .; Evain, M. Acta. Crystallogr. B
2000, 56, 972-979.

5. Julienpouzol, M.; Jaulmes, S. Acta Crystallogr. C 1995, 51, 1966-1968.
6. Onoda, M.; Chen, X. A.; Ishii, M.; Wada, H. Solid State Ionics 2000, 136, 415-418.
7. Onoda, M.; Wada, H.; Sato, A.; Ishii, M. J. Alloys Compd. 2004, 383, 113-117.

8. Gave, M.; Bilc, D.; Mahanti, S.; Breshears, J .; Kanatzidis, M. Inorg. Chem. 2005,
5293-5303.

9. Jade 6. 0, Materials Data, Inc. Livermore, CA.: 2003.

10. SMART, SAINT, SHEIXTZ V-5 and SADABS, Bruker Analytical X-ray Instruments,
Inc.: Madison, 1998.

11. Evain, M.; Boucher, F.; Gourdon, O.; Petricek, V.; Dusek, M.; Bezdicka, P. Chem.
Mot. 1998, 10, 3068-3076.

12. Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements; 2nd ed.; Butterworth-
Heinemann: Boston, MA, 1997; pp 1176, 1205.

13. Barley, J. W. Am. Mineral. 1950, 35, 337-364.

14. Gave, M.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46, 11063-11074.

168

CHAPTER 7

Cs4PZSe10: A NEW COMPOUND DISCOVERED WITH THE APPLICATION OF
SOLID STATE AND HIGH TEMPERATURE NMR.

169

INTRODUCTION
Chalcophosphates are compounds with oxidized phosphorus and at least one P-Q
bond, where Q = S, Se, or Te. These compounds exhibit an impressively rich structural

diversity because of the large number of stable [Pny 2‘ building blocks that can be

stabilized and the variety of binding modes in which they can engage.1 They are
important in the areas of catalysis, ceramics, glasses, and molecular sieves}4
Previously, solid state nuclear magnetic resonance (NMR) has been useful in providing
structural information about chalcophosphate materials.5'10 Most notably, in
selenophosphate materials there is a general correlation between the chemical shifts
(CSs) of anions and the presence or absence of a P-P bond. Compounds containing a PP
bond resonate between 25 and 95 ppm whereas those that do not contain a P-P bond
resonate between -115 and 5 ppm.11 Additionally, the spectral intensity of a peak in a
NMR spectrum is proportional to the number of nuclei that have the corresponding
chemical environment. 12 Therefore, NMR spectroscopy is a useful tool to evaluate phase
purity in cases where competing compounds have the same NMR-active nucleus.

There is little experimental data about the species that exist at high temperature
although it has been proposed that different [Pny 2' anions exist in the melt in various
equilibria.l Since there is 31P chemical shift discrimination between P-P and non P-P
bonded selenophosphate species in room temperature solids, in-situ 31P NMR studies of
high temperature syntheses can potentially provide information about the identity of 31P-
containing species that exist at high temperature. The known compound Cs4P2Se913 was

initially chosen for a high temperature NMR study because: (1) the synthesis temperature

could be reached by our NMR probe; (2) the compound contained the interesting Se-

170

bridged [P2Se9]4' anion; and (3) the compound melts congruently which suggests that

spectra of subsequent high temperature remelts should be similar to one another and that
some correlation can be made between the remelt spectra and the spectrum of the solid
compound.

The new compound Cs4PZSe10 which was discovered following high temperature
in-situ NMR synthesis intended to produce Cs4PZSe9 is reported here. Room temperature

solid state 31P NMR spectra of the synthetic products clearly showed that Cs4P2Se9 was
not the dominant crystalline product and the structure of Cs4P2Se10 was subsequently

solved with X-ray diffraction of single crystals obtained from the NMR synthesis.
Synthetic conditions were varied to ascertain the important factors leading to dominant

Cs4P28e9 or Cs4P2Se10 products and high temperature 31P NMR spectra of synthetic

melts were obtained and analyzed.

EXPERIMENTAL SECTION
Reagents

Chemicals were used as obtained unless otherwise noted: phosphorous (MCB
Reagents, Gibbstown, NJ, amorphous red), selenium shot (Tellurex Inc., Traverse City,

MI, 99.999%) cesium metal, (Aldrich Chemical Co., Inc., St. Louis, MO). Phosphorous

was freeze dried and CSZSe was prepared by a modiﬁed literature preparation.14t15
P2Se5 was prepared by combining a stoichiometric amount of the elements in a pyrex

ampoule, ﬂame-sealing at a reduced pressure of 100 mTorr, then heating to 300 °C over 6
h, holding there for 6 h, and then quenching in air. Manipulations of all starting materials

were performed under a nitrogen atmosphere in a dry box (Vacuum Atmospheres).

171

Synthesis

The synthesis that produced Cs4P2Se10 as the dominant P-containing product was
performed by combining CSZSe (0.186 g, 0.53 mmol), PZSeS (0.124 g, 0.27 mmol), and
Se (0.192 g, 2.4 mmol) which were subsequently ground together in an agate mortar and
pestle to increase homogeneity. This resulted in a reaction stoichiometry with the
nominal composition “Cs4PZSe16” or an excess of six equivalents of Se relative to the

Cs4PZSe10 stoichiometry. The extra Se in the reaction stoichiometry was added to favor

the production of Cs4PZSe10 rather than Cs4P2Se9.13 Approximately 75 mg of this

mixture was loaded into a ﬁlsed silica ampoule, sealed at ~100 mTorr, and placed in the
high temperature NMR probe. The sample was then heated at a rate of 10 °C/min, with
pauses in the temperature ramp to collect isothermal NMR spectra. After reaching the
maximum temperature of 600 °C and collecting an NMR spectrum, the sample was
cooled back to room temperature at a rate of 10 °C/min, again pausing for spectral
collection. The total time spent above room temperature was ~6 h. A room temperature
31P NMR spectrum following the in situ synthesis contained a dominant isotropic peak at

—52.8 ppm and integration of all the spectral peaks indicated that Cs4P2Se10 was >99% of

the total P-containing compounds, Figure 7-1A. Powder X-ray diffraction indicated that
elemental Se also remained in the ﬁnal products. It was also possible to prepare

Cs4P2Se10 with a conventional high temperature fumace using the same stoichiometry
and a heating/cooling proﬁle of a similar length. Crystals of Cs4P2Se10 were air and

moisture stable over a period of at least several hours.

172

(A) (B)

 

 

* * + +
MAM-i“ _A_'JL____..~..jo\ JLAArL—m
O O
100 50 0 .60 -100 -100 -200 100 50 0 0'0 -100 400 -200
* ppm
(C) (D)

 

 

":3

 

 

 

0
+1: 1 +*
W
100 50 0 00 -100 -1'50 -200 100 50 0 0'0 .100 4'50 -200
ppm

 

 

Figure 7-1. Ambient temperature 31P magic angle spinning NMR spectra of the ﬁnal products of: (A) a
“CS4P28e16” reaction stoichiometry indicating CS4P23e10 as the dominant 31P-containing species at -
52.8 ppm; (B) a “CS4P2Se8” reaction stoichiometry indicating CS4P28e9 as the dominant 31P-containing
species at -39.9 ppm; and (C) a “CS4P2Se12” reaction stoichiometry that produced CS4P2Se9 and

C84P2Se10 in a ~3:5 ratio. Panel D displays the static 31P NMR spectrum of a 350 0C melt of the products

in C. The peak with a chemical shift of -22.0 ppm in panels B and C was not assigned and the peak at -99.8
ppm in panel B was not assigned. For spectra A-C, the spinning ﬁ'equency was 14 kHz and spinning
sidebands are grouped with the corresponding isotropic peak using like symbols with * a CS4P2Se10, + a
CS4P2Se9, and o a unassigned peak with -22.0 ppm isotropic shift

Cs4P2Se9 was produced in high purity by combining CsZSe (0.281g, 0.82 mmol),
P28e5 (0.186 g, 0.41 mmol), and Se (0.032 g, 0.41 mmol) in a fused silica ampoule. This
resulted in a “Cs4P2Se8” reaction stoichiometry and under these Se starved conditions,

Cs4P2Se9 is favored. The ampoule was subsequently sealed under a reduced pressure of

173

~100 mTorr and allowed to react in a furnace. The sample was heated to 700 °C over 12
h, held at 700 0C for 12 h, and then cooled to room temperature over 12 h. The room
temperature 31P NMR spectrum had a dominant isotropic peak at -39.9 ppm and
integration of all the spectral peaks indicated that Cs4P2Se9 was >90% of the 31P-
containing compounds, Figure 7-1B. There were also peaks at -22.0 ppm and -99.8 ppm
that accounted for the remainder of the spectral intensity but were not assigned. The
powder X-ray difﬁaction pattern contained only peaks that corresponded to Cs4P2Se9.
Vitriﬁcation of the mixture with “Cs4P2Se16” stoichiometry was accomplished by
sealing a sample portion in a fused silica ampoule under a reduced pressure of ~100

mTorr, ﬂame heating until molten, and then rapidly quenching in ice water. Complete

vitriﬁcation was conﬁrmed by powder X-ray diffraction.

Single Crystal X-ray Diffraction

Intensity data for single crystals of Cs4P2Se10 were collected on a Bruker
SMART platform CCD diffractometer using Mo K01 radiation operating at 40 kV and 40
mA, Table 7-1. Fractional atomic coordinates are displayed in Table 7-2 and selected
bond lengths and angles are displayed in Table 7-3. A full sphere of data was collected
and individual ﬁames were acquired with a 10 8 exposure time and a 03° (2 rotation. The
SMART software was used for data collection, and SAINT software was used for data
extraction and reduction. An analytical absorption correction to the data was performed
and direct methods were used to solve and reﬁne the structures with the SHELXTLl6

software package.

174

Powder X-ray Diffraction

Powder X-ray difﬁ'action was performed with a Rigaku-Denki RW400F 2

(Rotaﬂex) difﬁactometer equipped with a Cu Ka rotating anode operating at 45 kV and

100 mA with a 2°/min scan rate.

Table 7-1. Crystallographic data for CS4P2Se10

 

Formula
FW

Space group
a (A)

b (A)

c (A)

9 (°)

13 (°)

7 (°)

v (A3)

Z

Cs4P2SelO
1383.18

P-l
7.3587(11)
7.4546(11)
10.142005)
85.938(2)
88.055(2)
85.609(2)

553.1004)
1

dcalc (99913)

p (mm-1)
7. (A)

Temp (K)
20 max (°)
Ra (l>2o(1))

Ra (all)

4.153

23.113
0.71073

298

28.30

R1 = 0.0395
wR2 = 0.0716
R1 = 0.0687
wR2 = 0.0789

 

a R = 2(IFOI - chI)/2IF0|. Rw = [2w(|1=0| - |Fc|)2/Ew|Fo|2] “2

Table 7-2. Fractional atomic corrdinates for CS4P2$810 with standard deviations in parenthesis.

 

 

atom x y 2

CSI 0.2210(1) 0.2754(1) 0.9686(1)
Cs2 0.1943(1) 0.1431(1) 1.3603(1)
Sel 0.2682(1) 0.7719(1) 0.9946(1)
Se2 0.3081(1) 0.0372(1) 0.6737(1)
Se3 0.0229(1) 0.6635(1) 0.7046(1)
Se4 0.5116(1) 0.5897(1) 0.7599(1)
Se5 0.4017(1) 0.6057(1) 1.4616(1)
P 0.2533(3) 07%» 0.7789(2)

 

Table 7-3. Selected bond lengths A and angles (°) for CS4P2Se6 with standard uncertainties in parenthesis.

 

Csl—Se2
Csl—Sell
Csl—Se4ll

Cal—Seliii

cs1——Se3i
Csl—Sel
Cal—Selii
Csl—Se4
Csl—Se3

3.5999(12) Se2—Csl—Seli 9840(2) Se5—Cs2—Seliii
3.648702) Sez—Csl—So4ii 136.97(3) Se2Vi—Cs2——Se1iii
3.676901) Seli—Csl—Se4ii 121.940) SeziV—csz—Sesvii
3.7348(12) Se2—Cs1—-Se1iii 6058(2) Se2V—Cs2—Se5vii
3.7543(12) Seli—csl—Seliii 8552(2) Ses—cs2-—Se5vii
3.7718(12) Se4ii——Cs1-—Se1iii 105.050) Sezvi—Csz—Sesvii
3.7762(12) Se2—Csl—Se3i 151.36(3) Seliii—Csz—Sesvii
3.7815(11) Seli——Cs1—Se3i 5945(2) SeziV—Csz—Se3i
4.0230(12) Se4ii—Cs1—Se3i 6258(2) Se2V—C82—Se3i

175

9131(2)
8220(2)
6472(2)
150.790)
3699(3)
5292(2)
106.06(2)
8189(2)
5686(2)

Table 7-3 continued

 

Seliii—Csl—Se3i 9729(2) Se5—Cs2—Se3i 174050)
Cs2—Se2iv 3.683502) SeZ—Csl—Sel 125.330) SeZVi—CSZ—Se3i 9824(3)
Cs2—Se2" 3.7433(12) Sell—Csl—Sel 104420) Seliii—Cs2—8c3i 94.650)
Cs2—Se5 3.7522(12) Se4ii—Csl—Sel 6204(2) Sesvii—Csz—Se3i 140010)
Cs2—Se2vi 3.8038(13) Seliii—Csl—Sel l66.51(3) SeziV—Cs2—Se3viii 5676(2)
Cs2—Seliii 3.8210(12) Se3i—Csl—Sel 8051(2) Se2V—Cs2—Se3viii 8029(2)
Cs2—8e5vii 3.8251(12) Se2—Cs1—Se1ii 8558(2) Ses—cs2—8e3viii 6582(2)
Cs2—Sc3i 3.8256(12) Seli—Cs1—8e1ii 164.68(3) Sezvi—Csz—Se3viii 117.210)
Cs2—8e3viii 3.8891(12) Se4ii—Cs1—Se1ii 5156(2) Seliii—Csz—st-3viii 145.710)
Cs2—864vi 4.0678(12) Seliii—cttl—Selii 8359(2) Sesvii—Cs2—8e3viii 7146(2)
Se3i—Csl—Selii 111.370) Se3i—Csz—Se3viii 108.890)
Sel—P 2.190(2) Sel—Csl—Selii 8484(2) SeziV—Csz—Se4vi 118.71(3)
Sez—Piii 2.159(2) Se2—Csl—Se4 7782(2) Se2V—Cs2—Se4Vi 126.19(3)
Se3—P 2.160(2) Seli—Cs1—8e4 134200) 8e5—Cs2—Se4vi 3410(2)
Se4—P 2.312(2) Se4ii—Cs1—Se4 83.330) Sezvi—Csz—So4vi 7212(2)
Seliii—Csl—Se4 127.310) Selii'L—csz—Se4vi 58.150)
Se4—Se5Vi 2.3125(13) Se3i—Csl—Se4 130.320) Sesvii—Csz—So4vi 5402(2)
Ses—Scsvii 2.4045(19) Sel—Csl—Se4 5085(2) Se3i—Cs2—So4vi 151.64(3)
Selii—Csl—Se4 6107(2) Se3viii—Cs2-——Se4vi 9908(2)
Se2—Csl—Se3 8170(3) Se4"i—Se5—Sc5vii 9915(6)
Seli—Cs1—8e3 7851(2) Sezx—P—Se3 113.1100)
Se4ii—Cs1—Se3 118.03(3) SeZX—P—Sel 116.6700)
Seliii—Csl—Se3 136.25(3) Se3—P—Sel 115.1700)
Se3i—Cs1—Se3 108.520) Sezx—P—Se4 107.7300)
Sel—Csl—Se3 5610(2) Se3—P—Se4 1094900)
Seli'L—Csl—Ses 116.770) Sel—P—Se4 9214(9)
Se4—Csl—Se3 5571(2)
ScziV—Csz—SezV 105.690) Se3—P—Cs1 7149(6)
SeziV—Csz—Ses 9266(3) Sel—P—Csl 6465(6)
Se2V—Cs2—Se5 123040) Se4—P—Csl 6451(6)
ScziV—Csz—Sczvi 7348(3) Se2x—P—Cslx 5799(6)
Se2V—C82—Se2Vi 154.340) Se3—P—Cslx 124.72(8)
Se5—Cs2—Sezvi 82.400) Sel—P—Cslx 6153(6)
Se2iV—Cs2-——Seliii 154.590) Se4—P—Cslx 125.46(8)
Se2V—Cs2—Se1iii 9300(2) Csl—P—Cslx 125.30(6)

 

(i) -x, l-y, 2-2; (ii) I-X. l-y. M: (iii) x, -l+y, 2; (W) x, y, 1+2; (V) -X. -y. 2-2; (vi) l-X. -y, 2-2; (vii) l-X. -l-
y, 3-2; (viii) x, -l+y. 1+2; (iX) -x, -y, 3-z; (X) x, 1+y, 2; (xi) x, y, -l+z; (xii) X. 1+y, -1+z.

NMR Spectroscopy

Room temperature 31P NMR spectra were collected on a 9.4 T spectrometer

(V arian Inﬁnity Plus) using a magic angle spinning (MAS) probe (V arian Chemagnetics)

176

and served as an analytical tool to assess the phase purity of P-containing compounds.
Samples were spun at 7 to 14 kHz in zirconia rotors of 4 mm outer diameter and a ~50 uL
sample volume. Bloch decay spectra were taken with a 4.6 [.18 11/2 pulse calibrated using

H3PO4 which also served as a chemical shift reference at 0 ppm. Four scans were
summed with a 5000 s recycle delay. The longitudinal relaxation time of Cs4PZSe10 at

ambient temperature was determined by ﬁtting the signal intensity as a ﬁrnction of
relaxation delay to an exponential buildup function.5

High temperature 31P NMR spectra were collected on a 9.4 T spectrometer
(Varian Inﬁnity Plus) using a static probe (Doty Scientiﬁc) with a 5.5 ps 7t/2 pulse
cahbrated at room temperature using H3PO4 which also served as a chemical shift
reference at 0 ppm. Several hundred to several thousand scans were averaged with
recycle delays ranging from 0.1 s to 10 s. Sample temperatures ranging ﬁom 250 °C to
600 °C were reached at a rate of 10 °C/min and were allowed to equilibrate for ~5 min
before data collection.

All NMR spectra were processed with line broadening and baseline correction.

Differential Thermal Analysis (DTA)
Differential thermal analysis (DTA) was performed on a Shimadzu DTA-50
thermal analyzer. Sample portions (~30 mg) were sealed under a reduced pressure of ~80

mTorr in fused silica, and an equivalent mass of A1203 in a similarly prepared ampoule

served as a reference. The sample was heated to 600 °C at a rate of 10 °C/minute and then
cooled at a rate of 10 °C /minute back to room temperature. The residues after DTA were

analyzed by powder X-ray diffraction.

177

RESULTS AND DISCUSSION

Identiﬁcation and Structure Description of Cs4P2Se10

The initial goal of synthesizing Cs4PZSe9 in a sealed tube in the NMR
spectrometer was to obtain high temperature 31P NMR spectra and to identify species
present in the synthetic melt. After one of the syntheses was complete, the room
temperature 31P NMR spectrum contained a dominant peak at -52.8 ppm in addition to
the -39.9 ppm shift of CS4P2869.“ The -52.8 ppm peak could not be assigned to any
known compound. Single deep reddish brown crystals were isolated, screened using X-
ray diffraction, and the structure of Cs4PZSe10 was subsequently solved. A summary of
the details of the reﬁnement are displayed in Table 7-1. All of the peaks in the powder X-
ray diffraction pattern of the sample prepared using the optimized synthetic method
corresponded either to the calculated Cs4P2Se10 pattern or to elemental Se. The 31P NMR

peak at -52.8 ppm was therefore assigned to Cs4P2Se10.

Cs4P2Se10 has the molecular anion [P2Se10]4‘ which is composed of two PSe4

tetrahedra that are connected by two central selenium atoms thus forming a tetraselenide

fragment, Figure 7-2A. The PSe4 tetrahedron is distorted with the Se-P-Se angles ranging

from 92.14(9)° to 116.67(10)°. The lengths of the P-Se bonds are typical and range from
2.159(2) to 2.312(2) A. Tables 7-2 and 7-3 contain the atomic and thermal parameters,

and selected bond lengths and angles, respectively.

178

 

 

Figure 7-2. Structure of Cs41>28ew (A) down the a-axis showing the [PzSe10]4' chains and (B) along the
b-axis with the weak interaction between the [P2Se10]4' chains emphasized with a dashed line.

The Se-Se contacts range from 2312503) to 2.404509) A, a large range of bond
lengths. This unusual behavior could be explained by the dihedral angles17 around the
bonds Se(4)-Se(5) and Se(5)-Se(5) of -83.78(1)° and 180.00(1)°, respectively. It is
noteworthy that the Se(5)-Se(5) bond is shared by two planar structural moities namely
Se(4)-Se(5)-Se(5)-Se(4) and Se(2)-'-Se(5)-Se(5)-"Se(2) to give maximized lone pair
repulsion in Se atoms and one of the longest Se-Se distances reported for a
selenophosphate ligand. Comparable or even longer Se-Se distances have been found in

the Senz' (n = 2, 3) subunits in the KZSeZ, Rb2S62,18 Sr4Sn2Se9, Sr4SnZSelol9 and

179

Eu3(Sn4Se14)(Se3)2 .20 A bond distance alteration in Q42‘ (Q = S, Se) has been reported
in CS4P2S10,21 APSe6 (A = K, Rb, Cs),22 and unligated Se42' ions.23

For Cs4P2Se10, unusually short intra- and intermolecular Se---Se nonbonding
interactions are observed at 3.243(4) and 3.399(4) A. Intermolecular nonbonding
interactions generate a pseudo-one dimensional inﬁnite chain including eight-membered
rings composed of P and Se atoms along the b-axis, Figure 7-2B.

The structure of Cs4PZSe10 is related to the structure of Cs4P2S1021 and both
contain the [P2Q10]4' anion. In CS4P2S10, there are two different conformations of the
[P2810]4' anion with dihedral angles about the central disulﬁde bonds of 89.60 and 0". In
CS4P2Se10, only one anionic conformation is present with a dihedral angle about the

central diselenide bond of 180° (vide supra). Additionally, there is positional disorder of

the S atoms in CS4P2S10, and this is not evident for the Se atoms in CS4P2Se10.

Differential Thermal Analysis (DTA)

For the glass with “Cs4P2Se16” stoichiometry, the ﬁrst DTA heating cycle
indicated an exothermic crystallization event centered at ~160 °C and broad endothermic
melting features centered at 315 °C, Figure 7-3. The subsequent cooling cycle contained
exothermic crystallization features that were centered near 275 °C. Subsequent DTA
cycles were similar to the ﬁrst cycle except that they did not contain the initial
crystallization peak near 160 °C. Possible assignments of the DTA features are: (1) 160

0C exothermic feature, glass to crystal phase transition of Cs4P2Se10 and Se; (2) 315 0C

180

endothermic feature/275 OC exothermic features, melting/crystallization of Cs4P2Se10
and Se.

The assignment of the 160 0C feature is supported by several arguments. First, the
feature is exothermic and is only observed during the ﬁrst heating cycle as would be
expected for a transition ﬁom a metastable glass to a thermodynamically stable crystal.
Second, annealing the glass near the putative glass to crystal transition temperature
resulted in formation of a mixture of crystalline materials that were predominately

CS4PZSe10 and Se. 31P solid state NMR and powder X-ray diffraction were the methods

used to analyze the annealed material. The glass to crystal transition temperature of the

CS4P25616 glass is lower than the corresponding temperatures that have been observed

for glasses that form related metal selenophosphate compounds.22424425 Materials with a
reversible glass to crystal phase transition have potential use in data storage
applications,22426'30 and ﬁlrther characterization of the phase change properties of

CS4P23610 may be of technological relevance.

181

0. 161°C

 

 

0 100 200 300 400 500 600
Temperature (°C)

Figure 7- 3. Differential thermal analysis (DTA) of a mixture with “CS4P2Se16” stoichiometry that was
quenched rapidly (tom the melt The displayed plot is for the ﬁrst heating/cooling cycle of this mixture.

The interpretation of this plot relied on the powder X-ray diﬂ‘raction pattern obtained following several
heating/cooling cycles. This pattern indicated that CS4P2Se10 and elemental Se were the major crystalline

products. In the displayed plot, the peak centered at ~l 60 °C is assigned to the glass to crystalline transition
of CS4P2Se10 and Se, the peak centered at at ~315 °C is assigned to the melting of this mixture, and the

peak centered at 277 °C is assigned to the subsequent recrystallization. The exothermic peak centered at
~160 °C was not observed for the second and subsequent heating/cooling cycles likely because the cooling
rate was slow enough that crystalline rather than glassy material was formed. Thermal events due to pure

Se were not observed perhaps because of insuﬂicient Se or because of comelting/cocrystallization of Se
With CS4P23e10.

182

Competition between CS4P2Se9 and Cs4PZSe10

Analysis of the products of furnace reactions done with different stoichiometries
showed that reaction stoichiometry had a signiﬁcant effect on the CS4P28e10:Cs4P2Se9
ﬁnal product ratio. Reduction by one equivalent of Se relative to Cs4P2Se9, i.e.
“CS4P2Se8” reaction stoichiometry, produced CS4P2Se9 as the dominant 31P-containing
product with no CS4P23610, Figure 7-lB. Syntheses with increased Se relative to
CS4P28610, i.e. “Cs4P28e10+x” with l < x < 5 reaction stoichiometries, resulted in
mixtures of CS4P2569 and CS4P23610. A representative spectrum of the products of the
reaction where x = 2 is displayed in Figure 7-1C. For the “Cs4PZSe16” reaction
stoichiometry, Cs4P28e10 was the only P-containing compound, Figure 7-1A. The Figure
7- 1B,C spectra also had small peaks at -22.0 and -99.8 ppm that accounted for <10% of
the total 31P signal intensity. To our knowledge, these shifts do not correspond to the

measured 31P shifts of any known cesium selenophosphate compound. The related

compound Cs4P2S10 was synthesized with the “Cs4P2S105” reaction stoichiometry and
although there was excess S, the as-yet unknown compound Cs4P2S9 was not

produced.21

The literature preparation of CS4P2Se9 was done with the “Cs5.4PZSe13.4”

reaction stoichiometry that contained an excess of both Cs and Se relative to Cs4P2Se9.13

Although an excess of elemental Se in the present study was found to favor the

production of Cs4P2Se10, this effect appeared to be counteracted in the literature

preparation by an increased Se2'/Se ratio that was a result of the excess C8236. The

183

Se2'/Se ratio correlates with the negative charge/Se atom ratio in the melt and increasing

these ratios might reasonably favor selenophosphate anions with shorter Se chains.31

Experiments were also done to understand the effect of the length of the
heating/cooling proﬁle on the Cs4P2Se10/CS4P2Se9 crystalline product ratio. The 600 0C
furnace syntheses were done with a “Cs4P2Se12” reaction stoichiometry and the

heating/cooling proﬁle length was varied between 18 h and 54 h. For all proﬁle lengths,

the ﬁnal product distribution showed that Cs4PZSe10 was more abundant than Cs4P2Se9
but there was a general trend towards smaller Cs4PZSe10:Cs4P2Se9 ratios with longer

heating/cooling proﬁles. This effect may provide some explanation for the discovery of

Cs4P28e10 under the relatively short 6 h heating/cooling proﬁle of the initial NMR

syntheses.

31P Room Temperature and High Temperature NMR Spectroscopy
A Herzfeld-Berger analysis32 of a room temperature 31P NMR spectrum of

CS4P23610 taken at 7 kHz spinning ﬁ'equency indicated 6“, 622, 533 chemical shift

anisotropy (CSA) principal values of 50, -41, and -167 ppm, respectively,33 and similar

values were obtained at 12 kHz spinning frequency, Figure 7-4. The 611 - 533 CSA
principal value difference of ~215 ppm was comparable to that of Cs4PZSe9 (~l80 ppm)
and other compounds with [PxSey]7-' anions containing (Se-Se)n linkages.11 The
longitudinal relaxation time T1 of CS4P2Se10 was 2100 :l: 600 s and was comparable to

the T1 values of other metal selenophosphate compounds.5 For both Cs4P2Se9 and

184

Cs4P2Se10 peaks, splittings were observed corresponding to P-Se spin-spin coupling with

coupling constant of ~270 Hz.

 

 

 

 

 

 

. 11.101

200 100 0 -100 -200 000
Chemical Shift (ppm)

 

Figure 7-4. The room temperature 31P NMR spectrum of a mixture with “Cs4P2Se16” stoichiometry that
was cooled over 24 h to produce crystalline CS4P2Se10 and C84P2Se9 . The MAS frequency was 7 kHz.
The intensities of peaks corresponding to CS4P2Se10 were used for the determination of CSA principal
values of this compound.

The high temperature 31P NMR spectra of the melts of syntheses that produced
Cs4P2Se10 or Cs4P2Se10/CS4P2Se9 mixtures contained one peak. Figure 7-lD displays a
representative spectrum of a 350 OC melt formed ﬁ'om Cs4P2Se9:Cs4P2Se10 products in a
3:5 ratio, Figure 7-1C. Spectra obtained at a given temperature were independent of both
the direction of the temperature ramp and the thermal history of the sample.

The -35 ppm shift at 350 0C was consistent with room temperature chemical shifts

of compounds with non-P-P bonded selenophosphate anions and suggested that the P-

containing high temperature species did not contain P-P bonds.11 The high temperature

shift was more positive than the crystalline Cs4P2Se9 or Cs4P2Se10 shifts observed at

room temperature perhaps because of the loss of deﬁned Cs+ ion coordination. An

185

alternate interpretation is that the 31P-containing species at high temperature were neither

the [PZSe9]4' nor the [P28e10]4‘ anions. At room temperature, the shiﬁs of Cs4PZSe9 and
Cs4PZSe10 differ by 13 ppm and it might be expected that if both [P28e9]4‘ and
[PZSe10]4' anions existed in high concentrations at high temperature, two peaks would be

observed in the high temperature spectrum. However, only one peak was observed in all
of the melts. The presence of two peaks in the room temperature spectrum but only one
peak in the high temperature spectrum (Figure 7-1C,D) shows that there is not a clear
NMR correlation between speciation in the crystalline products and speciation in the high
temperature melt. It is possible that fast chemical exchange at high temperature prevented
observation of discrete chemical shifts for the different species; however, such chemical
exchange was not obvious. The ﬁnal product spectra indicated an increase in the

Cs4PZSe10:Cs4PZSe9 ratio with increasing Se but there was no clear trend of the 31P

chemical shift in the high temperature melts as a function of Se stoichioimetry. For a
single temperature, the overall chemical shift variation in the melts as a function of
stoichiometry was less than 10 ppm In the future, spectra could be obtained at lower
temperatures and broader line widths at these temperatures would provide evidence for
chemical exchange.

The peak widths of the 31P NMR spectra at 350 0C were typically ~7 ppm or ~l

kHz which was less than the CSA magnitudes of Cs4PZSe10 (~215 ppm, 35 kHz) and

Cs4PZSe9 (~18O ppm, 30 kHz).11 The relative sharpness of the line at high temperature

indicated that most of the anisotropy had been averaged out by random tumbling.
Therefore, the frequency of motion must be >30 KHz, or have a rotational correlation

time of <5 us/radian.34 In the context of the Stokes-Einstein—Debye equation for a sphere

186

tumbling in a liquid, this correlation time 1 is proportional to the viscosity 1) and is
inversely proportional to the sphere vo lume V:

r = We T V (1)
where k3 is the Boltzmann constant and T is temperature in K units.35 Using a known T
and estimated 1, Eq. 1 could be used to calculate either 17 or V given a value for the

complementary property. Although neither 1] nor V in selenophosphate melts is known
with certainty, there is some information to assess these quantities: (l) ﬂame-heated
melts readily ﬂow when the sample tube is inverted; and (2) the P-containing species are
likely small molecular rather than large polymeric species because the ﬁnal crystalline

products contain discrete [PZSe9]4' or [PzSe10]4' anions. In the next paragraph, (1) V is
estimated from the dimensions of the [P28e10]4' anion; (2) 17 is calculated using this

value of V; and (3) comparison is made between this calculated 17 and visual assessment
of the melt viscosity. This approach tests whether a viscosity calculated with the
assumption of small P-containing anions is close to the viscosity estimated from visual

observation of the melts.

The radius of the [P2Se10]4' anion was estimated to be r z 8 A ﬁom the longest
Se-Se distance within a single [PZSe10]4‘ anion and the Se2' ionic radius. Using V =
4103/3, T = 1000 K (estimated temperature of ﬂame-heated melt), and r = 5 us, Eq. 1
yielded 17 ~ 15 Pa-s which is comparable to the viscosity of honey at room temperature36

or glycerol at 0 °C.37 This calculated 1] is likely an upper limit because 1 = 5 us is an

upper limit and because the P-containing anions might be smaller than [P2Se10]4', e.g.

[PSe4]3' with corresponding smaller V. Despite these caveats, there is general consistency

187

between the calculated 7] and visual observation of the melt viscosity. The approach

therefore yields qualitative support for the existence of small P-containing species.22

CONCLUDING REMARKS

The in situ NMR synthesis targeting Cs4P2Se9 led to the discovery of Cs4P2Se10
which can be formed with excess Se and a rapid heating/cooling proﬁle. The Cs4P2Se10

structure contained the novel discrete molecular anion [P2Se10]4'. The chemically related
compound Cs4P2810 is non-isomorphic, and it is possible that there are multiple
polymorphs of both the thio- and seleno- compounds. The optimum Cs4P2Selo synthesis
was performed in the high temperature NMR probe with an excess of Se relative to
Cs4PZSe10 and with a ~6 h heating/cooling proﬁle. The Cs4PZSew:Cs4PzSe9 ﬁnal
product ratio from furnace syntheses was dependent on stoichiometry and on the length
of the heating/cooling proﬁle. Cs4PZSe9 and Cs4PZSe10 were favored with Se-starved and
Se-rich stoichiometries and with long and short heating/cooling proﬁles respectively.

Room temperature 31P NMR spectra provided a facile means of evaluating the
relative amounts of the P containing compounds made in furnace and NMR syntheses. In
addition, some of these spectra presented in the paper have peaks that have not yet been
assigned to known crystalline compounds. Careﬁil screening of crystals ﬁom these
reactions may therefore result in new compound identiﬁcation.

The high temperature static 31P NMR spectra of the synthetic melts contained one
relatively narrow peak that may be the result of a rapidly tumbling selenophosphate anion
that does not contain P-P bonds, i.e. a P5+ species. Future time-dependent studies of high

temperature spectra may provide information about reaction rates. As illustrated by the

188

discovery of Cs4P2S610, future high temperature NMR syntheses may also aid discovery

of new compounds.

189

REFERENCES

l.

2.

.‘0

Kanatzidis, M. G. Curr. Opin. Solid. St. M 1997, 2, 139-149.

Bridenbaugh, P. Mater. Res. Bull. 1973, 8, 1055-1060.

. Rogach, E. D.; Sviridov, E. V.; Amautova, E. A.; Savchenko, E. A.; Protsenko, N. P.

Zh. T ekh. Fiz. 1991, 61, 201-204.

. Carpentier, C. D.; Nitsche, R. Mater. Res. Bull. 1974, 9, 1097-1100.

Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

. Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1993, 115, 4747-4753.
. Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1994, 116, 682-689.

. Maxwell, R; Eckert, H. J. Phys. Chem. 1995, 99, 4768-4778.

Mutolo, P. F.; Witschas, M.; Regelsky, G.; Guenne, J. S. A. D.; Eckert, H. J. Non-
Cryst. Solids 1999, 257, 63-72.

10. Schmedt, A. J.; Meise-Gresch, K.; Eckert, H.; Holand, W.; Rheinberger, V. Glass.

Sci. T echnol. 2000, 73, 90-97.

11. Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.

12. Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Longman Scientiﬁc and

Technical: New York, NY, 1983; pp 87.

13. Chondroudis, K; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 5401-5402.

14. Fehér, F. Handbuch der Praparativen Anorganischen Chemie; Ferdinand Enke

Verlag: Stuttgart, Germany, 1954; Vol. 1, pp 280-281.

15. McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 1257-1267.

16. SMART, SAINT, SHELXT L V-5 and SADABS, Bruker Analytical X-ray Instruments,

Inc.: Madison, 1998.

17. Hordvik, A. Acta. Chem. Scand. 1966, 20, 1885-1891.

18. Bottcher, P.; Getzschmann, J.; Keller, R. Z. Anorg. AlIg. Chem. 1993, 619, 476-488.

19. Pocha, R; Johrendt, D. Inorg. Chem. 2004, 43, 6830-6837.

190

20. Evenson, C. R.; Dorhout, P. K Z. Anorg. Allg. Chem. 2001, 627, 2178-2182.

21. Aitken, J. A.; Canlas, C.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2001, 40,
6496-6498.

22. Chung, I.; Do, J.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2004,
43, 2762-2764.

23. Brese, N. B.; Randall, C. R.; Ibers, J. A. Inorg. Chem. 1988, 27, 940—943.

24. Wachter, J. B.; Chrissaﬁs, K; Petkov, V.; Malliakas, C. D.; Bilc, D.; Kyratsi, T.;
Paraskevopoulos, K M.; Mahanti, S. D.; Torbrugge, T.; Eckert, H.; Kanatzidis, M. G.
J. Solid State Chem. 2007, 180, 420-431.

25. Chrissaﬁs, K; Kyratsi, T.; Paraskevopoulos, K M.; Kanatzidis, M. G. Chem. Mater.
2004, 16, 1932-1937.

26. In Advanced Data Storage Materials and Characterization Techniques, MRS
Proceedings, 2003; Hesselink, L.; Mijiritskii, A., Eds. 2003; p 803.

27. Edwards, A. H.; Pineda, A. C.; Schultz, P. A.; Martin, M. G.; Thompson, A. P.;
Hjalmarson, H. P. J. Phys. .° Condens. Matter 2005, 17, L329-L335.

28. Alan, D. H.; Kang, D. H.; Cheong, B. K; Kwon, H. S.; Kwon, M. H.; Lee, T. Y.;
Jeong, J. H.; Lee, T. S.; Kim, I. H.; Kim, K B. IEEE EIectr. Device L. 2005, 26, 286-
288.

29. Gidon, S.; Lemonnier, 0.; Rolland, B.; Bichet, 0.; Dressler, C.; Samson, Y. Appl.
Phys. Lett. 2004, 85, 6392-6394.

30. Ohta, T. J. Optoelectron Adv. Mater. 2001, 3, 609-626.
31. Kanatzidis, M. G.; Sutorik, A. C. Progr. Inorg. Chem. 1995, 43, 151-265.
32. Herzfeld, J.; Berger, A. E. J. Chem. Phys. 1980, 73, 6021-6030.

33. Eichele, K; Wasylishen, R. E. HBA 1.5, Dalhousie University and Universitat
Tﬁbingen: 2006.

34. Harris, R. K Nuclear Magnetic Resonance Spectroscopy; Longman Scientiﬁc and
Technical: New York, 1983; pp 86.

191

35. Cavanagh, J.; Fairbrother, W.; Palmer III, A.; Skelton, N. Protein NMR
Spectroscopy : Principles and Practice; Academic Press: San Diego, 1996; pp 17-19.

36. Juszczak, L.; Fortuna, T. J. Food Eng. 2006, 75, 43-49.

37. Weast, R. C.; Astle, M. J., CRC Handbook of Chemistry and Physics. 60th ed.; CRC
Press, Inc.: Boca Raton, 1980; p F-55.

192

CHAPTER 8
NUCLEAR MAGNETIC RESONANCE AND ELECTRON PARAMAGENTIC

RESONANCE EVIDENCE FOR FREE RADICALS IN CHALCOGENIDE
MELTS AND SYNTHESES.

193

INTRODUCTION

There has been much recent interest in various chalcophosphate compounds
because of potentially useﬁil physical properties such as ferroelectric ordering, 1'6
reversible glass-to-crystalline phase changes,7'll and magnetic phenomena. 12'” The
structures of these compounds contain an assortment of [Pny 2' (Q = S, Se, Te) anions,
and some ability to choose the type of anion present in the ﬁnal compound has been
demonstrated by varying the reaction stoichiometry.7,15’l6 It was proposed that the

products formed in these reactions may be affected by the equihbrium between [Pny 2'

anions.15
The different [Pny]7-' anions in the ﬁnal crystalline products have different
phosphorous bonding environments, and it has been shown that NMR spectroscopy

provides useful structural information on these compounds. For example, the 31P chemical

shift is strongly dependent on the presence or absence of a P-P bond in metal
selenophosphates, and 31P-31P dipolar coupling strength has been used to determine
thiophosphate anion identity. 17,18 Therefore, if the hypothesis that the anions remain intact
at high temperature is true, it is poss1ble that species in high temperature metal
chalcophosphate reactions could be characterized using 31P NMR spectroscopy. There are
also technical reasons for the choice of 31P NMR and the metal chalcophosphate system
First, 31P NMR spectroscopy has high sensitivity because the 31P isotope has 100%
abundance, spin 1/2, and a large gyromagnetic ratio. Second, many metal chalcophosphates

have been prepared using temperatures within the range available to a cormnercially

available high temperature NMR probe (200 0C — 600 0C). Third, the physical state of the

metal chalcophosphate mixture at high temperature is often a melt, i.e. a viscous liquid in

194

which small species will tumble fairly rapidly. This tumbling will provide signiﬁcant spatial
averaging of the anisotropic chemical shift and dipolar couplings experienced by the 31P
nucleus and should result in sharper 31P signals than would be observed in the solid state.
Although magic angle spinning (MAS) of the sample can also average these anisotropic
interactions, MAS at kHz ﬁequencies is technically challenging for the sealed glass
reaction ampoules used to contain the reaction mixtures. Flame sealing typically produces
an ampoule shape with some non-cylindrical character and the consequent wobbling
motion during MAS impedes spinning at high frequencies.

In this study, a combination of NMR and EPR spectroscopies was used to detect a
signiﬁcant population of free radicals in chalcophosphate and chalcostannate melts. The
presence of ﬁ'ee radicals can have a pronounced effect on many types of chemical
reactions including those related to polymerization,l9 organic synthesis,20'22 atmospheric
chemistry”,24 and biology and medicine”,26 Free radicals in pure S and Se melts were
ﬁrst identiﬁed in the 1950s,27'33 but to our knowledge, the presence of ﬁ'ee radicals in
chalcogenide-containing solid state reactions has not been previously considered. EPR
signals have previously been observed in solid chalcophosphate materials, and it was
proposed that the rapid longitudinal relaxation (20 — 80 s) of some room temperature
crystalline compounds could be explained by the presence of some residual ﬁee radicals.34
The identiﬁcation of signiﬁcant radical concentrations in such melts indicates that radical

based reaction mechanisms merit serious consideration.

195

EXPERIMENTAL SECTION
Synthesis

Samples of a particular nominal stoichiometric composition were prepared by
combining the appropriate ratio of the starting materials in fused silica ampoules, sealing
in vacuo and then heating in computer controlled ﬁlmace. The reactants were either the

elements or, in the case of the alkali metal-containing samples, A2Q (A = Na, K, Rb, Cs;

Q = S, Se).35 This mixture was typically heated to 400 °C over 4h, held there for 12h, and
then the ﬁimace was turned off. Sample portions of the resultant ingot were rescaled in

ﬁised silica ampoules and then either used for NMR or EPR spectroscopy. The
compounds K2Cu2P4Se1036 and K1.5Bi2.5(PS4)337 were synthesized according to their
respective literature preparations. Samples denoted in quotation marks such as “P85” refer
to an initial mixture of non-crystalline starting materials with the given stoichiometry and
samples denoted without quotation marks such as K2Cu2P4Se10 refer to a single

crystalline compound starting material.

Nuclear Magnetic Resonance (NMR)

The high temperature 31P NMR spectra were collected on a 9.4 T spectrometer
(Varian Inﬁnity Plus) using a single resonance static probe (Doty Scientiﬁc). Sample
portions of the previously described ingots weighing ~75 mg were sealed in ﬁlsed silica
ampoules in vacuo and placed in a protective ﬁised silica sheath at the center of the coil in
the probe. The samples were heated to the desired temperature at a rate of 10 °C / min and

allowed to equilibrate for 5 min before data collection. The 1: pulse width was typically

12-16 us and was cahbrated to within 0.2 as at each temperature where a spectrum was

196

spectrum was collected with the transmitter set at the resonant ﬁequency of the observed
spectral peak. The 1r/2 pulse duration was calculated as half of the 1t pulse width
Longitudinal relaxation times were measured using an inversion recovery pulse sequence

(rt-t-tt/Z-acquisition) and the data were ﬁt to the equation34

I = (210)—11 -(1—e“"”)) (I)
where I was the observed spectral intensity at pulse delay time 1', T1 was the ﬁtted
longitudinal relaxation time, and 10 and I 1 were ﬁtting parameters that represented the
signal intensity at 2' = O and T = 00. The spectra were processed with minimal line
broadening and baseline correction and referenced to 85% H3PO4 at 0 ppm. The total

time spent above room temperature was ~4 h.

Room temperature 31P NMR spectra were collected on the same spectrometer but
using a Varian Chemagnetics double resonance magic angle spinning (MAS) probe. The
sample volume in the 4 mm diameter zirconia rotors was ~50 11L and the MAS ﬁequency
was in the 8 — 14 KHz range. Each sample was ground to a ﬁne powder before being

packed in the rotor. The 1t/2 pulse width was ~4-6 us and was calibrated using H3PO4,

and spectra were processed identically to those at high temperature.

Electron Paramagnetic Resonance (EPR)

EPR spectra were collected on an X-band Bruker ESP300e instrument with a
Bruker perpendicular mode 4102ST cavity. The microwave ﬁ'equency was monitored by
an EIP Microwave model 25B frequency counter. The samples were prepared by sealing
in vacuo a known mass (~45 mg) of the desired starting material in a 3 mm outside

diameter ﬁised silica ampoule. The samples were then placed in a ceramic holder and

197

inserted in a furnace. The fumace was heated over 4 h to reach either 550 or 1000 0C and
held there for 12 h. The samples were then removed from the furnace and bolder and
immediately quenched in ice water and placed into the EPR cavity. Single spectra were
collected with ~200 mW X-band microwave radiation.

Approximate concentrations of radicals were determined using a calibration curve
generated using known masses of 2,2-diphenyl-1-picrylhydrazyl (DPPH). After achieving
critical cavity coupling, the iris of the resonator was ﬁxed, and retuning for each
subsequent reference and chalcophosphate or chalcostannate sample was performed
without changing the iris setting. Spectra used for quantitation were collected using
identical parameters, i.e. modulation amplitude, modulation frequency, time constant,
sweep width, and temperature. Signal intensity was determined by ﬁtting the background
with up to a cubic polynomial ﬁt and then performing a double integration. Quantitation of

radicals in the 1000 °C quenched PS5 sample was also performed using a CuSO4 standard

of known concentration. The samples used for cahbration were prepared by standard

dilution of CuSO4 in a 1:1 mixture of glycerol and water, and then adding a known

volume of the resultant solution to an EPR tube. The standard samples were quenched in
liquid nitrogen before inserting into the EPR cavity which was maintained at 5 K during
the measurement. The chalcophosphate samples were also maintained at 5 K during data
collection which was otherwise performed identically to the procedure described above.
For each sample, the radical concentration was converted to one unpaired electron spin
per some number of atoms by dividing the experimentally derived radical concentration by
the approximate number of atoms in the sample determined using the sample mass and

composition.

198

Powder X-ray diffraction

Powder X-ray diﬂr‘raction patterns of the crystalline compounds were collected
before and after high temperature NMR measurements using a Rigaku—Denki RW400F2
(Rotaﬂex) powder diffractometer. The data were collected using a copper rotating anode

operating at 45 kV and 100 mA.

RESULTS AND DISCUSSION

The original motivation for this study was to test the hypothesis that an equilibrium
between the various [Pny 2' anions exists at high temperature, as was previously
proposed. 15 During setup of the NMR spectrometer to study a prototypical
chalcophosphate sample, it was serendipitously discovered that a cumulative Bloch decay
signal could be obtained with pulsing with only ~100 ms delay between acquisitions. This

result was initially surprising because of the typical ~1000 5 T1 value of 31F in room

temperature solid metal chalcophosphates, Table 8-1.34 Although fast relaxation had been

previously observed in high temperature NMR measurements of related materials,38'42 we

are not aware of a systematic study to understand this phenomenon or assign its cause.

NMR evidence for free radicals in chalcophosphate melts
Figure 8-1 displays representative spectra at different temperatures for a melt with

“PS 5” stoichiometry and Figure 8-2 graphically displays the temperature dependences of
the peak chemical shifts line widths, and longitudinal 1/ T1 relaxation rates. Table 8-1

summarizes the high temperature NMR parameters for this and other melts and for some

199

of the NMR spectra before and after heating and cooling. Figure 8-3 displays two other

representative high temperature NMR spectra of K2Cu2P4Se10 and “KPSe7”. The
crystalline compounds K2Cu2P4Selo, K1.5Bi2.5(PS4)3, and KH2P04 melted congruently
so the initial and ﬁnal compounds were the same as determined by comparisons of the

powder X-ray diffraction patterns and the 31P NMR spectra before and after thermal

treatment .

600 °C

550 °C

500 °C

450 °C

 

q

7 I I

300 200 100 0 $00
Chemical Shift (ppm)

’9

Figure 8-1. NMR spectra of a melt mixture with the nominal ratio “P85 showing the increase in

chemical shift and peak width as a function of temperature. Spectra at 450, 500, 550, and 600 0C were
acquired with acquisition delays of 2 s and represent the sums of 16 scans.

200

Table 8-1. NMR parameters in high temperature melts and room temperature solids. Samples in
quotation marks represent nominal compositions, and those without quotation marks were pure crystalline
compounds at room temperature. The 22 °C data listed above and below the high temperature data were
collected before and after the high temperature data, respectively.

 

 

 

 

 

 

 

 

 

 

Temp. Peak Chemical Peak Width ”“31“!“
Sample (°C)a Shift (p m)b (KHz)° 1Relaxation
p Time (ms)d
350 -l3.2 1.7 11.7
KPSe7 450 3.4 2.2 2.5
550 30.6 2.7 0.47
22 35.0, -21.8 1.0, 4.8 n.d.
CS4P25616 350 -34.8 1 n.d.
450 -21.4 1.5 n.d.
550 7.06 2.3 n.d.
K4P6T625 550 355.8 2.5 0.7
K4P6qu 550 77.2 4 0.3
350 80.4 4.1 115
P85 450 86.9 2.5 37
550 98 5.7 9
22 55.7 0.1 1,0x106
400 93.3 3 4.8
KZCu2P4Seloe 500 109.9 3.7 2.5
600 130.9 5.1 1.5
22 55.7 1 1.4x105
K1 _53i2.5(1>s4)3f 550 76 0.9 15.7
22 88.8, 72.4 1.0, 0.8 5.7x105
22 4.2 0.5 1,3x106
KH2P04 350 -7.2, -l4.7 0.6, 0.5 6.7x103
22 4.5 0.7 7.0x105
KH2P04S 350 -7.2, -13.2 1.0, 1.5 343

 

a The uncertainty in temperature is estimated to be :I: 5 °C.

1’ Chemical shifts were determined by ﬁtting the spectral peaks to a 50% Gaussian 50%
Lorentzian line function and the reported chemical shift was the peak center ﬁtting parameter.
The uncertainty is estimated to be :h 0.5 ppm.

C Peak widths were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian
line function and the reported line width measurement was the full width at half maximum
(FWHM). A 50% Gaussian 50% Lorentzian line was chosen because it was in closer
agreement with the experimental spectrum than a Gaussian or a Lorentzian line alone. Room
temperature linewidths were measured under MAS conditions and high temperature linewidths
were measured under static conditions. The estimated uncertainty in the line width is :t 5%.
dThe method of measuring the longitudinal relaxation time is described in the text. The
uncertainty is estimated to be :1: 12 %.

e Chondroudis, K; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2098-2099.
f Gave, M.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007,46, 11063-11074.

201

 

 

 

 

 

110- (A) ’9'”
3 +42 I
G.

3' 100- k? "’
a: e:- -10 2.
2
m 94 @ é
90~ an *8 ~-
23 e- + 3'
E 9* “i‘ + ~6
5 80‘ + 4 E
E] + ”5‘ V
70 i *P T . . . 2
400 450 500 550 600
Temperature (°C)

0.35. (B)

0.30- .1.
£0254
1.0.204
Eors- {>-

.0.104 +
FIN
000 m "' ...
' 400 450 500 550 600
Temperature (°C)

Figure 8-2. (A) Peak 31F chemical shifts and peak widths as a function of temperature in a melt with the
nominal composition “PS5”. Chemical shifts and peak widths were determined by ﬁtting the spectral
peak to a 50%/50% Gaussian and Lorentzian line ﬁmction. The chemical shifts and peak widths are

represented by open and closed squares, respectively. (B) 31F longitudinal NMR relaxation rates for
“P85” as a function of temperature.

202

(A)

K2Cu2P4Se o

 

 

360 200 160 o -160

 

 

360 260 160 o -1bo
Chemical Shift (ppm)

Figure 8-3. High temperature 31P NMR spectra of (A) K2Cu2P4Se10 at 550 °C and (B) a sample with
the nominal composition “KPSe7” at 350 °C. The spectra were acquired with acquisition delays of 10 ms
and 200 ms, respectively and represent the sums of 50 scans.

In all samples, a single 31P NMR peak was observed in the melt, Figures 8-1 and
8-3. In samples for which spectra were obtained at multiple temperatures, the peak
chemical shift moved downﬁeld with increasing temperature and the line width and
longitudinal relaxation rates increased with increasing temperature. In a single sample, the
peak shifts, line widths, and longitudinal relaxation rates had non-linear dependence on

temperature. Determination of the transverse relaxation time T2 was attempted using a
Carr-Purcell-Meﬂaoom—Gill echo pulse sequence“,44 but accurate determination of T2
when T 2 < 100 as is impossible because it is not possrble to have pulses separated by less
than 100 us. The typical decay of the echoes occurred in <100 us and this suggests an
upper limit of ~100 us on 31P T2 values in the melt.

For samples containing S, Se, or Te, the typical 31P T1 in the high temperature
melt was ~10 ms and was ~105 times shorter than the typical room temperature T1 of

crystalline metal chalcophosphates. The effect was much less pronounced in the KH2P04

203

melt and the T1 for this melt at 350 0C was ~150 times shorter than the T1 of the

crystalline compound. The T1 = 340 ms in the “KH2P04S” melt at 350 0C can be
compared to the T1 = 6700 ms in the KH2PO4 melt and shows that the presence of
chalcogen in the melt has a substantial effect on T1. There does not appear to be different
chemistry with addition of S because the X-ray powder pattern of the ﬁnal ambient
temperature “KH2P04S” sample was a superposition of the KH2P04 crystalline powder
pattern and an amorphous background, presumably ﬁ'om elemental S.

The rapid longitudinal relaxation rates (1/ T 1 s) in the melts and the increases in
1/ T1 3, line widths and chemical shifts with increasing temperature are all potentially
attributable to the presence of unpaired electron spins. The presence of radicals in melts of

the pure S and Se has been well documented23,30,31,45 so it is reasonable to consider that
there are also radicals in chalcophosphate melts. Unpaired electrons typically increase
nuclear spin relaxation rates because the electron spin has a large dipole moment and

electron spin transitions can occur near the NMR ﬁ'equency of the detected nucleus.
Stable free radicals such as [Cr(H20)6]ClO446 and Gd(dpm)3 (dpm = bis(2,2,6,6-
tetramethyl-3,5-heptadionate))47 have been intentionally added to solutions to decrease

the nuclear spin T148 and to allow more rapid collection of the NMR data. At sufﬁciently

high concentrations (i.e. >10'2 M), the unpaired electron spins will also increase the
nuclear spin transverse relaxation rate and deshield the nucleus.46,47 These effects are

consistent with the relatively short T 2s and broad line widths of the 31P NMR melt spectra

and the downﬁeld 31P chemical shifts. Because free radicals are high energy species, it is

reasonable that their concentration would increase with increasing temperature. This

204

hypothesis correlates with the larger 31P longitudinal relaxation rates, line widths and peak
chemical shifts at higher temperatures.

The dependence of the chemical shift and peak width is correlated with a non-
linear increase of radical concentration as a ﬁrnction of temperature. This nonlinear
dependence of the chemical shift, and peak width is modeled in more detail in Chapter 9.
The paramagnetic contribution to the magnetic shielding depends linearly on the radical
concentration.28 Therefore, a non-linear change in chemical shift and peak width results
from a non-linear increase in free radical concentration. A similar temperature dependence

was also observed for pure elemental Se.28

Comparison of chalcophosphate melts to pure Se by high temperature NMR

The 77Se T ls of Se melts were 2 ms at 400 °C and ~0.2 ms at 600 °C28 and were
comparable to the high temperature chalcophosphate relaxation times summarized in
Table 8-1. Over this same temperature range (400 - 600 °C) the transverse relaxation
time, T 2 of the 77Se melts changed ﬁ‘om ~0.l ms at 600 °C to ~08 ms at 400 °C, or
linewidths in the T 2 broadened limit of ~3.2 KHz and ~0.4 KHz, respectively. These are
comparable to the linewidths of the chalcophospate linewidths summarized in Table 8-1.

The similarity of the 31P NMR parameters in chalcophosphate melts and 77Se NMR
parameters in selenium melts as well as similarity of the parameter variations with
temperature suggest that the parameters are governed by the same underlying
phenomenon. For the pure Se melts, this phenomenon has clearly been shown to be a

signiﬁcant concentration of unpaired electron spins.

205

The change in the 77Se chemical shift that was observed in a melt of pure Se28 is
similar to the change in the 31P chemical shift in chalcophosphate melts in frequency units.

Comparison of the 77Se spectra at 400 and 600 °C showed a chemical shift difference 8600
— 5400 z 85 ppm, or 6.5 KHz assuming a 76.3 MHz 77Se resonant ﬁ‘equency.28 For the

melts of K2Cu2P2Se10, the 31P NMR spectra had 5600 — 8400 z 37.6 ppm, Table 8-1, or
6.1 KHz. The average hyperﬁne coupling strength between the electron and nucleus is
independent of applied ﬁeld, and is proportional to the number of unpaired spins that are
present and proximate.49 The similarity of the contact shifts in ﬁequency units is evidence
that the concentrations of ﬁee radicals in pure Se melts is comparable to that in the
chalcophosphate melts. At 600 °C, pure elemental Se was reported to contain 1 unpaired
electron per ~350 atoms.28 For this comparison to be made, however, the assumption
must be made that the hyperﬁne coupling between the electron and the 77Se nuclei in the
pure chalco gen melts is equivalent to the electron to phosphorous hyperﬁne coupling in

the chalcopho sphate melts.

High temperature NMR of a tellurophosphate-containing melt

The 31P NMR spectrum of “K4P6Te25” was similar to the other chalcophosphate
melts studied in that there was a single resonance with a rapid relaxation rate, but the
chemical shift was signiﬁcantly ﬁrrther downﬁeld than any of the other melts studied,

Table 8-1. This indicated that the 31P local environment in this melt was ﬁmdamentally
diﬁ'erent than those in the other 0, S, or Se-containing melts. To date, there are only

several known tellurophosphates with P-Te bonds, whereas there are thousands of

members with P-O, P-S, or P-Se bonds. The chemical shifts of crystalline compound

206

K4P8Te4 were -4.3, -l6.9, and -34.6 ppm These chemical shifts are similar to those

observed in various oxo-, thio- and selenophosphate compounds. The combination of the
NMR and synthetic results suggests that the species that exist in P/T e containing melts
may be very different than those in P/O, P/S, and P/Se containing melts. The
electronegativity of Te (2.1 Pauling units) is only slightly less than the electronegativites of
Se (2.4 Pauling units) and S (2.5 Pauling units),50 so the difference in the high

temperature chemical shifts is unlikely to be explained solely by these electronegativities.

Room temperature electron paramagnetic resonance

Room temperature EPR spectra of rapidly quenched chalcophosphate melts were
collected to test for the presence of radicals that were present at high temperature and did
not recombine during the quenching process. This approach had previously been used to
study radical concentration in quenched Se and S melts and it was estimated that there
was ~100-fold loss in radical concentration during the quenching process presumably due
to radical recombination.51

Figure 8-4 displays EPR spectra of “PS5” samples quenched from 550 and 1000
°C and Figure 8-5 displays EPR spectra of “SnS5”, “PSe5”, and K1_5Bi2_5(PS4)3. The
spectral lineshapes and intensities of the S-containing samples at a given temperature were
approximately independent of the chemical stoichiometry. The 550 °C samples contained a
symmetric line centered near g z 2.00 and with an ~ 5 G peak width, Figures 8-4B and 8-
5A. The S-containing samples quenched from 1000 °C were also similar, and contained

an asymmetric line centered at g = 2.02 with an ~230 G peak with, Figures 4A, 5C.

Possible reasons for the broad line width and complex line shape of the 1000 °C samples

207

include: (1) the dominant radical species had a large hyperﬁne coupling between the

unpaired electron spin and the 31P nucleus, (2) the dominant radical species had a large g-

anisotropy (3) there was a superposition of signals from chemically different radical
species or (4) the dominant species was a biradical.28 The “PSe5” sample contained at

least two different resonances centered at g = 2.007 and g 2 1.933 with line widths of 11

G and 75 G, respectively, Figure 8-5B.

(B) g= 2.011

 

235 5 G

Figure 8-4. EPR spectra at room temperature of rapidly quenched “P85” samples. Prior to quenching, the

sample temperature in the furnace was (A) 1000 °C or (B) 550 °C. Acquisition parameters were: (A) 10
G modulation amplitude, 2000 G sweep width and (B) 0.1 G modulation amplitude, 50 G sweep width.

   

7= 1.933

g = 2.009

  

233 G

Figure 8-5. Room temperature EPR of (A) “SnS5” quenched ﬁ'om 550 °C (B) “PSe5” quenched from 550
°C and (C) K1.5Bi2_5(PS4)3 quenched from 1000 °C. Acquisition parameters were: (A) l G modulation

amplitude, 40 G sweep width; (B) l G modulation amplitude, 500 G sweep width; and (C) 10 G
modulation amplitude, 3000 G sweep width. Each spectrum was a single acquisition.

208

Comparison of integrated spectral intensities with those of DPPH and CuSO4

standards yielded one unpaired electron per ~105 atoms in the 550 0C sample of Figure 8-
4B and l unpaired electron per ~104 atoms in the 1000 oC sample of Figure 8-4A. The

550 °C quenched “SnSS” sample of Figure 8-5A also contained 1 unpaired electron per

~104 atoms. Samples quenched from 1000 0C melts formed from the pure crystalline
compound K1.SBi2.5(PS4)3 had comparable line shapes, line widths, and radical
concentrations to that of “P85” quenched from the same temperature, Figure 8-5C .
Radical quantitation of the “PSe5” sample was unreliable because of signiﬁcant
nonlinearity in the baseline and a partial overlap of the signals.

These data suggest that signiﬁcant radical concentrations are a general feature of
chalcogen-containing melts. The measured concentrations of l unpaired electron per 104-
105 atoms in the quenched melts likely represents a lower limit on the radical
concentrations in the high temperature melts because of loss of radicals during the
quenching process presumably from radical recombination. A loss of at least IOO-fold was
reported in an earlier study on quenched Se and S melts.51

It was previously reported that pure Se at 600 °C and 100 bars contained 1
unpaired electron per ~350 atoms and pure Se at 1000 0C and 100 bars contained 1
unpaired electron per ~20 atoms.28 If there is a lOO-fold loss in radical concentration

during the quenching process, the EPR results on the quenched melts correlate with

radical concentrations in the thiophosphate and thiostannate melts of one unpaired

electron per 103 atoms at 550 0C and one unpaired electron per 102 atoms at 1000 °C.
These concentrations are comparable to the reported radical concentrations in the pure Se

melts near these temperatures. To our knowledge, there is no similar literature on the

209

quantitation of elemental S, but our quantitation of a 550 °C quenched pure S melt was

found to contain 1 unpaired electron per ~104 atoms. These concentrations are again
comparable to the thiophosphate samples studied here, and also the concentration of
radicals in pure Se, assuming the lOO-fold loss of radicals upon quenching.

Most EPR spectra were obtained within minutes of quenching but for the “PS5”

sample quenched from 1000 °C, an additional spectrum was obtained four hours after
quenching. The g values, line shape, and line width did not change with time but there was
a ~10% loss in signal intensity after four hours that was evident as a change in the absolute
intensity of the signal. This change can likely be ascribed to radical recombination that

occurs at room temperature following the quench.

Chemical implications of free radicals

Free radical polymerizations of organic monomers occurs in the presence of S 10’6
M radical concentration, or, assuming a dilute monomer concentration, about one radical
per 108 atoms.29’52 The radical concentration in chalcophosphate and chalcostannate

melts appears to be 103-106 times greater and it is reasonable to consider that the radicals
play some role in the melt chemistry and the eventual formation of ambient temperature

crystalline products. One speciﬁc hypothesis is that the S-S or Se-Se bonds in complex
anions53 such as [P2810]4‘ or [P28e10]4' in crystalline compounds could be formed by
radical recombination. The NMR T1 data suggest that the concentrations of radicals are

much higher in chalcophosphate melts than in oxyphosphate melts and this difference

could be correlated with the larger variety of bonding motifs of chalcophosphate anions in

crystalline compounds. For example, compounds with anions such as [P2S10]4' and

210

[P4Se10]4' have been observed but there are no reports to-date of compounds with the
corresponding [P2010]4' and [P4010]4' anions even though thousands oxyphosphate
compounds are known. One further possrbility is that the differences that are observed
between reactions performed at intermediate temperatures (i.e. 200 — 600 °C) and those at

higher temperatures (i.e. > 600 °C)54 may be correlated with the increase in radical

concentration that occurs over those temperatures.

CONCLUDING REMARKS
31P NMR and EPR studies provide compelling evidence for a radical concentration

in chalcophosphate and chalcostannate melts of one unpaired spin per 102 — 105 atoms
and the radicals may play a signiﬁcant role in the melt chemistry and formation of the ﬁnal

crystalline compounds. The radicals also have a signiﬁcant impact on the application of

31P NMR to these systems: the short relaxation times allow for spectra to be acquired
rapidly, but also give rise to broad spectral lines. Future EPR studies will perhaps be

useful for determining the structure of the radical-containing species.

211

REFERENCES

l. Bourdon, X.; Maisonneuve, V.; Cajipe, V. B.; Payen, C.; Fischer, J. E. J. Alloys
Compd. 1999, 283, 122-127.

2. Carpentier, C. D.; Nitsche, R. Mater. Res. Bull. 1974, 9, 1097-1100.

3. Gave, M.; Bilc, D.; Mahanti, S.; Breshears, J.; Kanatzidis, M. Inorg. Chem. 2005,
5293-5303.

4. Maisonneuve, V.; Cajipe, V. B.; Simon, A.; VonDerMuhll, R.; Ravez, J. Phys. Rev. B.
1997, 56, 10860-10868.

5. Simon, A.; Ravez, J .; Maisonneuve, V.; Payen, C.; Cajipe, V. B. Chem. Mat. 1994, 6,
1575-1580.

6. Vysochanskii, Y. Ferroelectr. 1998, 218, 629-636.

7. Chung, I.; Do, J.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2004,
43, 2762-2764.

8. Chrissaﬁs, K; Kyratsi, T.; Paraskevopoulos, K. M.; Kanatzidis, M. G. Chem. Mat.
2004, 16, 1932-1937.

9. Gidon, S.; Lemonnier, 0.; Rolland, B.; Bichet, 0.; Dressler, C.; Samson, Y. Appl.
Phys. Lett. 2004, 85, 6392-6394.

10. Ohta, T. J. Optoelectron. Adv. M 2001, 3, 609-626.

11. Wachter, J. B.; Chrissaﬁs, K; Petkov, V.; Malliakas, C. D.; Bilc, D.; Kyratsi, T.;
Paraskevopoulos, K M.; Mahanti, S. D.; Torbrugge, T.; Eckert, H.; Kanatzidis, M. G.
J. Solid State Chem. 2007, 180, 420-431.

12. Chondroudis, K; Kanatzidis, M. G. Inorg. Chem. Commun. 1998, 1, 55-57.

13. Odile, J. P.; Steger, J. J.; Wold, A. Inorg. Chem. 1975, 14, 2400-2402.

14. Taylor, B.; Steger, J.; Wold, A.; Kostiner, E. Inorg. Chem. 1974, 13, 2719-2721.

15. Kanatzidis, M. G. Curr. Opin. Solid. St. M 1997, 2, 139-149.

16. Chung, L; Karst, A. L.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2006, 45,
2785-2787.

17. Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.

212

18. Gunne, J.; Eckert, H. Chem-Eur. J. 1998, 4, 1762-1767.

19. Hutchinson, R. Handbook of Polymer Reaction Engineering; Wiley-VCH Verlag
GmbH & Co. KGaA: Weinheim, Germany, 2005; Vol. 1, pp 153-212.

20. Wille, U. Chem. -Eur. J. 2002, 8, 341-347.

21. Zhuikov, V. V. Russ. J. Gen. Chem. 1999, 69, 1796-1800.

22. Kaim, W. Z. Naturforsch. B 1985, 40, 61-66.

23. Monks, P. S. Chem. Soc. Rev. 2005, 34, 376-395.

24. Mellouki, A.; Le Bras, G. Mater. Sci. Forum. 1999, 301, 47-67.

25. Aruoma, O. I.; Halliwell, B. Molecular biology of ﬁee radicals in human diseases;
OICA International: Sant Lucia, London, 1998.

26. Halliwell, B.; Gutteridge, J. M. C. Free radicals in biology and medicine; 4th ed.;
Oxford University Press: New York, NY, 2007.

27. Sampath, P. I. J. Chem. Phys. 1966, 45, 3519-3525.
28. Warren, W. W.; Dupree, R. Phys. Rev. B. 1980, 22, 2257-2275.

29. Carswell, T. G.; Hill, D. J. T.; Hunter, D. S.; Pomery, P. J.; Odonnell, J. H.; Winzor,
C. L. Eur. Polym. J. 1990, 26, 541-544.

30. Koningsberger, D.; Van Wolput, J.; Rieter, P. Chem. Phys. Lett. 1971, 8, 145-147.
31. Gardner, D. M.; Fraenkel, G. K J. Am. Chem. Soc. 1956, 78, 3279-3288.

32. Gardner, D. M.; Fraenkel, G. K J. Am. Chem. Soc. 1954, 76, 5891-5892.

33. Massen, C. H.; Weijts, A. G. L.; Poulis, J. A. T. Faraday Soc. 1964, 60, 317-&.

34. Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

35. Fehér, F. Handbuch der Praparativen Anorganischen Chemie; Ferdinand Enke Verlag:
Stuttgart, Germany, 1954; Vol 1, pp 280-281.

36. Chondroudis, K; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2098-2099.
37. Gave, M.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46, 11063-11074.

38. Maxwell, R.; Eckert, H. J. Phys. Chem. 1995, 99, 4768-4778.

213

39

40

41

42

43.

45.

46.

47.

48.

49.

50

51

52

53

54

. Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1993, 115, 4747-4753.
. Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1994, 116, 682-689.
. Maxwell, R.; Erickson, H.; Eckert, H. Z. Naturforsch. A 1995, 50, 395-404.

. Schmedt, A. J.; Meise-Gresch, K; Eckert, H.; Holand, W.; Rheinberger, V. Glass.
Sci. T echno]. 2000, 73, 90-97.

Carr, H. Y.; Purcell, E. M. Phys. Rev. 1954, 94, 630-638.

Melboom, S.; Gill, D. Rev Sci Instrum 1958, 29, 688-691.

Tveryanovich, Y. S.; Borisova, Z. U. Inorg. Mater. 1979, 15, 1664-1667.
Freeman, R; Pachler, K G. R; Lamar, G. N. J. Chem. Phys. 1971, 55, 4586-4593.
Dechter, J. J .; Levy, G. C. J. Magn. Reson. 1980, 39, 207-215.

Sharp, R.; Lohr, L.; Miller, J. Prog. Nucl. Mag. Res. Sp. 2001, 38, 115-158.

Wertz, J. B.; Bolton, J. R. Electron spin resonance; elementary theory and practical
applications; McGraw-Hill: New York,, 1972; pp 38-48.

. Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements; 2nd ed.; Butterworth-
Heinemann: Boston, MA, 1997; pp 662, 754.

. Bagley, B. G.; Waszczak, J. V.; Disalvo, F. J. Solid State Commun. 1972, 11, 89-92.

. Shen, J. C.; Tian, Y.; Zeng, Y. G.; Qiu, Z. W. Makromo]. Chem-Rapid 1987, 8, 615-
620.

. Aitken, J. A.; Canlas, C.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2001, 40,
6496-6498.

. Kanatzidis, M. G.; Sutorik, A. C. Progr. Inorg. Chem. 1995, 43, 151-265.

214

CHAPTER 9

ON THE SPECIATION OF CHALCOPHOSPHATE-CONTAINING MELTS.

215

INTRODUCTION

The presence of a signiﬁcant concentration of free radicals in chalcogen-
containing melts has recently been demonstrated.1 The existence of the radicals was
inferred from the 31P high temperature NMR spectra that had chemical shifts, peak
widths, and relaxation times that were strongly temperature dependent and these effects
can be readily interpreted on the basis of the presence of free radicals.2'4 The presence of
the free radicals was conﬁrmed by the presence of strong EPR signals on chalcogen-
containing samples that had been rapidly quenched from the melt.1’5

Previously, it was shown that solid state NMR provides useful structural
information of diamagnetic crystalline selenophosphate compounds. Most notably, the
31P chemical shift of selenophosphate-containing crystalline compounds is correlated
with the presence or absence of a P-P bond. Compounds that contain a selenophosphate
anion with a P-P bond typically have positive chemical shifts, and compounds with a
selenophosphate anion without a P-P bond typically have negative chemical shifts. For

example, the compounds RbZCdPZSe6 and Pb2P28e6 which contain the molecular
[P2Se6]4‘ anion have chemical shifts of 62.0 ppm and 29.1 ppm, respectively.6

Conversely, compounds without a P—P bond typically have a negative chemical shift.

Examples include Cs4PZSe9 and Cu3PSe4 which contain the molecular [PzSe9]4' and
[PSe4]3' anions, respectively, and resonate at -39.9 ppm and -83.3 ppm.6 The high

temperature 31P NMR spectra are not trivially interpreted in a similar fashion because:
(1) crystal structures are available for the solid compounds but not for the liquid melts
and therefore the 31F chemical environments have not been previously determined; (2)

the paramagnetic contact shift that occurs with increasing radical concentration is capable

216

of moving the 31P spectral feature from a region that could be attributed to a non P-P
bonded species to a region ascribed to P-P bonded species.

The chemical shift and peak width have a non-linear temperature dependence, and
this dependence is correlated with the effects of the paramagnetic radical species. In this
study, the effects of the free radicals on the chemical shift and peak width were modeled,
and this effect was removed so that the chemical shift and peak width of the diamagnetic
species could be determined. Some correlation existed between the modeled chemical
shift and the chemical shift that would be expected based off of the 31P chemical
environment in the known solids. This provides some evidence that the phosphorous
coordination in the crystalline solid may be closely related to the chemical environment
in the liquid melt. In each case, the extrapolated peak width was signiﬁcantly larger than
is observed for typical crystalline compounds but smaller than what is observed in a

typical amorphous glass.

EXPERIMENTAL SECTION
Reagents

Chemicals were used as obtained unless otherwise noted: phosphorous (MCB
Reagents, Gibbstown, NJ, amorphous red), selenium shot (T ellurex Inc., Traverse City,
MI, 99.999%), sulfur (Alfa Aesar, Ward Hill, MA, sublimed ﬂowers), copper (Fisher
Scientiﬁc, Hampton, NH, electrolytic dust), silver powder,7 bismuth (Tellurex Inc.,
Traverse City, MI, 99.999%), potassium (Aldrich Chemical Co. Inc., St. Louis, MO), and

thallium (ABCR, Karlsruhe, Germany, 99.999%). Selenium and bismuth were ground to

217

~100 mesh in an agate mortar and pestle. K2Se8»9 and P28e510 were produced as

described elsewhere.

Synthesis

The crystalline compounds K1.5Bi2.5(PS4)3,11 K2Cu2P4Se10,12 TlBiPZSe6,13 and
RbPSe614’15 were prepared according to their respective literature preparations.
ngPZSeé was produced by combining Hg (0.9356 g, 4.7 mmol), P (0.0627 g, 2.0
mmol), and Se (0.4728 g, 6.0 mmol) in a fused silica ampoule, then ﬂame sealing under a
reduced pressure of ~100 mTorr. This ampoule was loaded into a protective ceramic
sheath, and then allowed to react by heating to 450 °C over 12h and holding there for 24h,
then cooling over 6h back to room temperature.

A ~l:l mole ratio of HgZPZSe6 and RbPSe6 was formed by combining 17.7 mg
and 26.6 mg, respectively. This sample was heated in the NMR spectrometer as
described below. The Samples in quotation marks (i.e. “PSS”) represented mixtures that
were not intended to produce a compound with a given stoichiometry. These mixtures
were formed by combining the appropriate stoichiometric ratio of the reactants, sealing
them in a fused silica ampoule, and then placing them into the high-temperature probe of

the NMR spectrometer. Typical heating proﬁles for these samples are described below.

31P High Temperature NMR Spectroscopy
The high temperature 31P NMR spectra were collected on a 9.4 T spectrometer
(Varian Inﬁnity Plus) using a single resonance static probe (Doty Scientiﬁc). Sample

portions weighing ~75 mg were sealed in fused silica ampoules in vacuo and placed in a

218

protective fused silica sheath at the center of the coil in the probe. The samples were
heated to the desired temperature at a rate of 10 °C / min and allowed to equilibrate for 5
min before data collection. The 1t pulse width was typically 12-16 as and was calibrated
for each sample to within 0.2 us at each temperature where a spectrum was collected with
the transmitter set at the resonant frequency of the observed spectral peak. The NZ pulse
duration was calculated as half of the it pulse width. Longitudinal relaxation times were
measured using an inversion recovery pulse sequence (tt—t-7t/2—acquisition) and the data

were ﬁt to the equation16
I =(210)—Il-(1—e('71/7)) (1)
where I was the observed spectral intensity at pulse delay time 1', T1 was the ﬁtted

longitudinal relaxation time, and 10 and 11 were ﬁtting parameters that represented the

signal intensity at r =0and 1' :00. The spectra were processed with minimal line

broadening and baseline correction and referenced to 85% H3PO4 at 0 ppm. The total

time spent above room temperature was ~4 h. The spectra that were obtained upon the
ﬁrst heat/cool proﬁle were indistinguishable from those obtained in subsequent heat/cool
cycles.

The transverse relaxation time (T2) was measured using a Carr-Purcell, Meiboom-
Gill pulse sequence”,18 (rt/2 (r/2, 7r, 172),, r/2). The intensity of the resulting echoes, I,
was ﬁt to the equation
I = Joe-””2 (2)
where m is the time, and T2 and 10 are ﬁtting parameters representing the transverse

relaxation time and echo intensity at r = 0, respectively.

219

Chemical Shift and Peak Width Extrapolation
The change in chemical shift and peak width were ﬁt to a standard exponential

growth equation:
y = yo + Ber/t (3)

where the T was the absolute temperature, and y was either the chemical shift or the peak

width. The variables yo, B, and t were ﬁtting parameters. This empirical equation was

chosen because it provided a reasonable ﬁt and because it provides a means to extrapolate
the observed chemical shift to other temperatures. The chemical shifts and peak widths
in the absence of signiﬁcant contributions from the paramagnetic free radicals were
estimated by calculating the value of y at T = 298 K, which is approximately room
temperature. The room temperature NMR spectra of crystalline chalcophosphate
materials typically did not demonstrate signiﬁcant contributions from a radical species,
and the intention of choosing 298 K was to provide a means to compare the NMR spectra
of the crystalline materials to the high temperature NMR spectra. This ﬁtting can be
considered a “correction” to the chemical shift and peak width in the absence of
signiﬁcant effects of the free radical species. This model assumes that the chemical shift
and peak width vary continuously as a function of temperature and that there is no
discontinuity at the crystallization temperature. A discontinuity may result from the
species in the melt being fundamentally different from those present in the crystalline
solid. An interesting test of this assumption would be to collect high temperature NMR

spectra under magic angle spinning conditions slightly above and below the melting point

220

and to then compare the experimental chemical shifts to the values predicted by this

model.
Empirically, it was noticed that the parameter t in the ﬁts of the chemical shifts of

“KP1_5PSe6.3” (150.7) and “Cs4P2Se16” (127.4) differed by about 20%. The average

value of t for these two samples (139.1) was used for subsequent ﬁts of the data from the
remaining samples in which sufﬁcient data was not available to justify ﬁtting to a non-
linear equation with three ﬁtting parameters. In those cases, the value of t was ﬁxed to

139.1, and the parameters yo and B were ﬁt. There is presently no physical justiﬁcation

for ﬁxing this parameter and more data would be required to determine if the ﬁxing of the
t parameter is physically valid. One criticism of this approach is that the parameter t was
not constant in the ﬁts of the peak widths for samples with a larger number of data points.
For this reason, the peak widths of the samples with fewer data points were not ﬁt to an
exponential equation and therefore extrapolated peak widths for these samples were not

determined.

Previously, the change in the chemical shift for molten Se was described by:2

where %is the change in chemical shift in ppm units, cs is the number of radicals per

atom of Se, (A) is the average hyperﬁne coupling between the unpaired electron and the
NMR active nucleus, 7,, and 7,, are the gyromagnetic ratios of the electron and nucleus,
respectively, S is the spin quantum number, k3 is Boltzmann’s constant, Tis the
temperature in Kelvin, and ochemis the chemical shift without the effects of the radicals.

It was not possible to interpret the chalcophosphate data of the present work similarly

221

because magnetic susceptibility data were not available. These data are necessary to

determine (A) .

Estimation of the Thermodynamics of Radical Formation
Assuming that the paramagnetic radical species [P] is in equilibrium with a
diamagnetic species [D]:
[D] <—-> [P] (5)
The equilibrium constant has the form:

K = [P]/[D] (6)
Assuming a dilute concentration of the radical species, the concentration of the
diamagnetic species is approximately constant and therefore,

K = [P] (7)
The equilibrium constant K and the free energy change AG are related by:

K = e-AG/RT (8)
where R is the ideal gas constant, and T is the absolute temperature. The free energy
change can be expressed as:

AG = AH —TAS (9)
where AH andAS are the change in enthalpy and the change in entropy, respectively. By

combining (8) and (9), it is evident that:

 

_—AG_-AH+£
RT RT R

an (10)

222

The observed chemical shift in the NMR spectra is temperature dependent and it is
proposed that this change occurs because of a change in the concentration of the radical
species. The change in chemical shift, A00, is proportional to the equilibrium constant K.
K = n(A0') (11)

where n is an unknown constant. The change in chemical shift at a given temperature is
deﬁned as the difference between the observed chemical shift, and the chemical shift
extrapolated to T = 0 K. The radical species presumably collides with the NMR-active
nucleus. This changes the local ﬁeld at the nucleus and gives rise to the measured
increase in chemical shift.19 An increase in the number of radicals present increases the
statistical probability that collisions will occur and therefore with an increased radical
concentration, the enhancement to the chemical shift is increased. By combining (10)
and (11),

-AG —AH AS
= +—

ln[n(A0')] = RT RT

 

(12)

and a plot of the natural logarithm of the change in chemical shift versus the reciprocal of

 

the absolute temperature will have a y—intercept of (%-ln(n)) and a slope of

Though AS cannot be calculated because n is unknown, it is possible to estimate AH in

this manner.

RESULTS AND DISCUSSION
A summary of the 31F high temperature NMR data is displayed in Table 9-1. All

of the Se-containing melts displayed a strong dependence of chemical shift on

223

temperature. Over a ~200 °C temperature range, these compounds had an average of a
41.9 ppm chemical shift difference. It was not possible to properly phase the spectrum of

TlBiPZSe6 at each temperature point, and therefore the peak positions could not be

compared as a function of temperature for that compound.
The S-containing samples had a smaller dependence of chemical shift on

temperature, for example, the chemical shift of the sample with “PS5” stoichiometry

changed 17.6 ppm over the same 200 °C temperature range. The chemical shift of the

compound K1.SBi2.5(PS4)3 changed by 1.5 ppm over a 50 °C temperature range.

224

Table 9-1. The 31P chemical shifts, longitudinal relaxation times (T1), and line widths of various

chalcophosphate-containing materials as well as the ﬁnal product determined from powder X-ray
diffraction.

 

31}: T1 Line
(00a Shift 0 Width Final Product
(ppm)b (m5) (Hz)d
350 -132 11.6 1680
“KPSe7” 450 3.4 2.54 2220 n/a
550 30.6 0.47 2730
350 -46.5 8.45 2110
“K2PSe7.5” 450 -359 2.16 2540 n/a
550 -14.7 0.8 2740
350 8.9 7.63 1880
“KP1,3Se5_2” 450 35.3 1.92 4080 n/a
550 60.7 0.48 5616
350 32.6 9.2 1920
“KP1.5566.3” 450 47.9 1.5 2620 n/a
550 77.2 0.28 4040
350 -34.8 n.d 950
“C54P25e16” 450 -22.7 n.d 1320 CS41’23610
550 8.3 n.d 2361
350 80.4 114.5 4130
“PS5” 450 86.9 36.71 2530 n/a
550 98.0 9.08 5780
550 76 15.7 933
K1,5Bi2,5(PS4)3 575 77.5 11 926 K1.sBi2.5(PS4)3
600 77.5 7.94 929
500 93.3 4.84 2990
K2Cu2P4Se10 550 109.9 2.47 3660 K2€U2P45610
600 130.9 1.47 5130
525 n.d. 12.4 n.d.

T .
Stoichiometry emp

 

TlBiP2396 550 n.d 10.2 n.d. TlBi-steé
600 n.d 8.76 n.d.

KH21’04 350 -72 6.98 600 KH2P04

“KH2P04S” 350 -11.4 0.34 1200 KH2P04

 

a The uncertainty in temperature is estimated to be :1: 5 °C

b Chemical shifts were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported chemical shift was the peak center ﬁtting parameter. The uncertainty in the
chemical shift is estimated to be i 0.5 ppm.

c The method of measuring the longitudinal relaxation time is described in the text. The uncertainty is
estimated to be :1: 12 %.

d Peak widths were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported line width measurement was the full width at half maximum (FWHM). A 50%
Gaussian 50% Lorentzian line was chosen because it was in closer agreement with the experimental
spectrum than a Gaussian or a Lorentzian line alone. Room temperature linewidths were measured under
MAS conditions and high temperature linewidths were measured under static conditions. The estimated
uncertainty in the line width is :1: 5%.

225

Although there was some difference in the change in chemical shifts of the S and
Se-containing samples as a function of temperature, both types of sample had a strong

dependence of the longitudinal relaxation time T1. At 550 — 600 °C, all of the S and Se-
containing samples had 0.28 — 15.7 ms longitudinal relaxation times. T1 times of about
500 — 3000 s are commonly observed in crystalline chalcophosphate materials at room
temperatureﬁvwi20

Similarly, the 31P spectral line widths are temperature dependent. The Se-
containing samples had peak widths that increased by an average of ~12 Hz per °C. The

peak width in the “PS5” sample increased by ~8 Hz per °C, but the K1_SBi2.5(PS4)3

sample had no appreciable dependence of the line width on temperature.

Chemical Shift Extrapolation
The contribution of the free radicals to the chemical shift and peak width for the

sample with “KPL58e63” stoichiometry was modeled by ﬁtting the chemical shift and

peak width at each temperature to an exponential growth equation. A ﬁt of the chemical

shift and peak width of the “KPLSSe63” sample is displayed in Figure 9-1, and the raw

data are displayed in Table 9-2.

226

801 - 4500

:1 704 #4000

E

g 60- -3500 g

d:

3; :hi- ~3000 é?

'3 5'

.‘g’ 40. “2500 E

0 V

63 :M}- -2000 "
20' r1500

 

 

 

623 673 723 773 823
Temperature (K)

Figure 9-1. Fit of the experimental 31P NMR chemical shift and peak width for the sample “KP1.5Se6_3”.
The ﬁt values of yo, B and t were 16.9, 0.26, and 150.7 for the chemical shift, and 1238, 8.89, and 143.4 for

the peak width. The respective R2 values for these ﬁts were 99.953% and 99.625%. The chemical shift and
peak width at T = 298 K were 18.7 ppm and ~1300 Hz. The temperature of 298 K was chosen to provide a
means of comparison between the room temperature and high temperature NMR data.

The ﬁt suggested that at 298 K, the chemical shift and peak width would be 18.7
ppm, and ~1300 Hz, respectively. This chemical shift is within the range of what would

be expected for a P—P bonded species.6

227

Table 9-2. The 31P chemical shift and peak width of the “KPl .58e63” sample over the 623 — 823 K
temperature range.

 

 

Temp. 31p Shift T] Line Width
(°C)a (ppm)b (ms)c (Hl)d
623 32.6 9.20 1924
648 35.6 6.00 1989
673 39.7 3 .77 2259
698 43.4 2.80 2437
723 47.9 1.50 2619
748 53.5 1 .06 2907
773 59.5 0.76 3124
798 68.1 0.51 3540
823 77.2 0.28 4037

 

a The uncertainty in temperature is estimated to be i 5 °C

9 Chemical shifts were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported chemical shift was the peak center ﬁtting parameter. The uncertainty in the
chemical shift is estimated to be t 0.5 ppm.

c The method of measuring the longitudinal relaxation time is described in the text. The uncertainty is
estimated to be t 12 %.

‘3 Peak widths were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported line width measurement was the full width at half maximum (FWHM). A 50%
Gaussian 50% Lorentzian line was chosen because it was in closer agreement with the experimental
spectrum than a Gaussian or a Lorentzian line alone. Room temperature linewidths were measured under
MAS conditions and high temperature linewidths were measured under static conditions. The estimated
uncertainty in the line width is d: 5%.

A ﬁt of the chemical shift and peak width of the “Cs4P2Se16” sample indicated a -

44.2 ppm extrapolated chemical shift and a ~9OO Hz extrapolated peak width, Figure 9-2.
This stoichiometry is known to produce the crystalline compound Cs4PZSe10 upon
cooling, which has a -52.8 ppm chemical shift.10 The closeness of the extrapolated
chemical shift to the experimentally determined chemical shift of the crystalline
compound suggested that the 31P chemical environment of the solid and the melt are very

similar, and it is plausible that [P4Se10]4' anions or some other species without P-P

bonding exists in the molten liquid.

228

 

 

40- —6000
I 20- ~5000
g -e
E? ~4000 ii
0' 2
a ..-
”’ ~3000 E?
g -201 E
o "2000 O
6 40.
'1000
-60

 

623 673 723 773 823 873
Temperature (K)

Figure 9-2. Pit of the experimental 31? NMR chemical shift and peak width for the sample “CS4P2Se16”.
The ﬁt values of yo, B and t were -45.2, 0.08, and 127.4 for the chemical shift, and 907, 0.014, and 70.7 for

the peak width. The respective R2 values for these ﬁts were 99.971 % and 99.432%. The chemical shift and
peak width at T = 298 K were -44.4 ppm and ~900 Hz.

Table 9-3. The 31F chemical shift and peak width of the “CS4PZSe16” sample over the 623 - 873 K
temperature range.

 

 

Temp. 31P Shift Line Width
(K)a (ppnnb (thoc
623 -342 950
673 -295 1070
723 -22 1350
773 -9.5 1810
823 6.3 2360
873 31.6 4160

 

a The uncertainty in temperature is estimated to be :1: 5 °C

9 Chemical shifts were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported chemical shift was the peak center ﬁtting parameter. The uncertainty in the
chemical shift is estimated to be i 0.5 ppm.

c Peak widths were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported line width measurement was the full width at half maximum (FW HM). A 50%
Gaussian 50% Lorentzian line was chosen because it was in closer agreement with the experimental
spectrum than a Gaussian or a Lorentzian line alone. Room temperature linewidths were measured under
MAS conditions and high temperature linewidths were measured under static conditions. The estimated
uncertainty in the line width is i 5%.

229

The best ﬁts of the 31P chemical shifts of the “Cs4P2Se16” and “KP1.5Se6_3”

samples had similar ﬁtting parameters for t. These were the only two Se—containing
samples for which more than three 31P NMR spectra at different temperature points was
available. Because three parameters are simultaneously ﬁt for the exponential buildup to
correct for the presence of radicals, it was not reasonable to ﬁt the data from the samples
where only three different temperatures were used for NMR spectral collection. It was
therefore assumed that the parameter t could be ﬁxed for these samples. The value of t
that was chosen for subsequent ﬁts was 139.1 which was the average value obtained from

the ﬁts of the “Cs4P28e16” and “KPl_5Se6.3” samples. The parameters yo and B were

subsequently ﬁt to the available data.

One justiﬁcation for ﬁxing this parameter is that the right side of equation (4) may
be approximately constant for each of the selenophosphate samples at a given
temperature. The identity of the radical species may be the same in each case at a given
temperature, and there is some evidence that the enthalpy of radical formation is similar
for these compounds (vide infra). An additional empirical justiﬁcation is that using this
method produces an extrapolated chemical shift that is close to the experimentally

determined one for the compound K2Cu2P4Se10. The extrapolated chemical shift at T =
298 K was 59.3 ppm and the experimental chemical shift for this compound is 55.7
ppm.6 This lends further support to the possibility that the species that is present in the

crystalline solid (i.e. the [P4Se10]4' anion) is also present in the liquid melt.

230

633862 8:
$3 .82:ch 8.5.29: 6 8 E a 88205 .ozmzmzw 263 $52. 838888 95 .5 0:0 2:0 82:82 REEF—Soc :2: 203 £80828 0:: 35 8:865 22.: m

 

 

8:886: G. 20:: 303: 02:82:: 885:: 5:6: 82:822.: 6. :2: ..mme.
6.: 8:3: 6.: 6.: 6.: 2:2 2:92.: 3.26:: 26:22:59.
6.: 8.2:..- 6.: 6.: 6.: 2.9: 28:22.: 8.22- ..uwomﬂxiz
6.: 883:. 6.: 6.: 6.: 3:2 2:6: 2.: 3:268- inseam?
6.: 2 56.3. 6.: 6.: 6.: 2:2 2::5 .: 2.96:- .826...
825:: 8:13.. 8.85:: 38:22:: 888: 8.215 28:28.: 3.2651 ..628930:
8:58: 6. 2:: 8:22:62 2.2:: 8:322 8.35:2 28:28.: 62:62 .8825...
ANEV 353
5623 6.82 can E: 2 m o» 2 m 2: 32:62:28,:
68:28:32 6828632 8.52:": 56;: 6:8: was": :2: 82:20

 

28.—Evan 5 52m 0.8 moueﬁtoog 6.3228: 05. .ESoEEam 805 were: 005.582.
6— wam 3 cos—026.58 533 Mao:— 98 £2: 32526 cam—862x... 65 0:: 3:63: 8.95:: on: go some .20.. E20888 misc 05 we began .612 03:8

1
3
2

It has been shown previously that an increased amount of P relative to the metal and
chalcogen tends to produce compounds with lower valency P centers, i.e. P3+ and
P4+.7v15 The anions [PZSe6]4' and [P6Se12]4‘ both contain phosphorus in the 4+
oxidation state and have been produced with stoichiometries rich in P. Therefore, it
would be expected that the samples with the stoichiometry “KPLSSe63” (18.7 ppm
extrapolated chemical shift) and “KPL3Se52” (-1.0 ppm extrapolated chemical shift)

would be the most likely samples to have a P-P bond because these samples had the
highest relative concentration of P.
The chemical shift of 18.7 ppm for “KP1.58e6.3” is within the range of what

would be expected for a P-P bonded species, but the -1.0 ppm shift of “KP1.3Se5.2” was

on the border between the regions where P-P and non P-P bonded species are found. One
possibility is that there was rapid chemical exchange between a P—P bonded and a non P-
P bonded species although there is no evidence for such exchange.

The chemical shift and peak width for the sample with the “PS5” stoichiometry

were also ﬁt, and it was determined that the extrapolated chemical shift was 80.5 ppm,
and the extrapolated peak width was ~2400 Hz. The raw chemical shift, longitudinal
relaxation time and peak width data are displayed in Table 9-5. A room temperature 31F

MAS NMR spectrum of “P89” contained a dominant peak at 110 ppm that was about
5000 Hz wide.21 The extrapolated chemical shift for “P85” is ~30 ppm away from the

experimentally determined peak for “P89”. Either these two compositions result in

232

fundamentally different 31P coordination environments or the extrapolation of the

chemical shift is inaccurate for this sample.

Table 9-5. The 31P chemical shift and peak width of the “P85” sample over the 623 - 873 K temperature
range.

 

 

Temp. 3 1P Shift T1 Line Width
00“ (ppm)b (ms)c (Hl)d
623 80.4 1 14.5 4130
673 84.4 72.02 25 20
723 86.8 36.71 2530
773 91.9 17.06 3530
823 98.0 9.08 5780
873 112.6 3.61 12350

 

a The uncertainty in temperature is estimated to be :1: 5 °C.

9 Chemical shifts were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported chemical shift was the peak center ﬁtting parameter. The uncertainty in the
chemical shift is estimated to be :1: 0.5 ppm.

c The method of measuring the longitudinal relaxation time is described in the text. The uncertainty is
estimated to be i 12 %.

d Peak widths were determined by ﬁtting the spectral peaks to a 50% Gaussian 50% Lorentzian line
function and the reported line width measurement was the full width at half maximum (FWHM). A 50%
Gaussian 50% Lorentzian line was chosen because it was in closer agreement with the experimental
spectrum than a Gaussian or a Lorentzian line alone. Room temperature linewidths were measured under
MAS conditions and high temperature linewidths were measured under static conditions. The estimated
uncertainty in the line width is :1: 5%.

233

Peak Width Extrapolation
In two selenophosphate samples it was possible to compare the extrapolated peak
width to a known peak width of a crystalline solid. The peak width of the crystalline

compound CS4P25610, was about 100 Hz.6,10 Compared to the extrapolated peak width
of the melt of “Cs4PZSe16” (900 Hz), it appears that the broadness of the spectral line at

high temperature cannot be ascribed to the presence of the free radicals alone. One likely
contribution to the line width is some residual chemical shift anisotropy that is not fully
averaged out by the random tumbling of the 31P species that is present in the melt. The

magnitude of the chemical shift anisotropy of Cs4PZSe10 is 215 ppm at room

temperature, and some of this anisotropy may remain even at 600 °C.

The extrapolated peak width of the “P85” sample was narrower than a comparable
thiophosphate glass at room temperature, i.e. 2400 Hz versus 5000 Hz for “P89”.21 This

is contradictory to the observations of the molten selenophosphate compounds where the
extrapolated high temperature peak widths were still signiﬁcantly broader than their
room-temperature crystalline counterparts.

In a room temperature glass, there is a range of chemical environments. Unlike in
a crystal where all of the coordination geometries are in the most energetically favorable
conﬁguration, in a glass, many different coordination geometries are present with slightly
different bond lengths and angles. This lack of order distinguishes amorphous solids
from crystalline solids. These energetically disfavored geometries are trapped because
motion in the solid is restricted. The range of coordination geometries gives rise to the
broadness of NMR spectral features of glasses because each geometry has a

corresponding chemical shift. The chemical shifts are not resolved but instead are

234

overlapped into a broad spectral feature. The narrowness of the high temperature NMR
peak relative to the room temperature NMR peak of the amorphous glass suggests that in
the melt, motional averaging is increased such that contributions from the less
energetically favorable conﬁgurations are somewhat diminished and sharper lines
result.22 The restricted coordination geometries of the amorphous solid are free to relax
into more energetically favorable arrangements. This rearrangement occurs because of a
decrease in the viscosity of the melt relative to the amorphous solid resulting in an
increase in motional freedom.

For the sample “KPSe-f’, the transverse relaxation time (T2) was measured, and
was approximately 420 us, 200 us, and 190 us, respectively, at 350 °C, 450 °C, and 550
°C. In the T2 broadened limit,23 the predicted line widths would be 760 Hz, 1550 Hz, and
1700 Hz. The experimentally determined line widths were 1680 Hz, 2224 Hz, and 2734
Hz, respectively. The T2 broadening therefore only accounts for 45%, 70%, or 62% of

the linewidth, respectively. The additional line broadening is likely to be the residual
chemical shift anisotropy or some contributions from less energetically favored

coordination geometries.

Estimation of the Enthalpy of Formation of the Radical Species

The enthalpy of the formation of the radical species in the Se-containing samples
was approximately 34.6 kJ/mol. The corresponding enthalpy of formation of S-
containing radicals was approximately 48.1 kJ/mol. These data are displayed in Table 9-

6, and an example ﬁt using the “KPL5Se63” sample is displayed in Figure 9-3. The

calculated bond energies of Se-Se bonds in Se6 was reported to be 50 kJ/mol and the

235

bond energy of S-S bonds in S6 was reported to be 42 kJ/mol.24 Although there is some

debate about the true value of the chalcogen-chalcogen bond energy in the literature, the
similarity of the reported bond energies to the enthalpy calculated in the present work
provide some indication that the observed paramagnetic shift is correlated with radicals
that are formed upon chalcogen bond scission. If a biradical species was formed (i.e.

[D] H 2[P]) as is thought to be the case for the pure chalcogens,25v26 the calculated

values if the enthalpy change would be half of the values reported here. This can be
determined by substitution of the above equilibrium into equation (6) which would

become
K'=[P12/[Dl (13)
where K’ is the equilibrium between [D] <—) 2[P] . This could also be expressed as
K'= K2 (14)
Equation (10) with the biradical equilibrium would therefore be modiﬁed to

—AG_—AH+9§
RT RT R

21nK'=

 

(15)

By combining (15) and (11), and plotting the natural logarithm of the change in chemical

shift versus the reciprocal of the absolute temperature, the slope of the line is

 

This results in a change in the reported enthalpies of formation by a factor of 2. For

example, the enthalpy of “PS 5” would change to 96 kJ/mol.

236

Table 9-6. The enthalpy of radical formation.

uncertainties are given in parenthesis.

The 31? chemical shift was extrapolated to 0 K. The

 

 

 

31p Shift
AH
(ppm>T= M B 1
0K (Id/mo)
“KP1,5Se6,3” 17.2 -3446(91) 8.23(13) 30.2(8)
“CS4P28616” 45.1 4256(9) 9.121(11) 36.4(1)
“KPSe-f’ -25.6 -3872(30) 8.729(40) 32.2(2)
“K2PSe7.5” -56.2 -3705(26l) 8.19(37) 31(2)
“KPL3Se52” -23 -4481(670) 9.67(95) 37(6)
K2Cu2P4Sc10 58.3 4923(9) 9.926(11) 40.9(1)
“PS5” 80.5 -5783(300) 9.917(40) 48(3)
4.2-
4.0a
3.8-
3.6-
7.; 3.4-
s
s. 3.2+
3.04
2.87
2.6“

 

 

0.0012 0.0013 0.0014 0.0015 0.0016

1rr (K’l)

Figure 9-3. Plot of the data of the sample with the stoichiometry “KP1.5Se6.3” with the ordinate as the

reciprocal of temperature and the abscissa as the natural logarithm of the change in chemical shift. The ﬁt
had a slope of -3446 and a y-intercept of 8.228 which indicated an enthalpy change of 30.2 kJ/mol. The R

value was 99.743%.

237

CONCLUDING REMARKS

The nonlinear change of the chemical shift and peak width in chalcogenide-
containing melts has been modeled, and the chemical shift and peak width in the absence
of this effect has been calculated for several different samples. In the case of the

crystalline compounds K2Cu2P4Se10 and Cs4PZSe10, the extrapolated chemical shift

from the high temperature data was near the known chemical shifts of the room
temperature crystalline sample. Additionally, this chemical shift was correlated with the
known presence and absence, respectively, of a P-P bonded species. This methodology
will likely be useful in further characterization of chalcophosphate melts. Furthermore, it

may be possible to look for samples that contain multiple types of anion (i.e. [PSe4]3' and
[P2Se6]4') and after correcting for the effects of the free radicals, characterization of such

an equilibrium may be rationalized.

Additionally, the change in enthalpy that was derived from the NMR spectra
shows reasonable agreement with one set of the reported bond enthalpies for chalcogen-
chalcogen bonds. This provides support for the hypothesis that the mechanism that
produces the radical species is a homolytic cleavage of a chalcogen-chalcogen bond.

One other NMR observable that was not considered in the present work was the

extrapolation of the longitudinal relaxation time T1. In the case of the chemical shift and

peak width, the extrapolated values were always within about an order of magnitude of
the high temperature values. In the case of the relaxation times, several orders of
magnitude separate the room temperature samples which have relaxation times ranging
from several hundred to several thousand seconds and the high temperature samples

which have relaxation times that are frequently sub-millisecond. The failure of the

238

extrapolation of the relaxation rates to a comparable value of the room temperature
relaxation rates is likely because relaxation rates depend strongly on motional rates and
amplitudes. Presumably, the differences in the relaxation rates at high temperature and at
room temperature could be explained considering these differences in motion and also the

presence of the free radicals.

239

REFERENCES

1. Gave, M.; Kanatzidis, M. G.; Weliky, D. P. 2007, In preparation.

2. Warren, W. W.; Dupree, R. Phys. Rev. B. 1980, 22, 2257-2275.

3. Freeman, R.; Pachler, K. G. R.; Lamar, G. N. J. Chem. Phys. 1971, 55, 4586-4593.
4. Dechter, J. J .; Levy, G. C. J. Magn. Reson. 1980, 39, 207-215.

5. Bagley, B. G.; Waszczak, J. V.; Disalvo, F. J. Solid State Commun. 1972, 11, 89-92.
6. Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.

7. Gave, M.; Bilc, D.; Mahanti, S.; Breshears, J .; Kanatzidis, M. Inorg. Chem. 2005,
5293-5303.

8. McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 1257-1267.

9. Fehér, F. Handbuch der Pr'aparativen Anorganischen Chemie; Ferdinand Enke Verlag:
Stuttgart, Germany, 1954; Vol. 1, pp 280-281.

10. Gave, M.; Canlas, C.; Iyer, R. G.; Kanatzidis, M. G.; Weliky, D. P. J. Solid State
Chem. 2007, 180, 2877-2884.

11. Gave, M.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46, 11063-11074.
12. Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2098-2099.

13. Gave, M.; Malliakas, C. D.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46,
3632-3644.

14. Chung, 1.; Do, J .; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2004,
43, 2762-2764.

15. Chung, 1.; Karst, A. L.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2006, 45,
2785-2787.

16. Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

17. Carr, H. Y.; Purcell, E. M. Phys. Rev. 1954, 94, 630-638.
18. Meiboom, S.; Gill, D. Rev Sci Instrum 1958, 29, 688-691.

19. Prosser, R. S.; Luchette, P. A. J. Magn. Reson. 2004, 171, 225-232.

240

20. Canlas, C. G. Ph.D. Thesis, Michigan State University, East Lansing, MI, 2004.

21. Mutolo, P. F.; Witschas, M.; Regelsky, G.; Guenne, J. S. A. D.; Eckert, H. J. Non-
Cryst. Solids 1999, 257, 63-72.

22. Adamy, S. T.; Grandinetti, P. J .; Masuda, Y.; Campbell, D.; Jonas, J. J. Chem. Phys.
1991, 94, 3568-3576.

23. Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Longman Scientiﬁc and
Technical: New York, NY, 1983; pp 70-71.

24. Huggins, M. L. J. Am. Chem. Soc. 1953, 75, 4124-4126.
25. Koningsberger, D.; Van Wolput, J .; Rieter, P. Chem. Phys. Lett. 1971, 8, 145-147.

26. Sampath, P. I. J. Chem. Phys. 1966, 45, 3519-3525.

241

CHAPTER 10

THE SOLID SOLUTIONS Cu3PSe4_xSx AND FeszseGXSx.

242

INTRODUCTION

Solid solutions are materials where at least one crystallographic site has a random
substitution of one element for another. For example, after cooling a molten liquid
containing various proportions of copper and silver, the copper and silver randomly
occupy the positions in the crystal structure.1 In other cases, this substitution does not
occur and rather, the room temperature solid contains domains of each metal. Hume-
Rothery developed a set of empirical rules to determine if a given set of atoms will
randomly substitute for one another: (1) the atomic sizes of the atoms should be within
15% of one another, (2) the atoms must not have too great of an electronegativity
difference, and (3) the valency must be similar.1 Solid solutions can, in many cases, be
targeted by choosing species that adhere to this set of rules.

Solid solutions are industrially relevant because control of the stoichiometry in a
solid solution often allows for tuning Of various physical properties. Chalcogenide-based
solid solutions have been shown to possess such tunability, and a number of potential
applications of these materials exist. For example, there have been a number of
chalcogenide-based light emitting diodes (LEDS) that have been produced, and the color
of the light emitted from those diodes can be varied by changing the stoichiometry of the
solid solution. Slight changes in the composition of the solid solution can have a marked
effect on the band gap, and therefore the color of emission can be tailored.24 Solid
solutions are also relevant in data storage applications because the temperature of glass to
crystalline phase transitions can be tuned by modifying the composition of the solid
solution. A glass-to-crystalline phase transition is the basis for storing binary data on

optical storage media like CDs and DVDS, and the ability to tune the melt and

243

crystallization temperatures is beneﬁcial in the pursuit of next-generation data storage

materials.5'7

In this study, solid solutions of Cu3PSe4_xSx and Fe/ZPZSe6_xSx were formed. The
parent compounds Cu3PSe48 and Cu3PS49 were chosen because they are isomorphic and
FezPZSe610 and Fe2P2S6” were chosen because they are non-isomorphic, Table 10-1.

Additionally, the parent compounds can be synthesized in pure form without signiﬁcant

contributions from competing impurity phases.

Table 10-1. Literature crystallographic parameters for the parent compounds of the solid solutions.
Cu3PSe4a Cu3PS4b FezPZSe6C Fe2P2s6d

 

a A 7.685(2) 7.2817(11) 6.265(6) 5.947(1)

b A 6.656(1) 6.3387(10) 6.265(6) 10.300(1)

c A 6.377(1) 6.0746(10) 1980(2) 6.7222(8)

61 ° 90 90 90 90

B ° 90 90 90 107.16(1)

7° 90 90 120 90
Space group Pmn21 Pmn21 R-3H C2/m

 

a Ma, H. W.; Guo, G. C.; Zhou, G. W.; Wang, M. S.; Lin, S. H.; Dong, Z. C.;
Huang, J. S., Chinese J. Struc. Chem. 2002, 21, 288-291.

b Pﬁtzner, A.; Reiser, s., z Kristall. 2002, 217, 51-54.
c Klingen, W.; Eulenber.G; Hahn, H., Z. Anorg. Allg. Chem. 1973, 401, 97-112.
d Ouvrard, G.; Brec, R.; Rouxel, J., Mater. Res. Bull. 1985,20, 1181-1189.

It was hypothesized that the Cu3PSe4_,‘Sx compounds would form a solid solution
wherein the cell parameters would vary linearly as a function of the chalcogen content.
This hypothesis is justiﬁed on the basis of the Hume-Rothery rules as follows: (1) S2' has
a radius of 184 pm, and Se2' has a radius of 198 pm (7% difference)12 (2) the S and Se
electronegativity difference is small (2.5 and 2.4 Pauling units for S and Se,

respectively),12 and (3) S and Se both carry a 2- charge in these compounds. Therefore,

this system is predicted to form solid solutions. The linear dependence of the cell

244

parameters is predicted by Vegard’s law.13 This law states that the Size of the unit cell
depends only upon the radii of the constituent atoms. For this law to hold true, it is
necessary that the bonding properties of the constituent elements do not vary as a
function of the stoichiometry, a condition that might reasonably hold true under tenets (2)
and (3) of the Hume-Rothery rules.

In the case of the Fe2P2Se6_xSx compounds, however, the symmetry of the unit
cell changes from R-3H (hexagonal) for Fe2P28e6 to C2/m (monoclinic) for Fe2P286.

The electronegativity difference between S and Se is constant as well as the divalency

and so by comparison of the FezPZSem‘Sx compounds to the Cu3PSe4,xSx system, the

effect of the similarity of the parent unit cells in terms of cell symmetry could be
explored. In this case, Vegard’s law cannot hold because the cell parameters must not
vary linearly because the substitution of S for Se results in substantial changes to the unit

cell including a change in symmetry.

EXPERIMENTAL SECTION
Reagents

All reagents were used as received unless otherwise noted. Chemicals were used
as obtained unless otherwise noted: selenium Shot (99.999%, Tellurex Inc, Traverse City,
MD, sulfur (sublimed ﬂowers, Alfa Aesar, Ward Hill, MA), phosphorous (amorphous
red, MCB Reagents, Gibbstown, NJ), electrolytic copper dust (Fisher Scientiﬁc,
Pittsburgh, PA), iron ﬁlings (CCI, Columbus, WI). The selenium shot was ground in an

agate mortar and pestle to ~100 mesh.

245

Synthesis

The compounds Cu3PSe4_xSx and FezPZSengx were formed by combining the
appropriate stoichiometric ratio of the elements in a fused silica ampoule. The total
sample mass was held constant at 0.5 g. After sealing under a reduced pressure of ~100
mTorr, the samples were loaded into protective ceramic sheaths and then allowed to react
by heating to 950 °C over 12 h, holding there for 12 h, and then cooling back to room

temperature over 12 h.

Powder X-ray Diffraction

Powder X-ray diffraction was performed with a Rigaku-Denki RW400F2
(Rotaﬂex) diffractometer equipped with a Cu K0. rotating anode operating at 45 kV and
100 mA with a 2°/min scan rate. The resultant powder patterns were indexed using the
Jade 6.0 software package.14 The indexing was performed by overlaying the powder
pattern of the compound that was known to be an isomorphic analog, then reﬁning the

cell parameters to graft the overlayed peaks onto the experimental powder pattern.

Single Crystal X-ray Diffraction of Cu3PSeS3 and Fe2P28e284

Single crystals of Cu3PSeS3 and Fe2P28e284 were manually isolated from the
reaction ampoule and were mounted onto a glass ﬁber. Intensity data were collected on a
Bruker SMART platform CCD diffractometer using Mo Ka radiation operating at 40 kV
and 40 mA. A full sphere of data was collected and individual frames were acquired with
a 10 3 exposure time and a 03° omega rotation. The SMART software was used for data

collection, and SAINT software was used for data extraction and reduction. An

246

analytical absorption correction to the data was performed and direct methods were used
to solve and reﬁne the structures with the SHELXTL15 software package. The details of
data collection and reﬁnement are summarized in Table 10-2. Fractional atomic
coordinates are displayed in Tables 10-3 and 10-4, and selected bond lengths and angles
are displayed in Tables 10-5 and 10-6 respectively for Cu3PSeS3 and Feszse/ZS4. These
structures closely resemble Cu3PSe4 and Fe2P2S6, respectively, with the only major
change being the isomorphic substitution of the chalcogen. Detailed descriptions of the

parent structures can be found elsewherelo»11

Table 10-2. Summary of the details of the crystallographic reﬁnement of Cu3PSeS3 and Fe2P2Se2S4.

 

 

CU3PSCS3 F62P288284
Crystal system Orthorhombic Monoclinic
Space group Pmn21 C‘2/m
A, A (Mo Kn) 0.71073 0.71073
a, A 7.381(4) 6.031(1)
b, A 6.397(3) 10.458(2)
c, A 6.136(3) 6.816(2)
11, ° 90 90
13. ° 90 106.972(3)
7, ° 90 90
z 2 2
Crystal dimensions, mm 0.8 x 0.4 x 0.2 0.4 x 0.3 x 0.1
Dcalc g/ern3 4.429 3.638
11. min-1 17.098 12.857
Rim. % 2.5 5.2
total reﬂections/independent 3023/714 2219/496
ﬁnal mm”, %a 2.7%/6.2% 3.5n.6
Largest peak/hole, 372,3 O.423/-O.708 O.686/-O.911

 

a R = 2(IF0I - ch|)/Z|Fol. Rw = [2w(lFol - chl)2/}3wlF0l2]1/2

247

Table 10-3. Fractional atomic coordinates for Cu3PSeS3 with standard uncertanties in parenthesis.

 

 

x y z U(eq)a Occupancy
Cu(l) 0 0.6440(2) 0.0955(2) 22(1) 100%
Cu(2) 0.7460(1) 0.1816(1) 0.0870(1) 21(1) 100%

P 0 0.3238(4) 0.5961(6) 7(1) 100%

Se(1)|S(1) 0 -0.683l(4) 0.2553(4) 19(1) 75%/25%

Se(2)|S(2) 0 0.6320(3) 0.7176(4) 16(1) 75%/25%

Se(3)lS(3) 0.7369(2) 0.8338(2) 0.2109(2) 8(1) 75%/25%
a U(eq) = (ZiZjUijai*aj*ai-aj)/3 x 1000

Table 10—4. Fractional atomic coordinates for Fe2P2SeZS4 with standard uncertanties in parenthesis.

 

 

x y z U(eq)a Occupancy
Fe(l) 0 0.3326(1) 0 17(1) 100%
Se(l)IS(l) 0.2509(1) 0.1677(1) 0.2522(1) 17(1) 33%l66%
Se(2)|S(2) 0.7467(1) 0 0.2498(1) 18(1) 33%l66%
P 0.0555(2) 0 0.1673(2) 13(1) 100%

 

a U(eq) = (EiZjUijai*aj*ai-aj)/3 x 1000

248

Table 10-5. Selected bond lengths (A) and angles (°) for Cu3PSeS3. Standard uncertainties are displayed

in parenthesis.

 

Cul—SellSli 2.311(3) SeIISIi—Cul—Sezls2ii 113.20(11)
Cnl—Se2182ii 2.320(2) Serlsli—Cul—Se3ls3iii 109.46(6)
Cur—Se3ls3iii 2.3970(17) Sezlszii—Cur—Seals3iii 108.19(6)
Cul—Se3ls3iv 2.3970(17) Sellsri—cur—Se3ls3iii 109.46(6)
CuZ—SellSIV 2.3081(16) Se2132ii—Cul—Se3ls3iii 108.19(6)
Ctiz—Se3ls3vi 2.3131(19) Se11S1i—ciir—Se3ls3iv 109.46(6)
Cu2—stezlszvii 2.3149(17) Sezlszii—cm—Sesls3iv 108.19(6)
Caz—Se3ls3viii 2.3523(18) Se3ls3iii—cnl—Se3ls3iv 108.22(9)
Seusri—Cul—Se3ls3iv 109.46(6)
P—SellSli 2.092(5) Sezlszii—Cul—Se3IS3iV 108.19(6)
P—Se2ISZ 2.108(4) Se3ls3iii—CuI—Se3ls3iv 108.22(9)
P—Se3ls3ix 2.137(2) Sel ISlV—Cu2—SC3IS3Vi 1 14.69(7)
P—Se3IS3X 2.137(2) Sellsrv—Cuz—Sezlszvii 106.81(7)
Se3ls3vi—Cu2—Se2lszvii 1 14. 19(8)
Seuer—Cuz—Sezlszvii 106.81(7)
Se3IS3Vi—Cu2—SeZIS2Vﬁ 114.19(8)
Seuer—Cu2—Se3ls3viii 103.50(8)
Se3IS3VL—ct12—Se3Is3viii 106.48(5)
Sezlszvii—Cuz—Se3ls3viii 110.67(8)
Se11s1V—czuz—Se3ls3viii 103.50(8)
Se3ls3vi—ctiz—Se3ls3viii 106.48(5)
Sezlszvii—Cuz—Se3lssviii 110.67(8)
SelISli—P—SeZISZ 111.90(18)
Seusli—P—Se3ls3ix 108.6402)
Se2lsz—P—Se3ls3ix 108.9401)
Se3IS3ix—P—Se3IS3x 109.7704)

 

(i) x, 1+y, 2; (ii) x, y, -l+z; (iii) l-x, y, 2; (iv) -l+x, y, 2; (v) 1+x, 1+y, z; (vi) 1.5-x, l-y, -0.5+z; (vii) 0.5-x,
l-y, -0.5+z; (viii) x, -l+y, 2; (ix) 0.5-x, l-y, 0.5+z; (x) -0.5+x, l-y, 0.5+z; (xi) -1+x, -1+y, z; (xii) l-x, -1+y,
z; (xiii) x, y, 1+2; (xiv) 1.5-x, l-y, 0.5+z; (xv) 1+x, y, z.

249

Table 10-6. Selected bond lengths (A) and angles (°) for Fe2P28e284. Standard uncertainties are
displayed in parenthesis.

 

Fel—SellSl 2.5885(8) Sezlszi—Fel—Sezlszii 9488(4)
Fer—Sells1iii 2.5885(8) Se2lS2i—Fel—Se1l81 8434(2)
Fer—Seusrii 2.5924(7) Se2182ﬁ—Fel—SeIISIm 8434(2)
Fel—SellSliV 2.5924(7) Sensr—Fer—Serlsriii 96.44(3)
Fel—Se2182i 2.5881(8) Se2|SZi—Fe1—Se1|Slii 9597(3)
Fel—Se2lSZii 2.5881(8) Se2|s2ii—Fe1—Serlsrii 84.14(3)
P—SellSl 2.0982(8) SellSl—Fel—SellSlii 9574(2)
P—Sezlszvii 2.096504) Sellsriii—Fer—SeIISIii 8416(2)
P—Seusrviii 2.0983(8) 8e11s1—Fe1—Se11s1ii 9574(2)
P—Pix 2.181(2) Se21s2i—Fe1—Se11S1iv 84.14(3)
Se2|s2ii—1=e1—Se11s1iv 9597(3)
SellSl—Fel—SellSliV 8416(2)
Se2lS2Vﬁ—P—SeIIS1 113.410)
Sezlszvii—P—Se1131viii 113.41(4)
Se11s1—P—Serlsrviii 113.44(6)

 

(i) -0.5+x, 0.5+y, 2; (ii) 0.5-x, 0.5-y, -z; (iii) -x, y, -z; (iv) -0.5+x, 0.5-y, z; (v) 1+x, y, 2; (vi) 0.5+x, -0.5+y,
2; (vii) -l+x, y, 2; (viii) x, -y, 2; (ix) -x, -y, -z.
Raman Spectroscopy

Raman (750-100 cm'l) spectra were collected using a BIO-RAD FT Spectrometer
equipped with a Spectra-Physics Topaz T10-106c 1.064 nm YAG laser and a Ge
detector. Samples were ground into a ﬁne powder and loaded into fused silica tubes and

64 scans were averaged.

NMR Spectroscopy

Room temperature solid-state NMR spectra of the Cu3PSe4_xSx compounds were

collected on a 9.4 T spectrometer (Varian Inﬁnity Plus) using a double resonance magic
angle spinning (MAS) probe. Samples were spun at frequencies ranging from 5 kHz to
15 kHz in zirconia rotors of 4 mm outer diameter and with a maximum of a ~50 11L
sample volume. Bloch decay spectra were taken with a 3.6 us 90° pulse (calibrated using

85% H3PO4) and with relaxation delays of 1000 s. Each spectrum was processed with S

250

100 Hz line broadening and up to a 10th order polynomial baseline correction. The

spectra were referenced to 85% H3PO4 at 0 ppm.
For the samples Cu3PSe3S, Cu3PSe2S2 and Cu3PSeS3, it was not possible to spin

the samples with the full ~50 1.1L of sample with the probe inserted into the NMR magnet.
The samples would spin stably without the external magnetic ﬁeld present, however,
when the probe was subsequently inserted into the magnet spinning became erratic or
stopped altogether. To circumvent this problem, approximately equal volumes of the

Cu3PSe4_xSx sample and talc were ground together in an agate mortar and pestle, and this

mixture was loaded into the zircona rotor. This resulted in stable spinning for all
compositions. For these samples, spectra with lower signal-to-noise were collected

because of the decreased sample mass.

RESULTS AND DISCUSSION
X-ray Diffraction of Cu3PSe4.xSx
Compounds with the stoichiometry Cu3PSe4_xSx were formed and the powder

patterns were indexed following powder X-ray diffraction. The X-ray diffraction patterns
are displayed in Figure 10-1, the indexed cell parameters are displayed in Table 10-7, and
a plot of the cell parameters as a function of composition is displayed in Figure 10-2. In
each case, every peak in the powder X-ray diffraction pattern was indexed to a single
compound, and there was no evidence for the presence of, for example, mixtures of

Cu3PSe4 and CU3PS4 at the intermediate compositions.

251

 

 

CU3PSe83

f]. I I I A A I CU3P36282

11am - 55.333898

CU3PSC4

 

LLL Lima 3 L349
Lesa

 

 

 

 

 

 

10 20 30 40 50 60 70 80 90
Angle (20)

Figure 10-1. The X-ray diffraction patterns of the Cu3PSe4_xSx compounds. Each peak could be indexed
to the parent structure of either CU3PSe4 or Cu3PS4 and the cell parameters were subsequently reﬁned.

Table 10-7. The cell parameters of each of the Cu3PSe4_xSx compounds as indexed from the powder X-ray

diffraction data.
CU3PS4 Cu3PSe3S Cu3P862$2 CU3PSCS3 CU3PSC4

 

a (A) 7.280(5) 7.34400) 7.489(9) 7.592(3) 7.691(3)

b (A) 6.317(4) 6.401(4) 6.313(11) 6.589(3) 6.663(2)

c (A) 6.069(4) 6.127(3) 6.203(5) 6.286(3) 6.384(2)
Symmetry Ortho. Ortho. Ortho. Ortho. Ortho.

 

The cell parameters do not smoothly vary from one extreme to the other as
predicted by Vegard’s law, and this indicates the possibility that the nature of the bonding
in these compounds does not remain unperturbed as the switch from Se to S occurs.
Instead of a linear dependence, there is a minimum in the cell parameter b and cell

volume at the compound Cu3PSe282.

 

 

 

 

 

 

 

 

3401
A j V
o 300_ /
s - v/
3 280- v/
> .
E, 260::
~35 ‘ ____._.—I
: 7.5- I/l/.
a ‘ l
S 7.0-
; o '
a. 6'5 . O .xgé; 4
a 6.0- A ‘
U 4
5.5 I l l ﬁr I
4 3 2 1 0
Value of x in CU3PSe4PXSx

Figure 10-2. The cell parameters a (squares), b (circles), and c (upwards pointing triangles) of each of the
Cu3PSe4_xSx compounds, as well as the cell volume (downwards pointing triangles). The uncertainties

were smaller than the indicators used.

In a true solid solution, the types of atoms forming the solid solution are randomly
substituted for one another. One possibility that could explain the nonlinearity of the cell
parameters was that a true solid solution was not formed, but rather that there was some
site preference for the S and Se atoms. The deviation from linearity was most
pronounced in the cell parameter b, and therefore, the possibility existed that some
ordering of the S and Se atoms along the b-axis might reasonably result in an alternation

of bond lengths. To explore this possibility, a single crystal of Cu3PSeS3 was chosen and

X-ray diffraction data were collected, solved, and reﬁned. The crystallographic
reﬁnement contained no indication that a site preference existed for Se or S, and that each
chalcogen atomic site was had a 25% probability of being occupied by an Se atom, and a
75% probability of being occupied by an S atom. No long range order in the form of
satellite reﬂections were observed, and signiﬁcant contributions of a structural

modulation are therefore unlikely.

253

An additional possibility that was explored was that there was no long range
ordering of the chalcogen atoms, but rather only short range ordering. From the single
crystal X-ray diffraction data, the probability of each chalcogen site being an Se or an S

atom is ﬁxed at 25%, and 75%, respectively. Therefore, most of the [PQ4]3' (Q = S or
Se) moities would have the formula [PSeS3]3'. In contrast, it was hypothesized that the

formation of [PSe4]3‘ and [PS4]3' may be favored over the mixed species, and it was

these moities that were distributed in a 25% to 75% ratio. The crystallographic solution

was incapable of distinguishing between these two possibilities.

X-ray Diffraction of FezPZSe‘HSX
The powder X-ray diffraction patterns of the FeszZSe6_xSx compounds are
displayed in Figure 10-3. The parent members Fe/ZPZSe6 and Fe2P2S6 have hexagonal

and monoclinic cells, respectively“),11 The intermediate members that contained both S
and Se could be indexed based off of one of these cells. The results of this indexing are
displayed in Table 10-8. The members FeZPZSeSS FezPZSeZS4 and Fe2P28e3S3 were
indexed to a monoclinic cell, and the remaining FezPZSe4SZ and FezPZSeSS were indexed
with the hexagonal cell. For the S-rich compounds FeZPZSeSS Fe2P2Se254 all of the

peaks in the diffraction pattern could be indexed. In the case of the Se—rich compounds

Fe2P28e4S2 and FezPZSe5S there were some peaks that could not be indexed with the

hexagonal unit cell.
One possibility is that the Se-rich compounds were not pure but were rather

mixtures of phases with different unit cells. The extra peaks are putatively assigned to a

254

more S-rich compound that has monoclinic symmetry. This suggests that the S-rich
anions are formed preferentially. This may be the result of increased stability of the P—S
bond over the P-Se bond, and electronic structure calculations may verify this prediction.
For the peaks that were successfully indexed for each compound, the size of the unit cell
per formula unit (i.e. V/Z) there was a trend towards a larger unit cell with increased Se
content, as would be expected for the substitution of the larger Se for the smaller S, Table

10-6.

FCzP S6 '

     
     

FCszSCzS4

 

 

  

F62P2863S3
136213256482

F62P2365S .
l- ‘1“ L“

50 60

 

 

 

 

40
Angle (20)

Figure 10-3. The powder diffraction patterns of the Fe2P28e6,XSx compounds. The two strongest peaks
are off scale to show the smaller peaks at higher angle 20.

255

Table 10-8. The cell parameters of each of the Fe2P28e6_xSx compounds as indexed from the powder X-
ray diffraction data.
F62 F62 F62 F62 F62 F62 F62
P236 P23635 P236234 P236333 P236432 P23653 P2366
a (A) 599(1) 603(1) 6.039(7) 611(5) 610(3) 6.203(1) 6.268(3)
b (A) 1029(2) 1037(2) 10.454(8) 1050(2) 6.10(3) 6.203(1) 6.268(3)
c (A) 677(2) 681(1) 6.803(4) 683(2) 1952(6) 19.753(l) 1983(4)

 

a 90 90 9O 90 90 90 90
B 108.0(3) 108.0(3) 106.6(3) 107.3(4) 90 90 90
y 90 90 90 90 120 120 120
V (A3) 396.9 404.9 411.6 418.8 629.1 658.3 674.8
2 2 2 2 2 3 3 3
VIZ 198.45 202.45 205.8 209.4 209.7 219.4 224.9
Sym. Mono. Mono. Mono. Mono. Mono. Hex. Hex.

 

Like the single crystal reﬁnement of the compound Cu3PSeS3, the reﬁnement of a
single crystal of FeZPZSeZS4 contained no indication that there was a site preference for

one chalcogen atom over the other, and each of the sites had an equal probability of

containing each chalcogen atom.

Raman Spectroscopy
In order to determine whether or not the chalco gen atoms were randomly

distributed in the Cu3PSe4_xSx compounds, Raman spectroscopy was used to determine
the identity of the chalcophosphate moities present. It has been shown that [PS4]3' and
[PSe4]3' have totally symmetric stretching modes that are commonly at ~420 cm'1 and
~240 cm'l, respectively.16v17 For the mixed [PSe4_xSx]3' it would be expected that with

the loss of the tetrahedral symmetry, that the totally symmetric stretching modes would

be fundamentally different and therefore readily distinguishable.

256

The Raman spectra for the Cu3PSe4_xSx are displayed graphically in Figure 10—4,
and a list of the strongest peaks is displayed in Table 10-9. The features at 390 cm‘1 and
213 cm'1 in the Cu3PS4 and Cu3PSe4 spectra can be assigned to the [PS4]3' and [PSe4]3‘
totally symmetric stretch, respectively.16r17 In the case of the Cu3PSeS3 and CU3PSC3S
compounds, a spectral peaks at 384 cm'1 and 218 cm'l, respectively, are observed, and
these are likely attributable to the totally symmetric Stretch of the [PS4]3‘ and [PSe4]3‘

anions. The slight energy change in the totally symmetric stretch is likely affected by
nearby anions that are not homogeneously occupied by either an S or an Se. This change
in energy is further evidence that the sample is not a mixture of various species, but

rather a homogenous crystalline solid solution.

Cu PS4

Cu3PSeS

(31131336432
C113P363S
A 4‘ Cu3PSe4

100 150 200 250 300 350 400 450 500
Wavenumber (cm'l)

Intensity (Arb. Units)

 

 

 

Figure 10-4. Raman spectra of the Cu3PSe4.xSx compounds. The spectra were sums of 64 scans.
The peaks at 345 cm'1 and 262 cm'1 in the Cu3PSe3S and the Cu3PSeS3 compounds,

respectively, might reasonably be assigned to mixed chalcogen-containing anions. In the

case of the Cu3PSe3S compound this may be the [PSe3S]4‘ anion, and in the case of the

257

Cu3PSeS3 compound it may be the [PSe3S]4‘ anion. It is unlikely that signiﬁcant

contributions from the [PSe/ZSZP' anion is present in either case because the Raman

spectrum for the compound where such a feature would be expected to be the most

pronounced, namely Cu3PSe2S2, contains only very weak intensity signals. Instead, the

spectrum for this compound contains some features that closely resemble those in the

spectra of the other compounds.

Table 10-9. The energy in wavenumber(cm'1)of the strongest spectral features in the Raman spectra of

 

 

the Cu3PSe4_xSx compounds.
CU3PS4 CU3PSCS3 CU3P3633 CU3P364
390 384 218 213
345 262

 

Nuclear Magnetic Resonance
In order to conﬁrm the Raman assignments, NMR spectra were collected, Figure

10-5. The chemical shifts are displayed in Table 10-10. The parent compounds Cu3PS4
and Cu3PSe4 contained Sharp 31F spectral features at 55.9 ppm and -82.7 ppm,
respectively. The chemical Shift of Cu3PSe4 has been reported previously, and is well
within the range of what is expected for a compound containing a [PSe4]3'anion.18r19
The spectrum of Cu3PS4 contained an crystalline impurity phase with a chemical shift of
66.5 ppm. One possible assignment of this peak is to the compound Cu7PS6. The

compounds Cu3PS4 and Cu7P86 have been simultaneously produced previously.20

258

Table 10-10. The 31P NMR spectral features observed for the Cu3PSe4_xSx compounds, in ppm. The
peaks corresponding to the Cu7PS6 impurity have been omitted for clarity. The peaks common to different

stoichiometry compounds have been grouped horizontally.
C113PS4 C113P3633 CU3P36232 C113P363S CU3P364

 

55.9 55.9 -82.7 -82.7
26.6 28.0
-34.8 -40.8
-5.5 -3 -6

 

The spectra of the mixed S and Se compounds had much broader spectral lines.
This broadness is evidence for microscopic inhomogeniety as a result of having a random
distribution of the different anions rather than a perfectly ordered crystalline arrangement.

The spectrum of Cu3PSeS3 contained the same two spectral features that have been
assigned to the [PS4]3' anions of Cu3PS4 and Cu7P86. It also contained a spectral feature
at 26.6 ppm and one at -5.5 ppm. These spectral features are reasonably assigned to the
[PSe3S]3‘ and the [PSe2S2]3' anions, respectively. The intensity of the 55.9 ppm peak is
comparable to that of the 26.6 ppm peak, and both of these peaks are signiﬁcantly more
intense than the -5 .5 ppm peak. The relative intensities of these three peaks are
comparable to the respective peaks in the Raman spectrum of for this compound, with the

weakest NMR spectral feature likely corresponding to a Raman feature that can barely be

seen above the level of the noise. A weakness of this assignment is that if the

concentrations of the [PS4]3' and the [PSeS3]3' anions are approximately equivalent with
only a small amount of [PSe282]3', it would not be possible to account for the target
stoichiometry of CuPSeS3 but instead would indicate some more Se-rich stoichiometry.

This discrepancy is presently unexplained.

259

A similar argument for the assignment of the NMR spectrum of Cu3PSe3S can be
made. The -82.7 ppm is assigned to the P in the [PSe4]3' anion, and the -40.8 ppm peak
can reasonably be assigned to the [PSe3S]3' anion. A weak third spectral feature is
present at -6.0 ppm, which is assigned to the [PSe2S2]3‘ anion. Again the NMR spectral
features agree with the previously proposed Raman assignment. The Cu3PSe282

compound contains two strong spectral features at 28.0 ppm and -3.0 ppm, with a weaker

feature at -34.8 ppm. These have already been assigned to the [PSeS3]3‘, the [P362S2]3'

and the [PSe3S]3' anions.

CU3PS4

 

 

Cu PSeS3

CU3P36232

CU3PSC3S
) K CU3P364
150 100 50 0 -50 -100 -150
Chemical Shift (ppm)

 

Figure 10-5. The 31P NMR spectrum of the Cu3PSe4_xSx compounds. The spectra were sums of 4

individual scans with a pulse delay of 1000 s and the spinning frequency was 12 kHz, and Spinning
sidebands were not observed.

260

Observation of multiple anions within a single crystalline compound

It is somewhat surprising that the anion that is predicted by the stoichiometry of
the compound (i.e. [PSeS3]3‘ in the compound Cu3PSeS3) is not necessarily the dominant
ion, but instead the anion with only one type of chalcogen atom is also present in
signiﬁcant amounts. The compounds Cu3PSeS3 and Cu3PSe3S compounds contain
signiﬁcant numbers of [PSe4]3' and [PS4]3' ions, respectively.

The P-S bond has an enthalpy of 444 kJ/mol, and the P-Se bond has an enthalpy
of 364 kJ/mol.21 The formation of a P-S bond is therefore more exothermic than the
formation of a P-Se bond, and perhaps there is a relative difference in thermodynamic
stability of the anions containing only P-S and only P-Se bonds. In the melt, P—S and P-
Se bonds may break and form again in a dynamic equilibrium. Because of the increased
enthalpy of the P-S bond over the P-Se bond, perhaps the P-S bonds are not as readily
sundered as the P-Se bonds, and therefore S-rich anions are preferentially formed.
Support for this hypothesis is provided by the NMR spectra wherein the peaks that are
assigned to the more S-rich species are more intense than the Se-rich Species whenever

both S and Se are present. For example, in the spectrum of Cu3PSeS3 the peak that is
assigned to the [PS4]3' anion is more intense than the one for the [PSe/ZSZP' anion. Also,
in the NMR spectrum of Cu3PSe282 the peak assigned to [PSeS3]3' is more intense than
the peak assigned to [PSe3S]3'. One weakness of this hypothesis is that for Cu3PSe3S,
the intensity of the peaks assigned to [PSeZSZP' and [PSe4]3' are nearly equivalent. The

exact peak intensities are not trivially measured given the signiﬁcant nonlinearity of the

baseline and the relatively low signal to noise ratio. These preliminary data should be

261

recollected to ensure that the effects observed are not the result of deviations from the
nominal stoichiometry. Additionally, the summation of greater than four spectra for each
sample will be useful to increase the signal to noise ratio and make a measurement of the
relative peak intensities possible.

In each case where both S and Se are present, broad lines are observed whereas in
the S-only and Se-only compounds, the lines are substantially narrower. The broadness
of the NMR spectral features for each compound containing both S and Se provide some
evidence that the microscopic inhomogeniety plays a signiﬁcant role in the crystalline
structure for these compounds. It is hypothesized that in the compounds with both S and
Se, the different anions are randomly distributed, thus making a range of possible 31P

chemical environments. For example, with the compound Cu3PSe2S2, the [PSeS3]3',

[PSe2S213', and [PSe3S]3' are perhaps randomly distributed around a given P atom.

Because the distribution is random, a different P atom that is not proximate to the former
may have a different distribution of the different anions and therefore have a different
chemical shift. There are many possible arrangements of each of the different anions
near a given P atom within the crystalline solid, and this range of chemical environments
results in a range of chemical shifts. This is manifested in the NMR spectra as broad
lines because the different chemical shifts for each arrangement are not resolved and are
rather Signiﬁcantly overlapped. This provides evidence for the hypothesis that the

different anions are randomly distributed is correct.

262

CONCLUDING REMARKS

The compounds with the formula Cu3PSe4_xS appear to form solid solutions as

predicted by adherence to the Hume-Rothery rules.l Interestingly, a signiﬁcant number
of single chalcogen-containing chalcophosphate anions are present in the mixed solids.
That is, the stoichiometry Cu3PSe3S does not produce a crystalline solid that contains
only the [PSe3S]3' anion, but also contains a signiﬁcant amount of [PSe4]3'. The
Fe2P28e6_xSx compounds may also demonstrate a similar preference to form the single
chalcogen anions, and further Raman and NMR studies could be used to test this
hypothesis. If such a propensity to preferentially form anions containing only a single
type of chalcogen, it would be expected that the FeszSe6,xSx compounds would also
demonstrate a similar behavior. One interesting possibility is that in addition to the
single chalcogen atom anions being preferentially formed, that the [P2Se3S3]4' ion might
be preferentially formed. This possibility is justiﬁed because a single type of chalcogen

atom might reasonably be attached to each P-atom, i.e. [Se3P-PS3]4'.

263

REFERENCES

1

2.

9.

Hume-Rothery, W. J. Inst. Metals 1926, 35, 295-348.

Abe, S.; Masumoto, K. J. Cryst. Growth 2000, 220, 126-129.

. Karczewski, G.; Klirnkiewicz, M.; Glass, 1.; Szczerbakow, A.; Behrendt, R. Appl.

Phys. A-Mater. 1982, 29, 49-52.

Ramos, A.; Levelut, C.; Petiau, J .; Villain, F. J. Phys.-Condens. Mat. 1993, 5, 3507-
3518.

. Ahn, D. H.; Kang, D. H.; Cheong, B. K.; Kwon, H. S.; Kwon, M. H.; Lee, T. Y.;

Jeong, J. H.; Lee, T. S.; Kim, I. H.; Kim, K. B. IEEE Electr. Device L 2005, 26, 286-
288.

. Ahn, D. H.; Lee, T. Y.; Lee, D. B.; Yim, S. S.; Wi, J. S.; Jin, K. B.; Lee, M. H.; Kim,

K. B.; Kang, D. H.; Jeong, H. J .; Cheong, B. K. Jpn. J. Appl. Phys. 2007, 46, 5719-
5723.

. Wachter, J. B.; Chrissaﬁs, K.; Petkov, V.; Malliakas, C. D.; Bilc, D.; Kyratsi, T.;

Paraskevopoulos, K. M.; Mahanti, S. D.; Torbrugge, T.; Eckert, H.; Kanatzidis, M. G.
J. Solid State Chem. 2007, 180, 420-431.

. Ma, H. W.; Guo, G. C.; Zhou, G. W.; Wang, M. S.; Lin, S. H.; Dong, Z. C.; Huang, J.

S. Chinese J. Struc. Chem. 2002, 21, 288-291.

Pﬁtzner, A.; Reiser, S. Z. Kristall. 2002, 217, 51-54.

10. Klingen, W.; Eulenber.G; Hahn, H. Z. Anorg. Allg. Chem. 1973, 401, 97-112.

11. Ouvrard, G.; Brec, R.; Rouxel, J. Mater. Res. Bull. 1985, 20, 1181-1189.

12. Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements; 2nd ed.; Butterworth-

Heinemann: Boston, MA, 1997; pp 662, 754.

13. Vegard, L. z. Phys. 1921, 5, 17-26.

14. Jade 6.0, Materials Data, Inc. Livermore, CA.: 2003.

15. SMART, SAINT, SHELXTL V-5 and SADABS, Bruker Analytical X-ray Instruments,

Inc.: Madison, 1998.

16. Evenson, C. R.; Dorhout, P. K. Inorg. Chem. 2001, 40, 2884-2891.

17. Evenson, C. R.; Dorhout, P. K. Inorg. Chem 2001, 40, 2875-2883.

264

18. Canlas, C. G. Ph.D. Thesis, Michigan State University, East Lansing, MI, 2004.

19. Canlas, C. G.; Muthukumaran, R. B.; Kanatzidis, M. G.; Weliky, D. P. Solid State
Nucl. Magn. Reson. 2003, 24, 110-122.

20. Blachnik, R.; Gather, B.; Andrae, E. J. Therm. Anal. 1991, 37, 1289-1298.

21. Weast, R. C.; Astle, M. J ., CRC Handbook of Chemistry and Physics. 60th ed.; CRC
Press, Inc.: Boca Raton, 1980.

265

CHAPTER 11

CONCLUSIONS AND FUTURE DIRECTIONS.

266

Prior to this work, there were many known examples of alkali metal-containing
chalcophosphate materials,“21 and now there are a number of new materials that contain
pseudo alkali metals (i.e. Cu, Ag, and T 0.2233 These new compounds were found to
have potentially relevant physical properties like Spontaneous antiferroelectric ordering
and glass to crystalline phase transitions (Chapter 2) and also to possess a wide range of
chemical bonding motifs (Chapter 3). This also led to the discovery of several new alkali
metal containing compounds during attempts to rationalize the pseudo alkali metal
chemistry with the alkali metal chemistry (Chapter 5). Several new argyrodite-like
compounds that contain a pseudo alkali metal were also discovered, and these
compounds were found to have a composition-dependent phase transition (Chapter 6).
Temperature dependent phase transitions for related argyrodite materials have previously
been observed,24‘29 and variable temperature neutron diffraction experiments on the new
argyrodite materials of this work are presently underway to look for similar phase
transitions.

There are undoubtedly many other compounds that are yet to be discovered that
contain a pseudo alkali metal and various chalcophosphate anions. Most of the new
pseudo alkali metal containing chalcophosphate compounds presented herein also
contained bismuth. The choice of another element in lieu of bismuth may be selected to
target particular structtn'al or physical properties. For example, targeting phases with Sb
instead of Bi may more readily form vitreous products. Another possibility is to replace
the Bi atom with various 2+ metal cations such as Ti2+ and Sn2+. This may result in
compounds that have structures similar to the other pseudo alkali metal chalcophosphate

compounds with bismuth except with site vacancies to maintain charge neutrality.

267

An additional contribution that this work makes is some elucidation of the
properties of the molten mixtures that ultimately result in the formation of crystalline
compounds. In Chapter 7, it was proposed that a small molecular species such as
[P2Se10]4' may exist in the melt,30 and ﬁrrther support was given to this hypothesis in
Chapter 9. Only very general conclusions about the nature of the melt, for example the
viscosity, approximate molecular radius of the species present, and likelihood of
containing a P-P bonded versus non P-P bonded species were discussed.31,32 The
presence of a signiﬁcant concentration of a radical species was also demonstrated,
although the chemical identities of the species are not yet known.

One signiﬁcant limiting factor to the high temperature NMR experiments is that
in almost all cases only a Single peak was present at high temperature.30 The high
temperature NMR spectrum of Hg2P28e6 is the only example SO far where multiple peaks
have been observed once the sample is molten, Figure 11-1. There were peaks noted in
the chemical shift regions that have, at room temperature, been assigned to P-P and non
P-P bonded Species. No known [PSe4]3' ternary compound with Hg has yet been
discovered, and this high temperature Spectrum may provide evidence that the synthesis

of such a material may be possible.

268

 

 

 

300 200 100 0 -100 -200 -300
Chemical Shiﬁ (ppm)

Figure 11-1. High temperature 31P NMR spectrum of ngPZSe6 at 525 °C. The melting point of the

crystalline compound is 510 °C. The spectral features are presently unassigned and this is the ﬁrst known
example where speciation was observed in a molten sample.

New compound discovery with the high temperature NMR experiments has
already been successful for the compound CS4P2361030 (Chapter 7) and it is likely that

this technique will allow for the discovery of other new compounds as well. NMR
spectral collection at high temperature is possible in a matter of minutes (Chapter 8) and
therefore screening various temperature ranges to look for new compound formation is
potentially orders of magnitude faster than traditional methodology that combines high
temperature synthesis and subsequent characterization by diffraction.

Another limitation to the high temperature NMR spectra is that other than the
present works, there is no means by which to interpret the high temperature spectra of
these melts. In the case of the room temperature NMR experiments, the interpretation of
the resultant spectra is greatly assisted by known X-ray crystallographic reﬁnements of

the structure of the compound. In this manner, it is feasible to, in many cases, assign the

269

31P NMR Spectra to the known chemical environments of the crystalline solid. In a
molten liquid, it is not possible to obtain crystal structures so the interpretation of the
high temperature NMR spectra is substantially hindered. One type of method that may
provide signiﬁcant insights into the high temperature NMR Spectra is ab initio electronic
structure calculations.

The software package Gaussian 03 contains a routine to estimate NMR chemical
shifts of small molecules-°’3 and such a calculation may be useful to interpret the high
temperature NMR spectra. A ﬁrst step would be to validate the method for these types of
materials to ensure that the experimental chemical shifts correspond to the predicted
ones. In preliminary trials, attempts to calculate the chemical shifts of the crystalline
solids have failed. Calculations of NMR parameters are currently not implemented with
3D periodic structures like the chalcophosphate compounds of the present study, so it was
necessary to use a molecular approximant of the 3D structure. In all cases, these

calculations crashed and no chemical shift was predicted.

Calculations of the relative chemical shifts of isolated [PZSe6]4‘ and [PSe4]3'

anions (Q = S or Se) also did not provide useful information. Compounds containing
these two anions resonate at chemical shifts separated by perhaps hundreds of Wm, but
the calculation predicted a difference in chemical shift of only several ppm for the
isolated anions. Calculations performed on larger clusters that also contained nearby

metals were not successful in predicting chemical shifts that were comparable to those of

the crystalline compounds. Again, the diﬁerence in chemical Shifts between [P2Se6]4'
and [PSe4]3' containing Species was much smaller than what is experimentally observed.

Further work on improving the modeling techniques and the method of calculation may

270

circumvent these problems, and if this can be done, this technique will likely be very
useful to interpret the high temperature NMR spectra.

There is also great potential to characterize the chalcophosphate melts by other
means. An EPR heating apparatus was designed and fabricated, and temperatures of
~400 °C have been measured at the sample location in the NMR cavity. Unfortunately, to
date high temperature EPR spectra have contained noise but not Signal. The EPR spectra
of Chapter 8 were all collected on samples that had been rapidly quenched ﬁ'om the melt.
The high temperature NMR data have conclusively Shown that radicals are present in the
melt, and yet, they have not been observed directly with EPR measurements.

One possibility is that thermal noise is preventing the Observation of a Signal.
Another possibility is signiﬁcant electron-electron spin coupling. This may result in
Spectral features that are extremely broad, perhaps several thousand Gauss.34 Electron-
electron interaction is plausible in the molten mixture because of the high concentration
of radicals, and the putative mobility of those species in a relatively non-viscous liquid.
Furthermore, it has been proposed that the pure chalcogens form biradicals chains and an
additional possibility is that in the melt, these chains can align in such a way that the
radicals on each end are proximate. This is consistant with the formation of the various
cyclo- species that are known to exist in the different room temperature chalcogen

allotropes (i.e. cycIo-Sg and cyclo—Se6). An interesting test of this hypothesis would be to

monitor the EPR peak width from room temperature to the melt. Presumably, with an
increase in the number and mobility of the radical species, the EPR spectra will

increasingly broaden as a function of temperature. It may be possible to collect spectra

271

below the temperature where this electron-electron coupling makes the lines
unreasonably broad, if this hypothesis is correct.

Optical spectroscopy may also provide some useful information about the species
that exist at room temperature. Sulﬁrr radicals trapped in Zeolite cages have been
characterized using visible spectroscopy.35'37 It has been shown that the length of the S
chain is correlated with the color of the radical species. This may allow the identiﬁcation
of the structure of the radical if it is indeed dominated by chalcogen chains of various
lengths. High temperature Raman spectroscopy may also be useful in determining the
species that exist in the melt. The Raman features in the high temperature melts are
likely to be Signiﬁcantly broader than the room temperature crystalline counterparts.
Glasses and aqueous solutions that do not have long range order have been Shown to have
Raman features that are broader than those in well ordered crystals.38 Although the
peaks would be signiﬁcantly broadened by the loss of crystalline ordering, it may be
possible to observe whether or not these species persist in the melt Since various

crystalline chalcophospate compounds have distinct Raman features.

272

REFERENCES

1. Belkyal, 1.; El Azhari, M.; Wu, Y. D.; Bensch, W.; Hesse, K. F.; Depmeier, W. Solid
State Sci. 2006, 8, 59-63.

2. Chondroudis, K; Hanko, J. A.; Kanatzidis, M. G. Inorg. Chem. 1997, 36, 2623-2632.
3. Chondroudis, K; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 5401-5402.

4. Chondroudis, K; Kanatzidis, M. G.; Sayettat, J .; Jobic, S.; Brec, R. Inorg. Chem.
1997, 36, 5859-5868.

5. Chondroudis, K; McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1996, 35, 840-844.

6. Chung, 1.; Do, J.; Canlas, C. G.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2004,
43, 2762-2764.

7. Chung, 1.; Karst, A. L.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2006, 45,
2785-2787.

8. Coste, S.; Kopnin, B.; Evain, M.; Jobic, S.; Payen, C.; Brec, R. J. Solid State Chem.
2001, 162, 195-203.

9. Evenson, C. R.; Dorhout, P. K. Inorg. Chem. 2001, 40, 2884—2891.
10. Gauthier, G.; Evain, M.; Jobic, S.; Bree, R. Solid State Sci. 2002, 4, 1361-1366.
11. Gieck, C.; Tremel, W. Chem-Eur. J. 2002, 8, 2980-2987.

12. Iordanidis, L.; Bilc, D.; Mahanti, S. D.; Kanatzidis, M. G. J. Am. Chem. Soc. 2003,
125, 13741-13752.

13. Iordanidis, L.; Brazis, P. W.; Kyratsi, T.; Ireland, J.; Lane, M.; Kannewurf, C. R.;
Chen, W.; Dyck, J. S.; Uher, C.; Ghelani, N. A.; Hogan, T.; Kanatzidis, M. G. Chem.
Mot. 2001, 13, 622-633.

14. Iordanidis, L.; Schindler, J. L.; Kannewurf, C. R.; Kanatzidis, M. G. J. Solid State
Chem. 1999, 143, 151-162.

15. Kanatzidis, M. G. Chem. Mat. 1990, 2, 353-363.
16. Loken, S.; Tremel, W. Eur. J. Inorg. Chem. 1998, 283-289.
17. McCarthy, T. Ph.D. Thesis, Michigan State University, East Lansing, MI, 1994.

18. McCarthy, T.; Kanatzidis, M. G. J. Alloys Compd. 1996, 236, 70-85.

273

l9. McCarthy, T. J.; Kanatzidis, M. G. Chem. Mat. 1993, 5, 1061-1063.
20. McCarthy, T. J.; Kanatzidis, M. G. J. Chem. Soc. Chem. Commun. 1994, 1089-1090.
21. McCarthy, T. J.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 1257-1267.

22. Gave, M.; Bilc, D.; Mahanti, S.; Breshears, J.; Kanatzidis, M. Inorg. Chem. 2005,
5293-5303.

23. Gave, M.; Malliakas, C. D.; Weliky, D. P.; Kanatzidis, M. G. Inorg. Chem. 2007, 46,
3632-3644.

24. Evain, M.; Gaudin, E.; Boucher, F.; Petricek, V.; Taulelle, F. Acta. Crystallogr. B
1998, 54, 376-383.

25. Gaudin, B.; Boucher, F .; Petricek, V.; Taullele, F.; Evain, M. Acta. Crystallogr. B
2000, 56, 402-408.

26. Gaudin, E.; Petricek, V.; Boucher, F.; Taulelle, F.; Evain, M. Acta. Crystallogr. B
2000, 56, 972-979.

27. Kuhs, W. F.; Nitsche, R.; Scheunemann, K Mater. Res. Bull. 1979, 14, 241-248.
28. Onoda, M.; Chen, X. A.; Ishii, M.; Wada, H. Solid State Ionics 2000, 136, 415-418.
29. Onoda, M.; Wada, H.; Sato, A.; Ishii, M. J. Alloys Compd. 2004, 383, 113-117.

30. Gave, M.; Canlas, C.; Iyer, R. G.; Kanatzidis, M. G.; Weliky, D. P. J. Solid State
Chem. 2007, 180, 2877-2884.

31. Canlas, C. G. Ph.D. Thesis, Michigan State University, East Lansing, MI, 2004.

32. Canlas, C. G.; Kanatzidis, M. G.; Weliky, D. P. Inorg. Chem. 2003, 42, 3399-3405.

274

33.

34.

35.

36.

37.

38.

Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. B.; Robb, M. A.;
Cheeseman, J. R.; Montgomery, J ., J. A.; Vreven, T.; Kudin, K N.; Burant, J. C.;
Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.;
Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.;
Toyota, K; Fukuda, R; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao,
O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken,
V.; Adamo, C.; Jaranrillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, 0.; Austin, A.
J .; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K; Voth, G.
A.; Salvador, P.; Dannenberg, J. J .; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.;
Strain, M. C.; Farkas, O.; Malick, D. K; Rabuck, A. D.; Raghavachari, K; Foresman,
J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.;
Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, 1.; Martin, R L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03,
Revision C. 02, Gaussian, Inc., Wallingford CT, 2004.

Vanvleck, J. H. Phys. Rev. 1948, 74, 1168-1183.

Kowalak, S.; Jankowska, A.; Laczkowska, S. Stud Surf Sci Cata12005, 158, 215-
222.

Kowalak, S.; Jankowska, A.; Laczkowska, S. Recent Advances in the Science and
Technology of Zeolites and Related Materials, Pts a - C 2004, 154, 1633-1640.

Kowalak, S.; Jankowska, A.; Zeidler, S.; Wieckowski, A. B. J. Solid State Chem.
2007, 180, 1119-1124.

Kapitan, J.; Baumruk, V.; Kopecky, V.; Pohl, R; Bour, P. J. Am. Chem. Soc. 2006,
128, 13451-13462.

275

APPENDIX A

OPERATION AND MAINTENANCE OF THE HIGH TEMPERATURE
FURNACES.

276

Replacement heating elements can be purchased from Applied Test Systems
(hgpzllwwwatspacoy) for the furnaces with a serial number beginning with C or CH.
For the furnaces with serial numbers beginning with an H, the heating element is a single
strand of nicrome wire wrapped around a large diameter mullite tube. Replacement K-
type thermocouples can be ordered from Omega Engineering (http://www.omega.com/).
The Omega website also includes detailed user manuals for more advanced controls not
described here. Reaction ampoules should always be placed in a protective sheath prior
to insertion into the furnace. This prevents any reactants from leaking out and
contaminating the furnace when a reaction ampoule loses structural integrity. Mullite
tubes for this purpose can be ordered from CoorsTek (http://www.coorstel_<.com/).
Additionally, ﬁre brick can be cut and drilled by the glass shop.

When inserting or removing samples from the furnace, it is essential that care be
taken to avoid contact with the exposed heating element. Over time, the ceramic that is
used to embed the heating element coils can wear away leaving the heating element coil
exposed, and contact with this exposed wire could result in a shock or burn. Whenever
the furnace is above room temperature, the ends of the furnace should be plugged up with
an appropriate insulation material such as quartz wool to prevent excessive heat from
escaping.

The C and CH furnaces have heating elements that are rated to 1200 °C, although
operation at this temperature signiﬁcantly decreases the lifetime of the heating element.
Temperatures in excess of 1000 °C should only be used minimally if heating element life
is to be maximized. The H furnaces have homemade heating elements and there is

therefore no rating of a maximum temperature. However, temperatures exceeding 800 °C

277

have been avoided entirely and this temperature is a reasonable maximum for these
furnaces.

Programming of the Omega CN4800 controllers can be completed as follows:
(1) Press and hold the leftmost key (select) to enter into programming mode.
(2) Press the down arrow key until “ProG” is displayed, and check that the value for
this variable is set to “off". If it is not “off” press the rightmost key (return), then use the
up or down arrows to change to “off” then press the return key again.
(3) Using the down arrow, the different programming points can be accessed. The
deﬁnitions of these are as follows. “Sv 1” is the ﬁrst temperature to reach in °C, “rnlr”
is the amount of time to reach the temperature of “Sv 1” in hours, and “rn Is” is the
amount of time to stay at that temperature in hours. There are 4 different points that can
be set, i.e. “Sv 2”, “m2r” and “rn2s” follow.
(4) To change the values, go to the program point to be modiﬁed and then press the
return key. Each column of the display is changed separately. To input 550 °C into “Sv
2”, press the return key twice to select the hundreds column, and then press the up arrow
to change the hundreds place to a ﬁve. Press the return key again to select the tens
column, and then change it to a 5. Then press the return key two more times to return out
of editing mode.
(5) When the program has been successfully entered, press the up arrow several times

until the display reads “ProG” and change “off” to “run”.

278

APPENDIX B

OPERATION AND MAINTENANCE OF THE DOTY HIGH TEMPERATURE
NMR PROBE.

279

The Doty high temperature NMR probe and temperature controller should be
treated with utmost care because experience has shown that both are extremely fragile.
Typically, the samples sealed in fused silica ampoules are inserted into a protective fused
silica sheath that also functions to hold the sample at the center of the coil. The bell jar
and probe casing than then be replaced. The probe should be inserted into the NMR
magnet prior to making any connections.

Connections from the controller to the probe should be made as follows. The
order that the connections is made is not important as long as all connections are made
prior to probe heating.

(1) Ensure that the temperature controller is powered off, and connect the heater and
sensor leads.

(2) Connect the “drive” gas output from the MAS controller to the “drive” port on
the probe. The pressure of the gas should be manually set to about 40 psi. WARNING:
failure to provide sufﬁcient gas ﬂow could result in permanent damage to the heating
elements in the probe.

(3) The “upperstack gas” and frame cooling gas should be attached to the “electronic
and bell dewar” and “body tube” ports on the probe and then a ﬂow rate of 200 SFCH
should be set manually. The hot gas outlet should also be equipped with the insulated
outlet tube.

(4) The cooling stack for the Chemagnetics probes should be removed, and the
bearing gas hose should be placed a few inches down into the bore of the magnet. A ﬂow

of gas will help to keep the bore cool.

280

 

(5) The proton and X-channel RF cables should be attached.

(6) After opening the controller software on the computer, the power switch on the
controller can be safely switched on. The temperature of all three sensors should be near
room temperature.

(7) The heating rate should be set to 10 °C / minute by selecting “setup -) VT
controller”. Signiﬁcantly faster heating rates Should be used with extreme caution
because heating quickly will require substantially more power to be passed through the
heating elements, and their life spans will be shortened.

(8) The “heater on” button can then be pressed, and the target temperature set by
clicking in the “setpoint” box and then typing the desired temperature. After pressing the

return key, the temperature display should begin changing within several seconds.

Because of a fault, one of the safety interlocks for the probe has been intentionally
defeated. There is a 4th thermocouple that monitors the temperature of the gas that enters
the probe. For reasons that are unclear, the temperature being read at this thermocouple
was excessive even when the probe was at room temperature, and so this thermocouple
has been shorted. An electrical short on pins 12 and 13 of the sensor cable provides a
room temperature reading for this thermocouple, and allows for operation of the probe. It
is likely that a break in the thermocouple is causing this problem, and if this
thermocouple is ever replaced the short at pins 12 and 13 should be removed to restore

this safety feature. The building N2 is very unlikely to ever reach unacceptable

temperatures, and so this safety feature is not essential.

281

...J

 

APPENDIX C

OTHER SYNTHETIC ATTMPTS.

282

In addition to the new compounds reported herein, there were several other
systems that were explored in attempts to produce new compounds, Table A-1. The
results listed here were in most cases only attempted once, and the identiﬁcation was only
quickly screened by attempting to overlay known powder patterns with the experimental

one. The purpose of these results is to provide a starting point for some related systems.

Table A-1. Other synthetic attempts with the reactants, maximum temperature, cooling time, lab notebook
page number, and the major products as identiﬁed by X-ray diffraction. In cases where the powder
diffraction peaks had no matches, the product is listed as “Unknown.” When the powder pattern matched a
known diffraction pattern but with slightly shifted peaks, the product is given in italics.

 

 

Max Temp Cool Notebook
Reactants (°C) time (h) Page Products
Cu/Ga/P/Se 800 48 1MG85 0108582, Unknown
Cu/Sb/P/Se 800 48 1MG85 Unknown
Ag/Ga/P/Se 800 48 1MG92 AgGaSez
Ag/Sb/P/Se 800 48 1MG92 Agszse3
Zn/Hg/P/Se 700 91 1MG187 H82P25¢6r ZnP
Cd/Hg/P/Se 700 91 1MGl87 0121,2536
Ni/Hg/P/Se 700 91 lMGl87 H821’25136
Pb/Hg/P/Se 700 91 1M0187 Pb2P2S¢6
Cu/Mg/Bi/P/Se 800 24 1MG202 CuBisteor Bi2393. M82P23€6
Cu/Ca/Bi/P/Se 800 24 1M0202 CuBisteor Bi2593. Ca
Ag/Mg/Bi/P/Se 800 24 1MG202 AgBiP2Se6i M82P2366
Ag/Ca/Bi/P/Se 800 24 1M0202 AgBisteo Ca
Cu/Ce/P/Se 750 140 1M0209 C6362: CU3PSC4
Cu/Eu/P/Se 750 140 1M0209 Cu31’594, Unknown
Cu/Yb/P/Se 750 140 1M0209 Yb2393, Cu31’364. C1158
Ag/Ce/P/Se 750 140 1M0209 A8C6P2366
Ag/Eu/P/Se 750 140 1M0209 Ag4P2396 Unknown
Ag/Yb/P/Se 750 140 1MG209 Unknown
Ce/Cr/P/S 750 140 1140213 CcPS4. C‘2P2396
Ce/Mn/P/S 750 140 1MG213 CePS4i M1121’2396
Ce/Fe/P/S 750 140 1MG213 CcPS4. Fe21’2598
Ce/Co/P/S 750 140 11110213 CePS4. Unknown
Cu/Mg/P/Se 800 48 1MG220 CU3PSC4. M82P2586 Unknown
Cu/Sm/P/Se 750 140 1M0223 51115618, Cu31’594
Cu/Fe/P/Se 800 48 1M0224 Cu3PS°4r Fe21’2'566 Se
Ba/Bi/P/S 950 48 1M0242 Unknown
Ba/Bi/P/S/BaBr2 950 48 1M0242 BaBrzr BiPS4 Unknown
BaS/Fe2P286 1050 12 11140278 Fe21’236’ F65. Unknown
BaS/C02P256 1050 12 1MG278 Unknown

 

283

In several cases, there were no powder patterns that could be found to match the
peaks in the experimental powder pattern, and in these cases, the products have been
listed in table A-l as “Unknown.” In several cases, the peaks in the experimental powder
pattern were closely related to a known compound, except there was some shift in the
peaks. This may be evidence for a new compound with a structure that is closely related
but with a different unit cell size. These products have been listed in the table as italics.
There were also several cases where the resultant crystalline products matched with
known crystalline products but that not all of the reactant elements could be accounted
for. One possibility is that the sample chosen for the X-ray diffraction experiments was
not representative of the bulk, and the other possibility is that the compound that was
produced was isostructural to the known compound, but contained all of the reactant
elements. In these cases The “Unknown” and italicized products are the ones that most
likely contain as-yet unidentiﬁed crystalline products and merit further consideration. It
is also worth considering the cases where some of the reactant elements could not be

accounted for to look for isostructural compounds with different sets of elements.

284

APPENDIX D

FEE LOCATIONS OF SPECTRA

285

Figure 3-16. 31P NMR spectra of (A) a-TlBiPZSe6 [I], (B) B-TlBiP28e6 [II], (C)
TIBiPise [III]. (D) TlgBi3(PS4)4 [IV]. (E)T1BiP287 [V]. (F) TlgBi(PS4)2 [VI]. (G)
T14Bi2(PS4)2(PZS6), and (H) KBiP2S7. Each spectrum was an average of four scans with

a delay between pulses of 5000 s. Each spectrum was processed with S 100 Hz line
broadening. Peaks and their corresponding spinning sidebands are grouped with like
symbols .............................................................................................................................. 86

(A) lhome/gavematt4b/data/aT1BiP28e6/T1buildup_052706
(B) /home/gavematt4b/datafb'1‘lBiP2Se6/T1buildup_061 106
(C) /home/gavematt4b/data/’I‘lBiP286/T1buildup_072706
(D) lhome/gavematt4b/data/I'13Bi3P4S 16/T1buildup_052006
(E) /home/gavematt4b/data/T13Bi3P2S8/T1buildup_052406
(F) /home/gavematt4b/datafI‘lBiPZS7/T 1buildup_052006

(G) /home/gavematt4b/data/'l‘l2BiP2S7/T1buildup_052106
(H) /home/gavematt4c/data/KBiP287/52x_1000s

Figure 4-2. 31P NMR spectra of Tl4Bi2(PS4)2(P2S6), Tl3Bi(PS4)2, and TlBiP2S7 obtained
at 9.4 T and 21 T. The experimental spectra are represented by solid lines and the best-ﬁt
Lorentzian functions used to deconvolve the spectra are represented by dashed lines. The
sums of the Lorentzians are also shown as dashed lines and ﬁtted well to the
corresponding experimental spectra. ............................................................................... 100

(A) lhome/gavematt4b/data/aTlBiP2Se6/T1buildup_052706
(B) /home/gavematt4b/data/leBiP2Se6/T1buildup_061106
(C) lhome/gavematt4b/datafl'lBiP2S6/I‘ 1buildup_072706
(D) lhome/gavematt4b/data/I'13Bi3P4S16/T1buildup_052006
(E) /home/gavematt4b/data/'I‘l3Bi3PZS8/T1buildup_052406
(F) /home/gavematt4b/data/'1‘lBiP287/T 1buildup_05 2006

(G) /home/gavematt4b/data/'I'l2BiP2S7/T 1buildup_052 106
(H) /home/gavematt4c/data/KBiP287/52x_1000s

Figure 5-11. 31P NMR spectra of (A) K1_5Bi2_5(PS4)3 and (B) K9Bi(PS4)4 with isotropic
peaks and their spinning sidebands grouped under like symbols .................................... 141

(A) /home/gavematt4c/data/K15Bi25/'l‘ 1buildup_061906
(B) /home/gavematt4c/data/K9BiP4S l6_T1buildup_061606_2

286

Figure 7-1. Ambient temperature 31P magic angle spinning NMR spectra of the ﬁnal
products of: (A) a “CS4P2Se16” reaction stoichiometry indicating Cs4P2Se10 as the
dominant 31P-containing species at -52.8 ppm; (B) a “CS4P2368” reaction stoichiometry
indicating CS4P2Se9 as the dominant 31P-containing species at -39.9 ppm; and (C) a
“CS4P2Se12” reaction stoichiometry that produced CS4P2Se9 and Cs4P2Se10 in a ~3:5
ratio. Panel D displays the static 31P NMR Spectrum of a 350 0C melt of the products in
C. The peak with a chemical shift of -22.0 ppm in panels B and C was not assigned and
the peak at -99.8 ppm in panel B was not assigned. For spectra A-C, the spinning
frequency was 14 kHz and spinning sidebands are grouped with the corresponding
isotropic peak using like symbols with * .=. CS4P23610, + E Cs4P2Se9, and o E unassigned
peak with -22.0 ppm isotropic shift ................................................................................. 173

(A) /home/gavematt4c/data/Cs4P2Sex_Fast/spec_121306

(B) /home/gavematt4c/data/Cs4P28e9/spec_l21206

(C) lhome/gavematt4c/data/CS4P2Sex/Cs4P28e12/spec4x_022507

(D) lhome/gavematt4c/data/HT/Cs4P2Sex/Cs4P28e 12/spec4k_350C/022807

Figure 7-4. The room temperature 31P NMR spectrum of a mixture with “CS4P2Se16”
stoichiometry that was cooled over 24 h to produce crystalline CS4P23610 and Cs4P28e9 .
The MAS frequency was 7 kHz. The intensities of peaks corresponding to CS4P23610
were used for the determination of CSA principal values of this compound .................. 185

/home/gavematt4c/Cs4P28ex/6_6_24/csa4x__0l 1907

Figure 8-1. NMR spectra of a melt mixture with the nominal ratio “P85” showing the

increase in chemical shift and peak width as a function of temperature. Spectra at 450,

500, 550, and 600 0C were acquired with acquisition delays of 2 s and represent the sums
of 16 scans ........................................................................................................................ 200

/home/gavematt4c/data/HT/PSS/tlinv_500C_052607
/home/gavematt4c/data/HT/P85/tlinv_550C_052607
/home/gavematt4c/data/HT/PSS/t1inv_600C_052607

Figure 8-3. High temperature 31P NMR spectra of (A) K2Cu2P4Se10 at 550 °C and (B) a
sample with the nominal composition “KPSe-f’ at 350 °C. The spectra were acquired

with acquisition delays of 10 ms and 200 ms, respectively and represent the sums of 50
scans ................................................................................................................................. 203

(A) lhome/gavematt4b/data/HT/K2Cu2P4Se10/T1inv_550_110406
(B) lhome/gavematt4c/data/HT/l 18/T 1buildup_350_092506

287

Figure 8-4. EPR spectra at room temperature of rapidly quenched “P85” samples. Prior

to quenching, the sample temperature in the furnace was (A) 1000 0C or (B) 550 °C.
Acquisition parameters were: (A) 10 G modulation amplitude, 2000 G sweep width and
(B) 0.1 G modulation amplitude, 50 G sweep width ....................................................... 208

(A) jlm-e680/home/mattg/071907_PS5 1000_3.spc
(B) jlm—e680/home/mattg/071907_PS5550_4.spc

Figure 8-5. Room temperature EPR of (A) “Sn85” quenched from 550 °C (B) “PSCS”
quenched from 550 °C and (C) K1.5Bi2.5(PS4)3 quenched from 1000 °C. Acquisition

paramaters were: (A) l G modulation amplitude, 40 G sweep width; (B) 1 G modulation
amplitude, 500 G sweep width; and (C) 10 G modulation amplitude, 3000 G sweep
width. Each spectrum was a single acquisition ............................................................... 208

(A) /jlm-e680/home/mattg/061 807_SnS5_945 1 .spc
(B) /jlm—e680/home/mattg/061807_PSe5_94485.spc
(C) /jlm-e680/home/mattg/un15Bi25_295mg.spc

288

 

 

Table 9-2. The 31P cherrrical shifts, longitudinal relaxation times (T1), and line widths of
various chalcophosphate-containing materials as well as the ﬁnal product determined

 

 

from powder X-ray diffraction ......................................................................................... 227
Stoichiometry TEE]? File Location
350 gavematt4c/data/HT/118fI‘1inv_350_092506
“@367" 450 gavematt4c/data/l-IT/l l 8/T l inv_450_092506
550 gavematt4c/data/HT/ l 18/T l inv_550_092506
350 gavematt4c/data/HT /21 8/T 1inv_350_092706
“K2PS€7,5” 450 gavematt4c/data/HT/218/I‘linv_450_092706
550 gavematt4c/data/HT/2l8/1‘1inv_550_092706
350 gavematt4b/data/HT/348/I‘1inv_350_110106
“KP1,3Ses,2” 450 gavematt4b/data/l-lT/348/Tlinv_450__l10106
550 gavematt4b/data/HT/348fl‘ 1 inv_550_1 10106
350 gavematt4b/data/HT/23 8/T 1inv_350_1 10306
“KP1 , 5 566,3" 450 gavematt4b/data/HT/238/T l inv_450_1 10306
550 gavematt4b/data/HT/23 8/T 1 inv_550_1 10306
350 gavematt4c/data/HT/2synth_Cs4P28ex/spec3k_cool350C_121606
“CS4P236 1 6” 450 gavematt4c/datalHT/2synth_Cs4P2Sex/Spec3k_cool450C_l 21606
550 gavematt4c/data/l-IT/2synth_Cs4P28ex/spec3k_c001550C_l21606
350 gavematt4c/data/HT /PSS/T1inv_350_052607
“P55" 450 gavematt4c/data/HT/PS5/I‘ l inv_450_05 2607
550 gavematt4c/data/HT/PSS/T 1 inv_550_05 2607
550 gavematt4c/data/HT/K15Bi25/T 1 inv_550_092606
K1 , 5152.5 (PS4)3 575 gavematt4c/data/HT/K1 5Bi25/T1 inv_575_092606
600 gavematt4c/data/HT/K15Bi25/T linv_600_092606
500 gavematt4b/data/HT/K2Cu2P4Se10/I‘linv_500_110406
K2CU2P43810 550 gavematt4b/data/HT/K2Cu2P4Se10/I‘linv_550_1 10406
600 gavematt4b/data/1-l'T/K20u2P4Se10/I‘linv_600_110406
525 gavematt4c/data/HT/T1B iP2Se6/Tl inv_5 25_092606
TlBiP2566 550 gavematt4c/datalHT/I‘lBiP23e6/I‘ linv_550_092606
600 gavematt4c/data/HT fI‘lBiP2Se6/I‘ 1inv_600_092606
KH2P04 350 gavematt4c/data/HT/KH2PO4II‘1buildup_350_092806
K1-12P04/ S 350 gavematt4c/data/HT/KHZPO4_S/I‘1inv_350_093006

 

Figure 10-5. The 31P NMR spectrum of the Cu3PSe4_xSx compounds. The spectra were

sums of 4 individual scans with a pulse delay pulse delay of 1000 s and the spinning
frequency was 12 kHz, and spinning sidebands were not observed ................................ 260

/home/gavematt4c/data/Carly/Cu3PSe4/spec4x_030907
lhome/gavematt4c/data/Carly/Cu3PSe3S/spec4x_030807
/home/gavematt4c/data/Carly/Cu3PS2Se2/spec4x_030807
/home/gavematt4c/data/Carly/Cu3PSeS3/spec4x_030807
/home/gavematt4c/data/Carly/Cu3PS4/spec4x_030807

289

 

Figure 11-1. High temperature 31P NMR spectrum of Hg2P2Se6 at 525 °C. The melting

point of the crystalline compound is 510 °C. The spectral features are presently
unassigned and this is the ﬁrst known example where speciation was observed in a
molten sample. ................................................................................................................. 269

lhome/gavematt4c/data/HT/Hg2P28e6/specl6x_525C_052507

290