3%

. 1.3.4.751

 

ham-Ln

. y. ‘i x. .,

x. 24mm. . .,
:. . . . mewvfij.

,“fifi..fia.. . .

74¢: .

:x}... .
. u,
«33".... £9

2. at? , ,.

aim.” _ ilfl. . rfi .: ..
I o z 3 ,
.wwmez A, a

ma. .\.W_.m.mfl.uab .fim %u;nfi.r#

1,: e: ‘, .1 .... . x
a i/ .bflm fimwadWWWN.

MA 3
{AW}!!! .
. ...,........;..kr
9.4%.?wa .. ‘ a.
. . ..
fi$§ fits»... 4...... ,.
mm. {53%
33?.
:5"

' {a
, , ”3.5:. ..
3.525%» . n... Mn... .
innuiuun. Vaufififih?

23.1

,5. 3!.
z .1. . .
. 13:12
13.331}.\):.f$f

4A.

A. r .l4.2pi ‘ .uvv

, In. v.5ii...6\

tlx

.; . ‘
«a»... thanks“. 43..“
: z.

3%.!!! 1..

». 5.:va

.. 3.. A
‘ .L D, t.‘
.) «W95»? .nfizfi...‘ .
‘ . . Q -
.f{. x a I}
.11. 2...? 3

‘fi

ob?

.

716:3?! 3L...»- 4..
o. .. ll

‘ LIBRARIES
7 MICHIGAN STATE UNIVERSITY
4 ‘ EAST LANSING, MICH 48824-1048

290:;

39/74/415

This is to certify that the
dissertation entitled

ROOM TEMPERATURE AND HIGH TEMPERATURE
31F SOLID-STATE NUCLEAR MAGNETIC RESONANCE
STUDIES OF METAL SELENOPHOSPHATE AND K/P/Se

FLUX REACTIONS

presented by
CHRISTIAN G. CANLAS

has been accepted towards fulfillment
of the requirements for the

PhD. degree in CHEMISTRY

9M; Mai

Major Professor’s Signature

I’L'jjI/okg

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClRC/DateDuepBS-p. 15

 

ROOM TEMPERATURE AND HIGH TEMPERATURE
31F SOLID-STATE NUCLEAR MAGNETIC RESONANCE STUDIES
OF METAL SELENOPHOSPHATE AND K/P/Se FLUX REACTIONS

By

Christian G. Canlas

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Chemistry

2004

ABSTRACT

ROOM TEMPERATURE AND HIGH TEMPERATURE
31F SOLID-STATE NUCLEAR MAGNETIC RESONANCE STUDIES
OF METAL SELENOPHOSPHATE AND K/P/Se FLUX REACTIONS

By
Christian G. Canlas

31F ambient and high temperature solid-state nuclear magnetic resonance (NMR)
spectroscopy was applied to understand the chemistry of metal selenophosphate
syntheses and potassium selenophosphate (KZSe/PZSe5/Se) flux reactions of varying
compositions. In order to understand what was happening in the melt at high
temperatures the first step was to take NMR spectra of known compounds at room
temperature and find correlations between NMR spectra and structure.

31F room temperature solid state nuclear magnetic resonance (NMR) spectra of
twelve metal-containing selenophosphates have been examined to distinguish between
the [P28e6]4‘, [PSe4]3', [P4Se10]4', [P28e7]4', and [P28e914- anions. There is a general
correlation between the chemical Shifts (CSs) of anions and the presence of a P—P bond.
The [PZSe6]4— and [P4Se10]4_ anions both contain a P—P bond and resonate between 15
and 95 ppm whereas the [PSe4]3-, [P28e7]4', and [P28e9]4" anions do not contain a P—P
bond and resonate between —1 15 and —30 ppm. The chemical Shift anisotropies (CSAS) of
compounds containing [PSe4]3" anions are less than 80 ppm which is Significantly
smaller than the CSA of any of the other anions (range: 135 — 275 ppm). The smaller
CSAS of [PSe4]3' anions are likely due to its unique local tetrahedral symmetry. Spin-

lattice relaxation times (T1) have been determined for the solid compounds and vary

between 20 and 4000 5. Unlike the CS, T1 does not appear to correlate with P—P bonding.

There is a qualitative positive correlation between squared dipolar couplings and 1/1“ 1,
suggesting that these interactions contribute to relaxation along with the presence of
unpaired electrons, which may accelerate the rate of 31F relaxation.

The synthesis of the compounds mentioned above was observed using in-situ 31F
high temperature NMR. In general, the 31P NMR Spectra at high temperatures are
correlated with the final reaction products and final product distributions except for the
Rb4Ti2P6Se25 compound. In the synthesis of CS4P2869, the formation of a new Cs/P/Se
phase was discovered (CS4P23610) and the high temperature NMR Spectrum at 500 °C
correlates with the final product distributions.

The flux reaction of varying compositions of KZSe/PZSe5/Se was Studied in order
to determine which factor affects the formation of a specific type of anion using 31P room
temperature and high-temperature NMR spectroscopy. In addition to elucidation of
reaction mechanisms, we are hoping to develop mechanistic predictions for exploratory
alkali metal selenophosphate syntheses. Flux systems in excess of K28e compared to
PZSes generally forms crystalline selenophosphate final products, otherwise a mixture of
crystalline and amorphous selenophosphate products are formed. The 31F high
temperature NMR measurements have shown that equilibrium exists at high temperatures
for these systems. From the fast exchange between species in the melt, equilibria at high
temperatures have been observed and identified for reactions with flux ratios of (2:128),
(4:128), (4:238) that correlate with the final product distributions. The effect of cooling

rates was shown to be important in the formation of products.

To Lola Ba, My Family, and my Matsin.

iv

ACKNOWLEDGEMENTS

This would not have been possible without the help of many people to whom I am
very grateful. I am indebted to my advisor, Professor David P. Weliky for all his
guidance and support throughout the last five years. I would like to thank him for giving
me a chance to work in such a wonderful project, for nominating me for several
fellowships in the department and for the support to attend several research conferences
all of which were invaluable in the course of my degree. Through his mentorship,
especially his patience and desire to seek the answer at any cost, I have learned much
about being a scientist.

I am also indebted to our collaborator, Professor Mercouri G. Kanatzidis for
giving me the opportunity to work in this wonderful project especially in allowing me to
do most of my laboratory synthesis work in his lab. I have learned a lot from him through
his ability of simplifying seemingly complex problems, and insights in the chemistry of
exploratory chalcophosphate synthesis in which he is regarded as one of the people in the
forefront of this field.

I would like to thank all the past and present members of the Weliky group for
their help and advice and having had the chance to work with them throughout my time
here at MSU: Charles Gabrys, Jun Yang, Rong Yang, Paul Parkanzky, Michelle Bodner,
Chris Wasniewskj, Zhaoxiong Zheng, Jayson Criscione, Vamshi Cotla, Robert Boes,
Greg Berg. And also, I would like to thank the members of the Kanatzidis group for all
the help and advise they have given me with Special thanks to Jen Aitken, R.G. Iyer, In

Chung, Kasthuri Rangan, Pantelis Trikalitis. I would also like to thank Professor John L.

McCracken and his group specially Rajendra Muthukumaran (Pearl) for assistance with
EPR measurements of my samples. I would also like to thank Professor Larry Beck (Univ
of Michigan) and his group especially Kathryn Hughes for assistance on NMR
measurements on their 7 T magnet.

I acknowledge the Max T. Rogers NMR facility, its faculty and staff for their
assistance.

I would also like to thank my past mentors, friends and colleagues who have
Showed me the way to what research really is with special mention to Professor Cynthia
Goh (Univ of Toronto) and all the members of the UP-APCRL group.

Life here at MSU would not have been enjoyable without the company of
wonderful friends especially the members of the MSU Filipino Club past and present.

I would never have succeeded without the love and support of my family- my
parents Sebastian and Virginia Canlas and my brothers: Crisunde, Carlo, and Carvie and
my girlfriend Ellen. Their love and support is what kept me going throughout my stay
away from home. Lastly, I would like to thank the Lord Almighty to whom I owe

everything.

vi

 

TABLE OF CONTENTS

LIST OF TABLES .................................................................................... x
LIST OF FIGURES .................................................................................. xi
LIST OF ABBREVIATIONS .................................................................. xviii
CHAPTER 1: BACKGROUND OF THE STUDY ..................................... l
1 . Introduction ........................................................ 2
2. References ........................................................ 10
CHAPTER 2: THEORETICAL OVERVIEW OF SOLD-STATE NUCLEAR

CHAPTER 3:

MAGNETIC RESONANCE (NMR) APPLIED TO THE
INVESTIGATION OF METAL SELENOPHOSPHATE

SYNTHESES .............................................................. 12
1. Zeeman and Radio Frequency Harniltonians ................ 13
2. Chemical Shift ................................................... 14
3. Dipolar Interaction Hamiltonian. . . .._ ......................... 18
4. Spin-spin coupling or J-coupling .............................. l9
5. Magic Angle Spinning .......................................... 20
6. NMR Experiments .............................................. 22

Single-pulse Bloch decay ....................................... 22

Variable High Temperature in-Situ (HTNMR)

Experiment ....................................................... 24
7. References ........................................................ 25

31F SOLID STATE NMR STUDES OF METAL
SELENOPHOSPHATES CONTAH\I1NG [P2866]4., [P4Selo]4',

[PSe4]3', [P23e7141 AND [P25e914' LIGANDS ........................ 26
l . Introduction ...................................................... 27
2. Experimental Section ........................................... 28
Synthesis ......................................................... 28
Physical Measurements ......................................... 30
Elemental Analysis .................................... 30

Powder X-ray Diffraction ............................. 30

NMR ..................................................... 30

3. Results and Discussion ......................................... 33
[P28e6]4' .......................................................... 39
[P4Se10]4' ......................................................... 40
[PSe413' ............................................................ 41
[PZSegf' .......................................................... 41
[P23e714' .......................................................... 42

vii

CHAPTER 4:

CHAPTER 5:

CHAPTER 6:

4.

Correlation of CS, CSA, and T1 with Ligand Type. . . . .....42
Impurity Identification with Solid State NMR .............. 45

References ........................................................ 50

INVESTIGATION OF LONGITUDINAL 31F RELAXATION IN

METAL SELENOPHOSPHATE COMPOUNDS .................. 51
1 . Introduction ...................................................... 52
2. Materials and Methods .......................................... 53
Synthesis ......................................................... 53
Physical Measurements ......................................... 55
Powder X-ray Diffraction ............................. 55

NMR ..................................................... 55

ESR ....................................................... 56

3. Results and Discussion ......................................... 57
Longitudinal relaxation rates .................................. 57
Significance of relaxation times in compound
identification ..................................................... 62
Dependence of relaxation rates on dipolar couplings. ......64
Relaxation due to unpaired electrons ......................... 68

4. References ........................................................ 71

31P HIGH TEMPERATURE NMR STUDIES OF METAL

SELENOPHOSPHATE COMPOUNDS .............................. 73
1 . Introduction ...................................................... 74
2. Materials and Methods .......................................... 75
Preparation of Samples for in-situ NMR ..................... 75
Physical Measurements ......................................... 76

High temperature solid-state
nuclear magnetic resonance spectroscopy .......... 76
Room Temperature NMR ............................. 77
3. Results and Discussion ......................................... 79
[P23e614' .......................................................... 79
[PSe4]3' ............................................................ 88
[P4Se1014' ......................................................... 94
[P25e914' and the new [P25e1014‘ ............................... 99
[P23e714' ......................................................... 104
[PSe6]' ........................................................... 106
4. References ...................................................... 109

31F HIGH TEMPERATURE NIVIR STUDIES OF VARYING
RATIO OF THE FLUX COMPOSITION KZSe:P2865:Se IN THE

FAST COOLING REGIME ........................................... 110
1. Introduction ..................................................... 1 1 1
2. Materials and Methods ........................................ 118

viii

CHAPTER 7:

APPENDICES

Preparation of Samples for in-situ NMR ................... 118

Physical Measurements ....................................... 1 18
Differential Thermal Analyses ...................... 118
Powder X-ray Diffraction ........................... 118
High temperature solid-state nuclear magnetic
resonance Spectroscopy ............................... 118
Room temperature NMR ............................. 120
3. Results and Discussion ....................................... 121
KZSeszSeS:Se flux ratios group (1) ......................... 121
Equilibria between [PSe4]3' and [P28e9]4. in the flux
KZSe/PZSe5/Se .................................................. 139
KZSezP28e5:Se flux ratios group (2) ......................... 142
Effect of Heating and Cooling Rates ........................ 151
Ternary diagram for the KZSe/PZSe5/Se flux system. .....156
4. References ...................................................... 160
SUMMARY AND FUTURE DIRECTIONS OF THE
RESEARCH ............................................................. 161
1. Ab-initio calculations of the 31P NMR chemical
shifts of metal selenophosphates ............................. 162
2. In-Situ high temperature 3 P and 77Se NMR studies ...... 163
3. In-situ high temperature 31F and 77Se NMR studies of other
selenophosphate flux systems ................................ 165
4. Room and high temperature NMR Studies of
metal.thiophosphates and thiophosphate flux systems....165
5. References ...................................................... 167

31P high temperature NMR Spectra of the flux system

KZSe/PZSe5/Se ........................................................... 169
Plots of Chemical shift vs temperature of the flux System
Kzse/stes/Se .......................................................... 177

ix

Table 2.1.

Table 3.1

Table 4.1

Table 5.1

Table 6.1

Table 6.2

Table 6.3

Table 6.4

LIST OF TABLES

NMR parameters for 31F nucleus ................................................ 13

31F Chemical Shifts (CS), Chemical Shift Anisotropy (CSA), and T1
Measurements for Metal Selenophosphates .................................... 35

31F Chemical Shifts (CS), Chemical Shift Anisotropy (CSA), and T1
Measurements for Metal Selenophosphates .................................... 59

Shows the trend of the change in 31P NMR linewidth with
temperature ....................................................................... 105

Known melting points for some potassium selenide salts .................. 111

Shows the trend of the change in 31P NMR linewidth with
temperature ....................................................................... 13 l

Summarizes the relative amounts of [PSe4]3' and [P28e9]4' anions present

in flux (2:1:8) at high temperature using the lever rule as shown in Figure
6.7 .................................................................................. 141

Summary of results of the high temperature 31P NMR measurements at
500 0C for the flux ratios of group (2) ........................................ 148

Figure 1.1.

Figure 2.1

Figure 2.2

Figure 3.1.

Figure 3.2.

LIST OF FIGURES

Shows the effect of cooling rates on the final products for the reaction of
the flux composition KZSe:PZSe5:Se (422:8). (a) Fast cooling— has a 31P
NMR chemical Shift of —40.1 ppm that corresponds to crystalline K4P28e9.
(b) Slow cooling- has 31P NMR chemical Shifts at 43.0 and 45.6 ppm that
corresponds to the two crystallographically inequivalent P atoms in

K4P2866 .............................................................................. 6

Spinning about the magic angle. Orientations of the individual interaction
principal axes relative to the magnetic field, Bo, are averaged to the
orientation B, the inclination of the rotation axes relative to magnetic field
direction. When [3 = 54.7°, the broadening effects in the Hamiltonian
vanish ............................................................................... 21

The single-pulse Bloch decay experiment ..................................... 22

Schematic structures for the various [PxSey]n‘ anions examined in this
work ................................................................................. 34

31F solid state MAS NMR spectra of selenophosphate compounds. Each
spectrum represents a Single Bloch decay acquired after a delay time much

longer than the 31P T1 of the compound. Each spectrum was processed
with s 100 Hz line broadening and up to a 10th order polynomial baseline
correction. Chemical Shift referencing was done using 85% H3PO4 at 0
ppm. The isotropic peaks are indicated by asterisks (*) and spinning

sidebands are marked as “ssb”.(a) szPzSCG. The isotropic peak is at 29.1
ppm, the spinning frequency is 12 kHz, and the delay time is 10000 s. (b)
RbZCdP28e6. The isotropic peak is at 62.0 ppm, the spinning frequency is
12 kHz, and the delay time is 300 S. (c) KszPZSe6. The isotropic peak is
at 63.3 ppm, the spinning frequency is 15 kHz, and the delay time is 300
s. (d) Ag4PZSe6. The spinning frequency is 12 kHz and the delay time is
15000 5. There are isotropic peaks at 77.6 and 91.8 ppm which represent
the averages of doublets whose splitting is due to P—P J-coupling. For

Ag4PZSe6, the two P atoms in the [P28e6]4_ anion are crystallographically
and magnetically inequivalent and this is reflected in the observation of
two C83.11 The peak at -51.9 ppm corresponds to Ag7PSe6 which co-
crystallized as an impurity with Ag4P28e6. (e) K2Cu2P4Se10. The isotropic
peak is at 55.7 ppm, the Spinning frequency is 12 kHz, and the delay time
is 15000 s. (f) Cu3PSe4. The isotropic peak is at —83.3 ppm, the Spinning

xi

Figure 3.3.

Figure 3.4.

Figure 3.5.

Figure 3.6.

frequency is 12 kHz, and the delay time is 10000 s. (g) RbePSe4. The
isotropic peak is at —74.9 ppm, the Spinning frequency is 12 kHz, and the
delay time is 15000 8. Some impurity phases can also be seen in the NMR
Spectrum. (h) KPbPSe4. The isotropic peak is at -74.3 ppm, the spinning
frequency is 12 kHz, and the delay time is 5500 S. Some impurity phases

can also be seen in the NMR spectrum. (i) K4Pb(PSe4)2. The isotropic
peak is at -95.5 ppm, the spinning frequency is 7 kHz, and the pulse delay
is 1500 S. (j) CS4P2369. The isotropic peak is at -39.9 ppm, the

spinning frequency is 9 kHz, and the pulse delay is 4200 S. (k)
Rb4Ti2P6Se25. The isotropic peaks are at —34.6, —47.6 and —67.7 ppm, the
spinning frequency is 12 kHz, and the delay time is 6000 s. The first two
peaks corresponds to the two [P28e9]4- units and the last peak corresponds
to the [P28e7]4_ unit ............................................................... 37

(a) Experimental szCdPZSeG spectrum at 11.76 kHz MAS frequency and
(b) simulated stick spectrum calculated from the best-fit CSA principal
values. The Spectra were scaled so that their isotropic peak intensities were
equal. The principal values were derived from Herzfeld-Berger fitting of
the experimental isotropic and spinning Sideband intensities ............... 38

Experimental data and fitting of relaxation times for Ag4PZSe6 (solid
circles) and Ag7PSe6 (open squares). After nulling the magnetization,

there was a variable relaxation delay (I) followed by a 90° pulse and data
acquisition. Signal intensities (in arbitrary units) were calculated as the
sum of the integrated isotropic peak and spinning sidebands. Fitting was
done using Eq. (1) and the best-fit curves are superimposed on the
experimental data .................................................................. 39

(a) Chemical shifts of selenophosphates containing the anions [P28e6]4_,
[PSe4]3-, [P4Se10]4_, [PZSe9]4_, and [P28e7]4-. The * indicates the
selenophosphate containing the [P4Se10]4' anion, the # indicates the
selenophosphates containing the [P28e9]4- anion and the + indicates the

selenophosphate containing the [P28e7]4- anion. The bracketsindicate that
the CS is diagnostic of the presence or absence of a P—P bond. (b)

Comparison between the CSAS for compounds with the [PSe4]3- anion
and the CSAS for compounds containing all the other anions. Each CSA

was calculated as the difference 511 — 533 ...................................... 43

Comparison of the experimental X-ray powder diffraction patterns of the
synthesized Ag4PZSe6 and the calculated patterns of Ag4PZSe6 and
Ag7PSe6. (a) Experimental X-ray powder pattern. (b) Calculated X-ray
powder pattern for Ag4PZSe6. (c) Calculated X-ray powder pattern for

xii

Figure 4.1:

Figure 4.2:

Figure 4.3:

Figure 4.4:

Figure 4.5:

Ag7PSe6. The * in (a) indicate peaks whose relative intensities are not
consistent with the calculated Ag4P28e6 pattern ............................... 47

Schematic structures for various [PxSey]n- anions examined ................ 58

(a) Experimental data and fitting of relaxation times for Ag4PZSe6 (solid
circles) and Ag7PSe6 (open squares). (b) Experimental data and
fitting of relaxation time for K2CdP28e6. After nulling the magnetization,

there was a variable relaxation delay (1:) followed by a 900 pulse and
data acquisition. Signal intensities (in arbitrary units) were calculated as
the sum of the integrated isotropic peak and Spinning sidebands. Fitting
was done using Eq. (1) and the best-fit curves are superimposed on the
experimental data. Among all of the compounds in the present study,
Ag4PZSe6 and KszP28e6 had the slowest and fastest relaxation rates,
respectively ........................................................................ 61

31P NMR spectra of products of Ag/P/Se syntheses. All of the spectra
were taken at ambient temperature. The spectra in (a) and (b) are from one
synthesis with relaxation delays of 100 s and 200 S, respectively. Each
spectrum represents the co-addition of 32 scans taken at 8 kHz MAS
frequency. The Spectra in (c) — (f) are from a different synthesis with
relaxation delays of (c) 1000 S, (d) 2000 S, (e) 5000 S, and (f) 10000 3.
Each of the (c) - (f) Spectra represents a single FID which was taken after
the listed relaxation delay and at 12 kHz MAS frequency. There are
isotropic resonances for Ag4PZSe6 in the 70 — 95 ppm range and an
isotropic peak for Ag7PSe6 at —51 ppm. Spinning sidebands are also

apparent for Ag4PZSe6. Spectra are displayed so that the Ag7PSe6 peak
always has the same height. At the short relaxation delays in (a) and (b),
the Spectra have a larger Ag7PSe6 peak while at the 10000 s delay in (1'),
this peak represents only about 20% of the total integrated signal intensity.
All spectra were processed with 100 Hz line broadening .................... 63

Plots of 31F 1/1‘1 vs. squared dipolar interaction. Data is displayed for (a)
compounds whose selenophosphate anion does not contain a P—P bond
and (b) compounds whose selenophosphate anion contains a P—P bond.
The inset plot in (b) shows the four points closest to the abscissa on an
expanded vertical scale whose unit is 10"4 s". In each plot, the abscissa
represents an approximate measure of the sum of the squares of the
individual dipolar interactions between a central 31F and the nearby
nuclear spins. The dipolar sum calculation uses Eq. (2) and distances from
known crystal structures ......................................................... 67

X-band ESR spectra of (a) KszPZSe6 and (b) Rb2CdPZSe6 at 4.2 K. For
KZCdPZSeG, the spectrometer settings were: 100 kHz modulation

xiii

Figure 5.1.

Figure 5.2.

Figure 5.3.

Figure 5.4.

Figure 5.5.

Figure 5.6.

Figure 5.7.

Figure 5.8.

Figure 5.9.

frequency, 0.276 G modulation amplitude, 10.24 ms time constant, and

41.943 S sweep time. For szCdPZSC6, the Spectrometer settings were: 100
kHz modulation frequency, 2.188 G modulation amplitude, 5.12 ms time
constant, and 41.943 s sweep time. No ESR Signals were observed for an
empty cavity ....................................................................... 70

31F variable high temperature NMR spectra of the synthesis of szPZSC6.
(note: d=rampdown in temperature; r = recycled temperature). Note:

phasing difficulty in the spectrum at 500 °C .................................. 81

Chemical shift dependence on temperature in the synthesis of PbP28e6.

(a)Ramp-up temperature increase versus 31F chemical Shift of Pb2P28e6
from 300 to 600 degrees C. (b) shows the rampdown dependence of
chemical shift versus temperature from 600 to 300 degrees C. (c) shows

the temperature dependence of 31F chemical shift during ramp-up from
300 to 600 degrees C .............................................................. 82

31F high temperature NMR spectra for the syntheses of Ag4P28e6. (note:
=rampdown in temperature; I = recycled temperature) ..................... 86

Temperature dependence of 31F chemical shifts corresponding to the 2 P
peaks for Ag4PzSe6 (P1 and P2) and the P peak for Ag7PSe6 (P3) in the

temperature range 300-600 0C. (a) First cycle ramp-up in temperature; (b)
Rampdown in temperature; (c) Ramp-up in temperature in the second
cycle ................................................................................. 87

31F HT NMR syntheses of KPbPSe4. (note: d=rampdown in temperature; r
= recycled temperature) .......................................................... 90

31P HTNMR syntheses of K4Pb(PSe4)2. (note: d=rampdown in
temperature; r = recycled temperature) ......................................... 92

Shows the temperature dependence of 31P chemical shifts of the synthesis
of K4Pb(PSe4)2.(a) First cycle in temperature.(b) Recycled temperature..93

31F HTNMR syntheses of K2Cu2P4Sejo ........................................ 95

31F MAS NMR Spectrum of the final product formed during the syntheses
of K2Cu2P4Se10. The (*) indicate isotropic peaks. Two impurity peaks are
present at -17.2, —18.9 ppm. The spectrum was taken with 8 kHz spin rate
and 20003 pulse delay ............................................................ 96

xiv

Figure 5.10.

Figure 5.11.

Figure 5.12.

Figure 5.13.

Figure 5.14.

Figure 5.15.

Figure 6.1.

Figure 6.2.

Figure 6.3.

31F high temperature NMR Spectra of pure KzCu2P4Se10. The K2Cu-

2P4Se10 remain unchanged during the cycle. (note: d=rampdown in
temperature; r = recycled temperature) ......................................... 98

31F HTNMR syntheses of CS4P2369. The final product is composed of two

different crystalline products- CS4PZSe9 and a new phase CS4P25610. (note:
d=rampdown in temperature; I = recycled temperature) ................... 101

31P deconvoluted NMR spectra of the syntheses of CS4P2389 at 400 0C
during rampdown. Two peaks at —32 and -35 ppm are present at high
temperature. The integrated intensities are: -32 ppm is 24.65%; and —35
ppm is 75.35% ................................................................... 102

The crystal structure of CS4P28610 Shown down the a-axis. The dark-
colored atoms are Se atoms, the light-colored atoms are P atoms, the light-
colored atoms with no bonds are the CS atoms ............................... 103

31F High temperature NMR syntheses spectra for Rb4Ti2P6Se25. The final

product formed is pure Rb4TizP6Se25. (note: d=rampdown in temperature;
r = recycled temperature) ....................................................... 104

31F high temperature NMR spectra of pure RbPSe6 as starting material.
(note: d=rampdown in temperature) ........................................... 108

31F MAS high temperature Spectra of the KZSezP28e5:Se flux reaction
(2: 1:8).(note: d=rampdown in temperature; r = recycled temperature). 122

Figure 6.2. (i) 31F MAS NMR spectrum for the final product of the
reaction of Flux (2:1:8) shows two different 31P peaks indicated by (*) at
—1 13 ppm (K3PSe4) and at —39.6 ppm (K4P28e9). The ratio of integrated

intensities of the K4P28e9 peaks to the K3PSe4 peak is about 1.5. The
Spinning frequency is 8 kHz. (ii) 3'P MAS NMR Spectrum for the furnace

re-synthesized K4P28e9 shows two different 31P peaks indicated by (*) at
—113 (K3PSe4) ppm and -40 ppm (K4P28e9). The ratio of integrated

intensities of the K4PZSe9 peaks to the K3PSe4 peak is about 3. The
spinning frequency is 10 kHz. (iii) Comparison of (a) experimental and (b)
calculated X-ray powder diffraction patterns of K4P28e9. The * mark
represents the existence of another phase (likely K3PSe4). . . . . . . . . . 126

Differential Thermal Analysis Data for recycled Flux (2: 1:8) ............ 128

XV

Figure 6.4.

Figure 6.5.

Figure 6.6.

Figure 6.7.

Figure 6.8.

Figure 6.9.

Figure 6.10.

Figure 6.11.

Figure 6.12.

Figure 6.13.

31F High temperature NMR spectra of the reaction of KZSezPZSeszse with
ratio 3:1:8. (note: d=rampdown in temperature; r = recycled
temperature) ...................................................................... 130

31F High temperature NMR spectra of the reaction of K28e2PZSe5:Se with
ratio 3:228. (note: d=rampdown in temperature; r = recycled
temperature) ...................................................................... 133

31F High temperature NMR spectra of the reaction of K28e2PZSe5:Se with
ratio 4:1:8. (note: d=rampdown in temperature; r = recycled
temperature) ...................................................................... 137

31P High temperature NMR spectra of the reaction of KZSezPZSeszse with
ratio 4:2:8. (note: d=rampdown in temperature; r = recycled
temperature) ...................................................................... 138

Fast Chemical Shift Exchange at High Temperature. The dashed lines
indicate the chemical shift positions of fluxes (4:1:8) and (4:2:8) relative
to flux (2:1:8). The arrows containing the deltas (8) are measures of
distance in terms of ppm. 53 is the distance between the chemical shift
peaks of flux (4:1:8) and (4:228), 61 and 82 are the distance between the
chemical shift peak of flux (2:128) with flux (4:128) and (4:228)
respectively. The corresponding 31P room temperature MAS NMR
Spectrum is presented on the right side of the figure ........................ 140

31F High temperature NMR Spectra of the reaction of KZSe:P28e5:Se with
ratio 122:8. (note: d=rampdown in temperature; r = recycled
temperature) ...................................................................... 144

Plot of temperature versus 31P NMR chemical shift for all the samples
studied in this chapter. The groups of data presented are the chemical
shifts during the ramp-down recycle in temperature. Group (1) flux ratios
are the ones below 0 ppm, and group (2) flux ratios are the ones above 0
ppm ................................................................................ 147

Trend of the effect of increasing or decreasing the atomic percentage of a
reactant (K, P, Se) on the 31F chemical shifts at 500 0C .................... 150

Final product of flux (4:2:8) furnace syntheses with slow cooling ....... 153

Final products of separate fast cooling reaction of flux (4:2:8) in the

furnace. (a) the final products are K3PSe4 (*) at —l 13 ppm, and K4P28e6
(+) at 43 and 45.6 ppm plus some unknown impurities. (b) the final

xvi

Figure 6.14.

Figure 6.15.

Figure 6.16.

products are K3PSe4 and a similar anion type (*) at -99.3 ppm, and the
two crystalline peaks at -18 and —19.8 ppm (+) .............................. 154

31F MAS NMR of the final product formed from slow cooling reaction of
flux (314:8). The isotropic peak (*) at 54.6 ppm corresponds to crystalline
K2P28e6. The satellite peaks besides the isotropic peaks are spinning
sidebands .......................................................................... 155

Structure of K2P28e6 viewed down the b-axis. The shaded circles with
bonds are the P atoms, the circles that are partly shaded are the Sc atoms,
and the lightly shaded circles with no bonds are the K atoms. (Courtesy of
In Chung) .......................................................................... 156

Shows a plot of our results for flux ratios which gives crystalline final
products (flux: (2: 1:8), (3: 1:8), (421:8), (4:2:8)) and flux ratios which gives
amorphous final products plus crystalline peaks at —18.0 and —19.7 ppm
(flux (1:128), (122:8), (1:428), (22228), (22328), (2:428), (3:228), (32428),
(42428)) in scheme 2. The corresponding crystalline products are Shown in
the inset. Line (1) corresponds to P28e5:KZSe ratio (122), and line (2)

corresponds to PZSe5:KZSe ratio (1:3) ......................................... 158

xvii

1D/2D

CS

CSA

DC

DMF

DTA

EDS

ESR

HTNMR

MAS

PAS

PPm
PXRD

ssb

SSNMR

LIST OF ABBREVIATIONS

One-Dimensional/ Two-Dimensional
Chemical Shift

Chemical Shift Anisotropy

Dipolar Coupling
N,N-dimethylformamide

Differential Thermal Analyses
Energy Dispersive Spectroscopy
Electron Spin Resonance

Free Induction Decay

Full-Width at Half Maximum

High Temperature Nuclear Magnetic Resonance
Magic Angle Spinning

Nuclear Magnetic Resonance
Principal Axis System

parts per million

Powder X-ray Diffraction

Radio Frequency

Spinning sidebands

Solid-State Nuclear Magnetic Resonance

xviii

CHAPTER 1
BACKGROUND OF THE STUDY

INTRODUCTION
Chalcophosphates are compounds with oxidized phosphorus and at least one P—Q
bond, where Q = S or Se. To date, no examples with Q 2 Te exist in the literature. These

compounds exhibit an impressively rich structural diversity because of the large number

of stable [Psz]n_ building blocks that can be stabilized and the variety of binding modes

in which they can engage.1 Thio- and selenophosphates are still a relatively small group

of compounds compared to the huge class of oxophosphates. The latter are important in
the areas of catalysis, ceramics, glasses, and molecular sieves. Many thio- and seleno-

phosphates however also exhibit promising and unique properties such as intercalation

chemistry, ion-exchange, and magnetic and optical phenomena.2
Among the first chalcophosphate compounds to be studied were the M2P2Q6
. . . 4— . 3
class (M = divalent metal) WhICh features the ethane—like [P2Q6] lrgand as well as
4 5 . 3— . . .
M3PQ4 and MPQ4 wrth [PQ4] lrgands (M = monovalent or trrvalent metal). Various
other examples of [Psz]n- containing materials also exist.6 Interesting properties and
uses exhibited by some [P2Q6]4— containing compounds are: ferroelectric pr0perties for
use in memory devices (Sn2P236, CuInP286);7 nonlinear optical properties (Mn2P286);26

. . 8 . . . . . 9
photoconductrvrty (In1.33PzSe6); and cathode material In secondary lrthrum batteries.

Although many chalcophosphate compounds have been synthesized using the
traditional high temperature solid-state method in which the corresponding metals and

chalcophosphates are combined stoichiometrically at high temperature, there was also a

crucial role for polychalcophosphate fluxes.l In flux methodology, excess chalcogens and

and excess alkali chalcogens AzQ (A: Na, K, Cs, Rb) form molten salts of [PyQZ]n-

ligands at relatively lower temperatures which then react with metal ions. All flux
component are generally dissolved in the flux. Because of the low temperature,
kinetically stable phases can sometimes be trapped in lieu of thermodynamically stable

phases. Another factor is the ability of molten cations to readily crystallize with
selenophosphate anions as the flux is cooled. Molecular assemblies like [Psz]n- , at
lower temperatures and in the presence of a flux, have been found in new solid state
. . . . 4— . 10 . .
structures. An Interesting example IS the anron [P431310] anron whrch rs a
cyclohexane-like selenophosphate anion. In traditional high temperature methods, the

resulting compounds contained mostly the ethane-like [P2Q6]4- or the tetrahedral [PQ4]3_
anionic building blocks.1 The Kanatzidis group at Michigan State University is currently

investigating the generation, control, and reactivity of flux-assembled [Psz]n- (Q: S,

Se) anions as well as their utility in the synthesis of novel chalcophosphate materials. The

acid/base characteristics of the Ax[Psz] fluxes are very different from those of AZQx

fluxes because of the increased basicity of the phosphofluxes due to the presence of
phosphorous forming more negatively charged species which act as Lewis bases. Based

on the characterization of the compounds formed in these fluxes, the equilibria of Scheme

1 have been postulated.1

 

 

 

 

 

 

 

 

 

 

'2
Se Se
Se
Se
Se Se \
Se Se g:
'. Pu, 2
2 ‘s T / 39 Se
’5 Se
Se Se j
Se
Se S
89,6‘ e
6"P—P \28e\ ’
’4 Se S
I,” \ Se 2 e
/ ‘ 59/ \
Se ESQ Se
2e
Se\P/Se -2e 4
2 2: Se
Se Se ,
I
Se Se Se
\ / \ /
Se

Se

m\\\\\\|
CD

/
I‘D ‘
\
/
8’
(c6)
m\\\\“‘m
cor-U

(I)
(D
0

Scheme 1.

The typical components of metal selenophosphate reactions are elemental metals

(usually powdered), elemental Phosphorous or amorphous P28e5, and chalcogens. For

alkali metal polychalcogenide flux reactions, the components are alkali chalcogenides,

elemental phosphorous or amorphous P2365, and excess amounts of the chalcogens.
Crystalline PZSeS is shown to possess the structure shown in Scheme 2, and contains
. . +5 11
already oxrdrzed phosphorous (P ).
Se/P\\Se

Lil.

The starting materials are sealed in evacuated Pyrex or quartz tubes. Most of the

Scheme 2. Crystalline P28e5.

reactants are handled inside an inert atmosphere (N2) glove-box since the alkali

chalcogenides are moisture-sensitive and readily oxidixe in air. Inside an oven, reactions

are typically run in the 400-600 0C temperature ranges Since most of the starting

materials are molten at these temperatures. The temperature is held constant for several

days followed by slow cooling (2 to 20 OC/hr) to allow crystallization. The final product

is then examined using a light microscope in order to assess crystallinity. Some crystals
can be isolated from the glassy excess flux by dissolving and washing away the flux in a

polar solvent such as N,N-dimethylformamide (DMF). The excess fluxes are usually

composed of excess Se, K2Se, and sometimes glassy and crystalline P-Se units. The

crystalline products are then analyzed by chemical and structural techniques like
elemental analysis, powder and Single-crystal x-ray structure determination.

The discovery of novel final products is the main goal in the art of exploratory
solid-state syntheses. The final products can depend on a lot of variables such as

temperature, composition, and rate of heating and cooling. Figure 1.1 illustrates the

importance of the effect of cooling rate on the final products for a flux with composition

KgSe:P28e5:Se with a ratio 4:2:8.

 

 

 

 

 

(a.) (U)
_ K,P,Se, I<,1>,se6
l l ,
II I I
L 'w" J .J ‘4......_.....L..... .. A“ emgmv ’HWLUMMW
i r i I l l i l I I I q
0 -200 200 0 -200
1’1”“ ppm

Figure 1.1 Shows the effect of cooling rates on the final products for the reaction of the
flux composition K2SezPZSe528e (4:2:8). (a) Fast cooling- has a 31P NMR chemical Shift

of —40.1 ppm that corresponds to crystalline K4P28e9. (b) Slow cooling- has 31P NMR
chemical shifts at 43.0 and 45.6 ppm that corresponds to the two crystallographically

inequivalent P atoms in K4P28e6.

Figure 1.1 (a.) is the final product after doing a high temperature in-Situ NMR

reaction on the flux that has cooling rate relatively faster than a typical furnace synthesis

(~100 OC/hour). Figure 1.1 (b.) is the product after a furnace synthesis with a slow
cooling rate (~2 OC/hour). It seems that K4P28e9 is kinetically stable and is formed by

fast cooling, while K4P28e6 is thermodynamically stable formed from slow cooling. It is

potentially very useful to study the processes/reactions happening in the melt at high
temperature which led the final products. Although many new compounds have been

isolated from fluxes, there are very few direct studies of the flux components at high

12 . . . .
temperature. This study Will also allow more control over the reactron and the resultrng

compounds.
High temperature solid-state nuclear magnetic resonance (NMR) spectroscopy is

the most important in-situ technique that will identify and quantify the various
. . 31 . 77

chalcophosphate specres present In the flux. P along wrth Se NMR spectroscopy

could best provide us with this information, 31P in particular has a nuclear Spin of 1/2, 100

percent natural abundance, high gyromagnetic ratio and is very suitable for NMR.

There has been one NMR study of flux composition by Eckert er al. In a Li-P-Se

melt at T > 370 0C, the high temperature 31P NMR spectrum showed two peaks which
provided evidence for the existence of the stable phases Li4PZSe6 and Li7PSe6. NMR
spectra of the solid products showed that Li4P28e6 contains the [P2866]4_ unit, while

Li7PSe6 contains [PSe4]3_ and Sez— units.12 There have been additional 31P NMR

studies of ternary phases of crystalline solid products formed from direct combination of
M + P + Se ( M = Cu, Ag, Cd, Hg, Pb, Sn, Ca, and In) and it was reported that the MP-

Se system has much less structural anion variety compared to the ternary sulfide

12 . . . .
systems. In these direct combination melts, hrgh temperature NMR data were
. . . 4— 3—
consrstent With the presence of only two anions, [P2866] and [PSe4] . Eckert et al.

have also studied pure P-S and P-Se glasses with high temperature 31F and 77Se NMR.12

These NMR studies have confirmed the rich chemistry involved in

chalcophosphate synthesis. Using the flux methodology, new chalcophosphate phases

and anions are still being discovered and raise interesting questions about the equilibria

. . . l3 . . .
happening In the melt at hrgh temperatures. In this work we present 1n-srtu NMR

studies of the synthesis of known metal selenophosphate compounds and the flux reaction

of K23e + P2865 + Se at varying compositions. In order to study the NMR properties of

these compounds and reactions at high temperatures, it is important to have an idea of the
NMR properties of known representative selenophosphate compounds at room

temperature. In Chapter 3, we present a wide compilation of NMR data on known

14 . . . .
selenophosphate compounds. Approprrate correlations wrth structure and bonding

should yield considerable new insights in understanding and characterizing

chalcophosphate compounds. Chapter 3 presents for the first time solid state 31P NMR

investigations of several recently discovered phases containing known anions such as

[P25e6]4- (K2CszSe6, and RbngPQSeé) and [PSe4]3- (KPbPSe4, RbePSe4, and

K4Pb(PSe4)2) in bonding modes and arrangements different than those studied in the
. 4- 4—

past, as well as new anions such as [P4Se10] (K2Cu2P4Se10), [P2567]

. 4— . 15 31
(Rb4T12P63625) , and [P2369] (CS4P28e9, and RD4T12P63625). In Chapter 4, P
spin-lattice relaxation times (T1) have been measured for the solid compounds in order to
look for correlations with T1, dipolar couplings, and presence of unpaired electrons.l6
After laying out the basic 31P NMR properties (chemical shifts, chemical Shift

anisotropy, T1) of these compounds, in-situ high temperature NMR syntheses of the

compounds in chapter 3 are presented in Chapter 5, and in-situ high temperature NMR

studies for varying ratios of the flux KZSe/P28e5/Se are presented in Chapter 6.

REFERENCES

(1)

(2)

(3)

(4)
(5)

(6)

(7)

(8)

(9)

(10)

(11)

Kanatzidis, M. G. Curr. Opin. Solid State Mater. Sci. 1997, 2, 139-149 and
references therein.

(a) Clement, R. J. Chem. Soc., Chem. Commun. 1980, 647-648.

(b) Michalowicz, A.; Clement, R. Inorg. Chem. 1982, 21, 3872-3877.

(c) Jansen, M.; Henseler, U. J. Solid State Chem. 1992, 99, 110-119.

(d) Bridenbaugh, P. M. Mater. Res. Bull. 1973, 8, 1055-1060.

(e) Lacroix, P. G.; Clement, R.; Nakatani, K.; Zyss, J .; Ledoux, 1. Science, 1994,
263, 658-660.

(a) Johnson, J. W. Intercalation Chemistry; Edited by Whittingham M. S.;
Jacobson A. J. Academic Press Inc: New York, 1982.
(b) Ouvrard, G.; Brec, R.; Rouxel, J. Mater. Res. Bull. 1985, 20, 1181-1189.

Garin, J.; Parthe, E. Acta Crystallogr. 1972, 328, 3672-3674.

Le Rolland, B.; McMillan, P.; Molinie, P.; Colombet, P. Eur. J. Solid State Inorg.
Chem. 1990, 27, 715-724.

(a) Evain, M.; Brec, R.; Whangbo, M-H. J. Solid State Chem. 1987, 71, 244-262.
(b) Evain, M.; Lee, S.; Queignec, M.; Brec, R. J. Solid State Chem. 1987, 71, 139-
153.

(c) Evain, M.; Queignec, M.; Brec, R.; Rouxel, J. J. Solid State Chem. 1985, 56,
148-157.

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

(b) Rogach, E. D.; Amautova, E. A.; Savchenko, E. A.; Korchagina, N. A.;
Barinov, L. P. Zhumal Tek. Fiz. 1991, 61, 164-167.

(c) Bourdon, X.; Grimmer, A.-R.; Cajipe, V. B. Chem. Mater. 1999, II, 2680-
2686.

(a) Katty, A.; Soled, S.; Wold, A. Mater. Res. Bull. 1977, 12, 663-666.

(b) Etman, M.; Katty, A.; Levy-Clement, C.; Lemasson, P. Mater. Res. Bull.
1982, 17, 579-584.

(a) Thompson, A. H.; Whittingham, M. S. US. Patent 4,049,879 1997.

(b) Brec, R.; Le Mehaute, A. Fr. Patents 7,704,519 1997.

(c) Thompson, A. H., Whittingham, M. S. Mater. Res. Bull. 1977, 12, 741.
Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2098-2099.

Boldt, K., Engelen B. Acta Cryst. 1994, C50, 659-661.

10

(12)

(l3)

(14)

(15)

(16)

(a.) Francisco, R. H. P.; Tepe, T.; Eckert, H. J. Solid State Chem. 1993, 107,
452-459.

(b.) Francisco, R. H. P.; Eckert, H. J. Solid State Chem. 1994, 112, 270-276.

(c) Mutolo, P. F.; Witschas, M.; Regelsky, G.; Schmedt auf der Guenne, J.;
Eckert, H. J. Non-Cryst. Solids 1999, 256 & 25 7, 63-72.

((1) Maxwell, R.; Eckert, H. J. Phys. Chem. 1995, 99, 4768-4778.

(e) Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1994, 116, 682-689.

(f) Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1993, 115, 4747-4753.

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

Canlas, C.G., Kanatzidis, M.G., Weliky, D.P. Inorg. Chem. 2003, 42, 3399.

Rb4Ti2P6Se25 has also been examined by H. Eckert and G. Regelsky (private
communication).

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

11

CHAPTER 2
THEORETICAL OVERVIEW
OF
SOLID-STATE NUCLEAR MAGNETIC RESONANCE (NMR)
APPLIED TO THE INVESTIGATION OF METAL

SELENOPHOSPHATE SYN THESES

12

This chapter presents a summarized theoretical basis for NMR spectroscopy done

in this work.

Zeeman and Radio Frequency Hamiltonians.l’2

When nuclei of non-zero nuclear spin angular momentum are placed in a

magnetic field, its nuclear spin States Split as described by the Zeeman interaction:
Hz=-M°B=-YillzBo (1)
Where the magnitude of the splitting depends on the nuclear gyromagnetic ratio (y) and

the strength of the applied magnetic field, BO (along the z-axis), M is the magnetic

moment associated with the nuclear spin angular momentum, Iz is the nuclear spin
angular momentum projected along the external magnetic field (2 direction). The nucleus
most studied by solid-state NMR in this project is 31P with the following parameters
presented in Table 2.1.

Table 2.1 — NMR parameters for 31F nucleus

 

 

 

Nucleus Spin Natural Gyromagnetic Ratio, y NMR resonance
Abundance (107 rad T‘1 3-1) frequency (MHz)*
31p 1/2 100% 10.841 161.3

 

 

 

 

 

* relative to a 9.40 T magnetic field

The magnitude of the energy splitting between nuclear spin states is in the radio
frequency range. Therefore in order to excite transitions between the 2I+1 states (where I
is a non zero nuclear spin state), a radio frequency (rf) pulse is sent to the sample. The

effect of this pulse is described by the radiofrequency Hamiltonian 2

13

 

Hrf= 11°31 =-H°Brx=-Yh1311x (2)

where the applied radio frequency pulse B] is perpendicular to the external field Bo.

Classically, the applied field Bl causes a torque on the nuclear spin magnetization and

flips it away from the z-axis into the x-y plane.

The rf pulse is typically applied such that the magnetization is flipped by 90°.

Because of the torque exerted by the B1 field, the result is a transverse component of the

nuclear Spin magnetization along the x and y plane, MX and/or My. The resulting

transverse magnetization precesses in time about the z-axis at the Larmor frequency, v =

y/27t B0. This time-dependent transverse magnetization is detected through the NMR coil.

Chemical Shiftl ’2

When a substance is placed in a magnetic field, it becomes magnetized and it
modifies the field. This modification in field is due to electrons in the sample which are
induced to circulate around the nucleus, producing induced currents. These induced

currents create local magnetic fields which either enhance or oppose the primary applied
field B0. The induced fields can oppose B0 creating an effective field B at the nucleus
given by the expression:

B = BO(1 - o) (3)

where 0' is the shielding constant which is a dimensionless small fraction usually listed in

parts per million (ppm).

14

The chemical shift Hamiltonian describes the perturbation of the Larmor

frequency caused by the local chemical environments due to the motion of electrons in

terms of the second-rank shielding tensor , o 2
Hcs= 11111030 (4)
The shielding tensor is expressed by:
G = Gdiamagnetic + Gparamagnetic (5)
where Gdiamagnetic is the diagmagnetic shielding arising from the effects of paired

electrons surrounding the nucleus and reduces the magnetic field at the nucleus.

Oparamagnetic is the paramagnetic Shielding tensor term and is due to electrons with

higher electronic states interacting with the ground state of the atoms under the influence
of the applied magnetic field. This causes a deshielding effect which increases the
magnetic field within the sample causing a high frequency shift. In addition, the chemical
Shielding depends on the orientation of molecular segments (e.g. functional group)
relative to the external magnetic field direction.

The shielding is expressed by a tensor formulation,

oxx O'xy O'xz (6)
“yx “yy (M
an c’zy 0'22

In solution, rapid motion averages the chemical shift to “isotropic = 1/3 ((5XX +

oyy + on).

15

In solid-state NMR and in general, 0 depends on the orientation of the molecular

segment relative to the B0 field. In the lab frame (LF), B0 = B0 (0, 0, 1) is in the z-

direction, and with the unit vector b0 = Bo/Bo along the field direction, we have the

identity:

0
o: (0,0,1)oLF 0 =bososbos (7)
1

where 0 IS the laboratory frame Shreldrng tensor and 3 rs any coordinate system.

Combining equations (1), (3), (4) and using 000: -yB0, the relevant Hamiltonian
for the chemical shift of the Larmor frequency of a certain spin is denoted by:

HZ+HCS=000(l-o) I2 (8)
Which leads to a slow precession of the magnetization with a frequency:

(Des = - (Do 0 (9)
relative to the Larmor frequency 000 of the unshielded nucleus. The tensor is most

conveniently manipulated in the principal axis system (PAS), which is a molecule-fixed
axis system in which the chemical shift tensor is diagonal:

(on 0 0)
0 a”, 0

 

 

o = (10)
x 0 0 0 )
Combining equations (7), (9), and (10) gives us:
PAS PAS PAS
wCS='mob0 0 b0 (11)

16

On the right hand Side of equation 11, the B0 field, direction b0 and the tensor 0’ are

expressed by their coordinates in the PAS. Expressing be (in terms of its polar

. . . PAS PAS PAS . .
coordinates (0,6)) and the prrncrpal values oxx , oyy , on In the PAS grves the

general expression of the NMR frequency in terms of chemical Shielding:
(”CS = - (1)O [GXXPAS (cos (1) sin 0)2 + onyAS (sin (1) sin 0)2 + GZZPAS (cos 0)2] (12)

The polar angles 0 and (i) are the angles describing the external magnetic field direction in
the principal axis system (PAS).

In the solution state, the angular terms in equation (12) are averaged out due to
motion. In the solid state, the interaction tensor is dependent on the angular terms in
equation (12). This angular dependence of the interaction tensor with the external field
yields chemical Shift anisotropy (CSA). In a Single crystal, the effect of CSA is to cause a
dependence of the NMR frequency on the orientation of the crystal relative to the
external field. In powder samples, the effect of CSA is linebroadening of the NMR peaks.

For a nucleus in an axially symmetric environment, the chemical shielding tensor

simplifies to a perpendicular component, cl = 011 = 022 and a parallel component,

0 = 033. In this case equation (12) simplifies to

(0C5 = 1113010150 + (2mm -— 0.) (3cos29 -1)/21 (13)
where 0 is the angle between the symmetry axis and the external magnetic field direction,
and 0150 = [(2/3) oj + (1/3) 0”] is defined as the isotropic component of the chemical

shielding tensor.

l7

In NMR, spectra are often presented in terms of chemical Shifts rather than

chemical shielding. The chemical shift is : 5: —o and typically is referenced to a

particular compound (e. g. 85% H3PO4 for 31F).

. . . 2
Dipolar Interaction Hamiltonian

Another important type of interaction that will also be important in the analysis of
NMR of phosphochalcogenide systems is the dipolar interaction. There are two types of
dipolar interaction : homonuclear and heteronuclear dipolar interactions.

The dipolar interaction is analogous to the classical interaction described for two

spins, I and J, with magnetic moments 1,11 and 11], respectively, and separated by rij. The
dipolar energy is given by 2

Edipole = - I 3011' rjk) (M1 ' rjk) - #1 MJ }/ rij3 (14)
Quantum mechanically, the magnetic moments are defined by 111: hYI I and u] = hYJ J.

In a strong magnetic field the homonuclear dipolar coupling Hamiltonian is:

HD =- hZYIYJ t3(I-r><.I-r-)-I-Ji/r.,-3 (15)

or HD - (Mo/4n)(h2mr/rn3) [((3c0829- 1)/2) (31sz -I-J)]

where 0 is the angle between the radial vector between the two dipoles and the external

field direction. NMR measurement of dipolar interaction can be related to internuclear
distances, and the square of the dipolar interaction can he sometimes correlated to the

relaxation rate.

18

Spin-spin Coupling or J -couplingl’2

The J coupling of nuclear spins has similar Spin dependence as the dipolar
coupling (both have interactions bilinear in the spin operators), but only amounts to a few
hundreds of Hz or less. The energy of the J-coupling between two Spins (either

homonuclear or heteronuclear) is given by:

EJ = hJ(lelx2 + Iley2 +121122) (16)
where the IX, 1),, and I2 are the x, y, and 2 Spatial components of nuclear spin angular

momentum. This interaction is mediated through bonds and bonding electrons. For two

nuclei with nonzero spin in the molecule having spins 11 and 12, the resonance of Spin 11
is split into 212+1 lines of equal intensity and that of Spin 12 is Similarly split into 211+1
lines. The nature of the splitting suggests that energies of 11 are modified by a small
interaction from molecules with differing values of m1 of spin 12. Its magnitude is

determined by a Spin-spin coupling constant and is usually written as J12 for interaction

between Spins 1 and 2.

Unlike the chemical Shift and dipolar coupling interactions, the J-coupling of
nuclear spins has no Spatial or angular dependence. But like the dipolar coupling, the
magnitude of J-coupling is unaffected by variation of the spectrometer operating

frequency, and the coupling constants are always expressed in Hz and never in ppm.

19

Magic Angle Spinning (MAS)1’2

Chemical shift anisotropy and dipolar coupling interactions Significantly broaden

NMR lines in a solid powder (see Equations 12 and 15). The corresponding Hamiltonians

for these perturbations are dependent on the angles 0 and (I), which represent angles
between the applied magnetic field B0 and the principal axes of the interaction tensor.

These Hamiltonians can be modulated by rapidly spinning the sample with the spinner
axis oriented 54.7° from the external magnetic field. This angle is called the magic angle.

The geometry of this experiment is shown in Figure 2.12

20

 

 

FIGURE 2.1 - Spinning about the magic angle. Orientations of the individual

interaction principal axes relative to the magnetic field, B0, are averaged to the
orientation B, the inclination of the rotation axes relative to magnetic field direction.
When B = 54.7°, the broadening effects in the Hamiltonian vanish.

By Spinning the sample at the magic angle, the dipolar coupling Hamiltonian is averaged

out to zero, while the CSA Hamiltonian is averaged out to the isotropic chemical shift,

0150- The J- or scalar coupling of nuclear Spins is not changed by magic angle spinning

(MAS).

If spinning frequencies lower than the breadth of anisotropic NMR frequencies
are used, then satellite resonances called spinning sidebands (ssb) are seen. These
Spinning sidebands appear symmetrically on either Side of the isotropic frequency and

separated from it by integral multiples of the Spinning frequency.

21

NMR Experiments

Single-pulse Bloch decayl

11/2

llll fl.

IIIIII MALL“

Figure 2.2 — The single-pulse Bloch decay experiment

   
 

  

Delay Acquisition

     
 

 

 

II Illlin‘...
ll 1""

     

Figure 2.2 shows the single-pulse Bloch decay measurement (under magic angle
spinning) applied for the majority of the investigations in this work. The experiment is
composed of waiting for a certain delay time to establish the equilibrium spin population

difference between spin states induced by the Static magnetic field, followed by applying

0 . .
a 90 radio frequency resonant pulse whrch converts the z-component of the sample

magnetization into the x-y plane where it is detected by the NMR probe coil. The signal
detected is in the form of a free induction decay (FID) which is a plot of transverse
magnetization vs. time. Because of spin-spin and spin-lattice relaxation, Boltzmann
equilibrium is partially restored and the cycle can be repeated in order to increase the
acquired signal. Fourier-transformation of the FID provides a spectrum in the frequency

domain. The resulting frequencies, measured with reference to the standard material, are

22

evaluated as chemical shifts that are characteristic of the various chemical environments

experienced by the nuclei:
6
5 (Ppm) = (V " Vrefy Vref X 10 (17)

In addition, if the relaxation delay is sufficiently long, the integrated intensity of a
peak is directly proportional to the number of nuclei present with that chemical Shift.

The best value for the relaxation delay is dependent on the spin-lattice relaxation time

(also known as longitudinal relaxation time) or T1 of the sample. T1 is the characteristic
time to return the z-component of magnetization MZ to the equilibrium magnetization,

M0 and is illustrated mathematically by the following Bloch equation:

sz

= —(Mz-M0)/Tr (18)
dt

where Mx, My are the x and y components of magnetization and the B1 field is oscillating

with angular velocity of to. The relaxation of Mx and My is due to Spin-spin or transverse

relaxation times (T2), it is typically shorter compared to T1 and will not be discussed

here. Longitudinal relaxation requires the change of spin state (B —> or) and exhange of
energy with lattice modes.
Typically, the delay time is set to at least two times the spin-lattice relaxation time

of the nuclei.

23

Variable High-Temperature in-situ NMR (HTNMR) Experiment

31P HTNMR studies were performed for various metal selenophosphates in this
work. The experiments utilized the single-pulse Bloch decay experiment which
monitored the chemical shifts of different selenophosphate anions. Most of the HTNMR
studies presented in this work are static experiments; few samples were measured by
magic angle spinning HTNMR.

The experiments were conducted by raising the temperature by 500C increments
at a rate of lOOC/minute from room temperature up to 400 to 6000C followed by a cool-
down to room temperature at the same increment. Most of the samples studied were then
subjected to a repeat cycle of 31F HTNMR in order to observe the 31P NMR spectra of

the formed final product in the first cycle. Overall, the time it takes from heating to
cooling down the sample is about 6-12 hours compared to an oven synthesis that takes
several days.

Static NMR spectra are taken after equilibrating the sample at the specified

temperature for 20 minutes.

24

REFERENCES

(1) Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Wiley: New York,
1987.

(2) Schmidt—Rohr, K. and Spiess, H.F. Multidimensional Solid-State NMR and
Polymers; Academic Press Limited, 1994.

25

Chapter 3
31P Solid State NMR Studies of Metal Selenophosphates Containing

[stea‘z [PISeni‘z [PSe4133 thsen‘“ and [P2599I4— Ligands

26

INTRODUCTION

The ultimate goal of the project is to understand the chemistry involved in the
syntheses of metal selenophosphates at high temperature using in-situ high temperature
NMR spectroscopy. But in order to understand what is happening in the melt at high

temperature, the first step is to take NMR spectra of known compounds at room

. . 31 .
temperature and find correlations between NMR spectra and structure. P solrd state
nuclear magnetic resonance (NMR) spectra of twelve metal-containing selenophosphates

. . 4— -
have been examined to distrngursh between the [P2Se6] , [PSe4]3 , [P4Se10]4_,
[P2Se7] , and [P2Se9] anions. There IS a general correlation between the chemical
. . 4— 4-

shrfts (CSS) of anions and the presence of a P—P bond. [P28e6] and [P486101 both

- 3— 4—
contarn a P—P bond and resonate between 25 and 95 ppm whereas [PSe4] , [P28e7] ,
and [P28e9]4_ do not contain a P—P bond and resonate between —1 15 and -30 ppm. The

chemical shift anisotropies (CSAS) of compounds containing [PSe4]3_ are less than 80

ppm which is significantly smaller than the CSAS of any of the other anions (range: 135 —

275 ppm). The smaller CSAS of [PSe4]3- are likely due to its unique local tetrahedral
symmetry. Spin-lattice T1 relaxation times have been determined for the solid
compounds and vary between 20 and 3000 3. Unlike the CSS, T1 does not appear to

correlate with P—P bonding. 31P NMR is also shown to be a good method for impurity

detection and identification in the solid compounds. The results from the current study

27

31 . . . . . . . . .
suggest that P NMR Will be a useful tool for anron rdentrfrcatron and quantrtatron In

high temperature melts.

EXPERIMENTAL SECTION

Synthesis.

P2Se5 was prepared by reacting stoichiometric amounts of the elements in an

evacuated Pyrex tube at 300 0C for one day, followed by a cool-down to 50 0C over two

hours. Purity was assessed by X-ray powder diffraction analysis.

Ag4PZSe6,l Pb2P23e6,2 and CU3PSe43were prepared by reacting stoichiometric
amounts of the metal (Ag, Pb, Cu) with PZSeS and elemental Se. The reaction took place

in an evacuated quartz tube at 600 0C for one day followed by a cool-down to 50 0C over

twelve hours.

KPbPSe4 and RbePSe44 were prepared from stoichiometric amounts of the
alkali selenide (KzSe or szse), Pb metal, PZSe5, and elemental Se. The reactants were

heated in an evacuted quartz tube at 600 0C for one day and cooled down to 50 0C over

twelve hours.

K2Cu2P48e105 was prepared from 0.3 mmol Cu, 0.9 mmol P, 0.3 mmol K2Se,
and 2.4 mmol Se. The reactants were heated at 570 0C for two days followed by cooling

at 21 OC/h. The residual flux was removed with N,N-dimethylformamide. After washing

28

the remaining solid with diethyl ether, red-irregularly shaped crystals were obtained

which were stable in air and water.

K4Pb(PSe4)24 was prepared as orange crystals from 0.15 mmol Pb, 0.225 mmol
P28e5, 0.6 mmol KzSe, and 1.5 mmol Se. The reactants were heated at 500 0C for three
days followed by cooling at 10 OC/h.

szCdP2$e6 and KszPZSeé6 were prepared from 0.25 mmol Cd, 0.75 mmol
PZSe5, 1.0 mmol Rb2Se or K2Se, and 2.5 mmol Se with the same heating profile as was

used for K4Pb(PSe4)2. The residual flux was removed with N,N-dimethylformamide.

After the remaining solid was washed with diethyl ether, dark yellow rod-like crystals

were obtained, which were stable in air and water.

CS4P23697 was synthesized from a mixture of 0.45 mmol P28e5, 1.20 mmol
C8286, and 3.0 mmol Se. The reactants were sealed under vacuum in a pyrex tube and

heated to 490 0C for four days followed by cooling to 150 0C at 10 OC/hr. The product

crystals were red and were sensitive to air and water.

Rb4Ti2P6Se257 was synthesized from a mixture of 0.2 mmol Ti, 0.4 mmol P28e5,
0.4 mmol Rnge, and 2 mmol Se. The reactants were sealed under vacuum in a pyrex

tube and heated according to the same heating profile as CS4P2369. The residual flux was

removed with N ,N-dimethylformamide. After the remaining solid was washed with ether,

black crystals were formed, which were stable in air and water. The original goal of this

29

synthesis was to make RbTiPSe5 but powder X-ray diffraction on the whole sample and

elemental analysis on selected crystals showed that Rb4Ti2P6Se25 was the only

crystalline product.

Physical Measurements.

Elemental Analysis: The elemental compositions of selected crystals of

Rb4Ti2P6Se25 and K2Cu2P4Se10 were confirmed by semiquantitative elemental analysis

using Energy Dispersive Spectroscopy (EDS) on a JEOL 6400 Scanning Electron
Microscope equipped with a Tracor Noran detector.

Powder X-Ray Diffraction: Powder X-ray diffraction (PXRD) analyses were
performed using an INEL CPS 120 powder diffractometer with graphite
monochromatized Cu K a radiation. To assess sample purity, we visually compared the
experimental powder diffraction pattern to a pattern calculated from a Single crystal
structure. Visual inspection using a light microscope was also performed to assess the
crystallinity of the sample.

NMR: The room temperature solid-state NMR measurements of these compounds
were taken on a 9.4 T NMR Spectrometer (Varian Infinity Plus) using a double resonance
magic angle spinning (MAS) probe. Samples were spun at frequencies between 7 and 15

kHz using zirconia rotors of 4 mm outer diameter. Bloch decay spectra were taken with

the excitation/detection channel tuned to 31P at 161.39 MHz, a 4.5 [IS 900 pulse
(calibrated to i 0.1 us with 85% H3PO4), and a relaxation delay between 5 and 15000 8.

Each Spectrum was processed with s 100 Hz line broadening and up to a 10th order

30

polynomial baseline correction. The spectra were referenced using 85% H3PO4 at 0 ppm.

If the relaxation delay prior to pulsing is comparable to or longer than 31P longitudinal

relaxation time, a high signal-to-noise Spectrum can be obtained with one scan on 50 mg
of material.

Best-fit chemical shift anisotropy (CSA) principal values were calculated with a
Herzfeld-Berger computer program whose inputs were the 31P NMR and spinning
frequencies and the experimental peak intensities of the isotropic resonance and the
Spinning sidebands.8 For compounds containing the [PzSe6]4-, [P4Se10]4_, or [P2Se9]4—
anion, there were typically four spinning sidebands in a spectrum and for compounds

containing the [P23e7]4— or [PSe4]3— anion, there were typically two spinning sidebands

in a Spectrum. To quantitatively evaluate uncertainties in CSA principal values, spectra
for each compound were typically taken at two or more spinning frequencies and the
CSA principal value analysis was done at each frequency. Uncertainty was also evaluated
by doing analyses with spectral intensities changed by amounts comparable to the
spectral noise. Comparison of the analyses Showed that the greatest variation of a best-fit
principal value was ~ 20 ppm.

We also calculated an overall CSA value by taking the difference between the two
extreme CSA principal values. This overall CSA gives the approximate range of CSS
which would be observed in the static powder pattern, i.e. for consideration of all

possible orientations of the selenophosphate anion relative to the magnetic field direction.

The CSA is also a measure of the shielding field range for the 31P. When analyses were

31

compared between spectra taken at different spinning frequencies or between spectra
whose intensities were changed by amounts comparable to the Spectral noise, the greatest

variation of CSA was ~ 20 ppm.
Each compound 31P longitudinal relaxation time (T1) was determined by fitting
the equation:

-(t/T1))

S(t) = So(l—e (1)

where t is the relaxation delay time before pulsing, S(t) is the integrated signal intensity

(sum of isotropic peak and spinning sidebands), and S0 is a fitting parameter representing

the signal intensity at infinite t. Before the first I, the magnetization was nulled with a

series of 900 pulses at 0.5-1.0 second intervals for a duration of 15-30 seconds. For each
compound, there were typically at least two data sets and the experimental T1 uncertainty
was calculated from the variation in best-fit T1 values among the different data sets. This

variation was generally larger than the fitting uncertainty in the T1 value of a Single data

SCI.

32

RESULTS AND DISCUSSION

Figure 3.1 summarizes the structural schemes for the different selenophosphate

anions examined in this paper. 31P solid-state NMR spectra are presented in Figure 3.2
and the measured chemical shifts (CSs), CSA principal values, CSAS, and T1 values are

presented in Table 3.1. Figure 3.3 displays an experimental szCszSe6 spectrum as

well as a simulated Stick Spectrum calculated from its best-fit CSA principal values. The
agreement between intensities in the experimental and simulated spectra is a qualitative

illustration of the accuracy of the principal value analysis. Figure 3.4 displays examples

of experimental and best-fit buildup curves which were used to derive T1 values. In

general, there was good agreement between the experimental and fitted curves. Finally, it

is noted that in some spectra, there were resolved spectral splittings due to scalar (spin-

spin) couplings which are isotropic and are not averaged by MAS. For K2Cu2P4Se10,
Ag7PSe6, KPbPSe4, RbePSe4, K4Pb(PSe4)2, K3PSe4 and CS4P2369, the splittings were

7
due to 31P— 7Se scalar couplings while for Ag4P2866, the splittings were due to 31P—31P

scalar couplings.

33

(a) 11328414"

(b) [PSe413'
Se Se Se
S
41 ./ l
.9], (SP
(0 [P.Se1014' (d) [Pesent
r .
8 Se S; Se
Se Se g
/P‘|‘P// E,
SE // i /Se “J3 P
Se Se T 39 \\\\~‘ \3/ \Se
Se Se
Se
4-
(e) [P2569]
3; Se
1 Se 213,
86? \Se/ \Se/ \
Se

. . . n— . . . .
Frgure 3.1. Schematrc structures for the varrous [PxSey] anions exarruned 1n thrs work.

34

 

Table 3.1. 31P Chemical Shift (CS), Chemical Shift Anisotropy (CSA), and T1
Measurements for Metal Selenophosphates.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Seleno- Anion CS CSA principal values CSA Tl (s)d
h h t a
, .. (ppm) (ppm) (jam
511 522 533
pbszSeé pzseé‘i- 29.1 97 49 —59 156 1700 (100)
szcdpzseé 1328664— 62.0 161 93 —69 230 80 (5)
KZCdpzse6 p25e64- 63.3 155 101 —66 221 23 (2)
Ag4p2566 1525664- 77.6 152 73 7 145 3000 (200)
91.8 166 106 3 164
K4P28e6 1125.364" 42-7 - - - - 3500 (200)
45.4

P251464” 37-7 - — - - 4400 (150)
Ba2P2866
AgBiP28e6 P28664~ 14-6 - - - - 2250 (200)
K201215436 10 P 4561:1- 55.7 126 53 —12 138 1050 (100)
szzseéc P2566} 54'4 ' ' ' ‘ -
Ag7PSCG P5643: 451.9 n.d n.d. n.d. n.d. 1500 (50)

2_

Se
01315564 pse43_ —83.3 n.d. n.d. n.d. n.d. 300 (10)
Rbpbpse4 pSe43- —74.9 —50 —61 —1 14 64 970 (75)
Kpbpse4 pSc43- —74.3 —53 —54 —1 18 65 1080 (100)
K4Pb(PSe4)2 pse43' -95.5 n.d. n.d. n.d. n.d. -
K3153e4 pse43_ —113.2 —80 —r 13 —147 67 1250 (100)
Cs 4152369 1325694- —39.9 60 —59 —121 181 800 (200)
Rb4'ri2p6sezsf P25694- —34.6 100 —30 —174 274 540 (50)

4. —47.6 52 —66 —129 181 630 (40)
P2589

 

 

 

 

 

 

 

 

 

 

35

 

1.25674- —67.7 11 —65 —149 160 690 (50)

 

Rbpseé PSeé- 4.3 45 23 -68 113 3600 (100)

 

 

 

 

 

 

 

 

 

n.d. = not determined because of negligible spinning sideband intensity.
8 Uncertainties are ~ :I: 0.5 ppm.
b . . .
Maxrmum uncertarntres are ~ 3: 20 ppm.
C CSA = 811 — 533, i.e. the approximate overall width of the Static CS powder pattern.
Maximum CSA uncertainties are ~ :1: 20 ppm.
(1 . . . .
Uncertarntres are grven 1n parentheses.

c This is a new K/P/Se crystalline phase discovered by I. Chung, M.G. Kanatzidis
(manuscript in preparation)

CS assignments were based on G. Regelsky, Ph. D. thesis, University of Muenster,
2000.

36

 

 

 

 

 

 

 

 

 

 

 

 

(a) (0
ssb Sib 53:) ssb 1 #
'k
(b) (g)
ssb
ssb
ah I l 55." ssb I
'k
(C) (h)
ssb ssb ..
1 “I 1 53L
*
(d) * (i)
ssb ssb
(e) (1)
ab 51‘" Si" 52'}: -
200 100 0 -100 -200 k
ppm ( )

Figure 3.2. 31P solid state MAS NMR spectra of selenophosphate compounds. Each
spectrum represents a Single Bloch decay acquired after a delay time much longer than

the 31F T1 of the compound. Each spectrum was processed with s 100 Hz line

broadening and up to a 10th order polynomial baseline correction. Chemical Shift

referencing was done using 85% H3PO4 at 0 ppm. The isotropic peaks are indicated by

asterisks (*) and spinning sidebands are marked as “ssb”. (a) Pb2P28e6. The isotropic
peak is at 29.1 ppm, the spinning frequency is 12 kHz, and the delay time is 10000 s. (b)

szCdPZSeG. The isotropic peak is at 62.0 ppm, the spinning frequency is 12 kHz, and

37

 

the delay time is 300 s. (c) K2CdP2Se6. The isotropic peak is at 63.3 ppm, the spinning

frequency is 15 kHz, and the delay time is 300 S. (d) Ag4P2Se6. The spinning frequency
is 12 kHz and the delay time is 15000 5. There are isotropic peaks at 77.6 and 91.8 ppm
which represent the averages of doublets whose splitting is due to P—P J-coupling. For

Ag4P2Se6, the two P atoms in the [P28e6]4- anion are crystallographically and
magnetically inequivalent and this is reflected in the observation of two CSS. The peak
at -51.9 ppm corresponds to Ag7PSe6 which co-crystallized as an impurity with
Ag4P2Se6. (e) K2Cu2P4Se10. The isotropic peak is at 55.7 ppm, the spinning frequency
is 12 kHz, and the delay time is 15000 s. (f) Cu3PSe4. The isotropic peak is at —83.3 ppm,

the spinning frequency is 12 kHz, and the delay time is 10000 S. (g) RbePSe4. The
isotropic peak is at —74.9 ppm, the spinning frequency is 12 kHz, and the delay time is

15000 8. Some impurity phases can also be seen in the NMR spectrum. (h) KPbPSe4. The
isotropic peak is at —74.3 ppm, the spinning frequency is 12 kHz, and the delay time is

5500 3. Some impurity phases can also be seen in the NMR spectrum. (i) K4Pb(PSe4)2.
The isotropic peak is at —95.5 ppm, the spinning frequency is 7 kHz, and the pulse delay

is 1500 s. (j) CS4P28e9. The isotropic peak is at —39.9 ppm, the spinning frequency is 9

kHz, and the pulse delay is 4200 s. (k) Rb4Ti2P6Se25. The isotropic peaks are at —34.6, —
47.6 and —67.7 ppm, the spinning frequency is 12 kHz, and the delay time is 6000 s. The

first two peaks corresponds to the two [P28e9]4- units and the last peak corresponds to
the [P2Se7]4_ unit.

(a)

(b)

 

200 100 0 - 100 -200
PPm

Figure 3.3. (a) Experimental szCdP2$e6 spectrum at 11.76 kHz MAS frequency and (b)
simulated Stick spectrum calculated from the best-fit CSA principal values. The Spectra
were scaled so that their isotropic peak intensities were equal. The principal values were
derived from Herzfeld-Berger fitting of the experimental isotropic and spinning sideband
rntensrtres.

38

 

 

 

 

 

2.5 H
’3
E; 2.0 e
.E'
m 1.5 -
r:
’a’
: 1.0 — D S
to
c
.9
a) 0.5 -
0.0 1 I I
0 5000 1 0000 1 5000

Relaxation delay (3)

Figure 3.4. Experimental data and fitting of relaxation times for Ag4P28e6 (solid circles)
and Ag7PSe6 (open squares). After nulling the magnetization, there was a variable

relaxation delay (I) followed by a 900 pulse and data acquisition. Signal intensities (in
arbitrary units) were calculated as the sum of the integrated isotropic peak and Spinning
sidebands. Fitting was done using Eq. (1) and the best-fit curves are superimposed on the
experimental data.

[P25e6]4_2 The 31F CSS of selenophosphates containing [P28e6]4_ anions occur

downfield in the 14-95 ppm range. Their CSAS range from 145 to 230 ppm. One of the

compounds, Pb2PZSe6, has monoclinic symmetry so that the two P atoms are
crystallographically and magnetically equivalent and have the same CS.9 Two of the

compounds, szCszSC6 and K2CdP28e6, are isostructural and this is reflected in CSS

and CSAS which are within 1.3 ppm and 9 ppm of one another, respectively. These two

compounds also have the Shortest T1 values of any of the selenophosphate compounds in

39

 

. 4- . .
this study. For Ag4P’ZSC6, the two P atoms in the [P28e6] anron are crystallographically

and magnetically inequivalent and this is reflected in distinct CSS at 77.6 and 91.8 ppm.9

In addition, the P atoms experience a P—P scalar coupling of 426 Hz. These CS and scalar

coupling assignments were confirmed by measurements on a 7 T NMR. There is an

additional peak in the spectrum at —51.9 ppm which is assigned to a 20% Ag7PSe6
. . 10 3— 2— . . . .
Impurity. Ag7PSe6 has the form Ag7(PSe4 )(Se )2 and rts CS rs drstrnct from those
. . 4— .
found for compounds containing [P28e6] unrts.
4- , 4— . . . 5

[P4Se10] : The unique [P4Se10] anron was discovered in K2Cu2P4Se10 and
it essentially derives from the fusion of two [P28e6]4_ units followed by the elimination
of two Sez- anions. As a ligand it possesses eight terminal Se sites available for
coordination (cf. Figure 3.1(c)). K2Cu2P4Se10 has two crystallographically inequivalent

. . 31 . . .
P atoms in rts crystal structure, but the P NMR Spectrum has only a Single Isotroprc

peak at 55.7 ppm. To understand this apparent discrepancy, the environments around the
two P atoms were visually examined in the crystal strucuture. It was observed that the

local interatomic bond distances and angles were substantially the same for the two

. . . . 4—
atoms, which results in chemically and magnetrcally equrvalent P atoms. The [P4Se10]
anron rs Similar to the [P2Se6] anron in that It contains P—P bonds and tetravalent P.
These Structural Similarities help to explain the observation that the CS of K2Cu2P4Se10

(55.7 ppm) is within the [PZSe6]4_ anions. The CSA of K2Cu2P4Se10 (138 ppm) iS also

40

close to the CSAS found for compounds with [P28e6]4— anions. In K2Cu2P4Se10, the

resolved P—Se scalar coupling is 694 Hz.

[PSe4l3—z The 31P CSS of selenophosphates containing [PSe4]3_ units occur
upfield in the —l 15 to —50 ppm range. Their CSAS are < 80 ppm, which is Significantly
smaller than those observed for [P23e6]4_ anions, and likely reflects the high tetrahedral
symmetry of [PSe4]3_ units. One of the compounds, CU3PSe4, has a wurtzite-related

normal tetrahedral Structure with crystallographically equivalent P atoms in its unit cell

and a single CS9 Two of the compounds, KPbPSe4 and RbePSe4, have the same

orthorhombic crystal structure with crystallographically equivalent P atoms in their unit

cells. This leads to a Single CS for each compound which differ by only 0.6 ppm. Both

KPbPSe4 and RbePSe4 exhibit a resolved P—Se scalar coupling of 435 Hz. K4Pb(PSe4)2

has a structure consisting of two [PSe4l3— ligands which bridge adjacent Pb atoms. All P

atoms are crystallographically equivalent, leading to a single CS and a resolved P—Se

scalar coupling of 400 Hz. The P—Se scalar coupling of Ag7PSe6 is 490 Hz.
4— . . 4-
[PZSeg] 2 Two compounds, CS4P2569 and Rb4T12P6Se25, contarn the [P28e9]

anion. This ligand is structurally composed of two Se-Sharing [PSe4]3_ subunits. Thus, it

is reasonable that the CSS observed for this ligand (—50 to —35 ppm) are close to the CS

range observed for compounds containing single [PSe4]3— anions. The CSAS of

4- . 4- .
[P2369] anrons are comparable to those observed for [P28e6] anrons and are

41

significantly larger than those observed for [PSe4]3_ anions. The latter finding likely
4- . . 3— .
reflects the lower symmetry of [P2369] units relatrve to tetrahedral [PSe4] unrts. In
CS4P2Seg, the resolved P—Se scalar coupling is 530 Hz.
4— . . 4— . . .
[P2Se7] : Rb4T12P68e25 also contains the [PzSe7] lrgand whrch 1s structurally
3— . 4— . .
composed of a two [PSe4] shanng a common Se. The CS of the [P28e7] anron rs —
. . . 3— . 4- .
67.7 ppm and rs wrthrn the CS range of [PSe4] anrons. The CSA of the [P2Se7] anron
, 4— 4- . . . . .
rs comparable to those of [P2866] and [P28e9] anrons and IS srgnrfrcantly larger than
those of [PSe4]3_ anions. The CS assignment for Rb4Ti2P6Se25 were made using the
11.12 . . . .
work of Eckert et al. There are srmrlar T1 values for the compounds contarnrng the
[PzSe7]4_ anion and/or [P28e9]4_ anions.

Correlation of CS, CSA, and T1 with Ligand Type: The chemical shift

discrimination for the selenophosphates is better than that of their thiophosphate

counterparts.ll Figure 3.5(a) displays all of the measured CSS on a single ppm scale. As

shown in the figure, there is a correlation between the presence of a P—P bond and a

downfield CS. The [P2Se6]4— and [P4Se10]4_ compounds have positive CSS while the
3- 4— 4- .
[PSe4] , [P2Se7] , and [P2Se9] compounds have negative CSS. It may also be that

the [P23e914' CSS are somewhat downfield of the [PSe4]3" and [P23e714‘ CSS.

42

S e/ P—P 1’
88/ \Se Se \Se c

 

* ###+
II 111 1 . 1111111,

100 0 -120 ppm

0))

PSef' All other anions

A M

m 1111 II 1
150

300 ppm

Figure 3.5. (a) Chemical shifts of selenophosphates containing the anions [P2866]4_,
[PSe4]3_, [P4Se1014-, [P2Se9]4—, and [P28e7]4_. The * indicates the selenophosphate
containing the [P4Se10]4— anion, the # indicates the selenophosphates containing the

[P28e9]4— anion and the + indicates the selenophosphate containing the [P28e7]4— anion.
The brackets indicate that the CS is diagnostic of the presence or absence of a P—P bond.

(b) Comparison between the CSAS for compounds with the [PSe4]3_ anion and the CSAS
for compounds containing all the other anions. Each CSA was calculated as the

difference 511 - 833.

43

For a given anion type, isostructural compounds which only differed in alkali

metal ion had 31P CSS within 2 ppm of each other and CSAS within 10 ppm of each other

(e.g. Rb2CdP2Se6/K2CszSe6 and RbePSe4/KPbPSe4). Thus, the CSS and CSAS

appear to have only minor dependence on substitution of a metal ion within a specific

group. By contrast, K4Pb(PSe4)2 has the same elements and anion type ([PSe4]3-) as
KPbPSe4 but their CSS differ by 21 ppm. All of these observations are probably

. 31 .
reasonable because the electronic structure around P and thus the correspondrng CSS

and CSAS Should largely depend on local molecular structure.

Figure 3.5(b) displays all of the measured CSAS on a single ppm scale. The

[PSe4]3_ CSAS are significantly smaller than those of other ligands, presumably because
of the unique tetrahedral symmetry of the [PSe4]3_ anion. For compounds containing

[P2Se6]4_ anions, the smallest CSAS were found for Pb2P23e6 and Ag4P2Se6. This may

be due to the presence of one cation type in these solids compared to two cation types in

szCszSe6 and K2CdP2866.
. 4— 4— .

The T1 values for compounds wrth [P28e7] and [P2Se9] unrts were clustered
between 600 and 800 5 while those for compounds with [PzSe6]4_ units ranged between
20 and 4500 s and those for compounds with [PSe4]3— units ranged between 300 and
1500 3. Similar to the CS and CSA, isostructural pairs such as szCszSedKszP28e6

and RbePSe4/KPbPSe4 exhibit T1 values which are close to one another. However,

44

unlike the CS, T1 does not appear to correlate with P—P bonding. There may be a variety
. . . . 31 . . .
of Interactrons whrch contribute to P relaxation, rncludrng homo- and heteronuclear

dipolar couplings.l3 The experimental correlation between UT] and the square of the

dipolar coupling is at most qualitative, not quantitative, so there are likely other factors
which affect relaxation. As one example, although none of the compounds are
paramagnetic per se, there could be small quantities of trapped paramagnetic impurities

or defects which contribute to relaxation. To test this hypothesis, we resynthesized the

fast-relaxing szCdPZSe6 with the idea that paramagnetic reactant concentrations could
vary between reactant batches, and thus produce different T1 in the final product. The T1

value for the resynthesized RbZCdPZSC6 is 70 s, which is close to but not the same as its

value in the first synthesis. In future studies, we will attempt to detect paramagnetic
impurities using electron Spin resonance.

Impurity Identification with Solid State NMR: Figure 3.2(d) displays the spectrum

of Ag4P2Se6 and Ag7PSe6 products formed in a single synthesis. Because the 15000 s
delay time was much longer than the T1 value of either compound, the integrated

intensities are quantitative and give a Ag4PZSe62Ag7PSe6 product ratio of ~ 521.

Observation of Sharp lines strongly indicates the presence of crystalline domains of each

compound rather than glass formation. In fact, an X-ray analysis of one of the crystals

produced in the synthesis gave the Ag4P2Se6 structure. Interestingly, the presence of the

large Ag7PSe6 impurity was also not evident in X—ray powder diffraction analysis of the

45

 

entire reaction product. Figure 3.6(a) displays the experimental powder diffraction pattern

while (b) displays the pattern Simulated from the known Ag4P2Se6 structure. The good

agreement between experiment and Simulation initially caused confusion in the

assignment of the NMR spectrum. After identification of the Ag7PSe6 impurity, the

powder diffraction data were re-examined and two peaks were found in the experimental

pattern whose anomalous relative intensities could be explained by a contribution from

Ag7PSe6 (cf. Figure 3.6(a, c)). However, because of the strong overlap of the Ag7PSe6
and Ag4PzSe6 powder diffraction patterns, the presence of the Ag7PSe6 impurity was
generally masked in the experimental pattern. In contrast, the presence of Ag7PSe6 is
clearly evident in the NMR spectrum because Ag7PSe6 contains a [PSe4]3— anion rather

than a [P28e6]4— anion and its CS is more than 100 ppm upfield of Ag4PZSe6. Thus, for

these selenophosphate syntheses, NMR can be a powerful complementary tool for

product identification and analysis.

46

(a)

**

(b)

 

Diffraction Angle (degrees)

Figure 3.6. Comparison of the experimental X-ray powder diffraction patterns of the
synthesized Ag4PZSe6 and the calculated patterns of Ag4P2Se6 and Ag7PSe6. (a)
Experimental X-ray powder pattern. (b) Calculated X-ray powder pattern for Ag4PZSe6.
(c) Calculated X-ray powder pattern for Ag7PSe6. The * in (a) indicate peaks whose
relative intensities are not consistent with the calculated Ag4PZSe6 pattern.

Interestingly, repeated attempts to synthesize pure Ag4P2Se6 always yielded ~
20% Ag7PSe6 impurity. This was the case even when the reaction was carried out using

the Ag4P2Se6 stoichiometry. These results suggest that an equilibrium exists between

[P2866]? and [PSe4]3— under the 600 0C synthesis melt conditions and that the

equilibrium constant is close to 1. Future high temperature NMR studies will determine

47

the conditions which favor formation of Ag4PZSe6 or Ag7PSe6 and will follow up on
. 4— 3- . .
work by Eckert et al. whrch showed the presence of [P2366] and [PSe4] unrts 1n M-P-

Se melts (M = Li, Ag).9

Our data Show other examples of the utility of NMR in analyzing the products of

a selenophosphate synthesis. For example, for Figure 3.2(k), the intention of the synthesis

was to produce RbTiPSe5 using a procedure and stoichiometry analogous to that
employed for KTiPSe5.7 The NMR spectrum revealed that a pure product was formed

but it was Rb4Ti2P68e25 rather than the intended compound. In the RbePSe4 synthesis,

impurities are clearly present in the spectrum (cf. Figure 3.2(g)) but their chemical
identification has not yet been done. Finally, for spectra from some failed syntheses,
broad Signals were observed which indicate glass rather than crystalline product

formation.

In summary, the main result of this study is that the 31F CS is correlated with the

presence of P—P bonding in selenophosphate compounds. There are non-overlapping CS
ranges of compounds with P—P bonds (20 to 100 ppm) and compounds without P-P

bonds (—120 to —30 ppm). Thus, in future high temperature studies of metal-

selenOphosphate melts, the 31P CS should be useful in ligand identification. It may also

be possible to use the CS to distinguish between Similar ions in a melt, such as [P2Se7]4_
4.

and [P2Se9] anions. CSA measurements demonstrated that the CSAS of compounds

. . 3— . . . . . .
contarnrng [PSe4] unrts were srgnrfrcantly smaller than those of compounds contarnrng

48

any other ion. These other compounds include those with complex ions such as [PZSe7]
4— . . . . 3— .

and [PZSeg] whrch are brrdged forms of drmerrc [PSe4] . These CSA observatrons are

most reasonably explained by the tetrahedral symmetry which is unique to [PSe4]3— units.

The T1 values of the selenophosphate compounds varied between 20 and 3000 s, but

there was no clear correlation with P—P bonding. Finally, 31P NMR was shown to be a

useful tool for impurity detection and identification in selenophosphate syntheses.

49

REFERENCES
(1) Toffoli, P. P.; Khodadad, P.; Rodier, N. Acta Crystallogr. 1978, B34, 1779-1781.
(2) Yun, H.; IberS, J. A. Acta Crystallogr. 1987, C43, 2002-2004.
(3) Garin, J.; Parthe, E. Acta C rystallogr. 1972, 828, 3672.
(4) Chondroudis, K.; McCarthy, T.; Kanatzidis, M. G. Inorg. Chem. 1996, 35, 840-844.
(5) Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2098-2099.
(6) Chondroudis, K.; Kanatzidis, M. G. J. Solid State Chem. 1998, 138, 321-328.
(7) Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 5401-5402.
(8) Herzfeld, J .; Berger, A. E. J. Chem. Phys. 1980, 73, 6021-6030.
(9) Francisco, R. H. P.; Eckert, H. J. Solid State Chem. 1994, 112, 270-276.
(10) (a) Kuhs, W. F.; Schulte-Kellinghaus, M.; Kramer, V.; Nitsche, R. Z. Naturforsch.
1977, 32b, 1100-1101.
(b) Maxwell, R.; Lathrop, D.; Franke, D.; Eckert, H. Angew. Chem. Int. Ed. Engl.
1990, 29, 882-884.
(c) Gaudin, E.; Boucher, F.; Evain, M.; Taulelle, F. Chem. Mater. 2000, 12, 1715-
1720.

(11) (a) Schmedt auf der Gunne, J .; Eckert, H. Chem. Eur. J. 1998, 4, 1762-1767.
(b) Regelsky, G. Ph.D. Dissertation, University of Muenster, 2000.

(12) Derstroff, V.; Tremel, W.; Regelsky, G.; Schmedt auf der Gunne, J .; Eckert, H.
Solid State Sci. 2002, 4, 731-745.

(13) Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Wiley: New York, 1987.

50

CHAPTER 4
Investigation of Longitudinal 31P Relaxation
In

Metal Selenophosphate Compounds

51

INTRODUCTION
In Chapter 3, we focused on correlations between 31P chemical shifts (CSS) or
chemical shift anisotropies (CSAS) and local structure and bonding in metal

. . l
selenophosphates. We observed an Important correlation between the 3 P CS and the

presence or absence of a P—P bond in the selenophosphate anion. In this chapter, we

. . . . . 31 . .
present our 1nvest1gatron of longrtudrnal P relaxation 1n metal selenophosphate

compounds and observe relaxation rates which range over two orders of magnitude.
Unlike the CS and CSA, there does not appear to be a correlation between local anionic
structure and relaxation rates. A combination of theoretical and experimental approaches
suggests that much of the variation in relaxation rates can be correlated with: (1) dipolar
couplings to surrounding nuclei; and (2) dipolar couplings to unpaired electrons

associated with as-yet chemically unidentified paramagnetic impurities.

52

MATERIALS AND METHODS

Synthesis.

P2565 was prepared by reacting stoichiometric amounts of the elements in an

evacuated Pyrex tube at 300 0C for one day, followed by a cool-down to 50 0C over two

hours. Purity was assessed by X-ray powder diffraction analysis.

Ag4PZSe6l, Pb2P2Se62, and CU3PSC43 were prepared by reacting stoichiometric
amounts of the metal (Ag, Pb, Cu) with PzSeS and elemental Se. The reaction took place

in an evacuated quartz tube at 600 0C for one day followed by a cool-down to 50 0C over

twelve hours.

KPbPSe4 and RbePSe44 were prepared from stoichiometric amounts of the
alkali selenide (K28e or RbZSe), Pb metal, P2Se5, and elemental Se. The reactants were

heated in an evacuted quartz tube at 600 0C for one day and cooled down to 50 0C over

twelve hours.

K2Cu2P48e105 was prepared from 0.3 mmol Cu, 0.9 mmol P, 0.3 mmol K236,
and 2.4 mmol Se. The reactants were heated at 570 0C for two days followed by cooling

at 21 OC/h. The residual flux was removed with N,N-dimethylformamide. After washing

he remaining solid with diethyl ether, red-irregularly shaped crystals were obtained

'hich were stable in air and water.

53

 

K4Pb(PSe4)24 was prepared as orange crystals from 0.15 mmol Pb, 0.225 mmol
P2Se5, 0.6 mmol K2Se, and 1.5 mmol Se. The reactants were heated at 500 0C for three
days followed by cooling at 10 OC/h.

szCszSe6 and K2CdP28e66 were prepared from 0.25 mmol Cd, 0.75 mmol
PZSeS, 1.0 mmol RbZSe or K256, and 2.5 mmol Se with the same heating profile as was

used for K4Pb(PSe4)2. The residual flux was removed with N,N-dimethylformamide.

After the remaining solid was washed with diethyl ether, dark yellow rod-like crystals

were obtained, which were stable in air and water.

CS4P23697 was synthesized from a mixture of 0.45 mmol P28e5, 1.20 mmol
C8286, and 3.0 mmol Se. The reactants were sealed under vacuum in a pyrex tube and

heated to 490 0C for four days followed by cooling to 150 0C at 10 OC/hr. The product

crystals were red and were sensitive to air and water.

Rb4Ti2P6Sez57 was synthesized from a mixture of 0.2 mmol Ti, 0.4 mmol P2Se5,
0.4 mmol szse, and 2 mmol Se. The reactants were sealed under vacuum in a pyrex

tube and heated according to the same heating profile as CS4P2869. The residual flux was

removed with N ,N-dimethylforrnamide. After the remaining solid was washed with ether,

black crystals were formed, which were stable in air and water. The original goal of this

synthesis was to make RbTiPSe5 but powder X-ray diffraction and elemental analysis on

selected crystals showed that Rb4TizP68e25 was the only crystalline product.

54

Physical Measurements.

Powder X-Ray Diffraction: Powder X-ray diffraction analyses were perforrrred
using an INEL CPS 120 powder diffractometer with graphite monochromatized Cu K a
radiation. To assess sample purity, we visually compared the experimental powder
diffraction pattern to a pattern calculated from a Single crystal structure. Visual inspection
using a light microscope was also performed to assess the crystallinity of the sample.

NMR: The room temperature solid-state NMR measurements of these
polycrystalline compounds were taken on a 9.4 T NMR Spectrometer (Varian Infinity
Plus) using a double resonance magic angle spinning (MAS) probe. Samples were spun at

8 to 15 kHz using zirconia rotors of 4 mm outer diameter and 50 111 volume. Bloch decay

. . . . 31
spectra were taken With the excrtatron/detectron channel tuned to P at 161.39 MHz, a

4.5 ,us 900 pulse (calibrated to i 0.1 us), and a relaxation delay which was varied over a

range from 5 to 15000 S. Each Spectrum was processed with s. 100 Hz line broadening

and up to a tenth order polynomial baseline correction. The spectra were referenced using
85% H3PO4 at 0 ppm.

Each compound 31F longitudinal relaxation time (T 1) was determined from fits to
the equation:

— (r/Tl)

sec) =Sp(1—e ) (1)

where 1: is the relaxation delay time before pulsing, S(‘c) is the integrated signal intensity

(sum of isotropic peak and spinning sidebands), and S0 is a fitting parameter representing

the signal intensity at infinite 1:. Before the first 1:, the magnetization was nulled with a

55

. o . .
series of 90 pulses at 0.5-1.0 second Intervals for a duratron of 15-30 seconds. For each
compound, there were typically at least two data sets and the experimental T1 uncertainty
was calculated from the variation in best-fit T1 values among the different data sets. This

variation was generally larger than the fitting uncertainty in the T1 value of a single data

set.

ESR: Spectra were obtained at X-band using a Bruker ER300E spectrometer
equipped with a TE102 ESR cavity. An Oxford ESR-900 cryostat and ITC 502
temperature controller were used to maintain constant sample temperature. The external
magnetic field strength was measured with a Bruker ER 035M NMR Gaussmeter and the
microwave frequency was determined with an EIP 25B frequency counter. The

polycrystalline samples were contained in vacuum-sealed quartz tubes at 4.2 K.

56

RESULTS AND DISCUSSION

Figure 4.1 summarizes the structural schemes for the different selenophosphate

anions examined in this paper. 31P solid state NMR data are presented in Table 4.1,
including the measured CS, CSA, and T1 values. The CS and CSA measurements are

described in Chapter 3.8 Briefly, Table 4.1 Shows that the 31P CS is correlated with the

presence of P—P bonding in metal selenophosphate compounds. There are non-
overlapping CS ranges of compounds with P—P bonds (14 to 100 ppm) and compounds

without P—P bonds (—120 to —30 ppm). CSA measurements demonstrated that the CSAS

of compounds containing [PSe4]3- were significantly smaller than those of compounds
containing any other ion. These CSA observations are most reasonably explained by the

tetrahedral symmetry which is unique to [PSe4]3—.

Longitudinal relaxation rates.

Figure 4.2 displays experimental and best-fit buildup curves for (a) Ag4P2Se6 and

Ag7PSe6 and (b) K2CdP2Se6. Fitting was done using Eq. (1) and yielded 31P T1 values
of 3000 S, 1500 S, and 20 S, respectively. Among all of the compounds in the present

study, Ag4P2Se6 and K2CdP2Se6 had the one of the Slowest and fastest relaxation rates,
respectrvely. The T1 values for compounds contarnrng [P28e7] and [PZSeg] were
clustered between 600 and 800 s while those containing [P28e6]4_ ranged between 20
and 4500 S and those containing [PSe4]3— ranged between 300 and 1500 s. Isostructural

pairs such as szCdPZSe6/K2CdP2Se6 and RbePSe4/KPbPSe4 exhibit T1 values which

57

are close to one another. Unlike the CS, T1 does not appear to correlate with P—P
bonding.

(a) [PpSepr' (b) [PSe413'
Se Se Se
89/; ‘
I’P—--P{ "p
/ \0/I/ Se \\\\\“ \Se
Se Se Se Se
(c) [P4Seppi4' (d) [PpSe714'
Se S
e Se
[L‘se /Se \ 8% Se
Se/ '1 // Se ’P EP,
/ - P 8‘
Se fSe\ 39$ \Se/ \Se
Se Se 39
(e) [PpSepr

Se Se
1 Se P,
Se? \ / \ / \
S§\ se 59 59

I\\\“\g)

. . . Il- . .
F1 gure 4.1: Schematic structures for varrous [PxSey] anrons examined.

58

Table 4.1. 31P Chemical Shift (CS), Chemical Shift Anisotropy (CSA), and T1
Measurements for Metal Selenophosphates.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Seleno- Anion CS CSA principal values CSA T d
phosphate type a b (Win 1 (S)
(ppm) (ppm) c
)
51 1 522 533
4-
13132132366 P2896 29.1 97 49 —59 156 1700 (100)
szcdpzse6 P23664- 62.0 161 93 —69 230 80 (5)
K2CdP2Se6 15236} 63.3 155 101 —66 221 23 (2)
Ag4p2366 p236;- 77.6 152 73 7 145 3000 (200)
91.8 166 106 3 164
K4P28e6 1225664— 42.7 - - - - 3500 (200)
45.4
ste64‘ 37-7 - - - - 4400 (150)
Ba2P28e6
AgBiPZSeé 1225664‘ 14-6 - - - - 2250 (200)
19921545610 P436104. 55.7 126 53 —12 138 1050 (100)
szzseee p25,? 54.4 - - - - -
Ag7pse6 p364}, —5 1.9 n.d n.d. n.d. n.d. 1500 (50)
2-
Se
(3.815534 155,643- —83.3 n.d. n.d. n.d. n.d. 300 (10)
Rbpbpse4 P364} —74.9 —50 —61 —1 14 64 970 (75)
KPbPSe4 ”643- —74.3 —53 -54 -118 65 1080 (100)
K4Pb(PSe4)2 PSe43- -95.5 n.d. n.d. n.d. n.d. -
[(315364 P564} —113.2 —80 —1 13 —147 67 1250 (100)
(3541:2569 1525694- —39.9 60 —59 —121 181 800 (200)
Rb4T12P6SCZ 5f P23694- -34.6 100 —30 —174 274 540 (50)
4. —47.6 52 —66 —129 181 630 (40)
P2569

 

 

 

 

 

 

 

 

59

 

 

1.23674- —67.7 11 —65 —149 160 690 (50)

 

RbPSe6 PSe6— 4.3 45 23 -68 113 3600 (100)

 

 

 

 

 

 

 

 

 

n.d. = not determined because of negligible Spinning sideband intensity.
a Uncertainties are ~ :1: 0.5 ppm.
Maximum uncertainties are ~ :1.- 20 ppm.
C CSA = 511 — 533, i.e. the approximate overall width of the Static CS powder pattern.
Maximum CSA uncertainties are ~ :1: 20 ppm.
d Uncertainties are given in parentheses.

c This is a new K/P/Se crystalline phase discovered by I. Chung, M.G. Kanatzidis
(manuscript in preparation)

CS assignments were based on G. Regelsky, Ph. D. thesis, University of Muenster,
2000.

60

 

(a)

 

2.5 —

2.0 -~

1.5-

1.0-

Slgnal intensity (a. u.)

0.5 -

 

 

 

0.0 T I l
0 5000 10000 I 5000

Relaxatlon delay (3)

 

 

 

 

 

’5‘
4';
1?
2
.E
2
9
tn
0 100 200 300
Relaxation delay (a)

Figure 4.2: (a) Experimental data and fitting of relaxation times for Ag4PZSe6 (solid
circles) and Ag7PSe6 (open squares). (b) Experimental data and fitting of relaxation time
for K2CdP2Se6. After nulling the magnetization, there was a variable relaxation delay (I)

followed by a 900 pulse and data acquisition. Signal intensities (in arbitrary units) were
calculated as the sum of the integrated isotropic peak and Spinning sidebands. Fitting was
done using Eq. (1) and the best-fit curves are superimposed on the experimental data.

Among all of the compounds in the present Study, Ag4P28e6 and K2CdP2Se6 had the
slowest and fastest relaxation rates, respectively.

61

Significance of relaxation times in compound identification.

. . . 31 . . . .
The 1mportance of quantifying P relaxation rates rs demonstrated 1n Frgure 4.3,

which presents the NMR analysis of the products of two Ag/P/Se reactions which were
both run under the conditions described in Materials and Methods. The Spectra in (a) and
(b) are from one run of the reaction and the Spectra in (c) — (f) are from the second run of
the reaction. For both reactions, the experimental X—ray powder diffraction pattern of the

products was approximately quantitatively consistent with the pattern Simulated from the

known single crystal structure of Ag4P2$e6 (data not shown).1 However, NMR data
clearly showed a significant Ag7PSe6 impurity, as displayed in Figure 4.3. In these

spectra, Ag7PSe(, resonates at —51 ppm while the two inequivalent 31P in Ag4P2Se6

resonate at 77.6 and 91.8 ppm.9’10’ll At the Short relaxation delays in (a) and (b), the

Spectra have a larger Ag7PSe6 peak while at the 10000 s delay in (1'), this peak represents
only about 20% of the total integrated signal intensity. Two significant conclusions can
be drawn from these data. First, for these syntheses, 31P NMR is a complementary tool to

powder X-ray diffraction for compound identification and impurity analysis. In this case,

the powder diffraction patterns of the two phases were very Similar and NMR was a

. . . . . . . 31
better technique to Investigate compound purity. Second, the large varratron In P

relaxation rates among different compounds can lead to quantitatively distorted Spectra at

short relaxation delays. In this case, the data at short relaxation delays initially led to

confusion in assignment of peaks to Ag4PzSe6.

62

(a)

1
1

 

 

 

 

M— L. A
(C)
I...
(d)
L.
(e)

(f)

 

I I I I l
150 100 50 0 ppm

31P chemical Shift

Figure 4.3: 31P NMR spectra of products of Ag/P/Se syntheses. All of the spectra were
taken at ambient temperature. The spectra in (a) and (b) are from one synthesis with
relaxation delays of 100 S and 200 s, respectively. Each spectrum represents the co-

addition of 32 scans taken at 8 kHz MAS frequency. The spectra in (c) — (f) are from a
different synthesis with relaxation delays of (c) 1000 s, (d) 2000 s, (e) 5000 S, and (f)
10000 s. Each of the (c) — (1) spectra represents a Single FID which was taken after the
listed relaxation delay and at 12 kHz MAS frequency. There are isotropic resonances for

Ag4PZSe6 in the 70 - 95 ppm range and an isotropic peak for Ag7PSe6 at —51 ppm.
Spinning sidebands are also apparent for Ag4PZSe6. Spectra are displayed so that the
Ag7P866 peak always has the same height. At the short relaxation delays in (a) and (b),

the spectra have a larger Ag7PSe6 peak while at the 10000 S delay in (f), this peak
represents only about 20% of the total integrated signal intensity. All Spectra were
processed with 100 Hz line broadening.

Dependence of relaxation rates on dipolar couplings.

. . . . . . 31 . .
There are a wrde variety of rnteractrons whrch could contrrbute to P longrtudrnal

relaxation including CSA and dipolar couplings.12 However, CSA does not appear to be

a major factor, as evidenced by the lack of positive correlation between CSA and

relaxation rate (UT 1). For example, RbePSe4, KPbPSe4, and K3PSe4 all have CSAS

which are approximately half those observed for Pb2P28e6 and Ag4P2Se6, yet the

relaxation rates of the former compounds are two to three times greater than those of the
latter compounds.
A second possibility is that homonuclear and heteronuclear dipolar couplings play

a dominant role in relaxation. If this were the case, then there Should be an approximately

positive linear correlation between the dipolar coupling squared (DC)2 and 1/T1.12 To

test for this correlation, the former quantity was calculated using the following equation:
(DC)2 = >3 3/2 ( 1124/4461 + 2 4 BJ 3,2 ( 11:2 162/446 ) (2)

This equation is based on an approximate model which considers that the total relaxation

rate of a phosphorus nucleus (P) is the sum of relaxation rates from individual P—X

64

dipolar interactions where X is a nearby nucleus. X could either be a phosphorus or

another element type (I). In the right Side of Eq. (2), the first and second terms represent

these homo- and heteronuclear squared dipolar couplings, respectively. In Eq. (2), rp_p
. . . 1-7
and rp_1 are rntemuclear drstances determined from crystal structures, J represents one

of the isotopes of element 1, B J is its fractional natural abundance, S] is its spin quantum

number, and yp and y] are the gyromagnetic ratios of 31F and J, respectively. The sums

are over all isotopes of nearby homo- and heteronuclei. Eq. (2) only considers the squared
dipolar coupling constants and neglects the angular dependences of the dipolar couplings,
although presumably the Significance of these anisotropic terms is reduced by powder
averaging. In addition, this model: (1) neglects the multiple magnitudes of azimuthal
quantum numbers for quadrupolar Spins and the resulting multiplicity of dipolar matrix
elements; and (2) assumes that all compounds have Similar ranges of motional correlation

times.

Figure 4.4 displays plots of III] vs. (DC)2 for (3) compounds whose

selenophosphate anion does not contain a P—P bond and (b) compounds whose
selenophosphate anion contains a P—P bond. For the inset plot in (b), the four points
nearest the abscissa are shown on an expanded vertical scale. In Figure 4.4(a), there is an
approximate positive linear correlation between the two variables. Interestingly, the
intercept on the ordinate axis corresponds to a non-zero relaxation rate, which is
physically reasonable and could be due to some interaction other than dipolar coupling.

In some contrast, if one considers all of the data in Figure 4.4(b), the data do not fit well

65

to a linear correlation. However, a clearer picture emerges if we consider the l/Tl vs.

(DC)2 data as three groups: (1) group A which corresponds to the data in Figure 4.4(a);

(2) group Bl, the four points nearest the abscissa of Figure 4.4(b) (shown on an expanded
vertical scale in the inset); and (3) group B2, the other two points in Figure 4.4(b). In the
subsequent section, we will discuss a possible mechanism for the very fast relaxation of
group B2. In this section, we present a unified model for groups A and B1. First, as
shown in the inset of Figure 4.4(b), the group Bl data also Show an approximate positive

linear correlation. Furthermore, if the large P—P dipolar couplings are not included in the

(DC)2 calculation, then these Bl data generally fall within the group A positive linear

correlation displayed in Figure 4.4(a). This latter observation suggests that modulation of
the large directly bonded P—P dipolar coupling does not contribute greatly to relaxation.

Because Se is 92% spin zero isotopes, P—Se couplings make only a small contribution to

(DC)2 for most compounds. Thus, groups A and B1 data are consistent with a large

component of dipolar relaxation which is predominantly due to fluctuations in P—M
couplings (M 2 metal ion). Although the static P—M couplings are smaller than the
directly bonded P—P coupling, the metal ions may have larger motional amplitudes and/or
more efficient motional correlation times than the selenophosphate anions, and thus
dominate dipolar relaxation. Metal ion NMR as well as temperature- and field-dependent
studies may provide more insight into the interactions and motions which are associated

with NMR relaxation.

66

10‘:3 3’1

Relaxation Rate

10-2 s-1

Relaxation Rate

Figure 4.42 Plots of 31P 1/T1 vs. squared dipolar interaction. Data is displayed for (a)
compounds whose selenophosphate anion does not contain a P—P bond and (b)
compounds whose selenophosphate anion contains a P—P bond. The inset plot in (b)
Shows the four points closest to the abscissa on an expanded vertical scale whose unit is

—4 —l . .
10 S . In each plot, the abscrssa represents an approxrmate measure of the sum of the

squares of the individual dipolar interactions between a central 31P and the nearby
nuclear Spins. The dipolar sum calculation uses Eq. (2) and distances from known crystal

SU'LICIUI'CS.

(a)

 

3-4

 

 

T I

10

(DC)2 1029 rad4 T-4 s-4 A-5

(b)

 

.k
1

2-1

 

 

10‘ .

 

 

E 210 215

 

 

(DC)2 1030 rad4 T-4 s-4 13:6

67

 

 

Relaxation due to unpaired electrons.

In Figure 4.4(b), the rapid relaxation of group B2 (szCdP2S66 and KszPZSe6)

suggests that there is an important relaxation mechanism for these compounds other than
dipolar coupling. One reasonable possibility is relaxation caused by unpaired electrons.
Although none of the compounds in this study is intrinsically paramagnetic, there is still

the possibility that there are small but significant quantities of trapped paramagnetic

impurities. To first test this hypothesis, we resynthesized the fast-relaxing RbZCdPZSe6

with the idea that paramagnetic reactant concentrations could vary between reactant

batches, and thus produce different T1 in the final products. The T1 value for the

resynthesized szCdPZSe6 was 70 S, which is close to but not the same as its 80 s value

in the first synthesis.

A more definitive probe of the existence of unpaired electrons is electron spin

resonance (ESR)13 and Figure 4.5 displays the ESR Spectra of (a) K2CdP2Se6 and (b)
szCdPZSeé. Both samples Showed significant ESR signals, including a sharp Signal in

the g = 2 region. These data imply that unpaired electrons could contribute to the 31P

relaxation rate. In future studies, we will quantitate the concentration of unpaired

electrons and with a Simple relaxation model, we Should be able to determine whether

. . . . . . 31 .
thrs concentratron rs suffrcrent to explarn the enhanced P relaxation rate. It may also be

possible to use electron-nuclear double resonance to determine the chemical identity of

the paramagnetic impurities.

68

In summary, the 31P T1 varies by two orders of magnitude among different metal

selenophosphate compounds. For compounds whose selenophosphate anion does not
contain a P—P bond, there is an approximate positive linear correlation between squared
dipolar couplings and relaxation rate. If we neglect P—P coupling, this correlation also

holds for several compounds whose anion contains a P—P bond. These data suggest that

. . . . . 31 . .
P—metal ron dipolar fluctuations make a large contnbutron to P relaxatron In these

materials. Two of the compounds, szCdPZSeé and K2CdPZSe6, appear to have a

significant relaxation mechanism in addition to internuclear dipolar couplings. For these

compounds, unpaired electrons were detected by ESR Spectroscopy and may be the

. . 31 . . 31 . .
ongrn of enhanced P relaxatron. Finally, P solrd state NMR rs shown to be an
important tool for impurity detection in these compounds. However, because of the large
variation in relaxation rates among different compounds, spectra taken with short

relaxation delays may be very non-quantitative, and this can cause difficulties in

assignment of NMR peaks to particular synthetic products.

69

(a)

 

I I I l l
3410 3420 3430 3440 3450
Magnetic field (G)

(D)

My

 

F.

 

 

1 1 1 r 1
2750 2950 3150 3350 3550
Magnetic field (G)

Figure 4.5: X-band ESR spectra of (a) K2CdPZSe6 and (b) Rb2CdPZSe6 at 4.2 K. For

K2CdP2Se6, the Spectrometer settings were: 100 kHz modulation frequency, 0.276 G
modulation amplitude, 10.24 ms time constant, and 41.943 S sweep time. For

szCdPZSe6, the Spectrometer settings were: 100 kHz modulation frequency, 2.188 G
modulation amplitude, 5.12 ms time constant, and 41.943 8 sweep time. No ESR signals
were observed for an empty cavity.

70

REFERENCES

1.

10.

11.

P. Toffoli, P. Khodadad, and N. Rodier, Crystal-Structure of Silver

Hexaselenohypodiphosphate, Ag4P23e6, Acta Crystallogr., Sect. B: Struct. Sci.
34, 1779-1781 (1978).

H. Yun, and J. A. Ibers, Structure of PbPSe3, Acta Crystallogr., Sect. C: Cryst.
Struct. Commun. 43, 2002-2004 (1987).

J. Garin, and E. Parthe, The crystal structure of CU3PSe4 and other ternary normal
tetrahedral structure compounds with composition 13564, Acta Crystallogr. B28,
3672 (1972).

K. Chondroudis, T. J. McCarthy, and M. G. Kanatzidis, Chemistry in molten
alkali metal polyselenophosphate fluxes. Influence of flux composition on

dimensionality. Layers and chains in APbPSe4, A4Pb(PSe4)2 (A=Rb, Cs), and
K4Eu(PSe4)2, Inorg. Chem. 35, 840-844 (1996).

K. Chondroudis, and M. G. Kanatzidis, Flux synthesis of K2Cu2P4Seloz A

layered selenophosphate with a new cyclohexane-like [P4Se10]4- group, Inorg.
Chem. 37, 2098-2099 (1998).

K. Chondroudis, and M. G. Kanatzidis, Group 10 and group 12 one-dimensional

selenodiphosphates: AZWZSC6 (A = K, Rb, Cs; M = Pd, Zn, Cd, Hg), J. Solid
State Chem. 138, 321-328 (1998).

K. Chondroudis, and M. G. Kanatzidis, Complex Multinary Compounds from
Molten Alkali-Metal Polyselenophosphate Fluxes - Layers and Chains in

A4Ti2(PZSe9)2(PZSe-;) and ATiPSe5 (A=K, Rb) - Isolation of [P2369]4-, a Flux
Constituent Anion, Inorg. Chem. 34, 5401-5402 (1995).

CG. Canlas, M.G.Kanatzidis, D.P. Weliky. Inorg. Chem. 2003, 42, 3399.

R. Maxwell, D. Lathrop, D. Franke, and H. Eckert, Heteronuclear X-Y
Decoupling in the Mas-Nmr Spectroscopy of Inorganic Solids, Ang. Chem. Int.
Ed. 29, 882-884 (1990).

R. H. P. Francisco, and H. Eckert, Compound Formation and Local-Structure in
Ternary Metal- Phosphorus-Selenium Systems, J. Solid State Chem. 112, 270-276
(1994).

E. Gaudin, F. Boucher, M. Evain, and F. Taulelle, NMR selection of space groups
in structural analysis of Ag7PSe6, Chem. Mater. 12, 1715-1720 (2000).

71

12. R. K. Harris, "Nuclear Magnetic Resonance Spectroscopy," Wiley, New York
(1986).

13. A. Carrington, and A. D. McLachlan, "Introduction to Magnetic Resonance with
Applications to Chemistry and Chemical Physics," Harper, New York (1967).

72

CHAPTER 5
3"P HIGH TEMPERATURE NMR STUDIES OF METAL

SELENOPHOSPHATE COMPOUNDS

73

INTRODUCTION:

In Chapters 3 and 4, we have looked into the room temperature 31P NMR
properties of metal selenophosphates such as chemical shifts, chemical shift anisotropies
and longitudinal relaxation times. The most significant result is the chemical shift
discrimination between compounds with a P-P bond and compounds without a P-P bond.
We are now ready to study the syntheses of these compounds at high temperature using
high temperature NMR spectroscopy.

In this chapter we present the high temperature NMR (HTNMR) spectra and

results for the syntheses of representative metal selenophosphate compounds as studied in

Chapter 3 with different resulting anions in the final product : [P28e6]4- ( Ag4P2866 and
PbZPZSeé), [P4S‘310I4- (K2CU2P43610), [P364]; ( KPbPSe4 and K4Pb(PSe4)2), [P236914-
(CS4P2869 and Rb4T12P63625), [P2567]4- (Rb4Ti2P65625), [P366]- (RbPSC6). and

possibly [PZSe10]4-. [PSe6]- and [P28e10]4- are new selenophosphate phases.

74

MATERIALS AND METHODS

Preparation of Samples for In-situ NMR.

PzSe5 was prepared by reacting stoichiometric amounts of the elements in an

evacuated Pyrex tube at 300 0C for one day, followed by a cool-down to 50 0C over two

hours. Purity was assessed by X-ray powder diffraction analysis.
All samples were prepared in a pyrex tube which serves as an NMR rotor insert
and sealed under vacuum. All sample reactants have a total weight of about 100 mg. All

weighing are accurate within 0.001 gram sample.

Ag4PZSe6,l Pb2P28e6,2 were prepared by combining stoichiometric amounts of
the metal (Ag, Pb) with P2Se5 and elemental Se. For the Ag4PZSe6 synthesis, 0.04315 g
Ag was added to 0.04567 g P28e5 and 0.007896 g Se. For the Pb2P28e6 synthesis,
0.04144 g Pb was added to 0.04567 g PZSes and 0.007896 g Se.

KPbPSe43 was prepared by combining stoichiometric amounts of Kzse, P2565
and elemental Se: 0.02072 g Pb, 0.0228 g P2Se5, 0.007857 g Kzse and 0.007876 g Se.

K2Cu2P4Se10 4 was prepared from 0.3 mmol Cu, 0.9 mmol P, 0.3 mmol K2Se,
and 2.4 mmol Se: 0.011439 g Cu, 0.016725 g P, 0.028284 g K286, and 0.1137 g Se.

K4Pb(PSe4)2 3 was prepared from 0.15 mmol Pb, 0.225 mmol PZSe5, 0.6 mmol

K2Se, and 1.5 mmol Se: 0.01036 g Pb, 0.0343 g P2865, 0.03143 g Kzse and 0.03948 g

Se.

75

CS4P28e9 5 was synthesized from a mixture of 0.45 mmol P2Se5, 1.20 mmol
C8286, and 3.0 mmol Se: 0.02055 g P28e5, 0.04137 g CSZSC, and 0.023688 g Se.

Rb4Ti2P6$e25 5 was synthesized from a mixture of 0.2 mmol Ti, 0.4 mmol
P2Se5, 0.4 mmol RbZSe, and 2 mmol Se: 0.0036 g Ti, 0.10276 g PzSe5, 0.037484 g
RbZSe, and 0.05922 g Se.

The syntheses of pure RbPSe6 has been described in detail elsewhere6. In short,
pure RbPSe6 was obtained from stoichiometric mixture of Rnge/PZSe5/Se = 1/1/6 under

vacuum in a silica tube at 3500C for 10 days.

Physical Measurements.

High temperature solid-state nuclear magnetic resonance spectroscopy. The
static high temperature solid state NMR measurements of these compounds were taken
on a 9.4 T NMR Spectrometer (Varian Infinity Plus) using a high temperature double
resonance magic angle spinning (MAS) probe manufactured by Doty Scientific (DSI-

866). Samples were vacuum sealed in a pyrex rotor insert. The experiments were

. . o . o .
conducted by ralsmg the temperature by 50 C increments at a rate of 10 C/minute from

room temperature up to 400 to 6000C followed by a cool-down to room temperature at

the same increment. Most of the samples studied were then subjected to a repeat cycle of

31F HTNMR in order to observe the 31P NMR spectra of the formed final product in the

76

first cycle. Overall, the time it takes from heating to cooling down the sample is about 6-
12 hours compared to an oven synthesis which takes several days.
Static NMR spectra are taken after equilibrating the sample at the specified

temperature for 20 minutes. Bloch decay spectra were taken with the excitation/detection

channel tuned to 31F at 161.39 MHz, a 5.75 [13 900 pulse, and a relaxation delay between

1 and 1000 5. Each spectrum was processed with 100-1000 Hz line broadening and up to

a 10th order polynomial baseline correction. The spectra were referenced using 85%

H3PO4 at 0 ppm. At some temperatures, a series of spectra were acquired with delay

times between 0.1 and 1500 s. The significance of pulse delay is to ensure that the NMR
spectra is quantitative. Because of the large variation in relaxation rates among different
compounds, spectra taken with short relaxation delays may be very non-quantitative, and
this can cause difficulties in assignment of NMR peaks to particular synthetic products.

Magic-angle spinning NMR spectra are also taken for some samples. There is a
difficulty however in achieving higher than 1000 Hz spinning frequency in using the high
temperature probe. This is due to weight imbalance about the spinning axis. NMR spectra
shown in this chapter are static spectra unless otherwise indicated.

Room temperature NMR. The room temperature solid-state NMR measurements
of products after cool-down were taken on a 9.4 T NMR Spectrometer (Varian Infinity
Plus) using a double resonance magic angle spinning (MAS) probe. Samples were spun at

7-12 kHz using zirconia rotors of 4 mm outer diameter. Bloch decay spectra were taken

with the excitation/detection channel tuned to 31F at 161.39 MHz, a 4.5 as 900 pulse, and

a relaxation delay between 5 and 15,000 s. Each spectrum was processed with 100-1000

77

Hz line broadening and up to a 10th order polynomial baseline correction. The spectra
were referenced using 85% H3PO4 at 0 ppm. The “doublet” final product observed in the

synthesis of K2Cu2P4Se10 was verified if the peaks are due to P-P scalar couplings by

taking MAS NMR measurements on a 7 T NMR spectrometer. If the two peaks are
scalar-coupled, the peak frequency difference between the two peaks should be the same
in both magnetic fields and will have a P-P type of bonding. If the peak frequency
difference between the two peaks is proportional to the strength of the magnetic field,

then the two peaks are two magnetically inequivalent peaks with no scalar coupling.

78

RESULTS AND DISCUSSION

[P25e6]4-. The room temperature 31F CS of selenophosphates containing
[P28e6]4_ occur downfield in the 20 to 100 ppm range as discussed in Chapter 3. One of
the compounds is Pb2P28e6 which has two P atoms which are crystallographically

equivalent and have the same 31F chemical shift (CS) at room temperature at 29.1 ppm.7

Figure 5.1 presents the high temperature static variable temperature NMR spectra of the

syntheses of Pb2PZSe6 from starting materials (Pb + PZSe5 + Se) with 50-100 degrees

Celsius increment from room temperature to 600 0C followed by cooling down with the

same increment to room temperature. The same sample is then subjected to another cycle

of 31P HT NMR measurements.
During the first cycle, the appearance of a single 31P peak starts at about 300 0C
at 122.7 ppm, which indicate the presence of a P+4 species or a compound with P-P type

of bonding. Subsequent increase in temperature up to 600 0C shows considerable

lineshape narrowing which maybe due to increased motional processes in the melt (the
broadening effects of dipolar coupling and chemical shift anisotropy are effectively

averaged by motion resulting in increased resolution of the resulting spectrum). The

simulated lineshape at 600 oC exhibits semi-Lorentzian lineshape (~50%

Gaussian/Lorentzian) upon deconvolution with a full-width at half maximum of 1376 Hz.

Though the melting point of pure Pb2P28e6 is 784 0C, the reaction formation of

79

Pb2P28e6 shows liquid-like NMR spectra at 600 OC due to motional narrowing. Upon
cooling down to 300 0C, linebroadening occurs and an upfield peak appears close to the

room temperature 31P CS of Pb2P28e6. Figure 5.2a and 5.2b shows a plot of the ramp-up

temperature and ramp-down temperature versus chemical shift respectively in the

temperature range 300 0C to 600 OC wherein one single sharp peak is observed. Figure

5.2a has a slightly higher slope than Figure 5.2b. The slope in Figure 5.2a involves the
start of the synthesis process and involves different chemical reactions and hence will

show a different chemical shift dependence on temperature. Upon cooling down to 300
0C, a broad feature starts to appear upfield. At 200 oC considerable broadening appears
and the central peak has a chemical shift of about 22 ppm, which is close to the room
temperature 31P chemical shift of PbZPZSe6.

The second cycle shows about the same information as the first cycle. Figure
5.2(c) shows the plot of the temperature dependence of the 31F CS from 300 to 500
degrees Celsius with about the same slope as Figure 5.2(b). This might be expected

because if Pb2P25e6 has formed during the first cycle (particularly on the ramp-down in

temperature in the first cycle), the second cycle will just show the melting of Pb2P28e6

and will show the same chemical shift dependence on temperature independent of cycling
with slight changes due to hysteresis. The lower signal to noise observed are due to

mechanical difficulties arising from having the sample pyrex insert slipping out of the

NMR rotor. The final product after the syntheses and re-cycling produced pure Pb2P25e6.

8O

 

W
N 1
h an in” "’1 l 111k 21'”
will 1.11! W i

lilting} 1111*W'11W

j cc
Willi“jlii‘lilwl‘n’lal" ”l"

j 500 0C

Mj‘lj’ll W “W

‘ 600 0C
11111118]LW‘W‘Wlilll
‘ 5 00 °Cd

J filth»
'0le “M lrrl'i'i’vfluv 1‘1 “WIN

 

' 400 °Cd
‘ l
W‘A‘n [#9 “I'M“; *lw'lgl‘if
r71T1 II ‘
500 O _500
pp”)

Figure 5.1. 31

[F1 11]
500 0

300 oCd

/ 200 °Cd

.1 .171. . ”.1, 1
«I WA! #114 It 4“! K‘L‘I MY 1‘ ”A! W"

200 °Cr
I11 ’1

final product
I L Pd: 25008
[I 11 1

-500
ppm

P variable high temperature NMR spectra of the synthesrs of Pb2P2Se6

(note d—rampdown in temperature, r = recycled temperature). Note phasmg difficulty in
the spectrum at 500 0Cd.

81

(a)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A 150 ~
y = 0.0726x + 100.58
145 “ 2
g 140 . R =o.9943
5 135 ~
g 130 —
g 125 «
o 120 I I I M
0 200 400 600 800
Temperature ln degrees C
(b)
... 150
§145~ y=0.(2)55x+111.65
‘7 R =o.9947
g 1401
m
a 135 .
5 130.
.C
O
125 , r .
O 200 400 600 800
Temperature in degrees C
(C)
138
E 136 _ y = 0:62X 4' 104.07
a
e 134 . R =0.9486
g 132 ~
(I: 130 «
g 128 . .
g 126 .
5 1241
122 I T V I I
0 100 200 300 400 500 600
Temperature In degrees C

 

 

 

Figure 5.2. Chemical shift dependence on temperature in the synthesis of PngSe6.

(a)Ramp-up temperature increase versus 31F chemical shift of Pb2P2Se6 from 300 to 600
degrees C. (b) shows the rampdown dependence of chemical shift versus temperature

from 600 to 300 degrees C. (c) shows the temperature dependence of 31F chemical shift
during ramp-up from 300 to 600 degrees C.

82

 

Another compound is Ag4PZSe6, which has two crystallographically inequivalent
. 31 . . 7
P atoms wrth P chemical sh1fts at 77.6 and 91.8 ppm at room temperature. In Chapter

3, the syntheses is accompanied by the formation of a crystalline impurity Ag7PSe6

which has a chemical shift of — 51.9 ppm. The formation of two products makes this

system interesting to study using HTNMR. Ag7PSe6 8 has the form Ag7(PSe43_)(Se2-_)2
and its CS is distinct from those found for compounds containing [PZSe6]4—. Figure 5.3
shows the 31F high temperature static NMR spectra of the syntheses of Ag4P28e6 from
its starting materials: Ag + PZSe5 + Se.

During the first syntheses cycle, the formation of a P+5 or a species with no P-P
bonding (the formation of Ag7PSe6) is already apparent at the upfield region (~ - 60.8
ppm) along with the formation of a P+4 or a species with a P-P bonding with a broad

central peak at 128.8 ppm (the formation of Ag4PZSe6). At 350 0C and higher, a splitting

can be seen for the downfield peak with a peak-to-peak difference of ~28 ppm (which is

about twice the peak to peak difference observed at room temperature) which indicates

the formation of the two inequivalent P atoms in Ag4P28e6. There are splittings because

the two P atoms are magnetically inequivalent even at these high temperatures. The

splitting is not obvious in some spectra (500 0Cd for example) due to low signal to noise.

The appearance of the apparent splitting of the two P atoms of [P28e6]4- at high

temperature is not commonly observed in the synthesis of other compounds presented in

83

 

 

IO

this work. Usually, at high temperatures the P atoms belonging to different species

undergo fast chemical exchange and a single peak is observed. It is possible that the

formed Ag4PZSC6 is in the solid state. Future 2-D NMR correlation experiments will

confirm if the two peaks observed at high temperature are due to the two magnetically

inequivalent P atoms in [P28e6]4- (P 1-P2) or are due to two independent P-P sites (P l'Pl
and Pz-Pg). A 40-50 ppm downfield difference is observed for the chemical shifts
between the room temperature and high temperature 31F spectra. Figure 5.4 shows the
temperature dependence of 31F chemical shifts corresponding to the 2 P peaks of
Ag4PzSe6 (P1 and P2) and the P peak of Ag7PSe6 (P3) in the temperature range
300—600 0C. The lineshape of the downfield peak is appears broad at high temperature

with a full width at half maximum of around 60 ppm at 600 0C, but careful inspection

shows that the superposition of two peaks makes it appear broad. The individual peaks

actually fit a Lorentzian lineshape. The chemical shifts are in the slow exchange limit.

The high temperature NMR spectra show the simultaneous formation of [P28e6]4- or P+4
species (that eventually forms Ag4PZSC6) and [PSe4]3- or P+5 species ( that eventually
forms Ag7PSe6 impurity). The integrated intensities of the spectrum at 600 0C shows that

the [PSe4]3_—type species in the upfield region is 15% of the total peak integration. The

room temperature NMR spectrum of the final product shows approximately 18%

84

Ag7PSe6 formed which is consistent with the concentrations in the high temperature

spectra.

A splitting can also be observed for the Ag7PSe6 upfield peak at lower
temperatures (300 0C and lower). A possibility for this to happen is maybe due to the
known phase transition of Ag7PSe6 from its B-phase (with cubic space group F43m) to its
y-phase (with cubic space group P213) upon heating to 162 0C.8 An interesting follow-up
experiment for this system would be substitution of P28e5 with elemental P as starting
material, which may inhibit the formation of the [PSe4]3-/P+5 impurity. This is because
the phosphorous in P2Se5 is already P+5 (see Chapter 1, Scheme 1) and by replacing this

with elemental P it mi ght be possible to favor formation of PM.

85

1

1 11111 180 0C

”‘1 1111 ‘l' J11}. ‘1'

11M 1‘,

1 ’

‘1‘51‘1‘1‘! W” 111‘ 1‘11” ‘1
11 1 600 °C
11 1

. ’ 11
i1 l \1 11
“(Viva I 'Vu' 1 V 11.! 1 l
V la‘ ‘ : l ~V' l1, ,‘1
W 111' ‘1"? i '1

500 °Cd

$101114 11111111111.1l11ll‘l1111,11

 

IIIII'IIIII

500 0 -500
ppm

Figure 5.3. 31

 

11 1 170 °Cd

1 1
11
‘J.'é1'i1‘111‘«1IJ L11 't ‘

1 111 1141-." 111911.}11‘151

11

11,1411 j E111‘1l1f
11.111111111111111

150 0Cr

“1 j 300 °Cr
' J ‘1 11 i1 1
WWW) i 11 I‘VM‘ WM

1}

11 500 °Cr
.. 1. . 1. ,
111,101,13111’1‘Jlll' WJWJ 1111111111111J‘111 11,1

.“1'1

After HTNMR
pd=3 800$

 

 

.11.

 

 

[IIMI IIII

500 0 -500

ppm

P high temperature NMR spectra for the syntheses of Ag4P28e6. (note:
=rampdown in temperature; r = recycled temperature).

(a)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CS vs Ramp-up Temp
E f I
a 100 1
"’ 6P1
g 50 1
a I P2
5 0 « APS
5 ‘50 '1 ‘ ‘ A ‘
o '100 r 1 r l
200 300 400 500 600
Temperature In degrees C
(b)
CS vs Ramp-down Temp
A 200
E 100 1 I . P1
5 50 1 I P2
E o « A R3
-50 -
g A A A
o -100 . , . 1
200 300 400 500 600
Temperature In degrees C
(C)
cs vs Ramp-up Temp Recycled
200
E 150 ‘ = ‘
fi 100 ~ 9 P1
2’: 50 I P2
5 0 ‘ A P3
5 ‘50 ‘ A A A
-100 , Y T .
200 300 400 500 600
Temperature In degrees C

 

 

 

Figure 5.4. Temperature dependence of 31P chemical shifts corresponding to the 2 P
peaks for Ag4P28e6 (P1 and P2) and the P peak for Ag7PSe6 (P3) in the temperature

range 300-600 0C. (a) First cycle ramp-up in temperature; (b) Rampdown in temperature;
(c) Ramp-up in temperature in the second cycle.

87

[PSe4l3— : The room temperature 31F CSs of selenophosphates containing
[PSe4]3_ occur upfield in the —l 15 to —50 ppm range. Their CSA are < 80 ppm, which is
significantly smaller than those observed for [P28e6]4fi, and likely reflects the high
tetrahedral symmetry of [PSe4]3_.

One of the compounds, KPbPSe4 has a 31F room temperature chemical shift of
—74.3 ppm. Figure 5.5 shows the 31F HT NMR syntheses of KPbPSe4. At the start of the
syntheses up to 400 0C, the spectra are characterized by broad peaks in the upfield region
(— 40 to — 113 ppm) which indicate the presence of P+5 species or compounds with no

P-P bonding. At 500 OC and above, a sharp single Lorentzian peak is observed due to fast

motional processes in the melt. Upon cooling, broad peaks starts to appear again starting

from 400 OC and at 300 to 200 0C the peak starts to have a static powder pattern. At 200

0C, an axially symmetric powder pattern is observed which is consistent with previous
CSA principal value calculations for this compound (see Table 3.1 in Chapter 3). For an

axially symmetric powder pattern, 511: 822 (see Chapter 2), the CSA prinicipal values

calculated are 511 = -53 ppm, 522 = -54 ppm, and 533 = -118 ppm. The width of the

powder pattern is also consistent with the CSAS calculated in Table 3.1. In comparison

with the Ag/P/Se melt at high temperature discussed above, there is a clear distinction in

chemical shifts between formation of [P28e6]4- or P+4 type species which have downfield

88

chemical shifts and formation of [PSe4]3- or P+5 type species which have upfield

chemical shifts.

The second cycle of HTNMR shows a similar result as the first cycle. The melting

point of KPbPSe4 is around 500 0C which agrees with the high temperature 31P NMR
spectra.3 The final product distribution shows almost pure KPbPSe4 with about 5%

. . . . . 9
impurities Wthh are mostly unknown except for a llttle presence of K3PSe4.

89

gel 200 °C

' )
bl.“,I‘Vl'lllf,'tlflill l ' M

NW 300 °C

will i...

l N
W 400 °C
MlflW’WA WWW

 

A 300 0Cd
“4.1le $1 Ania l/WUWF
"l 200 °Cd

A
3 fat" Nl la ,
m . ”llvh‘W,

El
l

\500 °Cr
WNW MM

 

550 °Cr

400 °Crd

h.
.4 "WWW

l
A} X300 °Crd
W MW

 

 

300 °Cr
500 0C k 200 oCl'd
lv'l‘uy‘ «will. Wt
550 °C 400 °Cr Final pdt
pd: 13008
M m,” V: 12 kHz
”Aug
500 °Cr T 7% 1 7 I 1.260]
400 °Cd ppm
T F l l T l I T ‘ T I I I ‘ r 1 I
0 —200 0 “200
ppm ppm

Figure 5.5. 31P HTNMR syntheses of KPbPSe4. (note: d=rampdown in temperature; r =
recycled temperature).

90

Another compound that has a [PSe4]3- anion and is isoelemental to KPbPSe4 is

K4Pb(PSe4)2. K4Pb(PSe4)2 has a structure consisting of two [PSe4]3_ ligands which

bridge adjacent Pb atoms. All P atoms are crystallographically equivalent, leading to a

single C8 of —95.5 ppm and a resolved P—Se scalar coupling of 432 Hz at room

temperature. Figure 5.6 shows the 31F HTNMR syntheses of K4Pb(PSe4)2.
During the initial ramp-up at 200 0C, the 31F spectrum is already characterized by

a relatively sharp peak. The sharp Lorentzian peaks and low T1 at relatively lower

temperature is maybe due to the occurrence of increased motional processes in the melt.

This is characteristic of a flux reaction where the starting material is composed of excess

alkali selenide (K286) and Selenium which lowers the melting point of the reaction

mixture because the alkali selenide and selenium mixture itself has a relatively low

melting point. Figure 5.7 shows the dependence of 31F chemical shift with temperature.

During the first ramp-up in temperature (Figure 5.7a), the slope is not linear which is due
to chemical reactions happening in the melt. During cool—down and recycle, the SIOpe is
relatively linear (Figures 5.7a, 5.7b) and close agreement of slopes with slight hysteresis
during ramp-down in the first cycle and the recycle in temperature suggests presence of a

common product at both cycles. The final product is about 95% pure with little amounts

of K3PSe4 ( chemical shift at —1 13 ppm) and other unknown impurities.

91

g

200 °C h 300 °Cd 500 °Cr

300 °C fi 200 °Cd 300 oCrd

400 °C A 200 °Cr 22 oCrd
M ' MW

after HTNMR
v = 7 kHz

A
F

F-

500 °C

 

FF
%—

l

l

 

llll llllll

 

 

200 0 ~250
‘ 400 °Cd ‘ 400 °Cr ppm
1 I l l I l l l l l l I W1 1 l l l I l
200 O -250 200 0 -250
ppm ppm

Figure 5.6. 31F HTNMR syntheses of K4Pb(PSe4)2. (note: d=rampdown in temperature; r
= recycled temperature).

(a)

 

 

 

     

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

 

Chemical shift vs Temp First Cycle

-45
’s‘
a .50 - + ramp-up
E d
a _55 - +ramp- own
.§ .
E _ 41 67 5 ------ Linear (ramp-
2 '60 - y—0.0 x- . down)
0 R2 = 0.9859(ram p-down)

'65 t

150 350 550
Temperature in degrees C
(b)
Chemical shift vs Temp Recycle

-45
A —-O—ramp-up
E y = 0.0311x - 64.61

- 2
:2- ~55 . x”,
"" —————— Linear ram u
g 60 y = 0.0395x - 68.05 ( p- p)

- -( 2
g R = 1(ramp-down) —————— Linear (ramp-

-65 . down)

150 350 550
Temperature In degrees C

 

 

 

Figure 5.7. Shows the temperature dependence of 31F chemical shifts of the synthesis of
K4Pb(PSe4)2. (a) First cycle in temperature. (b) Recycled temperature

93

4- . 4— . . .
[P453101 : The unique [P4Se10] was discovered in K2Cu2P4Se105 and 1t

. . . 4— . . .
essentially dCI‘lVCS from the fusron of two [P28e6] followed by the elimination of two
Se2_ anions. As a ligand, it possesses eight terminal Se sites available for coordination

(cf. Figure 3.1(c)). KzCu2P4Se10 has two crystallographically inequivalent P atoms in its

crystal structure. Nevertheless, a careful examination of the local environment around

these two P atoms reveals that they are in fact chemically and magnetically equivalent

and result in a single 31P NMR isotropic peak at 55.7 ppm. [P4Se10]4- is similar to
[P28e6]4_ in that it contains P—P bonds and tetravalent P. These structural similarities
help to explain the observation that the CS of KZCu2P4Se10 (55.7 ppm) is within the
[P28e6]4_ CS range. The CSA of K2Cu2P4Se10 (138 ppm) is also close to the CSA found
for [PzSe6]4-. In KzCu2P4Se10, the resolved P—Se scalar coupling is 694 Hz.

The 31? high temperature NMR synthesis of K2Cu2P4Se10 is shown in Figure
5.8. Upon increasing the temperature to 400 0C, four distinct chemical shifts are observed
while at 450 0C, there is one peak perhaps due to formation of a new species or

occurrence of fast exchange between peaks observed at 400 0C. At 500 0C, two dominant

peaks are observed with peak centers at 90.2 and 40.3 ppm. Broad peaks are observed

during cool-down. The final product has three components with isotropic chemical shifts

at 56.0, -l7.2, -18.9 ppm. The peak at 56.0 is the K2Cu2P4Se10. The two peaks upfield

94

are two magnetically independent phosphorous sites, as determined from further NMR

measurement on a 7 T NMR magnet. This unknown “doublet” peak can also be found in

flux reactions using K25e (see Chapter 6). Figure 5.9 shows the expanded 31F MAS

NMR spectrum of the final product formed.

 

”W" WWW“ WWWW will (fulfills “W Wl'lll‘lllm W I}

 

 

250 °C a 500 .C
l I I'M
ling“ ’I(..'I(IL; Willi l . . (I... (11“.! Wm (IlfMIIlIl(IIllll" I'IIi‘(I(II(.(IIMI
l 400 °C ( final pdt
l
Illll 'l'll
lIlIlIIIIIJII'I/WV1.1....(IIIIHWII WW I I I I
5(l)0T I I I (I) I I I l—sbo 5A0 pém -5l)o

9131“

Figure 5.8. 31p HTNMR syntheses of K2Cu2P4Se10

95

**

 

 

 

III

llTllTllllITlll'lllllllllIlllllllllllllll
200 150 100 50 0 —50 «100 -150 ~200
PPm
Figure 5.9. 31F MAS NMR spectrum of the final product formed during the syntheses of

 

KZCu2P4Se10. The (*) indicate isotropic peaks. Two impurity peaks are present at -17.2,
-l8.9 ppm. The spectrum was taken with 8 kHz spin rate and 20003 pulse delay.

In order to isolate K2Cu2P4Se10, the final product mixture is washed with DMF

to remove excess flux and selenium.

Figure 5.10 illustrates the high temperature 31F NNIR spectra of pure
K2Cu2P4Se10 as the starting material. The purpose of this experiment is to have an idea
which peak corresponds to KzCu2P4Se10 in the high temperature syntheses in Figure 5 .8.

Upon heating up to 400 0C, the spectrum features a broad powder pattern with a
center around 50—60 ppm. At 500 OC and higher, the spectrum becomes sharp due to

melting of K2Cu2P4Se10. The melting point of KZCu2P4Se10 is 470 OC.4 The chemical

96

shift at 500 0C is 95.4 ppm and at 570 0C the chemical shift is 107.2 ppm. Upon cool-
down, the same pattern arises showing the broad powder pattern feature at 300 oC. Based
on this experiment, it seems that the K2Cu2P4Se10 peak in Figure 5.8 at 500 0C is the

more downfield peak at around 90.2 ppm. K2Cu2P4Se10 remain unchanged after the

cycle.

97

I“, 200 °C

I‘ l #‘l‘
tr‘W‘W' * M III III.

I”; 300 °C

III.
("{I‘I'M'H‘IIAIJ ”KW-r ‘2 ”Fl I lma

i

II.

”I 400 °C

I‘li' '} 1L! . .(' H
W i 51“.”.(5l “it“ (if IA (”(i1((|(w(~| (WI

, J 1
WWW" MWWW

I‘, 570 °C

 

[ l l l l I l l I l

500 O -500

ppm

6 - 500 °Cd

J

 

(( 400 °Cd
.l
l i
\er‘ w. ‘ W.;.“ i“:
A 300 oCd
A 200 °Cd
w ‘I
{If Ll
l
P“(;(!(x't')rv'1~w M ”\Wfly‘ \Wn’“ “A Ire/(If.
after HTNMR
pd: 13008

 

 

“Ii. m

 

[llllTlllll
500 O -500

PPm

Figure 5.10. 31F high temperature NMR spectra of pure KZCu2P4Se10. The

K2Cu2P4Se10 remain unchanged during the cycle. (note: d=rampdown in temperature; r

= recycled temperature).

[PZSe9]4_ and the new [P25610143 Two compounds, CS4P2869 and
Rb4Ti2P6Se25, contain the [P28e9]4_ anion. This ligand is structurally composed of two
Se-sharing [PSe4]3_ subunits. Rb4Ti2P6$e25 is composed of 2 P peaks corresponding to
[PZSeg]4' at —34.6 and -47.6 ppm, and 1 P peak corresponding to [P28e7]4-. We will
discuss the high temperature syntheses of Rb4Ti2P68e25 later in the [PZSe7]4- section.

Figure 5.11 shows the 31F HTNMR syntheses of CS4P2569. During initial heating
to 400 0C, two distinct 31? peaks are observed. Further heating to 490 OC gives a single

peak due to fast exchange between the two species formed. The peak at 490 0C is also

Lorentzian which indicates fast motional processes characteristic of the melt. The peaks

can actually be deconvoluted as shown in Figure 5.12. Upon cool-down to 300 OC, two
31F peaks are present. The recycling provides the same information. The final product
formed shows two isotropic peaks at —40.2 (the CS4P2569), and —52.8 ppm. The
CS4P23€9 formed is 25% and the peak at —-52.8 ppm is 75%. In furnace synthesis, pure

CS4P25€9 is always made using the same stoichiometry (but different heating/cooling

profile — see Chapter 3). A relatively fast cooling profile may have kinetically stabilized

the formation of the anion at —52.8 ppm. Further study of the peak at —52.8 ppm at room

temperature has revealed a new Cs-P-Se phase in CS4P28610. The P28e104_ is structurally

99

composed of two Se-Se-sharing [PSe4]3- subunits similar to the structure of CS4P2310.10

Figure 5.13 shows the crystal structure of CS4P28610.

The deconvoluted peaks using a Lorentzian simulation in Figure 5.12 show the
same trend, with the peak at ~32 ppm having 24.65% relative integrated intensity

compared to the peak at —35 ppm having relative integrated intensity of 75.35%. We can

therefore assign the more downfield peak at high temperature to be [P28e9]4- and the
upfield peak to be [P28e10]4-. The high temperature linewidth of [P28e9]4- is 840 Hz and

that of [P23e1014' is 630 Hz.

100

3‘3: 100 °C

'H I- H"
HIHHHIMHH J! I‘M)“ I
H 150 °C
”’11” J I!
J'I {H1
3»st II. 3H" “I ~

III 221°C
: H nil.
L H! b IdeM H
3W3” HHH L‘

I 300 °c

HH‘I‘HHHH HHHHH

Jar

[IIIIIIIII'

250 O ~250
DD”1

 

I 400 °Cd
I3

IIHI 300 °Cd
\‘n
3H” ”H?" H" HH‘HH

 

“HI [It'lJ W' UH.
HHIH 200 CI‘
H3: II, I h H I
i ”I'IIIHHAIIII/r' I'll/l Hl‘l
"I I i HI
H 300 °Cr
I: "H“HHI l III-I.
H3'HHH'H1HH" W
H
H 300 °Cr
HIM-H MHHHM
I I I I I I I I I I I
250 O -250
DD”1

400 °Cr

l
I 500 0Cr
M
400 °Crd

.__.H.._.

300 °Crd

 

HHHHHHNM HHHH'HIHIH

33200 °Crd

HHIVHI‘III I‘
[H 3 “HM. I
II~ L'HJ’JII‘IA!‘ I III c hII'I IVH‘IIJ

RT final pdt
V: 12 kHz

.JHL

IITIIIIIIIT

43H

 

 

250 O -250

pp”1

Figure 5.11. 31F HTNMR syntheses of CS4P2369. The final product is composed of two

different crystalline products- CS4P2869 and a new phase CS4P28610. (note: d=rampdown
in temperature; r = recycled temperature).

101

 

 

l I
-15 -25 -35 45 -55
ppm

Figure 5.12. 31F deconvoluted NMR spectra of the syntheses of CS4P2369 at 400 0C
during rampdown. Two peaks at -32 and —35 ppm are present at high temperature. The
integrated intensities are : -32 ppm is 24.65%; and —35 ppm is 75.35%.

102

0)

 

O
01

 

0}

A

Figure 5.13. The crystal structure of CS4P28€ 10 shown down the a-axis. The dark-colored
atoms are Se atoms, the light-colored atoms are P atoms, the light-colored atoms with no
bonds are the Cs atoms.

103

[P25e714'z Rb4Ti2P6Se25 also contains the [P28e714‘
3_

composed of a two [PSe4]
is within the CS range of [PSe4]3-

Figure 5.14 shows the 3

3H 33 200 °C
W h

H3 HH
HWMHHNHH HH/H HH‘HHVEH’H‘IA
3
3
I33 300 ”C
l
3

H'HH‘WI‘HVWH II“ 1.3%1‘3H‘H‘HW

|
3
{3 400°C
31
IH‘HH‘HH" HWW'H’HV/ HHWHHMHHH

H

 

I3 500°C
Wl/HH W
H 400 °Cd
33
«W WW.»
I Izgd I I I OI Iszso
PP"1

H 300 °Cd
H

“J L.....
H 200 0Cd
33

W L...
I

I3 200 °Cr

3 3

Mvwwr/ HHW ”an
IHI 300 °Cr

MJIHHM

3400 °Cr

.JHL.

IIIIIII(I)IIII2IS

250

 

ppn1

ligand which is structurally

sharing a common Se. The [P28e7]4_ CS is —67.7 ppm and
. All the P atoms are in the P+5 oxidation states.

1P high temperature NMR syntheses of Rb4Ti2P6Se25.

33 500 °Cr
I3 3
iflhvde/ Hthumnuuw
H 400 °Crd
3‘
H 300 °Crd
I
II 200 °Crd
ML“...
final pdt
pd=2000$

v: 8kHz

IIIIIIIITj—FII

250 O -250
PP”1

 

 

 

Figure 5.14. 31F High temperature NMR syntheses spectra for Rb4Ti2P6Se25. The final

product formed is pure Rb4Ti2P6Se25. (note: d=rampdown in temperature; r = recycled

temperature).

104

A single peak at high temperature starting from 400 0C is apparent which is

indicates fast exchange between species present. The lineshape at high temperatures is
also Lorentzian which indicates high motional processes happening in the melt. There is

an interesting observation about the linewidths at high temperatures. The linewidths at

relatively lower temperatures are slightly sharper than the linewidths at 500 0C as shown

in Table 5.1.

Table 5.1. Shows the trend of the change in 31P NMR linewidth with temperature.

 

 

 

 

 

Temperaturea F ull-width at half
0 C maximum(FWI-IM)
500 3982
400 37 12
300 3330
200 1909

 

 

 

 

a .
the temperatures shown are the temperatures during the recycle rampdown

This is may be due to different extent of exchanging species happening at each

temperature wherein equilibrium might exist as:
4- 4-
132367 :=‘ P2369 (l)

The sharpness of the peaks is related to the relative molecular tumbling motions of the

species present in the melt. The higher the motions, the sharper the peak is. At lower

temperatures the 31F linewidths are slightly sharper, which might be due to higher
content of [PzSe7]4- anion which has relatively higher symmetry and smaller molecular

size compared to the [P28e9]4- anion in equation (1) and hence lower CSA (see Chapter

3, Table 3.1) and faster molecular tumbling motion. This can also be thought of in terms

105

of Brownian motion wherein motion is dependent on particle size and temperature. As

. . . . 31
temperature increases, Brownian motion increases. Hence we expect a sharp P NMR

peak at higher temperatures, which is not the case in this example. Molecular size might
be more important in this case. The presence of relatively smaller molecular sizes at
lower temperatures might explain the slight sharpness in peaks relative to the peaks at
higher temperatures. Another reason might be different viscosity at different
temperatures, but there is no direct viscosity measurements done on these kinds of
samples. Eckert et al have also observed this broadening effect at higher temperatures in

their study of P-S-Se glasses which they say is due to depolymerization of the glass
matrix followed by radical formation occurring appreciably at very high temperatures.11
The formation of the radical electron may have increased the rate of transverse relaxation
(1/T2 oc FWHM), and resulted in line broadening.

Upon cooling and recycling, the chemical shift of the peaks is in the downfield

region relative to 0 ppm. This does not follow the trend observed for chemical shifts of

P+4 and P+5 species. The final product formed is almost pure Rb4Ti2P6Se25. There is no

reasonable explanation for the behavior of the chemical shift at high temperature
compared to the room temperature product. It might be possible that the high temperature

melt and the final products are not the same. In the future, HTNMR will be run on pure

Rb4TI2P68625 to provide further information.
[PSe6] ': The new [PSe6]- (APSe6 where A: K, Rb) anion is composed of a PSe4

tetrahedra with diselenide linkages that forms polymeric chains of this anion. The 31F

106

 

NMR room temperature chemical shift of RbPSe6 has a chemical shift of 4.3 ppm. This
shift is ~ 40 ppm higher than a typical shift of [PzSe9]4', a comparable discrete unit with
bridging Se. It has a 31F chemical shift anisotropy (CSA) of ~115 ppm which is
intermediate to those observed for [PSe413’ (CSA ~65 ppm), and [P25e914' (CSA ~200
ppm).6

Figure 5.15 shows the 31F high temperature NMR spectra of pure RbPSe6 as the

starting material. This compound melts around 315 0C based on the differential thermal

analysis (DTA) data for this compound. The spectra in Figure 5.15 show that below 315

OC, broad powder patterns are present, and above 315 0C, a sharp peak is present due to

. . . o . 31 .
increased motional processes in the melt. At 350 C, the static P spectrum has a Single

resonance with —21 ppm peak chemical shift and 4 ppm full-width at half maximum

(FWHM). The shift is typical of PSe4-type bonding, and the narrow linewidth is
diagnostic of rapidly tumbling small molecular species such as [PSe6]- ring-based

molecules, which would result from depolymerization of [PSe6]- infinite chains.6

107

 

‘H3 350 °c
‘33 200 °C
33%
I. 3.33 3 H3
Hail IlslH HH 3 II It“
HHII VHNHH H 'IHHHHHIHHHHIHHHHI H1333“
III 260 °C 250 °Cd

HIHHHHHVHII HHHIHHHHHHWIH II IIIIIIIIIIIIII II III33 '33II3I3‘I III II III H3" HH

3
250 O —250 250 0 -250
PPm ppm

 

 

Figure 5.15. 31F high temperature NMR spectra of pure RbPSe6 as starting material.
(note: d=rampdown in temperature).

In summary, the main result of this study is that the 31F CS at high temperature of

the syntheses of metal selenophosphates is correlated with the final reaction products

except for the Rb4Ti2P6Se25 compound.

108

REFERENCES

(l)
(2)
(3)

(4)
(5)
(6)

(7)
(8)

(9)

(10)

(11)

Toffoli, P. F.; Khodadad, P.; Rodier, N. Acta Crystallogr. 1978, B34, 1779-1781.
Yun, H.; Ibers, J. A. Acta Crystallogr. 1987, C43, 2002-2004.

Chondroudis, K.; McCarthy, T.; Kanatzidis, M. G. Inorg. Chem. 1996, 35, 840-
844.

Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1998, 37, 2098-2099.
Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 5401-5402.

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

Francisco, R. H. P.; Eckert, H. J. Solid State Chem. 1994, 112, 270-276.

(a) Kuhs, W. F.; Schulte-Kellinghaus, M.; Kramer, V.; Nitsche, R. Z. Natwforsch.
1977, 32b, 1100-1101.

(b) Maxwell, R.; Lathrop, D.; Franke, D.; Eckert, H. Angew. Chem. Int. Ed. Engl.
1990, 29, 882-884.

(c) Gaudin, E.; Boucher, F.; Evain, M.; Taulelle, F. Chem. Mater. 2000, 12,
1715-1720.

((1) Evain, M.; Gaudin, E.; Boucher, F.; Petricek, V.; Taulelle F. Acta Cryst.
1998, 854, 376-383.

Dickerson, C.A.; Fisher M.J.; Sykora, R.E.; Albrecht-Schmitt,T.E.; and Cody,
J .A. Inorg. Chem. 2002,41, 640-642.

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

Mutolo, P.F. Ph.D. Dissertation, University of California at Santa Barbara, 2000.

109

CHAPTER 6
31P HIGH TEMPERATURE NMR STUDIES OF VARYING RATIO
OF THE FLUX COMPOSITION K2Se:P2Se5:Se

IN THE FAST COOLING REGIME

110

INTRODUCTION:

Despite the impressive synthetic utility of the flux methodology, the speciation Of

molten Azse (Azalkali metal) salts at high temperatures remains unclear. For molten
Azse salts, it is believed that the polyselenide fluxes are formed by the simple fusion Of
A28e and Se to form A28ex or Sexz- chain fragments Of various lengths of which were
isolated and identified.1 The fusion Of AZSe/P28e5/Se results in the formation of

Ax[Psz] species that may co-exist with Sex} fragments. The [Psz]n' ligands become

solubilized in the melt and behave as ligands in the presence of metal ions. The

compounds that form from these fluxes are affected greatly by the stoichiometry of the
. 2,3
reaction.
Adjusting the Azse/P2Se5/Se ratio by removal or addition of excess Se in effect

increases or decreases the acidity Of the resulting flux mixture. In general, there are two

reasons why a selenophosphate flux reaction contains excess Se: (i) The melting point Of

the resulting flux from the reaction of A28e and Se decreases as the amount Of Se

. . . . . 4
increases; (Table 6.1 shows known melting pomts for potaSSIum selenide salts)

Table 6.1. Known melting points for some potassium selenide salts.

 

K2362 K2363 K2864 K2365

 

 

Melting points 460 380 205 190
(°C)

 

 

 

 

 

111

 

and (ii) Se acts as an electron acceptor for metals dissolved in the flux. Selenium is in

Group 6A of the periodic table and is generally an oxidizing agent. The materials

. . . . . . 2
resulting from these reactions generally possess metals in hi gh ox1dation states.

There are two effects to be considered when varying the amount and nature of

A28e in the resulting flux mixture: change 1n the concentration Of [Se ]; and alkah metal

coordInation. Fluxes high in A28e are very baSlC due to the presence Of Se 1n

solution/melt. The choice of the alkali metal, A, can also affect the reaction because Of its

size. As the size of the alkali metal decreases, the basicity from A28e also decreases

which consequently might decrease the probability Of the alkali metal being incorporated

in the final product. This can be explained more clearly in terms of the hard and soft acids

. . . 2- . .
and bases concept - hard bases stabilize hard acrds. Se 18 conSIdered to be a hard base.
. . . .+ + + + + .
The extent Of being a hard ac1d decreases in the order L1 <K <Na <Rb <CS w1th
increasing atomic size (Cs+ is considered to be a soft acid). Hence LiZSe is considered to
be more stable compared to C8286. Consequently the probability Of incorporating Li+ in
the final product is less compared to Cs+. Increasing the amount of P28e5 also in turn
increases the basicity Of the flux because in general, the formed [Psz]n- anions are more

. 2- 3
baSlC than Sex .

Because of the rich chemistry involved in metal selenophosphate flux reactions, it
is highly desirable to do physicochemical Studies designed to investigate the nature Of

selenophosphate fluxes. It is Of considerable interest to study thermodynamic and kinetic

112

processes and speciation equilibria in the flux since changes in these processes dictate the
reaction pathway. The proposed reaction equilibria by Kanatzidis as shown in Scheme 1

summarize possible equilibria happening in the melt based on a wide compilation Of

. 2
synthetic results:

113

 

 

3-
Se Se 89
394,‘ Se \P/
o, :
'p—p, + Se. —-— 2
‘04 Se ’89
Se 8989
Se
83 Se \
Se Se g:
'0 P u 2'
2 a ‘——— / 36 Se
‘5 Se
Se Se
Se
39 Se
Seél‘
6"P—P %
’9 39 Se
‘04 \ Se 2
Se/ Se 86 \P/Se\ i3,
/ a 59/ \
Se E’Se 36
S S 26
e e
-2e
\E/
2 =2 Se
Se 'ée
Se Se Se Se
\ / \ / \ /
F; Se 1"
Se Se / Se Ea S§ Se
Se Se

Scheme 1

There has been one NMR study of flux composition by Eckert et al. In a Li-P-Se

melt at T > 370 0C, the high temperature 31P NMR Spectrum provided evidence for the

e"‘iStence Of the stable phases Li4P28e6 and Li7PSe6. NMR Spectra Of the solid products

114

 

 

showed that Li4P28e6 contains the [P28e6]4- unit, while Li7PSe6 contains [PSe4]3_ and

Se2_ units.5 There have been additional 31P NMR studies of ternary phases Of crystalline

solid products formed from direct combination of M + P + Se ( M = Cu, Ag, Cd, Hg, Pb,

Sn, Ca, and In) and it was reported that the M-P-Se system has much less structural anion

variety compared to the ternary sulfide systems.6 In these direct combination melts, high
temperature NMR data were consistent with the presence Of only two anions, [P28e6]4-
and [PSe4]3_. Eckert et al. have also Studied pure P-8 and P-Se glasses with high

31 77 7
temperature P and Se NMR.

This chapter initiates the study of metal selenophosphate flux reactions of the

composition: AZSe/PZSe5/Se. In addition to elucidation Of reaction mechanisms, we are

hoping to develop mechanistic predictions for exploratory alkali metal selenophosphate

syntheses. In this chapter, the flux reactions of varying compositions of KZSe/P28e5/Se

are studied in order to determine which factor affects the formation of a specific type Of
. 31 . . .
anion. P room temperature and high-temperature NMR spectroscopy are applied in

these Studies. The ternary diagram shown in Scheme 2 summarizes the region of relative

composition ratio of the respective K23e2P28e5:Se studied in this chapter. The

compositions studied are in the excess Se region, wherein we have set the relative amount

of Se to be constant. The ratio is such that K2$ezP28e5:Se is to x:y:8 where x and y are

varied from l to 4, holding the relative ratio of Se to 8.

115

°/oK2$e

 

90 . 10
80 " 20
7o ' 30
60 .40
50 ’ ~ , 50
4o _' . ‘60
30 . , 4.. 70
' ~ .0. .’ ’O
20 y . .» 7 ~ 80
10 . . ". 1 90
A A A A A A A A A
90 so 70 60 50 4o 30 2o 10
%P2$e5 %Se

Scheme 2. Ternary diagram for the different flux compositions of KzseszSe5:Se studied
in this chapter.
It has been observed that in the selenium system, increased ratios Of A23e (more
baSIC) favor the formation of P + containing spec1es while reactions lower in A2Se
. . 4+ . 4- .
content (less baSic) yield P speCIes, most commonly the [P2Se6] anion. It has also

been Observed that high ratios of P2$e5 also favor the [P2Se6]4- anion. These

Observations were based on synthesis studies and the mechanisms involved are not well

116

2,8 . . . .
understood. The purpose Of the above synthetic investigations was to be able to make

some conclusions as to which flux conditions favor specific anions as building blocks.
Equilibria at high temperature may not be reflected in the final product distribution
because various factors can have different effect such as heating and cooling rates during
the synthesis, and concentration Of reactants. In this study, we hope to answer a few
questions about the equilibria happening at high temperature using in-situ high

temperature NMR spectroscopy technique.

117

MATERIALS AND METHODS

Preparation of Samples for In-Situ NMR.

P28e5 was prepared by reacting stoichiometric amounts of the elements in an

evacuated Pyrex tube at 300 0C for one day, followed by a cool-down to 50 0C over two

hours. Pun'ty was assessed by X-ray powder diffraction analysis.
All samples were prepared in a pyrex tube which serves as an NMR rotor insert

and sealed under vacuum. All sample reactants have a total weight of less than 100 mg.

K2Se, P2Se5 and Se were combined to give the following K2Se2P28e5:Se ratio:

(1:128), (1:228), (124:8), (2:128), (2:228), (223:8), (2:428), (32128), (3:228), (3:428), (4:1:8),

(42228), and (4:428). For one example, flux ratio (12228) was made by combining

0.007858 g KZSe, 0.04568 g P28e5, and 0.031584 g Se. All weighing are accurate within

0.001 grams.
Physical Measurements.
Difi‘erential Thermal Analyses. Differential thermal analyses were performed on a

Shimadzu DTA-50 thermal analyzer. Two heating and cooling cycles were done from
O
100 to 600 C.

Powder X-Ray Dzfiraction: Powder X-ray diffraction (PXRD) analyses were
performed using an INEL CPS 120 powder diffractometer with graphite
monochromatized Cu K a radiation.

High temperature solid-state nuclear magnetic resonance spectroscopy. The

static high temperature solid state NMR measurements of varying ratios of

118

K28e2P28e528e compounds were taken on a 9.4 T NMR Spectrometer (Varian Infinity

Plus) using a high temperature double resonance magic angle spinning (MAS) probe

manufactured by Doty Scientific (DSI-866). Samples were vacuum sealed in a pyrex

. . . . o
rotor insert. The experiments were conducted by raiSing the temperature by 50 C

increments at a rate Of 100C/minute from room temperature up to 400 to 6000C followed

by a cool-down to room temperature with the same increment. Most of the samples

studied were then subjected to a repeat cycle of 31F HTNMR in order to observe the 31P

NMR Spectra of the formed final product in the first cycle. Overall, the time it takes from
heating to cooling down the sample is about 6-12 hours compared to an oven synthesis
which takes several days.

Static NMR Spectra are taken after equilibrating the sample at the Specified

temperature for 20 minutes. Bloch decay spectra were taken with the excitation/detection

channel tuned to 31F at 161.39 MHz, a 5.75 ,us 900 pulse, and a relaxation delay between

1 and 1000 S. Each spectrum was processed with 100-1000 Hz line broadening and up to

a 10th order polynomial baseline correction. The spectra were referenced using 85%

H3PO4 at 0 ppm. At some temperatures, a series of spectra were acquired with delay

times between 0.1 and 1500 s. The significance Of varying the pulse delay is to ensure
that the NMR spectra are quantitative. Because Of the large variation in relaxation rates
among different compounds, spectra taken with short relaxation delays may be very non-
quantitative, and this can cause difficulties in assignment of NMR peaks to particular

synthetic products.

119

Magic-angle Spinning NMR spectra are also taken for some samples. There is a
difficulty however in achieving higher than 1000 Hz spinning frequency in using the high
temperature probe. This is due to weight imbalance about the spinning axis. The NMR
spectra shown in this chapter are static spectra unless otherwise indicated.

Room temperature NMR. The room temperature solid-state NMR measurements
Of products after cool-down were taken on a 9.4 T NMR Spectrometer (Varian Infinity
Plus) using a double resonance magic angle spinning (MAS) probe. Samples were spun at

7-12 kHz using zirconia rotors of 4 mm outer diameter. Bloch decay spectra were taken

with the excitation/detection channel tuned to 31F at 161.39 MHz, 3 4.5 as 900 pulse, and

a relaxation delay between 5 and 15,000 3. Each spectrum was processed with 100-1000

Hz line broadening and up to a 10th order polynomial baseline correction. The spectra

were referenced using 85% H3PO4 at 0 ppm. MAS NMR measurements on a 7 T NMR

spectrometer were taken for samples that gave mixtures Of crystalline and amorphous
products to verify the magnetic equivalence of the "doublet" crystalline final product.
Changing the magnetic field strength will change the chemical shift frequency separation
but will not change the J-coupling frequency separation. If the peak frequency difference
between the two peaks is proportional to the strength of the magnetic field, then they are
magnetically inequivalent. If the two peaks are scalar-coupled, the peak frequency
difference between the two peaks should be the same in both magnetic fields and

indicates a P-P type Of bonding with scalar coupling between the two P atoms.

120

RESULTS AND DISCUSSION

The flux ratios studied in this chapter will be presented into two groups: group (1)
are the flux ratios which produced crystalline final products [(2:1:8), (3:128), (3:228),
(4:128), and (4:2:8)], and group (2) are the flux ratios which produced a combination of
amorphous and crystalline final products [(1:128), (1:228), (1:4:8), (2:228), (223:8), (224:8),
(3:2:8), (3:428), and (4:4:8)]. The flux reaction with ratio (322:8) also gave both
amorphous and crystalline products, but we placed it in group (1) because its high

temperature NMR properties coincide with most Of the trends for this group.
KzSe2PZSe5:Se flux ratios group (1): (2:1:8), (3:128), (3:228), (4:128), and (4:2:8).

This group Of flux ratios, which forms mostly crystalline final products (no non-

, crystalline products which contain P) after reactions at high temperature, is characterized

by having a greater mole ratio Of K28e compared to P28e5. The final products formed in

each ratio are distinct and will be discussed separately.

Figure 6.1 shows the MAS 31F high temperature NMR spectra Of K2Se2P28e528e

flux reaction with ratio (2:1:8).

121

258 °C 500 °C 500 °Cr

 

 

 

 

283 °C 390 °Cd
kw 400 °Crd
221 °Cr

 

 

W A ##1##! l I l “I

 

 

 

 

 

 

 

 

 

 

400 0C 325 °Cr final product
l V: 7kHz
MW WW. .iimzooos
[ rI I I l i I I I l r F
o o 200 0 ~200

450 C 400 Cr ppm

[lrllI'llll'f IIII IIII II
200 0 -200 260 (I) -2(l)0
ppm ppm

Figure 6.1. 31P MAS high temperature spectra of the KZSezPZSe528e flux reaction
(2: 1:8). (note: d=rampdown in temperature; r = recycled temperature).

122

During initial ramp-up in temperature, sharp lines become apparent above 200 0C,

which indicates motional processes at these high temperatures. Lorentzian lineshapes are

Observed at 500 0C which is characteristic Of a liquid state spectrum.9 During the recycle,
similar patterns are Observed wherein sharp lines start to appear at 200 OC. Flux (22128)
has 31F chemical shifts that are upfield relative to 0 ppm (Figure 6.1). This indicates the
presence Of PSe4-type or P+5 species at high temperatures.10 The final product of flux
(2:1:8) yields two 31F peaks at —39.6 ppm and —113 ppm. Figure 6.2 (i) shows an
expanded version Of the 31P NMR spectra of the flux (2:1:8) final product. The peak at
—113 ppm is due to K3PSe4,11 and is about 40 % of the final product distribution. The
peak at —39.6 ppm is due to K4P2Se9 and is 60 % Of the final product distribution. The
K4P28e9 assignment is based on comparison of the 31F chemical shift and CSA of
CS4P2569 (see Chapter 3).12 K4P23e9 and CS4P2569 are isostructural. The assigment of

the peak at —39.6 ppm was confirmed to be K4P28e9 after synthesizing this compound
(done by In Chung in Prof. Kanatzidis’ group) and verifying that a single orange crystal
was indeed K4P28e9. The overall sample also contained red K3PSe4 crystals. The 31P
solid-state NMR (SSNMR) spectrum (Figure 6.2 (ii)) also showed a large peak at —40
ppm (K4P2Se9) and a smaller peak at —113 ppm (K3PSe4). The K4P28egzK3PSe4
integrated intensity ratio was ~ 3. Figure 6.2 (iii) shows the experimental PXRD from the

re-synthesized sample and the calculated PXRD for K4P2Se9. There is reasonable

123

correlation between Bragg angles in the experimental PXRD and calculated PXRD

patterns. There is also a large peak in the experimental pattern which is absent from the

calculated pattern and which may be due to K3PSe4 (known from solid state NMR to be a

25% impurity in the sample). In addition, the relative intensities in the experimental
powder pattern are different from those in the calculated powder pattern because the
PXRD sample preparation (e.g. grinding the sample and application of the sample to the
glass plate) may preferentially disrupt crystalline order along particular Miller planes.

This inhomogeneity can be particularly pronounced in a layered Structure such as

K4P28e9. Initial unsuccessful attempts to synthesize K4P2$e9 yield interesting results

and are discussed later in the chapter (see the section on the effect of heating and cooling
rates). The formation of the two crystalline peaks is maybe due to an equilibrium process

at high temperature. The equilibrium can be proposed as:
P2Se94- + SJ" :2 zse + 2PSe43' (1)

The appearance of the two crystalline products for flux (22128) is confirmed by

differential thermal analysis (DTA) data. The DTA data shown in Figure 6.3 for flux

(2:1:8) shows that melting starts at 142-160 0C with subsequent melting Of excess
selenium at 220 0C followed by another melting peaks at 279.3 0C and 314 0C. Sharp

NMR peaks are Observed above 314 0C due to adequate motional processes at these

temperatures which effectively average the broadening effects of dipolar coupling and

chemical shift anisotropy. During cool-down, the DTA shows crystallization peaks at

275, 227.9 and 141.3 0C. Consequently, the 31P NMR peaks start to broaden below

124

O . . . . . .
300 C due to reduction in motional processes from crystalline solid formation. The

. . o . . . .
crystallization at 227.9 C is due to excess Selenium. The two other crystallization

temperatures are due to the final products observed for flux (2:128).

125

(i)

SSb
SSb

 

ssb ssb

MWM‘WW‘WWNJ lat/W16) \le WMWW‘W

*

 

 

 

(ii)

SSb
SSb

 

SSb

WETLJLJLMJULJW

l | l l l
150 50 ~50 -150 -250
PPm

 

 

 

 

126

(iii)

 

 

 

(a)

 

 

 

 

 

 

 

ll (b)
, flu in .Li..i-.._.-iii....- -.....9_...
20

10 30 40 so 60 70
26(deg)

Figure 6.2. (i) 31P MAS NMR spectrum for the final product of the reaction of Flux
(2:1:8) shows two different 31P peaks indicated by (*) at —113 ppm (K3PSe4) and at
—39.6 ppm (K4P28e9). The ratio of integrated intensities of the K4P28e9 peaks to the
K3PSe4 peak is about 1.5. The spinning frequency is 8 kHz. (ii) 31F MAS NMR spectrum
for the furnace re-synthesized K4P28e9 Shows two different 31P peaks indicated by (*) at
-113 (K3PSe4) ppm and —40 ppm (K4P28e9). The ratio of integrated intensities of the

K4P28e9 peaks to the K3PSe4 peak is about 3. The Spinning frequency is 10 kHz. (iii)
Comparison of (a) experimental and (b) calculated X-ray powder diffraction patterns of

K4P28e9. The * mark represents the existence Of another phase (likely K3PSe4).

127

 

DTA flux (2:1:8) recycle

 

uV

-20 _

-25 _

 

-30 z

 

 

 

‘35 I I I I l l
0 1 00 200 300 400 500 600 700

Temp C

 

 

 

Figure 6.3. Differential Thermal Analysis Data for recycled Flux (221:8).

The 31F high temperature NMR spectra of the flux reaction (3:128) in Figure 6.4
shows magic angle spinning (MAS) spectra with Spinning speeds of about 500 Hz. At
221 0C, MAS spectra are apparent with isotropic peaks of —99 and -—1 13 ppm. In order to

positively assign the MAS peaks with greater precision, the spinning speed has to be

increased, but it is currently difficult to do that because of symmetry problems of the

glass insert containing the sample. Peaks at —99 and —l 13 ppm are present up to 250 0C.
At 300 0C, two peaks are present at —53.7 and -114 ppm with the presence of spinning

sidebands. At 400 0C, a single sharp peak with full width at half maximum (FWHM) Of

351 Hz is Observed due to fast exchange between species present and increased motional

processes in the melt. The evidence of fast exchange at this temperature is based on the

128

chemical shifts of the final products formed. Cooling down to 200 — 100 0C shows MAS

peaks in the upfield regime from 99-120 ppm with peaks not clearly resolved due to

spinning. The 31F chemical shifts at high temperatures are in the upfield regime relative

to 0 ppm, similar to the chemical shifts at high temperatures Observed for flux (2:1:8)

above. The final product formed during the reaction is also crystalline in nature. The final

product shows three crystalline 31F peaks with chemical shifts at 5.0, -99.3 and —112.6
ppm. The peak at —112.6 ppm is due to K3PSe4 and the peak at —99.3 ppm is a product

with a [PSe4]3- anion based on its chemical shift and low CSA. The peak at 5.0 ppm is

due to some form of phosphorous anion, which is still unknown but could possibly be

[PSe6]— due to comparison with RbPSe6 with a 31F chemical Shift of 4.3 ppm.13

129

 

 

 

 

 

 

 

III 221°C 400 °Cd
IIIIIII II
llIII'IIIIIII'III [IIIi Mr“ l'll'l‘Illm l WWII-l _: ‘_ VJHL?’ 72.. __i w
250 °C
I 300 °Cd
«III. III I; .
«*IIMWIW 'I’II‘IIIIIIWI Elle - r
300 °C
II 200 °Cd
III III pd: 1000s
I W IIIII III III
400 °C 100 °Cd
III pd=1000s
[IH'I

 

II’IIII IN WWII VIII! III

II

 

 

 

 

I‘MI’MII MIWHWWIII I.‘ I
I II final product
I 500 0C II V: 71(HZ
II II I pd=20008
.......J a...“ .I WIMW WW
I l T l I l 7 I I I r l l l l l I I
0 - 100 -200 0 - 100 -200
ppm ppm

Figure 6.4. 31P High temperature NMR Spectra of the reaction Of K2Se2P28e5:Se with
ratio 3:128. (note: d=rampdown in temperature; r = recycled temperature).

130

The high temperature NMR spectra for the reaction of Flux (3:2:8) are shown in

Figure 6.5. The 311’ peaks are already sharp during initial ramp-up from 200 0C, which

indicates increased motional processes in the melt. The 31F chemical shifts observed at
high temperatures are in the upfield regime relative to 0 ppm which indicates the
presence of P+5 species in the melt. This seems to follow the same trend as the chemical
shifts at high temperatures of the flux reactions (2:1:8) and (3:1:8) which have mole ratio
of K28e greater than the mole ratio of PZSe5. The final product distribution is semi-

crystalline and is composed of the two magnetically inequivalent peaks at —18.1 and —
19.9 ppm, which are the same peaks observed with those of the group (2) flux ratios as

will be discussed later, accompanied by amorphous features. The amorphous region

resonates around —50 ppm, which may indicate some amorphous forms of [P28e7]4.,
4- 4- . . . . . . 10

[P28e9] , and [P286101 based on 1ts chemical shift dlstnbution (see Chapter 5). A

slight broadening of the lines are observed at higher temperatures as shown in Table 6.2:

Table 6.2. Shows the trend of the change in 31P NMR linewidth with temperature.

 

 

 

 

 

Temperaturea Full-width at half
0 C maximum(FWHM)
500 1711
400 1347
300 l 189
200 1129

 

 

 

 

a .
the temperatures shown are the temperatures during the recycle rampdown

This broadening of the lines at higher temperatures was also observed in the synthesis of

Rb4Ti2P6Se25 (see Chapter 5) wherein the final product was a mixture of different

131

selenophosphate anions. At high temperatures, it is possible that equilibrium exists

between species for example: PSe43--‘= PZSe74- :3 P28e94. ;-—-‘ P28e104-.
The sharpness of the peaks is related to the molecular tumbling motions of the

species present at that temperature. The higher the motions, the sharper the peak is. The

sharper peaks observed at relatively lower temperatures compared to the peak at 500 OC

might be due to higher amount of smaller sized selenophosphate anions present at low
temperature. The broadening of lines at higher temperatures is also observed in other flux

ratios discussed below.

132

I

II 200 °C
.I
I I
' II
WWI-(I’M I W‘I'II‘IIIIWW‘M;
I 300 °c
I I

II
M L...“
II
II 500 °c

H

I.
JIM”

II 400 °Cd

II

I l I

I I I I l 1
0 -200
ppm

 

I
II 300°Cd
II

II
.aJLW
"'6"'12(I{
PP"1

 

I
I
_...J W

W

300 oCrd

 

I 200 °Crd

I
II

final product
v: 12 kHz

 

 

I IIAQOOOS

 

Illr
O

IrIj
-200
PP"1

Figure 6.5. 31P High temperature NMR spectra of the reaction of K28e2P28e5:Se with
ratio 3:2:8. (note: d=rampdown in temperature; r = recycled temperature).

133

Flux (4:1:8) has the most amount of K28e studied in this chapter. The 31F high

temperature NMR spectra of the reaction of flux (4:1:8) are shown in Figure 6.6. During

the ramp-up in temperature, broad lines up to 350 OC characterize the spectra. At 400-500

0 . . .
C, narrower lines around —65 ppm are observed due to motional processes at these high

temperatures although there is still a little broad peak downfield of the sharper peak
which is still in slow exchange with the sharp peak. Most of the flux ratios discussed in
this chapter exhibit fast chemical exchange between species at high temperatures such

that the NMR spectrum is composed of a single peak which is a weighted average of the

. . . . 31
resonances of the spec1es taking part in the exchange as discussed below. The P

chemical shifts at high temperatures are in the upfield region relative to 0 ppm which

suggests formation of PSe4-like anions or P+5 species. The chemical shifts at high

temperatures are also consistent with the trend discussed above for fluxes with greater

KZSezPZSes mole ratios, which also suggests crystalline products of the PSe4-type (e.g.
compounds having [PSe4]3- or [stex]4- anions where x = 7,8,9,10...) are formed. The
final product formed after the reaction is >95% pure K3PSe4 with a 31F chemical shift of
—1 13.0 ppm.11 Impurity peaks (<5%) are observed at -18, —19.7 ppm which are the same
peaks observed for flux ratios having K28e mole ratio less than P28e5 (will be discussed

below). Based on the final product formed in this reaction, it seems that the [PSe4]3-

anion is the favored species formed at high temperatures. This is reasonable since the

134

. . . 3- .
amount of K2Se 1s four times greater than P28e5, and the reaction to form [PSe4] lS

stoichiometrically feasible:

P2865 + 3K236 —-> 2K3PSB4
2- 3—
Or P2365 + 356 —> ZPSC4 . (2)
Flux (4:2:8) is another case wherein the mole ratio of K23e is greater than the
mole ratio of P28e5. The 31? high temperature NMR spectra for the reaction of flux

. . . 0
(4:2:8) are shown in Figure 6.7. Sharp narrow peaks are observed starting at 300 C
during the initial ramp-up in temperature due to increased motional processes at these

31 . . . . . .
temperatures. The P chemical sh1fts at high temperatures are in the upfleld region

. . . +5 . .

relative to 0 ppm Wthh suggests formation of PSe4-type or P unrts and are consrstent
with previous data for fluxes having the mole ratio of K28e greater than the mole ratio of

PZSeS. The peaks at higher temperature (500 0C) are characterized by Lorentzian

lineshapes and are slightly broader (~20% broader compared to the spectrum at 400 0C)

than peaks at lower temperatures due probably to different exchanging species at those

temperatures. The product formed after the reaction has a 31F chemical shift of -40.4

ppm and is >95% of the phosphorous-containing products. Small impurities are observed

as well at chemical shifts —18.0 and —19.7 ppm, which are the same peaks observed for

flux ratios having greater amounts of PZSeS (will be discussed below). The peak at —40

135

 

ppm is due to formation of K4P28e9 as observed in the flux ratio (2:1:8) discussed above.
The formation of a [P28e9]4- species is also reasonable stoichiometrically:
2- 4-
PZSeS + 28c + ZSe —> P28e9 (3)

since the amount of K2Se is two times the amount of P2865 in this flux reaction.

136

II 230 °C
I'IIIIII' IIIIIIIIII
III 270 °C

ILIIFIIIIIIIII‘IIIIIII/ I IIIIIIIIIIIII

IIII3 00 °C
“WNW/I? WWII ll!

350°C

III‘II
IIIWWM W ‘I‘IIII'I‘III

400 °C

MM

IlTlllllllIlllllllllI

500 250 0 -250 _500
ppm

 

 

 

500 °C

 

MIMI...

400 °Cd

 

I'JIIIIWIIIWHM W‘VI‘I‘WIWIII'I

MIMI 300 °Cd
IIWI IIIIIIIII II

final product
V= 12kHz
pd=1000s

A4

 

 

I'WIWIWIII...,

500 250 0 —250 -500

PM“

Figure 6.6. 31P High temperature NMR spectra of the reaction of KZSezPZSe5:Se with
ratio 4: 1:8. (note: d=rampdown in temperature; r = recycled temperature).

137

o 500 °Cr
200 °C 300 Cd
300 °C II 200 °Cd 400 °Crd
400 °C 200 °Cr 300 °Crd

0 200 oCrd
500 °C 300 Cr

0 final product
400 °Cd 400 Cr v=12kHz

I pd-[3000S

rIIItIIrI IITIIII'

I
0 -200 0 -200 0 -200
PPm PPm ppm

 

 

 

 

 

 

 

 

Figure 6.7. 31F High temperature NMR spectra of the reaction of KZSe:PZSe5:Se with
ratio 4:2:8. (note: d=rampdown in temperature; r = recycled temperature).

138

Equilibria between [PSe413- and [PZSe914- in the flux KgSe/P25e5/Se. . An

equilibrium based on possible anions present in the melt can be deduced from the flux
ratios which gives crystalline final products (flux: (2:1:8), (3:1:8), (4:1:8), (4:2:8)) as
discussed above. In order to identify a possible equilibria in the melt at high temperature,

we need to have an idea of what are the compositions of the melt in equilibrium.

The final product of the flux reaction of flux (2:1:8) gave 2 crystalline 31P NMR
peaks at —40.1 (which is K4P28e9 and 59.7% of the product distribution) and —1 13.0 ppm

(which is K3PSe4 and 40.3% of the product distribution) as shown in Figure 6.2. The

formation of these two crystalline species is maybe due to an equilibrium process

happening at high temperature. At high temperatures (say 500 0C), the rates of chemical

exchange become fast on the NMR timescale and only a single narrow feature is

observed in the spectrum: the weighted average of the resonance of [PSe4]3- and
[P28e9]4- taking part in the exchange (see Figure 6.1 at 500 0C). In this fast-exchange
regime, the temperature dependent averaged chemical shift, v ave is observed:

Vavezpa Va+(l ‘Pa) Vb (4)
where pa is the fractional population of species a and va, vb is the Larmor frequency of

. . 9
spec1es a and b respectively.

The final product of the flux reactions of flux (4:1:8) and flux (4:2:8) yields

almost pure K3PSe4 and K4P28e9 respectively. Based on the final products, the high

139

temperature 31P NMR spectra for these reactions are due to the resonances of pure
[PSe4]3' and [P28e9]4- as shown in Figures 6.6 and 6.7. We can then make use of the
lever rule in order to identify and quantify the spectra at 500 0C for flux (2:1:8) in
comparison with the 500 0C spectra of flux (4:1:8) and (4:2:8). Figure 6.8 shows the 31P

NMR spectrum at 500 0C for the flux ratios (2:1:8), (4:1:8), (4:2:8):

K,Se:P,Se5:Se at T = 500°C K,Se:P,Se5:Se at T = 20°C

K3PSe4

 

 

 

 

  

 

  

 

 

K3I’Se4
(221 28) l K4P2S‘39
a —+' ‘
MW : Una-Mu
I I I I
O 200 o 200

ppm

Figure 6.8. Fast Chemical Shift Exchange at High Temperature. The dashed lines indicate
the chemical shift positions of fluxes (4:1:8) and (4:2:8) relative to flux (2:1:8). The
arrows containing the deltas (6) are measures of distance in terms of ppm. 53 is the

distance between the chemical shift peaks of flux (4:1:8) and (4:2:8), 51 and 52 are the

140

distance between the chemical shift peak of flux (2:1:8) with flux (4:1:8) and (4:2:8)

respectively. The corresponding 31P room temperature MAS NMR spectrum is presented
on the right side of the figure.

From the lever rule, the relative amounts of [PSe4]3' and [P28e9]4- can be
calculated at each temperature. 53 is the distance between the chemical shift peaks of flux

(4:1:8) and (4:2:8), 61 and 52 are the distance between the chemical shift peak of flux

(2:1:8) with flux (4:1:8) and (4:2:8) respectively. Therefore the mole ratios can be

calculated by:
Mole ratio [PSe4]3- = 52/53 (5)
Mole ratio [PZSexf' = 251/53 (6)

Table 6.3 summarizes the results for the relative amounts of each species in the flux

(2:1:8) reaction.

Table 6.3. Summarizes the relative amounts of [PSe4]3- and [P28e9]4. anions present in
flux (2:1:8) at high tern erature using the lever rule as shown in Figure 6.7.

 

 

 

 

 

 

Temperature in M01 ratio [PSe4]3- Mol ratio [P28e9]4- FWHM (Hz)
degrees C
500 0.25 0.75 434
400 0.26 0.74 219
300 0.5033 0.497 139
Room Temperature 0.43 0.597 -
MAS spectra

 

 

 

 

Table 6.3 shows that the concentrations of the [PSe4]3- and [P28e9]4- anions in

the final product distribution are consrstent With the concentrations of the [PSe4] and

141

 

[P2Se9]4- anions at high temperatures in the fast exchange limit with best agreement at
300 0C. This data for flux (2:128) also suggest that at higher temperatures (400—500 0C),
the predominant species is [P28e9]4- anion but at lower temperatures, the amount of

3- . . . . .
[PSe4] anron increases resulting in narrower peaks at lower temperatures. ThlS

narrowing of peaks is due to increased molecular tumbling motion of species in the

. . . . . 3- . .
sample, Wthh in this case rs due to increased amount of [PSe4] anion which has a

smaller CSA and smaller molecular size compared to [P28e9]4- and hence greater

motion. A relatively sharp linewidth is also observed in the flux reaction (32128) (FWHM
~ 350 Hz as discussed above) in which [PSe4]3' anion is present at high temperature and
in the final product distribution.

The difference in linewidths of the 31P high temperature spectra at 500 0C in

Figure 6.8 is due to the difference in motional process happening in the melt for each
ratio. Flux (2:1:8) spectrum has the sharpest linewidth because it has the lowest melting

point which means lowest viscosity due to a relatively higher ratio of selenium and low

ratio of PzSe5.
KzSe/PgSe5/Se flux ratios group (2): (12128), (12228), (12428), (2:228), (2:328),

(2:428), (3:228), (3:428), and (4:428). This group is characterized by having final products

composed of both glassy features around —10 ppm and crystalline 31P NMR peaks at —

18.1 and —19.7 ppm. The two crystalline peaks upfield are two independent phosphorous

sites, as determined from further NMR measurement on a 7 T NMR magnet. The

142

chemical shift anisotropy (CSA) for the two crystalline peaks is in the range ~ 400 ppm

which is greater than the CSAs observed for [PSe4]3- anions (~60 ppm) and [P2369]4-

anions (~ 200 ppm). These anions or compounds may have lower or comparable

molecular symmetry compared to PZSex-type bridging species. Future experiments will

determine the anions for these peaks. Initial attempts to synthesize the two anions using
slow cooling in the furnace of representative ratios in group (2) were unsuccessful but

yielded interesting results as discussed later (see section on effect of cooling rate below).

A more promising way maybe is to synthesize K4P28e9 (from flux (4:2:8) (see Figures

6.7 and 6.8) in higher amount (> 1 gram) wherein the two anions occur as impurities.

Then the crystals can be physically isolated and identified by single crystal x-ray

. . 31 . .
diffraction and subsequent P solid-state NMR measurement. The two unknown anions
are also present as impurities in the high temperature NMR syntheses of K2Cu2P4Se10

(see chapter 5) and slow-cooling furnace synthesis of K4P28e6 (see discussion later

below — Figure 6.12). Another way is to do fast cooling and quenching experiments for

group (2) in large amount (>1 g). The glassy feature around —10 ppm is maybe due to

amorphous forms of PSe4-type such as [PSe6]' based on its chemical shift (see Chapter

5).13 Future NMR experiments will determine if the amorphous peak has P-P type of

bonding using dipolar coupling NMR experiments. Figure 6.9 shows representative high

temperature 31P NMR spectra of the reaction of flux (1:228). The spectra for all the other

flux ratios are shown in the appendix for Chapter 6.

143

N
O
O
O

I
, . . \ ' .
MnflfiWtfi/J II “\MJWMLh’I-hifi‘y‘

. II

I I 400 °C

mud/I I‘WW

II 300 °Cd
II
II 200 °Cd

WW

I I I I I

 

I I T I I I
7
PDT“

I
-00 O -200

I 200 °Cr

I
II 300 °Cr

WW WWW

I 500 °Cr

IIII~,:.,=IM IIIIymIII-vIwII.

I
I 300 °Crd

M W
200 0Crd

 

 

 

 

“JIM.

final product
v: 11 kHz
= 30003

I I I I I I I I I l

200 0 -200
PPm

  
  

 

Figure 6.9. 31P High temperature NMR spectra of the reaction of KzSezPZSeszSe with
ratio 122:8. (note: d=rampdown in temperature; r = recycled temperature).

The high temperature 3'lP NMR spectra of this group of flux ratios have common

features. At temperatures just below the melting point of any of the reactants present,

broad featureless lines are seen in the static spectrum. Beginning at temperatures about

0 . . . . . .
200-300 C, motional processes in the melt increase to rates that are sufficrent to give

narrowing of the spectral lines because the broadening effects of dipolar coupling and
chemical shift anisotropy are averaged by these motional processes. As the temperature
increases, the rates of chemical exchange become fast on the NMR timescale, and only a
single sharp peak is observed due to fast chemical exchange between species present in

the melt. This peak corresponds to the temperature dependent weighted average of the

resonances of species taking part in the exchange (see Figure 6.9 above 300 0C in the

initial ramp-up).9 The comparison of chemical shift data at high temperatures for all

samples studied in this chapter offer valuable information about the different chemical

environments of the phosphorous atom. Figure 6.10 shows the plot of 31P NMR chemical

shift for all the flux ratios studied versus temperature for the temperature range wherein

fast exchange between species is present. Group (2) flux ratios are the points with

.. 31 . .
posmve P chemical shifts.

The 31F chemical shifts for all group (2) flux ratios at high temperatures are in the

positive ppm (downfield region) relative to phosphoric acid at 0 ppm which suggests

. +4 .
formation of P -type compounds. As the amount of P28e5 increases at constant amount

of Kzse, the 31P chemical shifts at high temperature become more downfield. During

145

recycle in temperature, the 31P lineshapes are sharper at lower temperatures (200 0C)

which suggests the presence of molecular P-Se units which may exhibit rapid

. . . . 31 .
reorientation that narrows the resulting lineshape to a greater degree. The P lineshapes

at higher temperatures (500 0C) are slightly broader compared to the lineshapes at lower
temperature, which is also observed in the high temperature NMR synthesis of
Rb4Ti2P6Se25 as discussed in Chapter 5. There are several factors for this effect. This is
may be due to different extent of exchanging species at each temperature. As an example:
4- - 3—
132366 “(=3 P566 ‘——‘ P864 (7)

The sharpness of the peaks is related to the relative molecular tumbling motions of the

species present in the melt. The higher the motions, the sharper the peak is. At lower

temperatures the 31F linewidths are slightly sharper, which might be due to higher

content of tetrahedral [PSe4]3- anion which has relatively higher symmetry and relatively

smaller molecular size compared to the other two anions in equation (7) and hence faster

molecular tumbling motion (see discussion corresponding to Table 6.3 above and the

discussion of Rb4Ti2P6Se25 in Chapter 5). Another reason might be different viscosity at

different temperatures, but there is no direct viscosity measurements done on these kinds

of samples.

146

 

 

 

 

 

 

 

 

 

 

 

o f118
120
‘ A _¢_ I f128
A + + “148
A 80 -. _ :I— e; _--__ ____ ____ __ ___
g ! xf218
a. ' 3
v ____: f 3 fig _f__ #h ‘_ __g xf228
E 40 r . x
.5 i i 0 I238
T" Oq _ i- 7 _ __ * .k_ __ __ +948
2 " -
E .- "' ‘ z I318
g .40 ~ — 1" i 7 A— i i "x' —— —--»x--—- — -f328
2 f ; . . f348
'5 -80 ~ ‘ -7. v — _ —* ____ —-— ”418
X A f428
'120 T . . xf448
150 250 350 450 550
Temperature In degrees C

 

 

 

Figure 6.10. Plot of temperature versus 31P NMR chemical shift for all the samples
studied in this chapter. The groups of data presented are the chemical shifts during the
ramp-down recycle in temperature. Group (1) flux ratios are the ones below 0 ppm, and
group (2) flux ratios are the ones above 0 ppm.

The numerical values of the slopes in Figure 6.10 for the flux ratios in group (2)

are presented in Table 6.4 including a summary of the high temperature NMR results at
500 0C. The slopes or the temperature coefficient (ppm/0C), due to the dependence of
31F chemical shift versus temperature was plotted for each flux ratios in the fast

exchange regime: from the lowest initial ramp-up in temperature until the lowest cool-
down in temperature during the recycle. The complete plots for each ratio are shown in

the appendix. The slopes reported in Table 6.4 are the slopes during the final ramp-down

147

in temperature (as shown in Figure 6.10), which are about the same slopes during ramp-

up in the recycle period with just slight differences due to hysteresis.

Table 6.4. Summary of results of the high temperature 31P NMR measurements at 500 0C

for the flux ratios of group (2).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Flux Spectra at 500 0C3 Spectra 0f figal Slopeb ppmC at FWHM‘r
t
ra 10 pI'OdUCI (25 C) ppm/0C 500 0C 5000C(HZ)
12128 0.1073 34.8 2221
, . .. . WW WWI
250 0 -250 I"""""
ppm 250 0 -250
ppm
12228 0.0871 75.2 2006
TTTT"I "I ""I""7
200 0 -200 200 0 -200
PPm Ppm
12428 I 0.0625 113.3 2089
200 I0Y 200 rI""I"'II
ppm 200 0 -200
, PPm
22228 I 0.1006 40.1 2315
I7 ' ' ' I ' ' I ' I VI ' ' I I ‘ I I ' I
250 0 -250 250 0 -250
1’1”“ ppm
223:8 WALL 0.0978 62.9 2527
MW
T—‘I I I I I I I T tfi r I
200 0 200 200 0 -200
1’1”” ppm

 

 

 

 

 

 

148

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2:428 I AIL 0.073 105.1 2342
I I I r T I I I I l I I
200 0 _200 l I I I I I I 1
PP!“
32228 0.0693 —11.21 1711
MIMI W
"rFTfT‘I' "'1
O -200 0p
PPm
32428 I 0.0957 69.5 2453
fI T T T I I I I T I I T 2] l r l rj—r VT—rj
200 0 -200 -200
Ppm
42428 I 250pr 0.1005 34.8 2917
I T F I I I I I I Ij I I F I I I I ij
200 O -200
ppm Ppm

 

 

 

 

 

 

a Spectra at 500 0C during recycle in temperature

b Slope of the plot of 31P chemical shift vs temperature. The slopes reported here are
during the ramp-down in temperature in the recycle.

C 31F chemical shifts at 500 0C during recycle in temperature corresponding to a.

d Full-width at half maximum (FWHM) for spectra at 500 0C during the recycle in
temperature. '

149

 

In the fast exchange regime for group (2), the numerical value of the slope in

general decreases with increasing amount of P2Se5 at constant amount of K2Se. This
31 . . . . .
means that the P chemical shift changes less With increasmg amount of P2Se5.
. 31 . . . . . .
Simultaneously, the P chermcal shift goes downfield With increasmg amount of P2Se5

at constant amount of K2Se and the data sets become well separated (see Figure 6.10).

Figure 6.11 shows the trend on the effect of increasing or decreasing the atomic

percentage of a reactant (K, P, Se) on the 31F chemical shifts at 500 OC.

 

 

 

 

 

 

 

CS at 500 C vs atom %
§ ’ o I' A‘
'2 80 -
.2 ‘9’, ‘A e
E E 30 - .l. A A A egiat 5000 vs atom
§ 3 I CS at 5000 vs atom
g _20 I I o A %p
5 .' . o A ‘ A CS at 500C vs atom
a. I O A %Se
3 '70 “LI 1 . r r ‘ fl

0 20 4O 60 80
atom %

 

 

 

Figure 6.11. Trend of the effect of increasing or decreasing the atomic percentage of a
reactant (K, P, Se) on the 31? chemical shifts at 500 0C.

. . . . 31 . .
Clearly, the spec1ation of the melts and the fluctuations in the P chemical enVironment
depends on the amount of K28e present in the flux at excess amount of Se. The

. . + 2- . . .
incorporation of K and Se in the phosphorous env1ronment as K2Se increases seems to

150

favor and stabilize the formation of crystalline P+5 type of species (Group 1) in the melt
and the final products while the stabilization of P+4-type species (Group 2) in the melt is

. . . . +4 .
favored by decreasmg amount of KZSe. However, it is noted that crystalline P spec1es

have not been observed in any of the final products in this study. There may be other

. . . +4 . . . .
factors that stabilize crystalline P products. Future cooling rate studies usmg high

temperature NMR will shed more light to this problem.

Eflect of Heating and Cooling Rates. The high temperature NMR experiments
conducted in this chapter involved relatively fast heating and cooling cycles compared to
traditional furnace syntheses of metal selenophosphates. Furnace syntheses usually take a

couple of days to weeks as shown in scheme 3 below.

1 to several days or week at 500-800

2days to several weeks of cooling to 200
6-12 hours ramp-up
6-12 hours cool-down to room temperature

Scheme 3.

 

To illustrate the importance of the difference between the heating and cooling
rates conducted using high temperature NMR and furnace syntheses, we repeated the

reaction of flux (4:2:8) with the purpose of growing bigger crystals of the unknown

151

product with a chemical shift of — 40 ppm (which was later found to be K4P28e9- see
discussion above) . The reaction was done at peak temperature of 500 0C for a day and
slow cooling at 10 0C per hour to 150 0C. At high temperature as shown in Figure 6.7,
the 31F high temperature NMR spectrum of this reaction at 500 0C is in the upfield

region relative to 0 ppm which suggests formation of a P+5 species. The final product

after the reaction is shown in Figure 6.12 (done by In Chung in Prof. Kanatzidis’ lab).

The spectrum is composed of two 31F peaks at 43.0 and 45.6 ppm which is due to the
known compound K4P28e6 which has two crystallographically distinct P+4 atoms and P-

P bonding.14 A little impurity can be seen at —18.0 and —19.7 ppm which is produced in

most of the flux reactions discussed above. The importance of this result is that none of

the flux reactions conducted using high temperature NMR gave a final product which has

a P+4 or a [P28e6]4- type anion. This result suggests that most of the products formed

during this NMR study are due to kinetically formed products and that K4ste6 is maybe

a thermodynamically stable product. It is therefore interesting to conduct high

temperature NMR experiments similar to a furnace syntheses with slow cooling rates.

152

K4PQSe6

Unknown “doublet” present in amorphous final products

/
HM“) LEJW

 

 

 

ppm

Figure 6.12. Final product of flux (4:2:8) furnace syntheses with slow cooling.

Since slow cooling of flux (4:2:8) did not produce K4P28e9, fast cooling

synthesis was also done in the furnace. A completely different final product distribution

was obtained as shown in Figure 6.13. The final product produced about 90% K3PSe4

and about ~10 % of unknown impurity at —99.3 ppm which is also due to PSe4-type

species because of its chemical shift and low CSA. This result is consistent with the high

temperature NMR data of flux (4:2:8) because P+5-type species were predicted to form at

high temperature.

153

(a)

 

 

 

 

+
lllllilllllllllllllfllllllllllTl
100 0 -lOO

PPm

(b)
*
*

+
illiiiiIIHIIHHIHWTHHIHH
100 O -100 -200

PPm

Figure 6.13. Final products of separate fast cooling reaction of flux (4:2:8) in the furnace.
(a) the final products are K3PSe4 (*) at —l 13 ppm, and K4P28e6 (+) at 43 and 45.6 ppm

plus some unknown impurities. (b) the final products are K3PSe4 and a similar anion type
(*) at —99.3 ppm, and the two crystalline peaks at —18 and —19.8 ppm (+).

154

For the flux ratios belonging to group (2), representative ratios [(2:4:8), (3:4:8),
and (41428)] were subjected to slow cooling synthesis in the furnace in order to crystallize

products at —18 and —l9.7 ppm and the glassy feature. The final product is a completely

different feature and all three ratios produced pure K2P28e615 (first discovered by In
Chung and Prof. Kanatzidis) which is a P+4 type species which seems consistent with the
high temperature NMR data. K2P28e6 is a new recently discovered K/P/Se crystalline
15 . . . . 31 . .
phase. It has a chemical shift of 54.6 ppm which is close to the P chemical shift of
. . . . +4 . 10
K2Cu2P4Se10 and is conSistent w1th compounds haVing P-P or P type of bonding as
shown in Figure 6.14. The structure of K2P28e6 is composed of chains of P28e6 units

connected together by Se from each P28e6 unit as shown in Figure 6.15 below.

*

 

 

 

 

 

 

 

Www

Illlllllll'lllflIllllllllllllIllllllTTllP
300 200 100 0 -100
ppm

 

. 31 .
Figure 6.14. P MAS NMR of the final product formed from slow cooling reaction of

flux (3:4:8). The isotropic peak (*) at 54.6 ppm corresponds to crystalline K2P28e6. The
satellite peaks besides the isotropic peaks are spinning sidebands.

155

"Q
.Q

til. .\'

 

 

 

 

 

Figure 6.15. Structure of K2P28e6 viewed down the b-axis. The shaded circles with
bonds are the P atoms, the circles that are partly shaded are the Sc atoms, and the lightly
shaded circles with no bonds are the K atoms. (Courtesy of In Chung)

The bonding in K2P28e6 is similar to the bonding in K2Cu2P4Se10 wherein
[P28e6]2- units are connected together by Se atoms attached to each P-P units. In
K2Cu2P4Selo, the P-P units form cyclohexane rings, while K2P28e6 forms helical chains

. 15
of P28e6 units.

Ternary diagram for the KzSe/PZSe5/Se flux system. The ternary diagram shown

in scheme 2 summarizes the KzSe/PZSe5/Se flux ratios studied in this chapter. Figure

6.16 shows a plot of our results for flux ratios which gives crystalline final products (flux:

156

(2:1:8), (3:1:8), (4:1:8), (4:2:8)) and flux ratios which give amorphous final products plus
crystalline peaks at —18.0 and —-19.7 ppm (flux (111:8), (1:2:8), (1:4:8), (2:228), (213:8),
(2:4:8), (3:2:8), (3:4:8), (4:4:8)) in scheme 2. The distinction between the region of

having crystalline products and the region of having amorphous products is clearly

illustrated. Crystalline products are formed when the amount of Kzse is greater than the
amount of P28e5 in the presence of excess Se. Mixtures of crystalline and amorphous

products are formed when the amount of P28e5 is greater than the amount of KZSe. This

can also be shown stoichiometrically using equations 2 and 3 above.

Equation (2) shows that every mole of P2Se5 requires 3 moles of Se} (from
K2Se) reactants in order to produce 2 moles of PSe43- anions. Tracking this ratio of
P23e5:Se2- as 1:3 in the ternary diagram gives line (2) in Figure 6.16. Equation (3) shows
that every mole of P28e5 requires 2 moles of Sez- (from KZSe) reactants in order to

produce 1 mole of PZSe94- anions. Tracking this ratio of P28e5:Se2‘ as 1:2 in the ternary

diagram gives line (1) in Figure 6.16. After careful inspection of Figure 6.16, the lines

shown give a boundary that corresponds to the final product formed at that composition.

Ternary compositions corresponding to line (2) and above gives crystalline K3PSe4, and

ternary compositions corresponding to line (1) gives crystalline K4P28e9 which agrees
well with equations (2) and (3). Amorphous products are generally formed in the

presence of excess P2Se5 in this fast cooling regime.

157

$K2$e

 

 

+ Crystalline + amorphous
final products

K4PZSe9
A K3PSe4 + K4PZSe9
- K3PSe4
. K3PSe4 + unknowns

 

 

 

 

A A A A A A
80 70 60 40

90
$132585 'loSe

LII
O
m
o >
N
o
a
0

Figure 6.16. Shows a plot of our results for flux ratios which gives crystalline final
products (flux: (2:1:8), (3:1:8), (4:1:8), (4:2:8)) and flux ratios which gives amorphous
final products plus crystalline peaks at —18.0 and ~19.7 ppm (flux (1:128), (1:2:8), (1:4:8),
(2:2:8), (223:8), (2:4:8), (3:2:8), (3:4:8), (4:4:8)) in scheme 2. The corresponding

crystalline products are shown in the inset. Line (1) corresponds to P28e5zKZSe ratio
(1:2), and line (2) corresponds to P2865:K2Se ratio (1:3).

158

In summary, the reactions involved in the flux system KZSe/PZSe5/Se has rich
chemistry. In the relatively fast-cooling regime, flux systems with excess of K28e

compared to P28e5 generally form crystalline final products; otherwise a mixture of

crystalline and amorphous products are formed. Therrnodynamically more stable
products can be formed by doing slow cooling rate experiments on future in-situ NMR

measurements (as exemplified by the section on the effect of heating and cooling rates

. 31 .
discussed above). The P high temperature NMR measurements have shown that

equilibrium exist at high temperatures for these systems (see Figure 6.8).

159

REFERENCES

(1)

(2)

(3)
(4)
(5)

(6)
(7)

(8)
(9)

(10)

(11)

(12)

(13)

(14)

(15)

(a.) Sharp, K.W.; Koehler, W.H. Inorg. Chem. 1977, I6, 9, 2258.
(b.) Clark, R.J.H.; Cobbold, D.G. Inorg. Chem. 1978, I 7, 11, 3169.

(a.) Kanatzidis, M. G. Curr. Opin. Solid State Mater. Sci. 1997, 2, 139-149 and
references therein.

Pearson, R.G. J. Am. Chem. Soc. 1963, 85, #2, pp. 3533-3539.
Sutorik, A.C.; Kanatzidis M.G. Prog. Inorg. Chem. 1995, 43, 151-265.

Francisco, R. H. P.; Tepe, T.; Eckert, H. J. Solid State Chem. 1993, 107, 452-
459.

Francisco, R. H. P.; Eckert, H. J. Solid State Chem. 1994, 112, 270-276.

(a) Mutolo, P. F.; Witschas, M.; Regelsky, G.; Schmedt auf der Guenne, J .;
Eckert, H. J. Non-Cryst. Solids 1999, 256 & 25 7, 63-72.

(b) Maxwell, R.; Eckert, H. J. Phys. Chem. 1995, 99, 4768-4778.

(c) Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1994, 116, 682-689.

((1) Maxwell, R.; Eckert, H. J. Am. Chem. Soc. 1993, 115, 4747-4753.

Aitken, J .A. PhD Dissertation, Michigan State University, 2001.

Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Wiley: New York,
1987.

CG. Canlas, M.G.Kanatzidis, D.P. Weliky. Inorg. Chem. 2003, 42, 3399.

Dickerson, C.A.; Fisher M.J.; Sykora, R.E.; Albrecht-Schmitt,T.E.; and Cody,
J .A. Inorg. Chem. 2002, 41, 640-642.

Chondroudis, K.; Kanatzidis, M. G. Inorg. Chem. 1995, 34, 5401-5402.

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

K4P28e6 has also been examined by H. Eckert and (private communication).

Chung, I.; Kanatzidis, M.G. manuscript in preparation.

160

 

CHAPTER 7
SUMMARY
AND

FUTURE DIRECTIONS OF THE RESEARCH

161

The main goal of this project in general is to understand the art of exploratory
syntheses of chalcogenides and chalcophosphates using room temperature and in-situ
high temperature nuclear magnetic resonance spectroscopy (NMR). Using in-situ high
temperature NMR measurements on these systems, it is possible to monitor the reactions
during syntheses and determine the optimal conditions that favor a certain product over
another. These conditions can be determined by varying the compositions of the
reactants, and by varying heating and cooling rates. In this dissertation, metal

selenophosphates has been the main focus of this wide area of study. Chapters 3 and 4 in

this dissertation have laid down the basic 31P NMR properties of metal selenophosphates

by measuring the NMR of known compunds. And these NMR properties were used in

order to understand what is happening during their syntheses at high temperatures

(Chapter 5). Varying ratios of the K2Se/PZSe5/Se flux systems were then studied at room

temperature and high temperature in order to create a ternary diagram that can summarize

the behavior of this system at different compositions (Chapter 6).
Ab-initio calculations of the 31P NMR chemical shifts of metal selenophosphates.
A study has been initiated in collaboration with Professor Harrison at Michigan State

. . 31 . .
UniverSity to calculate the P chemical shifts of known structures of metal

selenophosphates using first principles. This will help in understanding how the P+4IP+5

chemical shieldings behave differently in different environments such as in the presence
of selenium or sulfur, and in the presence of metals of the same and different groups. This
study will give more depth in understanding the bonding chemistry of these class of

compounds.

162

 

I 77 . .
In—situ high temperature 3 P and Se NMR studies. Simultaneous 77Se and 31P

NMR at high temperatures might give meaningful assignments of peaks especially in the
fast exchange regime. Speciation and equilibrium models at high temperature can be
developed and tested. We know that the NMR peak during fast exchange at high

temperature is due to weighted average of the resonances of species taking part in the

exchange:1
5=X151+X252+X353+X484+... (1)
where xn (n=l,2,3...) are fractions of components 1,2,3...; 6“ (n=l,2,3...) are the
chemical shift of component xn.
In the study of the K/P/Se flux system, the known phases are: K4P28e6, K2P28e6,

K4P28e9, K3PSe4, and K6P88e132. By doing high temperature 31P and 77Se NMR on all

pure known phases to date, we can build a library of chemical shifts for pure
selenophosphate anions at high temperature. This is only feasible if the compounds do
not change in composition during the melting and cooling process. Based on the
experiments presented in Chapter 6, the chemical shifts at high temperature during the
initial cycle are about the same as the chemical shifts at high temperature during the
recycle period (with ~l-3 ppm difference due to hysteresis). This indicates that the

chemical composition of the formed product in the first cycle is conserved during the

recycle. Another example is the high temperature NMR of pure K2Cu2P4Se10 (presented

in Chapter 5, Figure 5.10) in which the chemical composition is conserved during the

process. The only limitation is that the melting points of the known phases should meet

163

 

the specifications of the high temperature NMR probe (maximum of 600 oC). Each

chemical shift at high temperature is unique for each anion which gives us values for on
(n=l,2,3...).
During fast chemical exchange for a certain flux ratio of KZSe/PzSe5/Se, the

chemical shift at high temperature is a weighted average of the resonances of all the

possible anions in the library:

5(T)= X(K4P2566 )5(K4P2366)+ X(K2P23€6)5(K2P2366) + X(K3PSC4)5(K3PSC4)+. . . (2)
. 31 77 . . . . .

where 8 is the P and Se chemical shift at high temperature. It is then poss1ble to solve

the values for the components xn in equation 2 using the equations from both 31P and

77Se high temperature NMR measurements. For example, we can propose a 2-component

equilibrium between anions [PxSey]n- and [Px’Seyt]m- with mole fractions x1 and x2 at

high temperature such as:

[PXSeyf' .—_x [Px~8ey~i"" (3)

And equation 2 becomes:

8 (31P)observed : X1 8 ([szeyIn-)P 4' (l'xl) 8 ([Px’sey’lm-» (4)
6(77Seiabsewea =x16(th8eyi“')3e + (1-X1)5([Px’sey’lm-)Se (5)
where 5([Px8eyi“'>p. 6([Px~Seyaim')p. 6([Px8eyi“')3e 6([Px~8ey»1m')3e are the

corresponding 31P and 77Se high temperature NMR chemical shifts of the pure

164

 

compounds in the library respectively. The values of XI and X; can then be solved

algebraically from equations 4 and 5.
A limitation of making a model is that unknown and still unidentified K/P/Se

phases might be present at equilibrium. Another limitation is that the number of

observables (MP and 77Se NMR) can only produce the same number of unknowns ( 2
species). 39K (spin 3/2) NMR might also provide useful information.
. . 31 77 ,
In-sUu high temperature P and Se NMR studies of other selenophosphate flux

systems. In this dissertation, the flux system KZSe/PZSe5/Se was studied. A wide

opportunity for study of fluxes is still available. To name a few are: (i) doing a high

temperature NMR experiment wherein slow cooling is done on the sample (ii) replacing
with other alkali metals such as Na, Rb, Cs. (iii) replacing P28e5 with elemental
phosphorous might be interesting because the latter has unoxidized P atoms and the
formation of PSe4 units might be reduced. (iv) addition of metals (transition metals and
possibly lanthanides) in a ternary flux system to produce quaternary compounds might be

interesting to discover new anions and phases (such as the ones found in Rb4Ti2P6Se25

etc.).
Room and high temperature studies of metal thiophosphates and thiophosphate

flux systems. Replacing Selenium by Sulfur exhibits a whole new chemistry to be studied.

. 2- . . . .
For example, the anion P286 which has a ring structure and no P-P bonding is more

common than having the selenium analog P2Se62-. Also, the 31F chemical shifts behave

165

 

differently with the replacement of Selenium by Sulfur. The chemical shift discrimination

between P+4 and P+5 is wider for selenophosphates (-120 to 100 ppm) compared to
thiophosphates (20 to 130 ppm)3. Therefore the assignment of peaks at high temperature

due to P+4 and P+5 will be more challenging.

In conclusion, the chapters presented in this dissertation may just be the beginning
of the story of the NMR of chalcophosphates and chalcogenides. Further studies on these
systems will definitely shed light on the art of exploratory synthesis of this class of
compounds. I expect that high temperature NMR spectroscopy will be useful for other

high temperature synthesis or reactions.

166

REFERENCES

(1) Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Wiley: New York,
1987.

(2) Chondroudis, K, Kanatzidis, M.G. Inorg. Chem. 1998, 37, 2582-2584.

(3) Regelsky, G. Ph.D. Dissertation, University of Muenster, 2000.

167

APPENDICES

168

The following figures shows the 31F high temperature NMR spectra for the reaction of

varying flux ratios of K2Se/P2865/Se as discussed in Chapter 6.

200 °C

A 400 °C

hmnuhmht fimwwwhfiww

   

500 oC

 

 

400 °Cd
300 °Cd
200 °Cr
r " *‘T 1*“zir ‘l
250 0 -250
PPm

300 °Cr
Ml...

400 °Cr

 

M
500 °Cr

w,

 
 
 
    
 

400 oCrd

300 °Crd

200 °Crd

 

 

r l' l I I l I T I I I
250 O -250

ppm

Figure A.1. 3)lP High temperature NMR spectra of the reaction of KZSezP28e5zSe with

ratio 1:1:8.

169

 

0

N
O
O
0

see

i

0

L»
O
O
('3

0

500 °C

H 400 °Cd
M 300 oCd

200 °Cd

%%

 

200 °Cr

300 °Cr

500 °Cr

300 °Crd

i

200 °Crd

 

 

 

 

L L
l

 
  
  

final product
V: 11 kHz
= 30003

 

I I I I I I I I I
200 0 -200

170

 

200 °C

%

“th

300 °C

400 °C

500 °C

400 °Cd

 

q i
i f

300 °Cd 500 °Cr

k A—A
VVV'W' V '7‘- ——~‘

200 °Cd 400 °Crd

fifi

 

300 °Crd

fl 200 °Crd

200 °Cr

 

 

 

 

 

300 °Cr
final product
v=12 kHz
400 °Cr l pd=45005
T I I I I I I I I I T I I I I I I I
200 0 -200 20° 0 ‘20“
ppm

PPm

Figure A.3. 31P High temperature NMR spectra of the reaction of KZSezP28e5zSe with

ratio 1:4:8.

171

 

200 °C 300 °Cd 500 °Cr

JEOO 0C 200 oCd 400 °Crd
' I‘Ml‘th‘vmsfl /

W
400 OC
I 200 °Cr 300 °Crd
500 °C
300 °Cr
I 200 oCI‘d

1......

final product
v: 12kHz

 

 

 

 

 

 

 
 
 
  

 

 

 

250 0 -250 WW MW pd=30003
m I l I l I I l I 1
PP 250 0 250 [HUIIHW
Ppm 250 o -250

P1331
Figure A.4. 31P High temperature NMR spectra of the reaction of KZSezPZSe5:Se with
ratio 2:2:8.

172

 

400 °Cr

WA 500 °Cr
WW “WWMW

400 °Crd

l 300 °Crd
200 oCrd

 

 

 

 

 

 

 

 

 

final product
v: 12 kHz
JEOOS
r I l l l I l l l
200 -200
1’1”“

Figure A.5. 31P High temperature NMR spectra of the reaction of KZSeszSe52Se with
ratio 2:328.

173

 

200 °C 300 °C d 500 Cr

 

400 °Crd

300 oCrd

L
f

 

 

 

 

 

 

 

 

 

 

200 °Crd
0 V: 12 kHz
400 Cd M pd: 4500s
IIIIIIIII‘II TITIITI'I'II IIIIIIIIIIII
200 O -200 200 0 -200 200 O -200
PPm PPm PPm

Figure A.6. 31P High temperature NMR spectra of the reaction of K28e:P2Se528e with
ratio 2:428.

174

 

500 °Cr

a a
O O
(i (i
O.- D-

400 °Crd

 

300 °Crd

 

 

 

 

 

 

 

 

200 °Cr I
L‘“

300 °Cr J 200 °Crd

0 final product
400 Cr v= 12kHz
WU pi: 45008
12(IX)IIII(I)IFFI260 IIIITTTIIIII
' 200 o -200
ppm
mm

Figure A.7. 31F High temperature NMR spectra of the reaction of K28e2P25e52Se with
ratio 324:8.

175

 

A 200 °C 1‘ 300 °Cd 500 °Cr

%

5:8

its
N
O
O
(i
D.

 

 

 

 

 

 

400 °Crd
that , l
400 °C 0
200 Cr 300 °Crd
MAW
I 500 °C H
300 °Cr j 200 °Crd
4 ° 1
00 Cd final product
400 °Cr v: 12kHz
llllIIITl pd:45008
200 0 -200 ,l / \ I |
PPm I l r 1 r I r 1 T 1 I [ I I r I I I I I I
200 pgm '200 200 0 -200
P9111

Figure A.8. 31P High temperature NMR spectra of the reaction of K28e2P286528e with
ratio 4:428.

176

 

The following figures as discussed in Chapter 6 are plots of 31P chemical shifts versus

Temperature for varying flux ratios of KZSe/PZSe5/Se from initial ramp-up in

temperature up to final recycle cool-down in temperature. The chemical shifts plotted are

in the fast exchange regime.

177

 

(a.)

 

 

 

 
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 
  
  

 

 

 

 

 

 

first cycle
40
y = 0.096x - 15.467 . ..
g 30 ‘ 2 +Imtial ramp-up
E Fl = 0.9999
0) 20 _ . ..
'5 +lnltlal ramp-down
E 10 ~
2 —— Linear (initial ram p-
0 o a down)
'10 7 r 1
150 250 350 450 550
Temp In degrees C
(b.)
Recycle
40
y = 0.1073X - 19.48 +ramp-up recycle
g 30 ‘ R2 = 0.9976(ramp-down)
if, 20 - +ramp-down recycle
8
“g 10 - — Linear (ramp-down
'2 Y = 0.1 135x ' 22.45 recycle)
0 0 ~ R2 = 0.9984(ramp-up) — Linear (ramp-up
recycle)
-10 w 1 1
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A.9. Plot of the dependence of 31F chemical shift with temperature for the flux

(121:8) reaction. (a.) corresponds to the initial reaction of KZSe, P2Se5, and Se. (b)
corresponds to the a temperature recycle after the initial reaction.

178

 

(a.)

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 
 
   
 

 

 

 

 

 

 

 

First Cycle

80

75 - y = 0.0808x + 34.47
5 70 - R2 = 0.9956
5 65 - +ramp-up
_§ 60 : +ramp-down
g 55 a —— Linear (ramp-down)
4‘ 50 —
0

45 —

40 r r 7

150 250 350 450 550
Temp in degrees C
(b.)
Recycle
80
75 - y = 0.0846x + 32.843
2
Fl = .

E 70 I 0 9998(ramp-down) ramp-up
m 65 -
_. +ramp-down
§ 60 A Linear (ramp-down)
g 55 a y = 0.0871x + 31.843 .
g 50 .. 2 — Linear (ramp-up)
0 R = 0.9987(ramp-up)

45 ~

40 1 . .

150 250 350 450 550
Temp ln degrees C

 

 

 

Figure A. 10. Plot of the dependence of 31F chemical shift with temperature for the flux

(1:228) reaction. (a.) corresponds to the initial reaction of K286, P28e5, and Se. (b)
corresponds to the a temperature recycle after the initial reaction.

179

(a.)

 

 

  
   

 

 

 

 

 

 

 

 

 

 

 

 

 
 
   
 

 

 

 

 

 

 

 

First Cycle
115
y = 0.0609x + 82.21
110 T 2
g R = 0.9979
‘2 105 ~ +mmp-up
3 + ramp-down
it 100 A ——Linear (ramp-down)
° 95 a
90 1 r 1
150 250 350 450 550
Temp in degrees C
(b.)
Recycle
115
y = 0.0637x + 81.08
3: 110 d R2 = 0.9977(ramp-up)
E + ramp-up
g 105 i +ramp-down
°' 2 — Linear ram -up)
5 100 y = 0.0625x + 81.35 . ( p
.c 2 —- Linear (ramp-down)
o 95 a R = 0.9913(ramp-down)
90 2 . r
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A. 1 1. Plot of the dependence of 31F chemical shift with temperature for the flux

(1:428) reaction. (a.) corresponds to the initial reaction of KZSe, P2Se5, and Se. (b)
corresponds to the a temperature recycle after the initial reaction.

180

 

 

Recycle

 

 

 
  

 

 

 

 

 

 

-30
-35 -<
t y = 0.0797x - 78.668
E -40 ~ 2 +ramp-up
2 R = 0.9986(ramp-up) + ramp-down
-45 ~
g Linear (ramp-down)
o -50 . ’ '
5 y = 0.0725x - 74.7 — “near (ramp Up)
'55 ‘ R2 = 0.9964(ramp-down)
-60 . T
250 350 450 550

Temp In degrees C

 

 

 

Figure A. 12. Plot of the dependence of 31P chemical shift with temperature for the Flux
(2: 1:8) that correspond to the temperature recycle after the initial reaction.

181

(a.)

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 
 
   
 

 

 

 

 

 

 

 

First Cycle
45
E 35 «
('71 +ramp-up
E 25 “ +ramp—down
é —— Linear (ramp-down)
5 15 A y=0.1063x-12.03
R2 = 0.9969(ramp-down)
5 I I I
150 250 350 450 550
Temp in degrees C
(b.)
Recycle plot
45
y = 0.1037x - 12.22
- 2
g 35 R = 0.9986(ramp-up) , rampup
m
a 25 - +ramp-down
'g — Linear (ramp-up)
e .
5 15 - y = 0.1006x - 11.01 —— “near (ramp-down)
R2 = 0.9954(ramp-down)
5 T r T
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A. 13. Plot of the dependence of 31P chemical shift with temperature for the Flux

(22228). (a.) corresponds to the initial reaction of KZSe, PZSe5, and Se. (b) corresponds
to the a temperature recycle after the initial reaction.

182

 

 

  
  

 

 

 

 

 

 

 

recycle plot
70
y = 0.1007x + 11.444
e 60 " R2 - o 9925(ram -u )
E ' ' p p +mmp~up
£32 50 ‘ +ramp-down
E 40 1 -— Linear (ramp-down)
0 ' -
5 30 _ y = 0.0978x + 13.334 — ”near (“mp up)
R2 = 0.9972(ramp-down)
20 I 7 I
150 250 350 450 550
Temp ln degrees C

 

 

 

Figure A.14. Plot of the dependence of 31F chemical shift with temperature for the Flux
(223:8) that corresponds to the temperature recycle after the initial reaction.

183

(a.)

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 
  
  

 

 

 

 

 

 

First Cycle
110
E
5 100 I +ramp-up
E -I—— ramp-down
as» 90 a — Linear (ramp-down)
5 y = 0.0739x + 68.01
R2 = 0.994(ramp-down)
80 1 1 1
150 250 350 450 550
Temp In degrees C
(b.)
Recycle plot
110
y = 0.0758x + 66.82
2
g 100 i R — 0.9982(ramp-up) ramp-up
% —I—— ramp-down
E ------ Linear (ramp-down)
o 90 a ______ , _
8 y = 0.073x + 67.85 “"95" (“amp Up)
R2 = 0.9925(ramp-down)
80 , 1 1
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A.15. Plot of the dependence of 31P chemical shift with temperature for the Flux

(22428). (8.) corresponds to the initial reaction of KZSe, P2Se5, and Se. (b) corresponds to
the a temperature recycle after the initial reaction.

184

 

First Cycle

 

 

I + ramp-down

 

Chemical SHift

 

 

 

0 200 400 600
Temp in degrees C

 

 

 

Figure A.16. Plot of the dependence of 31P chemical shift with temperature for
the Flux (3:1:8) that corresponds to the initial reaction of KZSe, P2885, and Se.

185

(a.)

 

 

 

    

 

 

 

 

 

 

 

 

 

 

 

  
 
  

 

 

 

 

 

 

 

First Cycle
-10

E

S

1’ '20 + ramp-up

§ —I-- ramp-down

E _ ' -

o R2 = 0.9913(ramp-down)

-40 T y r
150 250 350 450 550
Temp ln degrees C
(b.)
Recycle
-10
y = 0.0703x - 47.122
5 R2 = 0.9904(ramp-up)
5 -20 1 e ramp-up
g —l— ramp-down
“g — Linear (ramp-down)
g '30 7 y = 0.0693X - 46.672 —- Linear (ramp-up)
R2 = 0.9899(ramp-down)
'40 r I 1
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A.17. Plot of the dependence of 31P chemical shift with temperature for the Flux

(32228). (a.) corresponds to the initial reaction of KZSe, P28e5, and Se. (b) corresponds to
the a temperature recycle after the initial reaction.

186

(a.)

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

  
  
   

 

 

 

 

 

 

 

First Cycle
70

g 60 -

12 + ramp-up

_§ 50 1 +ramp-down

g y = 0 0864x + 22 91 — Linear (ramp-down)

‘ . .

° 40 R2 = 0.9995

30 1 1 1
150 250 350 450 550
Temp In degrees C
(b.)
Recycle plot
70
y = 0.1002x + 18.73
2

g 60 R — 0.9969(ramp-up) ramp-up
_ +
g 50 q ramp-down

E —— Linear (ramp-down)
0 R = 0.9957(ramp-down)

30 1 1 1
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A.18. Plot of the dependence of 31F chemical shift with temperature for the Flux

(3. 42 8). (a. ) corresponds to the initial reaction of K2Se, P2865, and Se. (b) corresponds to
the a temperature recycle after the initial reaction.

187

 

 

 

 

 

 

 

 

 

First cycle
-60

E
5 ~80 «
'5 -+- ramp-up
E -I- ramp-down
g -100 «
0

'1 20 I r r

150 250 350 450 550
Temp ln degrees C

 

 

 

Figure A.19. Plot of the dependence of 31P chemical shift with temperature for the Flux

(42128). (a.) corresponds to the initial reaction of KzSe, PZSes, and Se. (b) corresponds to
the a temperature recycle after the initial reaction.

188

(a.)

 

First Cycle

-20
-251
.30-

 

y = 0.065x - 59.8
R2 = 0.9937(ramp-up)

 

 
 
 
   

—O— ramp-up
+ ramp-down
Linear (ramp-down)
y = 0.0572x - 56.27 """" Linear (ramP'UP)
R2 = 0.9909(ramp-down)
-50 1 1 1

150 250 350 450 550

Temp In degrees C

 

Chemical Shift
do
01

 

 

 

 

 

 

 

 

 

(b-)

 

Recycle plot

 

 

_25 _ —O-—ramp-up
y = 0.0702x - 60.92

 
    

E

(f) '30 7 R2 = 0.9922(ramp-up) ' ramp-down
8 -35 - _

E , , , ------ Linear (ramp-
g -40 — down)

0

y = 0.0584x - 55.69

2 —————— Linear (ramp-Up)
n = 0.993(ramp-downl

 

 

 

 

'50 1
1 50 250 350 450 550

Temp in degrees C

 

 

 

 

 

Figure A.20. Plot of the dependence of 31P chemical shift with temperature for the Flux

(4:228). (a.) corresponds to the initial reaction of KZSe, P28e5, and Se. (b) corresponds to
the a temperature recycle after the initial reaction.

189

(a.)

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
     
 
  
 

 

 

 

 

 

 

First Cycle
40
y = 0.0976x - 15.26
g 30 ‘ R2 = 0.9988(ramp-down)
‘2 20 4 + ramp-up
§ —I— ramp-down
g 10 I —— Linear (ramp-down)
.:
O 0 g %
-10 1 , _
0 200 400 600 1
Temp in degrees C I
(b.) .
Recycle Plot L
40
2y = 0.104x - 17.7 I ramp-up
E 30 ‘ R = 0.9979(ramp-up)
('7) 20 4 , —--I-—ramp-down
g 10 . Y = 0.1005x - 16.3 ------ Linear (ramp-
3 R2 = 0.9949(ramp-down) down)
0 0 . ------ L1near(ramp-up)
-10 1 1 1
150 250 350 450 550
Temp in degrees C

 

 

 

Figure A.21. Plot of the dependence of 31F chemical shift with temperature for the Flux

(4:428). (a.) corresponds to the initial reaction of Kzse, PZSeS, and Se. (b) corresponds to
the a temperature recycle after the initial reaction.

 

190

IIIIIIIIIIIIIIIIIIIIIIIIIIIII

IIIIIIIIIIIIIIIIIIm