LANTHANI‘DE OXYCARBIDES:
PHASE AM} EQUILIBRIUM STUDIES

Thesis for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY
A. DUANE BUTHERUS
1969

 

 

This is to certify that the

thesis entitled

LANTHANIDE OXYCARBIDES:
PHASE AND EQUILIBRIUM
STUDIES

presented by

A. DUANE BUTHERUS
has been accepted towards fulfillment

of the requirements for

PhoDo degree in Chemistry

:z/ew/I/

%r professor

Date August 6, 1969

 

0-169

 

Michigfzn 3 Cafe

 

1 L113] ,‘
Universit
L ’J

A

"5'
J

ABSTRACT

LANTHANIDE OXYCARBIDES:
PHASE AND EQUILIBRIUM
STUDIES

by

A. Duane Butherus

Introduction'

 

Oxycarbide phases of uranium have been reported
by several workers. A report from this laboratory, however,
first showed the existence of a ternary oxycarbide phase in
a lanthanide system. In this thesis, studies on two
distinctly different Ln(O,C) phases (Ln = lanthanide) will
be reported in detail. The Ln202C2 phase has been prepared
only for Ln = Nd, La. The LnuOBC phase appears more general
and has been prepared for Ln = La, Nd, Gd, Ho, Er. The
Ln 0 C system has been studied quite thoroughly thermo—

2 2 2

dynamically, though less so structurally. The LnuO C system

3

on the other hand has been more thoroughly studied structurally.

Techniques

 

Preparative methods included induction heating
in various crucibles and in sealed bombs, arc melting,
electron-beam zone refining and arc zone refining. Pre-
parative techniques were developed with the aid of phase
purity studies, accomplished via X-ray powder diffraction

methods and micrography.

Thermodynamic studies were effected through static
methods, by utilizing induction heating to maintain sample
temperatures. Carbon monoxide, the only gaseous component

in the equilibrium system described by the equation

Ln203(s) + 3C(s) = Ln202C2(s) + CO(g), (1)
exhibited an equilibrium pressure in the range of l to 150
torr for reaction temperatures of l350° to 1850°. Both
second- and third-law procedures of data reduction were
employed.

Analytical procedures included gravimetric analysis
for the metal, combustion for carbon, and vacuum fusion for
oxygen. Structural studies utilized X-ray powder diffrac-

tometry.

Results of Study

 

The equilibrium system described by equation (1)

gave the following thermodynamic data:

AH§98 (Ln = Nd) = 75.9 _ 3.0kca1/gfw
AH§98 (Ln = La) = 82.2 i 1.9kca1/gfw
AS° (Ln = Nd) = 33. i

298 3 [lane‘U/gfw

A8298 (Ln = La) 37.7 i 3.6eu/gfw

The standard enthalpies of formation at 298° are computed

to be

AH; (Nd2O202(s)) = -329.8

l+

3 kcal/gfw,

H-

AHf (La202C2(s)) = —320.0 2 kcal/gfw;

and the standard free energies of formation at 298° are

AGf (Nd202C2(s))

-309.9 i 3.9 kcal/gfw

AG; (La20202(s)) = —315.6

H-

2.9 kcal/gfw.

The NduO3C phase is shown to possess a NaCl-type
fee structure having a lattice parameter a0 = 5.1406 i 0.0007A,
and appears to have oxide and methanide ions distributed

randomly in the anionic lattice sites.

LANTHANIDE OXYCARBIDES:
PHASE AND EQUILIBRIUM
STUDIES

f by

A“
\L
AX Duane Butherus

A Thesis

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

DOCTOR OF PHILOSOPHY

Department of Chemistry
1969

ACKNOWLEDGMENTS

My appreciation goes first to Dr. Harry Eick,
whose wisdom, tolerance, guidance and patience made this
work possible. Much invaluable help was also extended by
John Haschke, whose discussions concerning some of the work
were a great aid. The support of the U. S. Atomic Energy
Commission (Contract AT(ll—l)—716) is also gratefully
acknowledged. Finally my most grateful thanks to my wife
without whose patience and understanding this work would

not have been started.

_ 11 -

TABLE OF CONTENTS

Abstract
Acknowledgments
Table of Contents
List of Tables
List of Figures
List of Appendices

Chapter I, General Introduction

Chapter II, La202C2 Phase

A. Introduction, La202C2
B. Theoretical Considerations
1. Thermodynamics
a. Relationships from the second
law of thermodynamics
b. Heat Capacity Effects
0. Relationships from the third
law of thermodynamics
2. Instrumental Theory
a. Optical Pyrometry
b. Absorption Corrections in
Pyrometry
c. Two-Color Pyrometry

C. Experimental Methods, Ln20202
1. Synthesis

. Starting materials

. General techniques
Containers
Synthesis I

Synthesis II
Synthesis III
Synthesis IV

. Synthesis V

2. Phase Purity

3. Analytical Procedures

. Metal Analysis

. Total Carbon Analysis
. Hydrolysis Experiments
. X-Ray Investigations
Infrared Studies
Equilibrium Studies

U'OQPUQQIOO‘W

WQQOU‘Q’

- iii -

Table of Contents (cont.)

page
D. Results of Ln 020 Investigations #7
1. Preparatige R sults 47
2. Analytical Results A9
3. Hydrolysis Results 51
A. X-Ray Diffraction Results 52
5. Infrared Results 55
6. Equilibrium Results 55
E. Ln 0 C , Conclusions and Discussion 68
l. eneral Discussion 68
a. Stoichiometry and Bonding 68
b. Structure 71

2. Relationship of Lanthanum and
Neodymium Analogues of Ln2O2C2 73
a. Structure 73
b. Thermodynamis 74
3. Phase Relationships 76
Chapter III, The LnuO3C Phase 77
A. Introduction 77
B. Experimental Methods 78
1. Preparation 78
2. Analysis 84
a. Metal Analysis 84
b. Carbon Analysis 85
c. Oxygen Analysis 87
d. X-Ray Analysis 92
3. Hydrolysis Experiments 93
A. Equilibrium Studies 94
C. Results, Ln 0 C 95
l. Prepara 08y Results 95
2. Analytical Results 98
3. X-Ray Analysis Results 99
4. Hydrolysis Results 100
D. Discussion LnuO3C Phase 102
Chapter IV, Future Research 110
A. Ln202C2 Phase 110
1. Structure Determination 110
2. Bonding Studies 110
B. LnuO3C Phase 112
1. Structure and Bonding 112
2. Thermodynamics 112

3. Proposed New Phases, Based

on Chapter III 112

-1V-

Table of Contents (cont.)

Chapter V, Bibliography

Chapter VI, Appendices

114

118

Table I:

Table II:

Table III:

LIST OF TABLES

Observed Nd O C X—Ray Powder Diffraction

2 2 2
Data-coco. ....... O ....... OOOOOOOOOOOOOIOOOOOOO page 53
Observed La202C2 X-Ray Powder Diffraction
Data-.0. 0000000000000000 OOOOOOOOOOOOOOOOOOOOOO page 54
Calculated and Observed Diffraction
Itensities for (111) and (002)
Planes for NduO3C.. ..... . ..... . .......... .... page 100

-V1-

Figure
Figure
Figure

Figure

Figure

Figure
Figure
Figure

Figure

6.

7.
8.

9.

LIST OF FIGURES

Arc Melter................ ..... ........

Combustion Apparatus.........

Double-Walled Graphite Crucible........

Pressure of Carbon Monoxide in

Equilibrium with Nd2 2 2.

O C .............

Pressure of Carbon Monoxide in

Equilibrium with La O C

2220....

Oxygen Analysis System......

Pressure Measuring System.............

Arc Zone Melter.................

Electron-Beam Zone Melter...

- vii -

. page

page

page

page

page
page
page
page

page

30
MO

59

6O

61
88
120
122

124

LIST OF APPENDICES

Appendix I: Source and Purity of Materials ............ page 118

Appendix II: High-Speed Pumping System and

Current Concentrator ...................... page 119

Appendix III: Tentative Indexing of La202C2 ............. page 121
Appendix IV: Arc Zone Melter ........................... page 122
Appendix V: Electron-Beam Zone Melter ................. page 123

Appendix VI: Calculated Intensities of
Diffraction Lines for Nd403C in
ZnS- and NaCl-Type Lattices ............... page 125

Appendix VII: Effective Anionic Radii ................... page 126

- viii -

CHAPTER I. GENERAL INTRODUCTION

A. Literature Concerning Lanthanide Oxides, Carbides,

 

and Oxycarbides

 

Since the original discovery of a lanthanide
compound (Arrhenius, 1788, described by B. R. Geijerl),
most of the chemistry of the lanthanide elements has
involved the oxide system or compounds easily derived
from the oxides by classical solution techniques. Such
compounds as nitrates, oxalates, carbonates and sulfates,
were used in the early separation techniques.2

Carbides of the lanthanides were first reported
3

by Pettersson, who described the preparation of the di-
carbides of La, Y, and Ce. The preparation was effected
by melting together a mixture of the oxide and carbon in
a high current electric arc. The formulas reported, based
Pettersson

on analytical work, were YC and LaC

1.87 1.85‘
was probably trying to reduce the oxides to form the
metals, and it is possible that the samples contained
considerable quantities of oxygen. Moissan et_§l,4’5’6
prepared and reported hydrolytic studies on the lanthanum,
neodymium and praseodymium dicarbides. Only within the
past decade, however, have questions been raised as to

the possibility of forming ternary metal-oxygen-carbon

phases.

_ 2 -

B. Carbides as Pseudo-Oxides

 

In the first approximation, new ionic solids
of a given metallic element can be thought of as being
formed from known compounds by exchanging the anions
present for different ones. Thus the halides of the
lanthanides may be thought of as being formed by sub—
stituting two halide ions for each oxide ion in the

sesquioxide represented by the formula LnO In the

1.5'
case of the carbides, the substitution may be considered
as the replacement of an oxide ion O: by an acetylide ion
02: accompanied by partial reduction of the metal.

Partial replacement of one type of anion by
another is well known in the lanthanide—oxygen—halogen
systems. The oxyhalides, LnOX, for example, may be considered
a replacement of an oxide ion by two halogen ions in the
formula 1/2 Ln203. This thesis prOposes that a similar
model of partial replacement of anions describes that lantha—
nide—oxygen—carbon system with carbide units replacing a

fraction of the oxide units in the lanthanide oxide. By this

model, carbide ions may be considered to be pseudo oxides.

C. Purpose of this Research

 

This thesis reports the further elucidation of

the work of G. L Buchel,7 which prOposed the existence

of ternary lanthanide-oxygen—carbon systems, and the
description of other phases discovered in the Ln(O,C)

ternary systems.

CHAPTER II. THE Ln O C PHASE

2 2 2
A. INTRODUCTION, Ln202C2

Equilibrium systems of the general type
MOX(s) + (x+y)C(s) = MCy(s) + xCO(g) (1)

have been studied by a numer of workers for the phases
formed when M = uranium.8"11 As early as 1925, Otto
Heusler,8 using a vacuum furnace, reported equilibrium
data over the temperature range of 1480° to l80l° for the
above reaction when x and y both equal 2. Further research
into the uranium—carbon system, however, disclosed the
possible presence of other uranium-carbon phases in this
temperature range. The existence of the monocarbide up
to temperatures of 1800° was confirmed in 1948,12 and
the existence of the sesquicarbide was confirmed, but
there is some doubt about the temperature range over
which it exists.13 In addition, several crystal modifi-
cations of the phases may be present. Complicating the
uranium-carbon system even more is the existence of
uranium—carbon-oxygen solid tenary phases which may also
be in equilibrium with carbon monoxide.lo’14’15 Because
of these complications, most current thermodynamic in-
vestigations in this system are accomplished by vapori-
l6

zation methods.

-5-

Since no oxycarbides had been reported for the
lanthanide elements, and since their carbon systems were
relatively well-known, with LnC2 being the most stable
phase over the general temperature range of IOOO° to 2OOO°,17
earlier workers in this laboratory initiated equilibrium
studies according to reaction (1) (in which M = Nd, x = 1.5,
and y = 2). It was thought that few competitive processes
would occur to complicate the study. The reaction was based
on the familiar preparation of the dicarbide under vacuum
first used by Moissan in 1896.4

This reaction was carried out in a closed, evacuated
vessel, and did indeed appear to be an equilibrium process
since the pressure of the system increased from zero to
some constant value for a given crucible temperature.
Investigation of the crucible contents by X-ray powder

diffraction, however, showed the existence of the Nd203(s),

18
2.

X-ray powder diffraction pattern that could not be indexed

but no lines of the well-characterized NdC Instead, an
on the basis of any known neodymium-oxide or neodymium-
carbide phase was observed.

This new phase, which was found to exhibit the
stoichiometry NdOC, and its isostructural lanthanum analogue,
LaOC are the basis for the experimental work reported in

Chapter II of this thesis.

B. THEORETICAL CONSIDERATIONS

l. Thermodynamicslg’ 20

 

The behavior of all heterogeneous physical systems

at equilibrium is described by the phase rule of Gibbs,
f = c - p + 2, (2)

which summarizes the relationship between the number of

phases present (p), the minimum number of components necessary

to describe the system (c), and the degrees of freedom

(f). The degrees of freedom specify the number of intensive,

independent variables which describe the system.21
The system described in this part of the thesis,

which will be shown to be

Ln2o3<s) + 3c<s> = Ln202c2(s> + co<g>, (3)

contains four phases (sesquioxide, graphite, oxycarbide,

and carbon monoxide) and three components (lanthanide

metal, carbon, and oxygen). Thus, there is but one degree

of freedom, or one independent variable in the set of intensive
variables describing the system. The activities of the

solids are invariant, as will be shown, so only the variables
of temperature and pressure are needed to describe the

system. Since only one of these latter two variables is

independent, a unique pressure of carbon monoxide must be
in equilibrium with the three solid phases at any given
temperature.

By measuring the equilibrium carbon monoxide
pressure as a function of temperature, the free energy
change accompanying reaction (3) may be determined; with
additional thermodynamic data, this measured free energy
change can be converted to the standard free energy change
for the given reaction. The standard enthalpy and free
energy of formation for the oxycarbide at 298° can also be
determined, by use of the measured free energy change and
additional reported thermodynamic data.

(a) Relationships from the Secondllmvof Thermodynamics

 

The standard free energy change for the reaction

 

tL + mM = qQ + rR (4)
may be represented by
.3 afi
-AG$ = RT 1n Keq = RT in t (5)
3L 8M
where as represents the activity of the equilibrium component

Q in equation (A) raised to the power q, etc. In equation (3),
all phases except carbon monoxide are solids having extremely
low vapor pressures; therefore they can be considered to

exist in their standard states, and to possess unit activities

(standard states for solids are usually defined as the pure
solid in the most stable form at the specified temperature
and a pressure of one atmosphere). If the gas pressures
never exceed 0.5 atm. over the temperature range of interest,
the pressure of the gas may be assumed identical with its

fugacity and equation (5) may be re-written

-AG$ = RT 1n PCO(g)° (6)

Combination of this with the equation

wag = AH; - TAs; (7)
yields the equation
1 o AST
11’] PCO(g) = " “T AHT + T. (8)

It is apparent that a graph of the logarithm of the carbon
monoxide equilibrium pressure versus the reciprocal of the
corresponding absolute temperature will produce a straight
line from whose lepe and intercept AH° and AS° respectively

T T

may be deduced by assuming AHE to be constant over the

temperature range of interest. The implications of this

assumption will be investigated in the following section.

(b) Heat Capacity Effects

 

The thermodynamic functions, AH and AS are always
measured relative to some reference state which in this
work will be the described system at a reference temperature
9 = 298.16°F. In the general case, the heat capacity change
ACP at constant pressure (for one mole of product formed)
is a function of temperature, and the relationships between

the standard enthalpy and entrOpy at some temperature T

and the reference temperature 9 are

AHT = AHG + .[ ACP dT (9)
9
and
o _ o 1
AST — Ase + ‘[ ACP T-dT. (10)

9
Equations (9) and (10) are usually simplied by one of

three assumptions. First, if AC = O, the

P

AH§ = AHg (11)
and

As; = A85, (12)

thus the plot of 13 PCO vs. %-does in fact define a straight
A
line of lepe -—:E. This assumption is probably valid only

_ 10 -

over small temperature ranges. For larger temperature
ranges, the plot of 1n P X§,%-often shows curvature, and
the assumption of ACP = O is inaccurate. Consequently,

the second assumption is that the heat capacity change

is not zero but is a constant function of temperature.

By using this assumption, equations (9) and (10) are readily

integrated, giving

AH; = AHg + ACP(T—9) (13)
and
AS$ = A8; + ACP ln(T/9). (14)

Equations (13) and (14) may be applied in the following
manner: If the ln P y§,%-graph is relatively straight, but
the mean temperature T is very much different from the
reference temperature 9, the measured slope and intercept,
representing data taken at T must be corrected to the
reference temperature by use of equations (13) and (14).
If, however, the ln P 122% slope is not linear, the change
in AH° and AS° resulting from heat capacity variation must
also be included in the slope and intercept calculation.
This is readily accomplished by the so-called "Z-Plot Method."
In this method, equations (13) and (14) are sub-

stituted into equation (8), which after collecting terms yields

_ 11 -

AH°
9 6 o
-R1n P + ACP [‘T + 1n T ] =-—T— - ASe + ACP(1+ln6). (15)
Defining
— R1 P AC §-+ 1 T ]
‘ n ' ' P T n
and

I = Asg = ACP [1+1n9]

and substituting these into equation (15) gives

-AHg
E:= T + I. (15)

If this assumption is correct, a plot of 2 vs. %- will

 

produce a straight line of slope AHg and intercept I, from
whose definition A85 may be deduced.

The third method involves writing AC as an

P
analytic function of temperature, followed by integration

of equations (9) and (10). The results of these integrations
are substituted into equation (8), followed by a treatment

analogous to (but more difficult than) that used in the

"Z-plot method." This treatment, though the most rigorous,
is usually impossible in high temperature chemistry because
of the lack of analytic functions for heat capacity relation-

ships.

-12...

(0) Relationships from the Third Law of Thermodynamics

 

Thermodynamic calculations may be made at various
temperatures with the aid of the so—called "free energy
functions," henceforth abbreviated fef, (where 6 is normally

298.16°K)

G° - H° H° — H°
_(. .) _ . (T .)
fef _ T _ —sT + T (17)

and eXperimentally-determined enthalpies and free energies of
formation. This method is particularly useful in calcu-
lations involving high temperature thermodynamics for

several reasons:

1. It provides a compact method of storing large
amounts of thermal data covering wide temperature
ranges, since the fef varies smoothly with
temperature. Hence only relatively few fef
points need be tabulated XE- temperature for
construction of a graph from which fef values
may be deduced.

2. If the required fef's are available for the
computation in question, the enthalpy calculated by
their use is independent of the enthalpy calculated
by the second law method outlined previously,
since the fef's are dependent on third—law cal-

culations. This dependency will be demonstrated.

_ 13 _

3. If all required fef's are available, the possi-
bility of complicating reactions such as
sample - crucible interaction can be investigated.
4. If an fef is unavailable, but can be approximated
by analogy, third law enthalpies can still be cal-
culated. These enthalpies, though possibly
inaccurate in absolute value, still establish an

independent check of second-law equilibrium data.

The dependence of this method on third-law
information can be seen by the fact that the right side

of equation (17) can be evaluated by the third—law method

ll

of integrating heat capacity data from T O to T = T, the

temperature in question. The first term may be evaluated

as follows:

T1 T2 T

C C AH C
_ P P tr P
O T

The summation term is an approximation to the integral

T1

t[ EP_dT. T1 is usually selected to be a very low temperature
T

0

(some tens of degrees) and the first term evaluated either

graphically or by a Simpson's rule summation. The second

-111-

term of the right side of equation (17) can be evaluated

by a similar computation.

 

T T T
1 1 2
T. L[ CP dT + AHtr + CP dT +...+ CP dT . (19)

9 T1 T1

In the thid-law method, problems arise when
a needed fef is unknown and cannot be estimated with any
degree of certainty, because of a lack of thermodynamic
information about related compounds. In such<nfies, a
model of the particular phase is postulated and an fef
based on that model is calculated. If the fef appears
reasonable, the enthalpy value computed from this fef can
be used as an independent corroboration of the value computed
by the second-law method. The amount of confidence placed
in the numerical values resulting from such a calculation
will be related to the information supporting the model

used.

2. Instrumental Theory

 

(a) Optical Pyrometry

 

Planck's general law of radiation for a black

body is

 

C1
JAT= 5
’ x exp(02/AT)-l (20)
where
2
Cl — 2U 1n 0
C2 = hc/k,

c is the speed of light, and A is the light wavelength.
This equation describes the intensity of a black body
radiator as a function of wavelength and temperature.

If the wavelength is held constant in equation (2cn,the
intensity JA,T will be a unique function of temperature.
This law is the basis for the monochromatic Optical
pyrometers used to obtain all the experimental temperatures
reported in this work. A red filter transmits a band of
wavelengths around 6500 A, and the intensity of this band
is compared with that of a standard filament viewed through
the same filter. The intensity of the standard filament
which is superimposed on the image of the source is ad—
Justed by a slide wire until the two are of the same
intensity. The voltage drop across the standard filament
is then compared with a standard cell and the temperature
read from the calibrated slide-wire. The measurements made
by such a instrument have been found reproducible to

i2° up to 1750°. When the instrument is calibrated against

NBS* standards, measurements are presumed accurate within the

 

*
National Bureau of Standards

_ 16 -

standard deviation of the measurement and the error limits
established by the NBS for its standards at the measured
temperatures. The radiation measured, is assumed to emanate
from a perfect black body radiator.*

(b) Absorption Corrections in Pyrometry

 

Correction of radiation for absorption by the
external optical system is most easily accomplished
empirically, since the constants needed are difficult to
calculate. A combination of Beer's law for absorbance and
Wien's law of radiation, which is an empirical law more
easily handled than Planck's, is used for the correction.
Wien's law predicts intensities which are within 1%
of the true values in the Spectral region utilized in the
monochromatic pyrometer.

The following relationship between true temperature
T and observed temperature Ta viewed through an absorber

is derived as indicated above:

 

This assumption can be supported by using the approximation
that the number of internal reflections occurring before
escape of a given quantum of radiation is represented by
the ratio of the area of origin to that of the cavity
interior. (In a typical crucible used thnaughout the

work reported in this thesis, this ratio had a value of
greater than 700.)

T T _.—C . , (21)

The transmittancy of the absorber is represented by T.
Since the transmittancy is invariant for a monochromatic
pyrometer, the entire right side of this equation is

constant, and equation (21) can be written

11_
T'- T_'_ C. (22)

This equation allows a very simple experimental determination
of C for a set of points defined by T and Ta.

(0) Two-Color Pyrometry

 

Since the intensity of a particular wavelength
of black-body radiation defines the temperature of that
body, the ratio of intensities of two different wavelengths
can also be shown to define a black body temperature. The
major advantages of two-color pyrometry are:

l. The temperature measured is independent of the
absorber so long as the absorber does not
selectively transmit one of the two wavelengths
used.

2. In the absence of a black body radiator, the
correction term 8, denoted the "emittance" of the

body, relates the flux characteristic of the

true temperature, J to the observed flux

)\,T,

of a non—black body radiator, Ji The equation

.T'
is

I

J)”T = eJx,T. (23)

In two-color pyrometry, the ratio of two fluxes

is measured as is indicated in equation (24),

J 5 J

812T = l A1,T (24)
J 8 J

A2,T 2 A2,1?

If it can be shown that 81 and 52 are equal,

the emissivity corrections cancel, allowing true black

body temperatures to be measured from surfaces.

_ 19 _

C. EXPERIMENTAL METHODS, Ln202C2

1. (Synthesis

 

Preparatory techniques used in synthesizing
the Ln2O2C2 phase were develOped over a period of several
years. Improvements in product purity, ease and speed of
preparation, and reproducibility were due almost entirely
to the availability of new instruments during the time
involved. For this reason, the various preparatory
methods will be described individually in chronological
order along with descriptions of the instruments employed
in that particular synthesis. Several general instrumental

techniques used in most of the work reported will be

described separately.

(a) Startinngaterials and Their Handling

 

Materials used in the preparation of the oxycarbide
phases were the lanthanide metals in ingot form, lanthanum
(99%), neodymium (99%); the correSponding powdered sesqui-
oxides of lanthanum (99.9%) and neodymium (99.9%), and
grade no. 38 graphite powder. Sources of these materials
are listed in Appendix I.

The sesquioxides tend to absorb carbon dioxide

from the atmosphere to form the metal carbonates. Since

_ 20 _

appreciable amounts of such carbonate impurity would lead
to incorrect amounts of the metal and oxygen in the
preparatory mixtures, the oxides were calcined at 900°

in platinum crucibles for a minimum of eight hours in a
muffle furnace. The crucibles were then cooled and
stored in desiccators. The metal ingots were stored in a
vacuum desiccator and were exposed to air for minimal
time when samples were removed from the stock material.
Metal samples of O.5—1.0g were removed from the ingots
either by cutting pieces of the approximate size required
with a clean and sharp cold chisel, or by filing with a
previously cleaned, coarse mill file. Samples prepared
by the filing method were subjected to magnetic cleaning
to remove possible contaminants chipped from the file.

To minimize contamination by surface oxidation, metal
samples were used either immediately or stored in a vacuum
desiccator for no longer than one day. The graphite
powder was used without further purification.

(b) General Techniques

 

Described here are techniques and instruments
which will be referred to throughout the remainder of this
thesis, and will be noted only briefly subsequent to
these descriptions. Other more specific techniques will

be described in the context of their use.

_ 21 _

(1) Handling materials in the Absence of Oxygen and
Moisture. Many manipulations had to be carried out in the
absence of atmospheric moisture and/or oxygen, due to the
great reactivity of some of the materials studied. For
such manipulations, a helium-filled glove-box was
employed. The helium in this glove box was dried over
phosphorous pentoxide and recirculated through a system
which employed molecular sieve drying columns. Samples
which would have hydrolyzed or oxidized very rapidly in the
air were brought into the dry box quickly after removal
from the apparatus in which they were prepared. Exposure
to air never exceeded about fifteen seconds for such
materials. In all following descriptions, the term "glove

box refers to this apparatus.

(ii) Temperature Measurements. Except for a few

 

relatively low temperature (600 - lOOO°) measurements

for which thermocouples were employed, all temperatures
reported in this thesis were measured by a model 8622—C
Leeds and Northrup disappearing—fllament—type pyrometer
(serial no. 1572579). This pyrometer had been calibrated
at the NBS, and its calibration was checked by comparing
it to a similarly calibrated instrument reserved for

comparison purposes. A tungsten strip lamp powered by

- 22 _

a voltage-stabilized source which had a six-step ouput
was used for comparative measurements. The temperature
of the strip lamp was measured by the two pyrometers

at each of the six temperatures produced by the stepped—
output transformer. Care was taken to measure the
temperature of an identical section of the tungsten strip
each time. From the correction data supplied by the NBS
a calibration curve was constructed to enable comparison
of the two pyrometers.

A similar technique was used for correcting the
observed temperature when readings were taken through
a prism and an Optical window. The temperature of the
tungsten strip was determined at each of six temperatures,
by viewing the strip, first directly, then through the
window—prism combination.

These data were used with equation (22) which
relates absorbancy to temperature change, and the constant
C for the prism-window combination was determined. With
this constant a curve which covered the temperature range
of 900 — 2200° was prepared to relate correction values
to the observed temperature. This curve was combined with
the NBS calibration curve for this pyrometer. This overall
correction curve was then used for all measurements re-

quiring that particular Optical arrangement.

_ 23 _

(iii) Metallogrephy (Micrography). Micrography is

 

a general technique used to prepare and examine cross-
sections of fused materials. Samples were mounted in
Met-A-Test cold mount material (Precision Scientific
Co., 3733 w. Cortland St., Chicago, Ill.). When the
mounts had hardened, the samples were ground on a
Precision Scientific CO. electric polisher, by using
successively numbers 180, 320 and 600 abrasive papers;
then polished on a fiber platten charged with number 600
abrasive powder. If magnifications of 800x or greater
were to be used, a final polish was effected on a Buehler
Vibromet polisher (Buehler Ltd., 2120 Greenwood St.,
Evanston, 111.) by using 1.0 micron alumina abrasive.

During grinding and polishing operations, the
sample was bathed continuously with a thin, anhydrous
mixture of paraffin oil in kerosene. This mixture was
flowed in a thin stream onto the abrasive surface in front
of the sample. This lubricant provided a solvent for the
abrasive operation and also prevented hydrolysis of the
sample surface by atmospheric moisture. This lubricant
was continuously filtered, dried and re—circulated by a
peristaltic-type pump.

The polished face of the sample was washed

thoroughly in a stream of the freshly-filtered kerosene

-211-

mixture, the excess was wiped off and a microsc0pe slide
cover glass was placed over the polished sample surface
to prevent hydrolysis. If a permanent sample was desired,
the cover glass was affixed with a drOp or two of Canada
balsam. Interpretation of micrographs will be discussed
later in this chapter.

(iv) X-Ray Powder Diffraction. These techniques,

 

henceforth referred to as "x-ray diffraction" in this
thesis, were used in the preparatory development, but will
be described in the analytical section.

(c) Containers

 

Except where otherwise noted, the containers
used both for preparatory methods and later for equilibrium
studies were cylindrically-shaped graphite crucibles of
2 cm diameter and 3 cm height, with walls of at least
5 mm thickness. The crucibles were covered with close-
fitting graphite lids of at least 5 mm thickness, with
a 0.8 mm hole drilled axially. This lid permitted gases
to escape and provided a line of sight into the crucible
cavity for temperature measurements. These crucibles
were outgassed in the system described in Appendix II to
a temperature of 2000-2050° and an accompanying pressure

of less than 5 x 10-6 torr for a period of not less than

_ 25 _

three hours. Graphite served as a good crucible material
for the following reasons:
1. Graphite contamination did not affect
subsequent analysis of the product.
2. In equilibration studies, graphite was
shown to be one of the solid components of the
equilibrium mixture, and the crucible therefore
could not contribute any interfering contaminants
to the process being studied.
Different crucible materials were used infrequently, and
they will be described in the context of their use.

(d) Synthesis I

 

This procedure was developed during the discovery
of the phase and is therefore somewhat crude. The pelletized
lanthanide sesquioxide and graphite mixture was placed in
an outgassed graphite crucible which was positioned on a boron
nitride support. A double-walled, water—cooled Vycor tube
fitted with an optical window at its top surrounded the
crucible and permitted attainment of a pressure of approxi—

mately 10-5

torr. An oil diffusion pump and a mechanical
backing pump completed the vacuum unit. Since it was found
that the reaction did not proceed at a measurable rate

below l300°, as evidenced by a stable pressure in the system

—26-

when it was evacuated, closed and maintained at that tempera—
ture, the reactants were outgassed by heating the crucible and
contents at 900 - 1000° under vacuum for 30 minutes. The
system was then sealed and the crucible quickly heated to the
desired operating temperature at which it was maintained for
3.5-5 hours. When the product had cooled, it was transferred
to the glove-box and a small representative portion of the
pellet was removed for x-ray diffraction analysis. If
sesquioxide was present, the remaining sample was placed back
into the heating system, and the previous procedure was
repeated; except that after the operating temperature had
been reached, some of the gases produced by the reaction

were pumped off. The pressure change due to gas removal was
monitored by a stainless steel absolute pressure manometer
(Wallace & Tiernan model FA-l45). Usually two such pump-offs
were required for a 3g sample before the intensity of x-ray
diffraction lines attributable to the sesquioxide faded to

a minimum value.

(e) Synthesis II

 

This method is a further develOpment of synthesis I,
made possible by the availability of a new equilibrium
system which permitted the achievement and maintenance
of internal gas pressures from 5 x 10'7 to 700 torr.
(This is the same system that was later used for equilibrium

eXperiments and its operation will be described fully in

_ 27 _

that section). The procedural modifications made possible

by this new apparatus were suggested by the concept that

an equilibrium system was being studied, and maintenance of
the gas pressure below the equilibrium value at the given
temperature should induce complete reaction and yield a
product free of reactant material. In addition, the pro-
cedure would be well—defined and reproducible (which synthesis
I was not). Analytical investigations on samples prepared

by synthesis I had indicated probable composition of LnOC.

Therefore samples of sesquioxide and graphite were mixed

according to the stoichiometry predicted by the equation

Ln203(s) + 3c(s)-+Ln20202(s) + CO(g). (26)

Procedures employed were similar to those of
synthesis I except that after the initial outgas of the
crucible and contents, the system was sealed and the
temperature was elevated to about l600°. The sample was
heated until the pressure had remained constant for at
least one hour. This pressure was assumed to define an
equilibrium state at that temperature. Carbon monoxide
was then slowly pumped from the system until the pressure
of carbon monoxide had fallen about 10 torr, after which

the system was again sealed and the pressure rise monitored.

—28—

This procedure was repeated until the pressure no longer
increased to the original value after removal of carbon
monoxide, and was indicative of a lack of equilibrium
conditions. The crucible was allowed to cool, the system
was first evacuated and then brought to atmOSpheric
pressure by the addition of helium, and the crucible and
its contents transferred to the glove box.

Samples prepared by this synthetic method ex—
hibited only extremely faint sesquioxide lines. On the
assumption that incomplete reaction was a result of inhibi-
tion of the reaction by slow diffusion across the graphite -
sesquioxide interface, the preparation was attempted at
increased temperatures but with little improvement. At
about l850°, a vigorous reaction which always occurred
between the sample and the crucible wall, formed a fused
gold—colored material and eroded (sometimes completely
through) the crucible walls. This erosion occurred
simultaneously with the fusion of the oxycarbide sample,
since portions of the product not in direct contact with
the wall were also fused. X-ray investigation of the fused
oxycarbide not in contact with the wall showed the com-
plete absence of sesquioxide x—ray diffraction lines and

suggested the use of some technique involving fusion.

_ 29 _

Availability of an arc melter (Fig. 1) made such a

synthesis feasible (synthesis III). Study of the fused,
gold-colored material formed by reaction of the sample

and the wall will be discussed in the context of equilibrium
studies (cf. pp. 57).

(f) Synthesis III

 

The are melt technique is in principle quite
simple. It consists of fusing a mixture of the desired
composition with a direct current arc, struck between
a negatively-charged, water-cooled copper hearth on which the
sample mixture sits and a water-cooled graphite or tungsten
electrode. As soon as the sample mixture began to fuse, its
conductivity increased sufficiently so that the arc could be
maintained between the electrode and the sample, thereby
producing the beneficial result of heating the sample to a
very high temperature without heating the copper hearth
excessively.

The electrode is attached to the arc melter
chamber by a flexible metal bellows, and if the pressure
differential between the chamber interior and the atmosphere
is more than about .1 atmOSphere, the electrode is either
jammed against the hearth or held a distance away from
it (more than 5 cm). Either way makes maintenance of
the prOper arc-length of 1 cm very difficult. For this
reason, internal gas pressures very near atmOSpheric had

to be used. In the arc melt preparations of Ln 0 C

2 2 the

2’

30

( water coolant
IL)

 

 

 

fi.JJnsulatinq material

L
metal bellows m
[viewing port
H

, j [. __.___l l__..

 

 

capper
container

water-cooled

 

 

 

 

 

 

 

 

 

 

 

probe
copper
”WT/hearth
(L4.
__ .1 ”__.—j

 

 

0
water coolant

vacuum and
gas-inlet system

ARC-MELTER ASSEMBLY

Figure 1.

- 31 _

chamber was filled to a pressure of one atmosphere with

carbon monoxide, thus maintaining both ease of electrode manip-
ulation and the caron monoxide pressure requirements of

the fused oxycarbide. The toxicity of this gas required

that the chamber be flushed (with either nitrogen or helium)
before opening, and that the vacuum pump be exhausted

into a hood.

The initial sample was prepared according to the
stoichiometry of equation (3), the sesquioxide and graphite
being pelletized as before and arc-melted. The x—ray
diffraction pattern of the preparations exhibited sesquioxide
lines which could be minimized (but not eliminated completely)
by diluting the carbon monoxide in the melter chamber with
successively larger fractions of helium. While the
sesquioxide x-ray diffraction lines were still visible
at low partial pressures of carbon monoxide, the Ln20202
x-ray diffraction lines disappeared completely and a set
of lines indicating cubic symmetry appeared together with
the sesquioxide pattern. The persistence of the ses—
quioxide phase was not surprising at the higher partial
pressures of carbon monoxide, since this behavior appeared
to be analogous to the equilibrium formation observed in
syntheses I and II, though the atmosphere of an arc can
hardly be considered an equilibrium system. The cubic

phase which appeared is considered in detail in chapter III

_ 32 _

of this thesis. The need for a preparatory procedure in
which no carbon monoxide is liberated, thus minimizing
equilibration, led to preparations based on the stoichio-

metries indicated by the equation

Ln(s) + Ln203(s) + 3C(s)'* 3/2 Ln202C2(s). (26)

Samples were prepared in either of two ways:
1. Metal filings were mixed with the sesquioxide
and graphite, then the mixture was pelletized.
2. The metal was used in Single (0.5 - 1.0 g)
pieces and melted together with pelletized
sesquioxide and graphite mixtures.
Samples prepared via alternative (2) were indistinguishable
both micrographically and by x-ray diffraction from
those prepared by number (1). Since metal filings were
assumed more susceptible to oxidation prior to use, and
were, in addition, more difficult to make from stock metal
ingot, the second alternative was preferred. 0n the basis
of phase purity studies (discussed in the following
section), this method was Judged to produce the best
samples of the Ln202C2 phase and was employed for preparation
of all samples used for further studies.

(g) Synthesis IV

 

Another preparatory scheme involved samples

mixed according to stoichiometries indicated by

-33-

Ln203(s) + LnC2(s) + C(s)‘* 3/2 Ln20202(s) (27)

and arc-melted under a one atmOSphere pressure of CO. This
starting mixture was difficult to handle, since the LnC2

and Ln had to be weighed, mixed and placed in the

203
pellet press in a glove box to prevent hydrolysis of the

dicarbide. The LnC was prepared by heating the lanthanon and

2
carbon in 1:2 mole ratios in a tantalum bomb at 17000 for
four hours. The bomb was Opened and the product stored in
the glove box. The pellet press was loaded in the glove box,
enclosed in several layers of thin plastic sheet, then was
removed from the glove box and the sample subjected to a
pressure of about 2500 psi. The pellet press was then
returned to the glove-box where the pressed pellet was
removed and sealed in a vial. The vial was fractured
immediately prior to evacuation of the arc-melter, and

the pellet of Ln203 and LnC2 were exposed to air for not

more than ten or fifteen seconds.

(h) Synthesis v

 

The final preparatory technique was inSpired
by questions concerning the possibility of metal carboyl
formation. The Ln metal was arc melted in a carbon
monoxide atmosphere to investigate this possibility. The
fused sample was fractured and re—melted several times to

promote homogeneity.

-311-

2. Phase Purity

 

Several experimental techniques are used to
obtain information concerning identity of phases present
in a particular material. Very few of these techniques
if used alone permit the complete characterization of
the phases. In the present work, it was necessary to
determine when a preparatory procedure produced a phase
that was monophasic within the limits required for
analysis. These limits require than an impurity, such as
graphite, which can be determined independently, is
acceptable only when present in amounts low enough that
indeterminancy in free—carbon determinations will not
affect the results of the other analyses. On the other
hand, the sesquioxide or the dicarbide cannot be
differentiated analytically from the phase to be determined,
thus they may be present only in amounts so small that
the error they introduce is less than the standard
deviation of the measurements. Thus most phase-purity
investigations were physical, since chemical separation
is excluded as the preceding discussion indicates.

X-ray diffraction (discussed in detail in the
analytical section) is a powerful qualitative tool for
phase determinations, but quantitatively the relative

pattern intensity of two phases cannot be related directly

_ 35 _

to the relative amounts of the phases present, since
diffraction line intensity is also related to other
factors, such as structure type, self-absorption, and
thermal motion. As long as a minor component is present
in an amount greater than 15%, the diffraction pattern

of that phase usually can be observed. Thus x-ray
diffraction is very useful only for initial phase analysis
within the stated limits.

Another useful tool, within limits, is micro-
graphy. As long as the phases are immiscible and are
or can be made differentiable by some visual method, the
phase purity can be investigated. If a solid solution
exists, its presence can be shown by variation of a lattice
parameter which is determined by x-ray diffraction.

In the preparatory investigations, x-ray diffraction
was utilized to enable gross purification of the oxycarbide
from the lanthanide dicarbide and lanthanide sesquioxide
phases. Further refinement of preparatory methods was
accomplished micrographically. The oxycarbide showed no

apparent shift in lattice parameter when the LnC or

2
Ln203 were shown to be present by x-ray diffraction: thus
it was judged that micrography could be used to show the
presence or absence of these phases if they could be

differentiated visually from the Ln202C2 phase. The LnC2

-35-

phase was identifiable by its gold color, in contrast

to the metallic grey of the oxycarbide. The sesquioxide
was identified by allowing the polished surface of the
sample to tarnish by exposure to air for a few seconds.

The color of the sesquioxide remained unchanged in contrast
to the colors of the dicarbide and oxycarbide which became
dull quickly.

Identification of these phases and visual
estimates of their apparent ratios in the samples as
determined by the area each covered in the micrographic
section allowed phase purity estimates to be made in the
various synthetic schemes. Such investigations were
also used to determine the effect various electrode
materials and 'time at are temperature' had on sample purity.

3. Analytical Procedures

 

Ternary phases such as Ln2O2C2 consisting of
a relatively heavy metal and two light non-metals pose
analytical problems of some magnitude, particularly when
one of the non-metals is oxygen. The problems are due

to:

l) cumulative errors inherent in any difference
determination,

2) difficulty in analyzing directly for oxygen
3) small errors in the relative percentages of the

nonmetals which produce large errors in the
stoichiometry deductions.

_ 37 _

Analysis of this particular phase was complicated
further by the rapidity with which it hydrolyzed. This
property required that all analytical samples be pre—
pared and weighed in the glove-box.

Fused ingots of the Ln20202 were fractured
initially with a hardened steel mortar which had a
retainer sleeve to prevent material loss. The sample
was then pulverized in a large Diamonite mortar. As
many as six 1 g samples were prepared at one time, and
then were taken into the glove box, where they were pul-
verized and mixed together intimately to produce one
homogeneous sample. Such a preparation furnished four
samples for metal analysis and at least three samples

for total carbon or oxygen analysis.

(1) Metal Analysis. In the glove box, four weighed

 

samples of 0.5 to 0.8 g were placed in 400 ml beakers
which were then covered with watch glasses to prevent
accidental contamination, and removed from the glove

box. Since these portions were to be hydrolyzed, no
attempt was made to exclude air. Hydrolysis was effected
by the slow addition of 125 m1 of 1N HCl; the samples
were digested on a hot plate for not less than four
hours, and allowed to cool. The contents of the

beakers were filtered through tared alundum filter

—38—

crucibles to remove free carbon; the crucibles were
dried subsequnetly at 100° for at least 6 hours, cooled
and the increase in mass considered as free carbon. The
crucibles were then fired at around 900° for a minimum
of 4 hours to oxidize the filtered carbon, cooled,and
weighed. The weight loss, interpreted as free carbon,
provided a check on the previous determination. Dilute
ammonium hydroxide was added to the filtered solutions
until the methyl orange end point was reached (usually
small amounts of hydroxide precipitate could be observed
at this point). Twenty—five ml of concentrated oxalic
acid was added slowly with stirring; the samples were
covered and digested on a hot plate for at least six
hours. Completeness of precipitation was determined by
adding an additional lml of oxalic acid to the cooled
supernatant solutions. If no precipitate formed (normally
the case), the solutions were filtered through tared
alundum filter crucibles, which were first dried, then
heated slowly by a Meker burner to prevent too rapid
evolution of carbon dioxide with attendant material
loss. The crucibles were finally ignited to 900° for

at least twelve hours, to convert the oxalate to the
sesquioxide. The crucibles were cooled in a desiccator

for two hours, then weighed.

_ 39 _

(ii) Total Carbon Analysis. Total carbon content

 

of the samples was determined by a combustion technique,

by use of the apparatus illustrated in Fig. 2. Oxygen (USP)
was admitted to the system at pressures of less than one
psi and at flow rates of 6-8 ml/minute. Any carbon-
containing impurities in the oxygen were oxidized over a
CuO-CeO2 catalyst maintained at about 750° in furnace (A),
and water or carbon dioxide formed by this pre-oxidation
were absorbed in U-tubes (B) and (C) containing respectively
Mg(ClOu)2 and Ascarite. A sample to be analyzed was placed
in furnace (D) and heated in the stream of pure, dry

oxygen. Any carbon in the sample was oxidized to carbon
monoxide, carbon dioxide or a mixture of the two gases, and

then swept over a CuO—CeO catalyst maintained at 750° in

2
furnace (E), where any carbon monoxide gas present was
oxidized to the dioxide. The oxygen-carbon dioxide gas
mixture was dried in the H280Al column (F) and the Mg(ClOu)2
filled U—tube (G), and then swept through the tared Ascarite
filled U-tube (H), in which the carbon dioxide was absorbed.
This U—tube was then weighed, and the total carbon content
in the sample was determined.

In a typical eXperiment a 0.6 - 0.8 g portion

of the original homogeneous sample was placed in the

4O

msumucmad cowumsnfiou .N mmDlo

meHm<om¢ eoHomz Naeoflovmz usamaomc zmuwxo

/\. /\ 1

 

 

mmvflzmbm NmDB

 

 

 

 

 

 

 

 

 

TL. .

NU4ZMDh Ema?

 

 

mmqmmsm comm:

mOBflqbumm

EMBmNm mHmMA¢Z< zommdo

 

-111-

approximate center of the Vycor tube in furnace (D),

which was initially at room temperature (in contrast

to furnaces (A) and (E), which were maintained at about
750°). The system was closed, and the oxygen flow was
initiated and maintained at about 6 ml/min. The sample
temperature was elevated over a 5—hour period to just over
lOOO°, then maintained at this level until the absorber
weight as indicated by hourly weighings was constant (or
the weight gain had decreased to the predetermined "blank
rate"). The accuracy of this analytical procedure and

the "blank rate" were determined using out-gassed graphite
and calcium carbonate as known carbon sources. The
absorber connections were arranged so that the inlet and
outlet tubes could be connected when the absorber was
weighed, thus preventing the absorption of atmospheric
moisture by the ascarite. This arrangement also allowed
rapid closing of the system. The absorber was always
attached and detached as quickly as possible, and was handled
with plastic—coated cotton gloves to minimize weight gains
due to fingerprints.

(iii) Hydrolysis Experiments. A sample of the oxy-

 

carbide was hydrolyzed by water vapor and the gases so
produced were collected. These gases were analyzed using
an F & M Scientific Corp. flame-ionization-type gas
chromatograph (Model 615) which had been standardized with

ethane and acetylene.

-42..

(iv) X-Ray Investigations: The oxycarbide phase was

 

examined both by Debye-Scherrer cameras and recording
diffractometers. In each case, sample preparation
stressed exclusion of atmospheric moisture to prevent
hydrolysis. Capillaries 0.3 mm in diameter were loaded
in the glove box, placed in sealed vials and flame sealed
immediately upon removal from the vials. Debye-Scherrer
cameras of 114.59 mm diameter were used, with copper Ka
radiation (Nan = 1.5418 A).

Samples were prepared in the glove box for dif-
fractometry by pressing a powdered sample into a rectangular
2.5 x 0.5 cm cavity machined 2 mm deep into a brass sample
holder. The sample was smoothed and covered with a strip
of Scotch "magic mender" tape. The tape was shown to
prevent noticeable hydrolysis for up to two hours. For
longer diffractometer runs, a thin coating of paraffin
oil was flowed over the tape. This procedure prevented
noticeable hydrolysis for up to eight hours. A Siemens
diffractometer was equipped with a scintillation
detector whose output was monitored by a pulse height
analyzer in an attempt to reduce the fluorescence back—

ground of the sample.

—43—

(v) Infrared Studies - An infrared spectrum was

 

obtained on a Unicam SP 200 spectrometer using the KBr
pellet technique. The KBr press was loaded in the glove
box, removed and the sample pressed under vacuum. The
disc was subjected immediately to infrared analysis.
Since atmosphefib moisture could not be excluded from the
sample compartment of the spectrometer, slight hydrolysis
occurred on the disc surface.

(vi) Equilibrium Studies — The system used for equi—

 

librium studies is described fully in Appendix (11).
Briefly, it consisted of a 100 mm diameter Pyrex tube

which confined a water-cooled current concentrator sus-
pended from a brass, water—cooled lid which contained an
optical window that could be magnetically shuttered from
exposure to the hot crucible. The crucible was supported in
the center of the copper donut on 1.52 mm diameter tungsten
legs about 2 cm long. The pressure measuring system con-
sisted of the Wallace and Tiernan dial manometer described
previously and an absolute pressure U-shaped mercury
manometer whose column height difference was determined
with a Gaertner Scientific Co. Model M-908 cathetometer.
The mercury manometer was employed to check the dial

manometer at low pressures. Compressed gas from a tank was

-1111-

admitted to the system through a Mg(0104)2 drying column
and a flow-meter. This gas-handling system allowed very
precise monitoring and control of gas pressures.

Temperature control was achieved for the
La202C2 study by use of a two-color pyrometer (Latronics
Corp., Latrobe, Pa.). This pyrometer monitored the total
emission of the crucible lid through the Pyrex tube at
an oblique angle so that temperature measurements of the
crucible cavity could be made without interrupting the
control achieved by the pyrometer. The output of the
two-color pyrometer was led into a Model 90 Leeds and
Northrup automatic controller, in which the pyrometer
signal was balanced against a Zener diode. Any excursions
from the null point due to temperature variation generated
a signal which altered the furnace output so as to reverse
the variation. The two-color pyrometer had internal
calibration checks on both sensing circuits. The cali-
bration was checked regularly; every thirty minutes at the
beginning of an equilibrium experiment and less frequently
thereafter as the circuitry stabilized.

This temperature control unit was not available
0 C

2 2 2
system. The control unit accounts for the improvement in

when equilibrium measurements were made on the Nd

the scatter of the La202C2 data, since previously it was

-115-

difficult to maintain the temperature at a constant value
for several hours.

A finely pulverized Ln 0 C sample (1.4 - 2g)

2 2 2

was inserted into a graphite crucible in the glove box,
then the crucible was placed in the equilibrium system
which was immediately evacuated. The crucible was sub—
jected to a final outgas by heating in vacuum at just under
1000° for approximately two hours, after which the equili-
brium system was valved from the pumping system. Sufficient
carbon monoxide (Matheson, CP) was added to the equilibrium
system to achieve a pressure of 70 — 100 torr, and the
crucible was then heated to some pre—determined temperature
above 1350°. When the carbon monoxide pressure in the
system had stabilized, portions of carbon monoxide were
removed or added, thereby reducing or increasing the

carbon monoxide pressure by 3-10 torr increments, and the
system was then allowed to re-stabilize. During this
entire procedure, the temperature was maintained at a
constant value. When a given pressure was re-established
to within iO.3 torr after both high and low pressure excur-
sions at a constant temperature, that particular pressure
and the corresponding temperature were assumed to define

a reversible state. Small gas samples were removed

—46-

periodically from the equilibrium system by eXpanding

a small amount of the gas into an evacuated 1 liter flask
via a manifold on the gas inlet line. This gas sample,
which was in equilibrium with the crucible contents, was
analyzed mass spectrometrically. If the gas sample proved
to be carbon monoxide, the reversible state demonstrated
at the particular temperature and pressure was assumed

to be an equilibrium state.

This procedure was repeated for a series of
temperatures, taken in a generally random order. Seven
to twelve pressure—temperature pairs defining equilibrium
states were recorded for each sample, and at least three
different samples were subjected to equilibration study
for both La20202 and Nd202C2.

Small portions of the crucible contents were
removed periodically after equilibrium conditions had been
exhibited to determine the solid phases which exist in
equilibrium with carbon monoxide gas. At gas pressures
exceeding approximately 130 torr, pressure fluctuations
were detected. By heating an empty crucible in similar
helium pressures, these fluctuations were shown to exist
independently of the chemical system being studied. The
fluctuations were damped by partially closing the needle

valve between the manometers and the equilibrium system.

-117-

D. RESULTS OF Ln202C2 INVESTIGATIONS

l. Preparative Results

 

The original preparatory technique developed
by G. L. Buchel and modified by the writer was un-
satisfactory for the following three reasons.

1. Because of the heterogeneous solid - solid
reaction involved, the sample couldn't be
prepared completely free of sesquioxide.

2. The technique involved trial and error methods
which couldn't be standardized or repeated to
prepare uniform samples.

3. Micrographic analysis was impossible without a
fused sample.

The arc melt technique proved to be a great
improvement over previous methods. It produced samples
which were free of sesquioxide x-ray diffraction lines,
it was reproducible, and up to eight grams of sample
showing good uniformity in x-ray and micrographic analysis
could be prepared rapidly. 0f the various preparations
used, the one mixed according to stoichiometries indicated

by the equation

Ln(s) + Ln203(s) + 3C(s)"> 3/2 Ln20202(s) (28)

—48-

gave a product which showed greatest phase purity by both
x-ray and micrographic analysis, and which also required
less exacting preliminary care since the starting mixture
could be handled with reasonable precautions in air. Slight
deposits of sesquioxide on the interior of the arc-melter
chamber indicated that metal and/or metal monoxide gas were
being lost preferentially by the sample during melting.
Micrographic examination showed small amounts of a black
phase (later shown to be graphite) distributed through the
oxycarbide. The amount of graphite could be minimized by
arc-melting the sample only long enough to produce
homogeneous specimens. Samples arc—melted for about 35
seconds, which were turned over and remelted for another

35 seconds showed minimal amounts of graphite. Graphite,
however, interfered with neither the analytical results,
since bound and unbound carbon were determined independently,
nor with equilibrium results, Since graphite was one of

the solid components in the equilibrium mixture.

The sample mixed according to the equation

Ln203(s) + LnC2(s) + C(s)‘* 3/2 Ln202C2(s) (29)

was also arc-melted according to the above procedures. This
method produced the oxycarbide but the product showed no

improvement in purity over the method previously described

_ Mg _

and the technique was discarded because of the already
mentioned difficulties in preparing the sample mixture.
That the Ln2O2C2 has narrow composition limits was shown
by the position invariance of the various back—reflection
lines in the x—ray diffraction patterns shown by samples
containing excess metal, oxide or graphite.

Arc-melting the metal in carbon monoxide produced
the Ln202C2 phase, though it was demonstrated micrographically
that the fused ingot had to be fractured and re—arced
several times before it showed homogeneity. The carbon
deposit on the hearth around the samples indicated that
the reaction may have involved carbon and oxygen species

formed from carbon monoxide by the very high arc temperatures.

2. Analytical Results

 

Tne results of Nd20202 analyses performed by
R. B. Leonard are reported here.22 Vacuum—fusion analysis
indicated that neither nitrogen nor hydrogen was present
in the sample, since all the gas evolved in the vacuum
fusion was found to be carbon monoxide. Similarly the

gases liberated when sesquioxide reacted with graphite

were found by use of the mass spectrometer to consist of greater

_ 5o -

than 99% carbon monoxide, with 0.3% H2, 0.1% N2 and
traces of other Species.

Mass spectrometric analyses of the gas samples
removed from the system during equilibration studies of
La20

2C2 indicate the following:

CO. 97.6 i 2%

H 1.02 i 0.5%

0.7%

2,
O

|+

2, 1.38

These data indicate that the reversible pressures measured
during equilibration experiments were within 2% of the
true carbon monoxide pressure, and the pressure and
temperature data recorded did indeed represent a series
of equilibrium states.

The percentages determined from 20 analyses
performed on seven different neodymium oxycarbide

preparations are as follows:

Nd, 82.9 i 1.0%
C, 6.3 i 0.9%
O, 10.8 i 1.1 (difference)%
8.43 i 0.21 (fusion)%
The carbon content was considered a minimum value since
the sample was always exposed very briefly to air. These
data, when reduced to molar ratios, determine a stoichi-

ometry of

- 51 -

Nd O

1.00 i .01 1.04 i 0.12 C

O.91 i 0.13,

based on the average of both fusion and difference oxygen
analyses. These data are judged reasonably consistent
with the composition NdOC.

The percentages determined from 12 metal and
free carbon analyses and six total carbon analyses on the

analogous lanthanum species are:

La, 83.4 i 0.6%

C. 7.5 i 0.9%
0, 9.1 i 0.7% (difference).

The improved standard deviations probably result from
refinement of the preparatory technique by use of the
arc melter. Reduced to molar ratios, these data determine

a stoichiometry of

O

La1.00 i .01 0.95 i .06

C1.04 i 0.12.

These data are again judged reasonably consistent with

the stoichiometry LaOC.

3. Hydrolysis Results

 

The graphic display of the chromatagraphic
analysis of the gaseous hydrolysis products of the

neodymium species indicated acetylene, ethene and ethane

_ 52 _

to be present. The relative peak heights (acetylene:
ethene: ethane) were 945:8:5. The gaseous hydrolysis
products of LaOC consisted of 96.3% acetylene, 4.1%
mixed ethane-type hydrocarbons and no methane.

The LnOC phases hydrolyze very rapidly in
the presence of even small amounts of water vapor, thus
all manipulations had to be carried out to exclude as
much atmospheric moisture as possible. Samples could
not be kept uncovered in the glove box for more than
one day without noticeable decomposition. Whenever
possible, they were stored in a vacuum desiccator or in
small paraffin—sealed vials which were kept in the
glove box prior to use.

The predominance of acetylene in the gaseous
hydrolysis products and the rapid hydrolytic behavior
indicate that the carbon is present in the oxycarbide as

CE ions. Based on this, the formula should be written

Ln202C2 rather than LnOC.

4. X—Ray Diffraction Results

 

The principal interplanar g-values and the
relative intensities of their corresponding lines in the

x-ray diffraction patterns are listed for Nd202C2 and

La202C2 in tables (I) and (II) respectively. The Nd202C2

ggvalues were determined from two independently prepared

samples, and the La2020 devalues are representative of four

independently-prepared samples, two of which were examined

_ 53 _

TABLE I

OBSERVED Nd2O2C2 X—RAY POWDER DIFFRACTION DATA

1 933 _I_ 2,3
w 7.04 w 1.655
m 3.55 vw 1.623
w,m 3.42 vvw 1.607
VW 3.36 VW 1.586
VS 3.11 w 1.554
s 2.968 w 1.540
W 2.6OO vw 1.496
m 2.370 vvw 1.467
m 1.989 vvw 1.452
m 1.968 vw 1.425
w 1.926 W 1.342
vw 1.890 VVW 1.322
w 1.859 VVW 1.305
vvw 1.793 vw 1.295
vvw 1.744 vw 1.282
vw 1.701 vvw 1.268

vvw 1.680 m 1.257

OBSERVED

Il-l

VVW

mw

VW

VW

VVS

ms

VW

VW

VW

VW

VVW

VVW

Ia202C

ICL
\o
:1>O

.16

.65

.53

.479
.356
.172
.042
.958
.613
.446

NNWUUUUUUUJUUUWN

.412

.269

[UNION

.025

.992
.962

|—’

H

[__J

.908

2

_ 54 _

TABLE II

X-RAY POWDER DIFFRACTION DATA

Il—l

VVW

VW

VW

VW

VW

VW

VW

VW

VW

VW

VVW

VW

VVW

d, A
1.834
1.748
1.705
1.669
1.651
1.638
1.620
1.594
1.577
1.533
1.491
1.455
1.403
1.378
1.362
1.286
1.258

VW

mw

VW

VVW

mw

VVW

VVW

9A

1.233
1.209
1.184
1.156
1.089
1.066
1.014

0.859

_ 55 _

by the Debye-Scherrer technique and the other two by
diffractometry. The precision of these gfvalues is estimated
to be i 0.01 A, A tentative indexing based on Weissenberg
photographs of a poor-quality Single crystal of La 0 C

2 2 2
indicates monoclinic or lower symmetry. (Appendix III)

5. Infrared Results

 

The infrared spectrum was virtually featureless,
with no peaks in the carbonyl region (2000 cm-1) or in
the acetylide region (2200 cm-l). One very faint peak
which was observed in the 700 cm.1 region could be
attributed to a EC—H rocking mode from the acetylene which
was present in small quantities since the KBr sample disc
was exposed to atmospheric moisture during the infrared

scan.

6. Equilibrium Results

 

Three or more sets of equilibrium data were
taken, with a minimum of eight pressure and temperature—
defined equilibrium points in each data set. X-ray
diffraction analysis of the solid phases involved in the
equilibrium and mass spectrometric analysis of the gas
phase indicated the equation describing the equilibrium
to be

Ln203(s) + 3C(s) :Ln202C2(s) + C0(g). (30)
This equilibrium could not be reversed completely to the

left when the starting material was arc—melt prepared

Ln202C2(s) nor could it be forced completely to the right

_ 56 _

once the equilibrium was established. These observations are
apparently a result of a rapid decrease in reaction rate
after a certain fraction of the material has reacted.
This initial reaction probably formed a protective layer
through which the carbon monoxide was absorbed or evolved
only very slowly. The fraction of material which had reacted
could be monitored by measuring pressure changes resulting
from the uptake or evolution of carbon monoxide since the
volume of the system had been calibrated previously.
The effective temperature value used in the gas equation
to calculate the amount of gas absorbed or evolved was
determined by adding a known amount of carbon monoxide
at room temperature to the system containing an empty
graphite crucible. The crucible was heated and the
pressure of the known amount of gas was measured as a
function of crucible temperature. From this the "effective
gas temperature" was determined over a range of crucible
temperatures. It was found that this "effective gas
temperature" changed very little for a given crucible
temperature when varying amounts of gas were present in
the system.

The equilibrium data were taken over a temperature
range of about 1325° — l850°. Below l325° the reaction
preceded at so slow a rate that attainment of equilibrium

was difficult, and at about l850° the powdered solid

_ 57 -

mixture began to fuse, producing a fused layer on the

sample that apparently reduced the surface area enough

to drastically limit the absorption of gas. Above l850°,
the sample would completely fuse and react with the crucible
with an accompanying drastic increase in carbon monoxide
pressure. The fused material, which appeared to erode
through the crucible, exhibited a lanthanide dicarbide
diffraction pattern and was probably a eutectic of an2
and graphite. Although G. L. Buchel in his thesis7
reproted measurements of pressures over predominantly
oxide samples at temperatures in excess of 2000°, the

0 C

2 2 2’
and it appears that another equilibrium system was being

writer was unable to repeat this work using La

studied above l850°; since a change in slope occurs in
the graph representing G. L. Buchel's equilibrium data
at this temperature.

After a prolonged (more than 48 hours) equilib-
ration study, small amounts of sesquioxide were always
found on the graphite crucible wall exterior, and the
graphite wall was pitted under sesquioxide. It appears
that the lanthanide metal migrated through the graphite,
possibly in the form of an oxycarbide. When the migrating
phase reached the exterior wall surface (which was always
a little cooler than the interior of the crucible), the

carbon monoxide pressure exceeded the equilibrium value

- 58 -

for the phase and converted that phase to the sesquioxide
and graphite. The metal was probably not present as a
dicarbide phase during transport through the wall

since attempts to react LnC with carbon monoxide to form

2
either the Ln202C2 or Ln203 were unsuccessful. Only very
small amounts of carbon monoxide were absorbed in such
experiments, since a fused layer was always formed on the
dicarbide when it was heated to temperatures at which

the small amounts of CO were absorbed. The absorbed layer
inhibited further reaction.

To prevent this exterior wall phenomenon from
competing for the available carbon monoxide, a double wall
crucible (Fig. 3) which eliminated the problem was devised,
presumably since the crucible interior was maintained at

a constant temperature over its entire surface.

Graphs of the equilibrium data for the reactions

Nd203(s) + 3C(s) 1: Nd20202(s) + C0(g) (31)

La203(s) + 3C(s) S La20202(s) + CO(g) (32)

are displayed in Figs. (4) and (5). The decreased scatter

in the equilibrium data for La20202 is attributed to the

improved temperature control described previously

59

AXIAL ORIFICE

/[

VK\\\\\ \ \

 

PRESS FIT

d— OUTER CRUCIBLE

 

 

E\\\\\\\

\

INNER CRUC IBLE

 

 

 

 

 

FIGURF 3. Double-Walled Graphite Crucible

60

 

 

 

 

 

\ ’I°.\<
o £31:\<: ‘-
‘u ‘\g
5' \‘Q. o A
0"”.
I I O
9\‘ ‘
10 1
\° '
A
.-. \ne
1
' -\‘A ‘. - 5°
CW 1:
[\u \9
Z 8 . \‘K:
O
._ \..
, i
- I
6
r RUN
a 1
I 2
4;} 2
4 D
o 2
O 7
a 8
4.6 5.0 s. ‘ 5. 8

FIGURE 4. Graph of 2 vs. 104/T for Nd 02c2 System tbllck).
Overlay Shows Same Data Ploéted as log P
in Equilibrium with Nd 20(5), C(s), and N80
Versus 10

292C2(3)

 

12

10

60

 

 

 

l .-
‘B\!§kr
A
°\‘ .
. X.
' a
be
D ‘ .
s°
"’x
10
. \k
\.a
l.
A 1
I .2
O 3
. 3} 4
O
o 2 \
Cl 7
U 8
(.6 $6— I 534 ° 33—“
FIGURE 4. Graph of t 25. lO‘/T for Nd 0202 System (black)

Overlay Shows Same Data Plo ted as log P (atm.)
in Equilibrium with Nd 28§8)' C(s), and N8202C2(s)
Versus 10 4/?

 

61

 

 

 

\ ' l ‘ l ' l
03
O
0.8 - \o
_. It?
1'21“" w
9
1.4 — o
o 1.6 '-
U
D:
U‘
0
A
| 1.8 _.
0
2.0 ..
2.2 -
2.4 _. O RUN A G ‘
Q RUN s
0
o RUN c
2.6 '— O 0.1
I I 1 I l 1 1
5.0 5.4 5.8 6.2
1/T x 104(°K)
FIGURE 5.

Pressure of C0 in Equilibrium with La202C2(s),
C(s), and La203(s)

 

-52-

The graph of equilibrium data measured for the
Nd20202 system appeared to possess a slight negative
curvature and hence was subjected to a so-called Z—plot
treatment, in which the ACP for the oxycarbide was assumed
constant over the temperature range of interest. The
heat capacity of the oxycarbide was assumed to be that
of the sesquioxide plus 7.5 cal/deg.—mole, the latter
figure being three fourths of the extrapolated heat

capacity differences between Ca0 and CaC at 1750°. From

2
a combination of this value with the heat capacities

for Nd203, C and C0, ACP

deg-mole. At a reference temperature of l750° with a

was estimated to be —l.7 cal/

constant AC the function 2, defined as Z = R tn P

P’ CO
-ACP(tnT + 1750/T), was graphed against lOu/T (Fig. 4).
Data series 1 - 3 were made in the original apparatus
(cf, page 25) with a starting mixture of Nd203 and
graphite, while series 4 — 8 were made in the second
apparatus (cf. App. II) with starting mixtures of
Nd202C2 and sesquioxide, or only Nd20202. A different
Specimen was used in each series. The arc-melted sample
of Nd202C2 absorbed and evolved carbon monoxide quite
.rapidly as long as not more than 15—20 mole percent of the

cfuarge had reacted with carbon monoxide. When more than

2%) rnole percent of the charge had reacted, the retardation

04? tflne pressure restoration rate after a pressure excursion

—63—

from the equilibrium value was probably due to the
necessity for the CO to diffuse through a thick layer of
solid material to reach the reaction interface.

The linear least-squares equation which describes
the equilibrium data for the Nd O C system in the

2 2 2
temperature region 1350 - 1850° is

 

2: 472.1193) x 103 + (46.60 i 0.68) (33)

From this equation, the following thermodynamic data

are calculated for reaction (31):

A

H-

l kcal/

H°1750°K = 72'? '3 sfw:

H
.1:

AS°17506K = 32.2 .3 eu.

The standard deviation reported for the enthalpy represents
only the scatter in the data; that reported for the

+
entropy includes an estimated error of _ 0.5 cal/deg-mole

in the heat capacity.
These data can be converted to those which
represent the equilibrium system at 298° by making use of

the tabulated enthalpy and entropy data for the ses—
23,24

quioxide, graphite, and carbon monoxide, and by

estdheting the enthalpy of the Nd O C

2 2 2 to equal that

- 64 _

of the sesquioxide plus three-fourths the difference
between the enthalpy values for CaC2 and CaO. From

these calculations,

AH6298 (second law) = 75,9 i 3.0 kcal/gfw

AS°298 = 33.3 i 4.“ cu.

The indicated errors are estimated from those listed
previously for the l750° data, since the heat capacity
correction for the oxycarbide is an estimate and thus
will increase the error in the 298° K data.

The graphic display of equilibrium data des-
cribing the La2O2C2 system appeared to have no discernible
curvature and its treatment by the Z-plot method, using
approximations similar to those described for the Nd202C2
data reduction, produced changes in the linear equation
which were within the standard deviations computed

assuming AC = zero over the temperature range. Thus

P
the Z-plot treatment was not applied. The set of measured
data points is described by the linear least squares

equation

 

3
tn P = ‘(79'6i1golx 10 + (36.6 i 3.5), (34)

CO

whose constants represent the equilibrium state of 1809°K,

the mean temperature.

_ 65 _

From this equation the following thermodynamic

data may be deduced:

H

79.6 1. kcal/gfw

”H 1809 = 0

ASOlBOg = 36.6 i 3.5 eu.

Heat capacity corrections similar to those applied to the

previous data were used

to correct these values to 298°K.

The corrected data are:
° = +
AH 298 82.2 _ 1.9 kcal/gfw
° = +
AS 298 37.7 _ 3.6 eu.

These data show increased
due to approximations used in

O C

errors over the 1809°K data

determining the heat

capacity of La2 2 2(s), and the reported errors in the

tabulated heat capacities used for the other components

of the equilibrium.
As a check on these
ations, third—law values were

cited previously for the free

and La 0 C

2 2 2. The free energy

second—law enthalpy determin-
calculated using tabulated data
20202

function of the oxycarbides

energy functions of Nd

were assumed to be twice the sum of the free energy

functions of the lanthanide metal, graphite,

and an

estimated free energy function of oxygen, whose contribution

was assumed to be the difference between the free energy

functions of CaO and Ca over the temperature range.

- 66 -

The third-law enthalpies show no significant
trends over the temperature range studied, and their mean

values, together with their standard deviations are:

H

AH298 (Nd20202, 3rd law)

84.46 i 0.63 kcal/gfw

AH298 (La 0 C 3rd law)

2 2 2, 80.8 i 0.71 kcal/gfw

9

The listed errors represent standard deviations of the
values rather than actual errors in the enthalpies. The
agreement of the second and third—law data will be discussed
subsequently. In further calculations the second—law value
will be assumed correct because of the approximations used
in the third-law determinations.

Combination of these second-law values with the
enthalpies and free energies of formation of Nd203(s),
La2O3(s),23 and CO(g),2br permits calculation of the follow-

ing enthalpies of formation of Nd O C (s) and La 0 C (s)

2 2 2 2 2 2
at 298°F:
AHf (Nd202C2(s)) = 329 i 3 Kcal/gfw

AH (La202C2(s)) = 320 i 2 kcal/gfw

and similarly the free energies of formation at 298°K

are:

_ 57 _

f (Nd20202(s)) = 309.9 i 3.9 kcal/gfw

AG° (La202C2(s)) 315.6 i 2.9 kcal/gfw.

The reported deviations represent the accumulated errors
resulting from the arithmetic manipulations on the standard

deviations of both the measured and published data.

-68-

E. Ln 02C

2 PHASE, CONCLUSIONS AND DISCUSSION

2

1. General Discussion

 

a. Stoichiometry and Bonding

 

The stoichiometry of this phase, determined to
be Ln20202, is of interest not only because of its
support of the model proposed in the introduction, but also
because of its implications.
First, the stoichiometry does in fact appear
to support the exchange of an oxide ion for an acetylide

ion, and can be considered the result of the operational

process

Ln2O3(s) + CZ'* Ln202C2(s) + O_. (35)

Second, the stoichiometry implies the formal
oxidation state of the lanthanide metal is tripositive.
This implication is somewhat unexpected as acetylide and
methanide ions have normally been seen to stabilize low
formal oxidation states for the lanthanides by facilitating
the formation of conduction bands which accept valence
electrons from the tripositive lanthanide ions.25 That
this apparently does not happen may be due to one or both

of the following postulates:

_ 69 -

(a) the acetylide ions are "diluted" by the oxide
ions, and their effect is thus nullified.

(b) the symmetry of the acetylide sites is such
that whatever combination or overlap of ionic orbitals
is needed to form conduction bands is effectively pre-
vented.

The first postulate is thought the less probable,
since the phases discussed in the following chapters
possess O/C ratios of three to one, yet are excellent
conductors. A strict comparison is not valid, however,
because of the different nature of the carbon units.

Conductivity measurements were not performed on
this phase, since small amounts of graphite were always
present in the preparations. The appearance of small
tranSparent crystals of La2O2C2, used in the Weissenberg
technique, does support the non—existence of conduction
bands in these phases, as metallic conductors normally
exhibit Opaque crystals with varying degrees of metallic
color (description of Weissenberg experiments in Appendix III).

The hydrolysis behavior reveals much about the
bonding and the ionic species in the crystal. The ap-
pearance of no hydrolysis products other than acetylene

is indicative of two facts:

_ 7o _

1. All bound carbon present in the Ln20202 phase is in
the form of acetylide ions, as any methanide units
would produce appreciable quantities of methane.

This information indicated the correct formula to
be Ln202C2 rather than the empirical LnOC formula,
also consistent with the analytical results.

2. The complete absence of higher molecular weight
hydrocarbons indicates that no oxidation or re-
duction occurred during the hydrolysis. If any of
the metal were in the Ln+2 oxidation state, it would
reduce some of the acetylene, and would form higher
molecular weight hydrocarbons.26 The metal is thus
deduced to be in the tripositive state in the crystal.

The rapidity of the hydrolysis has been con-
sidered to be an indication of the ionic character of
the metal-carbide bond.27 By this criterion, the Ln—C2
bond is very ionic, since the hydrolysis of Ln20202
occurred with extreme rapidity in even very small amounts
of atmospheric moisture.

From the preceding arguments, it is apparent
that the Ln202C2 phase is a crystalline material of
highly ionic character, consisting of tripositive

lanthanide cations, oxide and acetylide anions. The

phase can be considered as a lanthanide sesquioxide

- 71 _

lattice with one-third of the oxide ions replaced by
acetylide ions.

b. Structure

 

If one were able to replace (as proposed in the
preceding model) one-third of the spherical oxide ions
in a sesquioxide lattice by acetylide ions, the resultant
species would undoubtedly be a distortion of the parent
oxide lattice, since the CE ion is markedly anisotropic.
Though the average C-C bond distance in the lanthanide

dicarbides is 1.285 A28 and the radius of the oxide

anion in metal oxides is of the order of 1.3 R, the actual
space filled by the acetylide ion is evidently much
greater. This can be demonstrated by comparing the
lattice parameters of the cubic forms of CaO and CaCE,
which are 4.799 E and 5.393 K respectively.29 The
difference is about 0.6 X. This comparison, however,
assumes the two sites to be equivalently spherical. In
addition to this size difference, the oxide ion is
spherical while the acetylide ion possesses a prolate
shape. A determination of the relative anisotropy of the
acetylide ion can be made by comparing the lattice
parameters (a = 5.49 K, c = 6.38 Z)29 of the tetragonal

CaC unit cell. The difference in length is due only to

2

_ 72 _

the orientation of the acetylide unit, therefore, one
can conclude that the space occupied by the anistropic
CZ unit is about 0.5 X greater in one dimension than the
other. The CaC2 was selected for this comparison rather
than some lanthanide dicarbide since the calcium undergoes
no apparent formal oxidation state change upon 0: hn-O=
substitution and is comparable in size/charge ratio to
La+3. It may be that the CE unit changes in dimensions
with conduction band formation in the LnC2 structures,
since the c-a difference for their unit cell dimensions
is of the order of 2.3 3.29

The combination of the size increase of the
acetylide ion over that of the oxide and the prolate
shape of the acetylide unit probably render impossible
any estimates concerning the effect of the model (acetylide-
for—oxide) substitution on the oxide structure, but
certainly would lead one to suspect that the structure
possesses fairly low symmetry compared to the sesquioxide
.or dicarbide. The probable monoclinic cell indexing
presented in Appendix III is consistent with these
observations.

An oxycarbide phase purported to be LnOC
(Ln = La, Ce, Nd, Sm, Gd) has been indexed on the basis

of a monoclinic cell by Leprince—Ringuet.30 Comparison

_ 73 _

of the gfvalues expected from their reported unit cell

for LaOC with the measured gfvalues for the lanthanum

phase reported in this work shows that the La2O2C2 is

probably indexed incorrectly, since too many diffraction
lines corresponding to devalues predicted by their unit
cell are not found in the La202C2 diffraction pattern.
Other comparisons with the French work are of
interest. They report the isostructural analogues of

this LnOC phase for samarium, cerium, and gadolinium, in

addition to neodymium and lanthanum. Attempts in this

laboratory to prepare the samarium and gadolinium
analogues of La20202 and Nd2O2C2 failed. The cerium—
oxygen-carbon system might be of interest, since one would
anticipate the acetylide-oxygen substitution would take
place in the cerium dioxide, rather than the sesquioxide
system, though the acetylide units might stabilize the
lower tripositive oxidation state of cerium.

The inability to form the Sm202C2 and Gd202C2
could possibly be explained in two ways:

1. The ionic radius change in the lanthanide metal between
neodymium and samarium might be sufficient to destabi-
lize the Ln2O2C2 phase which exists for the lighter
lanthanides. Surely this is the principal difference

between neodymium and samarium compounds with oxygen

and carbon.

-74-

2. Samarium shows some stability in the dipositive state,
and the carbon might be able to promote conduction-
band formation, and consequent phase change from the
neodymium to the samarium analogue.

The apparent invariance of any lattice para-
meters in the Ln20202 phase when prepared with an excess
of any component is indicative of narrow composition
limits. It appears that the C/O ratio is relatively
fixed, a behavior quite different from other oxycarbide

10,14 and zp(o,c),31 which show variable

phases U(O,C)
O/C ratios. Ternary phases of Ln-O-C are also quite
different from the Ln-O-N ternary phases which appear to
be solid solutions of nitride and oxide of the lanthanide
used and possess widely variable N/O ratios.32’33 This
seeming invariance of the C/O ratios in Ln20202 might be
accounted for on the basis of the anisotropic shape of the
acetylide ion. Apparently the acetylide ions occupy ordered
lattice sites (not equivalent to the oxygen sites), and

any change in the C/O ratio outside of very narrow limits
varies this anionic ordering and destabilizes the lattice.

2. Relationship of La and Nd Analogues of Ln 0 C

2 2 2

a. Structure

 

Comparison of the x-ray diffraction patterns

of La 02C

2 and Nd

2 202C2 showed that the lines of the

_ 75 -

two patterns corresponded on a one to one basis, with
slightly smaller d-values being calculated for the

Nd202C2 than La202C2. From this, it can be concluded

that the phases are indeed isostructural, and the slightly
smaller despacings for any given set of planes in the
Nd202C2 than the La2O2C2 are consistent with the 0.066 334
smaller ionic radius of Nd+3. The observed appearance

of a new phase on the attempted preparation of the

Sm202C2 and Gd2O2C2 has already been noted. The
diffraction patterns for the Sm(OC) and Gd(OC) are
unrelated to the patterns of the La2O2C2 and Nd20202.

This phase change between the oxycarbide of neodymium

and samarium is paralleled by the high temperature
behavior of the lanthanide sesquioxides, in that the
hexagonal lattice found for the lighter lanthanides is
replaced by the monoclinic lattice as the most stable

high temperature form as one moves from Nd to Sm. Nothing
quantitative can be stated from this comparison, but the
parallelism of the oxycarbides and sesquioxides is not
inconsistent with the model proposed, which is based on
the sesquioxide lattice.

b. Thermodynamics

 

A comparison of the enthalpies observed for the

equation (3) is of interest. For the reaction, AH298

-75-

changes from -82. + l. kcal/gfw for Ln = lanthanum

2 _ 9
to -75.9 for Ln - neodymium. Similarly, the heat of
formation of the oxycarbide at 298° varies from

—329.8 i 3.0 for Nd202C2 to -320.0 i 2.0 for La20202,

or an increase in stability of about ten kcal/gfw of the
neodymium over the lanthanum analogue. This behavior
again parallels the oxide system, when the increase in
-AH; of the Nd203 over that of the La2O3 is about 3.6
kcal/gfw.35

3. Phase Relationships

 

The lack of stability of the Ln202C2 phase in
a carbon container above l850° and yet its formation by
are melting at temperatures well in excess of 2000°
indicate that eutectic formation with graphite is
apparently the limiting factor in reactions studied with
graphite present. A small amount of graphite, howeven
is always present when the system is in equilibrium at
high temperature, and this presumably will complicate the
equilibrium study of the system above the fusion tempera-
ture since the graphite present at those temperatures will
undoubtedly form a eutectic. This eutectic apparently has
a much higher equilibrium pressure of carbon monoxide than
the pure Ln202C for a given temperature and thus forms

2

the LnC2.

CHAPTER III. LnuOBC PHASE

A. INTRODUCTION

In an attempt to prepare phase analogues of
La2O2C2 and Nd2O2C2, the methods of the previous chapter
were applied to the samarium, gadolinium, holmium, and
erbium systems. In all systems the attempt failed and
a new phase appeared. This new phase forms the basis

for the work reported in the second half of this thesis.

- 77 _

-78-

B. EXPERIMENTAL METHODS

1. Preparation

 

This phase was first prepared by arc melting
a pressed pellet of a lanthanide metal heavier than
neodymium, its sesquioxide, and graphite in the molar
ratios (Ln:O:C) of 1:1:1 under 1.1 - 1.2 atmOSpheres
of CO. As has been noted, this preparation was carried
out in an attempt to prepare the heavier lanthanide
analogues the Ln202C2 phases, and the procedure used in
the arc melt technique was identical with that discussed
in Chapter (II).

The samples prepared by the technique described
previously were shown by x—ray diffraction to contain no
Ln202C2 (or so little as to be undetectable by this method).
Instead, the samples exhibited the diffraction patterns
characteristic of two different phases:

(1) the readily—identified sesquioxide of the
lanthanide metal

(2) a new phase of apparent cubic symmetry, which
could not be indexed on the basis of any known
oxide or carbide of that particular lanthanide.

The first objective in the study of this phase

was its preparation in a state of purity adequate for

_ 79 _

analysis. In an attempt to eliminate the sesquioxide
impurity, the Ln203/Ln ratio was decreased by one half.
The resulting x-ray diffraction pattern was almost identical
with that of the previous preparation, but micrographic
examination of the arc-melted ingot showed small amounts
of the metal. On the assumption that this cubic phase
was some oxycarbide which was in equilibrium with carbon
monoxide, carbon and sesquioxide at high temperatures,
the reduction of carbon monoxide pressure during arc
melting might reduce the amount of sesquioxide present
by shifting the reaction equilibrium. Consequently,
the partial pressure of carbon-monoxide was reduced
by diluting it with helium in a 1:1 ratio. The product
prepared in this manner showed a marked decrease in the
intensity of the sesquioxide diffraction patter compared
to that of the cubic phase, implying that a further
reduction in the partial pressure of carbon monoxide
might allow preparation of a pure product. The partial
pressure of carbon monoxide was further reduced until
the samples were are melted in pure helium. This
procedure resulted in the complete disappearance of the
Ln203 diffraction pattern.

In addition to the cubic pattern, however, some

other very faint lines not indexable on the basis of any

-80-

known compound of the lanthanide were now seen in the
diffraction pattern. In an attempt to prepare samples
free of this impurity, a series of mixtures having varying
Nd/O/C molar ratios were arc melted under similar cond-
ditions. The fused ingots were examined both micro-
graphically and by x-ray diffraction in an attempt to
develop a preparatory scheme which would produce the cubic
phase having the greatest phase purity. The amounts of
oxygen and carbon in the sample mixtures (Ln:O:C) were
varied in a systematic way by fifteen mole-percent in-
crements over a range of lOO:l5:85 to lOO:85:l5. Another
series of starting mixtures of very low carbon content
(O:C of 100:0 to 85:15 in 5 mole-percent increments)

was also prepared, based on the postulate that the cubic
phase was a carbon stabilized monoxide of the lanthanide.
0

As with the previous oxycarbide phase, Ln C

2 2
this cubic phase hydrolyzed very rapidly in atmospheric

2’

moisture, particularly when in a finely divided state.

This rapidity was demonstrated by the perphoric hydrolysis
of the finely—divided oxycarbide Spattered on the hearth

of the arc melter. The hydrolysis occurred within seconds
after the arc-melter was Opened following the preparation
of samples of the cubic phase. The larger droplets

hydrolyzed more slowly and without pyrolysis. The fused

_ 81 _

ingots were transferred into the glove box as quickly
as possible after preparation and all subsequent sample
preparation for x-ray diffraction or other analytical
methods was accomplished in the glove box. The samples
were stored in massive form either in paraffin—sealed
vials in the glove box or in a vacuum desiccator, thereby
minimizing the hydrolysis of the product. Unlike the
Ln202C2 phase, the cubic phase hydrolysis rate appeared
to slow after an initial coat of hydrolysis products
covered the surface.

Fractured pieces of the prepared samples which
appeared by micrographic analyses to be the most monophasic

were then annealed in a ZrB TaC or graphite crucible

2,
which had been outgassed previously at temperatures of
l850° to l900°. Annealing was accomplished by heating
the sample to 1650° for not more than thirty minutes in
a helium atmosphere of approximately 10-2 torr. Small
amounts of gold—colored material were found on the bottom
of the crucible after the annealing procedure, though
the samples themselves remained unfused.

One sample was cooled quite slowly (approximately
10° per minute) from the annealing temperature to room

temperature by gradual diminuation of the furnace output.

The sample subjected to this protracted cooling was compared

-82-

by x-ray diffraction with a similar sample which had
been cooled as rapidly as possible from annealing
temperature by suddenly shutting off the furnace. This
procedure was effected to determine if the cubic phase
was a high-temperature modification quenchable by rapid
cooling rather than a phase stable from room to fusion
temperature.

Postulation that the gold-colored fused material
observed in the annealing process was an eutectic impurity
suggested that a pure, monophasic material might be
prepared utilizing a zone-refinement technique.

A mechanism which allowed the electrode of the
arc-melter to move laterally over a rod-shaped ingot of
sample to effect a zone-refinement of the ingot was
develOped. (This method and the apparatus is described
in Appendix IV.)

Another preparatory method was the reaction of
a mixture of appropriate molar ratios in a sealed bomb.
Molybdenum was chosen as the bomb container material since
it was found relatively inert to both graphite and carbon
monoxide at the temperatures (up to l850°) and heating
periods (three to four hours) employed. This effective
inertness had been established by x-ray diffraction

examination of scrapings from a molybdenum surface in

5V

—83-

contact with graphite for the temperature and time involved
and by x-ray diffraction examination of a piece of
molybdenum metal fused under carbon monoxide in the arc
melter.

Since it was hoped that this could be related
to the previously described work, the neodymium phase was
studied initially. Neodymium, its sesquioxide and graphite
in the appropriate ratios were placed in a previously
outgassed molybdenum bomb, prepared by heliarcing a
plug in the end of a 6.35-mm 0.D. molybdenum tube about
3.5 cm long. After the bomb was loaded, it was sealed
by a second plug which was heliarced to the open end of
the tube. This tube was then heated to about 18500
by an induction furnace under a pressure of less than
1 x 10'5 torr for not more than four hours. The tube
was transferred to the glove-box when cool and the fused
sample removed for analysis.

A further refinement of the bomb technique
involved zone-refining a sample enclosed in a tubular
bomb about 8 cm long. The loaded bomb was preheated to
l850° in the induction furnace as described previously,
prior to zone refining. Since the fused sample always
occupied less volume than the reaction mixture, typical

tube bombs were about one-third to one—half full after

—84—

induction heating. The tube bomb containing the sample to
be zone-refined was placed in the zone—refiner in a
position inverted with respect to the position in which
it was pre—reacted. This orientation allowed any
relatively low-melting impurities to be carried into the
empty portion of the bomb upon zone—refining.

The apparatus used for the refining process
was 3 NBC electron-beam zone-melter model EZB-94 (described
in more detail in Appendix V). The surface temperature
of the heat zone was maintained at around l700°, which
is equivalent to a correct temperature of approximately
l885° based on a spectral emissivity of 0.37.36 The heat
zone was twice swept down the tube at a rate of 0.33
mm/min.

2. Analysis

 

a. Metal Analysis

 

Samples for analysis were weighed in the glove
box from a finely powdered and thoroughly mixed specimen.
As with Ln202C2, a homogeneous sample was used so that
combination of results from analyses using different
portions of this sample was possible. The hydrolysis of
the samples was carried out differently from the method
used for metal analysis of the Ln 0 C phase. If the

2 2 2
HCl were added directly to the powdered sample, the

-85-

hydrolysis proceeded pyrophorically. Samples treated
in this manner contained large amounts of flocculent
black material in solution in contrast to solutions
formed when the hydrolysis was accomplished without
pyrolysis. To prevent this behavior, the tared samples
were partially hydrolyzed prior to addition of the HCl.
At first, water vapor from a steam source was directed
on the samples to effect the pre—hydrolysis, but later
the sample was treated with about 10 ml. of distilled
water. The remainder of the metal analysis was accom-
plished exactly as described in Chapter II, by preci—
pitating the metal as the oxalate and firing it to
form the sesquioxide.

b. Carbon Analysis

 

The free carbon analysis was accomplished as
before, by filtering the digested solution resulting from
the hydrolysis to collect the unbound carbon.

The total combustion method was employed for
carbon analyses as described previously, but with the
following modifications. Whereas the oxidation of all the

carbon in the Ln202C phases was always completed within

2
six hours as demonstrated by the constant weight of the
absorber after that time, complete oxidation of the cubic
phase was accomplished far more slowly, require more

than twenty-four hours. It is assumed that this very slow

overall oxidation rate is due to diffusion control of the

_ 86 _

reaction, with the oxide coating formed initially inhibiting
further oxidation. This behavior paralleled the
hydrolytic behavior of non-pulverized samples. The very
long oxidations necessarily introduced unwanted error into
the total carbon determinations, since the blank rate for
the absorber became a significant factor for oxidations
requiring more than twelve hours. The oxidation time was
shortened by introducing a few ml of distilled water by

a long pipette into the quartz sample boat prior to
heating. The system was closed very quickly to prevent
escape of any hydrolysis products and the flow of oxygen
initiated, carrying hydrolysis products fromed into the
oxidizer section where the hot CeO2 - CuO catalyst
facilitated the oxidation to carbon dioxide and water.
After about thirty minutes, the temperature of the

sample was elevated. The complete oxidation of carbon,
which was evidenced by the absorber attaining constant
weight, was now achieved within eight hours.

Because the addition of water into the oxidation
chamber would add mass to the ascarite absorber if it were
not completely removed, the H2804 and Mg(ClOu)2 dessiccants
were replaced prior to each total carbon analysis. This
procedure assured the presence of a fresh H20 absorption

system for each experiment.

_ 87 _

c. Oxygen Analysis

 

Oxygen analysis of the cubic phase was accom-
plished via the vacuum—fusion method of Gregory and

37 which will be described in some detail. The

Mapper,
apparatus used (Fig. 6) consists of four principal
sections:

(1) fusion furnace
(2) transfer pump
(3) volumetric analysis system
(A) main diffusion pump with its mechanical

backing pump.

The fusion furnace contains a graphite crucible in which
the fusion occurs (5) and is equipped with an optical
window for temperature measurement (6) and magnetically
activated magazines for the addition of platinum (7) or
samples (8) to the crucible. The transfer pump (2) is a
mercury diffusion pump which will maintain a low pressure
(of the order of 10"5 torr) in the fusion furnace with
relatively high backing pressures (up to l torr) in the
volumetric analysis system. The volumetric analysis
system (3) can be closed off from the main pump (4) and
used for quantitative determination of gases by means of

the multiple-range McCleod gauge (9). The system volume,

which can be varied by valving the fixed volumes (10) in or

88

mmwmw
hASOhwz

I..I'l1l.‘ll’ . I . ,".ll.l

masm
mcwxomm

GHQ: .:

., . IIII I'll

 

“I..--.----.-.._

masm
scamsmmfln
can: .:

A} .

 

'

 

madm
anommmoo< .ma

 

omwso
comaooz .m

 

 

 

 

 

Bopmhm
mwmhama<

.m

Bopmzm mwmhawn< nomhxo .0 .mHm

 

 

 

 

 

oasao>
Hesofipoo .oH

coachdh coamsm .H

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

zones: H838

 

 

0 238.5 .m o
x .
O x. _ 0
no .\ 0.
Au nu
mafia
nowadays .m
animus Ass
X ./
/ i i
/‘ .\
Eopm>m_ /, \
vmaqH .HH 0

 

.w

_ 89 _

out of the volumetric system is calibrated by admitting
known amounts of gas through the inlet (11). The main
pump (4) is used to evacuate the entire system or any part
of it by the prOper combination of valve settings. The
valves used throughout the volumetric system are mercury
float valves, which prevent the contamination of the gas.
The external side of the mercury float valves is evacuated
by the backing pump (12) which is entirely independent of
the rest of the vacuum system.

To effect the analysis, the oxygen-containing
sample is dropped into a molten platinum bath confined
in the graphite fusion crucible. At temperatures at or
above the melting point of platinum (l770°), considerable

1.38

carbon is present in the fused meta This carbon reacts
with any metallic oxide, nitride, or hydride, forming
either the metal or the metal carbide and evolving the
particular non-metal in the form of the elemental gas for
the nitride and hydride or as carbon monoxide in the case
of an oxide. The quantity of the gas evolved is deter-
mined volumetrically by expanding it into the evacuated,
closed analytical system whose volume is known and

by noting the pressure dr0p in the volumetric system.

By combining these pressure—volume relationships with

the gas temperature data, the quantity of gas present

_ 90 _

can be determined. Four samples of the cubic phase were
prepared for oxygen analysis by weighing approximately
25 mg of pulverized sample into cup-shaped containers
formed from 1 cm x 1 cm sheets of 0.13 mm thick platinum.
These containers were then folded to prevent sample loss
and pressed into compact, cylindrically—shaped pellets
of approximately 5 mm in diameter and length in the glove
box. The pellets were loaded into the sample magazine
behind three similarly—sized samples of freshly-calcined
Nd203 which were to be a standard oxygen source to check
the accuracy and reproducibility of the analysis system.
From 10 to 11 g of platinum, in the form of tight rolls
of platinum formed from 5 X 40 mm strips of 0.5 mm thick
sheet, was placed in the platinum-bath magazine and a
previously outgassed graphite fusion crucible (Ultra
Carbon Corp., Bay City, Mich.) was positioned in the
fusion furnace in a graphite powder bed, which serves

as a support for the crucible.

The system was evacuated and both diffusion
pumps turned on. When a pressure of less than 10‘5 torr
was attained in the system, the crucible was heated slowly
to a temperature of 2100° and outgassed at this temperature
for 12 hours. At the same time, the volumetric analysis

portion of the system was also outgassed, by use of heating

_ 91 _

tapes wrapped around the glass section to reduce further
the so—called "blank rate" describing the pressure
stability when the volumetric analysis system was isolated
from the pump. The blank rate was considered acceptable
when the pressure of the volumetric system did not increase

6 A

from 10- torr to more than 5 x 10' torr in a fifteen
minute interval from the time it was isolated from the
pump. When an acceptable blank rate was achieved, the
crucible temperature was decreased to about l200°, and the
analysis system was closed off from the pump and its
pressure recorded. A sample was dr0pped into the crucible
and its temperature was increased to 18000 - l825° to
effect extraction of the carbon monoxide. The pressure

in the volumetric analysis system was checked periodically,
and when it attained a stable value (usually within

twenty minutes), the temperature was recorded. From the
change in pressure upon evolution of carbon monoxide,

the volume of the analysis system and the temperature of
the gas (assumed identical with room temperature twenty
minutes after its evolution), the amount of the carbon
monoxide was determined using the ideal gas law (valid

6 to 10"1 torr for the two

in the pressure range 10'
significant figure measurements made). The percentage of
oxygen in the original sample is calculated from the amount

of carbon monoxide formed.

- 92 _

d. X-Ray Analysis

 

The initial identification of this cubic oxy-
carbide phase and most of the analytical work necessary
for development of the preparatory techniques was effected
by Bebye-Scherrer x-ray diffraction analysis, employing
cameras of 114.6 mm diameter and copper (Aau = 1.5418 A)
radiation. The use of the Debye-Scherrer method in phase
purity determination was facilitated by the relative ease
with which the cubic pattern could be identified and the
fairly good intensities of the diffraction lines, in con-
trast to the intensities of the sesquioxide and the Ln202C2
patterns. Capillaries of 0.3 mm diameter for use in the
cameras were filled in the glove box and flame-sealed
immediately upon their removal from it.

More detailed x-ray diffraction measurements
were made on Siemens and Norelco diffractometers. The
powdered diffractometric Specimens were mounted on glass
slides by spreading evenly a very thin layer of Canada
balsam-sample mixture over the center of the glass surface.
When this mixture had hardened and the surface of the
sample was dry, another very thin coat of the Canada
balsam, diluted with xylene, was flowed over the sample

surface to provide a barrier against hydrolysis by

atmOSpheric moisture. These samples appeared to remain

_ 93 -

unaffected by exposure to the atmosphere for 4—6 hours,
after which the slight change of color of the surface
indicated the onset of hydrolysis. The diffractometric
patterns were analyzed for faint superstructure lines
which would be indicative of some ordering of lattice sites.
In addition, intensity measurements on the peaks represent-
ing (111) and (002) planes in the diffractometer pattern
were effected by measuring the areas subtended by those
peaks with a polar planimeter.

3. Hydrolysis Experiments

 

An open, wide-mouthed vial containing a specimen
of this cubic oxycarbide was placed carefully into a
500 m1. flask fitted with a side arm to which a stopcock
was attached. Care was taken to insure that the vial
remain upright throughout the experiment. The flask was
sealed with a rubber st0pper and evacuated through the
side arm. Ten ml of distilled water was then added to
the evacuated flask by filling an external tube leading
to the stopcock, opening it carefully and letting the
water enter the evacuated flask very slowly. Care was
taken to insure that water did not come in contact with the
sample. After two days of hydrolysis, helium was admitted

to bring the internal gas pressure up to about one

_ 94 _

atmosphere. Gas samples of the contents were withdrawn with
a syringe inserted through the rubber stopper, and subjected
to gas chromatographic analysis in an F and M model 810 gas
chromatograph, which utilized six foot columns filled with
100-120 mesh Porapack Q (Waters Assoc., Inc., Framingham,
Mass.). Flame ionization detectors were used. Retention
times had been calibrated by use of samples of CP methane

or CF acetylene.

4. Equilibrium Studies

 

Attempts were made to determine if this phase
formed a component in any equilibrium systems involving
carbon monoxide, and was thus possibly suitable for study
by the same methods used in the Ln202C2 equilibrium

measurements.

RESULTS, LnuO3C PHASE

1. Preparatory Results

 

The mole ratios of Nd203:Nd:C which yielded an
arc-melted ingot judged on the basis of micrographic
and x-ray diffractometric analysis to possess the highest
phase purity was l:2.5:l, and the metal:oxygen:carbon ratios
were 4.5:3:1. Very little quantitative information can be
deduced from these starting ratios, however, since
considerable amounts of carbon were deposited on the
hearth around each sample (along with small pieces of the
sample spattered there by the arc), and a thin layer of
sesquioxide was usually found on the interior surfaces of
the arc—melter chamber.

Micrographic examination of this "best" sample
prepared according to ratios indicated above showed small
amounts of a gold-colored minor phase distributed through
the gray-silver major phase. The "herringbone" pattern
exhibited by this minor phase indicated that it was

probably the lower melting of the two components, and

the annealing experiments were not only used to sharpen

-95..

_ 96 _

the x—ray diffraction pattern by allowing quenched—out
crystal disorders to be relieved, but also as an attempt
to melt this minor phase and hopefully to remove at least
part of it from the major phase.

That this indeed occurred was demonstrated by
the micrographic analysis of the samples after annealing.
"Herringbone" shaped voids, which were similar in size
and shape to the patterns shown by the minor phase prior
to annealing, appeared in the unfused major phase. The
identity of the minor phase could not be determined
definitely, however. It appeared that the fused material
in the crucibles after annealing was an eutectic of

NdC but it might have been a different phase which was

2,
converted to an NdC2 eutectic after exiting from the major
phase. The identification of small amounts of impurity
phases often presents a formidable problem.

Attempts to prepare samples of greater apparent
phase purity by using the arc-melter as a zone-refiner
(Appendix IV) failed. However, larger amounts of ses-
quioxide were found on the interior of the arc melter
after an attempt at zone refining than after a normal
preparation.

The unfused major phase after annealing exhibited

very small amounts of the minor phase, usually in the portion

- 97 _

of the sample which was in contact with the bottom of
the crucible. Consequently, these materials were judged
to be sufficiently mon0phasic for analysis.

Preliminary analytical results indicated an
approximate stoichiometry of "Nd302C." It was apparent
that if some preparative technique were devised in which
no material was lost from the sample, the above stoichio-
metry could be checked independently by preparing a
sample according to its indicated mole ratios, then
analyzing the phase purity of the product.

Since molybdenum does not react measurably
with carbon monoxide (an anticipated decomposition
product of any oxycarbide) and only very slightly with
carbon at the temperature and reaction times employed,

a sample in the above ratio was sealed in a molybdenum
bomb and heated.

Preparations based on the postulated "Nd3020"
reactant ratio were found by micrographic analysis to
have very small amounts of the gold-colored minor phase
present. However, the ability to confine the phase to
a sealed bomb did suggest use of a zone refining tech-
nique.

The zone—refinement technique described in the

previous chapter proved successful. Small amounts of

_ 98 -

material were separated from the sample by the zone
refining and were shown micrographically to contain
appreciable quantities of the gold-colored minor phase
in some of the cubic phase. The amounts were still too
small for x-ray diffraction identification, however.
The remaining sample exhibited no minor phase upon
micrography, but, curiously, still possessed small
voids.

Specimens prepared in this manner were used for
the oxygen analysis; much of the carbon analysis, and all

of the diffractometric investigations.

2. Analytical Results

 

The analytical results, reported as weight
percentages and including the standard deviations of the

measurements are as follows:

Nd, 90.50 i 0.23%
0, 7.90 i 0.02%
c, 1.82 i 0.7%

These data indicate a stoichiometry of

Nd

1.00 i .02 00.787 i .002 0.24 i 0.08

or approximately Nd4030.

-99..

3. X-Ray Powder Diffraction Analysis Results

 

The diffractometer diffraction patterns exhibited
only peaks which were assignable to a face centered cubic
symmetry. No indication of any superstructure was found.
This lack of superstructure implies:

1. either a lack of, or a statistical distribution
of, lattice vacancies, or

2. a probable lack of ordering of the carbon sites.

The ratio of the measured intensities of the
(111) and (002) peaks was compared with the values cal-
culated for this ratio for both an NaCl and ZnS lattice

type by assuming that an Nd+3

ion and an oxygen atom
located at the reSpective lattice sites approximate

the electron density of the Nd403C Species. The scattering
39

factors reported by Cromer and Waber were used for these
calculations. The temperature factor of oxygen was
assumed to be 1.5 A2 and that of neodymium 0.6 A2, and the
absorption correction was considered to be negligible.

The complete set of calculated results (Appendix VI)

shows that the differences in relative intensities

between the two structures are significant for only the

(111) and (002) planes. With the exception of the

intensities of these two planes, the observed values

- 100 -

were in general agreement with those calculated for both
structures. The calculated and experimental data for

these two planes are compared in Table III.

TABLE III

Calculated and Observed Itensities for
the (111) and (002) planes of Ndu03C

 

Structure Type Plane
(111) (002)
ZnS 100.0 33.2
*-
Ndl.OOOO.79CO.24 100.0 66.4 i 1.2
NaCl 100.0 68.6

*-
Standard Deviation of Measurement

0n the basis of the data presented in this
table, the oxycarbide is considered to possess a NaCl—
type fcc structure having a lattice parameter a0 = 5.1406 1
0.0007 A, where the error reported is the standard deviation
of a least squares fit of the powder data.

Samples prepared using excess amounts of metal,
oxygen, and carbon showed invariant lattice parameters, and

indicate a narrow homogeneity range for this phase.

4. Hydrolysis Results

 

Graphic output of the vapor phase chromatograph

diSplayed a very intense peak which was attributable to

— 101 -

methane, and a very weak peak which could be assigned to
acetylene. The acetylene peak tailed off on the longer
retention time side, diSplaying two small shoulders,
presumably due to hydrolysis products similar to acetylene.

Planimeter measurements of the areas subtended
by the two peaks, corrected for the difference in the
flame-ionization detector reSponse to methane and acetylene,
indicated that the hydrolysis products consisted of
96 i 3% methane and 4 i 2% of a gas mixture of which
acetylene comprised the major part.

An attempt was made to determine the conductivity
of the Nd4O3C in a qualitative manner, by use of the
pressed pellet technique. The conductivity appeared
equivalent to that of an ingot of Nd metal of comparable

size.

— 102 -

D. Discussion, Ln40 C Phase

3

This ternary oxycarbide phase appears to be
a carbon stabilized lanthanide monoxide. The apparently

fixed C/0 ratio is markedly different from the variable

ratios exhibited by the previously discussed ZrOl_EC€31

and U0 0 10’14

l—x x phases which might be considered analo-

gous to this Nd403C system. Fixed 0/C ratios have been

observed in the Ln202

these Ln20202 phases, the fixed O/C ratio was postulated

to result from ordered anisotropic anion sites in the

C2 phases discussed previously. In

lattice, occupied by acetylide ions. These sites were
considered different from the other spherical anion
lattice sites which were occupied by oxide ions. This

argument does not seem applicable in the Ndu0 0 case,

3

since the carbon units appear to be Spherical methanide

entities (discussed in the following section), and no

evidence was found by x-ray diffraction for their ordering

in the crystal. If the methanide units are indeed statis—

tically distributed over the available anion sites, the

C/0 ratios might be varied over at least a moderate range,

paralleling the behavior of the Ce(0,N)32’33’uO and the
Sm(0,N) systems.41 (This analogy between nitride, carbide
and oxide ions is based on the fact that they form an

isoelectronic series.)

_ 103 _

A definitive explanation does not exist,
therefore, for the apparently anomalously invariant C/O
ratios found in this phase, but a number of postulates
can be made whose consequences appear consistent with
observed facts. If one examines the binary ionic compounds
of the tripositive lanthanides formed with the anions of
the isoelectronic series C-M, N_3, and 0‘2, he sees
that the only stable 1:1 solid phase is that of LnN, as
neither the LnC nor the LnO phase has been characterized.
(Reports of solid LnO phases will be discussed subse-
quently.) This stability of the LnN can be postulated as
being principally due to two factors:

1. Optimum anion radius

2. Apparent formal charge neutrality.
Consequently, it can be postulated that the stability
of any solid phase of a tripositive lanthanide with

any combination of the isoelectronic series 0-2, N-3

-u

C will be a maximum when both the formal charge

)

neutrality and the average anionic radius approach that
of the corresponding lanthanide nitride.

This postulate is supported by the Nd4O3C phase.
The lattice parameter for this phase is a0 = 5.1406 i
0.0007 A, which compares very favorably with the lattice

parameter of NdN, a0 = 5.151 A.29

— 104 —

The effective anionic radius in the phase

Nd403C calculated by the method described in Appendix VII,

is
reff = .75(1.40) + .25(2.60)
reff = 1.70 A
based on the Pauling radii for 0= = 1.40 A and C'LL = 2.60 A.

This calculated re compares favorably with the Pauling

ff
radius for N-3 = 1.71 3.42
The formal charge on the neodymium in Ndu03C
is 2.5 rather than 3, so the second of the factors is not
completely satisfied. In this case, however, the two
criteria cannot be satisfied simultaneously, since the

stoichiometry demanded for formal charge neutrality,

Nd20C would possess an effective radius

reff = .5(.40) + .5(2.60) = 2.00 X,
which is much larger than that required by the first
criterion. Thus, in the Nd403C phase, at least, the
requirement of formal charge neutrality is apparently
the less stringent of the two factors. This is not
inconsistent with the fact that many crystalline carbides
of the lanthanides possess conduction bands which
accept electrons from the lanthanide species, thereby

producing a low oxidation state for the metal.

_ 105 -

This general postulate has a number of other
implications which will be discussed in the chapter
describing future research.

There are several possibilities for placement
of the methanide ions in the crystal lattice. They may
be situated either as (a)C-u methanide ions occupying
one fourth of the octahedral anion sites, (b) randomly
oriented C52 acetylide units which occupy one eighth
of the octahedral sites (and thereby leave one eighth
of the anion sites vacant) or (c) a combination of the two.

Since the hydrolysis products consist almost
entirely of methane, with only a trace of acetylene and
its analogues, the carbon units appear to be present as
the methanide ions. Even though the hydrolysis was
effected principally by water vapor, some minor reaction
of methanide ion to form acetylene, followed by sub-
sequent reduction to more highly hydrated Species cannot
be precluded. In addition, the complete absence of
trace amounts of a dicarbide impurity cannot be proven,
but most of the carbon present must be considered to be
in the form of methanide ions. Thus the second, and
for the most part the third alternative schemes for carbon
placement in the lattice may be eliminated. The hypothesis
of methanide ions is not inconsistent with the x-ray

diffraction results. That the cubic form appears to be

- 106 -

crystallographically stable at both room and elevated
temperatures and not just a quenched, high temperature
modification substantiates further this hypothesis, since
acetylide-containing carbides usually exhibit symmetry
different from cubic at room temperature due to the
presence of the anisotropic acetylide species.43

Furthermore, if the carbon were to exist in
this phase in the form of acetylide ions, the formal
oxidation state of the metal would be 1.75, compared with
the value of 2.5 based on the methanide assumption. The
higher formal oxidation value seems the more likely.

It seems desirable to compare the Lnu03C phase
with the various reported lanthanon monoxides. The best
characterized monoxide, EuO, is not strictly comparable,
because of the much greater stability of the dipositive
eur0pium compared with that of the dipositive neodymium.
The nitride—stabilized samarium monoxide is probably the
best characterized "monoxide" if EuO is excluded. The small-
est lattice parameter reported for the SmO (a0 = 4.978A)4#
presumably describes the lattice of a sample least affected
by nitride ion contamination. The difference between this

parameter and that of Nduo C is 0.162 A. This difference

3

can be accounted for in a relatively straight-forward

-107-

manner. To twice the increase in the ionic radius from

Sm+3 to Nd+3(0.031 A)3Ur must be added the increase of

the effective anion diameter over that of oxygen (2.80 A)42
due to the presence of the larger methanide ions which
occupy one fourth of the sites. This increase (c.f. p. 104)
is 0.30 A, and the calculated difference in the lattice
parameter is 0.33. On the other hand, if one uses a
methanide radius determined to be one fourth of the unit
cell face diagonal of the methanide-type antifluorite

phase Be2C,29 corrected for the change in coordination

from eight in the antifluorite to six in the NaCl structure
by use of Pauling's formulation with an approximated Born
exponent of 9,42 the increase in the effective anion

radius is 0.06 A, and the calculated difference in the

lattice parameter for the SmO and Nduo C is 0.12 A.

3
The measured difference of 0.16 A is bracketed by the two
calculated values, but is somewhat close to the latter.

It appears from the difference in the methanide ionic radii
in the two computations that this ion is probably somewhat
diffuse and readily distorted. It does appear, however, that

the two phases are similar, the "samarium monoxide" being

stabilized by nitrogen, thereby forming an oxynitride and the

— 108 -

"neodymium monoxide" being stabilized by carbon, forming
an oxycarbide. 0n the basis of the proposed postulate
concerning phases of this type, the oxynitride should
become less and less stable with a decrease in the N_3/O=
ratio. Evidence of this behavior is demonstrated by the
reported inability to prepare the samarium monoxide in
the complete absence of nitrogen.b'5

0n the basis of this work, any "monoxide"
except possibly Eu0 prepared by reduction of the ses-
quioxide with graphite might in truth be an oxycarbide.
Additionally, the postulates deduced from this work
imply that many of the other "monoxides" prepared by
methods other than carbon reduction are oxynitrides,
having variable O/N ratios.

The high conductivity of the Nd403C phase is
similar to that observed in most lanthanide carbide phases,
leading to the assumption that the bonding is similar to
that postulated for the carbides. Assuming this hypothesis
to be true, the metal ion should be in a tripositive
oxidation state and should supply electrons into conduction

bands. The lattice parameter observed for the Nd40 C is

3

_ 109 -

consistent with the existence of Nd+3 ionic units in the
cell, as is demonstrated by the comparison of its observed
lattice parameter with that of NdN, in which the Nd+3 unit
has been established quite conclusively.

In this oxide carbide phase, each neodymium
ion presumably could donate one half an electron to a
conduction band (assuming all carbon units to exist as

C-4 ions in the stoichiometry Ndu0 C). Indeed, the

3
silver-grey color of the phase is indicative of band
occupation, but further conduction and magnetic studies

are needed before definite statements can be made con-

cerning the bonding.

CHAPTER IV. FUTURE RESEARCH

A. Ln202C2 PHASE

 

Two areas of great importance, structure and
bonding, have yet to be investigated.

1. Structure Determination

 

The structure determination is dependent
on the possession of a small high quality single crystal
of the Ln202C2. Some Single crystals of uncertain quality
were separated from the ingot of an arc melt, and some
from the fused crust of the Ln202C2, Ln203 and graphite
equilibrium mixture. These crystals were the basis for
some preliminary eXperiments as detailed further in
Appendix III.

The vapor-phase transport method might also be

used. In equilibration attempts with the Nd40 C phase,

3
small crystals were found growing on the underside of

the crucible lid. Most of these crystals showed the

gold color typical of the dicarbide, but toward the axial
center of the crucible lid, some silver grey crystals were
observed. These grey crystals were too small to permit
any x—ray diffraction examination, but showed hydrolytic
behavior typical of an oxycarbide upon exposure to atmos-

pheric moisture. If, as it appears, this reaction can

be described by the equation

- 110 -

- 111 -

LnO(g) + 0(8) " Ln(O,C) (S), (36)

oxycarbide crystals can be grown by passing LnO gas over
a graphite surface whose temperature varies. (A carbon
monoxide atmosphere has to be maintained in such a system
to prevent decomposition of the oxycarbide.) Presumably
oxycarbide crystals would be formed at that point on the
graphite surface where the temperature was such that an
equilibrium was maintained between the crystals formed
and the carbon monoxide atmosphere. With suitable
crystals such typical single crystal x-ray methods as

the Weissenberg or precession method could be employed.

2. Bonding Studies

 

Studies into the fundamental nature of the
bonding are, like the structure determination, dependent
on pure crystalline samples. If such samples can be made,
information concerning the oxidation state and resultant
bonding of the oxycarbides could be collected by classical

magnetic susceptibility and conductivity measurements.

— 112 -

B. Lnu0 C PHASE

3

1. Structure and Bonding

 

The investigations outlined in the previous
section for the Ln202C2 phase are applicable here also.
The molybdenum bomb which was described in Chapter III
might be the best method for growing suitable crystalline

material.

2. Thermodynamic Studies

 

Thermodynamic information about this phase
might be accessible either by static or flow techniques.
Preliminary experiments indicated a much lower equilibrium
carbon monoxide pressure than that observed for the Ln202C2
phase, thus a flow technique might be the preferred method.

3. Phases Based on Model Proppsed in Chapter III, part D.

 

In the discussion section in Chapter III, a
prOposal was made that the cubic phases composed of

11

lanthanide ions and anions of the isoelectronic series C- ,
N-3, 0"2 were most stable when the "effective anionic
radius" approached that of the N-3 ion and charge neutrality
was maintained. This prOposal was used to explain why
Ln(O,N) phases Showed varying N70 ratios, becoming more
stable as the N/O ratio increased; and why the Ln4O3C

showed very narrow composition limits.

- 113 -

It would be of interest to extend this pro-
posal to include F-, another member of the iso-
electronic series. Such a substitution would suggest
that a cubic oxyfluoride phase would have a variable O/F
ratio, and that its stability would increase as the oxide
anion concentration increased.

Another prOposed phase would be Ln(C,F), a
lanthanide carbide fluoride. Since one of the anions
is larger and the other smaller than the N'3 ion, the
proposed model would suggest that a fixed C/F ratio would
be found, with the ratio such that the "effective anionic
radius" would be equivalent to that of the N-3 ion. A

prOposed preparation method is described by the reaction
Ln(s) + CF(s) -> Ln(CF),

where the solid reactants are melted together, either in
the arc melter or a suitable container such as platinum
bomb.

N223: The graphite fluoride is thought to be ideally
(CF)n’ and has been prepared with C:F ratios up to 1:0.99,

and is a white, non—conducting solid.

1.

10.

ll.

12.

13.

l4.

15.

CHAPTER V. BIBLIOGRAPHY

B. R. Geijer, Crell Ann., I. 229 (1788).

 

R. C. Vickery, Chemistry of the Lanthanons, Academic
Press, N. Y., 1953.

 

0. Pettersson, Bericht., 28, 2419 (1895).

H. Moissan, Compt. Rend., 123, 148 (1896).

 

Ibid., 1312 595 (1900).

H. Moissan, A. Etard, Compt. Rend., 122, 573 (1896).

 

G. L. Buchel, Thesis, Michigan State University
(1963).

0. Heusler, Z. Anorg. u. Allgem. Chem., 154, 353
(1926).

 

J. R. Piazza, M. J. Sinnott, J. Chem. Eng. Data, 7,
451 (1962).

 

R. F. Stoops, J. V. Hamme, J. Am. Ceram. Soc., 57,
59 (1964)-

 

I
A. Heiss, M. Dode, Rev. Hautes Temp.Refract., 3,
245 (1966).

 

L. M. Litz, A. B. Garrett, F. C. Croxton, J. Am. Chem.
Soc., 79, 1718 (1948).

 

M. W. Mallett, A. F. Gerds, D. A. Vaughan, J. Electrochem.

 

292-: 98, 505 (1951).

P. Chiotti, W. C. Robinson, M. Kanno, J. Less-Common
Metals, 19, 273 (1966).

 

 

L. Blum, J. P. Morlevat, Rev. Hautes Temp. Refract.,
2

P.
3. 53 (1966).

- 114 -

“‘rWE—Hflfiww—*

l6.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

_ 115 _

E. K. Storms, Thermodynamics, 1, International Atomic
Energy Agency, Vienna (1966) p. 309.

 

W. A. Chupka, J. Berkowitz, C. F. Giese, M. G. Inghram,
J. Phys. Chem., 62, 611 (1958).

 

F. H. Spedding, K. Gschneider, A. H. Daane, J. Am.
Chem. Soc., 89, 4499 (1958). ‘“"'

 

I. M. Klotz, W. A. Benjamin, Chemical Thermodynamics,
N. Y., (1961).

 

C. N. Lewis, B. Randall, Thermodynamics, 2nd edition,
revised by K. S. Pitzer, L. Brewer, MEGraw-Hill,
New York, N. Y. (1961).

 

A. Findlay, The Phase Rule, ninth edition, revised
by F. N. Campbell, N] 0. Smith, Dover, N. Y. (1951).

 

A. D. Butherus, R. B. Leonard, G. L. Buchel,
H. A. Eick, Inorg. Chem., 5, 1567 (1966).

 

E. J. Huber, C. E. Holley, Jr., J. Am. Chem. Soc., 75,
3594 (1953)-

 

F. D. Rossini, D. D. Wagman, W. H. Evans, S. Levine,
K. Jaffe, U. S. National Bureau of Standards Circular
500, Series I, U1 S. Government Printing Office,
Washington, D. C., 1952, p. 99.

 

R. C. Vickery, R. Sedlacek, A. Ruben, J. Chem. Soc.,
503 (1959).

 

G. J. Palenik, J. G. Warf, Inorg. Chem., 1, 345
(1962).

 

M. J. Bradley, L. M. Ferris, Inogg. Chem., 1, 683
(1962).

 

M. Atoji, J. Chem. Ppys., 35, 1950 (1961).

 

 

 

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

- 116 -

A. Taylor, B. J. Kagle, Crystallographic Data on Metal
and Alloy Structures, Dover, New York, N.Y., (1963).

 

F. Leprince—Ringuet, Compt. Rend., 264, 1957 (1967).

 

F. Leprince—Ringuet, A. Lejus, R. Collongues,
Compt. Rend., 258, 221 (1964).

 

H. A. Eick, K. Manske, Report at American Chemical
Society Convention, Miami Beach, Florida, April, 1967.

H. A. Eick, N. C. Baenziger, L. Eyring, J. Am. Chem.
Soc., 78, 5147 (1956).

 

D. H. Templeton, C. H. Dauben, J. Amer. Chem. Soc.,
79; 5237 (1954)-

 

J. O'M Bockris, J. L. White, J. D. McKenzie,
Ppysioco-chemical Measurements at High Temperatures,
Academic Press, N. Y. (1959).

 

Handbook of Chemistry and Physics, Edition 48,
p. E-164, Cleveland (1967).

 

J. N. Gregory, D. Mapper, Analyst, 89, 225 (1955).

H. A. Sloman, C. A. Harvey, O. Kubaschewski,
J. Inst. Metals,_8g, 391 (1951-2).

 

D. T. Cromer, J. T. Waber, Acta. Cryst.,_18,
104 (1965).

 

H. A. Eick, unpublished work.

H. A. Eick, "Rare Earth Borides and Nitrides,"
Rare Earth Research, E. V. Kleber, Ed., The
Macmillan 00., New York, (1961).

 

L. Pauling, The Nature of the Chemical Bond, p. 514,
Third ed., Cornell Univ. Press, Ithaca (1960).

 

 

_ 117 -

43. A. L. Bowman, N. H. Krikorian, 0. P. Arnold,
T. E. Wallace, N. G. Neverson, Presented at 6th
Rare Earth Conf., Gatlinburg, Tenn., May 1967.

44. L. Eyring, Private communication.

45. W. Klemm, G. Winkleman, Z. Anorg. u Allgem. Chem.,
288, 87 (1956).

CHAPTER VI. APPENDICES

APPENDIX I. SOURCES AND PURITY OF MATERIALS

 

 

La Metal, 99% Purity, Research Chemicals, Div. of

Nuclear Corp. of America, Box 14588, Phoenix,

Arizona.

Nd Metal, 99% Purity, Michigan Chemical Co.,

St. Louis, Mich.

La203 99.9+% Purity, Research Chemicals, Div. of
Nuclear Corp. of America, Box 14588, Phoenix,

Arizona.

Nd203, 99.9% Purity, Michigan Chemical Co.,

St. Louis, Mich.

Graphite, No. 38 Powder, Fisher Scientific Co.,

1 Reagent Lane, Fair Lawn, N.J.

- 118 -

'7

P I;
I.

- 119 _

APPENDIX II. HIGH SPEED PUMPING SYSTEM
AND CURRENT CONCENTRATOR

The pumping system used (Fig. 7) consists of a
Cenco Hyvac No. 45 mechanical pump (Central Scientific Co.,
2600 Kostner Ave., Chicago, Ill.) backing a four inch oil
diffusion pump (Consolidated Vacuum Co., Rochester 3, N.Y.)
and a four inch, baffled liquid nitrogen cold trap.

The heating portion of the system was enclosed in
a 35 cm long 100 mm ID Pyrex pipe, which sat on a flat
silicone rubber gasket in a flanged fitting which lead to
the pumping system. The current concentrator hung from
a flat, water-cooled lid, which was attached by a similar
gasket to the base flange. The heating system was sealed
by atmOSpheric pressure on both flat gaskets.

The water-cooled current concentrator consists of
a c0pper donut which acts similarly to the core of a
transformer and forces the toroidally-shaped field to pass
through the hole in the donut. Such a shaping of the
field causes it to couple very efficiently with the
crucible.

The entire concentrator was prevented from
coupling to the induction field by a vertical cut which
prevented establishment of a closed conduction loop in

the copper.

- 120 -

AV)

 

Sukm>m
GZEZDQ

 

 

 

 

 

 

 

mwhwzozS).

>

 

 

 

 

 

 

Q

       

 

 

 

\I/

 

 

w03<0

 

 

 

 

 

mmnmmwma

   

nappy
popes

 

 

 

 

 

 

 

 

: PEDIm Ill

Pmom oz:sm.>

Ewhm>m ozEDmdm—z mmammmma

 

PRESSURE MEASURING SYSTEM

FIGURE 7.

- 121 -

APPENDIX III. TENTATIVE INDEXING OF L3202C2

Based on Weissenberg X-ray studies of a poor-
quality single crystal of La202C2, a monoclinic unit cell
was pr0posed and all possible d-values from that proposed
cell were calculated and compared with the observed
values. Some modification of the prOposed cell proved to
be necessary, since some of the axes used in the Weissenberg
studies were apparently not principal axes. The final

parameters of the proposed monoclinic unit cell for

La 0 C are:

2 2 2
a = 7.289 g
b = 3.529 A
c = 7.028 A

a = 84.370

- 122 —
APPENDIX IV. ARC ZONE MELTER

The arc zone melter (Figure 8) consists of an
assembly that is fastened to the hearth of the arc melter
previously described (Figure l). The electrode was
mechanically attached to a traveling block, which was
driven horizontally by an electrically-driven lead screw.
The arc traversed a 10 cm long groove milled in the
hearth. The samples were pelletized, laid in the hearth
groove and then arced. The electrode could be driven

at various speeds.

WATER-COOLED

PROBE \[7 /
~r—H—L—.

‘1 l
l r
.1, I [#4,

LEAD SCREW

 

 

 

 

 

 

 

TUNGSTEN ELECTRODE

 

L. _______________ I
WATER-COOLED COPPER HEARTH

 

 

 

FIGURE 8, ARC ZONE MELTER

II.

(«I

- 123 _

APPENDIX V. ELECTRON BEAM ZONE MELTER

The Model EZB-94 zone melter (c.f. page 84)
consists of a vacuum chamber and pumping system in which
was placed the mechanical drive mechanism and separate
from this assembly, a high potential DC power supply
(Figure 9).

The sample was mounted vertically inside the
vertical traverse of a circular thoriated tungsten filament,
which could be driven mechanically at speeds as low as
1 mm per hour.

After the chamber had been evacuated by the
pumping system, the filament was heated and a milliamp-
range current was induced to flow from the filament to the
sample by a potential in the 15 to 50 kilovolt range. The
electrons boiled off the filament were accelerated by this
high potential and transferred this energy to the sample
upon collision. This technique heated a rather flat
cross section of the sample. This heated zone was swept

down the sample by the mechanical drive.

Bell Jar

 

124

 

 

 

 

 

 

 

 

 

 

 

 

High Potential
Direct Current
Power Supply

 

 

 

Filament Heater
Power Supply

 

 

 

11

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample

 

Pumping
System

 

Lhoriated Tungsten Filament

Figure 9. Electron—Beam Zone Melter.

 

 

 

-125-

APPENDIX VI. CALCULATED INTENSITIES FOR X-RAY POWDER
DIFFRACTION LINES OF Ndu03C IN ZnS and
NaCl LATTICE TYPES

 

 

 

(h,K,t) Relative Itensity Calculated
NaCl Lattice ZnS Lattice
111 100.0 100.0
002 68.6 33.2
022 8.8 6.7
113 16.7 15.5
222 6.8 3.7
004 2.8 2.1
133 5.0 4.5
024 5.8 3.4
224 3.6 2.8
333, 115 2.2 1.9
004 1.0 0.7
135 1.7 1.5
244, 006 0.1 0.1

—126-
APPENDIX VII. EFFECTIVE ANIONIC RADII

In solid phases which contain two or more types of
anion, two possibilities exist for the distribution of the
anion sites. They may be regularly situated so as to form
a super lattice of some type, or may be statistically
distributed.

In the first case, such behavior can usually be
detected in the x-ray diffraction patterns. In the second
case, no extraneous lines appear in the x-ray diffraction
pattern, though the line intensities sometimes are different
from what one would anticipate from either anion singly
occupying the lattice sites. This is due to the differences
in scattering of the x-rays by the two anions.

In the statistically-distributed anion case, we
define the effective anionic radius to be a linear com-
bination of the two separate radii. This procedure
yields a composite radius with the contribution of each
anion proportional to the fraction of anionic sites it
occupies.

For example, in the general case for a phase

MXaYb

with the anions statistically distributed in the lattice,

the effect anionic radius re would be the linear com-

ff
bination of the two anionic radii rX and ry

r _ a b
Eff—mrx'i'm. I'y

. .111111111

3 129