Nonstoichiometry in Chalcogenide Systems - Advances in Chemistry

Chapter. Previous Chapter · Next Chapter ... Chapter 15, pp 170–178. DOI: 10.1021/ba-1964-0039. ... CHEMICAL SOCIETY. Peer Reviewed Book Chapter ...
0 downloads 0 Views 1MB Size
15 Nonstoichiometry in Cholcogenide Systems J. S. PRENER General Electric Research Laboratory,

Schenectady, Ν. Y.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

This paper reviews experimental and theoretical studies of nonstoichiometry

in metallic chalco-

genides exhibiting both small and large devia­ tions from stoichiometry.

These deviations result

from the presence of lattice disorder arising from the presence of structural defects.

The statistical

thermodynamic theory is outlined which gives the equilibrium concentration of these defects, and hence the stoichiometry, as a function of the values of the pressure, temperature, and chemical potential of one of the constituents.

The experi­

mental methods used to determine the concentra­ tion of defects and the limits of stability of the single-phase

region

are

presented.

For com­

pounds having extended ranges of homogeneity, Anderson's statistical thermodynamic treatment is presented, in which defect interaction is included. Finally, some structural studies of nonstoichiomet­ ric transition metal chalcogenides are reviewed.

• n principle, at any temperature above absolute zero every crystalline compound can exist as a single phase over a range of composition. A m o n g the metallic chalcogenides are two broad classes of compounds: (1) that group of chalco­ genides, particularly those of the transition elements, w h i c h from chemical analysis and structure determinations b y x-ray diffraction have been found to exist as stable homogeneous phases over a considerable range of composition (these are the compounds w h i c h exhibit large stoichiometric deviations), and (2) chalcogenides such as the II to V I compounds—e.g., ZnS, CdTe—and the compounds of P b whose deviations from a 1:1 stoichiometry are extremely small. These two classes are discussed separately, since the experimental and theoretical methods used i n study­ ing them have been different. Chalcogenide

Compounds

with Small Stoichiometric

Deviations

In compounds such as C d T e (26), C d S (23), P b S ( 5 ) , or P b T e ( 7 ) , the very small deviations from 1 to 1 stoichiometry result from lattice disorder arising from the presence, i n small concentrations, of vacancies or interstitials. Thus, i n C d T e it is believed that the predominant structural defects are C d interstitials 170 In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

15. PRENER

171

Chalcogenide Systems

(Cdj) and C d vacancies ( V ) , and an excess of one over the other gives C d - r i c h or Te-rich C d T e . I n C d S it was presumed that C d vacancies and sulfur vacancies ( V ) are the predominant defects and here again nonstoichiometric compositions arise from an excess of one type of defect over the other. Other combinations of interstitials and vacancies are possible and i n B i T e (15) indications are that stoichiometric deviations arise from the presence of one type or atom upon the sublattice of the other ( B i and T e ) . T h e maximum deviations from stoichiometry i n C d T e have been found to be about 1 0 excess T e or C d atoms per c c , and i n P b T e about 4 Χ 1 0 excess P b a n d 1 0 excess T e atoms per cc. Because these deviations from a 1 to 1 stoichiometry are so small (5 X 1 0 ~ or less), they cannot be studied b y chemical analysis, x-ray diffraction, or density measurements. Rather the extent of non­ stoichiometry and the nature of the defects must be inferred from such quantities as obtained from electrical, optical, magnetic, and self-diffusion studies. T o make the connection between the observed electrical, optical, etc., behavior and the type and concentration of point defects i n a solid compound, one must first have some qualitative ideas as to the expected properties of a solid arising from the presence of a particular type of defect. A simple example is discussed. Further details are given b y Kroger and V i n k (20, 22). Secondly, some theory must be available w h i c h relates the concentration of the various defects to the independent intensive variables of the solid phase and the parameters w h i c h char­ acterize the particular compound under study. This theory and some of its ap­ plications are discussed below. C i l

s

2

T e

3

B i

1 7

1 8

1 9

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

4

Properties of Point

Defects

The qualitative properties of a defect such as a sulfur vacancy i n ZnS are fortunately independent of the type of bonding i n the compound. If we consider first that ZnS is an ionic compound composed of Z n + and S ions, the removal of a neutral S atom to the gas phase to form S molecules leaves behind a neutral sulfur vacancy, V ° , since charge neutrality must be preserved i n the crystal. T h e two electrons left behind can be considered as being trapped i n the vicinity of the vacancy and can be removed one at a time into the conduction band of the solid b y thermal ionization. These processes can be written as ordinary chemical equations : 2

-

2

2

s

V° a

V

+

a

V

+

+

B

V

+2

a

e~ — energy e~ — energy

+

It is therefore possible, i n a compound like ZnS, to have not only neutral but also singly and doubly positively charged S vacancies and conduction electrons. The b i n d i n g energy of the electrons to V + and V ° arises, i n the simplest model due to Bethe (3), from the electrostatic attraction between the positively charged defects V + and V + and the negatively charged electrons moving i n a solid characterized b y its static dielectric constant. If we n o w consider the other extreme of covalent bonding, ZnS is made of Z n (. . . 3d 4s4pP) and S + (. . . 3 s 3 p ) , each w i t h the tetrahedral sp* configuration w h i c h gives rise to the 4 covalent bonds of the zinc blende structure ( 9 ) . T h e removal of a neutral S (. . . 3 s 3 p ) to the gas phase removes six of the eight bonding electrons, leav­ ing, as before, two electrons bound to the V ° w h i c h is formed. s

2

s

s

s

-

2

10

2

3

2

4

s

T h e above description of a sulfur vacancy is summarized b y the statement that such a defect may behave as a doubly ionizable donor. Evidence for such In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

172

ADVANCES IN

behavior i n ZnS was observed by adding to the solid a small amount of C u w h i c h acts as a singly ionizable acceptor (2). A transfer of two S atoms from the solid phase to the gas phase (to give S gas), leaves two V ° defects, w h i c h , since there are C u acceptor impurities present, can transfer a total of four electrons to these impurity atoms. The concentration of C u acceptors empty of electrons, as meas­ ured by the intensity of the infrared fluorescence from these defects, should then vary as the +V4 power of the sulfur pressure, when Z n S : C u is equilibrated at some h i g h temperature under various sulfur pressures. The data presented by A p p l e and Prener (2) show this quarter-power dependence measured over four decades of sulfur pressure. In a similar fashion, a C d i i n C d T e can act as a double donor ( C d ^ , C d i + , C d i + ) and a V as a double acceptor (V °, V c d ~ V — ) . The change from η to ρ type conductivity w i t h changes i n stoi­ chiometry has been observed i n the I I - V I compounds (5, 7, 23, 26). However, since both a C d i and V can act as doubly ionizable donors, a decision as to the type of defect present cannot be made on the basis of electrical or optical meas­ urements but must rather be inferred from other evidence, such as self-diffusion data or consideration of atomic sizes. F o r the chalcogenide compounds exhibiting small deviations from stoichiometry, the identification of the predominant struc­ tural defects is by no means certain at the present time and remains as a challeng­ ing problem. N o doubt the application of such techniques as electron paramag­ netic resonance w i l l contribute m u c h toward a solution of this problem. 2

s

2

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

CHEMISTRY SERIES

C d

C d

C d

2

T e

Statistical

Thermodynamics

of Nonstoichiometric

Chalcogenides

It can be shown generally that for a binary compound the number of degrees of freedom are three, independent of the nature and number of defects (21, 27). The concentration of all defects, and therefore the stoichiometry of the compound, are determined when any three independent intensive variables of the solid phase are fixed. Conveniently these are the temperature, pressure, and chemical po­ tential of one of the constituents of the compound. The problem of determining the quantitative relationships between the concentrations of defects (where these are small compared to the number of lattice sites) and the values of the three variables was essentially solved some 30 years ago by Wagner and Schottky (31). Since that time, their work has been extended by including within the framework of the original theory present-day concepts of the donor and acceptor properties of defects and the band theory of solids (6, 28), and by the introduction of graphical methods for obtaining solutions to the resulting equations (22). Using C d T e as an example, we can obtain a set of eight equations from w h i c h the con­ centrations of the defects (in number per cc.) C d ^ , C d i + , C d i + , V ^ , V , and V as w e l l as the free charge carrier electrons, e~, and holes, h+, can be de­ termined as functions of the three independent intensive variables Τ, P, and μ^ . T h e last equation gives the condition for charge neutrality i n the solid. 2

C d

0

C

d

_

- 2

ά

kT In a(V °)

=

Cd

-MCd +

G

MCd +

G

k Τ In e ( C d i ) = 0

kT kT

+

Cd

[In a(V ~*)

+

[ln û ( C d i + ) + kT

=

+

+

[In û ( C d i ) + 2

In a(h )]

2 l n a(h )]

+

ca

kT kT

[In a(V -)

=

-/z d+ - μ

l n a(e~)] =

2 l n a(e~)\ =

[ln a(h+)

+

G

C

+

+

G

+

AKM +

G

+

0

ά

/xcd +

l n a (e~)]

=

G

+

G G

+

G

G G

+

G

C

1

In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

15. PRENER

173

Chalcogenide Systems 2[Cdi+2] +

[Cdi+] +

2[V ~*]

[Λ+] =

+

Cd

[F

C d

-]

+ [e~\

At low defect concentrations: a{V »)

=

Ca

a(e-)

[V °]/N Cd

=

a(Cd°)

2}

[e~]/N

a(h )

=

+

Ci

=

[Cd-^/N,

[h ]/N , +

v

etc.

If / x rather than μ is chosen as the third independent variable, the GibbsD u h e m equation for small defect concentrations gives: αα

Te

/*cd + 1

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

μτβ =

Cs

The various constants, C . . . C , are functions of the temperature and pressure of the solid phase only and contain the various parameters characteristic of CdTe— namely, the thermal band gap, the thermal ionization energies of the donor and acceptor levels, the cohesive energy of C d T e relative to a reference state of i n ­ finite dispersion of C d a n d T e i n their lowest quantum states, and the energies required to remove a C d atom from the solid to the reference state and that required to introduce an interstitial C d atom. They also contain the pressurevolume work terms associated w i t h the various processes and the vibrational entropy terms. Since the pressure-volume terms are very small for condensed phases, the dependence of the C's on pressure can generally be neglected. T h e quantities N , N N , and N are, respectively, the number of lattice sites per cubic centimeter, the number of interstitial sites per cubic centimeter, and the effective density of states near the valence and conduction band edges of C d T e . T h e specific forms of the C's are not given here, since they can be found i n many of the references already cited. I n experimental work it is most convenient to fix the values of μ or / x b y allowing the solid to reach equilibrium w i t h a gaseous phase containing C d or T e at a given partial pressure. The relationships are: i9

s

8

v

c

€ά

T e

kTlnpcd

MCd = M ° ( C d ) + MTe = M ° ( T e ) + 2

kT l n p

T

e

2

m

The /x°'s can be found in standard texts (10). Experimentally, the quantities that can be determined are [e~~] and [h+] as obtained from suitable electrical measurements as the H a l l effect; these measure­ ments are made on crystals w h i c h have been equilibrated at some high tempera­ ture and quenched. Details as to h o w such measurements can lead to estimates of the quantities C and thus to the concentration of defects obtained b y solving graphically the set of equations are given b y de N o b e l (26). T h e stoichiometry of the solid at a given T, ?, and / x is given b y : Cd

*cd =

[Cd]/[Cd] +

[Te]

where [Cd]

= N

[Te]

=

a

-

![7

C d

°] +

[V ~] Cd

+

[V -*]} Cd

+

[Cdi°] +

[Cdi ] + +

[Cdi 2] +

N

B

Stability Limits of a Nonstoichiometric

Compound

The group of nonstoichiometric metallic chalcogenides discussed here have very narrow limits of stability. O n either side of the solid phase region i n the T-x diagram above the eutectic temperatures there coexists a l i q u i d phase ( C d + T e , P b + S, etc.). The limits of stability of the nonstoichiometric compound can be In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

ADVANCES IN CHEMISTRY SERIES

174

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

determined experimentally by the procedure given above, if the solid phase is now equilibrated w i t h the coexisting phase, contact between the two condensed phases being made by way of the vapor phase. Such a three-phase, two-component system has only one degree of freedom, the temperature. The equilibrium con­ centration of defects found corresponds to that at the stability limit of the solid phase at the particular temperature. This is the method used to determine the composition limits of P b T e ( 7 ) . Alternatively, if the liquidus P-T phase diagram is determined, then the concentration of defects at the stability limits of the solid phase can be evaluated from the set of equations given above. This is the method used to determine the composition limits of C d T e (26). Maximum

Melting

Point

T h e available data on a number of chalcogenides show that the maximum melting point (the temperature at w h i c h solid and l i q u i d phases have the same composition) does not occur at the stoichiometric composition (5, 7, 8, 26). Any theoretical analysis of the solid composition at the maximum melting point must involve some model for the l i q u i d phase, since the condition for equilibrium be­ tween two phases is given b y the equality of the chemical potentials of each component i n the two phases. The analysis of Hodgkinson (16), w h i c h was based on the properties of the solid phase alone, cannot therefore be completely valid. In the case of the P b compounds, the defects present i n the solid at high temperatures have been taken to be V , P b i + , e~, and h+. F r o m the set of equations given, we find that the chemical potential of lead at the stoichiometric composition ( x = / ) of the solid as a function of temperature is given by: P

s

P b

1

_

b

2

kT

M

s

= -j-

+

l n (NVNJNM

72 ( G +

C

3

-

C

2

-

C ) 4

F o r the l i q u i d phase the chemical potential of P b can be written as: /z

L

= C (T,P)

+

9

kTln

T h e quantities C and y are properties of the l i q u i d phase alone. W h e n the two phases are i n equilibrium, ^ = μ . Therefore, the composition of the l i q u i d phase can i n principle be found and it is not difficult to see that the value X p = / w o u l d be an exception rather than the rule. T h e temperature at w h i c h the l i q u i d phase is i n equilibrium w i t h the stoichiometric solid may be a great many degrees below the maximum melting point and X p may be very far from V . 9

L

s

Ώ

L

b

L

with Large Stoichiometric

2

2

b

Chalcogenides

1

Deviations

M a n y of the chalcogenide compounds of the transition elements can exist as homogeneous phases over rather large composition ranges. This range usually, but not always, encompasses some simple stoichiometric composition. A s w i t h the compounds discussed above, the composition range is made possible b y the presence of native structural defects built into the crystalline lattice. A theo­ retical analysis of the dependence of the concentration of defects and hence the stoichiometry upon the independent intensive variables must take into account additional energy terms arising out of interaction between the defects. These interactions have been taken into account i n a simple model of a nonstoichiometric solid and are discussed below. F r o m a chemical viewpoint one frequently speaks of the presence of metal In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

15. PRENER

175

Chalcogenide Systems

atoms i n two valence states i n these transition metal chalcogenides. F o r example, a composition such as F e S can be considered to be Fe 7^ Fe S. This ionic picture may be too simple, for i n the N i A s structure of this solid there are short metal-metal distances along the trigonal axis. It is therefore possible that the positive hole instead of being localized about the F e + moves i n a partially filled 3d band arising from the overlap of the 3d orbitals of the metal atoms along the trigonal axis. M o r i n (25) has discussed the electrical properties of the transition metal oxides from this point of view. Unfortunately, as practically nothing is known about the energy band structures or the electrical transport properties of the transition metal chalcogenides, there obviously exists a tremendous gap i n our knowledge of these compounds and of the solid state. O n the other hand, the very fact that we are dealing here w i t h compounds having large single-phase composi­ tion ranges has made it possible to study stoichiometric deviations b y the com­ bined measurements of density a n d lattice parameter. These have resulted i n the determination of the predominant structural defects responsible for the deviations from stoichiometry. Thus, i n F e ^ g S it has been shown that there are metal vacancies i n the N i A s structure rather than excess interstitial sulfur. 0/

0 9

2

+3

0m2

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

2

Statistical

Thermodynamics

with Large Stoichiometric

of

Chalcogenides

Deviations

T h e statistical treatment assumed that the defect concentration was so small that the interaction energy between defects could be neglected. However, w i t h increasing concentrations of defects i n compounds exhibiting large stoichiometric deviations this interaction energy must be taken into account. Anderson (1) used the approximation that the total interaction energy among defects could be represented as a sum of the contributions of those pairs of like defects that occur as nearest neighbors i n a completely random distribution. This assumption of a random distribution of defects cannot, of course, be completely accurate, since the number of nearest neighbor defects i n a given lattice depends on the concen­ tration of defects a n d also on the value of LJkT, where Ε is the interaction energy. T h e approximation is nevertheless sufficiently good at the present time i n view of the scarcity of good experimental data, w i t h w h i c h the theory m a y be tested. W h e n the defect interaction energy is m u c h larger than the thermal energy, it can lead to an ordering of defects into superlattice structures and to the appearance of phases having ordered arrays of defects. Other interactions m a y also become important as the spin-dependent interactions between the d elec­ trons i n the F i _ g S system (24). W e shall not consider these order-disorder phenomena, since they are discussed b y Wadsley (30). Some of the structuralconsequences of ordering are considered below. {

In Anderson's treatment, no account is taken of changes i n the electronic disorder of the compound arising from changes i n the stoichiometry. I n the sense of the notation used previously this is equivalent to considering the presence of only neutral defects. F o r a binary compound exhibiting only F r e n k e l disorder i n the metal lattice, the defects are therefore V ° a n d M ^ , w i t h no defects i n chalcogenide lattice. The presentation given here is equivalent to that of Anderson, since we can write: m

(-ju

m

+

(/*m +

Ci)/kT C )kT 2

= ln

a(V «) m

= l n a(M^)

In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

176

ADVANCES IN CHEMISTRY SERIES

In this case of large stoichiometric deviations and interaction between like pairs of defects, a(V °) is not [ V ° ] / N , but is given b y : m

m

s

a(V °)

= y(V o)[V »]/(N

-

[V °])

a(M-°)

= y(M^)[M^]/{Ni

-

[Mi ])

m

m

m

a

m

0

F o r the assumed random distribution of defects, the activity coefficients, y, are given b y the temperature and concentration dependent expressions : kT In y(V °)

= dE(V °

m

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

kT\ny(M«)

-

m

= dE(M?

-

= Z [F °]e(F °

V °)/d[V °] m

m

s

M-?)/d[M-«]

m

m

= Ζ^Μ^Μ-,

0

-

V^)/N

-

M-°)/Ni

s

where Ε is the total interaction energy per unit volume, e is the interaction per pair, a n d Z , Z are the number of nearest neighbor metal and interstitial sites. F i n a l l y defining 0 = [ V ° ] / N and θ = [M °]/N we get: i

s

V

m

{

s

i

i

These are the results given b y Anderson, w i t h the difference that he used μ as the independent variable and took: χ

Mm + μχ =

Cs(T,P)

As can easily be confirmed b y the application of the G i b b s - D u h e m equation, this again is only approximately true when θ is not negligible w i t h respect to unity. It was shown that for temperatures greater than —Ze/4k (e is negative for attractive forces between defects), there is one value of θ for each value of μ but at lower temperature there are two physically significant values. This repre­ sents physically t w o solid phases i n equilibrium w i t h the defect concentration given b y (θ )ι, (0ι)ι, and ( 0 ) and ( 0 ) , b u t w i t h the same structure and a random distribution of defects. This theory then gives the dependence of the defect concentrations on the intensive variables μ (or μ ), Τ and Ρ i n the non­ stoichiometric single-phase region, and predicts the range of stability of the phase. It also predicts a complete single phase region between two adjacent compounds, A B a n d A B , above a certain critical temperature w h i c h depends on the defect interaction energy. T h e equations were applied b y Anderson to the P t - P t S - P t S system, among others, based on the isothermal pressure measurements of Biltz and Juza (4). I n spite of the approximations of the theory the experimental data are reproduced w i t h critical temperatures of about 690° C . for P t S - P t S a n d 1427° C . for Pt-PtS. Chalcogenide systems are k n o w n i n w h i c h complete misci­ bility has been found experimentally between M X and M X . γ

V

2

4

2

ΤΆ

P

χ

r

2

2

2

Structural Studies of Nonstoichiometric

Chalcogenides

Studies of the structures of the phases appearing i n transition metal chal­ cogenides have been very numerous and cannot all be reviewed here. There are, however, a number of transition metal chalcogenides w i t h compositions near M X w h i c h exhibit, i n certain temperature ranges, the nickel arsenide structure. A m o n g these are F e S ( I I ) , V S e (17), C o T e ( 2 9 ) , N i T e (19), a n d C r S , CrSe, and C r T e (14). O n the other hand, there are a number of compounds w i t h compositions near M X w h i c h crystallize i n the cadmium iodide structure. A m o n g these are V S e (17), C o T e ( 2 9 ) , and N i T e (19). B o t h of these lattices are of 2

2

2

2

In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

15. PRENER

\77

Chalcogenide Systems

the hexagonal class. I n the N i A s structure there are metal atoms at (0, 0, 0 ) , (1, 0, 0 ) , (0, 1, 0 ) , ( 1 , 1, 0 ) , and another similar layer at c — / and 1. T h e two chalcogenide atoms are located at ( / , / , / ) a n d ( V , / , / ) . T h e C d l lattice is derived from the N i A s lattice b y removal of the layer of metal atoms at c — / . I n the nonstoichiometric compounds M _ g X having excess chalcogenide, x-ray and density studies have shown that vacancies appear i n the metal lattice of the N i - A s structure, and i n the nonstoichiometric compounds M X _ g having a metal excess, interstitial metal atoms (at c — / ) appear i n the C d l structure. These systems are therefore structurally equivalent to the case treated above, except that w i t h respect to M X , the metal vacancies i n M X are not randomly distributed over all the metal sites but are confined to every second layer. 1

2

2

3

3

1

4

3

2

1

3

3

4

2

1

2

x

1

2

2

2

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

2

It is possible to pass continuously from the N i A s to the C d l lattice b y first forming metal vacancies i n the M X compound randomly distributed over a l l the metal sites. A s the concentration of these defects becomes larger, the increasing interaction energy leads to an ordering of these defects. They tend to segregate into every second metal layer, leading finally to the C d l lattice at the composition M X ( 2 9 ) . This process allows one to pass from one phase to another of a different structure without a first-order phase transition, a n d is equivalent to a continuous change i n composition above the critical temperature as discussed above. Behavior of this type has been found i n the systems C o T e - C o T e (29) and N i T e - N i T e (19) prepared at elevated temperatures. T h e V S e - V S e (17) system behaves similarly, except that between V S e and V S e an ordering of defects into a phase of lower symmetry occurs. Beyond V S e a further ordering into the C d l lattice sets i n . A n x-ray study of the C r - S system at room temperature shows the type of behavior expected below the critical temperature i n w h i c h small regions of homogeneity exist near ideal stoichiometric compositions w i t h two-phase regions i n between (18). T h e compounds found were C r S , C r S , C r S , C r S , a n d CrS a n d at least three of these ( C r S , C r S , a n d C r S ) exist over a range of composition as homogeneous phases. A l l these compounds (except C r S ) have structures intermediate between those of N i A s and C d l . C r S , having the lowest metal vacancy concentration, has a random distribution of vacancies i n every second layer of the N i A s structure. A t higher temperatures the range of homogeneity about the composition C r S is m u c h broader and the defects apparently become completely disordered over both metal layers of the N i A s lattice (12, 13, 14). I n C r S , C r S , a n d C r S , w h i c h have higher defect concentrations, the lattices are supercells derived from the N i A s lattice b y confining the vacancies to every second metal layer and ordering these within the layers as well. This leads to phases of lower symmetry but still related structurally to the N i A s and C d l lattices. Nonstoichiometry i n these lower symmetry phases arises from adding or removing metal atoms randomly from the ordered array i n every second metal layer; these layers are incompletely filled when referred to the N i A s lattice. These observations a n d many more i n the literature on the range of homogeneity of binary compounds a n d order-disorder phenomena, are at least qualitatively i n accord w i t h the discussion i n the previous section. M o r e quantitative studies are needed on the changes i n the limits of homogeneity w i t h temperature and of the partial pressures of the chalcogenide (and therefore its chemical potential) over the nonstoichiometric solid over its entire range of stability. M o r e data are also needed on the dependence of order-disorder phenomena on the temperature a n d composition of nonstoichiometric chalcogenides. Finally, knowledge of the electronic 2

2

2

2

2

2

1 2

1 6

1 6

2

2

2

3

3

3

4

2

7

5

6

3

4

2

3

7

4

5

6

7

8

8

7

8

8

3

2

In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.

ADVANCES IN CHEMISTRY SERIES

178

structure of these materials must be gained b y studies of various electro-magnetothermo-optical effects upon single crystals of nonstoichiometric chalcogenides.

Downloaded by UNIV OF MICHIGAN ANN ARBOR on August 9, 2013 | http://pubs.acs.org Publication Date: January 1, 1963 | doi: 10.1021/ba-1964-0039.ch015

Literature Cited (1) Anderson, J. S., Proc. Roy. Soc. A185, 69 (1946). (2) Apple, E. F., Prener, J. S., J. Phys. Chem. Solids 13, 81 (1960). (3) Bethe, H., MIT Radiation Laboratory, Rept. 43-12 (1942). (4) Blitz, W., Juza, R., Z. anorg. u. allgem. Chem. 190, 161 (1930). (5) Bloem, J., Kroger, F. Α., Ζ. physik. Chem. (Frankfurt) 7, 1 (1956). (6) Brebrick, R. F.,J.Phys. Chem. Solids 4, 190 (1958). (7) Brebrick, R. F., Allgaier, R. S., J. Chem. Phys. 32, 1826 (1960). (8) Brebrick, R. F., Gubner, E., Ibid., 36, 170 (1962). (9) Coulson, C. Α., "Valence" p. 263, Oxford University Press, London, 1952. (10) Fowler, R. H., Guggenheim, Ε. Α., "Statistical Thermodynamics," p. 170, Cam­ bridge University Press, London, 1939. (11) Haraldsen, Η., Z. anorg. u. allgem. Chem. 231, 78 (1937); 246, 169 (1941). (12) Ibid., 234, 372 (1937). (13) Haraldsen, H., Mehmed, F., Ibid., 239, 369 (1938). (14) Haraldsen, H., Neuber, Α., Ibid., 234, 353 (1937). (15) Harman, T. C., Paris, P., Miller, S. E., Goering, H . L.,J.Phys. Chem. Solids 2, 181 (1957). (16) Hodgkinson, R. J., J. Electronics and Control 1, 612 (1956). (17) Hoschek, E., Klemm, W., Z. anorg. u. allgem. Chem. 242, 49 (1939). (18) Jellinek, F., Acta Cryst. 10, 620 (1957). (19) Klemm, W., Fratini, Ν., Z. anorg. u. allgem. Chem. 251, 222 (1943). (20) Kroger, F. Α., J. Phys. Chem. Solids 7, 277 (1958). (21 ) Kroger, F. Α., Stieltjes, F. H., Vink, H. J., Philips Res. Repts. 14, 557 (1959 ). (22) Kroger, F. Α., Vink, H . J., "Solid State Physics," F. Seitz and R. Turnbull, eds., Vol. 3, pp. 307-435, Academic Press, New York, 1956. (23) Kroger, F. Α., Vink, H. J., van den Boomgaard, J., Z. physik. Chem. 203, 1 (1954). (24) Lotgering, F. K., Philips Res. Repts. 11, 190 (1956). (25) Morin, F. J., "Semiconductors," Ν. B. Hannay, ed., pp. 600-33, Reinhold, New York 1959. (26) Nobel, D. de, Philips Res. Repts. 14, 430 (1959). (27) Prener, J. S., J. Appl. Phys. 33, 434 (1962). (28) Schottky, W., "Halbleiterprobleme," W. Schottky, ed., Vol. 4, pp. 235-81, Fried­ rich Vieweg und Sohn, Braunschweig, Germany, 1958. (29) Tengner, S., Z. anorg. u. allgem. Chem. 239, 127 (1938). (30) Wadsley A. D., ADVAN. C H E M . SER., NO 39, 23 (1962).

(31) Wagner, C., Schottky, W., Z. physik. Chem. B11, 163 (1931). RECEIVED September 6, 1962.

In Nonstoichiometric Compounds; Ward, R.; Advances in Chemistry; American Chemical Society: Washington, DC, 1963.