HAL Id: jpa-00225135
https://hal.archives-ouvertes.fr/jpa-00225135
Submitted on 1 Jan 1985HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
LOCAL PROBES OF MEDIUM RANGE ORDER IN
NETWORK GLASSES
P. Boolchand
To cite this version:
J O U R N A L DE PHYSIQUE
Colloque C8, suppliment au n012, Tome 46, d i c e m b r e 1985 page C8-51
L O C A L PROBES OF M E D I U M RANGE ORDER I N NETWORK GLASSES P . Boolchand
University o f Cincinnati, Cincinnati, Ohio 45221-0011, U.S.A.
R@sumB - Les possibilit6s des spectroscopies Plijssbauer et R m n , corn m6thodes compltinentaires de ll&tude de l'ordre a myenne distance dans les verres, sont discut6es.
Abstract
-
Mossbauer site and Raman bond spectroscopy ascomplementary probes of medium range order in network glasses are discussed.
INTRODUCTION
Experimental approaches to the structure of glasses can be broadly classified into four groups: Diffraction Methods, Vibrational Spectroscopy, Photoelectron Spectroscopy and Hyperfine Interaction Methods. The methods in the first three groups are generally appealing because they can be applied to any glass system of interest. These methods have been widely used and also reviewed in literature from time to time. The last group includes local methods like NMR, NQR and Mossbauer spectroscopy and form the focus of the present paper. It would appear at first glance that local methods are restricted in their application since they require a specific probe atom. This restriction is however less severe than generally realized, since quite a variety of nuclear probes is available in glass forming materials. Such investigations have spanned a wide range of materials including conducting, semiconducting and insul- ating glasses./l/ To attempt to review this subject in this limited space would not be tractable. For the interested reader, we refer to several review articles/l/ where local methods have provided new insights on short range atomic structure in glasses.
Experience has shown that whenever a local method can be used, detailed microscopic information on the local chemistry of probe atoms is accessible from measurements of relaxation times, chemical shifts, electric-field gradient tensors, magnetic hyperfine fields and isomer-shifts. When elements of a glass forming system do not have suitable nuclear probes, one can extend the range of investi- gations, in principle, by isovalent atom substitution. Furthermore, in favorable instances when the nuclear hyperfine structure from chemically inequivalent sites can be completely resolved in the spectra, glass compositional variation of site intensity ratios can serve as a powerful probe of medium ranse order. In this paper we illustrate this important principle by discussing specific examples
on chalcogenide glasses: g-GeSe2, g-GeS2 and g-Gel-xSnxSe2
alloys. Our choice of these glasses 1s based on several factors. Chalcogenide glasses are more amenable to the present investigations
C8-52 J O U R N A L DE PHYSIQUE
than oxide glasses because the backbone networks of the former can be prepared over a broader range of compositions. Secondly, the morphological structure of these prototypical glasses is a subject of profound current interest and debate.
11. MORPHOLOGICAL STRUCTURE OF GeSe2 and GeS2 GLASSES
Since neither Ge nor the chalcogens S and Se offer the prospect of a suitable Mossbauer probe atom, we have used isovalent Sn- and Te-atom substitution to probe the cation and anion chemical order in these prototypical glasses.
Fig. 1 shows l19sn Mossbauer spectra of g-(GeO.ggSnO.O1)xSel-x alloys taken near the stochiometric composition x = 1/3. The spectra show/2/ two well defined and chemically inequivalent Sn sites (A,B) populated at x = 1/3 with the site intensity ratio IB/ (IA+IB) = 0.16 (1) The experiments also show that Sn probe atoms select the tetrahedral A (Ge (Sell2) * ) and non-tetrahedral
B sites (Ge2(Se112) arising from Ge-Ge signatures) in the glasses in a random fashion, i. e., no site preference is displayed by the Sn dopant. Under this circumstance the ratio IB/I provides a direct measure of the non-tetrahedral Ge site fraction, which is defined/2/ as the degree of broken chemical order (DBO) of the net- work. Fig. 1 shows a plot of IB/I(x) deduced from the spectra. One finds from the plot of Fig. 1 that the growth rate of B-sites,
J
- 4 -2 0 2 L
V E L O C I T Y (mm/s)
i.e., d/dx (IB/I) = 32(2) at x = 1/3 to be much too steep to identify the B sites (Ge-Ge bonds) as isolated bonding (point) defects in a chemically ordered continuous random network (COCRN).
The rapid growth of B sites, on statistical ground requires that these sites form part of a cluster (line defect) which is identified as quasi-1 dimensional ethane-like chains. The site intensity ratio I d 1 in the corresponding sulfide glasses has also been studied/3/
(see Fig. 2) and it reveals a higher degree of broken chemical order ( I d 1 = 0.29(2)) in g-GeS2 and this is a point we return to later.
Fig. 2 Site intensity ratios In/I where n=A, B or C and
I = IA+IB+IC obtained from deconvolution of Mossbauer spectra of Sn doped g-GeXS1-, alloys (top) and g-GexSel-, alloys (bottom) plotted as a function of x. The various sites represent Sn replacing Ge sites in 3 distinct types of molecular clusters. See ref./3/ for details.
C8-5 4 JOURNAL DE PHYSIQUE
neighbors, i. e.
,
the chemically ordered sites (A) or replace chalcogen sites that have one Ge and one chalcogen near neighbor, i. e., the chemical1 disordered sites (B).
Upon nuclear trans- mutation (129Tem+1251 because of a change in chemical valence, one-fold coordinated 1291 sites that are o -bonded to either a Ge near neighbor (A site) or to a chalcogen near neighbour (B site) result. Because of the drastically different EFGs one can easily discriminate the two types of sites. Compositional variation of Mossbauer site intensity ratios IB/IA(x) in g-GeSe2-2,Te2, alloys reveals a power-law variation/4/. On statistical grounds this variation requires that these sites occur as part of clusters. Analysis/4/ of these results show that Te exhibits a high preference (factor of 13 or more) to replace the B over the A sites and that the DBO on anion sites in g-GeSe2 of about 0.19 to be quite compat- ible with the one deduced earlier/2/ for cation sites.Fig. 3 Mossbauer emission spectra of 129~em sources doped (<1/2 at. % ) in indicated hosts analyzed using a NaI absorber. A qualitative improvement in the fit to the g-GeSe2 and g-GeS2 spectra result in going from one-site fit to a two- site fit. This figure is taken from ref./4/.
111. CORRELATION OF MOSSBAUER SPECTROSCOPY AND RAMAN SCATTERING AS PROBES OF MEDIUM RANGE ORDER IN NETWORK GLASSES
Raman line/5/ (AIC) with x in g-GeySel-, alloys. These variations, it has been suggested/6/, are signatures for the existence of extended atomic correlations, i.e., clusters in g-GeSe2. In Raman
scattering,. these clusters in previous discussions have been
identified/7/ with 12-membered and more recently with 4-membered rings in a COCRN. .On the other hand, Phillips and co-workers/8,9/ were one of the first to invoke clusters that were only partially chemically ordered to account for the glass forming tendency and the AIC Raman line in g-GeSe2. These clusters, also called 'outrigger raftsf (OR) are visualized as fragments of the layered form of c-GeSe2 @-GeSe2) whose edges are reconstructed to have chemically disordered Se-Se sites (Fig. 4).
The proposal of OR as a cluster in g-GeSe2 provides several attrac- tive features to understand the Mossbauer spectroscopy results. Although the original proposal/8/ of a 2-chain OR leads to a DBO of 0.4, i.e., substantially larger than the observed value of 0.16, we have suggested/2/ that a raft having 3- or 4-fold larger lateral extention (i.e., a 6- or 8-chain raft) will quantitatively explain the observed DBO. Secondly, in the rafts the chemically disordered B sites reside at cluster edges while the chemically ordered A sites reside in cluster interior. This geometry is particularly appealing because it provides a natural way to understand the high site preference/4/ for the oversized chalcogen (Te) impurity to select the B over the A Se sites. When a Te dopant atom replaces a Se B site, it can satisfy its longer covalent bond length requirements by relaxing in the interfacial region (Van der Waals gap between clusters) without accumulation of much local strain. Te replacement at a Se A site on the other hand, can only occur at the expense of substantial strain energy as the network is locally stressed. Looked at this way, the high Te site preference energy for the disordered sites can be visualized as surface segregation of the impurity at internal surfaces of the clusters.
Fig. 4 Elements of medium range order in g-GeSe2 consist of a
molecular fragment of 6 -GeSe2 which has been reconstructed at its edges to have chemically disordered Se-Se bonds. The cluster shown consists of 2 corner sharing chains while the cluster proposed for g-GeSe2 has 8 corner sharing chains, i.e., has a 4-fold larger lateral width.
C8-56 JOURNAL DE PHYSIQUE
0
scale (40 to 80 A) into large chalcogen-rich clusters and narrow Ge-rich chains. Because the DBO in g-GeS2 is substantially larger than in g-GeSe2, we have suggested/3/ that the size of chalcogen- rich cluster in the S-containing glass is approximately a factor of two smaller.
Murase and coworkers/lO/ were one of the first to recognize signa- ture of chemically disordered bonds (S-S) in Raman scattering. Specifically, their vibrational density of states calculations showed that the 440 cm-l mode in g-GeS2 can be identified with S-S stretch of S-dimers present at the edges of ORs. These calculations also demonstrated that corresponding Se-Se stretch mode in g-GeSe2 occurs at 247 cm-l and is much weaker in strength. The higher broken bond chemical order inferred from Raman scattering lead Murase and coworkers/lO/ to suspect that the size of clusters in g-GeS2 is smaller than in g-GeSe2. The Mossbauer results confirm and extend these ideas on molecular phase separation by actually identifying the size of chalcogen-rich clusters in these stoichio- metric glasses.
IV. STRUCTURAL PHASE TRANSITION IN NETWORK GLASSES
The concept of chemically ordered continuous random networks as originally described by Zachariasen/ll/ in 1932, has been widely
Fig. 5 Mossbauer tetrahedral Sn percentage T in g-Gel-xSnxSe2 (filled circles) and g-Gel-xSnxS2 (open circles) alloys plotted as a function of x. Note that T = 100(1-Ig/I). The T(x) reveal a 2-peak structure whose origin is discussed in ref./l4/. Note that T I 1 at x = 0.35 in the
used as the most natural description of diffraction/l2/ and vibra- tional spectroscopy experiments/l3/ on stochiometric glasses. Our Mossbauer experiments on g-GeSe2 and g-GeS2 discussed here clearly show that this is actually not the case. Stimulated by the dis- covery that g-GeSe2 network is intrinsically heterogeneous, we considered the possibility of fusing the characteristic A and B molecular clusters to compact the network and realize a completely polymerized network. This has been successfully demonstrated in three separate experiments now showing that the Zachariasen glass can actually be realized in the laboratory in special cases.
Mossbauer spectra of g - G e ~ - ~ S n ~ S e ~ alloys studied/l4/ as a function of x reveal that the DBO appears to nearly vanish at x = Xmax = 0.35
(Fig. 5). At this composition all Sn sites are present only in Sn(S1/2)4 local tetrahedral units. Note in Fig. 5 that the tetrahe- dral Sn fraction T = 1-Id1 1 at x = 0.35. The ternary glass of this particular composition can be broadly described as a COCRN of Ge(Sel/2)4 and Sn(Sel/2)4 tetrahedral units in the approximate ratio
- a ,,- 7
A
RAMAN SHIFT(cm-I)
Fig. 6 Raman scattering in g-Gel-,SnXSe2 alloys showing extinc- tion of mode scattering strength with x approaching 0.35 (on the left). The normalized scattering strength of the AIC mode is plotted as a function of x on the right. See ref./l4/ for additional details.
C8-58 JOURNAL DE PHYSiQUE
collaborators/l6/ have succeeded in reversibly pumping the chemical order of melt-quenched GeSe2 glass optically by absorption in the Tauc edge of the glass. The optically pumped glass network which is characterized by the lack of Ge-Ge and Se-Se bond signatures in Raman scattering, represents a third example of a network that is chemically ordered.
The theoretical justification for the structural phase transition in g-Gel-xSnxSe2 alloys has been given in terms of topological principles using the idea of force-field constraints advanced by Phillips/9,17/. It is well known that although Si and Ge prefer to be fourfold coordinated, the energy difference between fourfold and sixfold coordination for the case of the heavier cation Sn is small at temperatures above room temperature where the white (metallic) Sn phase is stable. This indicates that bond-bending forces for Sn are weak. One can visualize alloying of GeSe2 with SnSe2 as a process that lowers the average number of constraints per atom in the overconstrained GeSe2-glass network.
Following ref./l7/, one may enumerate the number of constraints Nc per Gep,xSnx (S or Se)2 formula unit as follows:
Nc = x m(Sn)/2
+
(1-x)F(Ge)+
2F(S or Se)where 2F (Z) = m(Z)
+
(Nd-1) (2m(Z)-
Ng),
m represents the coordin- ation number and Z = S,Se,Ge or Sn. Slnce Nd = 3 and there are NA =3 atoms per formula unit of Gel-,SnX, (S or Se)2 the Phillips glass condition/9/ for a strain free elastlc network tb form requires
At x = 2/5 the constraint theory thus shows that the number of con- straints per atom in the present ternary glass equals three, the dimensionality Nd of the space in which the network is embedded. The process of cation-tuning of force-field constraints by alloying thus provides a good quantitative basis to understand the structural phase transformation observed in the present ternary. This trans- formation is only partial in character for the corresponding S glasses and this is discussed elsewhere/l8/ in some detail.
V. CONCLUDING REMARKS
Mossbauer experiments designed to probe both the cation and anion sites in chalcogenide glasses reveal evidence for substantial broken chemical order in even stoichiometric glasses like GeSe2 and GeS2. Such a result is understood in terms of mol- ecular phase separation of glass network into characteristic Ge-rich and chalcogen-rich clusters, i.e., a molecular cluster network (MCN). Such a model is quite compatible with vibrational Raman spectroscopy which also provides evidence for broken bond
the cluster surfaces fusing progressively. At x = 0.35, a COCRN characterized by little or no broken bond (site) chemical order is obtained. This observation may well represent one of the first realisation of a Zachariasen glass stable at ambient pressure and temperature. The theoretical basis of this structural phase transition from a MCN to a COCRN is understood in terms of topologi- cal principles using the idea of force-field constraints in elastic networks. According to this theory the phase transition is pre- dicted to occur as a function of composition at x = 0.4 and this is in remarkable accord with the experimental results.
VI
.
ACKNOWLEDGEMENTSI have benefitted immensely from conversations with Dr. J. C. Phillips, Dr. John P. deNeufville, Dr. J. G. Hernandez and Profes- sors P. Suranyi and P. Esposito. It is also a pleasure to acknow- ledge contributions of several individuals including Wayne Bresser, Ray Enzweiler, George H. Lemon, Mark Stevens, Dave Ruffolo, and Jeff Grothaus who have actively participated in some of the experiments described here. This work was supported by the National Science Foundation under grant DMR-82-17514.
REFERENCES
/1/ Several applications of local methods as structural probes of non-crystalline solids are discussed in Proceedinas of the International Conference on Amorphous systems investiaated by Nuclear Methods, Balatonfured, Hungary September 1981, pub- lished in Nucl. Instru. and Methods, 199. (1982).
Recent review articles focussing specifi,cally on NMR and Mossbauer spectroscopy of glasses are as follows:
W. Muller-Warmuth and H. Eckert, Physics Reports, 88, (1982), 91-149.
P. P. Seregin, A. R. Regel, A. A. Andreev and F. S. Nasredinov Phys. Status Solidi A, 2, (1982) 373.
P. Boolchand in Physical Properties of Amorphous Systems, Ed. David Adler, Brian B. Schwartz and Martin C. Steele Plenum, 1985, 221-260.
U. Gonser and R. Preston in Topics in Applied Physics, Vol. 53 Glassv Metals 11, Ed. H. Beck and H. J. Gintherodt, Springer- Verlag 1985, p. 93-126.
Several experiments utilizing local methods to probe glass structures not discussed above include:
B. Lamotte Phys. Rev. Lett. 53 (1984) 576; also see ibid 54, (1985) 1205 who discusses evidence for ortho-H2 in a-Si-H alloys using Magic angle p-NMR;
J. Szeftel Phil. Mag. B Q , (1981) 549 who discusses the structure of glasses from measurement of the asymmetry para- meter of 7 5 ~ s EFG in As2X3 glasses (X = S, Se and Te) using Zeeman perturbation of NQR;
C8-60 JOURNAL DE PHYSIQUE
Francis Inc. 1983, p. 234, who have discussed structure of amorphous FexSnl,x alloy films by magnetization and Mossbauer spectroscopy.
/2/ P. Boolchand, J. Grothaus, W. J. Bresser and P. Suranyi, Phys. Rev. B25. (1982) 2975; P. Boolchand and M. S. Stevens, Phys. Rev. B a (1984) 1.
/3/ P. Boolchand, J. Grothaus, and J. C. Phillips Solid State Comm. 45 (1983) 183; also see P. Boolchand and J. Grothaus in Proceedinss of the 17th International Conference on the Physics of Semiconductors, Ed. J. D. Chadi and W. A. Harrison, Springer-Verlag, New York Inc. 1985, p. 833.
/4/ W. J. Bresser, P. Boolchand, P. Suranyi and J. P. deNeufville, Phys. Rev. Lett. 46 (1981) 1689; also see P. Boolchand, W. J. Bresser, P. Suranyi and J. P. deNeufville, Nucl. Instru. and Methods 199, (1982) 295; also see P. Boolchand in ref. 1. /5/ P. Tronc, M. Bensoussan, A. Brenac, and C. Sebenne, Phys. Rev.
B8, (1973) 5947.
/6/ R. J. Nemanich, S. A. Solin, and G. Lucovsky, Solid State Commun. 21 (1977) 73.
/7/ J. E. Griffiths, J. C. Phillips, G. P. Espinosa, J. P. Remeika and P. M. Bridenbaugh, Phys. Status Solidi (b) 122 (1984) K11; also see R. J. Nemanich, G. A. N. Connell, T. M. Hayes, R. A . Street, Phys. Rev. B u (1978) 6900.
/ 8 / P. M. Bridenbaugh, G. P. Espinosa, J. E. Griffiths, J. C. Phil- lips and J. P. Remeika, Phys. Rev. B a (1979) 4140; also see
K. Murase, T. Fukunaga, K. Yakushiji, T. Yoshimi and I. Yunoki,
J. Non-Cryst. Solids 59 and 60 (1983) 883.
/9/ J. C. Phillips, J. Non-Cryst. Solids 34 (1979) 153 and also ibid 43 (1981) 37.
/lo/ K. Murase, T. Fukunaga, Y. Tanaka, K. Yakushiji and I. Yunoki, Physica 117B and 118B (1983) 962.
/11/ W. H. Zachariasen, J. Am. Chem. Soc. 54 (1932) 3841.
/12/ A. C. Wright and A. J. Leadbetter, Phys. and Chem. of Glasses 17 (1976) 122.
-
/13/ G. Lucovsky, R. J. Nemanich, F. L. Galeener in Proceedinss of the 7th International Conference on Amorphous and Liquid Semiconductors, Edinburgh, Scottland 1977 Ed. W. E. Spear
(G. G. Stevenson, Dundee Scottland), p. 125.
/14/ Mark Stevens, P. Boolchand, J. G. Hernandez, Phys. Rev. B3l, (1985) 981; and also Solid State Commun. 47 (1983) 199.
/15/ K. Murase and T. Fukunaga in A.I.P. Conference Proceedings 120 (1984) 449 Ed. P. C. Taylor and S. G. Bishop, American Insti- tute of Physics, NY.
/16/ J. E. Griffiths, G. P. Espinosa, J. P. Remeika and J. C. Phil- lips, Solid State Commun. 40, (1981) 1077 and Phys. Rev. B a
/17/ J. C. P h i l l i p s Solid State Comm. 47 (1983) 203; also see G. H. Dohler, R. Dandoloff and H. Bilz, J. Non-Cryst. Solids 42
(1980) 87; and J. C. Phillip and M. F. Thorpe Solid State Commun. 53 (1985) 699.
