- Email: [email protected]
Layer composition of bismuth tellurium sulfide
1 Phys Chrm
Sohds. 1975. Vol 36, pp 624-625.
LAYER
Pergamon Press
Printed m Great Bntarn
COMPOSITION
OF BISMUTH
TELLURIUM
SULFIDE*
KARL R. WILHELM and HAROLD D. BALE PhysicsDepartment,Universityof NorthDakota,GrandForks, ND 58202,U.S.A. (Received 16September 1974;in revisedform 23 December 1914)
Bismuth tellurium sulfide (B&Te&) consists of quintuple layers stacked in the rhombohedral[lll] direction [l]. Because of the composition of B&Te,S, at least one layer of the quintuple must contain two types of atoms. The X-ray diffraction results presented in this paper are in agreement with quintuple layers of the form Y(l)-Bi-Y(2)-Y(2)-Bi where the Y(1) layer contains only sulfur and the Y(2) layers each contain tellurium and sulfur in a suitable proportion to satisfy the crystal composition. The paper also points out that the above model for the quintuple layers requires that the sulfur and tellurium atoms making up the Y(2) layer do not form an ordered arrangement but rather are randomly distributed at the Y(2) lattice sites. Large crystals of a single phase for the Bi-Te-S system can only be grown for composition near B&Te&[l]. Crystals of this composition have the same structure as the mineral tetradymite (BitTezS) which exhibits (Rj m) symmetry[2]. For tetradymite with atomic positions S atom at 000 Bi atom at ? (uuu) u = 0,392 Te atom at 2 (uuu) u = 0.788 each layer of the quintuple contains only one type of atom which cannot be the case for B&Te& The question also arises for B&Te& as to whether the sulfur and tellurium atoms form an ordered arrangement or are randomly distributed at the lattice sites in Y(1) or Y(2) layers. Redin has determined that there appears to be only one way of accommodating the composition B&Te,SJ to an ordered arrangement of sulfur and tellurium in Y(1) and Y(2) layers and still preserve the (RJ m) symmetry. For this ordered model the Y(1) layers contain a 3 : 1 ratio of sulfur to tellurium and the Y(2) layers contain a 3 : 1 ratio of tellurium to sulfur. A second distribution which appears plausible is one for which the Y(1) sites contain only sulfur and the Y(2) sites contain the remaining sulfur distributed randomly among the tellurium atoms. This random model is similar to that proposed for the system BizTe3-,Se, which for x = 1 is isomorphic with tetradymite[4,5]. For x > 1 the Y(1) sites contain only *Work supported by National Science Foundation Grant No. GH-34561.Part of the material in this paper is taken from the thesis presented by Karl R. Wilhelm in partial fulfillment of the requirements for the degree of Master of Science at the University of North Dakota. 624
selenium and the Y(2) sites contain the remaining selenium atoms distributed randomly among the Te atoms. This model is supported by the greater affinity of bismuth for selenium than for tellurium. Hence selenium atoms prefer sites in Y( 1) where they can be bound to six nearest bismuth atoms. This argument is also valid if the selenium atoms are replaced by sulfur atoms. X-ray integrated intensity measurements of the (111) family of planes were made using the symmetrical Bragg crystal setting. Data were taken on four diflerent crystals using CuKa radiation and on one crystal using MoKa radiation. Major reliance was placed on the copper radiation because the calculated structure factors for the two models diiered most for the low order retlections where the CuKa radiation gave more reliable results. Eleven orders of Bragg reflections were observed with CuKa radiation and twenty orders with MoKa radiation. A single temperature factor was used to correct the intensity equation for lattice vibrations and Zachariasen’s method]61 was used to correct for extinction. The real and imaginary parts of the atomic scattering factors were included in the calculations of the structure factors. The crystals grown by the Bridgman method usually have a small region at the top (cap) which separates from the crystal proper when the crystal is cleaved. The composition of the cap and the crystal were determined by X-ray fluorescence. Compared to the melt composition (BisTe,&) the cap has a composition rich in tellurium and deficient in sulfur while the crystal proper has a slight deficiency of tellurium and an excess of sulfur with an average composition near B&Te&&. These results are consistent with Soonpaa’s density measurements[7] which showed the cap to have a greater density than the calculated density for B&Te& while the crystal proper has a density lower than the calculated density. An extinction factor and the two position parameters, u and u, were adjusted to yield the best agreement (minimum R value) between experiment and each of the theoretical models. Table 1 shows the minimum R values obtained and the corresponding u and u values for the random model. The minimum R values determined for the Table I
Radlatlon
ILvalue
”
Y
MOK”
0.082
0.3931
0.7870
C”Ktl
0.037
0.3929
0.7865
CuKa
0.040
0.3928
0.7865
CuKu
0.049
0 3931
0.7866
CLlKo
0.047
0.3928
0.7865
Technical Notes
c
12
a-
1. Projection of the electron density difference, Ap, (experimental minus model density) on the rhombohedral [1111direction. The projection distance of 5A corresponds to approximately one-half quintuple layer. An area of 15 mm2is equivalent to one electron. Solid line represents the density difference with respect to the random model and the dashed line is with respect to the ordered model. The positions for the three atomic layers is indicated by vertical line segments. Fig.
ordered model were always considerably larger than for the random model. For the four crystals studied with copper radiations this difference was more than a factor of two. The u and u values shown in Table 1 differ slightly from the values reported for tetradymite[2]. Figure 1 is a projection on the [ll l] direction of the difference between the experimental electron density and the electron density of a proposed model. This difference synthesis avoids the series-termination errors which often result when a Fourier series is utilizedPI. The figure clearly shows the failure of the ordered model in the region of the Y(1) layer. In fact even with respect to the random model there is still a deficiency of 2-3 electrons in this region. At least a portion of this deficiency may be attributed to the six covalent bonds which the atoms in Y(1) form with nearest bismuth atoms. The peak between the Y(1) and Bi layers is evidence of this bonding. Likewise the bonding between the Bi-Y(2) layers contributes to the excess density in the region between
625
these layers. In the region of the Y(2) layer the two curves show opposite deviations and differ by approximately 3.5 electrons which is in agreement with the expected difference of the two models in this region. With respect to the random model the curve shows a deficiency of about one electron ii~the region of Y(2) which is probably due to a shift of electrons to the region between the Bi-Y(2) layers. One would expect the bonding to result in a deficiency of electrons in the immediate region of the Bi layer. The slight excess density observed in this region is probably due to error in the atomic scattering factors or treatment of the data. We have shown that the layer structure of BisTe& is similar to that found in the Bi-Te-Se system. The two systems differ however, in that BizTe3-,Se, forms a continuous range of solid solutions; whereas it appears that crystals of a single phase for the Bi-Te-S system can only be grown for composition near B&Te&[ll. Crystals grown by the Bridgman method from other melt compositions contained X-ray diffraction lines corresponding to two or three phases. For example, crystals grown from a melt composition Bi,Te# gave diffraction lines for BhTej, Bi2Te2Sand BisTeWs. Acknowledgements-We wish to express our sincere appreciation to Professor Henn Soonpaa for supplying the crystals used in this study and for his assistance and advice throughout the investigation, and to Professor Frank Kamer for his assistance with the fluorescence measurements. REFERENW 1. Soonpaa H. H., Technical Report No. 1940,Mechanical div. General Mills, Inc. AD-235354(1960). 2. Harker D., Zeits. Krist. 89, 175 (1934). 3. Redin R. D., Progress Report No. 1, NSF Grant GH-34561 (1972). 4. Drabble J. R. and Goodman C. H. L., J. Phys. Chem. Solids 5, 142 (1958). 5. Wiese J. R. and Muldawer L., J. Phys. Chem. Solids 15, 13 (1960). 6. Zachariasen W. H., Acta Cryst. 23, 558 (1%7). 7. Soonpaa H. H., private communication. 8. Azaroff L. V., Elements of X-ray Crystallography,p. 324. McGraw-Hill, New York (1968).
Sohds. 1975. Vol 36, pp 624-625.
LAYER
Pergamon Press
Printed m Great Bntarn
COMPOSITION
OF BISMUTH
TELLURIUM
SULFIDE*
KARL R. WILHELM and HAROLD D. BALE PhysicsDepartment,Universityof NorthDakota,GrandForks, ND 58202,U.S.A. (Received 16September 1974;in revisedform 23 December 1914)
Bismuth tellurium sulfide (B&Te&) consists of quintuple layers stacked in the rhombohedral[lll] direction [l]. Because of the composition of B&Te,S, at least one layer of the quintuple must contain two types of atoms. The X-ray diffraction results presented in this paper are in agreement with quintuple layers of the form Y(l)-Bi-Y(2)-Y(2)-Bi where the Y(1) layer contains only sulfur and the Y(2) layers each contain tellurium and sulfur in a suitable proportion to satisfy the crystal composition. The paper also points out that the above model for the quintuple layers requires that the sulfur and tellurium atoms making up the Y(2) layer do not form an ordered arrangement but rather are randomly distributed at the Y(2) lattice sites. Large crystals of a single phase for the Bi-Te-S system can only be grown for composition near B&Te&[l]. Crystals of this composition have the same structure as the mineral tetradymite (BitTezS) which exhibits (Rj m) symmetry[2]. For tetradymite with atomic positions S atom at 000 Bi atom at ? (uuu) u = 0,392 Te atom at 2 (uuu) u = 0.788 each layer of the quintuple contains only one type of atom which cannot be the case for B&Te& The question also arises for B&Te& as to whether the sulfur and tellurium atoms form an ordered arrangement or are randomly distributed at the lattice sites in Y(1) or Y(2) layers. Redin has determined that there appears to be only one way of accommodating the composition B&Te,SJ to an ordered arrangement of sulfur and tellurium in Y(1) and Y(2) layers and still preserve the (RJ m) symmetry. For this ordered model the Y(1) layers contain a 3 : 1 ratio of sulfur to tellurium and the Y(2) layers contain a 3 : 1 ratio of tellurium to sulfur. A second distribution which appears plausible is one for which the Y(1) sites contain only sulfur and the Y(2) sites contain the remaining sulfur distributed randomly among the tellurium atoms. This random model is similar to that proposed for the system BizTe3-,Se, which for x = 1 is isomorphic with tetradymite[4,5]. For x > 1 the Y(1) sites contain only *Work supported by National Science Foundation Grant No. GH-34561.Part of the material in this paper is taken from the thesis presented by Karl R. Wilhelm in partial fulfillment of the requirements for the degree of Master of Science at the University of North Dakota. 624
selenium and the Y(2) sites contain the remaining selenium atoms distributed randomly among the Te atoms. This model is supported by the greater affinity of bismuth for selenium than for tellurium. Hence selenium atoms prefer sites in Y( 1) where they can be bound to six nearest bismuth atoms. This argument is also valid if the selenium atoms are replaced by sulfur atoms. X-ray integrated intensity measurements of the (111) family of planes were made using the symmetrical Bragg crystal setting. Data were taken on four diflerent crystals using CuKa radiation and on one crystal using MoKa radiation. Major reliance was placed on the copper radiation because the calculated structure factors for the two models diiered most for the low order retlections where the CuKa radiation gave more reliable results. Eleven orders of Bragg reflections were observed with CuKa radiation and twenty orders with MoKa radiation. A single temperature factor was used to correct the intensity equation for lattice vibrations and Zachariasen’s method]61 was used to correct for extinction. The real and imaginary parts of the atomic scattering factors were included in the calculations of the structure factors. The crystals grown by the Bridgman method usually have a small region at the top (cap) which separates from the crystal proper when the crystal is cleaved. The composition of the cap and the crystal were determined by X-ray fluorescence. Compared to the melt composition (BisTe,&) the cap has a composition rich in tellurium and deficient in sulfur while the crystal proper has a slight deficiency of tellurium and an excess of sulfur with an average composition near B&Te&&. These results are consistent with Soonpaa’s density measurements[7] which showed the cap to have a greater density than the calculated density for B&Te& while the crystal proper has a density lower than the calculated density. An extinction factor and the two position parameters, u and u, were adjusted to yield the best agreement (minimum R value) between experiment and each of the theoretical models. Table 1 shows the minimum R values obtained and the corresponding u and u values for the random model. The minimum R values determined for the Table I
Radlatlon
ILvalue
”
Y
MOK”
0.082
0.3931
0.7870
C”Ktl
0.037
0.3929
0.7865
CuKa
0.040
0.3928
0.7865
CuKu
0.049
0 3931
0.7866
CLlKo
0.047
0.3928
0.7865
Technical Notes
c
12
a-
1. Projection of the electron density difference, Ap, (experimental minus model density) on the rhombohedral [1111direction. The projection distance of 5A corresponds to approximately one-half quintuple layer. An area of 15 mm2is equivalent to one electron. Solid line represents the density difference with respect to the random model and the dashed line is with respect to the ordered model. The positions for the three atomic layers is indicated by vertical line segments. Fig.
ordered model were always considerably larger than for the random model. For the four crystals studied with copper radiations this difference was more than a factor of two. The u and u values shown in Table 1 differ slightly from the values reported for tetradymite[2]. Figure 1 is a projection on the [ll l] direction of the difference between the experimental electron density and the electron density of a proposed model. This difference synthesis avoids the series-termination errors which often result when a Fourier series is utilizedPI. The figure clearly shows the failure of the ordered model in the region of the Y(1) layer. In fact even with respect to the random model there is still a deficiency of 2-3 electrons in this region. At least a portion of this deficiency may be attributed to the six covalent bonds which the atoms in Y(1) form with nearest bismuth atoms. The peak between the Y(1) and Bi layers is evidence of this bonding. Likewise the bonding between the Bi-Y(2) layers contributes to the excess density in the region between
625
these layers. In the region of the Y(2) layer the two curves show opposite deviations and differ by approximately 3.5 electrons which is in agreement with the expected difference of the two models in this region. With respect to the random model the curve shows a deficiency of about one electron ii~the region of Y(2) which is probably due to a shift of electrons to the region between the Bi-Y(2) layers. One would expect the bonding to result in a deficiency of electrons in the immediate region of the Bi layer. The slight excess density observed in this region is probably due to error in the atomic scattering factors or treatment of the data. We have shown that the layer structure of BisTe& is similar to that found in the Bi-Te-Se system. The two systems differ however, in that BizTe3-,Se, forms a continuous range of solid solutions; whereas it appears that crystals of a single phase for the Bi-Te-S system can only be grown for composition near B&Te&[ll. Crystals grown by the Bridgman method from other melt compositions contained X-ray diffraction lines corresponding to two or three phases. For example, crystals grown from a melt composition Bi,Te# gave diffraction lines for BhTej, Bi2Te2Sand BisTeWs. Acknowledgements-We wish to express our sincere appreciation to Professor Henn Soonpaa for supplying the crystals used in this study and for his assistance and advice throughout the investigation, and to Professor Frank Kamer for his assistance with the fluorescence measurements. REFERENW 1. Soonpaa H. H., Technical Report No. 1940,Mechanical div. General Mills, Inc. AD-235354(1960). 2. Harker D., Zeits. Krist. 89, 175 (1934). 3. Redin R. D., Progress Report No. 1, NSF Grant GH-34561 (1972). 4. Drabble J. R. and Goodman C. H. L., J. Phys. Chem. Solids 5, 142 (1958). 5. Wiese J. R. and Muldawer L., J. Phys. Chem. Solids 15, 13 (1960). 6. Zachariasen W. H., Acta Cryst. 23, 558 (1%7). 7. Soonpaa H. H., private communication. 8. Azaroff L. V., Elements of X-ray Crystallography,p. 324. McGraw-Hill, New York (1968).