Journal of ELECTRONIC MATERIALS, Vol. 39, No. 9, 2010
DOI: 10.1007/s11664-010-1260-y 2010 TMS
Structural and Thermoelectric Properties of Chimney–Ladder Compounds in the Ru-Mn-Si System NORIHIKO L. OKAMOTO,1,2 TATSUYA KOYAMA,1 KYOSUKE KISHIDA,1 KATSUSHI TANAKA,1 and HARUYUKI INUI1 1.—Department of Materials Science and Engineering, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan. 2.—e-mail:
[email protected]
We have investigated phase relationships of the sesquisilicide alloys in the Ru-Mn-Si system. A series of chimney–ladder phases Ru1 xMnxSiy (0.14 £ x £ 0.97, 1.584 £ y £ 1.741) are formed over a wide compositional range between Ru2Si3 and Mn4Si7. We also investigated thermoelectric properties of the directionally solidified Ru1 xMnxSiy alloys as a function of Mn content and temperature. The dimensionless figure of merit ZT for alloys with high Mn content (x ‡ 0.75) increases as the Mn content increases. The alloy with x = 0.90 exhibits ZT as high as 0.76 at 874 K. Key words: Transition-metal silicide, transmission electron microscopy (TEM), valence electron count (VEC), thermal conductivity
INTRODUCTION The high-temperature (HT) phase of ruthenium sesquisilicide (Ru2Si3) possesses the tetragonal Ru2Sn3-type structure, which is known as one of the chimney–ladder structures.1 Some of the family of compounds with the chimney–ladder structures, including high manganese silicides (Mn4Si7, Mn11Si19, Mn15Si26, and so forth2–4), are known to exhibit a high Seebeck coefficient and low thermal conductivity. The chimney–ladder compounds have been investigated as candidate thermoelectric materials,5–8 because thermoelectric performance is evaluated by the dimensionless figure of merit ZT = a2T/(qk), where a, q, k, and T are the Seebeck coefficient, electrical resistivity, thermal conductivity, and temperature, respectively. The chimney– ladder compounds expressed by the general chemical formula MnX2n m (M: transition-metal element, X: group 13 or 14 element, n, m: integers) possess a particular tetragonal crystal structure in which the unit cell consists of M (Ru) and X (Si) subcells. The M subcell possesses the atomic arrangement of the b-Sn-type structure (chimney), whereas the X subcell possesses the atomic arrangement of coupled helices (ladder). Both subcells (chimney and ladder) (Received July 9, 2009; accepted April 27, 2010; published online May 20, 2010)
1640
are stacked along the c-axis of the tetragonal unit cell.1,9,10 We recently found that substitution of Ru (group 8) in Ru2Si3 by Re (group 7) stabilizes the HT phase with the chimney–ladder structure to appear at low temperatures, so that a series of chimney–ladder phases Ru1 xRexSiy are formed over a wide composition range between Ru2Si3 and Re4Si7.11,12 It is interesting to note that the Si/M values for these chimney–ladder phases increase with increased Re content, so that the valence electron count (VEC) per M atom in the compounds is maintained (VEC = 14).13 In the present study, we investigated the phase relationships and crystal structures in Mn-substituted ruthenium sesquisilicide alloys by using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). We also investigated the thermoelectric properties of directionally solidified crystals of the chimney–ladder compounds in the Ru-Mn-Si system as a function of Mn content and temperature. EXPERIMENTAL PROCEDURES Polycrystalline specimens with nominal compositions of Ru1 xMnxSi1.5 and Ru1 xMnxSi1.75 (x = 0.20, 0.40, 0.60, and 0.80) were prepared by arcmelting elemental Ru, Mn, and Si under Ar gas flow. Microstructures of the as-grown samples were
Structural and Thermoelectric Properties of Chimney–Ladder Compounds in the Ru-Mn-Si System
1641
examined by using SEM and TEM. The chemical compositions of the phases of interest were estimated by energy-dispersive x-ray spectroscopy (EDS) in both SEM and TEM. After determining the stable composition range for the chimney–ladder phases in the Ru-Mn-Si system, directional solidification was applied for some Ru1 xMnxSiy alloys by the Czochralski method (x = 0.50, 0.55, and 0.75) and by the optical floating zone method (x = 0.85 and 0.90) at a growth rate of 5 mm/h under Ar gas flow. The Seebeck coefficient and electrical resistivity were measured in the temperature range from 323 K to 1023 K by static direct-current (DC) and four-probe methods, respectively. Thermal conductivity was estimated from the values of thermal diffusivity and specific heat measured in the range from room temperature to 1073 K by the laser flash method. RESULTS AND DISCUSSION Phase Equilibria and Structures of the Chimney–Ladder Phases Figure 1a–d shows SEM backscattered electron images (BEIs) of as-arc-melted samples with nominal alloy compositions of Ru1 xMnxSi1.5 (x = 0.20, 0.40, 0.60, and 0.80). Two phases are identified for all samples. The major phase is the (Ru,Mn)Siy sesquisilicide phase, while the secondary phase is the (Ru,Mn)Si monosilicide phase (as indicated by white arrows). The volume fraction of the monosilicide second phase increases as the Mn content increases, indicating that the Si content in the sesquisilicide phase increases as the Mn content increases. The major (Ru,Mn)Siy sesquisilicide phase is always accompanied by contrast variation. EDS analysis in a SEM reveals that the contrast variation corresponds to compositional variation in the (Ru,Mn)Siy sesquisilicide phase, as observed previously for chimney–ladder phases in the RuRe-Si ternary system.11–13 Thus, a range of composition is indicated for the (Ru,Mn)Siy sesquisilicide phase in the ternary phase diagram of Fig. 1e for the Ru1 xMnxSi1.5 alloys. Although the measured compositions for the chimney–ladder and monosilicide phases deviate from those indicated by dashed lines in Fig. 1e because of insufficient accuracy of SEM EDS, the increase in Si content of the sesquisilicide phase has been confirmed by electron diffraction, as described in the next paragraph. Figure 2a shows a bright-field TEM image of a sample with the nominal composition of Ru0.40Mn0.60Si1.75. A grain of the (Ru,Mn)Siy sesquisilicide phase is located next to a grain of the Si phase. TEM EDS analysis reveals that the Mn content in the sesquisilicide phase decreases as the distance from the Si grain increases. The accuracy in the Mn/M ratio is several atomic percent, and the ratio depends on the sample thickness. This effect was mitigated by performing EDS analyses at positions with a similar thickness. Selected-area electron diffraction (SAED) patterns taken from the sesquisilicide phase exhibit
Fig. 1. (a–d) SEM BEIs of as-arcmelted samples with nominal alloy compositions of Ru1 xMnxSi1.5 (x = 0.20, 0.40, 0.60, and 0.80). (e) Chemical compositions measured by SEM EDS for Ru1 xMnxSi1.5 alloys.
characteristics of the chimney–ladder structure consisting of diffraction spots from the M subcell and those from the Si subcell, as shown in Fig. 2b ([120]M incidence when referred to the M subcell). However, the spacing of satellite spots from the Si subcell in the 00l systematic row increases as the Mn content decreases (Fig. 2c-i). This indicates that some different chimney–ladder phases with different Si subcell sizes are formed simultaneously. The Si/M atomic ratios (y) in the corresponding chimney–ladder phases, which were determined from the ratio of the spacing of satellite spots from the Si subcell to that from the M subcell,10 are plotted in Fig. 3a against the Mn content determined by TEM EDS analysis. The values of the Si/M ratio increase as the Mn content increases, and those for most chimney–ladder phases (0.14 £ x £ 0.87) are larger than those expected from the VEC = 14 rule. Chimney–ladder phases are thus confirmed to exist in a wide compositional range of 0.14 £ x £ 0.97 and 1.584 £ y £ 1.741 when expressed in the form Ru1 xMnxSiy.
1642
Okamoto, Koyama, Kishida, Tanaka, and Inui
Fig. 2. (a) Microstructure of the Ru1 xMnxSi1.75 alloy with x = 0.60 and (b) SAED pattern of the [120]M incidence taken from the sesquisilicide phase. (c–i) Variation of the 00l systematic rows of SAED patterns of the [120]M incidence. The Mn/M, cSi/cM, and Si/M ratios are also indicated.
VEC = 14 Rule and Atomic Packing Factor As described in the ‘‘Introduction,’’ chimney– ladder compounds are known to obey the VEC = 14 rule.9,14–16 However, these chimney–ladder compounds in the Ru-Mn-Si system do not obey the VEC = 14 rule precisely, as shown in Fig. 3b, where the VEC values experimentally determined by TEM and EDS are plotted as a function of Mn content. The VEC values for the chimney–ladder phases with compositions close to Mn4Si7 (0.9 £ x £ 0.97) are slightly smaller than 14. This is reasonable, since the chimney–ladder phase in the Mn-Si binary system is reported to be Si deficient so as to form a series of compounds with VEC < 14.3,17 On the other hand, the VEC values for the chimney–ladder compounds with compositions 0.14 £ x £ 0.9 are larger than 14. The larger VEC values of these chimney–ladder compounds indicate that Si atoms exist in their lattices in excess of what is expected from the VEC rule. The amount of excess Si atoms (i.e., the VEC deviation from 14) in (Ru,Mn)Siy compounds increases with Ru content up to x = 0.5 (Fig. 3b). The cell volume of the M subcell vM (= a2M 9 cM) determined by powder x-ray diffraction increases with Ru content also up to x = 0.5 (data not shown). Thus, the excess Si atoms are introduced to maintain the atomic packing (defined as the ratio of the volume of atoms located in the unit metal subcell to the volume of the unit metal subcell) of the chimney–ladder structure, as observed in the Ru-Re-Si system.13 The values of the atomic packing factor calculated for actual compositions of the experimentally observed chimney–ladder phases are plotted in Fig. 3c as a function of Mn content, together with those estimated for the corresponding VEC = 14 compositions. Goldschmidt radii (0.134 nm, 0.126 nm, and 0.117 nm, respectively, for Ru, Mn, and Si) are used in the calculation. Indeed, the values of the atomic
packing factor for the experimentally determined compositions vary with the composition less significantly than those estimated for the corresponding VEC = 14 compositions. This variance indicates that the chimney–ladder phases are stabilized by maintaining the atomic packing factor in a certain range. Upon further alloying of Mn4Si7 with Ru (0.14 £ x £ 0.5), the VEC values of the experimentally observed chimney–ladder phases do not change significantly with Ru content. In the alloy composition range (0.14 £ x £ 0.5), the change in the cell volume of the M subcell is not significant (not shown). This insignificance indicates that the amount of excess Si atoms to be introduced to maintain the atomic packing factor may not be so large in the alloy composition range. Thus, the deviation of VEC values of chimney–ladder compounds from the ideal value (VEC = 14) is in their upper limit; the introduction of excess Si atoms in their lattices may destabilize the chimney–ladder crystal structure. Thermoelectric Properties of Directionally Solidified Alloys We selected five different nominal compositions (Ru1 xMnxSiy: x = 0.50, 0.55, 0.75, 0.85, and 0.90) for crystal growth by directional solidification based on the results described in the previous section. In contrast to the case of chimney–ladder compounds in the Ru-Re-Si system,13 it was impossible to obtain single crystals of chimney–ladder compounds in the Ru-Mn-Si system. However, crystal orientation analysis by x-ray back-reflection Laue method revealed that crystals grow preferentially along the c-axis of the tetragonal chimney–ladder phases. The microstructure observed in longitudinal sections of the grown crystal with x = 0.55 is shown in Fig. 4, in which the growth direction is parallel to the horizontal edge of the figures. All crystals exhibit
Structural and Thermoelectric Properties of Chimney–Ladder Compounds in the Ru-Mn-Si System
1643
Fig. 4. SEM BEI of a directionally solidified alloy with nominal composition Ru0.45Mn0.55Si1.66.
Fig. 3. (a) Si/M ratios, (b) VEC values per M atom, and (c) atomic packing factors for Ru1 xMnxSiy chimney–ladder phases plotted against Mn content.
Ru2Si3 were evaluated along the growth direction. Values of Seebeck coefficient and electrical resistivity for these alloys are plotted in Fig. 5a and b, respectively, as a function of temperature. The alloys with x = 0.50 and 0.55 exhibited negative values in the whole temperature range investigated, whereas those with x = 0.75, 0.85, and 0.90 exhibited positive values. The considerably small values of the Seebeck coefficient for the Ru1 xMnxSiy alloy with x = 0.55 and 0.75 are considered to be due to the canceling out of the contributions from the n-type (Mn-poor) and p-type (Mn-rich) phases. While the values of electrical resistivity for the alloy with x = 0.75 decrease as temperature rises, those for the alloys with x = 0.50, 0.85, and 0.90 increase as temperature rises. Values of thermal conductivity for alloys with x = 0.85 and 0.90 are plotted in Fig. 5c as a function of temperature. The values of thermal conductivity for both alloys increase with temperature. Values of the dimensionless figure of merit (ZT) calculated for alloys with x = 0.85 and 0.90 are plotted in Fig. 5d as a function of temperature, together with similar plots for the Ru0.40Re0.60Si1.663 alloy, which is the best alloy in the Ru-Re-Si ternary system.13 The highest ZT value for the alloy with x = 0.90 is as high as 0.76 at 874 K, which is much higher than that obtained for the Ru0.40Re0.60Si1.663 alloy (ZT = 0.56 at 973 K).13 CONCLUSIONS
virtually a sesquisilicide single-phase microstructure. The sesquisilicide phase in directionally solidified alloys exhibits compositional variation due to the formation of many different chimney–ladder phases, as observed in the as-arcmelted samples (Fig. 1a–d). Thermoelectric properties of directionally solidified Ru1 xMnxSiy alloys as well as single-crystalline
The HT Ru2Si3 phase with the chimney–ladder structure is stabilized to appear at low temperatures by substitution of Ru by Mn to form a series of chimney–ladder compounds Ru1 xMnxSiy over a wide composition range between Ru2Si3 and Mn4Si7. The compositions of these chimney–ladder compounds deviate slightly from the compositional line connecting Ru2Si3 and Mn4Si7, corresponding to the ideal compositional line satisfying the VEC = 14 rule. The occurrence of this compositional
1644
Okamoto, Koyama, Kishida, Tanaka, and Inui
Fig. 5. Temperature dependence of thermoelectric properties of directionally solidified alloys Ru1 xMnxSiy: (a) Seebeck coefficient, (b) electrical resistivity, (c) thermal conductivity, and (d) dimensionless figure of merit (ZT). The highest values of power factor and ZT obtained among the Ru-Re-Si chimney–ladder compounds are also indicated in (d).13
deviation is explained in terms of the valence electron counting rule and atomic packing. The Ru1 xMnxSiy chimney–ladder phases exhibit n- and p-type semiconducting behaviors, respectively, at low and high Mn concentrations, at which the Si/M values are, respectively, larger and smaller than those expected based on the VEC = 14 rule. The values of thermoelectric power factor and ZT for alloys with high Mn content (x > 0.75) increase with Mn content. The ZT value for the alloy with x = 0.90 is as high as 0.76 at 880 K, which is higher than the highest value obtained for the ternary alloys in the Ru-Re-Si system. ACKNOWLEDGMENTS This work was supported by a Grant-in-Aid for Scientific Research (A) (No. 21246101) and Young Scientists (Start-up) (No. 20960049) from the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), Japan, and in part by the Global COE (Center of Excellence) Program of International Center for Integrated Research and Advanced Education in Materials Science from the MEXT, Japan. One of the authors (N.L.O.) expresses his appreciation to Yoshida Foundation for Science and Technology. The authors would like to thank Mr. Koyanagi, Osaka Titanium Technologies Co. Ltd., for supplying titanium sponge as oxygen getter.
REFERENCES 1. D.J. Poutcharovsky, K. Yvon, and E. Parthe, J. LessCommon Met. 40, 139 (1975). 2. U. Gottlieb, A. Sulpice, B. Lambert-Andron, and O. Laborde, J. Alloys Compd. 361, 13 (2003). 3. O. Schwomma, A. Preising, H. Nowotny, and A. Wittmann, Monatsh. Chem. 95, 1527 (1964). 4. H.W. Knott, M.H. Mueller, and L. Heaton, Acta Crystallogr. 23, 549 (1967). 5. B.A. Simkin, Y. Hayashi, and H. Inui, Intermetallics 13, 1225 (2005). 6. L. Ivanenko, A. Filonov, V. Shaposhnikov, G. Behr, D. Souptel, J. Schumann, H. Vinzelberg, A. Plotnikov, and V. Borisenko, Microelectron. Eng. 70, 209 (2003). 7. D. Souptel, G. Behr, L. Ivanenko, H. Vinzelberg, and J. Schumann, J. Cryst. Growth 244, 296 (2002). 8. V.K. Zaitsev, CRC Handbook on Thermoelectrics (New York: CRC, 1994), p. 299. 9. H. Nowotny, The Chemistry of Extended Defects in Non-Metallic Solids (Amsterdam: North-Holland, 1970), p. 223. 10. H.Q. Ye and S. Amelinckx, J. Solid State Chem. 61, 8 (1986). 11. B.A. Simkin, A. Ishida, N.L. Okamoto, K. Kishida, K. Tanaka, and H. Inui, Acta Mater. 54, 2857 (2006). 12. A. Ishida, N.L. Okamoto, K. Kishida, K. Tanaka, and H. Inui, Mater. Res. Soc. Symp. Proc. 980, 235 (2007). 13. K. Kishida, A. Ishida, T. Koyama, S. Harada, N.L. Okamoto, K. Tanaka, and H. Inui, Acta Mater. 57, 2010 (2009). 14. W. Jeitschk and E. Parthe, Acta Crystallogr. 22, 417 (1967). 15. D.C. Fredrickson, S. Lee, R. Hoffmann, and J.H. Lin, Inorg. Chem. 43, 6151 (2004). 16. D.C. Fredrickson, S. Lee, and R. Hoffmann, Inorg. Chem. 43, 6159 (2004). 17. R.D. Ridder, G. Vantendeloo, and S. Amelinckx, Phys. Status Solidi A 33, 383 (1976).
