NASA Aeronautics Research Institute · 11/12/2014 · Thermal expansion measurements were conducted on hot-pressed CrSi 2, TiSi 2, WSi 2 and a two-phase Cr-Mo-Si intermetallic alloy

1 23

Journal of Materials Engineering andPerformance ISSN 1059-9495Volume 24Number 3 J. of Materi Eng and Perform (2015)24:1199-1205DOI 10.1007/s11665-015-1390-8

Comparison of the Thermal ExpansionBehavior of Several Intermetallic SilicideAlloys Between 293 and 1523 K

S. V. Raj

1 23

Your article is protected by copyright and

all rights are held exclusively by ASM

International. This e-offprint is for personal

use only and shall not be self-archived in

electronic repositories. If you wish to self-

archive your article, please use the accepted

manuscript version for posting on your own

website. You may further deposit the accepted

manuscript version in any repository,

provided it is only made publicly available 12

months after official publication or later and

provided acknowledgement is given to the

original source of publication and a link is

inserted to the published article on Springer's

website. The link must be accompanied by

the following text: "The final publication is

available at link.springer.com”.

Comparison of the Thermal Expansion Behaviorof Several Intermetallic Silicide Alloys Between 293

and 1523 KS.V. Raj

(Submitted November 12, 2014; in revised form December 19, 2014; published online January 21, 2015)

Thermal expansion measurements were conducted on hot-pressed CrSi2, TiSi2, WSi2 and a two-phaseCr-Mo-Si intermetallic alloy between 303 and 1523 K during three heat-cool cycles. The corrected thermalexpansion, (DL/L0)thermal, varied with the absolute temperature, T, as

DL=L0ð Þthermal¼ A T � 293ð Þ3þB T � 293ð Þ2þC T � 293ð Þ þ D

where, A, B, C, and D are regression constants. Excellent reproducibility was observed for most ofthe materials after the first heat-up cycle. In some cases, the data from first heat-up cycle deviatedfrom those determined in the subsequent cycles. This deviation was attributed to the presence ofresidual stresses developed during processing, which are relieved after the first heat-up cycle.

Keywords CTE, disilicides, intermetallic alloys, thermal expan-sion

1. Introduction

Intermetallic silicides have found applications as ohmiccontacts in the semiconductor industry (Ref 1–5), thermoelec-tric materials (Ref 6), heating elements (Ref 7, 8), protectivecoatings (Ref 9), and they have been proposed for structuralapplications (Ref 10). A knowledge of the thermal expansionbehavior of these silicides is important in these applications inorder to reduce or eliminate thermal stresses between theintermetallic silicide and the substrate. Although a compilationof the thermal expansion data on many intermetallic silicides isavailable (Ref 11), the data are fairly old and it is unclearwhether they were influenced by impurities and processingmethods. Verkhorobin and Matyushenko (Ref 12) reportedlimited data on the coefficients of thermal expansion (CTE) ofCrSi2, MoSi2, NbSi2, TaSi2, and VSi2 determined from x-raydata. A more complete thermal expansion data on transitionmetal silicides also determined by x-ray diffraction (XRD) havebeen reported by Engstrom and Lonnberg (Ref 13). In contrast,there are only limited thermal expansion data on hot-pressedsilicides generated by dilatometric measurements.

The present investigation was undertaken to determine andcompare the thermal expansion behavior of hot-pressed CrSi2,

TiSi2, WSi2, and a two-phase Cr-30(at.%)Mo-30%Si* inter-metallic alloy between 303 and 1523 K by dilatometricmeasurements. It is noted that there is no previous CTE datafor the Cr-30Mo-30%Si alloy.

2. Experimental Procedures

Commercially produced powders (�325 mesh) of CrSi2,TiSi2,WSi2, and aCr-30%Mo-30%Si alloywere hot-pressed into25.4 mm long and 9.5 mm in diameter cylindrical specimens.The Cr-30%Mo-30%Si alloy was procured from ATI PowderMetals, Pittsburgh, PA as gas atomized powder. Table 1 gives thetotal purity and major impurity content of the powders. Thepowders were hot-pressed using conditions given in Table 2.Three CrSi2 specimens were fabricated in separate hot-press runs(582, 619, and 620) in order to evaluate the effect of batch-to-batch variabilities on the thermal expansion data. The two facesof hot-pressed specimensweremachined to ensure that theywereflat and parallel. The thermal expansion measurements wereconducted using a NETZSCH Dilatometer Model DIL 402Cequipped with a high purity alumina as a calibration standard.Measurements were made over three heat-cool cycles to (a)minimize the effects of compositional, microstructural andprocessing variables on the data, (b) determine the extent ofscatter in the data, and (c) to evaluate a statistical average of thecoefficients for the regression curve. The specimen was placed ina sample holder and aligned with a single push-rod with anapplied constant load of 0.2 N. The specimens were heated from303 to 1523 K at 10 K/min in the first cycle and cooled to 373 Kat 10 K/min. in the first cool-down cycle. Subsequent cyclesconsisted of heating and cooling between 373 and 1523 K at10 K/min. All measurements were conducted in aHe atmosphere

S.V. Raj, NASA Glenn Research Center, MS 106-5, 21000 BrookparkRoad, Cleveland, OH 44135. Contact e-mail: [email protected].

*Unless specifically stated, all compositions are reported in at.% in thispaper.

JMEPEG (2015) 24:1199–1205 �ASM InternationalDOI: 10.1007/s11665-015-1390-8 1059-9495/$19.00

Journal of Materials Engineering and Performance Volume 24(3) March 2015—1199

Author's personal copy

Table 1 Chemical compositions of the silicide powders in wt%

Powder Al Cr Fe Mo Ni Si Ti W Total purity

CrSi2 (Alfa Aesar) 0.05 46.5 0.38 … 0.13 52.4 0.33 … 98.9Cr-Mo-Si (ATI) … 35.2 0.038 50 0.015 14.7 … 0.06 99.9TiSi2 (Cerac) … … … … … … … … 99.5WSi2 (Materion) 0.005 0.001 0.005 … … 22.9 … 76.6 99.5

Table 2 Processing data for hot-pressing the silicide powders

Silicide powder Run # T, K P, MPa Time, h Environment

CrSi2 582 1183 68.9 2.0 ArCrSi2 619 1523 89.6 0.25 ArCrSi2 620 1523 89.6 2.0 ArCr-Mo-Si 621 1723 89.6 4.0 ArTiSi2 641 1523 89.6 0.67 ArWSi2 622 1573 89.6 2.0 Ar

Fig. 1 (a-f) Low and high magnification optical micrographs of the transverse cross-sections of hot-pressed CrSi2 specimens fabricated in threeruns

1200—Volume 24(3) March 2015 Journal of Materials Engineering and Performance


flowing at 60 cm3/min. The length changes were recorded by acomputerized data acquisition system. The experimental strain,DL/L0, where DL is the differential change in length, L� L0, L isthe instantaneous length, and L0 is the original length of thespecimen at room temperature, were measured.

3. Results and Discussion

3.1 Microstructures

Microstructural observations of the hot-pressed specimensrevealed that most of them were well consolidated although the

extent of homogeneity varied from specimen to specimen.Figure 1(a-f) show optical micrographs of the transverse cross-sections of three hot-pressed batches of CrSi2. While themicrostructures for specimens from hot-pressing runs 619 and620 show a great degree of homogeneity and a distribution of finegrain boundary porosity, themicrostructure for run 582 shows theboundaries of the powder particles and extensive porosity. It isnoted that run 582 was hot pressed at 1183 K while runs 619 and620 were hot pressed at 1523 K. The optical microstructures ofthe hot-pressed Cr-30Mo-30Si alloy were homogeneous andfully consolidated with little porosity (Fig. 2a-b). In contrast, thehot-pressed microstructures of the TiSi2 (Fig. 3a-b) and WSi2(Fig. 4a-b) showed a greater amount of grain boundary porosity.

Fig. 2 (a-b) Low and high magnification optical micrographs of the transverse cross-sections of a hot-pressed Cr-30%Mo-30%Si specimen

Fig. 3 (a-b) Low and high magnification optical micrographs of the transverse cross-sections of a hot-pressed TiSi2 specimen

Fig. 4 (a-b) Low and high magnification optical micrographs of the transverse cross-sections of a hot-pressed WSi2 specimen



3.2 Thermal Expansion of CrSi2

Figure 5(a-c) show the variation of the thermal expansion,(DL/L0), with increasing absolute temperature, T, for the hot-pressed three specimens from runs 582, 619, and 620. On closeexamination of the results, it is evident that DL/L0 for the first

heat-up run is higher than the values for subsequent cool-downand heat-up runs for all three specimens with the differencebeing more pronounced for run 582. This discrepancy in thevalues of DL/L0 between the first and other cool-down and heat-up cycles has been attributed to the relieving of residual stressesdeveloped in the specimens during hot-pressing of the powdersduring the first heat-up cycle (Ref 14). The measurements fromthe first cool-down to the third cool-down cycles were fairlyreproducible for runs 619 and 620 but a little more scattered forrun 582 presumably due to the extensive porosity andheterogeneous microstructure (Fig. 1a and b). Thus, neglectingthe data from the first heat-up cycle, the data from the first cool-down to the third cool-down cycles were fitted with Eq. 1:

DL=L0ð Þthermal¼ A T � 293ð Þ3þB T � 293ð Þ2þC T � 293ð Þ þ D

ðEq 1Þ

where (DL/L0)thermal is the magnitude of the DL/L0 withoutany residual processing strains, A, B, C, and D are regressionconstants. Table 3 compiles the values of the regression con-stants and the corresponding coefficients of determination, Rd

2,for the three CrSi2 specimens. Figure 6 compares the regres-sion plots for the three specimens, where it is seen that thecurves are reasonably close thereby suggesting that batch-to-batch variability is relatively small. Unfortunately, a sparsityof published data on polycrystalline CrSi2 did not permit ameaningful comparison with the present results.

3.3 Thermal Expansion of Cr-30%Mo-30%Si

Figure 7 shows the variation of DL/L0 with increasing T forhot-pressed Cr-30%Mo-30%Si. Unlike the observations onCrSi2, the curves almost overlap for all thermal cyclesincluding the first heat-up cycle. This observation suggeststhat the amount of residual strains developed during hot-pressing of this specimen was negligible. The constants givenin Eq. 1 were determined by regression analysis of all the dataincluding those measured during the first heat-up cycle. Thevalues of these constants are shown in Table 4. Since the Cr-30%Mo-30%Si alloy was prepared by substituting Mo for Crand Si in Cr3Si to improve its oxidation resistance (Ref 15),published data for Cr3Si are shown in Fig. 7 for comparison(Ref 11). The two sets of data increasingly deviate from eachother with increasing temperature, where the deviation isrelatively small. It is unlikely that this observed deviation in thethermal expansion can be attributed to the addition of Mo sincethe microstructure of the Cr-30%Mo-30%Si alloy consists of(Cr,Mo)3Si and (Cr,Mo)5Si3 (Ref 15). Therefore, this deviationis attributed to experimental scatter.

3.4 Thermal Expansion of TiSi2

The cyclic thermal expansion characteristics of hot-pressedTiSi2 are similar to the Cr-30%Mo-30%Si alloy (Fig. 8). Thecurves from the first heat-up to the third cool-down cycle are inexcellent agreement with negligible scatter in the data. Once

Fig. 5 Temperature dependence of the thermal expansion behaviorof CrSi2 during three heat up-cool down cycles between 303 and1523 K. The data are shown for three hot-pressed specimens fromruns (a) 582; (b) 619; and (c) 620

Table 3 Values of the regression constants describing the thermal expansion of CrSi2 between 293 and 1523 K

Hot press run no. Cycle description A, %K23 B, %K22 C, %K D, % Rd2

582 1st cool-down to 3rd cool-down 2.09 10�10 1.29 10�07 1.09 10�03 �1.89 10�01 0.993619 1st cool-down to 3rd cool-down 4.29 10�10 �1.49 10�07 1.1910�03 �1.49 10�01 0.999620 1st cool-down to 3rd cool-down 4.29 10�10 �1.89 10�07 1.1910�03 �9.39 10�02 0.999



again, it can be concluded that the residual stresses werenegligible during hot-pressing of the specimens. Table 5 givesthe values the regression coefficients from the first heat-up tothe third cool-down cycle. As shown in Fig. 8, the regressionline describes the experimental data very well. Comparison ofthe present results with literature data on TiSi2 (Ref 11) revealsthat the two sets of data are in reasonable agreement althoughthey increasingly deviate with increasing temperature.

3.5 Thermal Expansion of WSi2

In contrast to the thermal expansion behavior of theCr-30%Mo-30%Si alloy (Fig. 7) and TiSi2 (Fig. 8), the behav-ior of WSi2 is similar to CrSi2 in that the magnitudes of DL/L0are larger in the first heat-up cycle than those in the subsequentthermal cycles (Fig. 9). However, the difference is smaller thanthe observations on CrSi2 (Fig. 5a-c). The regression coeffi-

cients for WSi2 are shown in Table 6, where the data from thefirst heat-up cycle were not included in the regression analyses.Figure 9 compares the present experimental results with therecommended values for polycrystalline WSi2 based oncalculations from ‘‘selected axial thermal expansion values’’(Ref 11). Although the latter results are slightly larger than thecurrent experimental regression curve, the two curves are inreasonable agreement with the limits of experimental scatter.

3.6 Comparison of the Thermal Expansion Behaviorof Different Silicides

Metal silicides are used in the semiconductor industry ascontact materials (Ref 1–5) as well as being considered forother potential applications, such as oxidation-resistant coatings(Ref 9) and matrix constituents in SiC fiber-reinforced com-posite materials (Ref 10, 16–19). Therefore, it is useful tocompare the present results with thermal expansion datareported in the literature for MoSi2 (Ref 11), polycrystallineSi (Ref 20), SiC (Ref 11), and Si3N4 (Ref 11) (Fig. 10). Anexamination of Fig. 10 reveals that the magnitudes of DL/L0 forthe metal silicides are higher than those for Si, SiC and Si3N4

with the deviation from the latter set of data increasing withincreasing temperature. The thermal expansion values aresimilar for both Si and SiC up to 1000 K but larger than thosefor Si3N4. While using metal silicide contacts on Si or SiC-based semiconductors are unlikely to debond from the substrateif they operate close to room temperature, Fig. 10 suggests thatdebonding is increasing likely at higher operating temperatures.Similarly, SiC fiber-reinforced silicide matrix composites arelikely to develop large thermal strains due to the significantdifferences in the magnitudes of DL/L0 between the silicidesand SiC. Nevertheless, it has been successfully demonstratedthat mixing adequate amounts of Si3N4 with MoSi2 to form 0/90 MoSi2-Si3N4/SiC(f) laminated composites lasted over 1000

Fig. 6 Comparison of the regression equations for the three CrSi2specimens between 303 and 1523 K

Fig. 7 Temperature dependence of the thermal expansion behaviorof Cr-30%Mo-30%Si during three heat up-cool down cycles between303 and 1523 K. The current results are compared with data forCr3Si (Ref 11)

Table 4 Values of the regression constants describing the thermal expansion of Cr-30%Mo-30%Si between 293 and 1523 K


621 1st heat-up to 3rd cool-down 1.39 10�10 �1.79 10�08 9.29 10�4 �2.59 10�2 0.9993

Fig. 8 Temperature dependence of the thermal expansion behaviorof TiSi2 during three heat up-cool down cycles between 303 and1523 K. The data reported by Touloukian et al. (Ref 11) are shownin the figure



thermal cycles at 773 K (Ref 18). This concept will be furtherdeveloped and investigated in a future paper.

4. Summary and Conclusions

Thermal expansion measurements were conducted on hot-pressed CrSi2, TiSi2, WSi2, and a two-phase Cr-Mo-Siintermetallic alloy between 303 and 1523 K during threeheat-cool cycles. The optical microstructures of the cross-sections of the hot-pressed specimens revealed that they werewell consolidated except in the case of one CrSi2 specimen,where there was a large amount of grain boundary porosity.Batch-to-batch comparisons with measurements made on twoother CrSi2 specimens revealed that the values of DL/L0 for thisspecimen were slightly lower. Unlike for the other materialsinvestigated in the present research, the first heat-up cycle datafor the CrSi2 and WSi2 deviated from those determined in thesubsequent cycles due to the presence of residual stressesdeveloped during processing. Once these stresses were relievedafter the first heat-up cycle, the data generated in the secondand third heat-up and cool-down cycles were in excellentagreement. The corrected thermal expansion, (DL/L0)thermal,varied with the absolute temperature, T, as

DL=L0ð Þthermal¼ A T � 293ð Þ3þB T � 293ð Þ2þC T � 293ð Þ þD

where, A, B, C and D are regression constants.

The present data were compared with those reported forMoSi2 (Ref 11), Si (Ref 20), SiC (Ref 11) and Si3N4 (Ref 11). Itis demonstrated that the thermal expansion for the silicides waslarger than those for Si, SiC, and Si3N4 with the deviationbetween the former and latter set of data increasing withincreasing temperature. It is concluded that the thermal strainswould increase with increasing temperature if silicides are usedas contact materials on Si or SiC, or in silicide-based SiC fiberreinforced composites.

Acknowledgments

The author thanks the late Ms. Anna Palczer for conducting thethermal expansion measurements. This research was supported bya generous Grant from NASA�s ARMD Seedling Fund Program,and this is gratefully acknowledged.

References

1. K.N. Tu and J.W. Mayer, Silicide Formation, Thin Films: Interdiffusionand Reactions, J.M. Poate, K.N. Tu and J.W. Mayer, Ed., Wiley, NewYork, 1978, p 359-405

2. S.P. Murarka, Silicides for VLSI, Applications, Academic Press, NewYork, NY, 1983

3. J.P. Gambino and E.P. Colgan, Silicide and Ohmic Contacts, Mater.Chem. Phys., 1998, 52, p 99–146

4. L. Miglio and F.M. D�Heurle, Eds., Silicides: Fundamentals andApplications, Proc. 16th Course of the Intern. School of Sol. Stat.

Table 5 Values of the regression constants describing the thermal expansion of TiSi2 between 293 and 1523 K


641 1st heat-up to 3rd cool-down 1.89 10�10 �1.79 10�07 1.29 10�03 �2.09 10�02 0.999

Fig. 9 Temperature dependence of the thermal expansion behaviorof WSi2 during three heat up-cool down cycles between 303 and1523 K. The present data are compared with those compiled by Tou-loukian et al. (Ref 11)

Table 6 Values of the regression constants describing the thermal expansion of WSi2 between 293 and 1523 K


622 1st cool-down to 3rd cool-down 1.29 10�10 �1.29 10�07 8.39 10�04 �8.09 10�02 0.999

Fig. 10 Comparison of the temperature dependence of the averagethermal expansion behavior of hot-pressed CrSi2, Cr-30%Mo-30%Si,MoSi2 (Ref 11), Si (Ref 20), SiC (Ref 11), Si3N4 (Ref 11), TiSi2 andWSi2 between 293 and 2000 K



Phys., Erice, Italy, June 5-16, 1999, World Scientific, Washington, DC,2000

5. L.J. Chen, Metal Silicides: An Integral Part of Microelectronics, JOM,2004, 57, p 24–30

6. A. Heinrich, Semiconducting Silicides: Thermoelectric Properties andApplications, Silicides: Fundamentals and Applications, Proc. 16thCourse of the Intern. School of Sol. Stat. Phys., Erice, Italy, June 5-16,1999, L. Miglio and F.M. D�Heurle, Eds., World Scientific,Washington, DC, 2000

7. A.S. Pogosyan and A.Kh. Badyon, Molybdenum Disilicide HeatingElements Manufactured by the High-Temperature Heating-ElementFactory of Kirovkan, Soviet Powder. Metall. Met. Ceram., 1970, 9,p 1023–1029

8. Z. Yao, J.J. Stiglich, and T.S. Sudarshan, Molybdenum Silicide BasedMaterials andTheir Properties, J.Mater. Eng. Perform., 1999,8, p 291–304

9. S.J. Grisaffe and S.R. Levine, Fused Silicide Coatings ContainingDiscrete Particles for Protecting Niobium Alloys, U. S. PatentUS3931447 A, 1976

10. A.K. Vasudevan and J.J. Petrovic, A Comparative Overview of Molyb-denum Disilicide Composites,Mater. Sci. Eng., 1992, A155, p 1–17

11. Y.S. Touloukian, R.K. Kirby, R.E. Taylor, and T.Y.R. Lee (Eds.),Thermal Expansion: Nonmetallic Solids, Thermophysical Properties ofMatter, Vol. 13, Plenum, New York, 1977, p 1195-1220

12. L.F. Verkhorobin and N.N. Matyushenko, The Thermal Expansion ofDisilicides of Some Transition Metals, Soviet Powder. Metall. Met.Ceram., 1963, 2, p 468–469

13. I. Engstrom and B. Lonnberg, Thermal Expansion Studies of the GroupIV-VII, Transition-Metal Disilicides, J. Appl. Phys., 1988, 63,p 4476–4484

14. S.V. Raj and A. Palczer, Thermal Expansion of Vacuum PlasmaSprayed Coatings, Mater. Sci. Eng. A, 2010, 527, p 2129–2135

15. S.V. Raj, An Evaluation of the Properties of Cr3Si Alloyed with Mo,Mater. Sci. Eng. A, 1995, 201, p 229–241

16. R.P. Messner and Y.M. Chiang, Liquid-Phase Reaction-Bonding ofSilicon Carbide Using Alloyed Silicon-Molybdenum Melts,J. Am. Ceram. Soc., 1990, 73, p 1193–1200

17. Y. Pan and J.L. Baptista, The Infiltration of Cobalt Silicides into SiliconCarbide Preforms, J. Eur. Ceram. Soc., 1998, 18, p 201–207

18. M.G. Hebsur, Development and Characterization of SiC(f)/MoSi2-Si3N4(p) hybrid composites, Mater. Sci. Eng. A, 1999, 261, p 24–37

19. X. Lin, L. Cheng, L. Zhang, Y. Xu, and C. You, Microstructure andOxidation Resistance of Carbon/Silicon Carbide Composites Infiltratedwith Chromium Silicide, Mater. Sci. Eng. A, 2000, 290, p 89–94

20. C.A. Swenson, Recommended Values of Thermal Expansivity ofSilicon from 0 to 1000 K, J. Phys. Chem. Ref. Data, 1983, 12,p 179–182



Documents

NASA Aeronautics Research Institute · 11/12/2014 · Thermal expansion measurements were conducted on hot-pressed CrSi 2, TiSi 2, WSi 2 and a two-phase Cr-Mo-Si intermetallic alloy