Lattice trends in Ti5Si3Zx (Z=B,C,N,O and 0<x<1)

Journal of Alloys and Compounds 296 (2000) 59–66

L

www.elsevier.com / locate / jallcom

Lattice trends in Ti 5 Si 3 Z x (Z5B,C,N,O and 0,x,1) a, b a Andrew J. Thom *, Victor G. Young , Mufit Akinc a

Ames Laboratory and Department of Materials Science and Engineering, Iowa State University, Ames, IA 50011, USA b Department of Chemistry, University of Minnesota, Minneapolis, MN 55455, USA Received 9 June 1999; accepted 9 June 1999

Abstract Several Ti 5 Si 3 Z x compositions (Z5boron, carbon, nitrogen and oxygen) were synthesized by arc-melting. Powder X-ray diffraction indicates the materials maintain the Mn 5 Si 3 -type structure of the binary compound Ti 5 Si 3 . Calculated cell constants were correlated to ternary composition based on chemical analysis. Nitrogen and oxygen additions to Ti 5 Si 3 promote a cell volume decrease, while boron additions promote a cell volume increase. Carbon additions cause a decrease in the a-cell constant and an increase in the c-cell constant with a concomitant cell volume increase. Room temperature X-ray single crystal structural analysis was performed on one composition for each ternary addition. Each analyzed composition has the P6 3 / mcm space group (No. 193) with the Mn 5 Si 3 -type structure. The ternary addition occupies the normally vacant interstitial site at the center of the trigonal antiprisms of titanium (Ti2) atoms (Ti 6 Z polyhedra), located in chains at the corners of the hexagonal unit cell. Bonding between the interstitial atoms and the titanium (Ti2) atoms in the Ti 6 Z polyhedra is indicated by a decrease in the Ti2–Ti2 and Ti2–Z atomic separations.  2000 Elsevier Science S.A. All rights reserved. Keywords: Titanium silicide; Ti 5 Si 3 Z x ; Lattice trends; Single crystal; X-ray diffraction

1. Introduction Previous experimental work by Corbett and collaborators [1–5] has shown that intermetallic compounds of the form A 5 B 3 which possess the Mn 5 Si 3 -type structure present a wealth of potential interstitial chemistries. Materials with the base A 5 B 3 composition can be alloyed with a wide variety and substantial amount of Z elements. The resulting A 5 B 3 Z x (x#1) compound retains the Mn 5 Si 3 type structure of the base A 5 B 3 compound. The element Z (hereto referred to as dopant) occupies an interstitial site within the structure that contributes to a change in the bonding characteristics of the material. This presents a unique opportunity to modify material properties by suitable interstitial additions. A 5 Si 3 silicide intermetallics [6– 10], which are candidate high temperature structural materials, can be tailored for potential use by suitable additions which improve critical material properties [11,12]. Otherwise inferior materials can be dramatically improved by appropriate additions. Transition metal silicides of the form M 5 Si 3 exist in three structures: tetragonal Cr 5 B 3 -type, tetragonal W5 Si 3 type, and hexagonal Mn 5 Si 3 -type. Some M 5 Si 3 silicides will only exist in the Mn 5 Si 3 -type structure in the presence *Corresponding author.

of ternary stabilizing elements such as boron, carbon, nitrogen, and oxygen. These hexagonally stabilized materials, formulated as M 5 Si 3 Zx , are called Nowotny phases [13]. In the absence of the stabilizer, the pure binary M 5 Si 3 compound has the W5 Si 3 -type structure. The voids within the W5 Si 3 -type structure are not large enough to accommodate a significant amount of Z interstitials, and the Mn 5 Si 3 type structure becomes favored because of its ability to incorporate these additions [14]. As the amount of interstitial is further increased, a ternary compound having the Cr 5 B 3 -type structure may form. Parthe´ summarized the compositional variation of several M 5 Si 3 Z x Nowotny phases [15]. Considering the transition metals from Group IVB to VIB, the amount of carbon required to stabilize the Mn 5 Si 3 -type structure phase increases with both the group and period of the metal. It is important to note, however, that the discovery of the interstitial stabilization effect by Nowotny was based on the accidental introduction of impurities. More recent investigations using purer starting materials and controlled synthesis techniques have revealed the existence of many true binary A 5 B 3 compositions with the Mn 5 Si 3 -type structure [16]. Fig. 1 shows an [001] projection of the Ti 5 Si 3 unit cell. This is a hexagonal structure (hP16) in space group P6 3 / mcm that is characterized by two distinct chains extending

0925-8388 / 00 / $ – see front matter  2000 Elsevier Science S.A. All rights reserved. PII: S0925-8388( 99 )00533-2

60

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

melting point of many of the Mn 5 Si 3 -type materials. The melting points of most Group IVB–VIB transition metal A 5 Si 3 silicides are in excess of 20008C [18,19]. Light element additions to Ti 5 Si 3 have a pronounced effect on properties such as oxidation resistance [20] and thermal expansion anisotropy [21]. In the preceding experimental work in which the base composition Ti 5 Si 3 was doped with ternary additions such as carbon, boron, nitrogen, and oxygen, it was assumed that the dopant occupied the interstitial site at (0,0,0) within the Mn 5 Si 3 type structure. In order to validate this assumption, powder and single crystal X-ray diffraction (XRD) studies were conducted. Fig. 1. [001] projection of the Ti 5 Si 3 unit cell that has the Mn 5 Si 3 -type structure. The large circles are Ti, medium circles are Si, and the small circles are interstitial sites. Height symbol key: (z50, ]12 ), (z5 ]14 ), (z5 ]34 ).

along the c-direction. The first chain is a linear chain of titanium atoms (Ti1) located at ( ]13 , ]23 ,z) where z50, ]12 . Ti1 atoms are coordinated by six silicon atoms (Si) which are intermediate between trigonal prismatic and trigonal antiprismatic. The second chain is composed of titanium atoms (Ti2) which form trigonal antiprisms. Within the cavity of the (Ti2) trigonal antiprism is an interstitial site at (0,0,0) and (0,0, ]12 ) which is normally vacant in the binary material. With two formula units per unit cell, up to two interstitial atoms can be accommodated per unit cell. This yields a formula for the fully-doped Mn 5 Si 3 -type structure of Mn 5 Si 3 Z 1.0 which is isostructural with the Ti 5 Ga 4 -type structure (hP18). The silicon homogeneity range of Ti 5 Si 3 was not investigated in this work. Based on the binary Ti–Si phase diagram [17] that indicates a 2.5 atomic percent homogeneity range at room temperature, the silicon-rich composition would be Ti 5 Si 3 Si 0.16 . Further silicon additions promote the formation of a two phase mixture of Ti 5 Si 3.16 and Ti 5 Si 4 . Tetragonal Ti 5 Si 4 has an unrelated Zr 5 Si 4 -type (tP36) structure. The Ti2 chains are located at the corners of the hexagonal unit cell. The trigonal antiprisms can also be considered face sharing octahedra. Silicon atoms are at the three edges of each shared face, bonding to two Ti2 atoms within the plane (in-plane) on the face edge and also bonding to two Ti2 atoms (out-of-plane), one each above and below the plane. The silicon atoms serve to bond together the structure by forming an (interchain) bond with the Ti2 atom of another trigonal antiprism chain within the Ti2 plane and also forming the distorted octahedral chains about the Ti1 atoms. An ABACA stacking sequence can be envisioned by considering an A plane of the interstitial sites and Ti1 atoms. The C plane, which contains the Ti2 and Si atoms, is rotated by 1808 with respect to the B plane. The high degree of stability of this structure promoted by the strong chain bonding is evidenced by the high

2. Experimental Compositions of Ti 5 Si 3 Z x (Z5B,C,N,O and 0,x,1) were synthesized by arc-melting the constituents under a protective argon atmosphere using a nonconsumable tungsten electrode. Table 1 gives the constituent materials for each of the compositions synthesized. Large buttons (125 g) of Ti 5 Si 3 were arc-melted from reagent grade Ti and Si and then divided into smaller pieces (40 g) as starting material for each Ti 5 Si 3 Zx composition. Reagent grade boron, graphite, TiN, and TiO 2 were also used with all material added as pieces and each composition triple melted to promote chemical homogenization. Carbon content was measured in a carbon and sulfur analyzer (Model EMIA-520, Horiba, Tokyo, Japan). Nitrogen and oxygen content were determined using the inert gas fusion technique (Model TC-436, Leco, St. Joseph, MO, USA). Boron content was analyzed by the inductively coupled plasma–atomic emission spectroscopy technique (Fisons 3410 Minitorch, Applied Research, Dearbon, MI, USA). Pieces of the arc-melted buttons were ground with an agate mortar and pestle and sieved to -325 mesh (,37 mm) for XRD. The powders were analyzed using powder XRD with CuKa radiation (XDS 2000, Scintag, Sunnyvale, CA, USA). Patterns were collected using a 0.028 step size with a 1 s count time over the 2u angular range of 208–1108. Peak positions were determined using the profile-fitting capabilities of the diffractometer software, modeling peak shapes using a Pearson VII profile and deconvoluting Ka 1 and Ka 2 peaks. Cell constants were calculated from the peak positions using a least-squares refinement program [22].

Table 1 Starting materials for synthesized Ti 5 Si 3 Z x Composition

Starting materials

Ti 5 Si 3 B x Ti 5 Si 3 C x Ti 5 Si 3 N x Ti 5 Si 3 O x

Ti 5 Si 3 1boron Ti 5 Si 3 1graphite Ti 5 Si 3 1TiN1Si Ti 5 Si 3 1TiO 2 1Si

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

61

Table 2 X-ray single crystal experimental results Refined formula

Ti 5 Si 3 O 0.09( 1 )

Ti 5 Si 3 B 0.45( 2 )

Ti 5 Si 3 C 0.31( 1 )

Ti 5 Si 3 N 0.42( 1 )

Ti 5 Si 3 O 0.43( 1 )

Crystal size (mm) Cell dimensions ˚ a (A) ˚ c (A) ˚ 3) Vol. (A Formula weight Density (g / cm 3 ) R1[I .2s (I)] a

0.1830.1230.04

0.5830.1030.07

0.2330.1030.10

0.5030.1030.10

0.2330.1030.10

7.4521(8) 5.1522(15) 247.79(8) 325.21 4.359 0.0108

7.464(1) 5.165(3) 249.2(2) 328.64 4.380 0.0092

7.4477(7) 5.153(1) 247.54(7) 327.49 4.394 0.0091

7.4309(7) 5.138(1) 245.71(7) 329.65 4.456 0.0092

7.4356(7) 5.131(1) 245.68(6) 330.65 4.470 0.0128

a

O

uuFou 2uFcuu R1 5 ]]] uFou

O

For selected compositions, a crystal was extracted from the respective arc-melted button and mounted on a glass fiber in the diffractometer (CAD4, Enraf-Nonius, Bohemia, NY, USA). Data were collected at 29361 K using monochromated MoKa radiation. The typical u range was from 3.168 to 29.968 with an index range of 210 # h # 10, 210 # k # 10, and 0 # l # 7. The v 2 2u scan technique was used to collect all X-ray data. The cell constants for the data collections were determined from reflections found from a random search routine. Lorentz and polarization corrections were applied. A nonlinear correction based on the decay in the standard reflections was applied to the data although there was no measurable radiation decay. A series of azimuthal reflections was collected for each specimen. An empirical absorption correction was applied to the data for each specimen. Results of the X-ray single crystal experiments are given in Table 2.

compositions, carbon ranging from 120–300 ppm and nitrogen ranging from 53–200 ppm. However, there is an interesting variation in oxygen content among the samples. For undoped, boron-doped, and nitrogen-doped Ti 5 Si 3 , the oxygen level is relatively constant at about 1000 ppm. For the carbon-doped material, the oxygen level significantly decreases with increasing carbon content. Carbon may act as an oxygen getter, producing a gaseous species which evolves from the material. The accompanying decrease in carbon content is inconsequential given the significantly larger level of carbon added to the sample. The crystallographic location of the various nondoped Z atoms within the structure is unknown. Oxygen is presumably in the form of both surface oxide and interstitial oxygen, while boron, carbon and nitrogen could be interstitial or perhaps substituted on another lattice site such as silicon. Minor second phase formation of borides, carbides, or nitrides is also possible, but such levels are well below the detection threshold of XRD.

3. Results

3.2. Room temperature cell constants 3.1. Chemical analysis The results of chemical analyses performed on pieces of each arc-melt composition are shown in Table 3. The level of nondoped Z is relatively constant between the various

The variation of cell constants with level of interstitial addition is shown in Figs. 2–5. A vertical error bar of ˚ was used for all powder XRD data. The 60.0045 A rationale for establishing this error bar is given in the

Table 3 Chemical analysis of arc-melted Ti 5 Si 3 Z x Nominal composition (Z x )

Carbon (x)

Nitrogen (x)

Oxygen (x)

Boron (x)

Undoped C 0.25 C 0.50 C 0.75 B 0.25 B 0.50 B 0.75 N 0.25 N 0.50 N 0.75 O 0.25 O 0.50 O 0.75

120 ppm x50.251 x50.579 x50.709 260 ppm 200 ppm 180 ppm 150 ppm 200 ppm 170 ppm 290 ppm 300 ppm 250 ppm

69 ppm 56 ppm 53 ppm 58 ppm 58 ppm 58 ppm 58 ppm x50.172 x50.376 x50.520 120 ppm 170 ppm 200 ppm

1100 ppm 674 ppm 469 ppm 186 ppm 750 ppm 1100 ppm 940 ppm 910 ppm 1100 ppm 1200 ppm x50.267 x50.455 x50.647

1 1 1 1 x50.183 x50.394 x50.746 1 1 1 1 1 1

1

Not analyzed

62

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

Fig. 2. Variation of a- and c-cell constants with level of boron content. Open symbols denote cell constants calculated from powder diffraction. Solid symbols denote cell constants calculated from single crystal diffraction.

Fig. 3. Variation of a- and c-cell constants with level of carbon content. Open symbols denote cell constants calculated from powder diffraction. Solid symbols denote cell constants calculated from single crystal diffraction.

Discussion. The addition of boron promotes a significant expansion in both cell constants. In contrast, the addition of oxygen promotes a strong contraction in both cell

Fig. 4. Variation of a- and c-cell constants with level of nitrogen content. Open symbols denote cell constants calculated from powder diffraction. Solid symbols denote cell constants calculated from single crystal diffraction.

Fig. 5. Variation of a- and c-cell constants with level of oxygen content. Open symbols denote cell constants calculated from powder diffraction. Solid symbols denote cell constants calculated from single crystal diffraction.

constants. For the introduction of nitrogen, the a-cell constant significantly contracts while the c-cell constant is relatively unaffected. Carbon has an intermediate effect in which the c-cell constant strongly expands while the a-cell constant undergoes a minimum. For carbon-doped, nitrogen-doped, and oxygen-doped Ti 5 Si 3 , the change in the cell constants with increasing additions appears to decrease and may indicate an increasing interaction between the interstitials. It is not clear if boron-modified Ti 5 Si 3 exhibits this decreasing effect since more composition points are needed to establish the trend. There is sparse data for the direct comparison of these measured cell constants. Kajitani et al. [23] studied the structure of Ti 5 Si 3 doped with hydrogen isotopes. For the composition of Ti 5 Si 3 D 0.9 , deuterium occupied the interstitial sites at (0,0,0). The c-cell constant expanded about 0.3% and the a-cell constant contracted about 0.16% compared to Ti 5 Si 3 . The effect of similar interstitial additions on the cell constants of Zr 5 Si 3 and Zr 5 Sn 3 has been measured [2,4]. These data are summarized in Table 4. The addition of carbon and oxygen to Zr 5 Si 3 causes behavior similar to that of Ti 5 Si 3 . Carbon additions cause a decrease in the a-cell constant and an increase in the c-cell constant with an overall volume increase. In contrast, oxygen additions decrease both cell constants which results in a volume decrease. The addition of boron and nitrogen to Zr 5 Sn 3 also causes behavior similar to that of boron-doped and nitrogen-doped Ti 5 Si 3 . Boron causes a strong increase in both cell constants with an increase in volume. Conversely, nitrogen decreases both cell constants to decrease volume. These analyses indicate that several atomic percent of boron, carbon, nitrogen, and oxygen can be added to Ti 5 Si 3 while still maintaining the Mn 5 Si 3 -type structure. These observations are also consistent with available ternary phase diagrams [24–26] which indicate that Ti 5 Si 3 Z x can accommodate up to 10 atomic percent of carbon, nitrogen, and oxygen at 10008–11008C.

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

63

Table 4 Unit cell values for A 5 Si 3 Z x compositions Da ˚ (A)

Compound

a ˚ (A)

Ti 5 Si 3 Ti 5 Si 3 C 0.50 Ti 5 Si 3 C 0.90 Ti 5 Si 3 O 0.75 Ti 5 Si 3 B 0.75 Ti 5 Si 3 N 0.75 Zr 5 Si 3.1 Zr 5 Si 3 C 0.5 Zr 5 Si 3 C Zr 5 Si 3 O Zr 5 Sn 3 Zr 5 Sn 3 B Zr 5 Sn 3 N

7.4543(4) 7.4424(3) 7.4495(2) 7.4305(4) 7.478(1) 7.4259(5) 7.9582(6) 7.9409(5) 7.9400(6) 7.9200(6) 8.4560(7) 8.4936(4) 8.4040(8)

‡

c ˚ (A)

20.0119 20.0048 20.0238 10.0237 20.0284 20.0173 20.0182 20.0382 10.0376 20.0520

5.1474(6) 5.1677(4) 5.1691(3) 5.1333(2) 5.1799(7) 5.1452(3) 5.5613(8) 5.6016(6) 5.6116(8) 5.5502(7) 5.779(1) 5.8029(5) 5.7698(8)

10.0203 10.0217 20.0141 10.0325 20.0022 10.0403 10.0503 20.0111 10.0239 20.0092

Volume ˚ 3) (A 247.70(3) 247.89(2) 248.43(2) 245.45(3) 250.87(6) 245.71(3) 305.03(6) 305.90(5) 306.38(7) 301.50(6) 357.86(9) 362.54(5) 352.91(8)

DVol. (%)

Ref. ‡

10.077 10.30 20.91 11.3 20.80 10.29 10.44 21.2 11.3 21.4

‡ ‡ ‡ ‡ ‡

[2] [2] [2] [2] [4] [4] [4]

This work

3.3. Single crystal results The space group P6 3 / mcm and the Mn 5 Si 3 -type structure were used for the initial structural model. All atoms were refined with anisotropic thermal parameters, and final refinements [27,28] were performed with SHELXL-93. The (010) reflection was omitted from the refinements for each sample. Inclusion of this low-angle reflection caused problems with refinement of anisotropic thermal parameters for the doped site. Fo was always about 20% higher than Fc for this reflection. The error is most likely related to inaccuracies in the extinction correction. Refinement calculations [27,29] were performed using the SHELXTLPlus and SHELXL-93 programs. The data were quite well fitted by the refined models. The R1 values were at or below 1% for all five analyses (Table 2), suggesting high quality single crystals. The calculated stoichiometry for each analyzed single crystal is given in Table 5. There are some significant differences between the refined composition and the chemically analyzed composition. However, refinement of fractional occupancies to estimate stoichiometry of the interstitial dopant is not particularly accurate for light atoms such as boron, carbon, nitrogen, and oxygen because of their low X-ray scattering cross-section. For this reason, the stoichiometry for the single crystals plotted in Figs. 2–5 was Table 5 Dopant stoichiometry of arc-melted single crystals Nominal composition a

Ti 5 Si 3 Ti 5 Si 3 B 0.50 Ti 5 Si 3 C 0.12 Ti 5 Si 3 N 0.25 Ti 5 Si 3 O 0.50 a

Dc ˚ (A)

Chemical analysis

Single crystal analysis

x50.022 x50.39 11 x50.17 x50.46

x50.09(1) x50.45(2) x50.31(1) x50.42(1) x50.43(1)

Chemical / single crystal analysis based solely on interstitial oxygen content 11 Not analyzed

assumed identical to the nominal composition of the arcmelted button from which they were extracted. The undoped crystal refined better when including interstitial oxygen in the structure (x50.09 formula units). Oxygen presence is not surprising since the arc-melting process does not establish an oxygen-free atmosphere. The added dopant does not completely displace all of the oxygen and some impurity oxygen remains interstitially within the structure. Bond lengths determined from the single crystal analyses are shown in Table 6, together with the results of a single crystal study of Ti 5 Si 3 O by Perchenek [30]. The results from this work agree well with those of Perchenek. In general, the data of Perchenek show a greater change in bond lengths, and this likely reflects the increased oxygen content of Perchenek’s material. The behavior of Ti 5 Si 3 C x and Ti 5 Si 3 O x as measured in this study follow the trend observed for Zr 5 Sn 3 C and Zr 5 Sn 3 O as determined in a crystal structure determination by Kwon and Corbett [4]. The addition of carbon caused ˚ and the the Zr2–Z interstitial cavity to decrease by 0.08 A, Zr2–Zr2 lengths of the Zr trigonal antiprisms to decrease ˚ All Zr2–Sn distances increased, with the by 0.10–0.12 A. interchain and out-of-plane distances about equally affected. The addition of oxygen to Zr 5 Sn 3 causes similar changes in bond lengths. The work by Williams et al. [31] also agrees with the bond lengths calculated from the single crystal analyses of the present study.

4. Discussion

4.1. Interstitially induced bonding changes The change in cell constants can be related to changes in local bonding caused by introduction of the interstitial atoms. The effect of the interstitial atoms is indicated by the Ti2–Z bond length, which scales with the Ti2–Z

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

64

Table 6 ˚ Bond lengths for various interstitial additions (Z) to Ti 5 Si 3 Z x (A) Bond type

Vacant

Z5B

Z5C

Z5N

Z5O

Ti 5 Si 3 O (Ref. [23])

Ti2–Z Ti2–Si Interchain Out-of-plane In-plane Ti1–Si Ti2–Ti2 Out-of-plane In-plane Ti1–Ti1

2.2645(4)

2.2461(5)

2.2305(3)

2.2024(4)

2.2105(4)

2.183(3)

2.6650(8) 2.7865(7) 2.5642(6) 2.6323(3)

2.6757(6) 2.8118(13) 2.5804(5) 2.6307(5)

2.6799(5) 2.8119(6) 2.5757(4) 2.6240(3)

2.6833(6) 2.8227(6) 2.5808(5) 2.6124(3)

2.6735(6) 2.8162(5) 2.5849(5) 2.6123(3)

2.644(3) 2.851(4) 2.609(3) 2.593(3)

3.1789(7) 3.2256 2.5761(7)

3.1697(12) 3.1829 2.5825(14)

3.1550 3.1537(6) 2.5766(6)

3.1306(6) 3.0988(7) 2.5691(7)

3.1341(5) 3.1181(7) 2.5655(5)

3.118(4) 3.056(4) 2.571(3)

interaction, and the octahedral nature of the Ti2 trigonal antiprismatic unit, indicated by the difference between the Ti2–Ti2 (out-of-plane) and the Ti2–Ti2 (in-plane) bond lengths. These values are given in Table 7. For all additions, contraction of the interstitial cavity occurs, suggesting that Ti2–Z bonding occurs. As the (Ti2–Ti2) difference approaches zero, the trigonal antiprismatic units should become more isotropic due to the increased symmetry of the bonding between (in-plane) and (out-of-plane) Ti2 atoms. This decrease in Ti2–Ti2 difference is accompanied by a smaller cavity contraction, and together these changes in Ti2–Ti2 bond behavior contribute to changes in the cell constants. Nitrogen and oxygen promote the strongest cavity contraction and thus a contraction is observed in both the cell constants. Boron promotes the smallest cavity contraction, and this occurs despite the increase in both cell constants, suggesting other significant changes in bonding character are occurring. Carbon promotes a smaller cavity contraction, and this is accompanied by a decrease in the a-cell constant despite an increase in the c-cell constant. The increased Ti2 bonding symmetry for carbon doping may contribute to the decreased anisotropy observed for carbon-doped Ti 5 Si 3 . The thermal expansion anisotropy (CTE c-axis / CTE a-axis ) of Ti 5 Si 3 C 0.85 is 1.9 and is significantly lower than 2.4 for Ti 5 Si 3 [21]. While the decrease in the Ti2–Z dimension suggests bonding between Ti2 and Z, it should be noted that bond strength does not necessarily correlate to bond length [32].

Table 7 Influence of interstitial additions on Ti 6 Z polyhedra Composition

(Ti2–Ti2) Difference a

(Ti2–Z) Contraction

Ti 5 Si 3 O 0.09 Ti 5 Si 3 N 0.42 Ti 5 Si 3 O 0.43 Ti 5 Si 3 C 0.31 Ti 5 Si 3 B 0.45

20.0467 10.0318 10.0160 10.0013 20.0132

0.0000 0.0621 0.0540 0.0340 0.0184

a

(out-of-plane)–(in-plane) distance

Nonetheless, interstitial additions are clearly modifying the nature of Ti2 bonding.

4.2. Single crystal vs. powder XRD cell constants The single crystal cell constants show similar trends (for example, oxygen additions promote contraction of both cell constants) as observed for the powder XRD data. Figs. 2–5 show the single crystal cell constants plotted as solid symbols, with interstitial content x equal to that of the respective nominal arc-melted button. There are some fairly significant differences between the cell constants determined from the single crystal study and the cell constants determined from powder XRD. There are several possible sources for the observed differences. One source of variation is the failure to use an internal standard such as silicon during the powder XRD experiments. These data were collected over a fairly short period of time and sample to sample experimental deviation due to changing instrumental parameters is not expected to be significant. Nonetheless, absolute comparison of cell constants between powder samples and between powder and single crystal measurements is complicated. The dominant source of observed differences between powder and single crystal X-ray measurements is the method used for synthesis. During the arc-melting process, rapid cooling occurs when the electrode is deenergized and the molten button solidifies in the chilled copper hearth. This results in variation in stoichiometry within the button. The arc-melted buttons were not annealed at an elevated temperature to promote chemical homogenization, due to concern of introduction of impurity nitrogen and oxygen from the furnace atmosphere. This inhomogeneous distribution of the interstitial dopant within the arc-melted button will contribute to differences in small samples extracted for preparing a powder sample. Similarly, variations are expected between the composition of individual single crystals and the nominal bulk composition of the arc-melted button.

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

The net effect is a vertical error bar associated with the uncertainty of the cell constant estimates from powder samples throughout the arc-melted button. In order to quantify the vertical error bars, four random powder samples taken from an undoped Ti 5 Si 3 starting material button. Table 8 gives the cell constants of these samples. For Ti 5 Si 3 , the variation in the a- and c-cell constant is ˚ and 0.003 A, ˚ respectively. These variations about 0.006 A are about an order of magnitude greater than the typical cell constant standard errors predicted from the leastsquares refinement. Given the strong contraction effect of interstitial oxygen (Fig. 5) on both cell constants, these variations likely reflect the differences in oxygen content between the four samples. The smaller lattice constants of samples 3 and 4 correspond to a higher oxygen content than samples 1 and 2. Consequently, the lattice constants for sample 1 were plotted as the undoped composition point in Figs. 2–5. The trends in Fig. 5 were used to estimate oxygen content, assuming lattice contraction for samples 2–4 was due entirely to interstitial oxygen incorporation. Sample 1 was assumed to contain x50.022 formula units of oxygen, equal to that determined from chemical analysis (Table 3). Table 8 shows that sample 4 must contain between 0.11–0.18 formula units of oxygen to account for its smaller cell constants. Both the presence and variation of oxygen content in the undoped Ti 5 Si 3 material is a consequence of arc-melting large buttons (125 g). These problems are mitigated by arc-melting much smaller samples of 10–15 g [24]. The relative uncertainty in the instrumental techniques used to measure carbon, nitrogen, and oxygen in this work is 2%, 10%, and 5%, respectively. Unfortunately, achieving a representative sample for chemical analysis is difficult given that only small samples (few hundred milligrams) are taken from a larger sample (40–125 g) having a known compositional inhomogeneity. Since undoped material was used as stock material for all other arc-melted compositions, this level of uncertainty in lattice constants due to oxygen content is possible for all estimates given in Figs. 2–5. Table 8 also gives the cell constants of two powder samples of Ti 5 Si 3 C 0.12 . The variation in lattice constants is much smaller than for the undoped material, indicating the combined effect of variation in carbon content and oxygen impurity level to also

65

˚ be lower. Therefore, the vertical error bar of 60.0045 A ˚ and 0.003 A) ˚ assumed for all (an average of 0.006 A powder data estimates in Figs. 2–5 is a conservative estimate. Based on these error bars, the lattice trends in Figs. 2–5 are significant. Since the vertical error bar mainly describes variability in interstitial content, a horizontal error bar was neglected. Reconsidering the five single crystal measurements, lattice estimates for the carbon-doped and oxygen-doped crystals fall within the error bars of the powder data. The boron-doped crystal is reasonably close, with the c-cell constant slightly outside the error bar. Both cell constants of the nitrogen-doped crystal are strongly contracted with respect to the powder estimates, and this suggests a significant level of oxygen in the crystal. In fact, the c-cell constant of the crystal is intermediate between the trend for oxygen-doped and nitrogen-doped powder estimates, clearly indicating the presence of oxygen in the nitrogen-doped crystal. For the undoped crystal, the c-cell constant is close to the powder data while the a-cell constant is much smaller than expected and falls well outside the error bar. An increase or decrease in oxygen content will not sufficiently explain the observed lattice constants for the undoped crystal. This may be an indication of an inadequacy in the model assumptions used for the single crystal refinement.

4.3. Is Ti5 Si3 a true binary Mn5 Si3 -type structure material? The refinement of the undoped crystal with interstitial oxygen (x50.09 corresponds to 4400 ppm oxygen) warrants further discussion. Given the insensitivity of XRD to light elements, the accuracy of this oxygen content estimate is suspect. However, this is further evidence of a nontrivial level of interstitial presence in the undoped single crystal, particularly given the unexplained contraction in the a-cell constant. Even for sample 1 which clearly possessed the lowest impurity interstitial content, no W5 Si 3 -type structure form of Ti 5 Si 3 was detected. Therefore, Ti 5 Si 3 is most likely a true binary Mn 5 Si 3 structure type material. To better answer these questions, further work is required. Pure Ti 5 Si 3 must be synthesized and appro-

Table 8 Replicated cell constant measurements Sample

Composition

˚ a (A)

˚ c (A)

Estimated oxygen content (x)

1 2 3 4

Ti 5 Si 3 Ti 5 Si 3 Ti 5 Si 3 Ti 5 Si 3

7.460 7.4584 7.4559 7.4543

5.150 5.1502 5.1478 5.1474

0.022 (chemical analysis) 0.022–0.062 0.098–0.12 0.11–0.16

5 6

Ti 5 Si 3 C 0.12 Ti 5 Si 3 C 0.12

7.4522 7.4508

5.1536 5.1509

– –

66

A. J. Thom et al. / Journal of Alloys and Compounds 296 (2000) 59 – 66

priately annealed while minimizing impurity interstitial content. Structural analysis using a technique such as neutron diffraction, which is more sensitive to light elements, is needed.

5. Conclusions Ti 5 Si 3 can incorporate several atomic percent of ternary additions such as boron, carbon, nitrogen, and oxygen. The complementary powder and single crystal XRD studies strongly suggest that these additions are accommodated in the interstitial sites (0,0,0) within the trigonal antiprismatic chains of Ti2 atoms of the Mn 5 Si 3 -type structure of Ti 5 Si 3 . Bonding between the interstitial atoms and Ti2 is suggested by cavity contraction. There are significant changes in bond lengths as a result of these additions. Most notably there is an increase in the Ti2–Si bond lengths and a decrease in the Ti2–Ti2 bond lengths.

Acknowledgements Ames Laboratory is operated for the US Department of Energy by Iowa State University under contract number W-7405-ENG-82. This research was supported by the Office of Basic Energy Science, Materials Science Division.

References [1] E. Garcia, J.D. Corbett, Chemistry of polar intermetallic compounds. Study of two Zr 5 Sb 3 phases, hosts for a diverse interstitial chemistry, Inorg. Chem. 27 (13) (1988) 2353–2359. [2] Y. Kwon, M.A. Rzeznik, A. Guloy, J.D. Corbett, Impurity stabilization of phases with the Mn 5 Si 3 -type structure. Questions regarding La 5 Sn 3 and Zr 5 Si 3 , Chem. Mater. 2 (5) (1990) 546–550. [3] E. Garcia, J.D. Corbett, Chemistry in the polar intermetallic host Zr 5 Sb 3 . Fifteen interstitial compounds, Inorganic Chem. 29 (18) (1990) 3274–3282. [4] Y. Kwon, J.D. Corbett, Chemistry in polar intermetallic compounds. The interstitial chemistry of Zr 5 Sn 3 , Chem. Mater. 4 (6) (1992) 1348–1355. [5] J.D. Corbett, E. Garcia, A.M. Guloy, W.M. Hurng, Y. Kwon, E.A. Leon-Escamilla, Widespread interstitial chemistry of Mn 5 Si 3 -type and related phases. Hidden impurities and opportunities, Chem. Mater. 10 (1998) 2824–2836. [6] D.L. Anton, D.M. Shah, D.N. Duhl, A.F. Giamei, Selecting hightemperature structural intermetallic compounds: The engineering approach, J. Metals 41 (1989) 12–17. [7] A.I. Taub, R.L. Fleischer, Intermetallic compounds for high-temperature structural use, Science 243 (1989) 616–621. [8] P.J. Meschter, D.S. Schwartz, Silicide-matrix materials for hightemperature applications. J. Metals (1989) 52–55. [9] D.M. Shah, D. Berczik, D.L. Anton, R. Hecht, Appraisal of other silicides as structural materials, Mat. Sci. Eng. A155 (1992) 45–57.

[10] K.S. Kumar, C.T. Liu, Ordered intermetallic alloys. Part II: Silicides, trialuminides, and others, J. Metals 45 (1993) 28–34. [11] A.J. Thom, M.K. Meyer, Y. Kim, M. Akinc, Evaluation of A 5 Si 3 Z x intermetallics for use as high temperature structural materials, in: T.S. Srivatsan, V.A. Ravi, J.J. Moore (Eds.), Min. Met. Mater. Soc. Symp. Proc., Warrendale, PA, Processing and Fabrication of Advanced Materials for High-Temperature Applications, Vol. III, 1994, pp. 413–438. [12] M. Akinc, M.K. Meyer, M.J. Kramer, A.J. Thom, J.J. Huebsch, B. Cook, Boron-doped molybdenum silicides for structural applications, Mater. Sci. Eng. A261 (1999) 16–23. [13] H. Nowotny, Alloy chemistry of transition element borides, carbides, nitrides, aluminides, and silicides, in: P. Beck (Ed.), Electronic Structure and Alloy Chemistry of the Transition Elements, Wiley, New York, 1963, pp. 179–220. [14] B. Aronsson, Borides and silicides of the transition metals, Arkiv. Kemi. 16 (1960) 379–423. ´ Contributions to the Nowotny phases, Powder Metall. [15] E. Parthe, Bull. 8 (1 / 2) (1957) 23–34. [16] A.M. Guloy, J.D. Corbett, The lanthanum–germanium system. Nineteen isostructural interstitial compounds of the La 5 Ge 3 host, Inorg. Chem. 32 (16) (1993) 3532–3540. [17] T.B. Massalski, in: Binary Alloy Phase Diagrams, Vol. 3, American Society for Metals, Materials Park, OH, 1990, p. 3370. [18] T.Y. Kosolapova, in: Handbook of High Temperature Compounds: Properties, Production, Application, Hemisphere, New York, 1990, p. 162. [19] T.B. Massalski, Binary Alloy Phase Diagrams, Vol. 3, American Society for Metals, Materials Park, OH, 1990, (several references therein). [20] A.J. Thom, M. Akinc, Effect of ternary additions on the oxidation resistance of Ti 5 Si 3 , in: H.M. Hawthorne, T. Troczynski (Eds.), Advanced Ceramics for Structural and Tribological Applications, Canadian Institute of Mining, Metallurgy, and Petroleum, 1995, pp. 619–629. [21] A.J. Thom, M. Akinc, O.B. Cavin, C.R. Hubbard, Thermal expansion anisotropy of Ti 5 Si 3 , J. Mater. Sci. Lett. 13 (1994) 1657–1660. [22] D.E. Appleman, H.T. Evans, NTIS Document No. PB-216188, 1973. [23] T. Kajitani, T. Kawase, K. Yamada, M. Hirabayashi, Site occupation and local vibration of hydrogen isotopes in hexagonal Ti 5 Si 3 H(D) 12x , Trans. Jpn. Inst. Metals 27 (9) (1986) 639–647. [24] J.I. Goldstein, S.K. Choi, F.J.J. Van Loo, G.F. Bastin, R. Metselaar, Solid state reactions and phase relations in the Ti–Si–O system at 1373 K, J. Am. Ceram. Soc. 78 (2) (1995) 313–322. [25] S. Arunajatesan, A.H. Carim, Synthesis of titanium silicon carbide, J. Amer. Ceram. Soc. 78 (3) (1995) 667–672. [26] S. Sambasivan, W.T. Petuskey, Phase chemistry in the Ti–Si–N system: Thermochemical review with phase stability diagrams, J. Mater. Res. 9 (9) (1994) 2362–2369. [27] G.M. Sheldrick, SHELXL93, Program for the Refinement of Crystal ¨ Structures, University of Gottingen, 1993. [28] All X-ray scattering factors and anomalous dispersion terms were obtained from the International Tables for Crystallography, C, 4.2.6.8 and 6.1.1.4. [29] SHELXTL-PLUS, Siemens Analytical X-ray, Madison, WI. [30] N. Perchenek, Metallreiche Titansulfide und Die Phase Ti 5 Si 3 O, ¨ Stuttgart, 1988. Thesis, Universitat [31] J.J. Williams, M.J. Kramer, M. Akinc, J. Mater. Res. (submitted). [32] J.D. Corbett, Chevrel phases: An analysis of their metal–metal bonding and crystal chemistry, J. Solid State Chem. 39 (1981) 56–74.