- Email: [email protected]
Magnetic influence on thermoelectric properties of CrO0.1N0.9
Available online at www.sciencedirect.com
Acta Materialia 59 (2011) 1134–1140 www.elsevier.com/locate/actamat
Magnetic influence on thermoelectric properties of CrO0.1N0.9 Petr Tomesˇ a, Dmitry Logvinovich a, Jirˇ´ı Hejtma´nek b, Myriam H. Aguirre a, Anke Weidenkaff a,⇑ a
Solid State Chemistry and Catalysis, Empa, Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600, Duebendorf, Switzerland b Institute of Physics of ASCR, v.v.i, Na Slovance 2, 182 21 Praha 8, Czech Republic Received 18 August 2010; received in revised form 20 October 2010; accepted 20 October 2010 Available online 17 November 2010
Abstract Thermoelectric polycrystalline chromium oxynitride CrO0.09(3)N0.90(7) (with cubic symmetry and space group Fm-3m) samples were successfully prepared by thermal ammonolysis of chromium oxide. This potential n-type thermoelectric material shows a temperature dependence of the Seebeck coefficient (S = 48 lV K1 at 300 K) that is typical for degenerate semiconductors; the electrical resistivity (q300 2.8 mX cm), however, is highly influenced by the grain boundaries and low density (60%) of the sample. The material exhibits a clear antiferromagnetic–paramagnetic transition simultaneously with an orthorhombic–cubic structural transition at TN = 262 K. The magnetic transition affected the material’s thermal conductivity, electrical resistivity, Seebeck coefficient, magnetic moment and specific heat. Heat capacity measurements at low temperatures revealed the predominance of the magnetic contribution due to spin waves. Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Chromium oxynitride; Electrical properties; Thermal conductivity; Magnetic properties; Specific heat
1. Introduction Conventional thermoelectric materials, such as tellurides, are limited for applications at elevated temperatures in air due to evaporation and corrosion processes. Thus, for the last few years alternative ceramic materials have been considered and developed for high-temperature thermoelectric applications [1–3]. Among them transition metal oxynitrides are a fascinating class of ceramic materials with interesting thermoelectric, magnetic and electronic properties [4]. They can be produced by thermal ammonolysis of oxide precursors [5]. Transition metal nitrides are chemical- and corrosionresistant and possess good mechanical properties [6]. In addition, they exhibit relatively low electrical resistivities associated with large Seebeck coefficients (e.g. CrN: S = 135 lV K1 at T = 300 K). Thus, CrN can be considered as a promising thermoelectric material [7]. 3d tran⇑ Corresponding author. Tel.: +41 79 751 6883; fax: +41 44 823 40 19.
E-mail address: anke.weidenkaff@empa.ch (A. Weidenkaff).
sition metal nitrides exist in different crystal lattice structures. Most of them (e.g. ScN, TiN, VN, CrN) have a rock-salt (RS) structure with a small lattice constant ˚ ); others (i.e. MnN, FeN and CoN) crystallize (a < 4.25 A in the zinc blende (ZB) structure with a larger lattice ˚ < a < 4.6 A ˚ ) [8–12]. The influence of the constant (4.2 A lattice constants and the cohesive energy on the magnetic properties of 3d transition metal nitrides was reported in Ref. [11]. Cohesive energy calculations confirm that early transition metal nitrides (ScN, TiN, VN, CrN) prefer a RS crystal structure, and the later metal nitrides (from MnN) a ZB structure. The influence of the lattice parameter on the magnetic moment is stronger in the RS than in the ZB crystal structure due to possible nanosized precipitates in the ZB phase inducing ferromagnetic (FM) ordering. 3d transition metal nitrides show interesting properties like antiferromagnetism, Pauli paramagnetism and superconductivity [13,8]. For example, thin films of solid solutions of Cr1xTixN have a large magnetoresistance [14]. In the range of 0.28 6 x 6 0.50 these compounds are ferromagnetic, which results from the competition between
1359-6454/$36.00 Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2010.10.046
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
antiferromagnetic (AFM) interactions of half-filled Cr t2g orbitals and interactions governed by the Cr (Ti)–N–Cr (Ti) double-exchange mechanism. Using epitaxial thin films, it was shown that the orientations of the thin films influence the structural transformation and thereby the material properties regardless of the substrate material [15]. CrN is the only nitride that shows an antiferromagnetic– paramagnetic transition with a Neel temperature of 273 K < TN < 286 K, a magnetic moment of 2.4 lB and metallic behavior [16–18]. At the same time, CrN undergoes a structural transition from a low-temperature orthorhombic structure (with a small orthorhombic distortion below TN) to a high-temperature cubic structure accompanied by a 0.59% change in the unit cell volume [16,18]. Using the local spin density functional theory, the magnetic structure of CrN was described as (1 1 0) FM layers with alternating spins every two layers, so that the system is AFM overall. The magnetic behavior of CrN indicates the competition between the superexchange (AFM) and the double-exchange (FM) mechanism [19]. The magnetic and structural transitions are also revealed by a clear hysteresis of the electrical and thermal transport properties of CrN. Thus, a similar behavior is expected of CrO0.09(3)N0.90(7). The aim of this work is to study the thermoelectrical properties of CrO0.09(3)N0.90(7), i.e. the Seebeck coefficient (S), the electrical resistivity (q) and the thermal conductivity (j) with regard to a possible thermoelectric application. The influence of the magnetic properties on the heat capacity and the transport properties is explained by drawing an analogy to the Cr2+/Cr3 mixed valence compounds Cr1xTixN and Cr1xVxN. 2. Experimental Cr(NO3)39H2O was dissolved in deionized water and precipitated in a NH3 (28–30% aq.) solution. The precipitate was washed twice with deionized water and dried at T = 373 K for 12 h and T = 723 K for 1 h to yield a single phase Cr2O3 as confirmed by X-ray powder diffraction (XRPD). The oxide was reacted with ammonia (PanGas, 99.98%, 200 ml min1) at T = 1173 K and T = 1268 K for 4 h and 2 h, respectively. The reactions were carried out in a rotation cavity quartz (SiO2) reactor [20] with an internal diameter of 30 mm. The ceramic pellets were obtained by uniaxial pressing of the ammonolized powder at 10 bar followed by sintering at T = 1268 K for 2 h under ammonia. The sample was cooled down to room temperature under ammonia at a cooling rate of 10 K min1. The bulk sample had a density of 60% of the theoretically achievable density. Structure, phase composition and size-strain parameters of the starting and product powders were studied by X-ray powder diffraction (XRPD) using a Phillips X’Pert PRO MPD H–H System equipped with a linear detector X’Celerator. Crystallographic parameters were obtained from Rietveld refinement of the XRPD data collected in angular
1135
range of 25° < 2H < 135° in 0.017° increments and a counting time of 250 s per step. The background was modeled by a sixth-degree polynomial function and the diffraction peak shape by a pseudo-Voigt profile. All the refinements were performed using FULLPROF [21]. Powder samples were dispersed in methanol, dropped on carbon-film coated copper grids and dried in air. Transmission electron microscopy (TEM) was performed using a Philips CM 30 microscope at 300 kV equipped with an EDX INCA detector from Oxford Instruments. The O/N content of the sample was determined by hot gas extraction (HGE) using a LECO TC500 analyzer [22]. About 50 mg of the sample was placed in a tin container positioned in a double graphite crucible and heated up to T = 3273 K for carbothermal reduction. The nitrogen was measured as N2 by a thermal conductivity cell and oxygen was measured as CO2 in an IR cell. Silicon nitride and silicon oxide were used as calibration standards for nitrogen and oxygen, respectively. The thermal stability was measured by thermogravimetric analysis/differential thermal analysis (TGA/DTA) using a Netzsch STA 409 CD thermobalance in the temperature range of 300 K < T < 1300 K with a heating rate of 10 K min1 under a synthetic air atmosphere (20 vol.% O2/He). The Seebeck coefficient, thermal conductivity and electrical resistivity measurements were carried out using a self-built system [23] within the temperature range of 3 K < T < 300 K. The four-point steady-state method with separated sensors and power contacts was applied. The principle and details of low-temperature measurements can be found in Ref. [23]. At high temperatures (300 K < T < 850 K) the Seebeck coefficient and the electrical resistivity were measured in air using the RZ2001i Ozawa Science, Japan measurement system. The magnetic susceptibility was measured using a vibrating sample magnetometer (VSM) adaptor for the physical property measurement system (PPMS) from Quantum Design in the temperature range of 4 K < T < 400 K. The measurements were performed applying a field of 0.1 T on zero field-cooled (ZFC) and on field-cooled (FC) samples. Magnetization curves were obtained at T = 10 K in the field range of 5 T < H < 5 T. The high temperature heat capacity was studied at a heating rate of 20 K min1 and a gas flow rate of 50 ml min1under a synthetic air atmosphere (20 vol.% O2/He). A heat capacity puck for the PPMS was used in the temperature range of 2.3 K < T < 335 K and a Netzsch differential scanning calorimeter (DSC) 404 C Pegasus in the temperature range of 373 K < T < 700 K. 3. Results The chemical composition of the sample was determined to be CrO0.09(3)N0.90(7). The oxygen content was estimated using EDX spectroscopy coupled with TEM. The exact concentration was measured by hot gas extraction resulting
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
1136
in an oxygen molar fraction of 0.09, which was used as input for the Rietveld refinement of the XRPD data. Since the X-ray powder diffraction (XRPD) pattern of chromium oxynitride resembles the pattern of CrN at room temperature, the CrN model was applied for the Rietveld refinement. The value of the initial occupancy factor of Cr was set to 1, for oxygen and nitrogen to those determined by hot gas extraction and the atomic displacement parameter (ADP) of Cr was refined isotropically. Refinement parameters are summarized in Table 1. The refinement converged with Rwp = 3.79, Rp = 3.04, v2 = 1.18 and is plotted in Fig. 1, the inset showing the unit cell of CrO0.09(3)N0.90(7). The compound has a cubic crystal structure with Fm-3m space group and a lattice parameter of ˚ . As comparison, for CrN lattice paramea = 4.1506(2) A ˚ , 4.135 A ˚ and 4.149 A ˚ were published ters of a = 4.13 A previously [14,17,18]. The increase of the lattice parameter might be attributed to an increase of the average ionic radius on the Cr site in CrO0.09(3)N0.90(7) compared ˚ ). The influence of the to Cr3+ in CrN (D = 0.01035 A cation size on the lattice parameter is, however, partially compensated by the slightly smaller average anionic radius of (O0.09N0.91)2.91 compared to the N3 anion ˚ ) [24]. (D = 0.0072 A Temperature stability of CrO0.09(3)N0.90(7) was assessed by TGA/DTA. The results are shown in Fig. 2. The mass increase above T 700 K is probably due to oxidation of CrO0.09(3)N0.90(7) to Cr2O3 [25]. The process continues up to T 1028 K. Above this temperature, the analysis suggest the formation of a b-Cr2N phase [25,26]. Hence, CrO0.09(3)N0.90(7) has a temperature stability limit at T 700 K exceeding that of the conventional Bi2Te3-based thermoelectric materials. Fig. 3 shows the TEM analysis of the (high-temperature) cubic structure of CrO0.09(3)N0.90(7) in diffraction (a), low magnification (b) and high resolution (c,d) images. The diffraction pattern and the indexing planes in Fig. 3a are in agreement with the atomic model obtained by Rietveld refinement. The small particles have polygonal shapes with a size varying from 30 to 150 nm. Fig. 3d shows the high resolution image of a particle along the [1–10] zone axis with the fast Fourier transformation in the inset. The temperature dependence of the electrical resistivity in the temperature range of 3 K < T < 850 K is shown in Fig. 4a. The measurements reveal semiconducting characteristics up to a temperature of T 600 K with a minimum value of q 2.8 mX cm at T = 300 K. The hysteresis shown in the inset of Fig. 4a, occurred during the electrical
Table 1 Structural parameters of CrO0.09(3)N0.90(7) refined from XRPD data. Space ˚ , a = b = c = 90°, Rwp = 3.79, group: Fm-3m (a = b = c = 4.1506(2) A Rp = 3.04, v2 = 1.22). ˚ 2 Occupancy factor Site Name x Y z Biso, A Cr1 N1/O1
0 1/2
0 1/2
0 1/2
0.23 (2) 0.4
1.000 (0) 0.9/0.1
4a 4b
Fig. 1. Rietveld refinement of the X-ray powder diffraction data of CrO0.09(3)N0.90(7). Space group: Fm-3m. The observed intensities, calculated profile, difference curve and Bragg positions are shown. The inset shows the a model of the cubic crystal structure of CrO0.09(3)N0.90(7).
Fig. 2. TGA/DTA curve of CrO0.09(3)N0.90(7) in the sample was measured under oxygen with a heating rate of 10 K min1.
resistivity cycle measurement using a self-built system within the temperature range of 3 K < T < 300 K, indicates a first order transition and is in agreement with data reported for CrN [17]. The structural and magnetic phase transition from a low-temperature AFM orthorhombic to a high-temperature paramagnetic (PM) cubic structure takes place in a martensitic transformation [7]. In Fig. 4b ln q is plotted against 1/T in the temperature region. The data was linearly fitted above and below TN using the equation q(T) = q0 exp(Ea/2kBT) [22]. Activation energies of Ea,1 34 meV above TN and Ea,2 24 meV below TN were obtained. Both values are smaller than the activation energy of CrN with Ea 70 meV [12,27]. The Seebeck coefficient, as a measure of entropy per charge carrier, is a crucial parameter for characterizing thermoelectric materials. Therefore, it is important to know how the entropy, transported by a charge carrier, is generated when applying a temperature gradient along a CrO0.09(3)N0.90(7) sample, and which type of charge carrier dominates in this process. The Seebeck coefficient of
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
1137
Fig. 3. (a) Diffraction pattern and indexing in agreement with the atomic model obtained by Rietveld refinement, (b and c) detail showing the size and morphologies of CrO0.09(3)N0.90(7) particles, and (d) high resolution image of one particle along the [1–10] zone axis with the fast Fourier transformation in the inset.
the sample is negative and increases almost linearly with increasing temperature (Fig. 5). This indicates that electrons are the main charge carriers as in a degenerate semiconductor. In the temperature range of 255 K < T < 265 K the Seebeck coefficient jumps from 56 lV K1 (below TN) to 46 lV K1 (above TN). At T = 300 K the absolute value of the Seebeck coefficient of CrO0.09(3)N0.90(7) (S = 48 lV K1) is significantly smaller than that of CrN (S = 135 lV K1) but similar to that of Cr0.9V0.1N (S = 83 lV K1) [7]. The low electrical resistivity and a Seebeck coefficient of S 70 lV K1 results in a power factor PF (PF = S2/q) of 1.7 104 W K2 m1 at T = 590 K. Another important requirement for a good thermoelectric material is a very low thermal conductivity. The data of the total thermal conductivity measurements are plotted in Fig. 6. Initially, the total thermal conductivity jtotal increases with temperature. In the temperature range of 255 K < T < 265 K; jtotal suddenly decreases by 35% from 4.2 W m1 K1 to 2.7 W m1 K1, reflecting the influence of the magnetic and structural transition on the thermal transport [7]. The lowering of thermal conductivity in this temperature range is desirable for a possible thermoelectric application near room temperature. According to the classical model jtotal = jel + jph the total thermal conductivity is composed of an electronic and a phononic part. The
electronic part jel has only a minor contribution to the total thermal conductivity (inset of Fig. 6). jel can be estimated using the Wiedemann–Franz law: jel = L0rT, where L0 = 2.443 108 W S1 K2 is the Lorenz number for a free electron gas, r is the electrical conductivity and T is the absolute temperature. It is the phononic thermal conductivity jph that makes the main contribution to jtotal. Although the stress in CrN associated with a change of spin ordering has a purely magnetic origin, it can partly be relieved by structural transitions [19]. Filippetti et al. reported that the magnetic behavior of this material is more complex because of the competition between the superexchange (AFM) and the double-exchange (FM) mechanism. The major magnetic contribution comes from the direct nearest neighbor Cr–Cr interactions mediated by the Cr t2g orbitals inducing antiferromagnetism. Ferromagnetism originates from the next nearest neighbor Cr–N–Cr interactions and makes a small contribution to the magnetic properties of CrN. Assuming a weak AFM–FM competition, a small orthorhombic distortion would result in a sufficient energy gain to stabilize one of the magnetic structures [19]. The magnetic studies of this work confirm an antiferromagnetic–paramagnetic transition in CrO0.09(3)N0.90(7) at TN = 262 K (Fig. 7). The field cooling (FC) susceptibility data were fitted in the paramagnetic region above
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
1138
Fig. 6. Temperature dependence of the thermal conductivity of the CrO0.09(3)N0.90(7) with the electronic contribution (jel) to the total thermal conductivity in the inset.
Fig. 4. (a) Temperature dependence of the electrical resistivity with a close-up around TN in the inset and (b) linearization by plotting ln q vs. 1000/T, revealing an activation energy of Ea,1 34 meV above TN and of Ea,2 24 meV below TN.
Fig. 7. Temperature dependence of the molar magnetic susceptibility measured at 0.1 T on field-cooled (FC) and zero field-cooled (ZFC) samples of CrO0.09(3)N0.90(7) and the inset of the magnetization (M) in dependence of the magnetic field intensity at different temperatures.
Fig. 5. Temperature dependence of the Seebeck coefficient with a close-up around TN in the inset.
TN using the modified Curie–Weiss formula:v ¼ v0 þ l2
eff 0:12512 T H ;where v0 is the temperature-independent susceptibility containing Landau diamagnetism and Pauli susceptibility contributions, T is the absolute temperature, H is the Weiss constant and leff is the effective magnetic moment expressed in Bohr magnetons lB. Assuming the local spins S = 1 (Cr2+ LS) and S = 3/2 (Cr3+), the theoretical effective magnetic moment of CrO0.09(3)N0.90(7) would be leff = 3.78 lB which is close to the experimental effective magnetic moment leff =
3.63 lB. The evaluated Weiss constant is H = 1448 K and the susceptibility v0 = 2.0 105 emu mol1 Oe1. The negative sign of the Weiss constant indicates the predominance of antiferromagnetic interactions. The deviant peak at T 54 K is related to a melting of residual molecular oxygen in the sample [28] and/or in the PPMS measurement chamber [20]. The magnetization M was measured in a temperature range of 180 K < T < 300 K and a magnetic field range of 5 T < H < 5 T. M vs. H curve acquired is centrosymmetric with respect to the point M = 0 lB and H = 0 T. The inset of Fig. 7 shows the positive part of the magnetization. The data show that magnetization is a linear function of the applied external magnetic field. An increase of the magnetization by a factor of 0.37 was induced around TN (240 K < T < 260 K) at H = 5 T. The temperature dependence of the heat capacity Cp along with a fitting curve is presented in Fig. 8a. A clear peak at T = 262 K indicates the PM–AFM and the structural transition. Curve fitting was based on the modified
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
Debye–Einstein (DE) model. The DE model contains one triple degenerated Debye mode described by three acoustic branches and three Einstein modes with nine optical branches [29]: ! 3 Z xD 3 X 3 x4D ex wEi xEi exEi Cp ¼ R dx þ 2 xEi 1Þ2 xD ðex 1Þ 0 i1 ðe þ TV a2 b1
ð1Þ
where xD = HD/T, xEi = HEi/T and wEi = 1, 5, 3 are the optical branches which have to be assessed empirically in order to obtain the best fitting. The last term represents the lattice dilatation, where a is the volume thermal expansion coefficient and b is the isothermal compressibility factor for the unit cell volume V = 21.531 106 m3 measured at T = 300 K. As an approximation we used the data corresponding to CrN: a = 5.75 106 K1 [6] and b = 7 1012 Pa1 [30]. Consequently, the contribution of Cdilat to the total heat capacity is 2.1 J mol1 K1 at T = 700 K. According to Dulong–Petit’s law, the limiting Cp at high temperatures (e.g. above 300 K) for a system containing n atoms per molecules is C DP p ¼ 3Rn, where R is the universal gas constant. With this we obtain a limiting heat capacity of Cp = 49.9 J mol1 K1 CrO0.09(3)N0.90(7).
1139
The experimental Cp approaches a limit of Cp = 50.3 J mol1 K1 at T = 700 K. The peak corresponding to the PM–AFM and the structural transition (220 K < T < 306 K) was excluded from the fitting calculations. The resulting parameters for CrO0.09(3)N0.90(7) are: HD = 663 K, HE1 = 287 K, HE2 = 568 K, HE3 = 1116 K (standard deviation k2 < 0.25). In Fig. 8b Cp/T is plotted vs. T2 in the low-temperature region. The fitting curve is given by Cp = cT + bT3 + dTn. The first term is the electronic part of the total heat capacity represented by the Sommerfield coefficient c. The second term is the lattice (phonon) part represented by 4 b ¼ 125 pHRN 3 , where N is the number of atoms per mol and D
HD is the Debye temperature. The last term dTn represents the magnetic contribution of the spin waves to the total heat capacity. The Sommerfeld coefficient c and Debye temperature were determined from the low-temperature fitting resulting in c = 0.23 mJ mol1 K1 and HD = 745 K. The small Sommerfeld coefficient, typical for metallic compounds, is in agreement with the small band gap demonstrated by the electrical resistivity data. However, the density of states (DOS) at the Fermi level, determined by LSDA + U calculations, shows that CrN might be at the transition to a charge-transfer insulator [31]. The Debye temperature determined from the low-temperature Cp data (HD = 745 K) is higher than that determined from the hightemperature data (HD = 663 K). This finding might be attributed to the involvement of localized spins in the stiffness of the lattice. The softening of the lattice seems to be closely related to the magnetic and structural transition. Thus, the magnetic contribution (dTn) to the heat capacity dominates at low temperatures with a calculated coefficient d = 0.26 mJ mol1 K3. The fitting converged to n = 2, which is a typical value for an A-type antiferromagnetic structure. The respective contributions to the total heat capacity at low temperatures are plotted in Fig. 8b. 4. Conclusions
Fig. 8. (a) Temperature dependence of the heat capacity (Cp) of CrO0.09(3)N0.90(7) fitted by the Debye–Einstein model. (b) Low-temperature region of the heat capacity (Cp) with the fitting function, lattice, electronic and magnetic contributions to heat capacity of CrO0.09(3)N0.90(7).
Polycrystalline CrO0.09(3)N0.90(7) was synthesized by ammonolysis of Cr2O3 to investigate the magnetic influence on thermoelectric properties of CrO0.09(3)N0.90(7) as a potential n-type thermoelectric material. The high-temperature phase (above TN) of the material has a cubic structure ˚ , in contrast to with space group Fm-3m (a = 4.1506(2) A ˚ a = 4.14 ± 0.01 A for CrN). The TEM studies show that the particle size is ranging from 30 to 150 nm. CrO0.09(3)N0.90(7) is stable up to T 700 K determined by TGA/DTA measurement exceeding the stability range of conventional Bi2Te3-based thermoelectric materials. An antiferromagnetic–paramagnetic transition associated with an orthorhombic–cubic structural transition at TN = 262 K was observed during the measurements of: (i) the electrical and thermal transport, (ii) the magnetic susceptibility, and (iii) the heat capacity. The PM–AFM and the structural transition have significant influence on the lowering of
1140
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
thermal conductivity, which is desirable for thermoelectric materials. The transition temperature TN = 262 K of CrO0.09(3)N0.90(7) is very similar, but slightly lower compared to values of 273 K < TN < 286 K reported for pure CrN due to the presence of oxygen in the lattice. As expected, the oxynitride shows similar properties as the nitride. The electrical resistivity corresponds to that of a semiconductor showing a minimum value of q 2.8 mX cm at T = 300 K. The electrical resistivity is influenced by the magnetic and structural transition and shows a hysteresis during the cooling and heating cycles, indicating a first order character of the transition. Our results show that CrO0.09(3)N0.90(7) has an activation energy of Ea,1 34 meV above TN and Ea,2 24 meV below TN. Both values are smaller compared to CrN (Ea 70 meV above TN). The Seebeck coefficient of the sample is negative suggesting electrons as the main charge carriers. In the temperature range of 255 K < T < 265 K the thermopower values abruptly change from 56 lV K1 (below TN) to 46 lV K1 (at TN) in response to the magnetic and structural transition. The Seebeck coefficient value of CrO0.09(3)N0.90(7) (S = 48 lV K1 at T = 300 K) is lower than that of CrN (S = 135 lV K1 at T = 300 K) confirming a significantly higher charge carrier concentration. Nevertheless, the power factor of 1.7 104 W K2 m1 at T = 590 K and the lowering of the thermal conductivity above TN shows that CrO0.09(3)N0.90(7) could be a good candidate material for the n-type leg of a thermoelectric converter, but more work has to be done to improve the rather low ZT values. The effective magnetic moment leff = 3.63 lB calculated from the magnetic susceptibility data above TN is in agreement with the theoretical value of leff = 3.78 lB. The Weiss constant is H = 1448 K, indicating the predominance of antiferromagnetic interactions in CrO0.09(3)N0.90(7). The analysis of the experimental heat capacity data in the temperature range of 2.3 K < T < 700 K using the DE model resulted in the parameters HD = 663 K, HE1 = 287 K, HE2 = 568 K and HE3 = 1116 K with negligible contribution of free electrons and spin waves, while data analysis of the heat capacity at low temperature shows that the magnetic part (spin waves) of the heat capacity dominates. Acknowledgements We thank the Swiss Federal Office of Energy and Swiss National Foundation for financial support, E. Sˇantava´ for
the heat capacity measurement and R. Figi for the O/N content analysis. References [1] Terasaki I, Sasago Y, Uchinokura K. Phys Rev B 1997;56:12685. [2] Limelette P, Hardy V, Auban-Senzier P, Je´rome D, Flahaut D, He´bert S, et al. Phys Rev B 2005;71:233108. [3] Tomesˇ P, Robert R, Trottmann M, Bocher L, Aguirre MH, Hejtma´nek J, et al. J Electr Mater 2010;39:1696. [4] Logvinovich D, Hejtma´nek J, Marysˇko M, Homazava N, Tomesˇ P, Ebbinghaus SG, et al. J Appl Phys 2009;105:023522. [5] Ebbinghaus SG, Abicht HP, Dronskowski R, Mu¨ller T, Reller A, Weidenkaff A. Prog Solid State Chem 2009:37173. [6] Janssen GCAM, Tichelaar FD, Visser CCG. J Appl Phys 2006;100:093512. [7] Quintela CX, Rivadulla F, Rivas J. Appl Phys Lett 2009;94:152103. [8] Suzuki K, Kaneko T, Yoshida H, Morita H, Fujimori H. J Alloys Compd 1995;224:232. [9] Miao MS, Lukashev P, Herwadkar A, Lambrecht WRL. Phys Status Solid C 2005;7:2516. [10] Suzuki K, Morita H, Kaneko T, Yoshida H, Fujimora H. J Alloys Compd 1993;201:11. [11] Suzukia K, Kaneko T, Yoshida H, Obi Y, Fujimori H, Morita H. J Alloys Compd 2000;306:66. [12] Ha¨glund J, Grimvall G. Phys Rev B 1991;43:14400. [13] Constantin C, Haider MB, Ingram D, Smith AR. Appl Phys Lett 2004;85:6371. [14] Inumaru K, Koyama K, Miyaki Y, Tanaka K, Yamanaka S. Appl Phys Lett 2007;91:152501. [15] Inumaru K, Koyama K, Imo-oka N, Yamanaka S. Phys Rev B 2007;75:054416. [16] Browne JD, Liddell PR, Street R, Mills T. Phys Status Solid A 1970;1:715. [17] Mavromaras A, Matar S, Siberchicot B, Demazeau G. J Magn Magn Mater 1994;134:34. [18] Corliss LM, Elliott N, Hastings JM. Phys Rev 1960;117:929. [19] Filippetti A, Hill NA. Phys Rev Lett 2000;85:5166. [20] Logvinovich D, Borger A, Dobeli M, Ebbinghaus SG, Reller A, Weidenkaff A. Prog Solid State Chem 2007;35:281. [21] Rodriguez-Carvajal J. Physica B 1993;192:55. [22] Gruner W, Fresenius J. Anal Chem 1999;365:597. [23] Hejtma´nek J, Jira´k Z, Marysˇko M, Martin C, Maignan A, Hervieu M, et al. Phys Rev B 1999;60:14057. [24] Shannon RD. Acta Cryst A 1976;32:751. [25] Lu F-H, Chen H-Y. Thin Solid Films 2001;398:368. [26] Suzuki K, Kaneko T, Yoshida H, Obi Y, Fujimori H. J Alloys Compd 1998;280:294. [27] Bhobe PA, Chainani A, Taguchi M, Takeuchi T, Eguchi R, Matsunami M, et al. Phys Rev Lett 2010;104:236404. [28] Freiman YA, Jodl HJ. Phys Rep 2004;401:1. [29] Svoboda P, Leitner J, Sedmidubsky´ D. Int J Mater Res 2009;9:1246. [30] Knı´zˇek K, Hejtma´nek J, Jira´k Z, Tomesˇ P, Henry P, Andre´ G. Phys Rev B 2009;79:134103. [31] Herwadkar A, Lambrecht WRL. Phys Rev B 2009;79:035125.
Acta Materialia 59 (2011) 1134–1140 www.elsevier.com/locate/actamat
Magnetic influence on thermoelectric properties of CrO0.1N0.9 Petr Tomesˇ a, Dmitry Logvinovich a, Jirˇ´ı Hejtma´nek b, Myriam H. Aguirre a, Anke Weidenkaff a,⇑ a
Solid State Chemistry and Catalysis, Empa, Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600, Duebendorf, Switzerland b Institute of Physics of ASCR, v.v.i, Na Slovance 2, 182 21 Praha 8, Czech Republic Received 18 August 2010; received in revised form 20 October 2010; accepted 20 October 2010 Available online 17 November 2010
Abstract Thermoelectric polycrystalline chromium oxynitride CrO0.09(3)N0.90(7) (with cubic symmetry and space group Fm-3m) samples were successfully prepared by thermal ammonolysis of chromium oxide. This potential n-type thermoelectric material shows a temperature dependence of the Seebeck coefficient (S = 48 lV K1 at 300 K) that is typical for degenerate semiconductors; the electrical resistivity (q300 2.8 mX cm), however, is highly influenced by the grain boundaries and low density (60%) of the sample. The material exhibits a clear antiferromagnetic–paramagnetic transition simultaneously with an orthorhombic–cubic structural transition at TN = 262 K. The magnetic transition affected the material’s thermal conductivity, electrical resistivity, Seebeck coefficient, magnetic moment and specific heat. Heat capacity measurements at low temperatures revealed the predominance of the magnetic contribution due to spin waves. Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Chromium oxynitride; Electrical properties; Thermal conductivity; Magnetic properties; Specific heat
1. Introduction Conventional thermoelectric materials, such as tellurides, are limited for applications at elevated temperatures in air due to evaporation and corrosion processes. Thus, for the last few years alternative ceramic materials have been considered and developed for high-temperature thermoelectric applications [1–3]. Among them transition metal oxynitrides are a fascinating class of ceramic materials with interesting thermoelectric, magnetic and electronic properties [4]. They can be produced by thermal ammonolysis of oxide precursors [5]. Transition metal nitrides are chemical- and corrosionresistant and possess good mechanical properties [6]. In addition, they exhibit relatively low electrical resistivities associated with large Seebeck coefficients (e.g. CrN: S = 135 lV K1 at T = 300 K). Thus, CrN can be considered as a promising thermoelectric material [7]. 3d tran⇑ Corresponding author. Tel.: +41 79 751 6883; fax: +41 44 823 40 19.
E-mail address: anke.weidenkaff@empa.ch (A. Weidenkaff).
sition metal nitrides exist in different crystal lattice structures. Most of them (e.g. ScN, TiN, VN, CrN) have a rock-salt (RS) structure with a small lattice constant ˚ ); others (i.e. MnN, FeN and CoN) crystallize (a < 4.25 A in the zinc blende (ZB) structure with a larger lattice ˚ < a < 4.6 A ˚ ) [8–12]. The influence of the constant (4.2 A lattice constants and the cohesive energy on the magnetic properties of 3d transition metal nitrides was reported in Ref. [11]. Cohesive energy calculations confirm that early transition metal nitrides (ScN, TiN, VN, CrN) prefer a RS crystal structure, and the later metal nitrides (from MnN) a ZB structure. The influence of the lattice parameter on the magnetic moment is stronger in the RS than in the ZB crystal structure due to possible nanosized precipitates in the ZB phase inducing ferromagnetic (FM) ordering. 3d transition metal nitrides show interesting properties like antiferromagnetism, Pauli paramagnetism and superconductivity [13,8]. For example, thin films of solid solutions of Cr1xTixN have a large magnetoresistance [14]. In the range of 0.28 6 x 6 0.50 these compounds are ferromagnetic, which results from the competition between
1359-6454/$36.00 Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actamat.2010.10.046
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
antiferromagnetic (AFM) interactions of half-filled Cr t2g orbitals and interactions governed by the Cr (Ti)–N–Cr (Ti) double-exchange mechanism. Using epitaxial thin films, it was shown that the orientations of the thin films influence the structural transformation and thereby the material properties regardless of the substrate material [15]. CrN is the only nitride that shows an antiferromagnetic– paramagnetic transition with a Neel temperature of 273 K < TN < 286 K, a magnetic moment of 2.4 lB and metallic behavior [16–18]. At the same time, CrN undergoes a structural transition from a low-temperature orthorhombic structure (with a small orthorhombic distortion below TN) to a high-temperature cubic structure accompanied by a 0.59% change in the unit cell volume [16,18]. Using the local spin density functional theory, the magnetic structure of CrN was described as (1 1 0) FM layers with alternating spins every two layers, so that the system is AFM overall. The magnetic behavior of CrN indicates the competition between the superexchange (AFM) and the double-exchange (FM) mechanism [19]. The magnetic and structural transitions are also revealed by a clear hysteresis of the electrical and thermal transport properties of CrN. Thus, a similar behavior is expected of CrO0.09(3)N0.90(7). The aim of this work is to study the thermoelectrical properties of CrO0.09(3)N0.90(7), i.e. the Seebeck coefficient (S), the electrical resistivity (q) and the thermal conductivity (j) with regard to a possible thermoelectric application. The influence of the magnetic properties on the heat capacity and the transport properties is explained by drawing an analogy to the Cr2+/Cr3 mixed valence compounds Cr1xTixN and Cr1xVxN. 2. Experimental Cr(NO3)39H2O was dissolved in deionized water and precipitated in a NH3 (28–30% aq.) solution. The precipitate was washed twice with deionized water and dried at T = 373 K for 12 h and T = 723 K for 1 h to yield a single phase Cr2O3 as confirmed by X-ray powder diffraction (XRPD). The oxide was reacted with ammonia (PanGas, 99.98%, 200 ml min1) at T = 1173 K and T = 1268 K for 4 h and 2 h, respectively. The reactions were carried out in a rotation cavity quartz (SiO2) reactor [20] with an internal diameter of 30 mm. The ceramic pellets were obtained by uniaxial pressing of the ammonolized powder at 10 bar followed by sintering at T = 1268 K for 2 h under ammonia. The sample was cooled down to room temperature under ammonia at a cooling rate of 10 K min1. The bulk sample had a density of 60% of the theoretically achievable density. Structure, phase composition and size-strain parameters of the starting and product powders were studied by X-ray powder diffraction (XRPD) using a Phillips X’Pert PRO MPD H–H System equipped with a linear detector X’Celerator. Crystallographic parameters were obtained from Rietveld refinement of the XRPD data collected in angular
1135
range of 25° < 2H < 135° in 0.017° increments and a counting time of 250 s per step. The background was modeled by a sixth-degree polynomial function and the diffraction peak shape by a pseudo-Voigt profile. All the refinements were performed using FULLPROF [21]. Powder samples were dispersed in methanol, dropped on carbon-film coated copper grids and dried in air. Transmission electron microscopy (TEM) was performed using a Philips CM 30 microscope at 300 kV equipped with an EDX INCA detector from Oxford Instruments. The O/N content of the sample was determined by hot gas extraction (HGE) using a LECO TC500 analyzer [22]. About 50 mg of the sample was placed in a tin container positioned in a double graphite crucible and heated up to T = 3273 K for carbothermal reduction. The nitrogen was measured as N2 by a thermal conductivity cell and oxygen was measured as CO2 in an IR cell. Silicon nitride and silicon oxide were used as calibration standards for nitrogen and oxygen, respectively. The thermal stability was measured by thermogravimetric analysis/differential thermal analysis (TGA/DTA) using a Netzsch STA 409 CD thermobalance in the temperature range of 300 K < T < 1300 K with a heating rate of 10 K min1 under a synthetic air atmosphere (20 vol.% O2/He). The Seebeck coefficient, thermal conductivity and electrical resistivity measurements were carried out using a self-built system [23] within the temperature range of 3 K < T < 300 K. The four-point steady-state method with separated sensors and power contacts was applied. The principle and details of low-temperature measurements can be found in Ref. [23]. At high temperatures (300 K < T < 850 K) the Seebeck coefficient and the electrical resistivity were measured in air using the RZ2001i Ozawa Science, Japan measurement system. The magnetic susceptibility was measured using a vibrating sample magnetometer (VSM) adaptor for the physical property measurement system (PPMS) from Quantum Design in the temperature range of 4 K < T < 400 K. The measurements were performed applying a field of 0.1 T on zero field-cooled (ZFC) and on field-cooled (FC) samples. Magnetization curves were obtained at T = 10 K in the field range of 5 T < H < 5 T. The high temperature heat capacity was studied at a heating rate of 20 K min1 and a gas flow rate of 50 ml min1under a synthetic air atmosphere (20 vol.% O2/He). A heat capacity puck for the PPMS was used in the temperature range of 2.3 K < T < 335 K and a Netzsch differential scanning calorimeter (DSC) 404 C Pegasus in the temperature range of 373 K < T < 700 K. 3. Results The chemical composition of the sample was determined to be CrO0.09(3)N0.90(7). The oxygen content was estimated using EDX spectroscopy coupled with TEM. The exact concentration was measured by hot gas extraction resulting
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
1136
in an oxygen molar fraction of 0.09, which was used as input for the Rietveld refinement of the XRPD data. Since the X-ray powder diffraction (XRPD) pattern of chromium oxynitride resembles the pattern of CrN at room temperature, the CrN model was applied for the Rietveld refinement. The value of the initial occupancy factor of Cr was set to 1, for oxygen and nitrogen to those determined by hot gas extraction and the atomic displacement parameter (ADP) of Cr was refined isotropically. Refinement parameters are summarized in Table 1. The refinement converged with Rwp = 3.79, Rp = 3.04, v2 = 1.18 and is plotted in Fig. 1, the inset showing the unit cell of CrO0.09(3)N0.90(7). The compound has a cubic crystal structure with Fm-3m space group and a lattice parameter of ˚ . As comparison, for CrN lattice paramea = 4.1506(2) A ˚ , 4.135 A ˚ and 4.149 A ˚ were published ters of a = 4.13 A previously [14,17,18]. The increase of the lattice parameter might be attributed to an increase of the average ionic radius on the Cr site in CrO0.09(3)N0.90(7) compared ˚ ). The influence of the to Cr3+ in CrN (D = 0.01035 A cation size on the lattice parameter is, however, partially compensated by the slightly smaller average anionic radius of (O0.09N0.91)2.91 compared to the N3 anion ˚ ) [24]. (D = 0.0072 A Temperature stability of CrO0.09(3)N0.90(7) was assessed by TGA/DTA. The results are shown in Fig. 2. The mass increase above T 700 K is probably due to oxidation of CrO0.09(3)N0.90(7) to Cr2O3 [25]. The process continues up to T 1028 K. Above this temperature, the analysis suggest the formation of a b-Cr2N phase [25,26]. Hence, CrO0.09(3)N0.90(7) has a temperature stability limit at T 700 K exceeding that of the conventional Bi2Te3-based thermoelectric materials. Fig. 3 shows the TEM analysis of the (high-temperature) cubic structure of CrO0.09(3)N0.90(7) in diffraction (a), low magnification (b) and high resolution (c,d) images. The diffraction pattern and the indexing planes in Fig. 3a are in agreement with the atomic model obtained by Rietveld refinement. The small particles have polygonal shapes with a size varying from 30 to 150 nm. Fig. 3d shows the high resolution image of a particle along the [1–10] zone axis with the fast Fourier transformation in the inset. The temperature dependence of the electrical resistivity in the temperature range of 3 K < T < 850 K is shown in Fig. 4a. The measurements reveal semiconducting characteristics up to a temperature of T 600 K with a minimum value of q 2.8 mX cm at T = 300 K. The hysteresis shown in the inset of Fig. 4a, occurred during the electrical
Table 1 Structural parameters of CrO0.09(3)N0.90(7) refined from XRPD data. Space ˚ , a = b = c = 90°, Rwp = 3.79, group: Fm-3m (a = b = c = 4.1506(2) A Rp = 3.04, v2 = 1.22). ˚ 2 Occupancy factor Site Name x Y z Biso, A Cr1 N1/O1
0 1/2
0 1/2
0 1/2
0.23 (2) 0.4
1.000 (0) 0.9/0.1
4a 4b
Fig. 1. Rietveld refinement of the X-ray powder diffraction data of CrO0.09(3)N0.90(7). Space group: Fm-3m. The observed intensities, calculated profile, difference curve and Bragg positions are shown. The inset shows the a model of the cubic crystal structure of CrO0.09(3)N0.90(7).
Fig. 2. TGA/DTA curve of CrO0.09(3)N0.90(7) in the sample was measured under oxygen with a heating rate of 10 K min1.
resistivity cycle measurement using a self-built system within the temperature range of 3 K < T < 300 K, indicates a first order transition and is in agreement with data reported for CrN [17]. The structural and magnetic phase transition from a low-temperature AFM orthorhombic to a high-temperature paramagnetic (PM) cubic structure takes place in a martensitic transformation [7]. In Fig. 4b ln q is plotted against 1/T in the temperature region. The data was linearly fitted above and below TN using the equation q(T) = q0 exp(Ea/2kBT) [22]. Activation energies of Ea,1 34 meV above TN and Ea,2 24 meV below TN were obtained. Both values are smaller than the activation energy of CrN with Ea 70 meV [12,27]. The Seebeck coefficient, as a measure of entropy per charge carrier, is a crucial parameter for characterizing thermoelectric materials. Therefore, it is important to know how the entropy, transported by a charge carrier, is generated when applying a temperature gradient along a CrO0.09(3)N0.90(7) sample, and which type of charge carrier dominates in this process. The Seebeck coefficient of
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
1137
Fig. 3. (a) Diffraction pattern and indexing in agreement with the atomic model obtained by Rietveld refinement, (b and c) detail showing the size and morphologies of CrO0.09(3)N0.90(7) particles, and (d) high resolution image of one particle along the [1–10] zone axis with the fast Fourier transformation in the inset.
the sample is negative and increases almost linearly with increasing temperature (Fig. 5). This indicates that electrons are the main charge carriers as in a degenerate semiconductor. In the temperature range of 255 K < T < 265 K the Seebeck coefficient jumps from 56 lV K1 (below TN) to 46 lV K1 (above TN). At T = 300 K the absolute value of the Seebeck coefficient of CrO0.09(3)N0.90(7) (S = 48 lV K1) is significantly smaller than that of CrN (S = 135 lV K1) but similar to that of Cr0.9V0.1N (S = 83 lV K1) [7]. The low electrical resistivity and a Seebeck coefficient of S 70 lV K1 results in a power factor PF (PF = S2/q) of 1.7 104 W K2 m1 at T = 590 K. Another important requirement for a good thermoelectric material is a very low thermal conductivity. The data of the total thermal conductivity measurements are plotted in Fig. 6. Initially, the total thermal conductivity jtotal increases with temperature. In the temperature range of 255 K < T < 265 K; jtotal suddenly decreases by 35% from 4.2 W m1 K1 to 2.7 W m1 K1, reflecting the influence of the magnetic and structural transition on the thermal transport [7]. The lowering of thermal conductivity in this temperature range is desirable for a possible thermoelectric application near room temperature. According to the classical model jtotal = jel + jph the total thermal conductivity is composed of an electronic and a phononic part. The
electronic part jel has only a minor contribution to the total thermal conductivity (inset of Fig. 6). jel can be estimated using the Wiedemann–Franz law: jel = L0rT, where L0 = 2.443 108 W S1 K2 is the Lorenz number for a free electron gas, r is the electrical conductivity and T is the absolute temperature. It is the phononic thermal conductivity jph that makes the main contribution to jtotal. Although the stress in CrN associated with a change of spin ordering has a purely magnetic origin, it can partly be relieved by structural transitions [19]. Filippetti et al. reported that the magnetic behavior of this material is more complex because of the competition between the superexchange (AFM) and the double-exchange (FM) mechanism. The major magnetic contribution comes from the direct nearest neighbor Cr–Cr interactions mediated by the Cr t2g orbitals inducing antiferromagnetism. Ferromagnetism originates from the next nearest neighbor Cr–N–Cr interactions and makes a small contribution to the magnetic properties of CrN. Assuming a weak AFM–FM competition, a small orthorhombic distortion would result in a sufficient energy gain to stabilize one of the magnetic structures [19]. The magnetic studies of this work confirm an antiferromagnetic–paramagnetic transition in CrO0.09(3)N0.90(7) at TN = 262 K (Fig. 7). The field cooling (FC) susceptibility data were fitted in the paramagnetic region above
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
1138
Fig. 6. Temperature dependence of the thermal conductivity of the CrO0.09(3)N0.90(7) with the electronic contribution (jel) to the total thermal conductivity in the inset.
Fig. 4. (a) Temperature dependence of the electrical resistivity with a close-up around TN in the inset and (b) linearization by plotting ln q vs. 1000/T, revealing an activation energy of Ea,1 34 meV above TN and of Ea,2 24 meV below TN.
Fig. 7. Temperature dependence of the molar magnetic susceptibility measured at 0.1 T on field-cooled (FC) and zero field-cooled (ZFC) samples of CrO0.09(3)N0.90(7) and the inset of the magnetization (M) in dependence of the magnetic field intensity at different temperatures.
Fig. 5. Temperature dependence of the Seebeck coefficient with a close-up around TN in the inset.
TN using the modified Curie–Weiss formula:v ¼ v0 þ l2
eff 0:12512 T H ;where v0 is the temperature-independent susceptibility containing Landau diamagnetism and Pauli susceptibility contributions, T is the absolute temperature, H is the Weiss constant and leff is the effective magnetic moment expressed in Bohr magnetons lB. Assuming the local spins S = 1 (Cr2+ LS) and S = 3/2 (Cr3+), the theoretical effective magnetic moment of CrO0.09(3)N0.90(7) would be leff = 3.78 lB which is close to the experimental effective magnetic moment leff =
3.63 lB. The evaluated Weiss constant is H = 1448 K and the susceptibility v0 = 2.0 105 emu mol1 Oe1. The negative sign of the Weiss constant indicates the predominance of antiferromagnetic interactions. The deviant peak at T 54 K is related to a melting of residual molecular oxygen in the sample [28] and/or in the PPMS measurement chamber [20]. The magnetization M was measured in a temperature range of 180 K < T < 300 K and a magnetic field range of 5 T < H < 5 T. M vs. H curve acquired is centrosymmetric with respect to the point M = 0 lB and H = 0 T. The inset of Fig. 7 shows the positive part of the magnetization. The data show that magnetization is a linear function of the applied external magnetic field. An increase of the magnetization by a factor of 0.37 was induced around TN (240 K < T < 260 K) at H = 5 T. The temperature dependence of the heat capacity Cp along with a fitting curve is presented in Fig. 8a. A clear peak at T = 262 K indicates the PM–AFM and the structural transition. Curve fitting was based on the modified
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
Debye–Einstein (DE) model. The DE model contains one triple degenerated Debye mode described by three acoustic branches and three Einstein modes with nine optical branches [29]: ! 3 Z xD 3 X 3 x4D ex wEi xEi exEi Cp ¼ R dx þ 2 xEi 1Þ2 xD ðex 1Þ 0 i1 ðe þ TV a2 b1
ð1Þ
where xD = HD/T, xEi = HEi/T and wEi = 1, 5, 3 are the optical branches which have to be assessed empirically in order to obtain the best fitting. The last term represents the lattice dilatation, where a is the volume thermal expansion coefficient and b is the isothermal compressibility factor for the unit cell volume V = 21.531 106 m3 measured at T = 300 K. As an approximation we used the data corresponding to CrN: a = 5.75 106 K1 [6] and b = 7 1012 Pa1 [30]. Consequently, the contribution of Cdilat to the total heat capacity is 2.1 J mol1 K1 at T = 700 K. According to Dulong–Petit’s law, the limiting Cp at high temperatures (e.g. above 300 K) for a system containing n atoms per molecules is C DP p ¼ 3Rn, where R is the universal gas constant. With this we obtain a limiting heat capacity of Cp = 49.9 J mol1 K1 CrO0.09(3)N0.90(7).
1139
The experimental Cp approaches a limit of Cp = 50.3 J mol1 K1 at T = 700 K. The peak corresponding to the PM–AFM and the structural transition (220 K < T < 306 K) was excluded from the fitting calculations. The resulting parameters for CrO0.09(3)N0.90(7) are: HD = 663 K, HE1 = 287 K, HE2 = 568 K, HE3 = 1116 K (standard deviation k2 < 0.25). In Fig. 8b Cp/T is plotted vs. T2 in the low-temperature region. The fitting curve is given by Cp = cT + bT3 + dTn. The first term is the electronic part of the total heat capacity represented by the Sommerfield coefficient c. The second term is the lattice (phonon) part represented by 4 b ¼ 125 pHRN 3 , where N is the number of atoms per mol and D
HD is the Debye temperature. The last term dTn represents the magnetic contribution of the spin waves to the total heat capacity. The Sommerfeld coefficient c and Debye temperature were determined from the low-temperature fitting resulting in c = 0.23 mJ mol1 K1 and HD = 745 K. The small Sommerfeld coefficient, typical for metallic compounds, is in agreement with the small band gap demonstrated by the electrical resistivity data. However, the density of states (DOS) at the Fermi level, determined by LSDA + U calculations, shows that CrN might be at the transition to a charge-transfer insulator [31]. The Debye temperature determined from the low-temperature Cp data (HD = 745 K) is higher than that determined from the hightemperature data (HD = 663 K). This finding might be attributed to the involvement of localized spins in the stiffness of the lattice. The softening of the lattice seems to be closely related to the magnetic and structural transition. Thus, the magnetic contribution (dTn) to the heat capacity dominates at low temperatures with a calculated coefficient d = 0.26 mJ mol1 K3. The fitting converged to n = 2, which is a typical value for an A-type antiferromagnetic structure. The respective contributions to the total heat capacity at low temperatures are plotted in Fig. 8b. 4. Conclusions
Fig. 8. (a) Temperature dependence of the heat capacity (Cp) of CrO0.09(3)N0.90(7) fitted by the Debye–Einstein model. (b) Low-temperature region of the heat capacity (Cp) with the fitting function, lattice, electronic and magnetic contributions to heat capacity of CrO0.09(3)N0.90(7).
Polycrystalline CrO0.09(3)N0.90(7) was synthesized by ammonolysis of Cr2O3 to investigate the magnetic influence on thermoelectric properties of CrO0.09(3)N0.90(7) as a potential n-type thermoelectric material. The high-temperature phase (above TN) of the material has a cubic structure ˚ , in contrast to with space group Fm-3m (a = 4.1506(2) A ˚ a = 4.14 ± 0.01 A for CrN). The TEM studies show that the particle size is ranging from 30 to 150 nm. CrO0.09(3)N0.90(7) is stable up to T 700 K determined by TGA/DTA measurement exceeding the stability range of conventional Bi2Te3-based thermoelectric materials. An antiferromagnetic–paramagnetic transition associated with an orthorhombic–cubic structural transition at TN = 262 K was observed during the measurements of: (i) the electrical and thermal transport, (ii) the magnetic susceptibility, and (iii) the heat capacity. The PM–AFM and the structural transition have significant influence on the lowering of
1140
P. Tomesˇ et al. / Acta Materialia 59 (2011) 1134–1140
thermal conductivity, which is desirable for thermoelectric materials. The transition temperature TN = 262 K of CrO0.09(3)N0.90(7) is very similar, but slightly lower compared to values of 273 K < TN < 286 K reported for pure CrN due to the presence of oxygen in the lattice. As expected, the oxynitride shows similar properties as the nitride. The electrical resistivity corresponds to that of a semiconductor showing a minimum value of q 2.8 mX cm at T = 300 K. The electrical resistivity is influenced by the magnetic and structural transition and shows a hysteresis during the cooling and heating cycles, indicating a first order character of the transition. Our results show that CrO0.09(3)N0.90(7) has an activation energy of Ea,1 34 meV above TN and Ea,2 24 meV below TN. Both values are smaller compared to CrN (Ea 70 meV above TN). The Seebeck coefficient of the sample is negative suggesting electrons as the main charge carriers. In the temperature range of 255 K < T < 265 K the thermopower values abruptly change from 56 lV K1 (below TN) to 46 lV K1 (at TN) in response to the magnetic and structural transition. The Seebeck coefficient value of CrO0.09(3)N0.90(7) (S = 48 lV K1 at T = 300 K) is lower than that of CrN (S = 135 lV K1 at T = 300 K) confirming a significantly higher charge carrier concentration. Nevertheless, the power factor of 1.7 104 W K2 m1 at T = 590 K and the lowering of the thermal conductivity above TN shows that CrO0.09(3)N0.90(7) could be a good candidate material for the n-type leg of a thermoelectric converter, but more work has to be done to improve the rather low ZT values. The effective magnetic moment leff = 3.63 lB calculated from the magnetic susceptibility data above TN is in agreement with the theoretical value of leff = 3.78 lB. The Weiss constant is H = 1448 K, indicating the predominance of antiferromagnetic interactions in CrO0.09(3)N0.90(7). The analysis of the experimental heat capacity data in the temperature range of 2.3 K < T < 700 K using the DE model resulted in the parameters HD = 663 K, HE1 = 287 K, HE2 = 568 K and HE3 = 1116 K with negligible contribution of free electrons and spin waves, while data analysis of the heat capacity at low temperature shows that the magnetic part (spin waves) of the heat capacity dominates. Acknowledgements We thank the Swiss Federal Office of Energy and Swiss National Foundation for financial support, E. Sˇantava´ for
the heat capacity measurement and R. Figi for the O/N content analysis. References [1] Terasaki I, Sasago Y, Uchinokura K. Phys Rev B 1997;56:12685. [2] Limelette P, Hardy V, Auban-Senzier P, Je´rome D, Flahaut D, He´bert S, et al. Phys Rev B 2005;71:233108. [3] Tomesˇ P, Robert R, Trottmann M, Bocher L, Aguirre MH, Hejtma´nek J, et al. J Electr Mater 2010;39:1696. [4] Logvinovich D, Hejtma´nek J, Marysˇko M, Homazava N, Tomesˇ P, Ebbinghaus SG, et al. J Appl Phys 2009;105:023522. [5] Ebbinghaus SG, Abicht HP, Dronskowski R, Mu¨ller T, Reller A, Weidenkaff A. Prog Solid State Chem 2009:37173. [6] Janssen GCAM, Tichelaar FD, Visser CCG. J Appl Phys 2006;100:093512. [7] Quintela CX, Rivadulla F, Rivas J. Appl Phys Lett 2009;94:152103. [8] Suzuki K, Kaneko T, Yoshida H, Morita H, Fujimori H. J Alloys Compd 1995;224:232. [9] Miao MS, Lukashev P, Herwadkar A, Lambrecht WRL. Phys Status Solid C 2005;7:2516. [10] Suzuki K, Morita H, Kaneko T, Yoshida H, Fujimora H. J Alloys Compd 1993;201:11. [11] Suzukia K, Kaneko T, Yoshida H, Obi Y, Fujimori H, Morita H. J Alloys Compd 2000;306:66. [12] Ha¨glund J, Grimvall G. Phys Rev B 1991;43:14400. [13] Constantin C, Haider MB, Ingram D, Smith AR. Appl Phys Lett 2004;85:6371. [14] Inumaru K, Koyama K, Miyaki Y, Tanaka K, Yamanaka S. Appl Phys Lett 2007;91:152501. [15] Inumaru K, Koyama K, Imo-oka N, Yamanaka S. Phys Rev B 2007;75:054416. [16] Browne JD, Liddell PR, Street R, Mills T. Phys Status Solid A 1970;1:715. [17] Mavromaras A, Matar S, Siberchicot B, Demazeau G. J Magn Magn Mater 1994;134:34. [18] Corliss LM, Elliott N, Hastings JM. Phys Rev 1960;117:929. [19] Filippetti A, Hill NA. Phys Rev Lett 2000;85:5166. [20] Logvinovich D, Borger A, Dobeli M, Ebbinghaus SG, Reller A, Weidenkaff A. Prog Solid State Chem 2007;35:281. [21] Rodriguez-Carvajal J. Physica B 1993;192:55. [22] Gruner W, Fresenius J. Anal Chem 1999;365:597. [23] Hejtma´nek J, Jira´k Z, Marysˇko M, Martin C, Maignan A, Hervieu M, et al. Phys Rev B 1999;60:14057. [24] Shannon RD. Acta Cryst A 1976;32:751. [25] Lu F-H, Chen H-Y. Thin Solid Films 2001;398:368. [26] Suzuki K, Kaneko T, Yoshida H, Obi Y, Fujimori H. J Alloys Compd 1998;280:294. [27] Bhobe PA, Chainani A, Taguchi M, Takeuchi T, Eguchi R, Matsunami M, et al. Phys Rev Lett 2010;104:236404. [28] Freiman YA, Jodl HJ. Phys Rep 2004;401:1. [29] Svoboda P, Leitner J, Sedmidubsky´ D. Int J Mater Res 2009;9:1246. [30] Knı´zˇek K, Hejtma´nek J, Jira´k Z, Tomesˇ P, Henry P, Andre´ G. Phys Rev B 2009;79:134103. [31] Herwadkar A, Lambrecht WRL. Phys Rev B 2009;79:035125.