Effect of Ti substitution on the thermoelectric properties ... - IEEE Xplore

APPLIED PHYSICS LETTERS

VOLUME 72, NUMBER 16

20 APRIL 1998

Effect of Ti substitution on the thermoelectric properties of the pentatelluride materials M 12 x Tix Te5 „M5Hf, Zr… R. T. Littleton IV, Terry M. Tritt, C. R. Feger, J. Kolis, M. L. Wilson, and M. Marone College of Engineering and Sciences, Clemson University, Clemson, South Carolina 29634

J. Payne and D. Verebeli Department of Physical Sciences, South Carolina State University, Orangeburg, South Carolina

F. Levy Institut De Physique Appliquee, Laussanne, Switzerland

~Received 22 December 1997; accepted for publication 23 February 1998! The thermoelectric properties ~resistivity and thermopower! of single crystals of the low dimensional pentatelluride materials, HfTe5 and ZrTe5 , have been measured as a function of temperature from 10 K,T,320 K. The effect of small amounts of Ti substitutional doping (M 12x Tix Te5 , where M 5Hf, Zr! on the thermoelectric properties is reported here. A resistive transition occurs in the pentatellurides, as evidenced by a peak in the resistivity, T P '80 K for HfTe5 and T P '145 K for ZrTe5 . Both parent materials exhibit a large positive ~p-type! thermopower near room temperature which undergoes a change to negative ~n-type! below the peak temperature. The thermal conductivity is relatively low ~'5 W/m K! for the M Te5 materials. The Ti substitution affects the electronic properties strongly, producing a substantial shift in the peak temperature while the large values of thermopower remain essentially unaffected. These results warrant further investigation of these materials as candidates for low temperature thermoelectric applications. © 1998 American Institute of Physics. @S0003-6951~98!03716-4#

needs to possess a high dimensionless figure of merit, ZT, where ZT5 a 2 s T/l, where a is the Seebeck coefficient, s the electrical conductivity, T is the temperature in K and l the total thermal conductivity (l5l L 1l E ; the lattice and electronic contributions, respectively!. Materials of both n-type ~typical thermopower! and p-type ~positive thermopower! are necessary if a thermoelectric device is to be fabricated. The current state of-the-art materials typically possess a ZT'1 at their peak application or operating temperature. In the quest for new thermoelectric materials, many new classes of compounds and new synthetic techniques are being investigated with some of these showing promising results, however only at temperatures well above room temperature.7–14 The most promising materials for thermoelectric use have been semiconductors with carrier densities n in the range of 1019 carriers/cm3. In most semiconducting materials at temperatures far from a phase transition, the electrical conductivity and thermopower are related to the electron density of states near the Fermi energy g(E F ). The conductivity is proportional to g(E F ) while a is proportional to (1/g)dg/dE at E F . Hence, as n ~or g! is increased s typically increases while a decreases. This opens up a common route for optimization of the thermoelectric properties through doping to tune the number of carriers. In this scheme, high mobility carriers are especially desirable so that s can be large without also having a large carrier density. Low dimensional systems could provide a new route to the optimization of ZT. These systems can be very susceptible to van Hove singularities ~or cusps! in their density of states. Systems that are candidates for such low temperature

A considerable amount of research has been performed on materials based on the Bi2Te3 and Si12x Gex systems for thermoelectric applications.1–3 These materials have been extensively studied and optimized for their use in thermoelectric refrigeration ~down to '250 K with Bi2Te3) and power generation applications ~T.700 K with Si12x Gex ) and are currently the state-of-the-art materials. However, despite the thorough and comprehensive investigation of the traditional thermoelectric materials, there is still substantial room for improvement. The need for lower temperature (T,200 K! thermoelectric materials is especially acute, where, for example, lower temperature thermoelectric cooling packaging for HgCdTe infrared detectors could provide increased sensitivity and faster response than current technology for thermal sensing.4 In addition, as the field of ‘‘cryoelectronics’’ and ‘‘cold computing’’ grows the need for lower temperature ~100–200 K! thermoelectric materials will become necessary. The advantages of cold computing are discussed in a recent article by Sloan,5 where he states that ‘‘cooling is the fundamental limit to electronic system performance.’’ A severe limitation to cellular phone communications technology using superconducting narrow-band spectrum dividers to increase frequency band utilization is a reliable-low-maintenance cooling and refrigeration technology. Thermoelectric energy conversion utilizes the Peltier heat transferred when an electric current is passed through a thermoelectric material to provide a temperature gradient with heat being absorbed on the cold side and rejected at the sink, thus providing a refrigeration capability. Conversely, an imposed DT will result in a voltage or current, i.e., small scale power generation.6 A good thermoelectric material 0003-6951/98/72(16)/2056/3/$15.00

2056

© 1998 American Institute of Physics

Appl. Phys. Lett., Vol. 72, No. 16, 20 April 1998

thermoelectric materials are heavy fermion materials, Kondo systems and quasi-one-dimensional ~1D! materials. Low dimensional materials are specifically susceptible to electronic phase transitions and exotic transport properties which can add structure in g(E) near E F . Doping can produce very substantial effects in these types of materials and can drastically change their electronic transport. Quantum well systems take advantage of this low dimensional character through physical confinement in thin film structures to enhance the electronic properties of a given material ~i.e., the power factor, a 2 s ).7 In this letter the effect of substitutional doping on the thermoelectric properties of a class of quasi-1D conductors based on HfTe5 and ZrTe5 will be reported. We have performed investigations of a very interesting class of quasi-1D conductors known as the pentatellurides, HfTe5 and ZrTe5 . Resistivity and thermopower data are shown in Figs. 1 and 2, for the undoped HfTe5 and ZrTe5 , respectively, as a function of temperature. Electrical contact was made using Au wires bonded to the crystal with Au paint. All the thermopower data were corrected for the small contributions from the leads. Single crystals of both the undoped and doped materials were grown in similar conditions to previously reported methods.15 A stoichiometric ratio of the materials was sealed in a fused silica tubing with iodine ~'5 mg/mL! and placed in a tube furnace. The starting materials were placed at the center of the furnace and at the other end of the reaction vessel near the furnace to provide a temperature gradient. Typical crystals of these materials were obtained which were approximately 1.5 mm long and 100 m m in diameter with the preferred direction of growth along the a axis, as determined by face indexing. These are long chain systems with a crystal structure orthorhombic.16 They exhibit very anisotropic transport properties and the high conductivity axis is the growth axis. As seen in Fig. 1~a!, the resistivity for HfTe5 at first decreases and then increases as the temperature is decreased from room temperature, exhibiting a peak at around T P '80 K after which the resistivity falls rapidly as the temperature is further reduced. The ZrTe5 material behaves in a similar manner except that the peak occurs at a higher temperature, T P '145 K. The room temperature electrical conductivity of these materials is 3 (mV cm) 21 for HfTe5 and 1 (mV cm) 21 for ZrTe5 which is comparable to the best known thermoelectric materials. The thermopower of these materials also shows behavior indicative of their potential for thermoelectric use ~Figs. 1 and 2!. At high temperatures (T@T P ) pentatellurides display a large positive ~p-type! thermopower, '1100 m V/K. Near T P , the thermopower undergoes a dramatic change, passing through zero before reaching a negative ~n-type! peak at '2100 m V/K. Thus, these materials exhibit thermopower that is relatively large over a broad temperature range: exhibiting n-type (T,T P ) and p-type (T.T P) behavior depending on the temperature. Several aspects of the electrical transport properties of these materials, specifically in relation to the resistive peak, were studied in the early 1980’s, albeit not in relation to the materials’ properties for applications in thermoelectrics.17–22 Early data suggested a possible charge density wave ~CDW!

Littleton IV et al.

2057

FIG. 1. ~a! The normalized resistivity, r (T)/ r ~273 K!, and ~b! the absolute thermopower, a , as a function of temperature for single crystal HfTe5 and Hf0.95Ti0.05Te5 .

peak but no evidence of a CDW transition or CDW behavior was found. DiSalvo et al.18 performed Zr and Ta substitution for Hf in the HfTe5 material and found the doping and substitutions were able to substantially change the peak temperature, while also affecting the magnitude of the resistivity of these materials. This indicated that the transition appears most likely electronic in origin. The thermopower was not reported in that study. An investigation by Jones et al.17 found that the magnitude of the thermopower was highly sample dependent, probably due to small amounts of trace impurities or differences in growth conditions. Substantial pressure effects23 have also been observed in the pentatelluride materials with the thermopower below the peak, the n-type, being changed by 150% or more to values of approximately 2240 mV/K in ZrTe5 at T5120 K and P512 kbar while the resistivity decreased by a factor of 4. These trends yield an enhancement to the power factor ( a 2 s ) by an order of magnitude. Smaller changes are observed in HfTe5 at similar pressures. Uniaxial stress measurements exhibit substantial effects in both parent materials.24 These results support the idea of an electronic phase transition. We have recently reported the effect of isoelectronic substitution on the thermopower of Zr for Hf (Hf12x Zrx Te5 ).25 Through tuning the Zr concentration, T P is shifted systematically between those of the two parent compounds with little effect on the magnitudes of the thermopower and resistivity. Here good agreement between T P and the zero thermopower crossing temperature T 0 have been found with each shifting systematically with dopant concentration (x) and tracking with each other. Ti atoms are substantially smaller than either Hf or Zr and hence should produce some slight compression of the

2058

Littleton IV et al.

Appl. Phys. Lett., Vol. 72, No. 16, 20 April 1998

on the thermopower, electrical resistivity, and thermal conductivity of HfTe5 and ZrTe5 have been performed. These data show that the pentatellurides comprise a promising family of compounds for future investigation as thermoelectric materials. It has been shown that, through tuning either the Zr or Ti concentration in HfTe5 , the observed resistive transition temperature can be shifted rather uniformly from 150 down to 35 K. Additionally, this shift occurs without significantly decreasing the magnitude of either the p-type thermopower (T.T P ) or the n-type thermopower (T,T P ). While current data yield a maximum dimensionless figure of merit ZT ;0.2 at 150 K, a great deal of work remains. Other doping possibilities exist in these pentatelluride systems and many of these are currently under investigation. This system of materials shows much promise for future work and potential as a candidate for a low temperature thermoelectric material. Enhanced behavior of the thermoelectric properties at higher temperature ~T.300 K! is suggested by the data in Figs. 1 and 2; thus extensions of these measurements to much higher temperatures are also in progress.

FIG. 2. ~a! The normalized resistivity, r (T)/ r ~273 K!, and ~b! the absolute thermopower, a , as a function of temperature for single crystal ZrTe5 and Zr0.90Ti0.1Te5 .

The authors acknowledge support from the U.S. Army Research Office ~No. DAAG55-97-1-0267!. Facilities at SC State were developed under a grant from the DOE ~DOE No. DE-FG05-93ER45493!.

H. J. Goldsmid, Electronic Refrigeration ~Pion Limited, London, 1986!. CRC Handbook of Thermoelectrics, edited by D. M. Rowe ~CRC, Boca Raton, 1995!. 3 C. W. Wood, Rep. Prog. Phys. 51, 459 ~1988!. 4 A. W. Allen, Detector Handbook, Laser Focus World, March issue 1997. 5 J. Sloan, Superconductor Industry, Fall 1996, p. 30. 6 T. M. Tritt, Science 272, 1276 ~1996!. 7 L. D. Hicks and M. S. Dresselhaus, Phys. Rev. B 47, 12727 ~1993!. 8 J. P. Fleuriel, T. Calliet, and A. Borshchevsky, Proceedings of the XIII International Conference on Thermoelectrics ~AIP, Kansas City, MO, 1995!, pp. 40–44. 9 B. C. Sales, D. Mandrus, and R. K. Williams, Science 272, 1325 ~1996!. 10 D. T. Morelli, T. Caillat, J. P. Fleurial, A. Borchevsky, J. Vandersande, B. Chen, and C. Uher, Phys. Rev. B 51, 9622 ~1995!. 11 G. A. Slack and V. G. Toukala, J. Appl. Phys. 76, 1635 ~1994!. 12 G. Nolas, G. Slack, D. T. Morelli, T. M. Tritt, and A. C. Ehrlich, J. Appl. Phys. 79, 4002 ~1996!. 13 T. M. Tritt, G. S. Nolas, G. A. Slack, D. T. Morelli, A. C. Ehrlich, D. J. Gillespie, and J. L. Cohn, J. Appl. Phys. 79, 8412 ~1996!. 14 G. Nolas, G. Slack, V. G. Harris, and T. M. Tritt, J. Appl. Phys. 80, 6304 ~1996!. 15 S. Furuseth, L. Brattas, and A. Kjekshus, Acta Chem. Scand. 27, 2367 ~1973!. 16 S. Okada, J. Phys. Soc. Jpn. 51, 487 ~1982!. 17 T. E. Jones, W. W. Fuller, T. J. Wieting, and F. Levy, Solid State Commun. 42, 793 ~1982!. 18 F. J. DiSalvo, R. M. Fleming, and J. V. Waszczak, Phys. Rev. B 24, 2935 ~1981!. 19 W. W. Fuller, S. A. Wolf, T. J. Wieting, R. C. LaCoe, P. M. Chaiken, and C. Y. Huang, J. Phys. C 3, 1709 ~1983!. 20 M. Isumi, K. Uchinokura, E. Matsuura, and S. Harada, Solid State Commun. 42, 773 ~1982!. 21 D. W. Bullett, Solid State Commun. 42, 691 ~1982!. 22 G. N. Kamm, D. J. Gillespie, A. C. Ehrlich, D. L. Peebles, and F. Levy, Phys. Rev. B 35, 1223 ~1987!. 23 G. N. Kamm, D. J. Gillespie, A. C. Ehrlich, T. J. Wieting, and F. Levy, Phys. Rev. B 31, 7617 ~1985!. 24 E. P. Stillwell, A. C. Ehrlich, G. N. Kamm, and D. J. Gillespie, Phys. Rev. B 39, 1626 ~1989!. 25 R. T. Littleton IV, M. L. Wilson, C. Feger, J. Kolis, M. Marone, and T. M. Tritt, Proceedings of the XVI International Conference on Thermoelectrics, edited by A. Heinrich ~IEEE, Dresden, Germany, 1997! ~in press!. 26 D. T. Verebelyi, Rev. Sci. Instrum. 68, 2494 ~1997!. 1 2

lattice, possibly similar to external pressure. The Ti substitution is also isoelectronic, hence should not directly alter the carrier concentration of the compounds. The resistivity and thermopower, respectively, are shown for the HfTe5 and Hf0.95Ti0.05Te5 materials in Fig. 1~a! and 1~b!. This small amount of Ti substitution ~'5! shifts the peak temperature substantially from 80 K for HfTe5 to T P 538 K for Hf0.95Ti0.05Te5 , but in contrast to previous results, T 0 ( a 50), occurred at much higher temperatures, T 0 550 K. A high resistance state appears to subsist below the peak in contrast to all the other pentatelluride materials studied. This low temperature-high resistance state is similar to that observed in HfTe5 under stress.22 Since the origin of the main transition is still unknown, it is superfluous to speculate on this high resistance state, at this time. Figure 2 shows similar resistivity and thermopower data for pure ZrTe5 and Zr0.90Ti0.1Te5 . A nominal Ti substitution of 10% for Zr shifts the peak temperature from 145 K for ZrTe5 to T P 5110 K for Zr0.9Ti0.1Te5 , and T 0 is now coincident with T P . The strong metallic behavior is again evident below T P . Thus, substitutional doping of Ti for either Hf or Zr, leads to a systematic variation of the peak temperature from 38 to 145 K, while maintaining the relatively large values of thermopower at low temperature. Preliminary measurements of l were performed on single crystal whiskers. We used a technique and apparatus previously reported26 and these measurements yield l T values of no more than 5 W/mK at temperatures below 200 K. Additionally, the thermal conductivity appears to decrease by a factor of 2 due to the addition of Ti into the lattice. All these results have strong ramifications on the potential development of these materials for low temperature thermoelectric applications. In summary, measurements of the effect of Ti dopants