Properties of Thermoelectric Nanocomposite Bi 2 Te 3 Layers Prepared by PLD

Transcription

1 Sensors & Transducers 2014 by IFSA Publishing, S. L. Properties of Thermoelectric Nanocomposite Bi 2 Te 3 Layers Prepared by PLD 1, * Radek ZEIPL, 1 Miroslav JELÍNEK, 1 Tomáš KOCOUREK, 1 Jan REMSA, 1, 2 Jan VANIŠ, 3 Milan VLČEK 1 Institute of Physics of the Academy of Sciences of the Czech Republic, v.v.i., Na Slovance 2, Prague, Czech Republic 2 Institute of Photonics and Electronics of the Academy of Sciences of the Czech Republic, v.v.i., Chaberská 57, Prague, Czech Republic 3 Institute of Macromolecular Chemistry of the Academy of Sciences of the Czech Republic, v.v.i., Heyrovského nám. 2, Prague, Czech Republic 1 Tel.: , fax: * zeipl@fzu.cz Received: 19 October 2014 /Accepted: 28 November 2014 /Published: 31 December 2014 Abstract: Some properties of thermoelectric nano-layers prepared by Pulsed Laser Deposition from hot pressed Bi 2 Te 3 target are presented. The layers were prepared under various deposition conditions that include the substrate temperature during the deposition and the laser beam density, to study its influence on the quality of the surface. Crystallinity, composition and morphology of the layers are presented. Transport and thermoelectric properties such as electrical resistivity, the Seebeck coefficient, power factor and the thermoelectric figure of merit for the smoothest layers prepared at substrate temperature of 200 C applying laser beam density 3 Jcm -2 are also given. Nano crystallites observed on the layer s surface are studied by X-ray Diffraction and by Atomic Force Microscope. The problematic of a new method for a relative thermal conductivity characterization of thin thermoelectric layers and multi-layered structures in nanometre range using a scanning thermal microcsope working in an active constant current mode and the first results are described. Copyright 2014 IFSA Publishing, S. L. Keywords: Thermoelectric materials, Thin layers, Pulsed laser deposition, Thermal conductivity measurement. 1. Introduction Thin layers and multi-layered thermoelectric structures are potential candidates for many thermoelectric applications and nano devices from solid-state coolers and generators, thermoelectric transducers, thermocouples to thermal sensors and detectors. A relative efficiency of thermoelectric material is expressed in terms of the dimensionless figure of merit (ZT). It depends on the Seebeck coefficient, the electrical conductivity, the thermal conductivity and on the absolute temperature. Despite of no theoretical ZT limit, it seems that ZT of about 1 is a practical limit for a majority of conventional thermoelectric materials [1-2]. Bismuth and tellurium compounds, such as Bi 2 Te 3 etc., prepared in a form of very thin layers (nano layers) have been well established candidates for many thermoelectric applications

2 Knowledge of thermoelectric properties of materials with lateral resolution of tens of nanometres is necessary and the problem of accurate characterization of materials for thermoelectric nano devices is extremely important from the nano physics, nano electronics, nano mechanics and nano medicine point of view of. We have been working on the development of a simple and reliable method for relative thermal conductivity characterization of thin thermoelectric layers and multi-layered structures using a scanning thermal microscope working in constant current active DC and AC modes already for some time. For our experiments we need high quality thermoelectric layers and nano layers with thermal conductivity in the range from about one-tenth to few tens of Wm -1 K -1 with a very smooth surface. The increased surface roughness might cause inaccuracy and create difficulties such as artefacts etc., when using any scanning probe microscope measurement technique, including thermal microscope. As a suitable thermoelectric material for our experiments we have chosen Bi 2 Te 3 nano layers which were successfully prepared by several deposition methods including Pulsed Laser Deposition (PLD) in the past [3-5]. In this paper we present the influence of the main deposition conditions that include the substrate temperature (T S ) and the laser beam density (D S ) on the quality of layer surface. The X-ray Diffraction (XRD) patterns, the Wavelength Dispersive analysis (WDX) composition and the Atomic Force Microscope (AFM) surface morphology of the prepared layers are presented. The study was already presented during the TechConnect World 2014 conference [6]. Transport and thermoelectric properties studied on the smoothest layers prepared at T S =200 C applying D S =3 Jsm -2 together with the first results of the relative thermal conductivity measurement using a scanning thermal microscope working in active current DC mode are also described. 2. Layers Preparation and Measurement To study the influence of PLD deposition conditions on the resulting layer surface quality with the goal to find the conditions providing the smoothest surface, two series of layers with thickness ranging from 230 nm to 500 nm were prepared on Si (100) substrates from a Bi 2 Te 3 target. The first series layers were prepared at the substrate temperature T S =360 C applying various laser beam density D S (1, 2, 3, 4 and 5 Jcm -2 ). The original choice of T S was selected based on previously published experiments providing smooth and crystalline Bi 2 Te 3 layers [3-4]. The second series layers were prepared with different T S (200 C, 250 C, 300 C and 400 C) applying the same D S =3 Jcm -2, which was chosen based on results observed on the first series layers. The starting polycrystalline Bi 2 Te 3 material was synthesized from Bi and Te elements of 5N purity in evacuated silica ampoules at 1073 K for 48 hours. After verification of homogeneity of the prepared compound by means of X-ray powder diffraction, the polycrystalline ingot was crushed using an agate mortar and sieved to obtain particle size below 100 μm. The target for PLD deposition of 20 mm in diameter and of 2 mm in height was prepared by the hot pressing method (temperature 500 C, pressure ~60 MPa for 1 hour). The measured density of pressed target reached about % of the theoretical density. The powder X-ray diffraction patterns of the Bi 2 Te 3 compound used for the target were collected in the Bragg-Brentano geometry on the Bruker D8 Advance diffractometer equipped with a secondary graphite monochromator. CuKα radiation was used. The XRD spectrogram of the PLD target compound is depicted in Fig. 1. Fig. 1. The XRD spectrogram of the PLD target with only peaks corresponding Bi2Te3 compound. The basic schema of the experimental apparatus for PLD is depicted in Fig. 2. Conceptually and experimentally, PLD is an extremely simple method, probably the simplest of all the thin film growth techniques. A high power pulsed excimer KrF laser (COMPexProTM 205 F) radiation (1) is used as an external energy source to vaporize materials of the target (5) and to deposit a thin film. A set of optical components is utilized to focus the laser beam on the target surface (2, 3). After the laser pulse irradiation the temperature rises very rapidly (10 11 Ks 1 ) and the evaporation becomes non-equilibristic. For our experiments the substrates were cleaned from mechanical dirt in an ultrasonic cleaner. Next the substrates were subsequently cleaned in ethyl alcohol, acetone and toluene. Cleaning in vapours of boiling ethyl alcohol then completed the process. Two types of square shaped substrates - fused silica glass and Si (100) approximately mm in dimension were used as substrates. The substrates were finally annealed in an oven at a temperature of around 250 C. Deposition took place in Ar atmosphere (13 Pa), the target to substrate distance was set to 40 mm and the base vacuum of the coating 104

3 system was Pa. The layers were prepared by PLD using a KrF excimer laser λ = 248 nm, τ = 20 ns, repetition rate of 10 Hz and laser spot 2 1 mm 2. Fig. 2. The basic scheme of PLD apparatus: (1) laser beam, (2) mirrors, (3) focusing lens, (4) quartz window, (5) target holder, (6) substrate holder, (7) vacuum pump, (8, 9) Pirani and Penning vacuum gauges, respectively. electrical conductivity, the Seebeck coefficient and thermal conductivity (that make up ZT) are measured at the same place and at the same time with electrical current flowing through the layer. For intrinsic ZT evaluation we used the so-called variable thickness approach mathematically equivalent to measurement with different current density [5, 8], where ZT can be obtained on a layer across which a temperature difference is developed by the Peltier effect using a quasi-steady-state current [7]. We have been developing a new method for relative thermal conductivity characterization of thin thermoelectric layers in the nanometer scale. By the method the thin layers of different material prepared on the same type of substrate can be quickly compared from thermal conductivity point of view. The general schema is depicted in Fig. 3. Thickness and roughness of the layers were measured by an Alpha-step IQ mechanical profilometer (KLA TENCOR Co.). The uncertainty in the thickness estimation is about 10 % in the examined range of thicknesses partly the fact that the layer thickness in the centre of the sample is higher than at its edge. The layer roughness and homogeneity were characterized by the Atomic Force Microscope (AFM) Solver NEXT (NT-MDT) operating in dynamic regime with HA_NC tips and in contact mode with CSG10 tips. The layers surface parameters were calculated from a μm area with software NOVA PX. Transport properties, such as the in-plain electrical resistivity and the Seebeck coefficient, were measured on the smoothest prepared layers in the temperature range from room up to 200 C. The power factor was then calculated. Four square shaped contacts for the measurements were prepared by evaporating Ti in the sample corners. Pressed Pt/PtRh thermocouples with a diameter of 0.07 mm were used as leads. A conventional DC van der Pauw s method was used for the electrical resistivity measurement. The experimental error of this method is about 10 %. The Seebeck coefficient was determined from the variation of the electromotive force for different temperature gradients (from about 1 C to 8 C) across the layer. Both sides of the sample were in thermal contact with an independent wire resistant sub-heater that supplies the heat and induces the sample temperature gradient. The thermocouple junctions were bonded to each corner of the square shaped sample. The experimental error of the Seebeck coefficient measurement is about 15 %. The test of thermoelectric figure of merit (ZT) was conducted at room temperature by the Harman technique [7], in which parameters related to Fig. 3. The general schema of the relative thermal conductivity measurement using thermal microscope. For these experiments additional samples were prepared at T S =200 C with D S =3 Jcm -2. The layers were deposited only on the part of the substrate, so it has been possible to measure pure Si substrate and the system Bi 2 Te 3 layer-si substrate without any manipulation with sample after the system calibration. The method uses a scanning thermal microscope working in the active constant current mode, the microscope s cantilever being heated by an electrical current during the measurement. When physical contact between the measured sample and the cantilever is developed, the probe temperature decreases due to power dissipation into the sample. The temperature change leads to resistance change in the probe which is evaluate using a Wheaston bridge. A result of the measurement is the surface height relief and its thermal map. To increase the stability and sensitivity of the measurement the probe was driven with DC current pulses instead of constant DC current, thus only the tip of the cantilever was periodically heated up and not the whole system. This utilization of the system increases sensitivity of the measurement for about two orders in comparison to the constant DC current mode and minimize the influence of the measurement on an ambient interferences. 105

4 For this study Atomic Force Thermal Microscope (AFMTh) Commercial SPM - Veeco MultiMode system with NanoScope IVa controller equipped with three exchangeable scanners AS-0.5MF ("A") with a range of μm, AS-12VMF ("EV") with a range of μm and V version leak resistance ("J") with a range μm has been used. The two terminated measuring silicon probes of 100 µm width and 2-3 µm in thickness and 200 µm in length with a Tungsten resistive path produced also by Picocal Ltd. were utilized. crystals peak less than 200 nm above the surface, as seen in the 3D surface visualization depicted in the Fig. 4. The second are so called PLD droplets. They are much more spread over the surface and are much less frequent in comparison to the first type. The problem is that they are unfortunately much larger. Their height reaches 400 nm and their diameter is in the range of 200 nm to 300 nm. 3. Results and Discussion The Energy Dispersive X-ray analysis (EDX) and the Wavelength Dispersive X-ray analysis (WDX) were performed on several layers prepared at T S =360 C applying D S =5 Jcm -2 to obtain information on layer stoichiometry. The excess of Bi (Bi/Te ratio ~ 0.76) in comparison to the PLD target Bi 2 Te 3 stoichiometry (Bi/Te ratio ~ 0.67) was observed in all explored layers. The standard deviation was found to be ±1.5 %. Different deposition conditions or a different target would have to be selected to get the exact Bi 2 Te 3 stoichiometry, which is not necessary in our experiment. Our goal was to prepare thermoelectric layers with the smoothest possible surface. Thus, more valuable for us was the observation of a very granular surface covered with crystals of different dimensions. The surface roughness of all prepared layers was studied by AFM (scanned area um) and checked by Alpha-step mechanical profilometer. The results of these measurements are summarized in Table 1. Fig. 4. AFM 3D surface visualization of the surface (detail area um) of the layer prepared at TS=300 C applying DS=3 Jcm -2 where crystals and droplets peaking over the surface are visible. The influence of the laser beam density on the surface roughness appears to be negligible, however the smoothest surface was observed applying D S =3 Jcm -2 (see Fig. 5). Table 1. The comparison table of surface roughness Sa (Ra) value for each prepared nano layer measured by Alpha-step mechanical profilometer. Deposition Profilometer Layer AFM condition thick. TS DS Sa (Ra) Sa (Ra) [nm] [ C] [Jcm -2 ] [nm] [nm] Generally there are two types of objects observed on the layer s surface. The first are regularly shaped crystals of Bi 2 Te 3 or Bi (2+N) Te (3+M) compounds ranging in size from about 100 nm to 500 nm. These create a basic surface and were proved to be crystalline by the XRD measurement. Usually Fig. 5. The layer roughness Sa (Ra) versus laser beam density DS for samples prepared at the same substrate temperature TS=360 C. The impact of the substrate temperature is more important (see Fig. 6). With increased T S the layers become rougher. This was obvious and visible by eye for the layer prepared at T S =400 C. This layer had a different colour (matt white) in comparison to the other layers which are usually silver and shiny. The same trend of dependencies was observed using AFM as well as a mechanical profilometer (see Fig. 5 and Fig. 6). The resulting values are different which expected, due to the use of different methods with varying accuracy. In general the prepared layers were homogeneous from a macro point of view (the AFM scans were performed in several different positions over the 106

5 whole surface and the roughness and granularity remains the same). Fig. 6. The layer roughness Sa (Ra) versus substrate temperature TS for samples prepared applying the same laser beam density DS=3 Jcm -2. The smoothest layer surface was obtained at T S =200 C applying D S =3 Jcm -2 is depicted in Fig. 7. The roughest measurable layer prepared at T S =360 C applying D S =4 Jcm -2 is shown in Fig. 8. Fig. 7. AFM surface morphology (scanned area um) of the smoothest nano layer prepared at TS=200 C applying DS=3 Jcm -2. It is important to emphasise that these deposition conditions might not be optimal from thermoelectric properties point of view, as they influence the layer stoichiometry. Once we identified proper deposition conditions for high quality surface (i.e. T S =200 C, D S =3 Jcm -2 ), additional series of layers was prepared to check the transport and thermoelectric properties. The in-plain electrical resistivity and the Seebeck coefficient were measured in the temperature range from room temperature up to about 200 C. The resultant temperature dependencies are depicted in Fig. 9. and Fig. 10. All characterized layers were N type of conductivity, which was determined from the sign of the Seebeck coefficient. The measured temperature dependency of electrical resistivity was found to be flat in the studied temperature interval with average value around 1 mωcm and is depicted in Fig. 9. Positive temperature dependency was measured for the Seebeck coefficient (increasing Seebeck coefficient with temperature) with room temperature absolute value of the around 60 μv/k (value is negative - N type) and is shown in Fig. 10. The both values correspond to published results for EDC prepared layers [9]. The Bi 2 Te 3 Bridgman s method prepared bulk mono-crystal room temperature values of electrical resistivity and the Seebeck coefficient measured perpendicularly to crystallographic c axis are about 1.3 mωcm and about 227 μv/ C, respectively [10]. Our layers are rather Bi (2+N) Te (3+M) compounds than Bi 2 Te 3 stoichiometry and so the measured properties are different. In Fig. 11 there is temperature dependency of power factor with the room value of about WK -2 cm -1 calculated. From the measured properties point of view the prepared layers are average thermoelectric material. Room temperature ZT values published for single thin Bi 2 Te 3 layers prepared by different depositions techniques [9] are in the range from 0.07 to We observed room temperature value ZT around 0.13, which is close to the method resolution limit. Fig. 8. AFM surface morphology (scanned area um) of the roughest measurable layer prepared at TS=360 C applying DS=4 Jcm -2. The layer prepared at TS=360 C applying DS=1 Jcm -2 was not measurable at all. Fig. 9. Temperature dependency of measured electrical resistivity for about 330 nm thick N-type layers prepared from the Bi2Te3 target at Ts=200 C with Ds=3 Jcm 2. Solid line mean value; dashed lines max/min measured values. 107

6 Fig. 10. Temperature dependency of the measured Seebeck coefficient for about 330 nm thick N-type layers prepared from the Bi2Te3 target at Ts=200 C with Ds=3 Jcm 2. Solid line absolute mean value; dashed lines max/min measured values. Fig. 12. Relative thermal conductivity AFMTh measurement done on the boundary of Bi2Te3 layer of about 139 nm in thickness deposited on the part of Si substrate. On the left side there is the measured topography. On the right side there is the corresponding thermal response. In the bottom there are corresponding profiles - height (nm) and measured electrical response (mv). Although the system shows its tremendous possibilities it is still under development and mainly numerical evaluation of the tip to sample interface needs to be done. 4. Conclusions Fig. 11. Temperature dependency of calculated power factor (S 2 /R or S 2 σ) for about 330 nm thick N-type layers prepared from the Bi2Te3 target at Ts=200 C with Ds=3 Jcm 2. The experimental relative thermal conductivity characterization setup was tested on the two samples with the different thickness of the Bi 2 Te 3 layer (about 131 nm and 750 nm in thickness) deposited on the part of Si substrate. For each type of the surface i.e. Si substrate and Bi 2 Te 3 layer, on the prepared samples a contact and a far withdraw reference position, defined as a 170 μm distance from the sample surface, were measured (a Wheaston bridge voltage value). Equalizing of measured ratios on reference Si substrate gives a difference in measured values on Bi 2 Te 3 layers with different thickness. The difference for our samples was equal to 2.5 mv. Such system response is repeatable over the different parts of the sample. This sensitivity of the system is demonstrated in the picture Fig. 12, where there is a boundary Si substrate - Bi 2 Te 3 layer depicted. On the left Fig. 12 side the surface topography is shown, on the right side the thermal response. In the bottom there are profiles from the region of the boundary shown. The influence of the deposition conditions - the substrate temperature and laser beam density - on the quality of the surface (homogeneity and roughness) was analyzed on thermoelectric nano layers prepared by PLD from Bi 2 Te 3 target. The smoothest layers were deposited at T S =200 C while applying D S =3 Jcm -2. Such deposition conditions are not optimal from thermoelectric properties point of view as was proved by characterization and prepared layers are rather average thermoelectrics; because the layer stoichiometry is influenced by the deposition conditions. The response of the experimental relative thermal conductivity measurement system is repeatable and measurable, so the approach seems to be functional. The method still under development and more experiments with different material are necessary. Acknowledgements The project has been supported by Czech Grant Agency under P108/ S. References [1]. X. F. Tang, Q. Zhang, L. Chen, T. Goto, T. Hirai, Synthesis and thermoelectric properties of p-typeand n-type-filled skutterudite RYMXCo4 xsb12 108

7 (R:Ce,Ba,Y;M:Fe,Ni), Journal of Applied Physics, Vol. 97, 2005, pp [2]. B. C. Sales, D. Mandrus, B. C. Chakoumakos, V. Keppens, J. R. Thompson, Filled skutterudite antimonides: Electron crystals and phonon glasses, Physical Review B, Vol. 56, 1997, pp [3]. A. Dauscher, A. Thomy, M. Scherrer, Pulsed laser deposition of Bi2Te3 thin films, Thin Solid Films, Vol. 280, No. 1-2, 1996, pp [4]. R. Zeipl, S. Karamazov, M. Jelínek, P. Lošťák, M. Pavelka, Sz. Winiarz, R. Czajka, J. Vaniš, F. Šroubek, J. Zelinka, J. Walachová, in Proceedings of the 22 nd International Conference on Thermoelectrics, La Grande Motte, France, August 2003, pp [5]. J. Walachová, R. Zeipl, J. Zelinka, V. Malina, High room-temperature figure of merit of thin layers prepared by laser ablation from Bi2Te3 target, Applied Physics. Letters, Vol. 87, 2005, pp [6]. R. Zeipl, M. Jelínek, J. Remsa, T. Kocourek, J. Vaniš, M. Vlček, Thermoelectric properties of nanocomposite Bi2Te3 layers prepared by PLD, in Proceedings of the TechConnect World Inovation Conference, Washington, DC, USA, Vol. 1, Chapter 5, [7]. T. C. Harman, J. H. Cahn, M. J. Logan, Measurement of Thermal Conductivity by Utilization of the Peltier Effect, Journal of Applied Physics, Vol. 30, 1959, pp [8]. R. Venkatasubramanian, E. Siivola, T. Colpitts, B. O. Quinn, Thin Film Thermoelectric Devices with High Room Temperature Figures of Merit, Nature, Vol. 413, 2001, pp [9]. L. M. Gancalves, C. Couto, P. Alpuim, D. M. Rowe, J. H. Correia, Thermoelectric properties of Bi2Te3/Sb2Te3 thin films, Materials Science Fórum, Vols , 2006, pp [10]. T. Plechacek, J. Navratil, J. Horak, P. Lošták, Defect structure of Pb-doped Bi2Te3 single crystals, Philosophical Magazine, Vol. 84, 2004, pp Copyright, International Frequency Sensor Association (IFSA) Publishing, S. L. All rights reserved. ( 109