Thermoelectric properties of nanostructured bismuth–telluride thin
films grown using pulsed laser deposition
Phuoc Huu Le
a, Chien-Neng Liao
b, Chih Wei Luo
c,1, Jihperng Leu
a,⇑a
Department of Materials Science and Engineering, National Chiao Tung University, Hsinchu 30049, Taiwan, ROC
b
Department of Materials Science and Engineering, National Tsing-Hua University, Hsinchu 30013, Taiwan, ROC
c
Department of Electrophysics, National Chiao Tung University, Hsinchu 30010, Taiwan, ROC
a r t i c l e
i n f o
Article history:
Received 4 February 2014
Received in revised form 21 May 2014 Accepted 2 July 2014
Available online 10 July 2014 Keywords:
Bi2Te3
Thermoelectric properties Nanostructures
Pulsed laser deposition (PLD)
a b s t r a c t
Nanostructured n-type bismuth telluride (Bi2Te3) thin films were grown on SiO2/Si (1 0 0) substrates at
argon ambient pressure (PAr) of 80 Pa by pulsed laser deposition (PLD). The effects of film morphologies,
structures, and compositions on the thermoelectric properties were investigated. At a substrate temper-ature (Ts) of 220–340 °C, stoichiometric films with highly (0 0 l)-oriented and layered structures showed
the best properties, with a carrier mobilitylof 83.9–122.3 cm2/Vs, an absolute Seebeck coefficient |
a| of 172.8–189.7lV/K, and a remarkably high power factor (PF) of 18.2–24.3lW cm1K2. By contrast, the
Te-rich films deposited at Ts6120 °C with (0 1 5)-preferred orientations and columnar-small grain
struc-tures or the Te-deficient film deposited at 380 °C with Bi4Te5polyhedron structure possessed poor
prop-erties, withl< 10.0 cm2/Vs, |a| < 54lV/K, and PFs 6 0.44lW cm1K2. The morphology of highly (0 0 l)
oriented-layered structures and the stoichiometry predominantly contribute to the substantial enhance-ment ofland |a|, respectively, resulting in remarkable enhancement in PF.
Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction
Thermoelectric (TE) materials are of interests for applications as heat pumps and power generators[1–4]. The performance of TE materials is evaluated in terms of a dimensionless figure of merit, ZT =
a
2r
T/
j
, in whicha
,r
,j
, and T are the Seebeck coefficient, the electrical conductivity, the thermal conductivity, and absolute temperature, respectively. To achieve a high ZT value, a TE material must exhibit a high power factor (PF),a
2r
, and low thermal con-ductivity,
j
. However, increasing the ZT value is challenging because of the coupling among the TE parameters[3]: the relation-ship betweena
and the carrier concentration n (expressed by |a
| n2/3[3]) limits the enhancement of the PF (=a
2r
), whereasthe proportional relationship between electrical conductivity and electronic thermal conductivity (the Wiedmann–Franz law) restricts the improvement of the
r
/j
ratio.Bismuth telluride (Bi2Te3)-based materials have been widely
exploited for Peltier-coolers and thermoelectric generators at low temperature regime [5–8]. Nanocrystalline and nanostructured Bi2Te3-based films conduct heat poorly because of extensive
phonon scattering at grain boundaries[9–12], but the electrical transport properties of the films are impaired because of lattice imperfections and grain-boundary defects[9], indicating that fur-ther investigation is required to determine how to improve PF or the electronic part of ZT. Currently, enhancing the PF of Bi2Te3
-based thin films is challenging. Besides the coupling among TE material properties[3], the control of film stoichiometry is a key factor for obtaining better TE properties[13–16]. Yet, it is a chal-lenge to grow stoichiometric films because of the tendency for re-evaporation of volatile elements (i.e. Te, Se) at elevated Ts
[15,16], and the low sticking coefficient Te (<0.6) at Tsbeyond
300 °C [17,18]. Numerous charge carriers arising from vacancy defects of volatile elements can constrain the enhancement of |
a
|; however, low carrier concentrations can suppress electrical conductivity if carrier mobility (l
) is poor.Substantial effort has been devoted to enhancing the PF and ZT values of Bi2Te3thin films grown using various vapor-deposition
techniques. Moreover, nanocrystalline and nanostructured Bi2Te3
thin films have recently attracted considerable attention because they exhibit superior TE performance [9,10,13,19–22]. The layered-hexagonal Bi2Te3 films fabricated using radio-frequency
magnetron sputtering possessed a PF of 8.8
l
W cm1K2 for an(0 1 5)-oriented film, and a PF of 33.7
l
W cm1K2 for a highly(0 0 l)-oriented layered film [19,21]. Furthermore, PFs of 27
l
W cm1K2 and 39.9l
W cm1K2 were measured forhttp://dx.doi.org/10.1016/j.jallcom.2014.07.018
0925-8388/Ó 2014 Elsevier B.V. All rights reserved.
⇑Corresponding author. Tel.: +886 35131420.
E-mail addresses:[email protected](C.W. Luo),[email protected]
(J. Leu).
1 Tel.: +886 35712121x56196.
Contents lists available atScienceDirect
Journal of Alloys and Compounds
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j a l c o msmooth-epitaxial- and hexagonal-Bi2Te3 films grown using
molecular beam epitaxy (MBE) [23] and co-evaporation [24], respectively. In the case of pulsed laser deposition (PLD), tightly controlling substrate temperatures (Ts) and ambient pressures
enables the morphologies and compositions of films to be manip-ulated extensively, which offers a new method for enhancing the TE properties of films[13,16,20,25,26]. For example, self-assembled Bi2Te3 films featuring well-aligned zero- to three-dimensional
nanoblocks have been fabricated, but the room-temperature PFs of these films remain low (61.9
l
W cm1K2) [20]. By contrast,Li Bassi et al. [13] obtained high room-temperature PFs of approximately 21.2 and 50.6
l
W cm1K2 for compact-smoothand layered-smooth Bi2Te3films, respectively. Therefore, the
inter-relationships between PLD processing conditions, microstructures, and TE properties of Bi2Te3-based thin films must be understood
comprehensively.
In this study, the advantageous features of the PLD technique were exploited and a wide range of Tsof 30–380 °C was used to
grow various Bi2Te3 thin films featuring well-defined
morpholo-gies and grain sizes ranging from the nanoscale to the microscale. The PFs of the films were substantially enhanced because the resulting improvement in crystal structure enabled attaining high
l
values and concurrently achieving stoichiometry, which lowered n and enhanced |a
| (obeying |a
| n2/3approximately). Acompre-hensive understanding of TE Bi2Te3-based thin films grown using
PLD is presented in this study.
2. Experimental details
Bismuth telluride thin films were deposited on SiO2 (300 nm)/Si (1 0 0)
substrates (size of 15 15 mm2) at T
sof 30–380 °C and PArof 80 Pa using PLD.
The substrate was heated and maintained at desired Tsusing a thermocouple and
a PID temperature controller. The thermocouple was buried inside a stainless-steel substrate holder which was heated by a tungsten lamp just behind the holder. The UV pulses (20 ns duration) from a KrF excimer laser (k = 248 nm, repetition: 5 Hz) were focused on a stoichiometric polycrystalline Bi2Te3target (purity 99.99%) with
fluence of 3.8 J/cm2
; the target-to-substrate distance was 40 mm. The number of laser pulses was 12,000 and deposition took 40 min. The average growth rate was approximately 0.52 Å/pulse. An approximately 300-nm-thick SiO2layer was
ther-mally grown on the Si substrates for electrical isolation purpose.
The orientation and crystallinity of Bi2Te3films were determined using X-ray
diffraction (XRD, Bruker D8) with Cu Karadiation (k = 1.5406 Å) in 2h xand rocking-curve (x-scan) configurations. Digital HRTEM images were obtained using high-resolution transmission electron microscopy (HRTEM, Philips Tecnai F20) operated at 200 kV, and Gatan 2 k 2 k charged couple device camera. The specimens were prepared using a standard procedure of mechanical thinning and Ar-ion milling. Surface morphology and film thickness were examined using field-emission scanning electron microscopy (SEM, JEOL JSM-6500) through plane-view and cross-sectional images, respectively. Film compositions were also analyzed using an Oxford energy-dispersive X-ray spectroscopy (EDS) equipped with the SEM instrument at an accelerating voltage of 15 kV, a dead time of 22–30%, and a collecting time of 60 s. The atomic percentage of each film was deter-mined by averaging the values measured in 5 or more distinct 13 18lm2
areas on the surface of films. The in-plane electrical conductivity, carrier concentration, and mobility were measured at room temperature using a Hall system (Bio-Rad HL5500PC) with van der Pauw geometry. Indium balls were used to improve ohmic contact on the films’ surface. The in-plane Seebeck coefficient at room temperature was determined from the slope of the voltage difference vs. the temperature difference curve, based on a static temperature-difference method[27].
3. Results and discussion
Fig. 1presents the cross-section and top-view SEM images of Bi2Te3thin films grown at Tsranging from 30 to 380 °C at PAr= 80
-Pa. Under these conditions, films of six well-defined morphologies featuring distinct grain sizes, shapes, and stacking characteristics were successfully prepared.Fig. 2presents the grain-size distribu-tion and the most probable size (MS) of these films, which were determined using SEM statistical analysis. First, at room tempera-ture (30 °C), 0D columnar nanoparticles (CNPs) were grown, which exhibited the smallest MS (57 nm) and a columnar structure that
was approximately 50 nm wide and 400 nm high. Second, at 120 °C, columnar nanoflowers (CNFs) were formed as a result of the stacking of 2D platelets (MS = 73 nm), which generated flower-like structures featuring columns that were approximately 75 nm wide and 500 nm high. Third, at 220 °C, nanodiscs (NDs) were formed that comprised numerous disc-like crystals, whose MS was 287 nm in diameter and 24.5 nm in thickness; the disc thickness was determined by performing 2D fast Fourier transform (FFT, the inset inFig. 1c) of the dashed-square area indicating a set of discs (Fig. 1c). Fourth, at 300 °C, a layered compact polycrystal-line (LCP) film was prepared that exhibited an MS of 477 nm. Intriguingly, this film was formed by compactly stacking of the 3D layered nanoblocks, but it exhibited a relatively rough surface. Fifth, at 340 °C, a layered-triangular-platelet (LTP) structure was prepared that exhibited the largest MS (846 nm, with a layer thick-ness of 53 nm) and the broadest grain-size distribution (reaching the micron size) (Figs. 1e and 2). Finally, the films deposited at 380 °C displayed a polyhedral (PH) structure that was composed of 3D triangular and polygonal crystals (MS of 785 nm) and exhib-ited a diminished density because of the presence of microvoids between crystals. The MS and the width of the distribution curves increased monotonically with increasing Tsfrom RT to 340 °C and
then decreased slightly at 380 °C, as shown inFig. 2and the inset ofFig. 2.
The Tsused can affect the nucleation and growth of films. At low
a Ts, the rate of supersaturation is high, which reduces not only the
critical size of the nuclei but also the magnitude of the nucleation energy barrier, and thus numerous, small nanoparticles grew at 30 °C and nanoflowers grew at 120 °C[16]. Moreover, the growth mechanism of the columnar structures (Figs. 1a and b) can be attributed to the combined effects of a high deposition rate and a low crystal growth rate: the high PArof 80 Pa tightly confines the
ablated plume along the direction of the substrate to increase the deposition rate[25], whereas the potentially adsorbed argon limits the mobility of adatoms to suppress the lateral growth of the crys-tals at low Ts(6120 °C)[28,29]. By contrast, the deposition is faster
on the top of islands than in the valleys between the islands with an oblique incident flux (the so-called shadowing effect)[28,30], which generates the separated or voided inter-grain boundaries of columnar structures. The columnar structures present in the Bi2Te3films grown here were similar to those in Bi2Se3films grown
at comparatively high temperatures and pressures (200–250 °C, 173 Pa helium)[16].
At a high Ts(P220 °C), however, a reduction in the
supersatura-tion rate increases the critical size of the nuclei and the nucleasupersatura-tion barrier. Consequently, the large nuclei can further create isolated islands and 3D crystal structures on the substrates to minimize the surface energy and interface energy[31,32]by means of sur-face diffusion, grain-boundary migration, and possible recrystalli-zation. Conversely, the LTP structure might be formed because of the anisotropic bonding configuration of Bi2Te3and the inevitable
deviations from a uniform growth environment[33–35]. Moreover, both the diffusion of atoms at high Ts(P220 °C) and the naturally
layered crystal structure of Bi2Te3result in the formation of layered
ND, LCP, and LTP structures[33].
Fig. 3a shows the normalized XRD patterns of the Bi2Te3target
and films. The polycrystalline rhombohedral Bi2Te3phase (space
group D53dR3m) with (0 1 5)-dominant orientation of the target
can be confirmed (JCPDS 82-0358). The films grown at Ts6340 °C exhibited the Bi2Te3phase but no other detectable phases.
How-ever, when Tswas increased to 380 °C, the PH film possessed Bi4Te5
phase (JCPDS 22-0115), which was associated with a composition of approximately 51.5 at.% Te (Fig. 4). Moreover, like the target, the dominant orientation of the CNP (30 °C) and CNF (120 °C) structures was (0 1 5). Typically, Bi2Te3 (0 1 5) is the preferred
supports not only stoichiometric growth but also regular structures that feature a bonding sequence of Te(1)–Bi–Te(2)–
Bi–Te(1) [37]. Because the adatom mobility increased at
Ts= 220–340 °C, the films featuring ND, LCP, and LTP structures
exhibited the highly preferred crystal orientation of {0 0 l}, which possesses the lowest surface energy as a result of the weak Te(1) Te(1)bond[37].
The crystallinity and grain orientation of the films were deter-mined by measuring X-ray rocking curves. As shown inFig. 3a, the full width at half maximum (FWHM) of the (0 0 6) peak in the Bi2Te3phase and the (0 0 11) peak in the Bi4Te5phase dropped
substantially, from 5.87° in the case of CNFs (at 120 °C) to 0.74° in
NDs (at 220 °C), indicating that NDs feature superior crystallinity and grain orientation compared with CNFs. Because of the pres-ence of disoriented grains and a rough surface, the FWHMs of the LCP (at 300 °C) and PH (at 380 °C) structures increased slightly, to 1.82° and 1.49° respectively. By contrast, the LTP film exhibited a small FWHM, 1.05°, which can be attributed to the large (micrometer-scale) grain size, flat surface, and layered structure of the film, reducing the disorientation of crystallites.
HRTEM images performed on the low
l
CNP film and the high-estl
ND film are shown in Fig. 3b and c, respectively. Clearly, Fig. 3b presents some nanoparticles (nanocrystals) with sizes >10 nm. Moreover, the lattice spacing of nanoparticles is approxi-mately 0.323 nm, which corresponds to the value of (0 1 5) inter-planar distance of the Bi2Te3 crystal. The white lines in Fig. 3bindicate the orientations of (0 1 5) planes. Intriguingly, although the overall (0 1 5)-orientation is randomly, it possesses some local preferred orientations as shown by the parallel white lines amongst some close nanocrystals. For the ND (220 °C) film, the lower inset inFig. 3c shows the film with uniform thickness of approximately 295 nm, and the SiO2 layer with thickness of
300 nm. Furthermore, an HRTEM image (Fig. 3c) obtained from the solid square area in the inset exhibits the projected periods of 0.508 nm along the c-axis correspond to the lattice spacing of the (0 0 6) planes. Thus, the c-axis lattice constant of the film is 30.48 Å of Bi2Te3, which agrees closely with the value (30.44 Å)
presented in JCPDS 82-0358. The other examined areas also pres-ent similar results. Consequpres-ently, this TEM results further demon-strated the highly (0 0 l)-orientated and crystallized structures of the ND, LCP, LTP films that should facilitate the transport of charge carriers.
The PArof 80 Pa was determined to be a suitable background
pressure because it allows stoichiometric films to be grown even
Fig. 1. Cross-section and top view SEM images of n-type Bi2Te3thin-films with different nanostructures deposited at various substrate temperatures (Ts) from 30 to 380 °C
under an argon background pressure (PAr) of 80 Pa. The inset in panel (c) shows the FFT patterns and distance profile of the dash square area in the SEM cross-section image.
Fig. 2. Grain size distributions and the most probable size (MS) of the films in
Fig. 1(a–f), which was statistically analyzed from at least 200 grains of top-view SEM images. The inset shows the Ts-dependent MS of the films.
when the Tsis high (up to 340 °C). We note that low pressures
typ-ically increased Te deficiency and elevated n, whereas high pres-sures commonly generated macroscopic droplets on film surfaces. These effects of pressure agree with previous studies on Bi2Se3films [16] and with the results described elsewhere[38].
In Fig. 4, the black squares indicate the Ts-dependent Te at.% of
the six films featuring distinct nanostructures that were deposited at 80-Pa argon. The films clearly exhibited Te enrichment, stoichi-ometry, and substantial Te deficiency at Ts6120 °C, 220 °C 6 Ts
6340 °C, and Ts= 380 °C, respectively. Because the vapor pressure of Te is higher than that of Bi (PTe
v=PBiv 105 at 300 °C [39]), Te reevaporates from heated substrates much faster than Bi does [15,39]. This also occurred in the case of Bi2Se3 films
[40], in which the film composition varied from being Se-rich to stoichiometric to Se-deficient with increasing Ts. The Te at.%
dropped sharply to approximately 51.5% at 380 °C (Fig. 4), which might be explained by a substantial increase in the Te reevapora-tion rate and a lowering of the Te sticking coefficient[18].
The variation in the n-type carrier concentration (n) and mobil-ity (
l
) as a function of Tsand the nanostructures is shown inFig. 4.The n values of Te-rich films were 6.9 1019cm3(CNPs, at 30 °C)
and 9.3 1019cm3(CNFs, at 120 °C), but decreased considerably
to range from 2.9 1019to 4.9 1019cm3in the case of
stoichi-ometric ND, LCP, and LTP films deposited at 220, 300, and 340 °C, respectively. However, in the highly Te-deficient PH-Bi4Te5 films
deposited at 380 °C, n increased dramatically, reaching approxi-mately 1.06 1021cm3. This agreed well with the result in Ref.
[13] that the n increased dramatically from 4.9 1019cm3 to
5.0 1020cm3with increasing Bi content from 40 at.%
(stoichi-ometry) to 45 at.% (Bi4Te5 phase). Therefore, the stoichiometry
plays a vital role in reducing the n of the films.
The
l
value was inversely proportional to the FWHM and the carrier concentration n (Figs. 3a and4). The CNP and CNF filmsFig. 3. (a) The normalized X-ray diffraction patterns of Bi2Te3and Bi4Te5thin films. FWHMs of X-ray rocking curves for (0 0 6) peak in Bi2Te3phase and (0 0 1 1) peak in Bi4Te5
phase. (b) An HRTEM image of the columnar nanoparticle (30 °C) film; the white lines indicate the (0 1 5) orientation of the nanograins. (c) An HRTEM and a low magnification TEM (inset) images of the nanodisc (220 °C) film.
Fig. 4. Ts-dependent Te at.% (black squares), carrier concentration (n, red
triang-ulars), and carrier mobility (l, blue spheres) of the Bi2Te3and Bi4Te5films. The
abbreviations: CNP, columnar nanoparticle; CNF, columnar nanoflower; ND, nanodisc; LCP, layered compact polycrystalline; LTP, layered triangular platelet; PH, polyhedron. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
grown at a low Ts(6120 °C) had a low
l
, <10 cm2/Vs, because ofthe strong grain-boundary scattering resulting from the columnar structures with small grains (MS of 57 or 73 nm) and the defective scattering centers (ionized impurities). However, in the ND, LCP, and LTP films grown at 220–340 °C,
l
increased substantially and ranged from 83.9 to 122.3 cm2/Vs because of the suppression ofthose scattering mechanisms, as suggested by the low FWHMs and n results. In addition, the highly (0 0 l)-oriented layered struc-tures with in-plane large crystallites provides a preferential way for electron transport along the ab-plane and thus promotes the carrier mobility. Recently, Deng et al.[21]observed that the
l
of the highly (0 0 l)-oriented layered Bi2Te3 film was approximately5 times higher than that of the ordinary (0 1 5)-oriented film. Despite featuring a large grain size (MS = 785 nm), the
l
in the Bi4Te5 films displaying the PH structure was small (2.5 cm2/Vs)because of an extremely high carrier concentration (n = 1.06 1021cm3), the difference in phase, and the ordinary
3D-voided structure.
Li Bassi et al.[13]reported that the
l
of Bi-Te films obtained a high value of approximately 100 cm2/Vs only for the stoichiometricBi2Te3film, meanwhile it remained low values of 10–30 cm2/Vs for
the other non-stoichiometric Bi–Te films and phases. This result suggests that the stoichiometry plays a certain contribution to the substantially enhanced
l
of the films grown at 220–340 °C. Since the high-l
preferred structures and the stoichiometry are obtained concurrently in these films, it is hard to fully extract the individual contribution of each factor for the enhancedl
. Nev-ertheless, under a similar deviation within 2.0 at.% from stoichiom-etry, thel
of the 220–340 °C films exhibit a small difference (below 31.5%,Fig. 4) compared to 81.2% for the compact films in Ref.[13]. This weaklyl
-dependence on the stoichiometry suggests that microstructure is the predominant factor contributed to the substantiall
enhancements of the present films.To explain the evolution of n, the antisite and vacancy defects must be considered. From the XRD (0 0 6) and (0 0 15) peaks, the averaged c-lattice constants of the films were determined using the hexagonal unit cell relation:
1 d2hkl ¼4 3 h2þ hk þ k2 a2 ! þ‘ 2 c2
As shown in the inset inFig. 5a, the c-lattice constant of the CNP and CNF films that were Te-rich is considerably smaller than the standard value of 30.44 Å (JPCDS 82-0358), suggesting the pres-ence of a high density of antisite TeBi(Te occupying a Bi site, a
donor-point defect) because of the smaller atomic radius of Te (1.4 Å) compared with Bi (1.6 Å) [41]. This is to be expected because the strain effect can be neglected for such thick films (thicknesses >530 nm,Fig. 1a and b), and also because TeBiexhibits
the smallest formation energy (approximately 0.5 eV) among point defects such as BiTe(Bi occupying a Te site) and VTe(Te vacancy)
under a Te-rich condition[42,43]. This result suggests that TeBiis
the dominant donor defect that generates the moderate n values of the Te-rich films.
The decrease of n in the stoichiometric ND, LCP, and LTP films was associated with a reduction in the donor defects TeBi and
VTe, because the c-lattice constant was close to the database value
(inset inFig. 5a). Moreover, the Tsof 220–340 °C should be
suffi-cient for atoms to move and drop to the lowest potential-energy sites and thus reduce the numbers of defects. The structure of the Bi4Te5 phase can be derived by stacking hexagonal Bi2 and
Bi2Te3 blocks, (Bi2)m(Bi2Te3)n, and (m:n = 1:5), where the –(Bi–
Bi)– blocks intercalate in van der Waals gaps between the – (Te(1)–Bi–Te(2)–Bi–Te(1))– blocks[44]. Furthermore, the PH-Bi
4Te5
film remained substantially Te-deficient (approximately 4.1 at.%) when compared with the Te at.% of the EDS results (51.5) and the ideal value of the Bi4Te5phase (55.6). Therefore, the dramatic
increase of n in the PH-Bi4Te5film can be attributed to the
domi-nance of VTeunder such Te-deficient (or Bi-rich) conditions, which
can also leave Bi interstitials in the lattice and thus generate a c-lattice constant that is slightly larger than the database value (JCPDS 22-0115).
Fig. 5a shows the n-dependent |
a
| of the films deposited at var-ious Ts. The stoichiometric ND, LCP, and LTP films featuring a low n(2.9–4.9 1019cm3) possessed superior |
a
| values, ranging from172.8 to 189.7
l
V/K. By contrast, both Te-rich (CNPs and CNFs) and Te-deficient (PHs) films featuring a high n possessed substan-tially lower |a
| values, which ranged from 32.6 to 53.6l
V/K (Fig. 5a). This can be described effectively by the relationship |a
| mn2/3 in degenerate semiconductors (i.e., the parabolicband, energy-independent scattering approximation [3,16]), as shown inFig. 5a, in which the solid lines are the plots of the for-mula in Fig. 5a [3], featuring various effective mass m values,
ranging from 0.4m0to 1.0m0(where m0is the free electron mass).
Very recently, Shin et al. [45] used m= 0.58m
0 whose value
was inferred from Ref. [4] to describe well their transport results of the Bi2Te3 nanowires with the similar n of
4.9 1019–1.8 1020cm3. The present mvalues in the range of
0.4m0–1.0m0was comparable or slightly larger than the expected
value (0.58m0) because various scattering sources, such as the
Fig. 5. (a) Absolute Seebeck coefficients (|a|) vs. n; the solid lines are the plots of the formula inFig. 5a with various effective mass m
from 0.4m0to 1.0m0(m0is the free
electron mass). Inset: Ts-dependent c-axis lattice constant of the Bi2Te3and Bi4Te5films. (b) Tsdependence of room temperature Seebeck coefficienta(red circles), electrical
conductivityr(blue triangulars), and power factor (PF =a2
r, black squares) of the Bi2Te3and Bi4Te5films. (For interpretation of the references to color in this figure legend,
grain boundary, lattice, and ionized impurity scatterings, are excluded in the approximation.
Fig. 5b shows the Ts-dependent
a
,r
, and PF (=a
2r
) of the films.The
r
value gradually increased from 34.5 ± 0.1 to 814.3 ± 1.5 S cm1when Tswas increased from 30 to 300 °C, and
then sharply decreased to 647.3 ± 0.4 S cm1 at 340 °C and
414.0 ± 1.2 S cm1 at 380 °C. The enhanced
r
(=647.3–814.3 S cm1) of the films grown at 220–340 °C originated from
the substantially enhanced
l
because the n exhibited a slight decrease (Fig. 4). Although the coupled relationship betweenr
(=ne
l
) and |a
| (n2/3) generally constrains the concurrentenhancement of
r
and |a
|, a reduction of n and a substantial increase ofl
in the same optimal range of Ts, 220–340 °C, couldlead to high values of both
r
and |a
|. Consequently, the PF of the stoichiometric ND, LCP, and LTP films reached remarkably high val-ues, ranging between 18.2 ± 0.25 and 24.3 ± 0.44l
W/cm K2,whereas the PF was low (60.44
l
W/cm K2) in the case ofnonstoi-chiometric films deposited at Ts6120 or 380 °C (Fig. 5b).
The composition, transport and TE properties at room tempera-ture of the optimal Bi2Te3films in this study and those in the
ear-lier relevant studies [13,19–21,23,24,46] are summarized in Table 1. The optimal PF value (24.3
l
W cm1K2) of alayered-compact polycrystalline film obtained in this study was considerably higher than those of the Bi2Te3 films featuring a
nanoparticle structure (PF = 1.9
l
W cm1K2) [20], a layeredstructure (PF = 8.8
l
W cm1K2) [19], a hexagonal structure(PF = 18.4
l
W cm1K2) [46], and a compact-smooth structure(PF 21.2
l
W cm1K2)[13]. However, it was slightly lower thanthe PF = 27
l
W cm1K2of a smooth epitaxial Bi2Te3film grown
using MBE[23]. Furthermore, the optimal PF of 24.3
l
W cm1K2was approximately 1.39-, 1.64-, and 2.08-times lower than the PFs 33.7, 39.9, and 50.6
l
W cm1K2of a highly (0 0 l)-orientedlayered [21], a hexagonal polycrystalline [24], and a layered smooth films[13], respectively. Generally, as illustrated inTable 1, a stoichiometric composition is necessary to obtain a reasonably low n (61.0 1020cm3) which in turn allows obtaining a high
|
a
| value. Moreover, a layered structure is mostly found to be the best morphology for excellent TE properties (Table 1). In this study, the structure combining both layered and compact features exhib-ited the highest PF value amongst our films owning to its highr
up to 814.3 ± 1.5 S cm1.The aforementioned results revealed that in this study, Bi2Te3
thin films prepared using PLD exhibited high PF values at elevated temperatures, at which the PFs of Bi2Te3and Bi2Se3thin films could
be suppressed because of nonstoichiometry and donor-point defects (i.e., vacancies VTe and VSe or antisites TeBi and SeBi)
[15,16]. In this study, a simple deposition strategy was adopted in which the ambient pressure used (80 Pa) was higher than that typically used in PLD depositions (seeTable 1), with the goal being to reduce the extent of doping and the Ts(220 °C 6 Ts6340 °C) for
high-
l
preferred structural growth. This approach not only allevi-ated the doping problem without the requirement of any extra engineering of the targets or engineering during the film growth, but also improved the structural quality of the films and thereby enhanced the charge-carrier mobility and substantially increased PFs. This PLD strategy could potentially be extended to fabricating high-PF thin films on excellent compounds such as Bi2xSbxTe3andBi2Te3xSex, promising for applications in TE devices.
4. Conclusion
Nanostructured n-type bismuth telluride thin films were fabri-cated using PLD. At Tsof 220–340 °C and PAr= 80 Pa, the
stoichiom-etric Bi2Te3films with highly (0 0 l)-oriented and layered structures
possess remarkably high PFs between 18.2 ± 0.25 and 24.3 ± 0.44
l
W cm1K2, that is attributed to the concurrentlysubstantial enhancements in
l
(83.9–122.3 cm2/Vs) and |a
|(172.8–189.7
l
V/K). It has been demonstrated that the morphol-ogy of highly (0 0 l) oriented-layered structures and the stoichiom-etry predominantly contribute to the substantial enhancement ofl
and |
a
|, respectively, resulting in remarkable enhancement in PF. AcknowledgementsOne of the authors (P.H. Le) is thankful for Dr. Hong Quan Nguyen for the help of taking HRTEM images, and Prof. K.H. Wu for the support of using the equipment. Financial support from the Ministry of Science and Technology of the Republic of China (Taiwan) under Contract Nos.: 101-2112-M-009-016-MY2 and 103-2923-M-009-001-MY3 is gratefully acknowledged.
References
[1]R. Venkatasubramanian, E. Siivola, T. Colpitts, B. O’Quinn, Nature 413 (2001) 597–602.
[2]D.M. Rowe, in: D.M. Rowe (Ed.), Thermoelectrics Handbook: Macro to Nano, CRC/Taylor & Francis, Boca Raton, FL, 2006.
[3]G.J. Snyder, E.S. Toberer, Nat. Mater. 7 (2008) 105.
[4]G.S. Nolas, J. Sharp, H.J. Goldsmid, Thermoelectrics Basic Principles and New Materials Developments, Springer, Berlin, 2001.
[5]H. Böttner, J. Nurnus, A. Gavrikov, G. Kühner, M. Jägle, C. Künzel, D. Eberhard, G. Plescher, A. Schubert, K.-H. Schlereth, J. Microelectromech. Syst. 13 (2004) 414–420.
[6]B. Huang, C. Lawrence, A. Gross, G.-S. Hwang, N. Ghafouri, S.-W. Lee, H. Kim, C.-P. Li, C. Uher, K. Najafi, M. Kaviany, J. Appl. Phys. 104 (2008) 113710. [7]G. Min, D.M. Rowe, Solid State Electron. 43 (1999) 923–929.
[8]M. Takashiri, T. Shirakawa, K. Miyazaki, H. Tsukamoto, Sens. Actuators A 138 (2007) 329–334.
[9]M. Takashiri, M. Takiishi, S. Tanaka, K. Miyazaki, H. Tsukamoto, J. Appl. Phys. 101 (2007) 074301.
[10] C.-N. Liao, Y.-C. Wang, H.-S. Chu, J. Appl. Phys. 104 (2008) 104312. [11]M. Takashiri, S. Tanaka, K. Miyazaki, H. Tsukamoto, J. Alloys Comp. 490 (2010)
L44–L47.
[12]H.-C. Chang, C.-H. Chen, Y.-K. Kuo, Nanoscale 5 (2013) 7017. Table 1
Morphology, growth conditions, method, carrier concentration n, carrier mobilityl, electrical conductivityr, Seebeck coefficienta, power factor PF (=a2
r), and Te content of the optimal Bi2Te3thin films in this study as compared to properties of Bi2Te3thin films reported in the literature. All the selected values were recorded at room temperature.
Morphology Ts(°C) PAr(Pa) Method n (1019cm3)l(cm2/Vs)r(S/cm) a(lV/K) PF (lW/cm K2) Te (at.%) Ref.
Nanodiscs 220 80 PLD 2.9 ± 0.1 122.3 ± 0.5 572.0 ± 0.06 178.2 ± 1.2 18.2 ± 0.25 60.5 ± 0.4 This work Layered compact polycrystalline 300 80 PLD 5.0 ± 0.2 102 ± 3.6 814.3 ± 1.5 172.8 ± 1.4 24.3 ± 0.44 59.7 ± 0.4 This work Layered-triangular platelets 340 80 PLD 4.8 ± 0.1 83.9 ± 2.0 647.3 ± 0.4 189.7 ± 2.3 23.3 ± 0.57 58.4 ± 0.6 This work Layered-smooth film 250 10 PLD 10.1 90.6 1464.0 186 50.6 – Bassi et al.[13]
Compact-smooth film 350 10 PLD 3.5 95 532.0 170 21.2 60 Bassi et al.[13]
Nanoparticles 300 20 PLD 9.7 14.8 230 91 1.90 60 Chang et al.[20]
Layered Structure 350 1.0 Sputtering 95 12.1 1840 70 8.8 56.6 Deng et al.[19]
Highly (00 l)-oriented layered film 350 2.0 Sputtering 10.0 52 840.0 200 33.7 60 Deng et al.[21]
Hexagonal polycrystalline 260 Co-evaporation 1.8 125 353.4 228 18.4 60 Silva et al.[46]
Hexagonal polycrystalline 260 Co-evaporation 6.5 75 769.2 228 39.9 Zou et al.[24]
[13]A. Li Bassi, A. Bailini, C.S. Casari, F. Donati, A. Mantegazza, M. Passoni, V. Russo, C.E. Bottani, J. Appl. Phys. 105 (2009) 124307.
[14]R.S. Makala, K. Jagannadham, B.C. Sales, J. Appl. Phys. 94 (2003) 3907. [15]L.M. Goncalves, C. Couto, P. Alpuim, A.G. Rolo, F. Völklein, J.H. Correia, Thin
Solid Films 518 (2010) 2816–2821.
[16]P.H. Le, C.-N. Liao, C.W. Luo, J.-Y. Lin, J. Leu, Appl. Surf. Sci. 285P (2013) 657– 663.
[17]N. Peranio, M. Winkler, D. Bessas, Z. Aabdin, J. König, H. Böttner, R.P. Hermann, O. Eibl, J. Alloys Comp. 521 (2012) 163–173.
[18]A. Mzerd, D. Sayah, J.C. Tedenac, A. Boyer, J. Cryst. Growth 140 (1994) 365. [19]Y. Deng, H.-M. Liang, Y. Wang, Z.-W. Zhang, M. Tan, J.-L. Cui, J. Alloys Comp.
509 (2011) 5683–5687.
[20]H.-C. Chang, C.-H. Chen, CrystEngComm 13 (2011) 5956–5962.
[21]Z. Zhang, Y. Wang, Y. Deng, Y. Xu, Solid State Commun. 151 (2011) 1520–1523. [22]Y. Zhou, L. Li, Q. Tan, J.-F. Li, J. Alloys Comp. 590 (2014) 362–367.
[23]N. Peranio, O. Eibl, J. Nurnus, J. Appl. Phys. 100 (2006) 114306. [24]H. Zou, D.M. Rowe, G. Min, J. Cryst. Growth 222 (2001) 82–87.
[25]A. Bailini, F. Donati, M. Zamboni, V. Russo, M. Passoni, C.S. Casari, A. Li Bassi, C.E. Bottani, Appl. Surf. Sci. 254 (2007) 1249–1254.
[26]L.S. Faraji, R.P. Singh, M. Allahkarami, Eur. Phys. J. Appl. Phys. 46 (2009) 20501. [27]L.R. Testardi, G.K. McConnell, Rev. Sci. Instrum. 32 (1961) 1067.
[28]J.A. Thornton, J. Vac. Sci. Technol. 11 (1974) 666. [29]J.A. Thornton, J. Vac. Sci. Technol. 12 (1975) 830.
[30]T. Karabacak, J.P. Singh, Y.-P. Zhao, G.-C. Wang, T.-M. Lu, Phys. Rev. B 68 (2003) 125408.
[31]M. Ohring, The Materials Science of Thin Films, Academic Press, San Diego, CA, 1992.
[32]D.L. Smith, Thin-Film Deposition: Principles & Practice, McGraw-Hill, New York, 1995.
[33]D. Kong, W. Dang, J.J. Cha, H. Li, S. Meister, H. Peng, Z. Liu, Y. Cui, Nano Lett. 10 (2010) 2245–2250.
[34]Y. Liang, W. Wang, B. Zeng, G. Zhang, J. Huang, J. Li, T. Li, Y. Song, X. Zhang, J. Alloys Comp. 509 (2011) 5147–5151.
[35]G. Hao, X. Qi, L. Yang, Y. Liu, J. Li, L. Ren, F. Sun, J. Zhong, AIP Adv. 2 (2012) 012114.
[36]Y. Deng, Y. Xiang, Y. Song, Cryst. Growth Des. 9 (2009) 3079–3082. [37]C.-N. Liao, H.-D. Shih, P.-W. Su, J. Electrochem. Soc. 157 (2010) D605–D608. [38]S.X. Zhang, L. Yan, J. Qi, M. Zhuo, Y.-Q. Wang, R.P. Prasankumar, Q.X. Jia, S.T.
Picraux, Thin Solid Films 520 (2012) 6459–6462.
[39]H. Noro, K. Sato, H. Kagechika, J. Appl. Phys. 73 (1993) 1252. [40]P.H. Le, K.H. Wu, C.W. Luo, J. Leu, Thin Solid Films 534 (2013) 659–665. [41]S. Cho, Y. Kim, A. DiVenere, G.K. Wong, J.B. Ketterson, J.R. Meyer, Appl. Phys.
Lett. 75 (1999) 1401.
[42]A. Hashibon, C. Elsässer, Phys. Rev. B 84 (2011) 144117.
[43]G. Wang, X.G. Zhu, Y.Y. Sun, Y.Y. Li, T. Zhang, J. Wen, X. Chen, K. He, L.L. Wang, X.C. Ma, J.F. Jia, S.B. Zhang, Q.K. Xue, Adv. Mater. 23 (2011) 2929.
[44]J.W.G. Bos, H.W. Zandbergen, M.-H. Lee, N.P. Ong, R.J. Cava, Phys. Rev. B 75 (2007) 195203.
[45]H.S. Shin, S.G. Jeon, J. Yu, Y.-S. Kim, H.M. Park, J.Y. Song, Nanoscale 6 (2014) 6158–6165.
