Research Article www.acsami.org
Controllable Electrical Contact Resistance between Cu and OrientedBi2Te3 Film via Interface Tuning Xixia Kong,† Wei Zhu,† Lili Cao,*,† Yuncheng Peng,† Shengfei Shen,† and Yuan Deng*,†,‡ †
School of Materials Science and Engineering, Beihang University, Beijing 100191, China Beijing Key Laboratory for Advanced Functional Materials and Thin Film Technology, Beijing 100191, China
‡
ABSTRACT: The contact resistance between metals and semiconductors has become critical for the design of thin-film thermoelectric devices with their continuous miniaturization. Herein, we report a novel interface tuning method to regulate the contact resistance at the Bi2Te3−Cu interface, and three Bi2Te3 films with different oriented microstructures are obtained. The lowest contact resistivity (∼10−7 Ω cm2) is observed between highly (00l) oriented Bi2Te3 and Cu film, nearly an order of magnitude lower than other orientations. This significant decrease of contact resistivity is attributed to the denser film connections, lower lattice misfit, larger effective conducting contact area, and smaller width of the surface depletion region. Meanwhile, our results show that the reduction of contact resistance has little dependence on the interfacial diffusion based on the little change in contact resistivity after the introduction of an effective Ti barrier layer. Our work provides a new idea for the mitigation of contact resistivity in thin-film thermoelectric devices and also gives certain guidance for the size design of the next-level miniaturized devices. KEYWORDS: Interface tuning, contact resistance, oriented microstructure, contact area, interfacial diffusion
1. INTRODUCTION The metal−semiconductor (M−S) junction, acting as a prominent component in electronic devices, would affect the device performance, such as the open-circuit voltage in solar cells, the on-state currents in field effect transistors, the switching speed in diodes, as well as the maximum cooling fluxes in thermoelectric (TE) devices.1−4 Hence, the issue of M−S contacts has caused extensive concern in semiconductor device manufacture. In particular, TE devices have long been considered as promising energy conversion devices satisfying the urgent need for clean energy conversion technology.5−9 Furthermore, the development of miniaturization technologies to fabricate a thin-film TE device has become critical to enhance the performance of the TE device. Remarkably, the performance of thin-film TE devices is not only determined by the properties of TE films but also the M−S interface, which may introduce an additional electrical resistance to a large extent, thus decreasing the effective ⟨ ZT⟩D of TE devices according to eq 1.8 ⟨ZT ⟩D =
(L ) ⟨ZT ⟩M (L + 2R cσ )
with high ZT value to enhance the overall performances of devices.10,11 The M−S contact resistance is affected by several factors including the work function of both materials, interfacial states, crystal structure, and so on.12 The methods of surface or interface treatment,13,14 metallization system design,15,16 and heavy doping by ion implantation10 are commonly adopted to mitigate contact resistance. For instance, Gupta et al.11 acquired quite low contact resistivity of approximately 10−7 Ω cm2 via surface treatment and post deposition annealing in the study of Bi2Te3−Ni/Co contacts. In addition, a significant reduction of contact resistivity (ρc) for a metal/(Bi,Sb)2(Se,Te)3 contact was obtained by increasing the interface carrier concentration through ion implanting.10 It is worth noting that the oriented microstructure can also affect the contact resistance,17−19 and a low contact resistivity has been achieved below 2 × 10−6 Ω cm2 between the evaporated Cr/Ni/Au electrodes and δ-doped Bi2Te3-based superlattice structures.4 Our preliminary work also revealed that the interface between (110)-oriented Bi0.5Sb1.5Te3 film with Cu electrode exhibited a relatively low contact resistivity of 1 × 10−5 Ω cm2.18 Ideally, an extremely low ρc of ≤10−7 Ω cm2 is required for widespread adoption of TE technology in applications.10 Thus, more studies are still required to further decrease the contact resistance between the thin-film TE materials and electrodes.
(1)
where L is thermoelectric leg length; Rc is contact resistance; σ is bulk conductivity; ⟨ZT⟩D is the device figure-of-merit, and ⟨ZT⟩M is the effective ZT of thermoelectric material between Th and Tc. Hence, it is of great significance to decrease the nonnegligible Rc between TE materials and electrodes, especially in scaled-down thin-film devices, meanwhile combining TE films © 2017 American Chemical Society
Received: April 19, 2017 Accepted: July 10, 2017 Published: July 10, 2017 25606
DOI: 10.1021/acsami.7b05460 ACS Appl. Mater. Interfaces 2017, 9, 25606−25614
Research Article
ACS Applied Materials & Interfaces
was 5 h; and the distance between targets and substrates was maintained at 90 mm. To realize the optimization of interface structure, we fabricated three different oriented Bi2Te3 films before the Cu deposition, marked by Sample 1, Sample 2, and Sample 3, respectively. The Bi2Te3 target was connected to a DC power with 18 W, while the Te target was connected to a radio frequency (RF) power with 40 W. Then, the Bi2Te3 films were fabricated with the compensation of Te at different temperature of 200, 300, and 350 °C for Sample 1, Sample 2, and Sample 3, respectively. Working pressure of Ar was 2.0 Pa, and sputtering time was 1 h. Nanostructured Bi2Te3 films with different orientations are shown as follows: (1) Sample 1: (015) orientation (2) Sample 2: (00l) + (015) (3) Sample 3: (00l) orientation 2.2. Characterization and Measurement. The crystal structure of the films was examined by X-ray diffraction (XRD, Rigaku D/MAX 2200 PC) using the Cu Kα radiation (λ = 0.154 nm). The morphology and composition of the thin films were observed on a field emission scanning electron microscope (FE-SEM) (FEI Sirion 200) equipped with energy-dispersive X-ray spectroscopy (EDS). The surface topography and surface toughness of Bi2Te3 films were observed by an atomic force microscope (AFM, Bruker ICON, USA). Seebeck coefficient (a) was measured on thin films by ZEM-3 (Ulvac Riko, Inc.). Carrier concentration (N) was characterized with a four-probe method using a Hall effect measurement system (ECOPIA HMS3000). For the quantitative analysis of the different elements, five or more different regions of each sample were analyzed. The electrical contact properties were characterized by the TLM method and threeprobe method using an Agilent B2912A Precision/Measurement Unit equipped with a set of self-assembly probe stations. 2.3. Contact Resistance Measurement. 2.3.1. TLM Method to Measure Specific Contact Resistivity (ρc). The schematic diagram of a transmission line model (TLM) method is shown in Scheme 2a. ρc can
As is known, great efforts have been made to elevate the performance of TE films by controlling the oriented growth.18,20 For instance, Deng’s group reported that the Bi2Te3 film with preferential growth of the (00l) plane possessed higher performance compared to that with preferential growth of the (015) plane.21 Additionally, the layered Bi2Te2.7Se0.3 film with ZT of 1.27 was also obtained at room temperature.22 Thus, it is equally essential to illustrate how the different oriented microstructures affect the contact properties between TE films and electrodes. Moreover, element diffusion at the interface may occur in multilayer films, leading to a rapid deterioration of the performance for TE materials or devices and affecting the contact resistance as well.8,19,23 Therefore, it is imperative to consider the diffusion problem and optimize the interfacial contact resistance simultaneously. In this paper, we aim at proposing an effective approach to obtain a low contact resistivity (ρc) in M−S contacts. Meanwhile, the interfacial diffusion problem is also considered to maintain excellent performance for TE films. Besides, it is still a challenge to explore an accurate measurement method suitable for contact resistance research in thin-film materials. The current measurements require strong binding force at the interface to maintain effective contacts or time-consuming processing like lithography to obtain the test patterns.12,24−26 Herein, in this work, we adopted an easymanipulating and effective method compatible with a maskassisted deposition technique to directly fabricate Bi2Te3−Cu contacts, which can be completed simultaneously without any predeposition steps. In addition, we measured the contact resistance at varied contact areas by the three-probe method and extracted the contact resistivity of different oriented Bi2Te3−Cu contacts through the transmission line model (TLM) method. Besides, a Ti barrier layer was also introduced to solve the interfacial diffusion problem. Finally, the electrical transport mechanism at the Bi 2Te 3−Cu interface was investigated as well.
Scheme 2. Principle Diagrams of the TLM Measurement Systema
2. EXPERIMENTAL SECTION 2.1. Film Deposition and Patterning. The insulating polished aluminum nitride (AlN) was used as a substrate, where Bi2Te3 film was deposited followed by the deposition of the Cu film pattern (Scheme 1). Before the deposition process, the AlN substrate was cleaned in
a
(a) Contact resistance probing approaches for equal-spaced contacts. The distance between the adjacent metal pads is 0.2 mm, and the length of each electrode (Z) is 1.2 mm. (b) An example of the total resistance (RT) vs distance (d) plot where the specific contact resistivity (ρc) can be extracted.
Scheme 1. Deposition Process Diagram of Bi2Te3−Cu Contacts
be extracted from the linear fitting curve of TLM.11 Four-probe TLM measurements were performed on the fabricated devices to eliminate the effect of the experimental setup.27 Assuming that contact resistance Rc does not change when adjacent metal pads d varied and the sheet resistance of Bi2Te3 is the same at different places, the total resistance measured between adjacent metal pads is given by28 detergent, deionized water, acetone, and alcohol, respectively, for 15 min in an ultrasonic bath. In order to obtain the TLM pattern, a set of mask plates (including two mask plates) was fixed to the substrate successively. The films were fabricated by using a magnetron sputtering system (JGP-450a, SKY Technology Development Co., Ltd., Chinese Academy of Sciences). Commercial 60 mm diameter hot-pressed Cu (99.99% purity), Bi2Te3 (99.99% purity), and Te (99.99% purity) targets (purchased from General Research Institute for Nonferrous Metals, China) were used in the sputtering process. The Cu target was connected to a direct-current (DC) power supplied with the power of 30 W. Cu films were prepared under identical experimental parameters: the substrate temperature was kept at 200 °C; the working pressure of Ar was fixed at 1.0 Pa; the sputtering time
RT = LT =
R sh 2L TR sk d+ W W ρc /R sh
(2) (3)
where RT is the total measured resistance; Rsh is the sheet resistance of Bi2Te3; Rsk is the modified sheet resistance of Bi2Te3 under the contact; and LT is the transfer length, which is related to the distance needed for the current to flow in or out of the contact and can be given by eq 3. W is the contact width of Cu on Bi2Te3, and d is the distance between the two measured metal pads. It can be clearly seen that RT is the function of d from eq 2. In TLM, the Rsk and Rsh values are assumed to be equal;29 thus, the intercept of 25607
DOI: 10.1021/acsami.7b05460 ACS Appl. Mater. Interfaces 2017, 9, 25606−25614
Research Article
ACS Applied Materials & Interfaces the plot at RT = 0 gives −d = 2LT, and the intercept at d = 0 gives Rc, which is given by eq 4. The Rsh can be also extracted from the slope of the plot, as shown in Scheme 2b. Rc =
L TR sk W
growth mode. Figure 1b shows the orientation factor F of the (015) plane and (00l) plane, which is calculated by the following formulas30
(4)
In our TLM system, W is 0.6 mm, and d varies with 0.2, 0.4, 0.6, 0.8, and 1.0 mm, respectively. The total resistance RT is measured by the current−voltage (I−V) test with the four-probe method. Thus, ρc can be extracted according to eq 3. 2.3.2. Three-Probe Method to Measure the Contact Resistance (Rc). The contact resistance between Bi2Te3 and Cu at different nominal contact areas (0.24, 0.32, 0.40, 0.48, 0.56, and 0.64 mm2) was measured by the I−V test, which is operated by the three-probe method. In the I−V test (e.g., in order to extract Rc2), as shown in Scheme 3a, the current flows from probe 1 to probe 2 sweeping from
F=
P − P0 1 − P0
(5-1)
P0 =
I0(00l) ∑ I0(hkl)
(5-2)
P=
I(00l) ∑ I(hkl)
(5-3)
The high F for Sample 1 (0.77) and Sample 3 (0.55) also implies a preference growth along the [015] direction and [00l] direction, respectively. Meanwhile, the increasing F of (00l) orientation and declining F of (015) orientation from Sample 1 to Sample 3 illustrate a preferential transformation from (015) to (00l) orientation. It has been reported that the direction of preferred orientation could affect the microstructure of the film (e.g., surface roughness Ra).31,32 Our results show that the highly (00l)-oriented Bi2Te3 film is denser with fewer defects and has larger Ra, which plays a key role in the contact area and hence in the contact resistivity.33,34 The different morphologies and surface roughness Ra of Bi2Te3 films are shown in Figure 2. As shown, the inclined-strip Bi2Te3 film of Sample 1 is very loose among columns (Figure 2a,b), while the surface of the film is rather smooth with the surface roughness Ra of 13.7 nm from its AFM topography image (Figure 2c). Compared with Sample 1, the Bi2Te3 film of Sample 2 tends to grow into a hexagonal-like sheet with an average size of about 600 nm, and the interior film is denser (Figure 2d,e). Besides, the surface of this film is rougher than the former (Figure 2f) since Ra increases to 47.2 nm. With the orientation factor of the (00l) plane increasing, the layered stacking of Sample 3 is more obvious, and the film is very dense inside (Figure 2g,h). Moreover, the surface fluctuation of the film is larger with the surface roughness of 63.9 nm (Figure 2i), which is beneficial to the increment of contact area at the interface with Cu film, thus leading to a lower contact resistance.35,36 3.2. Interfaces of Bi2Te3−Cu Multilayer Films. In order to study the electrical transport properties of the Bi2Te3−Cu interface, Cu films were deposited on Bi2Te3 films, which appeared as crown-shaped clusters composed of nanoparticles. The SEM images of the Bi2Te3−Cu multilayer films are shown in Figure 3. The interfaces are tough in all our samples due to the different oriented microstructures of our Bi2Te3 films (e.g., Ra), which indicates the possibility of the influence of surface
Scheme 3. Principle Diagrams of the Three-Probe Method Measurement Systema
a
(a) The probing approaches of the three-probe method. The numbers 1∼4 refer to the position of probe 1 ∼ probe 4. (b) The circuit principle diagram of the three-probe method, Rt, is the contact resistance between the probe and sample; Rc is the contact resistance between Cu and Bi2Te3; and Rs is the Bi2Te3 sheet resistance. −20 to 20 mA with the step of 0.02 mA, and the voltage drops between probe 3 and probe 4. It represents the voltage of Uc2, and hence, Rc2 can be extracted from Rc2 = Uc2/I (Scheme 3b).
3. RESULTS AND DISCUSSION 3.1. Oriented Microstructure of Bi2Te3 Films. The Bi2Te3 films with different microstructures were first fabricated, and the crystal structures were investigated by XRD, as shown in Figure 1a. The intense and sharp XRD peaks of the Bi2Te3 films are typical signatures of a high degree of crystallinity. Besides, the Bi2Te3 film of Sample 1 exhibits a highly preferential orientation along the (015) plane since the intensity of the (015) peak is the strongest. In Sample 2, there is an increasing orientation of (00l) but a decreasing orientation of (015), indicating a transition mode of the preference growth. Additionally, the Bi2Te3 film of Sample 3 is highly (00l) oriented. The high crystallinity and high degree of orientation of Sample 1 and Sample 3 indicate an ordered
Figure 1. (a) XRD patterns of Bi2Te3 films for the three samples and (b) orientation factor F of (015) and (006) crystal planes for different samples. 25608
DOI: 10.1021/acsami.7b05460 ACS Appl. Mater. Interfaces 2017, 9, 25606−25614
Research Article
ACS Applied Materials & Interfaces
Figure 2. Surface (a, d, g), cross-sectional view (b, e, h), and 3-D AFM topography image (c, f, i) of Bi2Te3 films for Sample 1, Sample 2, and Sample 3, respectively. The surface roughness Ra is marked at the bottom of every AFM image.
Figure 3. Surface (a) and cross-sectional view (b, c) of Bi2Te3−Cu multilayer films of Sample 1; Surface (d) and cross-sectional view (e, f) of Sample 2; and Surface (g) and cross-sectional view (h, i) of Sample 3.
states on the contact resistance.37 Additionally, it can be observed that many defects such as pores generate at the interface between the two layers of Sample 1 (Figure 3a−c), which may block the transport of electrons and thus result in a large contact resistance. However, the interface of Sample 2 is relatively dense (Figure 3d−f), and the highly (00l)-oriented Bi2Te3 of Sample 3 is more closely connected with Cu film (Figure 3g−i). The denser film connections can increase the contact area and adhesive strength, which is similar to the growth of Cu film on the alumina substrate with different surface roughness, thus providing a constant and stable thoroughfare for electrical transport.35 As reported, Bi2Te3 atoms with different orientations have different lattice misfit (Δa/aBi2Te3) with Cu atoms at the interface,18 which can be calculated by38
Δa a Bi 2Te3
=
|aCu − a Bi 2Te3| a Bi 2Te3
× 100% (6)
where aCu is the lattice constant of Cu and aBi2Te3 is the lattice constant of Bi2Te3. When the lattice mismatch |Δa/aBi2Te3| is less than 5%, the two layers are coherent to a great degree. By means of calculation, the layered structured Bi2Te3 film with (00l) orientation has a lower misfit of 2.8% at the interface with the Cu electrode, which indicates that the coherent layers have fewer defects at the interface. However, the larger mismatch 35.4% of Cu film on (015)-oriented Bi2Te3 film illustrates a complete mismatch at the interface, which allows for strain between the layers and induces crystallographic defects, therefore generating greater resistance to the transportation of carriers or phonons or macroscopic defects, such as buckling, cracking, and delamination.39 The results are in accordance 25609
DOI: 10.1021/acsami.7b05460 ACS Appl. Mater. Interfaces 2017, 9, 25606−25614
Research Article
ACS Applied Materials & Interfaces
suggest that oriented microstructures largely affect the interfacial electrical contact performances. Remarkably, the S in our paper is the designed nominal contact area, rather than the effective conducting contact areas for Bi2Te3−Cu contacts. The effective contact area S′ is defined as the real area of contact, which is conductive to the electrical transport.40 Actually, the effective contact area is not equivalent to the nominal one on most occasions because of the complexity for the interfacial contacts between films; thus, it is very difficult to detect the concrete data.41 To compare the effects of the different oriented microstructures on contact resistance, we measured area-dependent contact resistances Rc at different nominal contact areas S and explored the relationship between them. The results are summarized in Figure 5. As shown in Figure 5a, the contact resistance of Cu on
with the morphologies in Figure 3, which complements the reason for different surface states at the interface. Besides, the lower lattice misfit at the (00l)-oriented Bi2Te3−Cu interface also indicates denser film connections, thus contributing to lower contact resistances and better interfacial electrical transport properties, which will be discussed in the following section. 3.3. Electrical Transport Properties. As a preliminary test, I−V tests were applied to the Bi2Te3−Cu contacts by the three-probe method to reveal the Ohmic vs Schottky nature of the contacts. As shown in Figure 4a−c, symmetric I−V curves
Figure 5. (a) Contact resistance histogram of different oriented Bi2Te3−Cu contacts at various contact areas (inset is the enlarged part of the red frame). (b) Relationship between Rc and S made by nonlinear curve fitting.
(00l)-oriented Bi2Te3 is very small at each nominal contact area, exhibiting a very good ohmic contact. In comparison, the Rc of each (015)-oriented Bi2Te3−Cu contact is the largest at every contact area. Additionally, the relationship between Rc and S is made by nonlinear fitting (Figure 5b). Results show that, for all the Bi2Te3−Cu contacts, the contact resistance increases nonlinearly with the decrease of contact area, which is in accordance with a nonlinear relationship of Rc = aSb (a and b are constants, a > 0, b < −1). However, it can be observed that the dependence of Rc on S is greater than the ideal condition since b < −1.41 The deviation from the ideal state indicates the existence of the near surface depletion region at the interface, which makes the conductive cross-section area much smaller than the ideal cross-section. A similar phenomenon was also observed in contacts between Si nanowires and electrodes.42,43
Figure 4. I−V characteristics for different oriented Bi2Te3−Cu contacts at various contact areas measured by the three-probe method: (a) (015) orientation, (b) (00l) + (015), and (c) (00l) orientation. The S in the Figure refers to the designed nominal contact area.
were obtained for all samples at various contact areas S. The curves exhibit very good linear relationship, indicative of ohmic contact formation.17 Since the three-probe method can directly measure the contact resistance of Cu film on Bi2Te3, the slope of every I−V curve is the contact resistance of Bi2Te3 and Cu at a certain contact area. Further comparing the three curves, the different slopes among different oriented Bi2Te3−Cu contacts 25610
DOI: 10.1021/acsami.7b05460 ACS Appl. Mater. Interfaces 2017, 9, 25606−25614
Research Article
ACS Applied Materials & Interfaces Further comparing the three curves in Figure 5b, the Rc of (015)-oriented Bi2Te3−Cu contact depends on S most; on the contrary, that of (00l)-oriented Bi2Te3−Cu contact depends on S least. This result indicates a smaller surface depletion region and larger conductive cross-section area with the gradual orientation variation from (015) to (00l). Obviously, the differences of the contact resistances among Cu on Bi2Te3 with different oriented microstructures are amplified at smaller contact areas (