NANO LETTERS
Atomic Scale Alignment of Copper-Germanide Contacts for Ge Nanowire Metal Oxide Field Effect Transistors
2009 Vol. 9, No. 11 3739-3742
T. Burchhart, A. Lugstein,* Y. J. Hyun, G. Hochleitner, and E. Bertagnolli Institute for Solid State Electronics, Vienna UniVersity of Technology, Floragasse 7, A-1040 Vienna, Austria Received June 16, 2009; Revised Manuscript Received August 7, 2009
ABSTRACT In this letter, we report on the formation, of copper-germanide/germanium nanowire (NW) heterostructures with atomically sharp interfaces. The copper-germanide (Cu3Ge) formation process is enabled by a chemical reaction between metallic Cu pads and vapor-liquid-solid (VLS) grown Ge-NWs. The atomic scale aligned formation of the Cu3Ge segments is controlled by in situ SEM monitoring at 310 °C thereby enabling length control of the intrinsic Ge-NW down to a few nanometers. The single crystal Cu3Ge/Ge/Cu3Ge heterostructures were used to fabricate p-type Ge-NW field effect transistors with Schottky Cu3Ge source/drain contacts. Temperature dependent I /V measurements revealed the metallic properties of the Cu3Ge contacts with a maximum current density of 5 × 107 A/cm2. According to the thermoionic emission theory, we determined an effective Schottky barrier height of 218 meV.
One dimensional nanostructures, such as nanotubes1-5 and nanowires6-8 have attracted a lot of attention as natural candidates for a wide range of novel devices having applications in nanoelectronics,9,10 bio/chemical sensors,11,12 light emitting devices with extremely low power consumption, and solar cells.13-15 After more than four decades of classical integrated electronic circuit technology that followed Moore’s famous law,16 particularly semiconducting NWs are in the center of interest as alternative building blocks for metal oxide field effect transistors (MOSFETs)17 as well as novel quantum devices.18 For both applications, germanium appeared to be a favorable material for two reasons. First, the high carrier mobility19 enables high performance devices; second, the large exciton Bohr radius implies that quantum confinement effects of Ge nanostructures will be more prominent.20 However, reliable electrical contacts to NWs are still a key issue that determines device performance and reliability. Up to now, mostly all investigations of NW-based devices were made with conventional metal contacts defined by e-beam lithography, metal layer deposition, and lift-off techniques. Thereby interface states between the NW and the metal contacts often lead to Fermi-level pinning and hence detrimental large Schottky barriers for the devices.21 For Si-NWs, one can overcome this problem * To whom correspondence should be addressed. E-mail: alois.lugstein@ tuwien.ac.at. 10.1021/nl9019243 CCC: $40.75 Published on Web 08/19/2009
2009 American Chemical Society
by the formation of nickel-22-24 or platinum-25 silicide source/drain contacts. For Ge-NWs, we explored a quasi-metallic coppergermanide-based process leading to atomically sharp interfaces with an expected very low barrier height of ∼0.06 eV with respect to the valence band of germanium enabling quasi-ohmic contacts to p-type Ge-NWs. Moreover we show that Cu3Ge, which can be formed at rather low temperatures of about 310 °C, has a very low resistivity compared to typical silicides for silicon CMOS26,27 and can sustain current densities of 5 × 107A/cm2. The Ge-NWs were synthesized via the VLS process in a hot wall chemical vapor deposition system using GeH4 (2% in He) as the precursor gas. As the NW growth promoting catalyst, a 2 nm thick Au layer was sputter deposited on Si(100) substrates pretreated by buffered hydrofluoric acid for native oxide removal. The vapor-liquid-solid (VLS) synthesis for 45 min at 300 °C and a total pressure of 75 mbar leads to rod like NWs with lengths of about 3 µm and diameters between 20 and 40 nm. The NWs grow epitaxially with an inclined angle of 35.27° with respect to the Si (100) surface indicating a 〈111〉 NW growth direction (see Supporting Information). High-resolution transmission electron microscopy (HRTEM) investigations proved the 〈111〉 growth direction and the NWs appeared to be single crystalline and defect free with a 2 nm thick native germanium oxide shell. To form, reliable
Figure 1. (a) SEM image of a Cu3Ge/Ge/Cu3Ge NW-heterostructure with an unreacted germanium segment of 500 nm. The inset shows the schematic illustration of the contacted NW. Brown represents the copper contact pads, red is the Cu3Ge NW segment, and green shows the remaining intrinsic Ge-NW. (b) Lattice-resolved HRTEM image of the Ge-NW segment with the [1-11] growth direction. The inset shows fast Fourier transform (FFT) pattern confirming the Ge [-112] zone axis. (c) HRTEM image showing the sharp interface between Ge and Cu3Ge. (d) Lattice-resolved HRTEM image of the Cu3Ge-NW with the [211] growth direction. The inset shows the FFT pattern confirming the Cu3Ge [-102] zone axis.
contacts appeared to be critically dependent on a complete removal of the native oxide layer prior to the metal deposition. Within the hydrogen halide acids, we obtained the best results for hydroiodic acid, which is in well accordance with the literature.28 Nevertheless the resistance of Au contacts with a titanium adhesion layer was determined to be about 3 orders of magnitude greater than the resistance of the Ge-NW. Four-probe measurements of such grown GeNWs with Ti/Au contact pads revealed an intrinsic resistivity of about 11 Ωcm. Moreover, gate-dependent current versus bias voltage measurements of back-gated NW-FETs have shown that the VLS grown Ge-NWs exhibit a field effect response characteristic of a p-type semiconductor (see Supporting Information). Since the Ti/Au contacts do not allow proper contacting, we explored a novel process forming reliable quasi-metallic copper-germanide contacts with an atomically sharp interface to the Ge-NWs. Since copper-germanide has a work function of 4.6 eV, we expect a very small energy barrier of about 0.06 eV to the valence band of Ge resulting in ideal ohmic contacts for p-channel Ge-MOSFETs. To investigate the crystal structure of the germanides formed, Ge-NWs were dispersed on a 50 nm thick silicon nitride membrane. Afterward Cu pads were processed to connect the Ge-NW. During annealing at 310 °C in hydrogen atmosphere, copper-germanide segments emerging from the Cu contact pads are formed through axial diffusion of copper (see scanning electron microscopy (SEM) image in Figure 1a). The extension of the copper-germanide region and accordingly the length of the remaining Ge-NW segment (∼500 nm in Figure 1a) critically depend on the annealing temperature, annealing time, and the diameter of the NWs. For tapered NWs, we observed that for increasing NW diameter the diffusion of Cu and thereby the coppergermanide phase formation became slower and stopped completely for NWs with diameters > 150 nm. Figure 1b shows the HRTEM micrograph of the remaining Ge-NW segment with the (111) atomic planes separated by 0.327 nm in good agreement with tabulated values.29 The corre3740
Figure 2. Cu3Ge phase formation during annealing at 310 °C monitored by in situ SEM imaging. (a) SEM image of the 1 µm long intrinsic Ge-NW with Cu contact pads before the germanidation process. (b) SEM snapshot of Cu3Ge/Ge/Cu3Ge heterostructures after 2 min annealing with a remaining length of the unreacted Ge-NW segment of about 180 nm. (c) SEM snapshot after 4 min annealing. The unreacted Ge-NW length is about 15 nm.
sponding reciprocal lattice peaks obtained from the Fourier transformation proves the 〈111〉 growth direction of the Ge-NW. The Cu3Ge appeared to be orthorhombic with lattice constants of a ) 0.528 nm, b ) 0.422 nm, and c ) 0.454 nm. The HRTEM image in Figure 1d shows the wellresolved (020) planes and (201) planes of Cu3Ge oriented NWs in the 〈211〉 direction. The crystallographic relationship between the Ge and Cu3Ge are Ge[-112]//Cu3Ge[-102] and Ge(1-11)//Cu3Ge(211). The atomically sharp interface between the Ge-NW and the Cu-germanide (see Figure 1c) might be possible due to increased tolerance in lattice mismatch of one-dimensional NWs. The Cu3Ge phase formation can be monitored and even controlled in situ when the annealing is performed in a SEM system with a heating stage. Prior to the annealing, the 35 nm thick and 1 µm long Ge-NW clamped in between the copper pads show a uniform contrast (see Figure 2a). For the annealing, the temperature was ramped up by about 10 °C/min to the final annealing temperature of 310 °C. Nano Lett., Vol. 9, No. 11, 2009
Figure 3. (a) I/V characteristic of a fully germanided Cu3Ge-NW and the Cu3Ge/Ge/Cu3Ge-NW heterostructures as a function of the temperature for a bias voltage of 20 mV (Cu3Ge-NW) and 100 mV (Cu3Ge/Ge/Cu3Ge-NW heterostructures). The inset shows the temperature-dependent normalized two-terminal resistance of individual Cu3Ge-NW; the resistance, R, is normalized by the value at 298.15 K, R0. (b) ln(I/T2) as a function of inverse temperature at various bias voltages for an individual Cu3Ge/Ge/Cu3Ge-NW. The upper right inset introduces a model of circuit diagram. (c) Band relation between the Cu3Ge contacts and the intrinsic Ge-NW. (d) Transconductance measurement for the Cu3Ge/Ge/Cu3Ge-NW heterostructure MOSFET device in back-gate configuration. The NW heterostructure was formed on a 200 nm silicon dioxide as gate dielectric on a highly doped Si substrate. The inset shows the semi log plot of the full ID vs VG measurement.
During the annealing, the right and the left end of the NW adjacent to the Cu pads became brighter in contrast. In situ EDX analysis proved the formation of Cu-germanide at both ends of the NW which in the subsequently performed HRTEM was specified as Cu3Ge. After 2 min maintaining at 310 °C, Cu3Ge/Ge/Cu3Ge heterostructures are formed with a remaining Ge-NW length of about 180 nm (Figure 2b). The velocity of diffusion at the final temperature of 310 °C appeared to be 40 nm/min. Finally, we stopped the annealing when the Ge-NW segment was about 15 nm. Thus we were able to control the geometric length of the intrinsic Ge-NW segment down to a few nanometers. Under the given experimental conditions, annealing of about 5 min results in a pure Cu3Ge-NWs, which means that all the Ge is converted to Cu-germanide. To investigate the electrical transport properties of such generated NW heterostructures, we performed temperature dependent I/V measurements. Figure 3a shows for comparison the current through a 2 µm long fully germanided NW (red line) and a Cu3Ge/Ge/Cu3Ge-NW heterostructures (green line) with an unreacted 1 µm long Ge-NW segment. For the pure Cu3Ge-NW, the current increases for lower temperatures. The inset shows in more detail that the resistivity decreases monolithically down to about 40 K and then Nano Lett., Vol. 9, No. 11, 2009
saturates as expected for a metal.30 The specific resistivity of such fully germanided NWs was determined to about 34 µΩcm at room temperature. Previous reported values for single crystalline Cu3Ge thin film structures show resistivities in the range of 5.5-24 µΩcm.31,32 The current through such a 30 nm thick Ge-NW typically exceeds 100 µA at a bias voltage of 10 mV. The ampacity of such Cu3Ge-NW is remarkably high. The highest measured current before electrical breakdown was 12 mA corresponding to a current density of 5 × 107 A/cm2. For the Cu3Ge/Ge/Cu3Ge-NW heterostructures, the current at room temperature appeared to be 3 orders of magnitude lower in comparison to the fully germanided NW although the bias voltage is five times greater. Also opposite to the Cu3Ge-NW the current decreases with decreasing temperature indicating that the semiconducting Ge-NW segment determines the electrical properties of the overall NW heterostructure. To determine the Schottky barrier height (SBH) of the Cu3Ge/Ge interface, we propose a model of two back-toback Schottky contacts which is commonly addressed to explain nonlinear I-V curves of semiconductor NW devices33 (inset Figure 3b). Assuming thermionic emission, the appropriate I-V-T relation is of a reverse-biased Schottky diode according to the following equation21 3741
I ) AA**T2 exp
( ) -qφB kBT
where A is the contact area, A** is the effective Richardson constant, and φB the effective barrier height. φB can be obtained from the slope of Figure 3b which is an activation energy plot of ln(I/T2) versus 1/T at various bias voltages. This holds only for high temperatures where A** and φB are temperature independent.34 This is in good agreement with the bias independent slope of the activation energy plot in the linear low-bias regime. The calculated SBH of 218 meV is about 50 meV below the theoretical value of 270 meV, which is shown for an intrinsic Ge-NW in the band diagram in Figure 3c. On the basis of such Cu3Ge/Ge/Cu3Ge-NW heterostructures, we fabricated back-gated MOSFETs, where the unreacted inner 700 nm long Ge segment forms the active channel connected to Cu3Ge source/drain extensions. The typical transconductance characteristic in Figure 3d shows that with increasing negative gate voltage the current through the wire increases appreciably, characteristic for a p-channel enhancement mode transistor. Above the threshold voltage of about -2.5 V the drain current increases nearly linear with VG. The logarithmic plot of ID versus VG (inset, Figure 3d) shows the ION/IOFF ratio to be better than 103 and the mobility was calculated to be 264 cm2 V-1 s-1 (for calculation, see Supporting Information). In summary we have grown epitaxially, single crystal GeNWs on Si(100) substrates. Thereof we formed Cu3Ge/Ge/ Cu3Ge-NW heterostructures with atomically sharp interfaces between the metallic Cu3Ge segments and the intrinsic GeNW. With in situ SEM process monitoring, we are able to control the length of the remaining Ge segment down to a few nanometers limited only by the resolution of our SEM. Electrical characterization of the Cu3Ge-NWs show a specific resistivity of 34 µΩcm and a high current breakdown density beyond 5 × 107 A/cm2. We have determined the effective Schottky barrier height for Cu3Ge contacts to Ge-NWs with temperature dependent I/V measurements according to the thermoionic emission theory. Furthermore we fabricated p-channel enhancement mode MOSFETs with the use of atomically sharp, defect free Cu3Ge/Ge/Cu3Ge-NW heterostructures. The formation of such perfect heterostructures might be a powerful tool to develop short channel metal oxide semiconductor transistors down to the sub-10 nm regime were the channel length is controlled by chemical reaction inside the NW and not restricted by a lithographical process. Acknowledgment. This work is partly funded by the Austrian Science Fund (Project Nos. P20937-N14 and L332-N16) and the Austrian Society for Micro- and Nanoelectronics (GMe). Technical Support by USTEM TU-Wien is gratefully acknowledged. Supporting Information Available: SEM image of epitaxially grown Ge-NWs on Si(100) with a schematic
3742
illustration, a single nanowire with four-probe contact pads, I-V curves of the four-probe measurement and the corresponding two-point measurement to determine the contact resistance, field effect measurements and calculation of the channel mobility, and drain current versus gate voltage curves in linear and logarithmic scales. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Rosenblatt, S.; Yaish, Y.; Park, J.; Gore, J.; Sazonova, V.; McEuen, P. L. Nano Lett. 2002, 28, 869. (2) Yao, Z.; Kane, C. L.; Dekker, C. Phys. ReV. Lett. 2000, 84, 2941. (3) Martel, R.; Schmidt, T.; Shea, H. R.; Hertel, T.; Avouris, Ph. Appl. Phys. Lett. 1998, 73, 2447. (4) Wind, S. J.; Appenzeller, J.; Martel, R.; Derycke, V.; Avouris, Ph. Appl. Phys. Lett. 2002, 80, 3817. (5) Martel, R.; Derycke, V.; Lavoie, C.; Appenzeller, J.; Chan, K. K.; Tersoff, J.; Avouris, Ph. Phys. ReV. Lett. 2001, 87, 256805. (6) Duan, X.; Huang, Y.; Cui, Y.; Wang, J.; Lieber, C. M. Nature 2000, 409, 66. (7) Cui, Y.; Duan, X.; Hu, J.; Lieber, C. M. J. Phys. Chem. B 2000, 104, 5213. (8) Huang, Y.; Duan, X.; Cui, J.; Lauhon, J.; Kim, K.-H.; Lieber, C. M. Science 2001, 294, 1313. (9) Duan, X.; Huang, Y.; Lieber, C. M. Nano Lett. 2002, 2, 487. (10) Javey, A.; Nam, S.; Friedman, R. S.; Yan, H.; Lieber, C. M. Nano Lett. 2007, 7, 773. (11) Cui, Y.; Lieber, C. M. Science 2000, 291, 891. (12) Zheng, G.; Patolsky, F.; Cui, Y.; Wang, W. U.; Lieber, C. M. Nat. Biotechnol. 2005, 23, 1294. (13) Gradecak, S.; Qian, F.; Li, Y.; Park, H.; Lieber, C. M. Appl. Phys. Lett. 2005, 87, 173111. (14) Barrelet, C. J.; Agarwal, R.; Lieber, C. M. Nano Lett. 2005, 5, 917. (15) Persson, A. I.; Pettersson, H.; Tra¨gardh, J.; Samuelson, L. Nano Lett. 2006, 6, 229. (16) Moore, G. E. Electronics. 1965, 38, 8. (17) Xiang, J.; Lu, W.; Hu, Y.; Wu, Y.; Yan, H.; Lieber, C. M. Nature 2006, 441, 489. (18) Lu, W.; Xiang, J.; Timko, B. P.; Wu, Y.; Lieber, C. M. Proc. Nat. Acad. Sci. U.S.A. 2005, 102, 10046. (19) Prince, M. B. Bell Teleph. Lab. 1953, 92. (20) Maeda, Y.; Tsukamoto, N.; Yazawa, Y.; Kanemitsu, Y.; Matsumoto, Y. Appl. Phys. Lett. 1991, 59, 3168. (21) Leonard, F.; Talin, A. A. Sandia National Laboratories. Private communication, 2008. (22) Weber, W. M.; Geelhaar, L.; Graham, A. P.; Unger, E.; Duesberg, G. S.; Liebau, M.; Pamler, W.; Chez`e, C.; Riechert, H.; Lugli, P.; Kreupl, F. Nano Lett. 2006, 6, 2660. (23) Wu, Y-.; Xiang, J.; Yang, C.; Lu, W.; Lieber, C. M. Nature 2004, 430, 61. (24) Yamamoto, K.; Kohno, H.; Takeda, S.; Ichikawa, S. Appl. Phys. Lett. 2006, 89, 083107. (25) Lin, Y.; Lu, W.; Wu, W.; Bai, J.; Chen, L. J.; Tu, K. N.; Huang, Y. Nano Lett. 2008, 8, 913. (26) Gaudet, S.; Detavernier, C.; Kellock, A. J.; Desjardins, P.; Lavoie, C. J. Vac. Sci. Technol., A 2006, 24, 474. (27) Chao, Y.-L.; Xu, Y.; Scholz, R.; Woo, J. C. S. IEEE Elec. DeV. Lett. 2006, 27, 7. (28) Onsia, B.; Conard, T.; Gendt, S. D.; Heyns, M.; Hoflijk, I.; Mertens, P.; Meuris, M.; Raskin, G.; Sioncke, S.; Teerlinck, I.; Theuwis, A.; Van Steenbergen, J.; Vinckier, C. Solid State Phenom. 2005, 103104, 27–30. (29) CaRIne Crystallography 3.1. (30) Song, Y.; Schmitt, A. L.; Jin, S. Nano Lett. 2007, 7, 965. (31) Krusin-Elbaum, L.; Aboelfotoh, M. O. Appl. Phys. Lett. 1991, 58, 1341. (32) Tawancy, H. M.; Aboelfotoh, M. O. J. Mater. Sci. 1995, 30, 6053. (33) Nam, C. Y.; Tham, D.; Fischer, J. E. Nano Lett. 2005, 5, 2029. (34) Sze, S. M. Physics of Semiconductor DeVices; Wiley: New York, 1981.
NL9019243
Nano Lett., Vol. 9, No. 11, 2009
