- About this Journal ·
- Abstracting and Indexing ·
- Aims and Scope ·
- Article Processing Charges ·
- Articles in Press ·
- Author Guidelines ·
- Bibliographic Information ·
- Citations to this Journal ·
- Contact Information ·
- Editorial Board ·
- Editorial Workflow ·
- Free eTOC Alerts ·
- Publication Ethics ·
- Reviewers Acknowledgment ·
- Submit a Manuscript ·
- Subscription Information ·
- Table of Contents
Advances in Materials Science and Engineering
Volume 2011 (2011), Article ID 316513, 16 pages
Formation and Device Application of Ge Nanowire Heterostructures via Rapid Thermal Annealing
1Device Research Laboratory, Department of Electrical Engineering, University of California, Los Angeles, CA 90095, USA
2Department of Materials Science and Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan
3Department of Electrical and Computer Engineering, Iowa State University, Ames, IA 50014, USA
Received 21 April 2011; Accepted 14 June 2011
Academic Editor: Joseph Lai
Copyright © 2011 Jianshi Tang et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We reviewed the formation of Ge nanowire heterostructure and its field-effect characteristics by a controlled reaction between a single-crystalline Ge nanowire and Ni contact pads using a facile rapid thermal annealing process. Scanning electron microscopy and transmission electron microscopy demonstrated a wide temperature range of 400~500°C to convert the Ge nanowire to a single-crystalline Ni2Ge/Ge/Ni2Ge nanowire heterostructure with atomically sharp interfaces. More importantly, we studied the effect of oxide confinement during the formation of nickel germanides in a Ge nanowire. In contrast to the formation of Ni2Ge/Ge/Ni2Ge nanowire heterostructures, a segment of high-quality epitaxial NiGe was formed between Ni2Ge with the confinement of Al2O3 during annealing. A twisted epitaxial growth mode was observed in both two Ge nanowire heterostructures to accommodate the large lattice mismatch in the NixGe/Ge interface. Moreover, we have demonstrated field-effect transistors using the nickel germanide regions as source/drain contacts to the Ge nanowire channel. Our Ge nanowire transistors have shown a high-performance p-type behavior with a high on/off ratio of 105 and a field-effect hole mobility of 210 cm2/Vs, which showed a significant improvement compared with that from unreacted Ge nanowire transistors.
As an important one-dimensional material, semiconductor nanowire has attracted enormous research interest for its unique electrical properties, which has promising applications as building blocks for nanoelectronics [1–5]. Since 2004, there have been a lot of efforts on studying the thermal diffusion of metal into a single-crystalline Si nanowire, in which a silicide/silicon/silicide nanowire heterostructure is formed by solid-state reactions between the Si nanowire and metal contacts. Many metals (contact pads or nanowires) have been studied as the diffusion source, such as Ni [6–8], Co , Pt , and Mn . One of the salient features in this nanowire heterostructure is the atomically sharp interface between the Si nanowire and the formed silicide nanowire. Such clean interface may help to avoid Fermi-level pinning effect, which is commonly observed in conventional metal-semiconductor contacts . Also, the nanowire heterostructure can be easily used to fabricate nanowire field-effect transistors (FETs) using the formed silicide regions as the source/drain contacts to the Si nanowire channel [6, 7, 10]. The channel length can be well controlled by the annealing time and growth length of silicide, therefore, can be aggressively shrunk down to sub-50 nm . Clearly, this process offers great advantages over modern high-cost and complex photolithography technology to fabricate short-channel transistors and may further facilitate the advance of scaled nanodevices.
Compared with Si nanowire, metal-Ge nanowire is a new system of interest, since Ge is an important complement to Si with even higher carrier mobilities for further device miniaturization compatible to the existing CMOS technology [1, 2]. For Ge, the atomically sharp interface is of particular interest to alleviate the Fermi-level pinning effect in the metal-Ge contact, since Ge has a high density of interface states due to native defects on the Ge surface . Experimentally, Yamane et al. have demonstrated the epitaxial growth of high-quality Fe3Si on Ge(111) substrate with an atomically controlled interface, which successfully depined the Fermi-level of Fe3Si/Ge contact . It is, therefore, highly desirable to achieve a metallic contact to Ge with a high-quality interface. On the other hand, many germanides, such as Mn5Ge3 and Ni3Ge, exhibit ferromagnetism above room temperature [20, 21], and thus offer great advantages over silicides for future applications in spintronics, such as realizing spin injection into semiconductor from a ferromagnetic contact. Therefore, there is an increasing research interest on the metal-Ge nanowire system, such as Ni-Ge [13–15] and Cu-Ge [16–18]. For comparison, Table 1 briefly summarizes the literature report of Si and Ge nanowire heterostructures formed by solid-state reactions between a semiconductor nanowire and metal contacts. It is noted that the annealing temperature of metal-Ge nanowire system is usually lower than that of metal-Si nanowire system, which is partially due to a lower melting point of Ge than Si. The low-temperature process in metal-Ge nanowire systems is favorable to reduce the thermal budget in forming nanowire heterostructures toward future process integration.
In this paper, we will first discuss the formation of single-crystalline Ge nanowire heterostructure by the solid-state reaction between a Ge nanowire and Ni contact pads. By a comparative study of with or without an Al2O3 capping layer during annealing, we study the effect of oxide confinement in the growth of germanide in a Ge nanowire. A detailed transmission electron microscopy (TEM) analysis including the epitaxial relationships is presented in this section. In the second part of this paper, we will introduce electrical characterizations of Ge nanowire back-gate FETs fabricated using the Ge nanowire heterostructures. Specifically, the effect of Al2O3 capping layer on the device performance is studied in this section. Finally, we will discuss the possible research directions for future work on the Ge nanowire heterostructures, especially for promising applications in spintronics and further study on the growth dynamics.
2. Experimental Results
The growth of Ge nanowires can be accomplished by a variety of techniques. In this study, two popular methods were employed to synthesize single-crystalline Ge nanowire with growth direction. One is the supercritical fluid-liquid-solid (SFLS) approach, in which Ge nanowires were produced in highly pressurized supercritical fluids enriched with organogermane precursors and metal nanocrystals as catalyst [22, 23]. The typical diameter of as-synthesized Ge nanowires is around 40–50 nm and the length could be more than 10 μm. The other approach is the vapor-liquid-solid (VLS) method, in which metal (such as Au) catalyzed Ge nanowires were grown on SiO2/Si(100) substrates by means of chemical vapor deposition (CVD) using a gaseous Ge precursor GeH4 . The VLS-grown Ge nanowires are typically 70–80 nm in diameter and have lengths over 10 μm. The reported carrier mobility of VLS-grown Ge nanowires is higher than SFLS-synthesized Ge nanowires [13, 14], while the latter method is claimed to provide a better control of the nanowire size and a higher product yield [23, 25]. In both two methods, Ge nanowires are not doped on purpose during growth, but unintentional doping usually occurs [26, 27].
To form NixGe/Ge nanowire heterostructures, SFLS-synthesized Ge nanowires diluted in isopropyl alcohol (IPA) were dispersed onto a SiO2/Si substrate. The top thermal SiO2 was about 330 nm thick. The Si substrate was degenerately doped with a resistivity of 1–5 ×10−3 Ω-cm, which served as a back gate for further device characterization. E-beam lithography (EBL) was used to define Ni contacts to Ge nanowires. Before e-beam evaporation of about 120 nm-thick Ni (with the purity of 99.995% and in vacuum at a pressure lower than 10−6 Torr), the sample was dipped into diluted hydrofluoric acid (HF) for 15 s to completely remove native oxide in the contact region. A field-emission scanning electron microscopy (JEOL JSM-6700 FESEM) was used to examine the sample morphology before and after the annealing process. Figures 1(a) and 1(b) show the device schematics before and after the thermal diffusion of Ni into the Ge nanowire. Figure 1(c) shows the SEM image of the as-fabricated Ge nanowire device showing a uniform contrast. Then the sample was annealed with rapid thermal annealing (RTA) in the ambient of N2 to allow Ni thermal intrusion into Ge nanowire and subsequently form NixGe/Ge heterostructures along the nanowire. In the previous study on the interfacial reactions of Ni thin film on Ge(111) substrate, the germanide phase formation sequence was found to be Ni2Ge and NiGe at increasing temperatures in the range of 160°C to 600°C . Various annealing temperatures ranging from 400°C to 700°C were used in this study to optimize the formation of nanowire heterostructures. It is found out that Ge nanowires were easily broken at a high annealing temperature (>550°C) due to the significant reduction of the melting point for Ge nanowires compared with that of bulk Ge . When the temperature decreased to 400°C–500°C, clear diffusion of Ni into the Ge nanowire was also observed and the formed germanide was identified to be Ni2Ge (refer to the TEM analysis later). Figure 1(d) shows the SEM image of the Ge nanowire device upon RTA at 500°C for 60 s, in which clear contrast was observed between the Ge nanowire and the formed nickel germanide nanowire due to the conductivity difference. The remained Ge region was easily controlled down to 650 nm, and it can be further reduced to sub-100 nm [11, 16]. Similar contrast was also observed after RTA at 400°C for 40 s, as shown in Figure 1(e).
In order to identify the phase of the formed germanide and the epitaxial relationship of germanide-germanium interface, in situ TEM was used to study the formation process and reaction kinetics. To prepare the TEM sample, the single-crystalline Ge nanowires were dispersed on the TEM grid with a square opening of a Si3N4 thin film. The low-stress Si3N4 film was deposited by low-pressure chemical vapor deposition (LPCVD). The Si3N4 film was about 50 nm thick, which provided a reliable mechanical support for Ge nanowire devices during the fabrication process and at the same time to assure it is transparent to the electron beam without interference with images of the nanowires. EBL-defined Ni pads were employed as the Ni diffusion source. A JEOL-2010 TEM (operated at 200 KV with a point-to-point resolution of 0.25 nm) attached with an energy dispersive spectrometer (EDS) was used to investigate the microstructures and to determine the compositions of the samples. To in situ observe the reactions of the Ni electrodes with Ge nanowires, the samples were heated inside TEM with a heating holder (Gatan 652 double tilt heating holder connected with a power supply to heat up the samples to the desired temperature) under a RTA mode with a pressure below 10−6 Torr. Figures 2–2 show high-resolution TEM (HRTEM) images of the formed NixGe/Ge interface upon 500°C annealing. According to the lattice-resolved HRTEM analysis, the formed germanide was identified to be single-crystalline Ni2Ge with an orthorhombic lattice structure and lattice constants nm, nm, and nm (space group 62). It was observed that a large lattice mismatch of 56.3% at the Ni2Ge/Ge epitaxial interface could result in the segregation of nanoparticles (see Figure 4). In Figure 2(b), a clean and sharp interface between Ni2Ge/Ge was observed with an approximately 1 nm GeOx shell surrounding both the Ge and Ni2Ge regions. The insets in Figures 2(a) and 2(c) illustrate the fast Fourier transform (FFT) patterns of Ni2Ge and Ge HRTEM images, respectively. The crystallographic epitaxial relationships between Ge and Ni2Ge are shown to be and Ge(1)//Ni2Ge(100), as we discussed. Figures 2(d) and 2(e) show the low-magnification TEM images of Ge NWs before and after annealing at 500°C, respectively. Figure 2(f) shows the EDS of the formed germanide nanowire region, showing that the ratio of Ni to Ge concentration is about 2 : 1, which further support the fact that the formed germanide phase is Ni2Ge. The signals of Si and N peaks are contributed from the Si3N4 window.
Real-time observation on the Ni2Ge growth in a Ge nanowire was performed using in situ TEM video, which allows us to obtain lattice-resolved TEM images of the epitaxial interface in progression and thus to estimate the growth velocity. Figure 3(a) shows the relation of Ni2Ge nanowire length versus the reaction time at 400 and 500°C, illustrating a potentially linear growth behavior of Ni2Ge in the Ge nanowires, while the detailed growth mechanism requires further study. Figures 3(b), 3(c), 3(d), and 3(e) show the in situ TEM images of the Ni2Ge nanowire growth at 400 and 500°C, respectively. The growth length of the Ni2Ge nanowire is 138.9 nm for 455 s at 400°C and 357.5 nm for 340 s at 500°C, respectively. Based on the data collected on more than three nanowires, the extracted growth velocities are about 0.31 nm/s at 400°C and 1.05 nm/s and 500°C, respectively. Using the Arrhenius plot , the activation energy of the Ni2Ge growth in the Ge nanowire is estimated to be 0.55 ± 0.05 eV/atom.
As mentioned above, due to a large lattice mismatch on the Ni2Ge/Ge epitaxial interface, the resulting huge strain in the Ni2Ge/Ge/Ni2Ge nanowire heterostructure could lead to the segregation of Ni2Ge nanoparticles on the Ni2Ge nanowire after a long-time annealing. Figure 4(a) shows the TEM image of an as-fabricated Ge nanowire device with EBL-defined Ni pads at room temperature. Figures 4(b)–4(d) are a series of in situ TEM images of the Ge nanowire device upon 400°C, 450°C and 500°C sequential annealing, respectively. The time clocks shown at the lower-right corner in each TEM image were captured in the form of hour: minute:second. Volume expansion and segregation to form nanoparticles were clearly observed. Figure 4(e) shows a TEM image of another Ge nanowire device annealed at 400°C for 30 min. The results in Figures 4(c)–4(e) clearly demonstrate that as a result of strain release, Ni2Ge nanoparticles were formed and segregated on the Ni2Ge nanowire as Ni diffused along the formed Ni2Ge nanowire.
Similarly, the segregation of nanoparticles during annealing was also observed in the Ni-Si nanowire system as reported by Weber et al. . For Si, a thin layer of high-quality native oxide is usually formed on the Si nanowire surface, while Ge does not have a stable native oxide. It is observed that the SiO2 shell has a substantial confinement effect on the nickel silicide growth along with phase transformation in a Si nanowire . We studied the effect of oxide confinement on the germanide growth in a Ge nanowire; an oxide layer, such as Al2O3, was deposited to cap the Ge nanowire device before the annealing process, as shown in the schematics of Figure 5(a). Figure 5(b) shows the SEM image of the as-fabricated Ge nanowire device showing uniform contrast. Prior to RTA, 20 nm thick Al2O3 was deposited on top by atomic layer deposition (ALD) at 250°C (Figure 5(c)). It was noticed that the diameter of the Ge nanowire was increased after the Al2O3 deposition since ALD provided a conformal coverage. After that the sample was then annealed with RTA in N2 ambient to allow for the thermal intrusion of Ni into the Ge nanowire and subsequently form the NixGe/Ge heterostructures along the nanowire. Clear diffusion of Ni into the Ge nanowire was observed in both SEM and TEM, and the formed germanide was analyzed in HRTEM, as to be explained further later. Figure 5(d) shows the SEM image of the NixGe/Ge/NixGe heterostructures after RTA at 450°C for 20 s. Clear contrast was observed between the Ge nanowire and the formed nickel germanide nanowire, which is again attributed to the conductivity difference of the two. Figure 5(e) schematically illustrates the formation of NixGe/Ge/NixGe nanowire heterostructure with the Al2O3 confinement. The red line indicates the position where the device was cut with focused ion beam (FIB) to study the cross-sectional structure, as to be explained in Figure 6.
Figure 6(a) shows the low-magnification cross-sectional TEM image of the NixGe/Ge nanowire heterostructure, which was cut with FIB from the Ge nanowire device fabricated on a SiO2/Si wafer, as shown in Figure 5(e). Before FIB cutting, a 200 nm-thick Pt film was deposited on top to protect the NixGe/Ge/NixGe nanowire heterostructure from the ion beam bombardment. When preparing the cross-section of NixGe for TEM analysis using FIB, we chose the position as close as to the NixGe/Ge interface. It is noted that in the cross-sectional view, there were some nanoparticles segregated on the SiO2 surface from the bottom of the formed nickel germanide, the region that was not confined by the Al2O3 capping layer. This result also explains the fact that we did not observe nanoparticles on the Al2O3 capped surface from the SEM image shown in Figure 5(d). Figure 6(b) shows the lattice-resolved HRTEM image of the interface between the formed NiGe nanowire and the segregated NiGe nanoparticle, which were both identified to be NiGe from the FFT pattern shown in the inset of Figure 6(b). NiGe has an orthorhombic lattice structure with lattice constants nm, nm, and nm (space group 62). In order to further confirm the germanide phase, an EDS line scan was performed through the Al2O3-capped NixGe nanowire to determine the profiles of Ge, Ni, Al and O atoms, as shown in the cross-sectional TEM image in Figure 6(c) as well as the individual line-scan profile in Figures 6(d)–6(g). From Figures 6(d) and 6(e), the Ni/Ge ratio was estimated to be about 1 : 1, which further proved the NiGe phase.
To study the epitaxial relationships between NixGe and Ge, the VLS-grown Ge nanowires were dispersed on the TEM grids with a square opening of a Si3N4 thin film, as described above. The sample preparation was similar to the process as explained above for the Ni2Ge/Ge/Ni2Ge nanowire heterostructure. The only difference is that before the annealing process, an Al2O3 film was deposited by ALD at 250°C to cap the device and confine the solid state reaction of germanidation during annealing. The Al2O3-coated devices were then annealed both in situ (inside TEM) and ex situ (in RTA). Figure 7(a) shows the low-magnification TEM image of the Ni-Ge nanowire device capped with 10 nm Al2O3 after annealing at 450°C for 30 s. The enlarged TEM image in Figure 7(b) shows that there are clearly two interfaces in the NixGe/Ge nanowire heterostructure. Figure 7(c) shows the EDS line-scan profile from the nanowire heterostructure. The line profile indicates two germanide phases in the formed NixGe region, which corresponds to the two interfaces observed in the NixGe/Ge heterostructure. The Ni/Ge ratio is about 1 : 1 in the small germanide region close to the NixGe/Ge interface, suggesting the formation of NiGe. This result is consistent with the line-scan profile in Figures 6(d) and 6(e). The length of the NiGe region can range from tens of nanometers to hundreds of nanometers in our experiments. On the other hand, the Ni/Ge ratio is about 2 : 1 in the other germanide region close to the Ni pad on the left, implying that the phase is Ni2Ge. It is also worth noting that the almost constant concentration of Ge along the heterostructure suggests Ni is the dominant diffusion species in this system . Figure 7(d) shows the lattice-resolved HRTEM image of the NixGe/Ge heterostructure, clearly exhibiting two interfaces. The FFT patterns at the Ni2Ge, NiGe, and Ge regions are shown in Figures 7(e)–7(g), which help further confirm the germanide phases. The crystallographic epitaxial relationships between the Ge/NiGe interface were determined to be and Ge(1)//NiGe(001), while those for the Ni2Ge/NiGe interface were Ni2Ge//NiGe and Ni2Ge(011)//NiGe(001). It is worth noting that a large lattice mismatch of 77.7% observed at the NiGe/Ge epitaxial interface could result in the segregation of nanoparticles (see Figure 6). In contrast, the epitaxial relationships in the Ni2Ge/Ge/Ni2Ge nanowire heterostructure formed without oxide confinement during annealing were found to be Ge//Ni2Ge and Ge(1)//Ni2Ge(100), as we discussed . Therefore, the Al2O3 capping layer plays an important role in confining the growth of germanides and also promoting the formation of NiGe to maintain high-quality epitaxial relationships between Ni2Ge and Ge.
Figure 8 schematically illustrates the epitaxial relationships of the Ni2Ge/Ge interface in the Ni2Ge/Ge/Ni2Ge nanowire heterostructure and the NiGe/Ge interface in the Ni2Ge/NiGe/Ge/NiGe/Ni2Ge nanowire heterostructure. According to the epitaxial relationships in both cases, the nanowire growth direction (along the Ge  direction) is not perpendicular to the epitaxial planes (parallel to the Ge (1) plane). This “twisted” growth mode of nanowires is substantially different from that in the typical epitaxial growth of thin films, in which the growth direction is usually perpendicular to the epitaxial planes . The presence of the oxide capping may alter the energy of the growth and thus change the twisted angle. Furthermore, it suggests that the twisting in nanowires may be used to accommodate substantially large lattice mismatches. This unique growth mode may be attributed to the fact in minimizing the total system energy in the presence of a large lattice mismatch in the interface. Further microscopic studies in simulation and experiment are required to understand the growth kinetics for this unique growth mode in one-dimensional systems.
The formed atomically sharp interface in the NixGe/Ge/NixGe nanowire heterostructures (both with and without Al2O3 capping during annealing) can be used to explore promising applications in nanoscale devices [6, 7, 10, 16]. As aforementioned, this NixGe/Ge/NixGe nanowire heterostructure can be easily used to fabricate Ge nanowire FETs, in which the germanide regions NixGe can be used as source/drain contacts to the Ge nanowire channel. The channel length can be simply controlled by the annealing time and growth length of the germanide nanowire; therefore, it can be well controlled to sub-100 nm using a convenient RTA process. Clearly, this simple process has a great advantage over traditional complex and expensive photolithography technology in achieving short-channel transistors.
Back-gate FETs were fabricated on the SiO2/Si substrate to study the electrical transport property of the Ni2Ge/Ge/Ni2Ge nanowire heterostructure. The Si substrate is degenerately doped to serve as a back gate. The device schematic is shown in Figure 9(a), and electrical measurements were performed using a probe station with a Keithley 4200 semiconductor parameter analyzer. For comparison, Figure 9(b) shows the logarithm plot of typical curves for the Ge nanowire FET at various drain voltages before and after RTA. They both show a p-type transistor behavior, although we intended to grow undoped Ge nanowires. This is mainly due to the Fermi level pinning at the Ge nanowire surface induced by Ge surface states, which tends to result in hole accumulation [26, 27]. The maximum current measured before RTA at V is about 30 nA, corresponding to a current density of A/cm2. The current density is relatively small due to a large Schottky barrier at the source/drain contacts before annealing. To extract the mobility, we first used the cylinder-on-plate model to estimate the gate capacitance coupling between the Ge nanowire and the back-gate oxide as (see ) where F/cm is the vacuum dielectric constant, is the relative dielectric constant for SiO2, and nm is the radius of the Ge nanowire. The Ge nanowire channel bis μm, and the thickness of the back-gate dielectric is nm. Given the above parameters, the estimated gate capacitance is F. The field effect hole mobility can be extracted from the curves using the transconductance () at a fixed drain bias
Using the maximum transconductance extracted from the curves, the hole mobility obtained falls in the range of 3–8 cm2/Vs. This is consistent with previous reported values (less than 10 cm2/Vs) for SFLS-synthesized Ge nanowires .
After RTA at 400°C for 15 s, however, electrical transport measurements on the Ge nanowire device show much improved transistor characteristics, as shown in Figure 9(b). The gate bias was scanned from 0 V to −40 V, the latter of which corresponds to a maximum vertical gate electrical field of V/cm. The curves show a p-type behavior with an on/off ratio larger than 103. The maximum current measured at V is about 0.7 μA corresponding to a current density of A/cm2, which was more than 20 times larger after annealing, in which the Ni2Ge contact to the Ge nanowire channel was developed. The maximum transconductance extracted from curves at drain bias V is 13.3 nS, giving rise to a field-effect hole mobility of 65.2 cm2/Vs. Although this mobility is still lower than the reported value from VLS-grown Ge nanowires [14, 16], it still shows about one order of magnitude improvement among SFLS-synthesized Ge nanowires , and this increase may be attributed to the atomically sharp contact of Ni2Ge to the Ge nanowire.
Similarly, the Ni2Ge/NiGe/Ge/NiGe/Ni2Ge nanowire heterostructures formed with Al2O3 capping during annealing can also be used to fabricate Ge nanowire FETs. To further improve the transistor performance, VLS-grown nanowires were used in this study because of their higher carrier mobility. Back-gate nanowire FETs were fabricated on the SiO2/Si substrate similarly as shown in Figure 10(a). Figure 10(b) shows the typical curves of a back-gate Ge nanowire FET after RTA at 450°C for 20 s, in which the on/off ratio is as high as 105. The maximum transconductance is obtained to be about 0.168 μS at V, which gives rise to a normalized transconductance of 2.4 μS/μm, assuming the effective channel length is equal to the nanowire diameter (70 nm) . The extracted hole mobility in our experiments is typically in the range of 150–210 cm2/Vs. Compared with SFLS-synthesized Ge nanowires, the significant improvement in the transistor performance here could be attributed to a better crystalline quality of VLS-grown Ge nanowires. Figure 10(c) shows the typical curves of a back-gate Ge nanowire FET at various gate voltages.
As explained above, the Al2O3 capping layer helps confine the growth of germanide in a Ge nanowire during thermal annealing. More importantly, it is of interest to study the effect of the Al2O3 capping layer on the device performance. In the previous study on the metal contacts to Ge substrate through a thin layer of Al2O3 tunneling oxide by Zhou et al., it is found that Al2O3 helps terminate the dangling bonds and passivate the Ge surface, therefore, alleviate Fermi level pinning effect . In this study, dual sweepings of gate bias in the curves were performed to investigate the charge trapping due to the Al2O3 capping layer on passivating the Ge nanowire surface. Figure 10(d) shows the curves under various conditions: before Al2O3 deposition (both in air and in vacuum) and after Al2O3 deposition at 250°C. The gate bias was swept from +40 V to −40 V then back to +40 V in steps of 0.5 V at a fixed drain bias of mV. The curve measured in air before annealing shows the biggest hysteresis, which is mainly due to the absorption of molecules from the ambient and the charge trapping on the Ge surface [26, 33]. The measured reduced hysteresis in a vacuum (less than 10−5 Torr), however, rules out the contribution from the ambient. Also, measurements in the vacuum help reduce scatterings from the molecules absorbed on the Ge surface as carriers transport along the Ge nanowire channel, which will reduce the resistance of Ge nanowires. Furthermore, the hysteresis was significantly reduced after the Al2O3 deposition, which unambiguously demonstrates the passivation effect of the Al2O3 layer on the Ge nanowire surface . Besides, the Al2O3 passivation could again reduce scatterings for carriers transport along the Ge nanowire channel and thus further reduce the resistance of Ge nanowires. The small hysteresis present after Al2O3 passivation, however, may arise from the charge trapping on the Ge surface between the Ge nanowire channel and the back-gate dielectric, the region that is not covered by the Al2O3 capping layer.
3. Conclusion and Discussion for Future Study
In summary, Ge nanowire heterostructures with atomically sharp interfaces have been demonstrated by the solid-state reaction between a single-crystalline Ge nanowire and Ni contact pads via a facile rapid thermal annealing process in a temperature range of 400–500°C. The crystallographic epitaxial relationships in the formation of the Ni2Ge/Ge/Ni2Ge nanowire heterostructure were determined to be Ge//Ni2Ge and Ge(1)//Ni2Ge(100). Back-gate FETs were fabricated using the formed Ni2Ge region as source/drain contacts to the Ge nanowire channel. The electrical measurement shows an on/off ratio over 103 and a field-effect hole mobility of about 65.4 cm2/Vs, which are superior to reported values for SFLS-synthesized Ge nanowires.
More importantly, the effect of oxide confinement on the formation of the Ge nanowire heterostructure was studied by capping the Ge nanowire device with a layer of Al2O3 before the annealing process. In contrast to the single germanide phase in the Ni2Ge/Ge/Ni2Ge nanowire heterostructure, a segment of high-quality epitaxial NiGe was then formed between Ni2Ge and Ge. The crystallographic epitaxial relationships were determined to be Ge//NiGe and Ge(1)//NiGe(001) for the Ge/NiGe interface and Ni2Ge//NiGe and Ni2Ge(011)//NiGe(001) for the Ni2Ge/NiGe interface, respectively. Similarly, back-gate FETs were fabricated on the formed Ni2Ge/NiGe/Ge/NiGe/Ni2Ge nanowire heterostructure, and the electrical measurements reveal a high-performance p-type behavior, showing a high on/off ratio of 105 and a field-effect hole mobility of about 210 cm2/Vs. Moreover, the Al2O3 capping was found to passivate the Ge nanowire surface as well as to provide an appreciable confinement during the growth of germanide and changes its composition to maintain the high-quality epitaxial relationships.
For further understanding the growth kinetics of Ge nanowire heterostructures, a more detailed study is required [8, 9, 34–36]. For instance, the size of the metal diffusion (contact) source may affect the formation of Ge nanowire heterostructure, especially the segregation of nanoparticles. In this paper, the metal source was defined with EBL and e-beam evaporation, in which the size of planar Ni contacts (typically in the μm range) is much larger than the diameter of Ge nanowires (typically below 100 nm) . On the other hand, in some other studies of Si nanowire heterostructures, another extensively-studied diffusion source was metal nanowires or dots (refer to Table 1), which led to a point-contact rather than a planar contact to the Si nanowire [8, 9, 37, 38]. Similarly, we have also studied the formation of Ge nanowire heterostructures through the point contact reaction between a Ge nanowire and a Ni nanowire. Figure 11(a) shows the TEM image of a Ge nanowire in point contact with two Ni nanowires on the TEM grid at room temperature. The Ni nanowires were prepared by the electrochemical deposition, and the diameter was typically 50–100 nm which depends on the size of the anodic aluminum oxide (AAO) template. Figure 11(b) shows the TEM image, similarly illustrating the diffusion of Ni into the Ge nanowire upon 450°C annealing. The red arrows indicate the growth tip of the NixGe nanowire. It is worth noting that the growth of the NixGe nanowire started from the point-contact region, similar to the planar contact case . However, this is different from the case of the NiSi/Si/NiSi nanowire heterostructure formed via the point-contact reaction between a Si nanowire and a Ni nanowire, in which the growth of NiSi actually started from both ends of the Si nanowire instead of the point-contact region . Further kinetics study is needed to understand the different phenomena in Ge and Si nanowire heterostructures. Figure 11(c) TEM image of another Ge nanowire in point-contact with a Ni nanowire at room temperature. Figure 11(d) shows the TEM image illustrating the diffusion of Ni into the Ge nanowire upon 450°C annealing. Clear consumption of Ni in the formation of the NixGe nanowire upon annealing was observed, as highlighted by the red circle. Moreover, it is surprising that the segregation of NixGe nanoparticles was not observed in the Ni-Ge nanowire point-contact system upon annealing . We believe that the much smaller Ni flux through such point contact compared with the planar contact may play an important role in preventing the segregation of nanoparticles. Therefore, the difference of various metal contacts in the formation of Ge nanowire heterostructures requires further study.
For the device applications, we have demonstrated in this study that the formed germanides with atomically sharp interface to Ge can be used as a good electrical contact. Indeed, Lin et al.  have successfully detected the electrical spin injection into a Si nanowire from the ferromagnetic MnSi contact in the formed MnSi/Si/MnSi nanowire heterostructure. However, the signal can be observed only at relatively low temperature, since MnSi has a low Curie temperature of about 30 K. On the other hand, there are many germanides, such as Mn5Ge3 and Ni3Ge, which exhibit room-temperature ferromagnetism [20, 21], and thus offer great advantages over silicides for promising applications in room-temperature spintronics, including spin injection, transport and detection.
In fact, a room-temperature ferromagnetic germanide phase, Ni3Ge , was developed at high reaction temperature, in which a high-concentration Ni vapor from a large-area Ni contact pattern surrounded the Ge NWs to form a fragmented Ni3Ge nanowire. Figure 12(a) shows the TEM image of the formed Ni3Ge/Ge/Ni3Ge nanowire heterostructure on the TEM grid upon 650°C annealing. Figure 12(b) illustrates the HRTEM image of the Ge nanowire heterostructure with a clean and sharp interface between Ni3Ge and Ge. The Ge region was controlled down to as small as 12 nm. The strained short Ge region in the Ni3Ge/Ge/Ni3Ge nanowire heterostructure is promising for high-performance FETs and spintronics applications [11, 39]. Figure 12(c) shows the lattice-resolved TEM image of the Ni3Ge-Ge interface, and the formed germanide was identified to be single-crystalline Ni3Ge with a face-centered cubic (FCC) lattice structure (Fd3m, space group 227 and JCPDS no. 65-7680) and a lattice constant of nm. Although a slight volume expansion was still observed, the Ni3Ge lattice was well fit with the Ge lattice, due to their same lattice structure and a very small lattice mismatch of only 1.5% (the lattice constant of Ge is nm). It is worth mentioning that the twisted growth mode, which was observed in both Ni2Ge/Ge/Ni2Ge and Ni2Ge/NiGe/Ge/NiGe/Ni2Ge nanowire heterostructures to accommodate the large lattice mismatch, did not occur in this Ni3Ge/Ge/Ni3Ge nanowire heterostructure due to such a small lattice mismatch. However, as mentioned above, the melting point of Ge nanowires is significantly reduced from that of bulk Ge . As a result, Ge nanowires were easily broken at high temperature, as shown in Figure 12(d). Therefore, a ferromagnetic germanide remains undeveloped at a relatively low temperature in order to study the spin transport in Ge nanowire.
This work was in part supported by Western Institution of Nanoelectronics (WIN) and Focus Center on Functional Engineered Nano Architectonics (FENA). The authors also acknowledged the support from National Science Council through Grant no. NSC 98-2221-E-007-104-MY3. J. Tang and C. Y. Wang contributed equally to the paper.
- D. Wang, Q. Wang, A. Javey et al., “Germanium nanowire field-effect transistors with SiO2 and high-κ HfO2 gate dielectrics,” Applied Physics Letters, vol. 83, no. 12, pp. 2432–2434, 2003.
- M. I. Van Der Meulen, N. Petkov, M. A. Morris et al., “Single crystalline Ge1-x Mnx nanowires as building blocks for nanoelectronics,” Nano Letters, vol. 9, no. 1, pp. 50–56, 2009.
- Y. Huang, X. Duan, Y. Cui, L. J. Lauhon, K. H. Kim, and C. M. Lieber, “Logic gates and computation from assembled nanowire building blocks,” Science, vol. 294, no. 5545, pp. 1313–1317, 2001.
- C. Thelander, H. A. Nilsson, L. E. Jensen, and L. Samuelson, “Nanowire single-electron memory,” Nano Letters, vol. 5, no. 4, pp. 635–638, 2005.
- Y. Cui, Q. Wei, H. Park, and C. M. Lieber, “Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species,” Science, vol. 293, no. 5533, pp. 1289–1292, 2001.
- Y. Wu, J. Xiang, C. Yang, W. Lu, and C. M. Lieber, “Single-crystal metallic nanowires and metal/semiconductor nanowire heterostructures,” Nature, vol. 430, no. 6995, pp. 61–65, 2004.
- W. M. Weber, L. Geelhaar, A. P. Graham et al., “Silicon-nanowire transistors with intruded nickel-suicide contacts,” Nano Letters, vol. 6, no. 12, pp. 2660–2666, 2006.
- K. C. Lu, W. W. Wu, H. W. Wu et al., “In situ control of atomic-scale Si layer with huge strain in the nanoheterostructure NiSi/Si/NiSi through point contact reaction,” Nano Letters, vol. 7, no. 8, pp. 2389–2394, 2007.
- Y. C. Chou, W. W. Wu, S. L. Cheng et al., “In-situ TEM Observation of repeating events of nucleation in epitaxial growth of nano CoSi2 in nanowires of Si,” Nano Letters, vol. 8, no. 8, pp. 2194–2199, 2008.
- Y. C. Lin, K. C. Lu, W. W. Wu et al., “Single crystalline PtSi nanowires, PtSi/Si/PtSi nanowire heterostructures, and nanodevices,” Nano Letters, vol. 8, no. 3, pp. 913–918, 2008.
- Y. C. Lin, Y. Chen, A. Shailos, and Y. Huang, “Detection of spin polarized carrier in silicon nanowire with single crystal MnSi as magnetic contacts,” Nano Letters, vol. 10, no. 6, pp. 2281–2287, 2010.
- Y. Zhou, M. Ogawa, X. Han, and K. L. Wang, “Alleviation of Fermi-level pinning effect on metal/germanium interface by insertion of an ultrathin aluminum oxide,” Applied Physics Letters, vol. 93, no. 20, Article ID 202105, 2008.
- J. Tang, C. Y. Wang, F. Xiu et al., “Single-crystalline Ni2Ge/Ge/Ni2Ge nanowire heterostructure transistors,” Nanotechnology, vol. 21, no. 50, Article ID 505704, 2010.
- J. Tang, C. Y. Wang, F. Xiu, et al., “Oxide-confined formation of Ge nanowire heterostructure for high-performance transistors,” ACS Nano, vol. 5, no. 7, pp. 6008–6015, 2011.
- N. S. Dellas, S. Minassian, J. M. Redwing, and S. E. Mohney, “Formation of nickel germanide contacts to Ge nanowires,” Applied Physics Letters, vol. 97, no. 26, Article ID 263116, 2010.
- T. Burchhart, A. Lugstein, Y. J. Hyun, G. Hochleitner, and E. Bertagnolli, “Atomic scale alignment of copper-germanide contacts for ge nanowire metal oxide field effect transistors,” Nano Letters, vol. 9, no. 11, pp. 3739–3742, 2009.
- T. Burchhart, C. Zeiner, Y. J. Hyun, A. Lugstein, G. Hochleitner, and E. Bertagnolli, “High performance Ω-gated Ge nanowire MOSFET with quasi-metallic source/drain contacts,” Nanotechnology, vol. 21, no. 43, Article ID 435704, 2010.
- T. Burchhart, A. Lugstein, C. Zeiner, Y. J. Hyun, G. Hochleitner, and E. Bertagnolli, “Nanowire-metal heterostructures for high performance MOSFETs,” Elektrotechnik und Informationstechnik, vol. 127, no. 6, pp. 171–175, 2010.
- K. Yamane, K. Hamaya, Y. Ando et al., “Effect of atomically controlled interfaces on Fermi-level pinning at metal/Ge interfaces,” Applied Physics Letters, vol. 96, no. 16, Article ID 162104, 2010.
- T. Izumi, M. Taniguchi, S. Kumai, and A. Sato, “Ferromagnetic properties of cyclically deformed Fe3Ge and Ni3Ge,” Philosophical Magazine, vol. 84, no. 36, pp. 3883–3895, 2004.
- M. Jamet, A. Barski, T. Devillers et al., “High-curie-temperature ferromagnetism in self-organized Ge1-xMnx nanocolumns,” Nature Materials, vol. 5, no. 8, pp. 653–659, 2006.
- H. Y. Tuan and B. A. Korgel, “Importance of solvent-mediated phenylsilane decompositon kinetics for high-yield solution-phase silicon nanowire synthesis,” Chemistry of Materials, vol. 20, no. 4, pp. 1239–1241, 2008.
- T. Hanrath and B. A. Korgel, “Supercritical fluid-liquid-solid (SFLS) synthesis of Si and Ge nanowires seeded by colloidal metal nanocrystals,” Advanced Materials, vol. 15, no. 5, pp. 437–440, 2003.
- Y. L. Chueh, C. N. Boswell, C. W. Yuan et al., “Nanoscale structural engineering via phase segregation: Au-Ge system,” Nano Letters, vol. 10, no. 2, pp. 393–397, 2010.
- A. D. Schricker, S. V. Joshi, T. Hanrath, S. K. Banerjee, and B. A. Korgel, “Temperature dependence of the field effect mobility of solution-grown germanium nanowires,” Journal of Physical Chemistry B, vol. 110, no. 13, pp. 6816–6823, 2006.
- T. Hanrath and B. A. Korgel, “Influence of surface states on electron transport through intrinsic Ge nanowires,” Journal of Physical Chemistry B, vol. 109, no. 12, pp. 5518–5524, 2005.
- S. Zhang, E. R. Hemesath, D. E. Perea, E. Wijaya, J. L. Lensch-Falk, and L. J. Lauhon, “Relative influence of surface states and bulk impurities on the electrical properties of Ge nanowires,” Nano Letters, vol. 9, no. 9, pp. 3268–3274, 2009.
- Y. F. Hsieh, L. J. Chen, E. D. Marshall, and S. S. Lau, “Partial epitaxial growth of Ni2Ge and NiGe on Ge(1 1 1),” Thin Solid Films, vol. 162, pp. 287–294, 1988.
- Y. Wu and P. Yang, “Melting and welding semiconductor nanowires in nanotubes,” Advanced Materials, vol. 13, no. 7, pp. 520–523, 2001.
- Y. C. Lin, Y. Chen, D. Xu, and Y. Huang, “Growth of nickel silicides in Si and Si/SiOx core/shell nanowires,” Nano Letters, vol. 10, no. 11, pp. 4721–4726, 2010.
- Y. Zhou, W. Han, Y. Wang et al., “Investigating the origin of Fermi level pinning in Ge Schottky junctions using epitaxially grown ultrathin MgO films,” Applied Physics Letters, vol. 96, no. 10, Article ID 102103, 2010.
- S. Ramo, J. R. Whinnery, and T. V. Duzer, Fields and Waves in Communication Electronics, John Wiley & Sons, New York, NY, USA, 3rd edition, 1994.
- D. Wang, Y. L. Chang, Q. Wang et al., “Surface chemistry and electrical properties of germanium nanowires,” Journal of the American Chemical Society, vol. 126, no. 37, pp. 11602–11611, 2004.
- Y. C. Chou, W. W. Wu, L. J. Chen, and K. N. Tu, “Homogeneous nucleation of epitaxial CoSi2 and NiSi in Si nanowires,” Nano Letters, vol. 9, no. 6, pp. 2337–2342, 2009.
- L. J. Chen and W. W. Wu, “In situ TEM investigation of dynamical changes of nanostructures,” Materials Science and Engineering R, vol. 70, no. 3–6, pp. 303–319, 2010.
- Y. C. Chou, W. W. Wu, C. Y. Lee, C. Y. Liu, L. J. Chen, and K. N. Tu, “Heterogeneous and homogeneous nucleation of epitaxial NiSi2 in [1 1 0] Si nanowires,” Journal of Physical Chemistry C, vol. 115, no. 2, pp. 397–401, 2011.
- K. C. Lu, K. N. Tu, W. W. Wu, L. J. Chen, B. Y. Yoo, and N. V. Myung, “Point contact reactions between Ni and Si nanowires and reactive epitaxial growth of axial nano-NiSi/Si,” Applied Physics Letters, vol. 90, no. 25, Article ID 253111, 2007.
- W. W. Wu, K. C. Lu, C. W. Wang et al., “Growth of multiple metal/semiconductor nanoheterostructures through point and line contact reactions,” Nano Letters, vol. 10, no. 10, pp. 3984–3989, 2010.
- W. W. Wu, K. C. Lu, K. N. Chen et al., “Controlled large strain of Ni silicide/Si/Ni silicide nanowire heterostructures and their electron transport properties,” Applied Physics Letters, vol. 97, no. 20, Article ID 203110, 2010.