IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 18 (2007) 105307 (5pp)
doi:10.1088/0957-4484/18/10/105307
Sulfur passivation for ohmic contact formation to InAs nanowires D B Suyatin1,2 , C Thelander1 , M T Bj¨ork3 , I Maximov1 and L Samuelson1 1 Solid State Physics/The Nanometer Structure Consortium, Lund University, Box 118, S-221 00, Lund, Sweden 2 Division of Microelectronics, Institute of Nuclear Physics, Moscow State University, Leninskie Gory, GSP-2, 119992, Moscow, Russia 3 IBM Research GmbH, Z¨urich Research Laboratory, S¨aumerstrasse 4, 8803, R¨uschlikon, Switzerland
E-mail:
[email protected]
Received 1 December 2006 Published 6 February 2007 Online at stacks.iop.org/Nano/18/105307 Abstract We have studied the formation of ohmic contacts to InAs nanowires by chemical etching and passivation of the contact areas in an ammonium polysulfide, (NH4 )2 Sx , water solution. The nanowires were exposed to different dilution levels of the (NH4 )2 Sx solution before contact metal evaporation. A process based on a highly diluted (NH4 )2 Sx solution was found to be self-terminating, with minimal etching of the InAs. The stability of the contacts was investigated with electrical measurements as a function of storage time in vacuum and air. (Some figures in this article are in colour only in the electronic version)
considerable reoxidation in air and aqueous solutions [12, 13]. It also provides both chemical and electrical passivation of a surface. Sulfur-adsorbed III–V compound semiconductor surfaces have been studied extensively by various surfacesensitive techniques and noticeable improvement of various device-related properties was observed [12, 13]. Surface passivation becomes progressively more important for a device as the surface-to-volume ratio of the structure increases. In spite of this, only a limited number of studies on passivation of semiconductor nanowire structures have been reported so far [8–11]. The key difference in chalcogenide passivation technology for nanostructures, compared to bulk materials, is the importance of etching rates [7]. In bulk, a certain amount of surface material can be sacrificed without the loss of device functionality. However, semiconductor nanostructures in general, and nanowires in particular, are extremely sensitive to removal of surface material due to their large surface-tovolume ratio. The aim of this study was to find optimum process conditions which provide ohmic contact formation to InAs nanowires together with minimal semiconductor material removal. We describe a method for passivation of InAs nanowires using both highly diluted and regular (NH4 )2 Sx
1. Introduction In the past few years, considerable progress in the control of semiconductor nanowire formation has been achieved, and already, a number of different nanowire-based applications have been demonstrated [1–4]. Nanowires of different materials are not only of interest as potential building blocks for future electronics and photonics, but they also let researchers access low-dimensional systems with interesting properties [5, 6]. A key to electrical transport studies on such nanometre-scale objects is the ability to form reliable contacts to individual nanowires. The surface oxides must be removed, and the semiconductor contact areas should ideally be passivated before contact metal deposition to prevent reoxidation. Semiconductor surface passivation was actively pursued in the 1960s in the development of III–V based high speed transistors. Since the mid 1980s, research on chalcogenide passivation for bulk III–V semiconductors has attracted a lot of attention. In particular, ammonium polysulfide, (NH4 )2 Sx , solutions have been found to remove native oxides and contaminants from III–V semiconductor surfaces, and to provide passivation with covalently bonded sulfur atoms. The passivation allows good short-term surface stability without 0957-4484/07/105307+05$30.00
1
© 2007 IOP Publishing Ltd Printed in the UK
Nanotechnology 18 (2007) 105307
D B Suyatin et al
water solutions. The treatment in highly diluted (NH4 )2 Sx solution was shown to be a self-terminating passivation process. An optimized treatment in such a solution leads to the formation of reliable ohmic contacts with no considerable etching effect. This treatment could also be adopted for other III–V materials, because the passivation mechanism of the (NH4 )2 Sx solution is reported to be the same for all III– V semiconductors [12]. The stability of the contacts was investigated with electrical measurements as a function of storage time in vacuum and air.
(a)
(b)
104
Untreated 7 min. 1.5M 12 min. 1.5M 17 min. 1.5M
103 102
(c)
pH
10.2
101 Original solution
100
9.8 10 M
2. Experimental details
9.4 0.1 1 10 100 Concentration (%)
In order to study and optimize the formation of contacts to InAs nanowires, a set of wires 50–220 nm in diameter and 3–4 μm in length were grown from Au aerosol particles deposited on InAs(111)B substrates. The chemical beam epitaxy (CBE) technique produces crystalline wurtzite InAs wires without tapering [14–16]. All measurements reported here were performed on nanowires grown in one growth run. After growth, the nanowires were mechanically transferred onto silicon substrates covered by a thermally grown 100 nm thick silicon dioxide top layer, on which reference markers and macroscopic metal pads were predefined [14]. The positions of individual wires were determined using an optical microscope in order to avoid potential contamination deposited when imaging with a scanning electron microscope (SEM). Nanowire sections with lengths greater than 1.5 μm were selected for the experiments. The samples were spin-coated with an e-beam sensitive polymethylmethacrylate (PMMA) resist. Following this, electron beam lithography was used to define contacts connecting individual wires to the macroscopic contact pads [14]. Previous measurements on InAs nanowires [6, 14] have shown that ohmic contacts do not function as non-invasive probes in a 4-probe configuration. No difference between 2-probe and 4-probe results was observed. For this reason, a 2-probe configuration was used in this investigation, with 300 nm wide contacts separated by 1 μm, as shown in figure 1(a). Electrical contacts were established to measure 10– 20 individual nanowires in each batch. Prior to the evaporation of metals, the samples were treated by oxygen plasma to remove PMMA contaminants from the contact region (the treatment was done for 12 s at 5 mbar oxygen pressure in a Plasma-Preen System II 862 from Plasmatic System Inc.). Following that, the samples were treated in a heated (NH4 )2 Sx water solution to remove native oxides from the nanowire surface and to passivate it with sulfur atoms. After the passivation, the samples were rinsed in an excess of deionized (DI) water for 30 s and then mounted in the metal evaporation ˚ Ni and then chamber. After thermal evaporation of 250 A ˚ Au, the sample processing was finalized by a lift-off 900 A process. No contact heat treatment (annealing) was performed. Current–voltage measurements with applied bias in the range ±100 mV were carried out at room temperature (RT) to characterize the contact quality. An SEM was used to inspect the etching effect of the passivation treatment. The SEM resolution was 1.5 nm. The nanowire etching effect was found to be nonuniform; therefore we performed only a qualitative characterization of the latter.
10-1
50
100
Diameter (nm) Diame
Figure 1. (a) Scanning electron micrograph of a CBE grown InAs nanowire with electrical contacts. An electrode separation of 1 μm and electrode width of 300 nm were used for all measurements presented here. The scale bar is 1 μm. (b) Measured two-point resistance as a function of diameter for different passivation times using a non-self-terminating passivation process (solution A). The measurements were done at room temperature in helium gas. The dotted curve marks the limit above which nanowires are observed to be non-ohmic. The solid curve shows the calculated theoretical lower limit of resistance for fully ballistic nanowires at room temperature. See text for more details. (c) pH value as function of (NH4 )2 Sx concentration in DI water. The lower curve shows pH values for solutions obtained from the stock solution with 10 M of additional sulfur; the upper curve displays pH values for solutions obtained from the stock solution without additional sulfur.
Prior to an electrical characterization of the contacts quality, etching tests on free-laying nanowires were performed: InAs nanowires were randomly deposited on a SiO2 surface and subsequently treated in ammonium polysulfide solution. The etching effect of the solution was checked by measurements of nanowire diameter in an SEM before and after treatment in the (NH4 )2 Sx solution for different times. The accuracy of the SEM measurements was defined to be about 5 nm.
3. Results and discussion A very high device resistance was obtained if no treatment of the nanowire surface was carried out before metal evaporation. This is illustrated in figure 1(b). Roughly 50% of the wires showed no conductance at all (R > 100 G) for the bias voltages used; these are not included in figure 1(b). It is known [12] that an InAs surface that has been exposed to oxygen is normally covered by a native oxide layer. Most probably, the oxide layer prevents the formation of ohmic contacts to unpassivated nanowires. The basic chemical for our experiments was a 20% ammonium sulfide, (NH4 )2 S, solution (stock solution) obtained from VWR International. Elemental sulfur was added to the stock solution to form an ammonium polysulfide, (NH4 )2 Sx , solution. In the rest of the text, the elemental sulfur concentration is presented in respect to the stock solution. The solution concentration in DI water is presented in respect to the mixture of the stock solution with additional elemental sulfur. 2
Nanotechnology 18 (2007) 105307 (a)
D B Suyatin et al
the effective nanowire diameter locally decreases. The curves in figure 1(b) illustrate how strongly the nanowire resistance depends on nanowire diameter. Once the diameter of an InAs nanowire is less than 35–40 nm, we observe a strong increase in resistivity with reduced cross-section area. In this case, we believe that the radial confinement becomes strong enough to deplete the wire from carriers, leading to a non-ohmic nanowire device behaviour. Figure 2(b) shows the result of InAs nanowire treatment with solution A for 17 min. In this case, the etching effect was negligibly small, below the measurement accuracy. Solution A can thus provide a high yield of ohmic contacts (see figure 1(b)). Nevertheless, it was observed that occasional variations in the etching effect can lead to substantial contact degradation. An ideal surface treatment process for the formation of ohmic electrical contacts to nanowires should, however, first of all provide efficient removal of the surface oxides together with a minimum etching effect, and secondly ensure the formation of a continuous passivation layer. In general, the etching rate can be controlled by adjusting the pH value of the solution; with lower pH values yielding lower etch rates. However, at too low pH values, effective removal of the oxides cannot take place [12]. The pH value of the solution depends on the dilution level and sulfur concentration. Figure 1(c) shows the pH values for the original solution and the solution with 10 M sulfur as a function of concentration. The figure demonstrates that an increase of sulfur content and decrease of ammonium polysulfide concentration in DI water leads to a decrease in pH. A higher concentration of sulfur ions provides a better supply of sulfur atoms for the surface passivation process. Therefore, a high concentration of sulfur ions together with a high dilution level should be favourable, both for the passivation process and to prevent etching of the semiconductor. However, as the dilution level increases, the highest possible sulfur molar fraction decreases for a stable solution. In a modified process, 3 M of additional elemental sulfur was added to the stock solution. This solution is, from here on, referred to as solution B. As the passivation treatment is a photoelectrochemical process [12], it should also be improved by increased illumination and temperature. We therefore used white light illumination during treatment with solution B (75 W incandescent light bulb). An elevated temperature also increases the reaction rate as described by Arrhenius law and decreases the pH value, as can be seen from the inset in figure 3. The temperature range for the treatment is limited, however, because thermal decomposition of ammonium sulfide solution starts at around 40 ◦ C (Technical Data Sheet, Merck KGaA, Darmstadt). High temperatures therefore lead to fast degradation of the solution. For solution B, we limited the treatment temperature to 62 ◦ C. This value is close to a highest value (65 ◦ C) reported for the treatment in an ammonium sulfide solution [18]. Sulfur passivated surfaces are known to have limited stability in air [12]. Before metal evaporation, samples are usually exposed to air for some time (such as during sample transportation and loading). In our experiments, passivated samples were transported in an opaque ceramic cup of DI water, and the sample handling before metal evaporation was performed at decreased illumination.
(b)
Figure 2. SEM micrographs of nanowire contact regions (a) with and (b) without a strong etching effect. The white arrows in (a) point towards partly etched nanowire. The nanowire was completely etched away in the contact region. The parts of the nanowire protected by PMMA were etched only partly due to (NH4 )2 Sx solution penetration under the resist. The black arrow in (b) shows the result of (NH4 )2 Sx solution treatment which did not cause a substantial etching effect. See text for more details. The scale bars are 200 nm for both figures.
The stock solution is impractical for contact processing, because of an extremely high etching rate. Diluting the stock solution ten times in DI water gives a controlled etch rate. Furthermore, in the text, a 10% mix of the ammonium polysulfide solution in DI water with 1.5 M of elemental sulfur is referred to as solution A. Surface treatment of nanowires with solution A at 40 ◦ C was carried out first. In figure 1(b), the room temperature two-terminal nanowire resistance is plotted as a function of wire diameter for three different treatment times, 7, 12, and 17 min respectively, together with the result for untreated contacts. The solid curve in figure 1(b) is a calculation of the lower limit of resistance for a fully ballistic wire of a certain diameter with a Fermi energy of 130 meV at room temperature. The Fermi energy relates to the bottom of the conduction band of the bulk material. The Fermi energy value was obtained from [17] for bulk InAs surfaces exposed to air. The calculation shows a theoretical minimum for the nanowire resistances. The dotted curve indicates the limit above which nanowires are observed to be non-ohmic. The curve was calculated with the assumption of diffusive electron transport by formula: R(A) = ρ Al , where R is the nanowire resistance, ρ is a measure of the nanowire resistivity, l is the nanowire √ length, always 10−6 m in our experiments and A = 827 d 2 is the hexagonal cross-section area of the nanowire, where d is the nanowire diameter4. A resistivity of 10−2 cm has been used to calculate the dotted curve; below this curve the nanowires are usually observed to have ohmic behaviour and the contact resistance is negligible compared to the total device resistance [6, 14]. Figure 1(b) shows that the yield of ohmic nanowire devices increases with increased treatment time. However, treatment times longer than 17 min result in substantial nanowire etching and thus overall device degradation. Figure 2 illustrates the etching effect on the diameter of the InAs nanowires. The SEM micrograph in figure 2(a) shows a nanowire with a diameter of 70 nm, which was completely etched away in the contact region. However, even a small etching effect will also lead to device degradation because 4
The nanowires have hexagonal cross-section [14]. The nanowire width measured by SEM is referred to in the text as the nanowire diameter. The width was always measured for nanowires lying on a surface.
3
Nanotechnology 18 (2007) 105307
D B Suyatin et al
25 C o 62 C
8,5
7,5 7 0
100
200
300
Time (min) 10 1
60
40
20
0
50
100
Diameter (nm)
100 150 Diameter (nm)
200
Figure 3. Measured resistance as a function of diameter for wires treated by an optimized self-terminating process (0.2%, solution B) compared to untreated wires. The dotted and solid curves are the same as in figure 1(b). All the measurements were done at room temperature in helium gas. The inset shows the time dependence of the solution pH value. The time dependence was measured for solution B with 0.2% concentration (see text) at 25 ◦ C (solid curve) and 62 ◦ C (dashed curve).
After 2 weeks in air
60
40
20
0
50
100
150
200
Diameter (nm) After processing After 1 month in vacuum
20
Resistance (kΩ)
50
Resistance (kΩ)
10 -2
Untreated 30 min. 3M s.t.
20
After processing
80
10 -1
40
(d) 25
(c)
10 0
60
0
200
150
After processing After 1 month in vacuum
80
Resistance (kΩ)
Resistance (kΩ)
pH
o
8
10 2
After processing After 2 weeks in air
80
9
10 3
(b)
(a)
10 9,5
10 4
15 10 5 0
70
80
90
100
110
70
80
90
100
110
Diameter (nm)
Diameter (nm) (e) 25 After processing After 1 hour in He
Resistance (kΩ)
20
The results of our experiments on free-laying nanowires indicate that the modified treatment in a highly dilute (2%) sulfur-saturated (NH4 )2 Sx solution yields a self-terminated passivation process for InAs nanowires: 300 min treatment time did not result in a larger decrease of nanowire diameter as compared to the 30 min long treatment. The 300 min treatment time for the highly diluted solutions at 62 ◦ C always resulted in substantial loss of the solution colour, the pH value gradually decreased with time and indicated neutrality of the solution at the end of the treatment (for the pH value time dependence see the inset in figure 3). This is a clear indication of the solution decomposition. Therefore, we did not perform the experiments with treatment times longer than 300 min. The nanowire treatment time for solution B was set to 30 min. The treatment time was less than, but comparable to, the observed solution degradation time for the highly dilute solutions, see the inset in figure 3. Therefore, the selftermination of the passivation process can be caused, at least partially, by the solution degradation during the processing. Our experimental data also show that the solution stability (and therefore treatment results) depends on solution volume and open surface. For this reason, the solution volume and open surface were fixed at 3 ml and around 1 cm2 respectively, and only freshly mixed solutions were used. The treatment was found to be reproducible and reliable, provided that all of the treatment parameters are kept constant. A set of different concentrations of solution B was tested in order to optimize the contact passivation process. Solutions with a (NH4 )2 Sx concentration 1% strongly etched the wires, resulting in poor contacts. A much smaller etching effect was observed after treatment in 0.5% solution, with a maximum of 3–5 nm removed from the wire, but in most cases nothing. The same results were obtained for two samples with 15 and 14 nanowires respectively. Using a 0.2% solution we could observe only a very small etching effect, this time a decrease
15 10 5 0
70
80
90
100
110
Diameter (nm)
Figure 4. (a) Resistance as a function of diameter for wires measured immediately after processing and after two weeks in air. (b) Resistance as a function of diameter measured for wires immediately after processing and after storage in vacuum for a month. (c) Magnification of a small diameter range in (a). The dot indicated by the arrow corresponds to a nanowire with a diameter of 70 nm. The dots corresponding to the wire were shifted for clarity. (d) Magnification of a small diameter range in (b). (e) Resistance as a function of diameter for wires measured immediately after processing and after 1 h in helium gas at room temperature. The data points with resistance for nanowires immediately after processing are the same for all of the panels in this figure. The grey lines in panels (c), (d) and (e) connect points corresponding to the same nanowire. The dotted curves in all of the figures are the same as the dotted curve in figure 1(b). All of the measurements were done at room temperature in helium gas.
in diameter of 2–3 nm, in most cases none. Again, the same results were observed for two independent samples with 10 and 18 wires respectively. Both treatments in 0.2% and 0.5% solutions gave 100% yield of ohmic contacts. Electrical measurement results for wires processed with the optimized self-terminating (0.2%) treatment process are shown in figure 3 together with those for the untreated wires. It is clear that the resistance for devices with passivated contact areas is much smaller than for devices fabricated without passivation treatment. The stability of the contacts was also investigated in different atmospheres. In figure 4(a), the measured resistance of wires of various diameters is shown both directly after processing and after two weeks in air. Figure 4(b) shows similar measurement results for wires immediately after processing and after one month of storage in vacuum 4
Nanotechnology 18 (2007) 105307
D B Suyatin et al
−3
(10 mbar). Figures 4(c) and (d) are based on the same data as figures 4(a) and (b), but show a smaller diameter range. It can be seen that roughly 50% of the wire devices that previously exhibited ohmic behaviour (initially the yield was 100%) irreversibly degraded after two weeks in air and were no longer ohmic. We believe that the degradation in air is entirely due to oxidation of the contact area. We base this assumption on the fact that we can obtain ohmic behaviour for nanowires that have been lying in air for several years. In contrast, storing the sample in vacuum for one month still gives ohmic contact behaviour, but there are some resistance fluctuations (figures 4(b), (d)). We attribute the fluctuations to charge rearrangements in nearby surroundings, either at the nanowire surface, or in the SiO2 on which the nanowire is lying. In fact, resistance measurements for wires immediately after processing and after 1 h in helium gas show that most of the wires exhibited a resistance decrease (figure 4(e)). The effect is most likely related to nanowire surface adsorbents, such as oxygen or water, which upon storage in He gas are slowly desorbed. The fact that the conductivity of nanowires is closely linked to the surface chemistry has been demonstrated in chemical sensing experiments [19]. The assumption of a charge rearrangement effect is also supported by the fact that the resistance fluctuations in time for the large diameter wires (figure 4) are substantially smaller than those for thinner wires; a higher carrier concentration in these wires more efficiently screens charges in the immediate neighbourhood, hence reducing the resistance fluctuations due to surface charge changes. Coating the nanowires, for example by nanowire shell growth, would probably minimize the observed resistance spread and fluctuations.
Naval Research (ONR), and the EU program NODE 015783. The authors gratefully acknowledge help from Kimberly Dick, Carina Fasth and Henrik Nilsson.
References [1] Appell D 2002 Wired for success Nature 419 553–5 [2] Samuelson L 2003 Self-forming nanoscale devices Mater. Today 6 22 [3] Lieber C M 2003 Nanoscale science and technology: building a big future from small things MRS Bull. 28 486–91 [4] Wang Z L 2005 Self-assembled nanoarchitectures of polar nanobelts/nanowires J. Mater. Chem. 15 1021–4 [5] De Franceschi S, van Dam J A, Bakkers E P A M, Feiner L F, Gurevich L and Kouwenhoven L P 2003 Single-electron tunnelling in InP nanowires Appl. Phys. Lett. 83 344–6 [6] Hansen A E, Bj¨ork M T, Fasth C, Thelander C and Samuelson L 2005 Spin relaxation in InAs nanowires studied by tunable weak antilocalization Phys. Rev. B 71 205328 [7] Wang S H, Mohney S E, Robinson J A and Bennett B R 2004 Sulfur passivation for shallow Pd/W/Au ohmic contacts to p-InGaSb Appl. Phys. Lett. 85 3471–3 [8] Park W I, Kim J S, Yi G-C, Bae M H and Lee H-J 2004 Fabrication and electrical characteristics of high-performance ZnO nanorod field-effect transistors Appl. Phys. Lett. 85 5052–4 [9] Cui Y, Zhong Z, Wang D, Wang W U and Lieber C M 2003 High performance silicon nanowire field effect transistors Nano Lett. 3 149–52 [10] Hanrath T and Korgel B A 2004 A comprehensive study of electron energy losses in Ge nanowires Nano Lett. 4 1455–61 [11] Seo K, Sharma S, Yasseri A A, Stewart D R and Kamins T I 2006 Surface charge density of unpassivated and passivated metal-catalyzed silicon nanowires Electrochem. Solid-State Lett. 9 G69–72 [12] Bessolov V N and Lebedev M V 1998 Chalcogenide passivation of III–V semiconductor surfaces Semiconductors 32 1141–56 [13] Petrovykh D Y, Yang M J and Whitman L J 2003 Chemical and electronic properties of sulfur-passivated InAs surfaces Surf. Sci. 523 231–40 [14] Thelander C, Bj¨ork M T, Larsson M W, Hansen A E, Wallenberg L R and Samuelson L 2004 Electron transport in InAs nanowires and heterostructure nanowire devices Solid State Commun. 131 573–9 [15] Bj¨ork M T, Ohlsson B J, Sass T, Persson A I, Thelander C, Magnusson M H, Deppert K, Wallenberg L R and Samuelson L 2002 One-dimensional heterostructures in semiconductor nanowhiskers Appl. Phys. Lett. 80 1058 [16] Jensen L E, Bj¨ork M T, Jeppesen S, Persson A I, Ohlsson B J and Samuelson L 2004 Role of surface diffusion in chemical beam epitaxy of InAs nanowires Nano Lett. 4 1961–4 [17] Baier H-U, Koenders L and M¨onch W 1986 Oxidation of cleaved InAs(1 1 0) surfaces at room temperature: surface band-bending and ionization energy Solid State Commun. 58 327–31 [18] Tao Y, Yelon A, Sacher E, Lu Z H and Graham M J 1992 S-passivated InP(100)-(1 × 1) surface prepared by a wet chemical process Appl. Phys. Lett. 60 2669–71 [19] Cui Y, Wei Q, Park H and Lieber C M 2001 Nanowire nanosensors for highly sensitive and selective detection of biological and chemical species Science 293 1289–92
4. Conclusion The formation of ohmic contacts to InAs nanowires using sulfur passivation in ammonium polysulfide, (NH4 )2 Sx , water solution has been studied. Surface passivation of InAs nanowires in a highly dilute water solution of (NH4 )2 Sx is shown to be self-terminating. An optimized self-terminated process provides ohmic contact formation to InAs nanowires together with minimal semiconductor material removal. Degradation of the contacts occurs for nanowires stored in air, whereas no degradation is observed after storage in low vacuum ambience. The self-terminated treatment reported here could potentially be adopted for surface passivation of other III–V semiconductor nanostructures, because the sulfur passivation mechanism of (NH4 )2 Sx is reported to be similar.
Acknowledgments This work was supported by the Swedish Foundation for Strategic Research (SSF), the Swedish Research Council (VR), the Knut and Alice Wallenberg Foundation, the Office of
5
