Controlling Heterojunction Abruptness in VLS-Grown Semiconductor

Jun 22, 2011 - Controlling Heterojunction Abruptness in VLS-Grown Semiconductor. Nanowires via in situ Catalyst Alloying. Daniel E. Perea,*. ,†. Nan...
0 downloads 0 Views 2MB Size
Download PDF
LETTER pubs.acs.org/NanoLett

Controlling Heterojunction Abruptness in VLS-Grown Semiconductor Nanowires via in situ Catalyst Alloying Daniel E. Perea,*,† Nan Li,† Robert M. Dickerson,‡ Amit Misra,† and S. T. Picraux*,† †

Center for Integrated Nanotechnologies and ‡Material Science and Technology-6, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States

bS Supporting Information ABSTRACT: For advanced device applications, increasing the compositional abruptness of axial heterostructured and modulation doped nanowires is critical for optimizing performance. For nanowires grown from metal catalysts, the transition region width is dictated by the solute solubility within the catalyst. For example, as a result of the relatively high solubility of Si and Ge in liquid Au for vapor liquid solid (VLS) grown nanowires, the transition region width between an axial Si Ge heterojunction is typically on the order of the nanowire diameter. When the solute solubility in the catalyst is lowered, the heterojunction width can be made sharper. Here we show for the first time the systematic increase in interface sharpness between axial Ge Si heterojunction nanowires grown by the VLS growth method using a Au Ga alloy catalyst. Through in situ tailoring of the catalyst composition using trimethylgallium, the Ge Si heterojunction width is systematically controlled by tuning the semiconductor solubility within a metal Au Ga alloy catalyst. The present approach of alloying to control solute solubilities in the liquid catalyst may be extended to increasing the sharpness of axial dopant profiles, for example, in Si Ge pn-heterojunction nanowires which is important for such applications as nanowire tunnel field effect transistors or in Si pn-junction nanowires. KEYWORDS: Si and Ge nanowires, axial heterojunction, VLS growth, alloy catalyst, in situ alloying, interface abruptness

C

ompared to compositionally uniform nanowires, axial heterostructured and modulation doped semiconductor nanowires have a much wider relevance for electronic,1 4 photovoltaic,5 and thermoelectric6 device applications. The interest in nanowire heterostructures lies in the ability to precisely tune the electronic transport properties by modulating the composition and/ or dopant type to create novel high-performance devices. For high subthreshold slope nanowire devices such as tunnel and avalanche field effect transistors, the performance is directly related to the junction abruptness. However, for group IV nanowires grown using the gold-catalyzed vapor liquid solid (VLS) growth mechanism,7 junction abruptness is generally limited in magnitude to the order of the nanowire diameter due to the liquid Au droplet mediating growth acting as a solute reservoir.8 Thus, establishing a method to create more abrupt axial heterojunctions, while maintaining VLS growth, is important to further exploit the inherent benefits of nanowires. The vapor liquid solid (VLS) growth mechanism is the favored and arguably most common method for synthesizing both compositionally uniform and heterojunction nanowires because of the ability to control growth conditions along the wire while achieving sufficiently long wires to be of interest for device applications. With the VLS method, a liquid nanosized droplet is formed on a substrate via a eutectic reaction between a metal catalyst nanoparticle (usually Au) and the semiconductor solute provided from a vapor-phase precursor in a chemical vapor deposition (CVD) system. Upon supersaturation of the liquid, r 2011 American Chemical Society

the semiconductor precipitates at the liquid substrate interface and the nanowire grows as material is added in a layer-by-layer process. The versatility of the VLS growth mechanism is exemplified in the ability to grow axial heterojunctions and modulation-doped nanowires.9 To create an axial junction, the vapor-phase reactants are simply changed during CVD growth. With this method, heterojunctions of different materials types have been realized for both group III V nanowires (e.g., InAs InP,10 GaP GaAs,11 GaAs InAs12) and group IV (e.g., Si Ge13 and Si SiGe14). While relatively abrupt interfaces have been observed for group III V systems upon switching the group V constituents, for example InAs InP heterostructures,15 where the group V solubility in the catalyst is small, more diffuse interfaces are observed for group IV heterostructures. In general, an abrupt change of the vapor-phase reactants does not necessarily correspond to an abrupt heterojunction. Consider the growth of a heterojunction from arbitrary semiconductor materials A and B, where initially the nanowire is composed of A. At the point of heterojunction growth initiation, the vapor-phase source of A is terminated and B is introduced. Assuming (1) both A and B have a finite solubility in the liquid, (2) both A and B have an equal probability for incorporation at the growing liquid solid interface, and (3) the liquid is homogeneous at all times, then as Received: April 4, 2011 Revised: June 13, 2011 Published: June 22, 2011 3117

dx.doi.org/10.1021/nl201124y | Nano Lett. 2011, 11, 3117–3122

Nano Letters

Figure 1. (a) Partial phase diagram for binary Au Ge (black, dashed), Au Si (red, dashed), and ternary Au50Ga50 Ge (black, solid), Au90.7Ga9.3-Si (red, solid) system. Vertical and horizontal dashed lines correspond to the eutectic composition and temperature, respectively. Adapted from ref 31. (b) Schematic illustrating the three main growth steps (I f III) of a Ge Si heterostructure grown from a liquid Au(1 x)Ga(x) catalyst alloy particle. The Ge Si composition profile (III) illustrates the SiGe alloyed region of arbitrary width.

nanowire growth precedes layer-by-layer, an interface transition region of both A and B will be created as species A is depleted from the liquid reservoir while B is replaced from the vapor phase. It then follows that the interface abruptness is proportional to the number of A atoms in the liquid reservoir, which in turn is proportional to the solubility of A atoms in the liquid. Ultimately, the interface abruptness of a VLS-grown nanowire is determined then by the solubility of the initial semiconductor species within the liquid. Thus lowering the semiconductor’s solubility in the liquid growth media should increase the heterojunction abruptness. One way to alter the solubility of the semiconductor species in the liquid Au catalyst is to introduce other metals to modify the phase diagram by alloying the Au catalyst. While there have been a number of recent studies of Si or Ge nanowire growth from Au alloys, this approach has not previously been used to alter the interface sharpness during VLS growth. However recently this concept was applied for solid phase growth of Si Ge axial heterostructures where, atomically abrupt compositional interfaces were demonstrated using solid Au2Al alloy nanoparticle catalysts. In this case the growth involved the vapor-solid solid (VSS) growth mechanism and the intrinsically low solubility of Si and Ge in the solid phase of the Au2Al alloy particles gave rise to the sharp interfacial abruptness.16 In order to circumvent the inheritably slow growth rate of the VSS growth mode and obtain practical wire lengths for device applications, those authors

LETTER

adopted a two step growth sequence: (1) growth of a long Si segment in VLS mode; (2) switching to a VSS growth mode of the Si Ge heterointerface by cooling of the catalyst alloy below the eutectic temperature to obtain a short Ge segment. However, to obtain a long Ge segment for practical applications it is necessary to switch back from solid phase to liquid phase growth, and this approach to achieve such structures has not yet been demonstrated. We describe here a new approach to controlling the heterointerface width by using a Au alloy catalyst for the growth media where the solubility of the semiconductor species is reduced by the alloying, enabling significantly more abrupt heterostructured interfaces. Further, we demonstrate the systematic tuning of the semiconductor solubility by in situ alloying of the metal catalyst, all while maintaining a VLS growth mode. In the approach described here, the heterojunction in the Ge Si nanowires was grown from a Au(1 x)Ga(x) metal alloy nanoparticle. We show, for the first time, that by varying the catalyst alloy composition in situ, the Ge eutectic solubility is systematically decreased which results in a decrease the heterointerface width between Ge and Si, while maintaining VLS growth from a liquid particle. The benefit of in situ catalyst alloying is that the alloy can be formed at any arbitrary point in the growth sequence by the introduction of a metalorganic precursor. This means that the nanowire growth can occur initially from one catalyst material such as Au (or other predefined metal or metal alloy), then by in situ alloying, growth can continue from a different catalyst material in which the eutectic solute solubility is altered, thus allowing tunability of the width of any heterointerface and/or pn-junction. Considering the relatively high solubility of Ge in Au (28%) relative to Si (18.6%), we chose Ga as a metal to alloy with Au since the eutectic solubility of Ge is reduced by greater than 6 times in a 1:1 Au:Ga alloy as compared to pure Au (Figure 1a). The nanowires were synthesized in a cold-walled CVD reactor using GeH4 and SiH4 as the Ge and Si sources, respectively, and H2 as the carrier gas. Nanowire growth occurred on Si(111) substrates that were initially solvent-cleaned and native oxide etched prior to dispersion of 30 and 60 nm diameter Au colloid nanoparticles (British Biocell) on separate substrates. Ge Æ111æ nanowire growth was initiated from Au colloid nanoparticles by nucleation at 340 C for 3 min followed by growth at 280 C for 20 30 min (Figure 1b, I). Following growth of the Ge nanowires from Au colloid nanoparticles, in situ catalyst alloying was accomplished during growth by simultaneously flowing vaporphase trimethylgallium (TMGa) from a bubbler line and GeH4 (Figure 1b, II). The presence of GeH4 during the alloying step was to maintain a supersaturated liquid state of the particle and prevent instability of the liquid catalyst growth seed or loss of the catalyst from the nanowire tip by diffusion along the nanowire sidewalls. In order to systematically vary the Au(1 x)Ga(x) alloy composition and thus the interface abruptness, the introduction of TMGa was performed at 380 C, while varying the exposure time from 1 to 120 s (see S1 in the Supporting Information). The Si segment of the heterostructure was grown after Au1 xGax alloy formation by removing both TMGa and GeH4 and introducing SiH4, while the temperature was ramped from 380 to 495 C (Figure 1b, III). A total pressure of 3 Torr was maintained throughout the Ge growth and alloy formation, while the total pressure of 0.5 Torr was maintained for the Si growth. We note that while some Ga atoms may be deposited on the surface of the Ge nanowire sidewalls during the in situ Au Ga alloying, Ga is a p-type dopant in Ge. However electronic transport 3118

dx.doi.org/10.1021/nl201124y |Nano Lett. 2011, 11, 3117–3122

Nano Letters

Figure 2. (a) SEM images of Ge Si heterostructure nanowires grown with increasing Ga concentration in the Au(1 x)Ga(x) catalyst for (1) x = 0.00, (2) x = 0.17, (3) x = 0.23, and (4) x = 0.34; 60 nm diameter Au colloid particles were used to catalyze the VLS growth for all wires. TMGa exposure temperature was 380 C and Si segment growth times were 17 min in each case. The Ge and Si segments correspond to the left and right of the white dashed line, respectively. Scale bar applies to all SEM images. (b) Composition profiles with corresponding interface abruptness across the Ge Si interface for 60 nm diameter nanowires grown from a pure Au (top) and Au(0.66)Ga(0.34) (bottom) catalyst. Top and bottom profiles were generated using an SEM and TEM, respectively.

properties are beyond the scope of the present paper; rather we focus here on the in situ catalyst alloy formation and its effect on altering the heterojunction width. Figure 2a shows Ge Si nanowires grown with increasing Ga content in the Au Ga alloy for wires 1 to 4. The Si segments for all nanowires were grown for the same amount of time, and thus the observed decrease in the Si segment length directly corresponds to a systematic decrease in the Si growth rate with increasing Ga content. Energy-dispersive X-ray spectroscopy (EDX) equipped on an FEI Inspect F scanning electron microscope (SEM), was used to measure both the composition profile of Ge and Si across the heterojunction (Figure 2b), as well as the Au Ga ratio in the alloyed catalyst17 (S2 in the Supporting Information). For nanowires with the smallest interface widths, EDX analysis was performed on a transmission electron microscope (TEM). We quantitatively define the interface transition region as the distance spanning 10 90% of the maximum X-ray intensity taken across the heterojunction and define this width as the interface abruptness. We note that in an earlier study by Lugstein et al., homogeneous Si nanowires were successfully grown in a hot-walled CVD reactor from a AuGa alloy that was deposited as a 1:1 bilayer thin film.18 However they reported significant tapering of the Si nanowires due to sidewall catalyzed growth by excess Ga, whereas we do not observe such excessive tapering of the Si segment after the heterojunction transition. Also, our EDX analysis suggests there is not significant loss of Ga from the alloyed Au catalyst during subsequent growth of the Si segment. Whether Ga is incorporated from the catalyst into the

LETTER

Figure 3. Ge Si heterointerface abruptness (10 90% composition change) measured by EDX as a function of increasing Ga concentration in the Au Ga alloy for nanowires grown from Au colloid particles of initial diameters: (a) 60 nm and (b) 30 nm. Horizontal dashed line indicates the minimum measured interface width that can be reported accurately using SEM analysis. The “open” data points were determined using TEM. Solid black lines are the simulated interface abruptness as described in the text.

Si or Ge segments at dopant level concentrations is an open question and will be addressed in future work using atom probe tomography.19 For the ∼60 nm diameter nanowire shown in Figure 2b (top), in which the interface was grown from pure Au, a width of 45 nm is measured. This abruptness is consistent with previous reports of the Si SiGe alloy heterointerface widths of nanowires of comparable diameter also grown from pure Au catalysts.8,20 In contrast, the interface abruptness for a wire grown from a Au0.66Ga0.34 catalyst shows a factor of ∼4 increase in the abruptness, with the width decreasing from 45 to 11 nm (Figure 2b (bottom)). Indeed, our measurements of heterointerface growth from a Au(1 x)Ga(x) alloy is found to consistently result in a significant increase in the heterointerface sharpness. We now focus on this change in heterointerface abruptness as a function of alloy composition and nanowire diameter. Increasing the Ga concentration in the Au Ga alloy results in a systematic increase in the Ge Si interface sharpness for wires grown initially from 60 nm (Figure 3a) and 30 nm (Figure 3b) Au colloids. The error bars for each point represent the standard deviation in the measured Au Ga ratio and interface abruptness as measured for three to seven wires. For wires grown from 60 nm (30 nm) Au colloids, the interface abruptness decreases from 45 nm (32 nm) for heterointerfaces grown from pure Au to 11 nm (6 nm) for heterointerfaces grown from a Au0.66Ga0.34 alloy. Due to the limited spatial resolution of the SEM which limits our ability to accurately measure widths less than ∼17 nm (see S3 in the Supporting Information), EDX equipped on a TEM (FEI Tecnai F30 or FEI Titan 80-300) was used for the “open” data points at high Ga catalyst compositions. By adapting the analysis described by Wen et al.,16 we note that TEM heterointerface analysis is not limited by spatial resolution using an electron beam energy of 300 keV, which results in a estimated beam broadening of ∼1 and ∼2 nm for a 30 and 60 nm diameter nanowire, respectively. Thus the increase in the Ga composition within the Au(1 x)Ga(x) alloy results in a steady increase in the heterointerface sharpness, and this increased sharpness can be 3119

dx.doi.org/10.1021/nl201124y |Nano Lett. 2011, 11, 3117–3122

Nano Letters

Figure 4. (a) Schematic of VLS layer-by-layer growth of an A B heterointerface. (b) Ge (blue) and Si (red) eutectic composition of a Au(1 x)Ga(x) alloy. Solid points are taken from the respective reported ternary phase diagrams. Dashed line is a linear extrapolation of the Si eutectic composition. (c) Ge composition profiles (solid) across the Ge Si heterointerface for a nanowire grown from Au (black) and Au(0.68)Ga(0.32) (red) alloy catalysts. Dashed lines are the simulated Ge composition profiles as described in the text. Nanowire diameter is 65 nm.

attributed to the decrease in Ge solute concentration in the eutectic liquid. To better understand the dependence of heterointerface abruptness on solute solubility during Au(1 x)Ga(x) catalyst VLS nanowire growth of axial heterostructures, we have applied a layer-by-layer growth model which we developed previously to predict the Ge Si composition transition profiles during pure Au catalyst growth.21 This growth model follows the same basic approach as a transition model for VLS heterostructure growth developed independently by Li et al.,20 where an analytical continuum solution was developed which yields an exponential decrease in the concentration of a growth species A with distance upon switching the growth from species A to B. This exponential decay solution was shown to provide a reasonable approximation to the Ge axial interface profile for Si1 xGex Si heterostructure growth for Ge concentration changes of 10 and 16%.8,20 In our model, we instead use a layer-by-layer simulation approach to solve for the axial composition profiles, which avoids the need for several limiting assumptions of the analytic approach, such as constant solute solubility. For example, in simulating a Ge to Si axial nanowire growth transition for a pure Au catalyst, we continuously adjust the solute solubility in Au for each atomic layer of growth, which changes from ∼28% for pure Ge to ∼19% for pure Si in the liquid Au1 x ySixGey alloy during a Ge Si transition. Other effects, such as the change in solute solubility with growth temperature, the change in solubility with Au catalyst alloy concentration, and a changing nanowire growth rate during the heterostructure growth transition, are also readily accounted for in this simulation approach. A heterointerface transition region results when nanowire growth is mediated by a catalyst material with a finite solubility so that the composition along the nanowire does not instantaneously change with a change in the gas phase growth species due to the liquid catalyst acting as a reservoir. Our layer-by-layer growth model for this growth transition region has two basic

LETTER

assumptions, (1) the liquid growth catalyst acts as a reservoir in quasi-equilibrium with the growth species according to their solubilities: this implies (a) the incorporation of the growth atoms across the vapor liquid interface is fast enough to maintain equilibrium across the vapor liquid interface and (b) the diffusion of the solute atoms within the liquid is sufficiently rapid to maintain a homogeneous liquid composition; (2) the probabilities for the growth species being incorporated at the liquid solid interface for each atomic layer of growth is given by the species’ segregation coefficient, kseg = Cl/Cs where C is the equilibrium solute concentration in the liquid (l) and solid (s), respectively. These assumptions are well satisfied for typical liquid-mediated, low-temperature VLS growth. Figure 4a illustrates schematically the situation for nanowire heterostructure VLS growth after switching from growth with species A to species B and the incorporation of A and B atoms into the growth of a new layer at the liquid solid interface. As A atoms are consumed at the liquid solid growth interface, they are replaced by B atoms across the liquid vapor interface and the concentration of A atoms in the liquid is gradually depleted. This layer-by-layer simulation model is well suited to the case of nucleation-limited VLS nanowire growth where the concentration builds up to a supersaturation level in the liquid and then, upon nucleation of a new layer, rapid growth occurs from the solute atoms in the liquid catalyst. As pointed out by Li et al. this model can be applied to both compositional and dopant transitions during nanowire growth.20 The growth model can also be used for switching between pure A and B atoms or any combination of AB nanowire alloy compositional changes. In the present case of Ge to Si nanowire heterostructure growth at low temperatures, we have used the eutectic compositions for the solubility of Ge and Si in the Au(1 x)Ga(x) growth catalyst and assumed the Ge and Si eutectic concentrations scale linearly with Ga concentration, x, in the Au(1 x)Ga(x) alloy (Figure 4b). This assumption is based on the near-linear dependence of the three published Si eutectic compositions and a similar inference for Ge between the available eutectic compositions. As the composition changes during the transition, the total solute concentration of Ge and Si in the liquid pseudobinary alloy is taken as the linear combination of that for pure Ge and pure Si. Also the incorporation rate for Ge and Si for each atomic layer grown is taken as proportional to the relative concentrations of Ge and Si in the liquid at each instant in time. A more detailed description of this simulation method based on our layer-by-layer growth model is given in S4 in the Supporting Information. With the above conditions, we compare in Figure 4c the simulated (dashed line) and measured (solid line) decrease in Ge concentration with distance along the nanowire axis at the Ge to Si transition region for pure Au and Au0.68Ga0.32 alloy catalysts. The good agreement in the shape of the profiles for both 30 and 60 nm diameter nanowires with no adjustable parameters suggests this model accurately predicts the solubility-dependent transition behavior for VLS nanowire axial heterostructure growth. Extending the model to simulate the interface abruptness as a function of increasing Ga composition in the Au(1 x)Ga(x) alloy (i.e., decreasing Ge solubility), we find that the simulated interface abruptness, shown by the solid lines in Figure 3, is also consistent with the measured widths. Nanowire growth from a liquid catalyst particle (i.e., VLS growth) is desirable over growth from a solid particle (i.e., VSS growth) in that VLS growth rate is in general much higher compared with VSS, which in turn allows long nanowire segment 3120

dx.doi.org/10.1021/nl201124y |Nano Lett. 2011, 11, 3117–3122

Nano Letters

Figure 5. (a) Ge eutectic temperature vs the AuGa alloy composition. The open circles are taken from the individual ternary phase diagrams, and the solid line is a linear interpolation between the points. (b) SEM image of AuGa catalyst on a Ge nanowire.

lengths required for devices to be reached in a reasonable amount of time. In general, for the catalyst nanoparticle to be in the liquid state, nucleation and initial nanowire growth should occur at or above the eutectic temperature. For a bulk binary system between Au and Ge, this temperature is approximately 380 C. However VLS growth of Ge nanowires from an undercooled liquid Au droplet is readily achieved at temperatures as low as 260 C.22 Considering our observed Au Ga alloy formation with Ge present at the TMGa exposure temperature of 380 C, we anticipate the alloy to remain in a liquid state to at least 10% Ga composition based on a linear extrapolation of the Au Ga Ge alloy eutectic temperature Ge (see Figure 5a). In fact, the SEM image of the catalyst tip region of a Ge nanowire after TMGa exposure at 380 C shown in Figure 5b, although bulbous, exhibits a hemispherical-like shape without any obvious surface faceting and is empirically consistent with the shape of Au catalyst nanoparticles known to remain liquid throughout growth.23,24 Moreover, measured heterojunction widths greater than 10 nm (Figure 3) also support VLS growth throughout the transition from Ge to Si for our Au Ga alloy conditions, where otherwise near atomically abrupt heterojunction widths may be expected should growth proceed via VSS growth.16 We note that in situ TEM growth studies of this Au Ga alloying approach to increase heterojunction abruptness may be useful in providing additional insights and confirming the composition range over which VLS growth for Au(1 x)Ga(x) alloys occurs. The addition of Ga to the Au catalyst also influences the nanowire growth kinetics. As shown in Figure 2a, a decrease in the Si segment length with increasing Ga composition in the AuGa alloy is observed and suggests that the presence of Ga decreases the SiH 4 decomposition kinetics at the vapor liquid interface and/or decreases the crystallization rate at the liquid solid interface. Due to the much poorer catalytic property of pure Ga compared to pure Au to decompose silane,25 silicon nanowire growth from a pure Ga catalyst metal requires the use of a RF plasma to support radical-assisted decomposition of silicon for VLS growth.25 28 This is despite a low Ga Si liquid eutectic temperature of ∼30 C. Also, Lugstein et al. have shown a decreased activation energy for Si nanowire growth from a 1:1 Au:Ga alloy as compared to pure Au.18 Thus we infer the observed reduction in Si nanowire growth rate is due to a reduction in the silane incorporation kinetics at the vapor liquid interface due to the presence of Ga in the liquid catalyst. A more detailed discussion of the effects of Ga on the growth kinetics will be given in future studies. Additional simulations were carried out as a function of nanowire diameter to explore the limits of abruptness through lowering the solute concentration by alloying. Shown in Figure 6 is the simulated Ge Si heterointerface width (as defined by the

LETTER

Figure 6. Simulated Ge Si interface abruptness as a function of nanowire diameter for varying AuGa alloy compositions; Ge eutectic composition in the respective AuGa alloys are: blue (top line), 28%; red (middle), 15%; black (bottom), 7%.

90 10% composition change) as a function of diameter and semiconductor eutectic composition in the catalyst alloy. For an interface grown from pure Au (blue triangle), the interface abruptness decreases linearly with decreasing nanowire diameter. A linear dependence of the interface width with diameter, d, is to be expected, since the number of solute atoms in the liquid catalyst scales as d3, whereas the corresponding number of solute atoms lost upon the growth of each atomic layer scales as d2. The slope in Figure 6 is found to be consistent with that predicted for Si Si(1 x)Gex nanowire heterojunctions reported by Li et al.20 and with the observations of Clark et al.16 for the limited concentration change of x ∼ 0 to 0.16. Extending the simulation to interfaces grown from a Au(1 x)Ga(x) alloy, the interface abruptness increases with decreasing Ge eutectic composition. As a function of nanowire diameter the slope again is predicted to decrease linearly with decreasing Ge solubility and the decrease in interface width is seen to become less pronounced over a given range of diameters for alloys with a decreasing Ge eutectic composition (i.e., decreasing Ge solubility). With decrease of the solute solubility in the liquid catalyst, these results unambiguously demonstrate that semiconductor nanowire heterojunctions can be made more abrupt by tailoring the catalyst composition via in situ alloying and that this effect can be very significant in increasing the abruptness of Si/Ge heterostructured interfaces. Although we have presented a new approach of in situ catalyst alloying to engineer the heterointerface width using the specific case of Au(1 x)Ga(x) catalysts, other metal alloys should offer additional opportunities for heterojunction engineering in nanowires. In general, axial heterojunction growth using sizecontrolled ex situ chemically synthesized metal alloy nanoparticles with tunable composition and diameter should also provide a rational route to heterojunction engineering.29 Although a number of previous group IV nanowire growth studies using many different transition metal solid catalysts have been demonstrated via the VSS growth mode,30 the present results motivate exploration of heterojunction formation and the resulting interface sharpness using alternative liquid catalyst metals and alloys. Also, given that the solid catalysts used for VSS growth tend to be transition metal elements, metal alloys with main-group elements may be more promising as alloys for heterojunction engineering via the VLS growth mode, as has been shown here with AuGa. In conclusion, we have shown the systematic increase in interface sharpness of a nanowire axial heterojunction grown via the VLS growth mechanism which can be achieved using catalyst alloying to decrease solute solubility. By in situ tuning of 3121

dx.doi.org/10.1021/nl201124y |Nano Lett. 2011, 11, 3117–3122

Nano Letters the catalyst composition, the Ge Si heterojunction width can be systematically controlled by decreasing the semiconductor solubility within the liquid metal AuGa alloy catalyst, and Ge Si interface widths as narrow as 6 nm have been demonstrated. The present approach can be extended to increasing the sharpness of axial dopant profiles in, for example, Si, Ge, or Si Ge pnheterojunction nanowires which is important for potential device applications such as nanowire tunnel field effect transistors and Si or Ge pn-junction structures.

’ ASSOCIATED CONTENT

bS

Supporting Information. Data supporting (S1) control of the catalyst alloy composition, (S2) EDX analysis procedure, (S3) determination of the SEM spatial resolution for EDX analysis, and (S4) heterostructure transition profile calculations. This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: (D.E.P.) [email protected]; (S.T.P.) [email protected].

’ ACKNOWLEDGMENT This research was funded in part by the Laboratory Directed Research and Development (LDRD) Program at Los Alamos National Laboratory and performed in part at the Center for Integrated Nanotechnologies, a U.S. Department of Energy, Office of Basic Energy Sciences user facility at Los Alamos National Laboratory (Contract DE-AC52-06NA25396). D.E.P. acknowledges funding from a LDRD-funded Postdoctoral Director’s Fellowship at Los Alamos National Laboratory (LANL). Electron microscopy and EDX analysis was carried out at the MST-6 Electron Microscope Laboratory LANL. The authors would also like to acknowledge helpful discussions with E. J. Schwalbach and P. W. Voorhees at Northwestern University and T. Lookman at LANL. ’ REFERENCES (1) Gudiksen, M. S.; Lauhon, L. J.; Wang, J.; Smith, D. C.; Lieber, C. M. Nature 2002, 415, 617–620. (2) Vallett, A. L.; Minassian, S.; Kaszuba, P.; Datta, S.; Redwing, J. M.; Mayer, T. S. Nano Lett. 2010, 10, 4813–4818. (3) Borg, B. M.; Dick, K. A.; Ganjipour, B.; Pistol, M. E.; Wernersson, L. E.; Thelander, C. Nano Lett. 2010, 10, 4080–4085. (4) Bjork, M. T.; Knoch, J.; Schmid, H.; Riel, H.; Riess, W. Appl. Phys. Lett. 2008, 92, 193504. (5) Kempa, T. J.; Tian, B. Z.; Kim, D. R.; Hu, J. S.; Zheng, X. L.; Lieber, C. M. Nano Lett. 2008, 8, 3456–3460. (6) Li, D. Y.; Wu, Y.; Fan, R.; Yang, P. D.; Majumdar, A. Appl. Phys. Lett. 2003, 83, 153186. (7) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (8) Clark, T. E.; Nimmatoori, P.; Lew, K. K.; Pan, L.; Redwing, J. M.; Dickey, E. C. Nano Lett. 2008, 8, 1246–1252. (9) Agarwal, R. Small 2008, 4, 1872–1893. (10) Bjork, M. T.; Ohlsson, B. J.; Sass, T.; Persson, A. I.; Thelander, C.; Magnusson, M. H.; Deppert, K.; Wallenberg, L. R.; Samuelson, L. Nano Lett. 2002, 2, 87–89. (11) Borgstrom, M. T.; Verheijen, M. A.; Immink, G.; de Smet, T.; Bakkers, E. P. A. M. Nanotechnology 2006, 17, 4010–4013.

LETTER

(12) Ohlsson, B. J.; Bjork, M. T.; Persson, A. I.; Thelander, C.; Wallenberg, L. R.; Magnusson, M. H.; Deppert, K.; Samuelson, L. Phys. E (Amsterdam, Neth.) 2002, 13, 1126–1130. (13) Dayeh, S. A.; Huang, J.; Gin, A. V.; Picraux, S. T. Synthesis, fabrication, and characterization of Ge/Si axial nanowire heterostructure tunnel FETs. In Proceedings from the 10th IEEE International Conference on Nanotechnology and Joint Symposium with Nano Korea 2010 KINTEX (IEEE-NANO), Seoul, South Korea, Aug. 17 20, 2010; IEEE: Piscataway, NJ, 2010; pp 238 241. (14) Redwing, J. M.; Lew, K. K.; Bogart, T. E.; Pan, L.; Dickey, E. C.; Carim, A. H.; Wang, Y. F.; Cabassi, M. A.; Mayer, T. S. Synthesis and properties of Si and SiGe/Si nanowires. In Proceedings from SPIE: Quantum Dots, Nanoparticles, and Nanoclusters, San Jose, CA, 2004; Huffaker, D. L. B., P., Ed.; SPIE: Bellingham, WA, 2004; pp 52 59. (15) Bjork, M. T.; Ohlsson, B. J.; Thelander, C.; Persson, A. I.; Deppert, K.; Wallenberg, L. R.; Samuelson, L. Appl. Phys. Lett. 2002, 81, 4458–4460. (16) Wen, C. Y.; Reuter, M. C.; Bruley, J.; Tersoff, J.; Kodambaka, S.; Stach, E. A.; Ross, F. M. Science 2009, 326, 1247–1250. (17) Despite an inferior spacial resolution and sensitivity, EDX analysis on an SEM was chosen for the greater analysis throughput compared to analysis on a TEM. (18) Lugstein, A.; Steinmair, M.; Hyun, Y. J.; Bertagnolli, E.; Pongratz, P. Appl. Phys. Lett. 2007, 90, 023109. (19) Perea, D. E.; Hemesath, E. R.; Schwalbach, E. J.; Lensch-Falk, J. L.; Voorhees, P. W.; Lauhon, L. J. Nat. Nanotechnol. 2009, 4, 315–319. (20) Li, N.; Tan, T. Y.; Gosele, U. Appl. Phys. A: Mater. Sci. Process. 2008, 90, 591–596. (21) Picraux, S. T. First reported at the Spring MRS Meeting, 2008. (22) Kodambaka, S.; Tersoff, J.; Reuter, M. C.; Ross, F. M. Science 2007, 316, 729–732. (23) Wu, Y. Y.; Yang, P. D. J. Am. Chem. Soc. 2001, 123, 3165–3166. (24) Ross, F. M. Rep. Prog. Phys. 2010, 73, 114501. (25) Sunkara, M. K.; Sharma, S.; Miranda, R.; Lian, G.; Dickey, E. C. Appl. Phys. Lett. 2001, 79, 1546–1548. (26) Jeon, M.; Tomitsuka, Y.; Kamisako, K. J. Ind. Eng. Chem. 2008, 14, 836–840. (27) Sharma, S.; Sunkara, M. K. Nanotechnology 2004, 15, 130–134. (28) Zardo, I.; Yu, L.; Conesa-Boj, S.; Estrade, S.; Alet, P. J.; Roessler, J.; Frimmer, M.; Cabarrocas, P. R. I.; Peiro, F.; Arbiol, J.; Morante, J. R.; Morral, A. F. I. Nanotechnology 2009, 20, 155602. (29) Connell, J. G.; Al Balushi, Z. Y.; Sohn, K.; Huang, J. X.; Lauhon, L. J. J. Phys. Chem. Lett. 2010, 1, 3360–3365. (30) Lensch-Falk, J. L.; Hemesath, E. R.; Perea, D. E.; Lauhon, L. J. J. Mater. Chem. 2009, 19, 849–857. (31) Butt, M. T. Z.; Bodsworth, C.; Prince, A. Au-Ga-Ge Phase Diagram; Okamoto, H., Cenzual, K., Eds.; ASM International: Materials Park, OH, 1989.

3122

dx.doi.org/10.1021/nl201124y |Nano Lett. 2011, 11, 3117–3122