Home
Search
Collections
Journals
About
Contact us
My IOPscience
A technique for large-area position-controlled growth of GaAs nanowire arrays
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2016 Nanotechnology 27 135601 (http://iopscience.iop.org/0957-4484/27/13/135601) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 139.80.14.107 This content was downloaded on 01/03/2016 at 03:07
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 27 (2016) 135601 (6pp)
doi:10.1088/0957-4484/27/13/135601
A technique for large-area positioncontrolled growth of GaAs nanowire arrays Christoffer Kauppinen1, Tuomas Haggren1, Aleksandr Kravchenko2, Hua Jiang3, Teppo Huhtio1, Esko Kauppinen3, Veer Dhaka1, Sami Suihkonen1, Matti Kaivola2, Harri Lipsanen1 and Markku Sopanen1 1
Department of Micro- and Nanosciences, Micronova, Aalto University, PO Box 13500, FI-00076 Aalto, Finland 2 Department of Applied Physics, Micronova, Aalto University, PO Box 13500, FI-00076 Aalto, Finland 3 Department of Applied Physics and Nanomicroscopy Center, Aalto University, PO Box 15100, FI-00076 Aalto, Finland E-mail: christoffer.kauppinen@aalto.fi Received 28 October 2015, revised 9 January 2016 Accepted for publication 28 January 2016 Published 22 February 2016 Abstract
We demonstrate a technique for fabricating position-controlled, large-area arrays of vertical semiconductor nanowires (NWs) with adjustable periods and NW diameters. In our approach, a Au-covered GaAs substrate is first coated with a thin film of photoresponsive azopolymer, which is exposed twice to a laser interference pattern forming a 2D surface relief grating. After dry etching, an array of polymer islands is formed, which is used as a mask to fabricate a matrix of gold particles. The Au particles are then used as seeds in vapour–liquid–solid growth to create arrays of vertical GaAs NWs using metalorganic vapour phase epitaxy. The presented technique enables producing NWs of uniform size distribution with high throughput and potentially on large wafer sizes without relying on expensive lithography techniques. The feasibility of the technique is demonstrated by arrays of vertical NWs with periods of 255–1000 nm and diameters of 50–80 nm on a 2×2 cm area. The grown NWs exhibit high long range order and good crystalline quality. Although only GaAs NWs were grown in this study, in principle, the presented technique is suitable for any material available for Au seeded NW growth. Keywords: nanowires, GaAs, MOVPE, laser interference, azopolymer, lithography, vapour– liquid–solid (Some figures may appear in colour only in the online journal) (MOVPE) or molecular beam epitaxy. The deposition of the seeds is usually done by applying a solution with particles of the desired size, annealing a thin metal film or by lithography [13–16]. The lithographic technique allows full control of the density and placement of the NWs, which is required for ordered arrays. However, the most common lithographic techniques used for NW arrays, electron beam lithography (EBL) and nanoimprint lithography (NIL), pose their own challenges. EBL is extremely slow and expensive [17], and thus not a suitable method for large scale NW fabrication [14]. NIL is also challenging as the equipment and stamps are expensive, and for each period and NW diameter a separate stamp is needed. Emerging lithography techniques like nanosphere lithography (NSL), hold great promise in
Introduction Semiconductor nanowires (NWs) are promising for many applications in nanoelectronics and optoelectronics, including solar cells [1], FETs [2], LEDs [3, 4], lasers [5], antireflection coatings [6], sensors [7, 8] and photochemical water splitting [9]. Direct band gap III–V semiconductor NWs, such as GaAs NWs, are especially interesting due to their high carrier mobilities and high optical absorption. For these reasons, growing periodic III–V semiconductor NW arrays has recently been under intense research [10, 11]. Vapour–liquid– solid (VLS) growth using Au catalyst particles as seeds [12] is the most popular method for growing III–V semiconductor NWs, either with metalorganic vapor phase epitaxy 0957-4484/16/135601+06$33.00
1
© 2016 IOP Publishing Ltd Printed in the UK
Nanotechnology 27 (2016) 135601
C Kauppinen et al
fabricating hexagonal arrays of seeds for NW growth on large areas [18, 19], and some demonstrations of using NSL for NW growth have been shown [20, 21]. In this paper, we present a technique for fabricating large arrays of vertical GaAs NWs, using azopolymer thin films with laser interference lithography (LIL) [22], dry etching and VLS growth in MOVPE. The diameter of the seeds can be modified, the period can be adjusted and the repeatability of the LIL patterning has been demonstrated [17]. Our technique for position-controlled NW growth has at least two orders of magnitude faster exposure than EBL [23]. The exposure can be conducted in rooms with normal lighting (without the yellow lamps that protect traditional photoresist from exposure) [22], and the technique requires only regular optics, commercially available chemicals, and a blue–green laser, which all make the technique low-cost. Previously LIL has been used without azopolymers to create many types of NW arrays, such as Si [24], ZnO [14, 16], GaN [25] and InSb [26], but we present the first GaAs NW arrays, where the position is controlled by LIL and the first arrays of vertical NWs fabricated using azopolymers. The feasibility of the technique is demonstrated with arrays of vertical GaAs NWs having periods from of 255 to 1000 nm and diameters from 50 to 80 nm on a 2×2 cm area. The grown NWs exhibit high long range order and good crystalline quality.
Figure 1. Fabrication process for NW arrays. (a) First the
azopolymer solution is spin coated on the gold coated sample. (b) Then the polymer film is exposed to a p-polarized interference pattern from an argon ion laser. (c) The formed surface relief grating structure is then thinned with oxygen using reactive ion etching (RIE). (d) The remaining polymer tops are used as a milling mask. This forms a periodic array of gold islands. (e) When the islands are annealed, the surface tension forms spheroids from the islands. (f) The gold spheroids are used as seeds in the VLS MOVPE growth, resulting in a periodic array of NWs.
Experiment The fabrication process began with evaporation of 20 nm of Au on a 2×2 cm (111)B GaAs sample surface with an e-beam evaporator. Figure 1 depicts the successive process steps. A two weight percent solution of poly(Disperse Red 1 acrylate) (pDR1a) in 1,2-dichloroethane was prepared, by first mixing the solution, then filtering it through a 0.2 micron polytetrafluoroethylene (PTFE) syringe filter. This solution was then spin coated on the sample (30 s at 5000 rpm), see figure 1(a). Then, the samples were dried in an oven at 80 °C for 1 h, resulting in a roughly 100 nm thick polymer coating determined by ellipsometry. The spin-coated samples were exposed to an interference pattern of p-polarized light, see figure 1(b), using an argon ion laser with the wavelength of 488 nm. The interference pattern causes the azopolymer molecules to move away from the areas of high irradiance, thus forming a surface relief grating (SRG) [27, 28]. The exposure setup is based on a Lloyd’s mirror interferometer, shown in figure 2(a). The period (d) is l determined by d = 2 sin , where λ is the wavelength of the q laser and θ is the angle of incidence relative to the sample normal, see figure 2(a). The period was controlled by changing the Lloyd mirror angle: 73° relative to the sample normal for the 255 nm period and 14° for the 1000 nm period. Two exposures were performed with each sample, with a 90° rotation of the sample in the plane of the sample for the second exposure. The exposure times were set to obtain a symmetric SRG, see figure 2(b). For the 255 nm period, the
Figure 2. (a) Laser interference exposure setup. The p-polarized
beam is spatially filtered, expanded in free space and collimated using a lens. The Lloyd’s mirror interferometer is then used to form an interference pattern on the sample. The white arrows represent two rays of light forming constructive interference with an incidence angle θ. One ray hits the sample directly, while the other reflects from the mirror. (b) Atomic force microscopy (AFM) image of an azopolymer thin film exposed two times orthogonally.
2
Nanotechnology 27 (2016) 135601
C Kauppinen et al
irradiance was 8.5 mW cm−2, and the exposure times 90 and 13 min. For the 1000 nm period, the respective values were 78.6 mW cm−2, 20 and 25 min. After the exposure the polymer layer was etched with oxygen plasma, depicted in figure 1(c). The etching was performed in a reactive ion etching (RIE) system for a total of 2 min for the 255 nm period and 1 min 30 s for the 1000 nm period, with the radio frequency (RF) power of 40 W and the oxygen flow of 40 sccm at the pressure of 2 Pa. The etching was conducted in 30 s cycles with 2 min cooling periods. An array of azopolymer islands on gold was obtained, see figure 1(c). The dimensions of these mask islands can be controlled by the total duration of the oxygen plasma etch. The gold layer was then patterned using the grid of polymer islands as an etch mask, seen in figure 1(d). Argon ion milling was used for the patterning, for a total of 7 min 30 s for the 255 nm period and 6 min for the 1000 nm period. The milling was conducted in 30 s cycles, with 2 min cooling periods. The RF power was 100 W, the argon flow was 25 sccm and the pressure was 2 Pa. The diameter of the resulting Au seeds is controlled by both the oxygen plasma etch time and the milling time. The milled gold structures were then annealed in H2 flow of 5 slm at 650 °C under tertiarybutylarsine (TBA) flow of 180 μmol min−1, see figure 1(e). The annealing was performed in an atmospheric pressure horizontal flow MOVPE system. Finally, epitaxial NWs were grown on the samples using the Au spheroids as seeds, illustrated in figure 1(f). The NW growth was performed in the same atmospheric pressure MOVPE system as the annealing. An oxide removal step of 10 min at 650 °C in H2 flow of 5 slm was first performed on the samples. Next, NW growth was initiated at 470 °C by simultaneously switching on precursor flows for Ga (trimethylgallium, 6.12 μmol min−1) and As (TBA, −1 42.12 μmol min ). The growth time was 2 min, after which the reactor was cooled down to 270 °C under TBA flow. Between annealing and epitaxy, the samples were taken out of the MOVPE for analysis, which is presented later. A Zeiss Supra 40 field emission scanning electron microscope (SEM) was used to investigate the morphology and dimensions of the Au seeds and NW arrays, while a Jeol JEOL-2200FS transmission electron microscope (TEM) was used to investigate the crystallinity of the NWs. Energy dispersive x-ray spectroscopy (EDX) was performed with the same TEM, to study the composition and possible contamination of the NWs.
Figure 3. SEM micrographs of Au seeds, the insets show a close up
of one of the seeds. (a) An array of gold seeds on GaAs after annealing. The period of the array is 255 nm. (b) Gold seeds with a period of 1000 nm on GaAs, after annealing and extra milling, to remove residual gold particles.
removal usually requires an extra etching step [15]. Small residual gold particles were observed on the annealed sample with the 1000 nm period. This residual gold could be removed by an additional ion milling step of 4 min before growth, with the same parameters as before. The resulting 1000 nm array is seen in figure 3(b). Additional milling decreases the size of the seeds and makes the seeds nonspherical, see inset of figure 3(b). NW arrays grown from the seeds in figures 3(a) and (b) are displayed in figures 4(a) and (b), respectively. The NWs have a smooth surface and they are well ordered. The long range order of the finished NW arrays can be inferred from the SEM image in figure 4(c). This would indicate, that the seeds on the substrate have not moved significantly at any stage. The clear tapering seen on the NWs in figure 3(b) is discussed later. The measured dimensions of the NWs are presented in table 1, the error given is the standard deviation and N is the number of counted seeds or NWs. The data is collected from several SEM images from different positions on the sample. The diameters are calculated from top side SEM images using the imageJ software [29]. For the tapered 1000 nm period sample the bottom diameter is given, as this way the counting can be automated to imageJ. The lengths of the NWs are calculated from the SEM images. The 1000 nm period seeds are smaller than the 255 nm period seeds, but this is explained by the considerably longer argon milling time, (7 min 30 s for the 255 nm period versus 6 min with additional 4 min of cleaning for the 1000 nm period). Both periods produced NWs with uniform diameters, as the corresponding small standard deviations suggest. The 1000 nm period seeds possess a fairly large standard deviation. As the NWs grown from these seeds exhibit a much smaller standard deviation, the difference might be due to ImageJ’s difficulties at calculating the diameters from the small nonspherical seeds. With the 255 nm period, the NW diameter and the seed diameter are remarkably well correlated, but for the 1000 nm period the
Results and discussion The periodic arrays of annealed Au nanoparticles are shown in figure 3. As can be seen in figure 3(a), the seeds with the 255 nm spacing are spherical after annealing. Further analysis of the insets in figure 3 show that the seeds appear to be clean with no polymer residues visible in the SEM images, though this does not rule out for example knock-on contamination of the substrate due to the milling. The absence of the polymer is important as this may affect VLS growth, and accordingly its 3
Nanotechnology 27 (2016) 135601
C Kauppinen et al
16 times longer than the 255 nm period NWs. As this is not the case, the 1000 nm period NWs are growing sublinerly with respect to the Ga increase. Thus the 1000 nm period Au seeds can not consume all of the increase in the available Ga per NW, and the excess Ga contributes to radial growth on the NWs thus causing the tapering, and to 2D growth on the substrate surface. Similar behavior of increasing NW tapering, with increasing availability of group III adatoms has been observed before [32]. A TEM image of a GaAs NW from the 255 nm period sample is shown in figure 5(a). The NW displays smooth side walls, as was observed in the SEM images. The results of the EDX measurements in figure 5(b) indicate that the seed contains gold and and that the NW contains GaAs. The copper signal in the EDX measurements in figure 5(b) is due to the copper TEM sample grid. The EDX detector is unable to detect any contaminants from the processing, as the accuracy of EDX is on the order of one atomic percent. This indicates, that most polymer was etched away before annealing. However, possible small contamination, that is not detectable by EDX, seems not to affect the NW growth as the crystal quality is high, as seen from the high resolution TEM images and the fast Fourier transform (FFT) pattern in figures 5(c) and (d). The FFT pattern can be indexed by a single-phase zincblende GaAs structure, and indicates that the NW grows along the [111] direction. Infrequent crystal twinning was observed in some NWs. The main limitations of the presented technique relate to the minimum period of the mask array in the LIL process, and the adjustment of the seed diameter. To obtain smaller periods smaller wavelengths must be used, as the incidence angle of the exposure can not be increased much over 70°. This causes a problem as the absorbance of pDR1a peaks approximately at 450 nm and at 400 nm the absorbance is down over 50% [33], leading to a weaker movement of the polymer. More blue absorbing substances could be used, but pDR1a was chosen as pDR1a has been used for making SRGs for over 20 years [27], and it has been used succesfully before in nanofabrication [17]. Regarding the adjustment of the seed diameter, when the array period is changed then the resulting oxygen etching and argon milling times in RIE change significantly. Obtaining a specific diameter with a fixed period requires trial-and-error optimization for obtaining the right etching times. The minimum achievable feature size (the seed diameter) is determined by both the oxygen etching and the argon milling steps. In principle very small diameter polymer mask arrays can be made with oxygen plasma in RIE, but these would be very thin, and not very effective at protecting the underlaying metal. To compensate, less metal can be evaporated in the beginning. The authors speculate that although smaller diameter NWs are possible with the presented technique, dramatic (one order of magnitude) reduction in diameter is not feasible. Larger diameter NWs should be easier to achieve. Although only GaAs NWs were grown in this study, in principle, the presented technique is suitable for any material available for Au seeded NW growth. The sample sizes were limited to 2×2 cm, as larger semiconductor pieces could not be fit into the small reactor
Figure 4. (a) Finished NW array with the period of 255 nm. The
sample is tilted 10° from the surface normal in this SEM micrograph. (b) NW array with the 1000 nm period. This sample is tilted 30.4° from the surface normal. (c) The same sample as in figure 4(a) imaged with another magnification.
NW bottom diameter is significantly larger than the seed size. figure 4 confirms this further, as the longer period NWs appear to be tapered, while the shorter period NWs are not. The tapering of the 1000 nm period NWs can be explained by the fact that the availability of Ga adatoms for each NW is sixteenfold in the 1000 nm period compared to the 255 nm period. This increase is due to the sixteenfold NW density in the 255 nm period compared to the 1000 nm period. The length of the GaAs NWs is proportional to the growth rate, and the GaAs NW growth rate increases linearly at high V/III ratios (low TMG flow), but saturates to a sublinear increase at low V/III ratios (higher TMG flows) [30, 31]. If the growth rate of the NWs would depend linearly on the availability of Ga the 1000 nm period NWs would be roughly 4
Nanotechnology 27 (2016) 135601
C Kauppinen et al
Table 1. Mean seed and NW dimensions together with the standard deviations and the number of NWs or seeds measured (N) in the two
samples with different periods. Period Seed diameter NW diameter NW length
d=255 nm 46.3 ± 2.8 nm, 49.8 ± 2.7 nm, 1841 ± 24 nm,
d=1000 nm N=132 N=646 N=13
37.3 ± 11.9 nm, 83.2 ± 5.7 nm, 2662 ± 102 nm,
N=25 N=58 N=11
well ordered one-dimensional nanostructures without using time consuming or expensive lithography techniques. Futhermore, as MOVPE allows the creation of quantum wells in the NW and the doping of the NW to be controlled, this enables parallel fabrication of nanodevices with an areal density of around 109 cm−2.
Acknowledgments The authors thank the Academy of Finland (project 13251864), the Moppi project of the Aalto Energy Efficiency Program and the Aalto ELEC Doctoral School for funding. The authors also thank Dr Marco Mattila for inspiring discussions.
References Figure 5. TEM images, EDX spectra and electron diffraction pattern
of the NWs. (a) A TEM image of a single NW. (b) EDX spectra from two points on the wire. The lines in figure 5(a) point the EDX measurement areas on the NW. (c) A high resolution TEM image of a grown NW. (d) A magnified image of the boxed area in figure 5(c). The inset shows the FFT of the magnified area.
[1] Garnett E, Brongersma M, Cui Y and McGehee M 2011 Annu. Rev. Mater. Res. 41 269–95 [2] Cui Y, Zhong Z, Wang D, Wang W U and Lieber C M 2003 Nano Lett. 3 149–52 [3] Minot E D, Kelkensberg F, van Kouwen M, van Dam J A, Kouwenhoven L P, Zwiller V, Borgström M T, Wunnicke O, Verheijen M A and Bakkers E P A M 2007 Nano Lett. 7 367–71 [4] Tchernycheva M, Lavenus P, Zhang H, Babichev A V, Jacopin G, Shahmohammadi M, Julien F H, Ciechonski R, Vescovi G and Kryliouk O 2014 Nano Lett. 14 2456–65 [5] Duan X, Huang Y, Agarwal R and Lieber C M 2003 Nature 421 241–5 [6] Diedenhofen S L, Vecchi G, Algra R E, Hartsuiker A, Muskens O L, Immink G, Bakkers E P A M, Vos W L and Rivas J G 2009 Adv. Mater. 21 973–8 [7] Offermans P, Crego-Calama M and Brongersma S H 2010 Nano Lett. 10 2412–5 [8] Zheng G, Patolsky F, Cui Y, Wang W U and Lieber C M 2005 Nat. Biotechnol. 23 1294–301 [9] Yang X, Wolcott A, Wang G, Sobo A, Fitzmorris R C, Qian F, Zhang J Z and Li Y 2009 Nano Lett. 9 2331–6 [10] Schuster F, Hetzl M, Weiszer S, Garrido J A, de la Mata M, Magen C, Arbiol J and Stutzmann M 2015 Nano Lett. 15 1773–9 [11] Munshi A M et al 2014 Nano Lett. 14 960–6 [12] Wagner R S and Ellis W C 1964 Appl. Phys. Lett. 4 89–90 [13] Pierret A, Hocevar M, Diedenhofen S L, Algra R E, Vlieg E, Timmering E C, Verschuuren M A, Immink G W G, Verheijen M A and Bakkers E P A M 2010 Nanotechnology 21 065305 [14] Wei Y, Wu W, Guo R, Yuan D, Das S and Wang Z L 2010 Nano Lett. 10 3414–9 [15] Mårtensson T, Carlberg P, Borgström M, Montelius L, Seifert W and Samuelson L 2004 Nano Lett. 4 699–702
chamber of the MOVPE system. In principle, there are no physical limitations that would inhibit patterning for example whole 2 inch wafers with the presented technique. The authors speculate, that the leap from the presented size to a whole 2 inch wafer should be fairly straightforward.
Conclusions To conclude, we have demonstrated a scalable and high throughput technique for fabricating large-area position-controlled GaAs NW arrays using LIL and MOVPE, where the diameter and spacing of the NWs is well defined. The method can be applied to any material that can be grown with Au VLS. Laser interference patterns on azopolymer thin films and dry etching were used to control the period and the wire diameters. Analysis of the SEM images revealed the largescale ordering of the wires, and TEM images indicated that the NWs were single-phase zincblende. Using MOVPE to control the growth of the NWs and the ability to determine the period and diameter of the NWs on large areas opens up a new avenue for creating large-scale uniform NW arrays at high troughput. This provides an exciting ability to fabricate 5
Nanotechnology 27 (2016) 135601
C Kauppinen et al
[25] Hersee Stephen D, Sun X and Wang X 2006 Nano Lett. 6 1808–11 [26] Vogel A T, de Boor J, Becker M, Wittemann J V, Mensah S L, Werner P and Schmidt V 2011 Nanotechnology 22 015605 [27] Rochon P, Batalla E and Natansohn A 1995 Appl. Phys. Lett. 66 136–8 [28] Kim D, Tripathy S, Li L and Kumar J 1995 Appl. Phys. Lett. 66 1166–8 [29] Schneider C A, Rasband W S and Eliceiri K W 2012 Nat. Methods 9 671–5 [30] Soci C, Bao X Y, Aplin D P and Wang D 2008 Nano Lett. 8 4275–82 [31] Verheijen M A, Immink G, de Smet T, Borgström M T and Bakkers E P A M 2006 J. Am. Chem. Soc. 128 1353–9 [32] Dick K A, Deppert K, Samuelson L and Seifert W 2007 J. Crys. Growth 298 631–4 [33] Brown D, Natansohn A and Rochon P 1995 Macromolecules 28 6116–23
[16] Kim D S et al 2007 Small 3 76–80 [17] Kravchenko A, Shevchenko A, Grahn P, Ovchinnikov V and Kaivola M 2013 Thin Solid Films 540 162–7 [18] Bechelany M, Brodard P, Elias J, Brioude A, Michler J and Philippe L 2010 Langmuir 26 14364–71 [19] Bechelany M, Maeder X, Riesterer J, Hankache J, Lerose D, Christiansen S, Michler J and Philippe L 2010 Crys. Growth Des. 10 587–96 [20] Fan H J, Fuhrmann B, Scholz R, Syrowatka F, Dadgar A, Krost A and Zacharias M 2006 J. Cryst. Growth 287 34–8 [21] Madaria A R, Yao M, Chi C, Huang N, Lin C, Li R, Povinelli M L, Dapkus P D and Zhou C 2012 Nano Lett. 12 2839–45 [22] Kravchenko A, Shevchenko A, Ovchinnikov V, Priimagi A and Kaivola M 2011 Adv. Mater. 23 4174–7 [23] Greve M and Holst B 2013 J. Vac. Sci. Technol. B 31 043202 [24] de Boor J, Geyer N, Wittemann J V, Gösele U and Schmidt V 2010 Nanotechnology 21 095302
6
Search
Collections
Journals
About
Contact us
My IOPscience
A technique for large-area position-controlled growth of GaAs nanowire arrays
This content has been downloaded from IOPscience. Please scroll down to see the full text. 2016 Nanotechnology 27 135601 (http://iopscience.iop.org/0957-4484/27/13/135601) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 139.80.14.107 This content was downloaded on 01/03/2016 at 03:07
Please note that terms and conditions apply.
Nanotechnology Nanotechnology 27 (2016) 135601 (6pp)
doi:10.1088/0957-4484/27/13/135601
A technique for large-area positioncontrolled growth of GaAs nanowire arrays Christoffer Kauppinen1, Tuomas Haggren1, Aleksandr Kravchenko2, Hua Jiang3, Teppo Huhtio1, Esko Kauppinen3, Veer Dhaka1, Sami Suihkonen1, Matti Kaivola2, Harri Lipsanen1 and Markku Sopanen1 1
Department of Micro- and Nanosciences, Micronova, Aalto University, PO Box 13500, FI-00076 Aalto, Finland 2 Department of Applied Physics, Micronova, Aalto University, PO Box 13500, FI-00076 Aalto, Finland 3 Department of Applied Physics and Nanomicroscopy Center, Aalto University, PO Box 15100, FI-00076 Aalto, Finland E-mail: christoffer.kauppinen@aalto.fi Received 28 October 2015, revised 9 January 2016 Accepted for publication 28 January 2016 Published 22 February 2016 Abstract
We demonstrate a technique for fabricating position-controlled, large-area arrays of vertical semiconductor nanowires (NWs) with adjustable periods and NW diameters. In our approach, a Au-covered GaAs substrate is first coated with a thin film of photoresponsive azopolymer, which is exposed twice to a laser interference pattern forming a 2D surface relief grating. After dry etching, an array of polymer islands is formed, which is used as a mask to fabricate a matrix of gold particles. The Au particles are then used as seeds in vapour–liquid–solid growth to create arrays of vertical GaAs NWs using metalorganic vapour phase epitaxy. The presented technique enables producing NWs of uniform size distribution with high throughput and potentially on large wafer sizes without relying on expensive lithography techniques. The feasibility of the technique is demonstrated by arrays of vertical NWs with periods of 255–1000 nm and diameters of 50–80 nm on a 2×2 cm area. The grown NWs exhibit high long range order and good crystalline quality. Although only GaAs NWs were grown in this study, in principle, the presented technique is suitable for any material available for Au seeded NW growth. Keywords: nanowires, GaAs, MOVPE, laser interference, azopolymer, lithography, vapour– liquid–solid (Some figures may appear in colour only in the online journal) (MOVPE) or molecular beam epitaxy. The deposition of the seeds is usually done by applying a solution with particles of the desired size, annealing a thin metal film or by lithography [13–16]. The lithographic technique allows full control of the density and placement of the NWs, which is required for ordered arrays. However, the most common lithographic techniques used for NW arrays, electron beam lithography (EBL) and nanoimprint lithography (NIL), pose their own challenges. EBL is extremely slow and expensive [17], and thus not a suitable method for large scale NW fabrication [14]. NIL is also challenging as the equipment and stamps are expensive, and for each period and NW diameter a separate stamp is needed. Emerging lithography techniques like nanosphere lithography (NSL), hold great promise in
Introduction Semiconductor nanowires (NWs) are promising for many applications in nanoelectronics and optoelectronics, including solar cells [1], FETs [2], LEDs [3, 4], lasers [5], antireflection coatings [6], sensors [7, 8] and photochemical water splitting [9]. Direct band gap III–V semiconductor NWs, such as GaAs NWs, are especially interesting due to their high carrier mobilities and high optical absorption. For these reasons, growing periodic III–V semiconductor NW arrays has recently been under intense research [10, 11]. Vapour–liquid– solid (VLS) growth using Au catalyst particles as seeds [12] is the most popular method for growing III–V semiconductor NWs, either with metalorganic vapor phase epitaxy 0957-4484/16/135601+06$33.00
1
© 2016 IOP Publishing Ltd Printed in the UK
Nanotechnology 27 (2016) 135601
C Kauppinen et al
fabricating hexagonal arrays of seeds for NW growth on large areas [18, 19], and some demonstrations of using NSL for NW growth have been shown [20, 21]. In this paper, we present a technique for fabricating large arrays of vertical GaAs NWs, using azopolymer thin films with laser interference lithography (LIL) [22], dry etching and VLS growth in MOVPE. The diameter of the seeds can be modified, the period can be adjusted and the repeatability of the LIL patterning has been demonstrated [17]. Our technique for position-controlled NW growth has at least two orders of magnitude faster exposure than EBL [23]. The exposure can be conducted in rooms with normal lighting (without the yellow lamps that protect traditional photoresist from exposure) [22], and the technique requires only regular optics, commercially available chemicals, and a blue–green laser, which all make the technique low-cost. Previously LIL has been used without azopolymers to create many types of NW arrays, such as Si [24], ZnO [14, 16], GaN [25] and InSb [26], but we present the first GaAs NW arrays, where the position is controlled by LIL and the first arrays of vertical NWs fabricated using azopolymers. The feasibility of the technique is demonstrated with arrays of vertical GaAs NWs having periods from of 255 to 1000 nm and diameters from 50 to 80 nm on a 2×2 cm area. The grown NWs exhibit high long range order and good crystalline quality.
Figure 1. Fabrication process for NW arrays. (a) First the
azopolymer solution is spin coated on the gold coated sample. (b) Then the polymer film is exposed to a p-polarized interference pattern from an argon ion laser. (c) The formed surface relief grating structure is then thinned with oxygen using reactive ion etching (RIE). (d) The remaining polymer tops are used as a milling mask. This forms a periodic array of gold islands. (e) When the islands are annealed, the surface tension forms spheroids from the islands. (f) The gold spheroids are used as seeds in the VLS MOVPE growth, resulting in a periodic array of NWs.
Experiment The fabrication process began with evaporation of 20 nm of Au on a 2×2 cm (111)B GaAs sample surface with an e-beam evaporator. Figure 1 depicts the successive process steps. A two weight percent solution of poly(Disperse Red 1 acrylate) (pDR1a) in 1,2-dichloroethane was prepared, by first mixing the solution, then filtering it through a 0.2 micron polytetrafluoroethylene (PTFE) syringe filter. This solution was then spin coated on the sample (30 s at 5000 rpm), see figure 1(a). Then, the samples were dried in an oven at 80 °C for 1 h, resulting in a roughly 100 nm thick polymer coating determined by ellipsometry. The spin-coated samples were exposed to an interference pattern of p-polarized light, see figure 1(b), using an argon ion laser with the wavelength of 488 nm. The interference pattern causes the azopolymer molecules to move away from the areas of high irradiance, thus forming a surface relief grating (SRG) [27, 28]. The exposure setup is based on a Lloyd’s mirror interferometer, shown in figure 2(a). The period (d) is l determined by d = 2 sin , where λ is the wavelength of the q laser and θ is the angle of incidence relative to the sample normal, see figure 2(a). The period was controlled by changing the Lloyd mirror angle: 73° relative to the sample normal for the 255 nm period and 14° for the 1000 nm period. Two exposures were performed with each sample, with a 90° rotation of the sample in the plane of the sample for the second exposure. The exposure times were set to obtain a symmetric SRG, see figure 2(b). For the 255 nm period, the
Figure 2. (a) Laser interference exposure setup. The p-polarized
beam is spatially filtered, expanded in free space and collimated using a lens. The Lloyd’s mirror interferometer is then used to form an interference pattern on the sample. The white arrows represent two rays of light forming constructive interference with an incidence angle θ. One ray hits the sample directly, while the other reflects from the mirror. (b) Atomic force microscopy (AFM) image of an azopolymer thin film exposed two times orthogonally.
2
Nanotechnology 27 (2016) 135601
C Kauppinen et al
irradiance was 8.5 mW cm−2, and the exposure times 90 and 13 min. For the 1000 nm period, the respective values were 78.6 mW cm−2, 20 and 25 min. After the exposure the polymer layer was etched with oxygen plasma, depicted in figure 1(c). The etching was performed in a reactive ion etching (RIE) system for a total of 2 min for the 255 nm period and 1 min 30 s for the 1000 nm period, with the radio frequency (RF) power of 40 W and the oxygen flow of 40 sccm at the pressure of 2 Pa. The etching was conducted in 30 s cycles with 2 min cooling periods. An array of azopolymer islands on gold was obtained, see figure 1(c). The dimensions of these mask islands can be controlled by the total duration of the oxygen plasma etch. The gold layer was then patterned using the grid of polymer islands as an etch mask, seen in figure 1(d). Argon ion milling was used for the patterning, for a total of 7 min 30 s for the 255 nm period and 6 min for the 1000 nm period. The milling was conducted in 30 s cycles, with 2 min cooling periods. The RF power was 100 W, the argon flow was 25 sccm and the pressure was 2 Pa. The diameter of the resulting Au seeds is controlled by both the oxygen plasma etch time and the milling time. The milled gold structures were then annealed in H2 flow of 5 slm at 650 °C under tertiarybutylarsine (TBA) flow of 180 μmol min−1, see figure 1(e). The annealing was performed in an atmospheric pressure horizontal flow MOVPE system. Finally, epitaxial NWs were grown on the samples using the Au spheroids as seeds, illustrated in figure 1(f). The NW growth was performed in the same atmospheric pressure MOVPE system as the annealing. An oxide removal step of 10 min at 650 °C in H2 flow of 5 slm was first performed on the samples. Next, NW growth was initiated at 470 °C by simultaneously switching on precursor flows for Ga (trimethylgallium, 6.12 μmol min−1) and As (TBA, −1 42.12 μmol min ). The growth time was 2 min, after which the reactor was cooled down to 270 °C under TBA flow. Between annealing and epitaxy, the samples were taken out of the MOVPE for analysis, which is presented later. A Zeiss Supra 40 field emission scanning electron microscope (SEM) was used to investigate the morphology and dimensions of the Au seeds and NW arrays, while a Jeol JEOL-2200FS transmission electron microscope (TEM) was used to investigate the crystallinity of the NWs. Energy dispersive x-ray spectroscopy (EDX) was performed with the same TEM, to study the composition and possible contamination of the NWs.
Figure 3. SEM micrographs of Au seeds, the insets show a close up
of one of the seeds. (a) An array of gold seeds on GaAs after annealing. The period of the array is 255 nm. (b) Gold seeds with a period of 1000 nm on GaAs, after annealing and extra milling, to remove residual gold particles.
removal usually requires an extra etching step [15]. Small residual gold particles were observed on the annealed sample with the 1000 nm period. This residual gold could be removed by an additional ion milling step of 4 min before growth, with the same parameters as before. The resulting 1000 nm array is seen in figure 3(b). Additional milling decreases the size of the seeds and makes the seeds nonspherical, see inset of figure 3(b). NW arrays grown from the seeds in figures 3(a) and (b) are displayed in figures 4(a) and (b), respectively. The NWs have a smooth surface and they are well ordered. The long range order of the finished NW arrays can be inferred from the SEM image in figure 4(c). This would indicate, that the seeds on the substrate have not moved significantly at any stage. The clear tapering seen on the NWs in figure 3(b) is discussed later. The measured dimensions of the NWs are presented in table 1, the error given is the standard deviation and N is the number of counted seeds or NWs. The data is collected from several SEM images from different positions on the sample. The diameters are calculated from top side SEM images using the imageJ software [29]. For the tapered 1000 nm period sample the bottom diameter is given, as this way the counting can be automated to imageJ. The lengths of the NWs are calculated from the SEM images. The 1000 nm period seeds are smaller than the 255 nm period seeds, but this is explained by the considerably longer argon milling time, (7 min 30 s for the 255 nm period versus 6 min with additional 4 min of cleaning for the 1000 nm period). Both periods produced NWs with uniform diameters, as the corresponding small standard deviations suggest. The 1000 nm period seeds possess a fairly large standard deviation. As the NWs grown from these seeds exhibit a much smaller standard deviation, the difference might be due to ImageJ’s difficulties at calculating the diameters from the small nonspherical seeds. With the 255 nm period, the NW diameter and the seed diameter are remarkably well correlated, but for the 1000 nm period the
Results and discussion The periodic arrays of annealed Au nanoparticles are shown in figure 3. As can be seen in figure 3(a), the seeds with the 255 nm spacing are spherical after annealing. Further analysis of the insets in figure 3 show that the seeds appear to be clean with no polymer residues visible in the SEM images, though this does not rule out for example knock-on contamination of the substrate due to the milling. The absence of the polymer is important as this may affect VLS growth, and accordingly its 3
Nanotechnology 27 (2016) 135601
C Kauppinen et al
16 times longer than the 255 nm period NWs. As this is not the case, the 1000 nm period NWs are growing sublinerly with respect to the Ga increase. Thus the 1000 nm period Au seeds can not consume all of the increase in the available Ga per NW, and the excess Ga contributes to radial growth on the NWs thus causing the tapering, and to 2D growth on the substrate surface. Similar behavior of increasing NW tapering, with increasing availability of group III adatoms has been observed before [32]. A TEM image of a GaAs NW from the 255 nm period sample is shown in figure 5(a). The NW displays smooth side walls, as was observed in the SEM images. The results of the EDX measurements in figure 5(b) indicate that the seed contains gold and and that the NW contains GaAs. The copper signal in the EDX measurements in figure 5(b) is due to the copper TEM sample grid. The EDX detector is unable to detect any contaminants from the processing, as the accuracy of EDX is on the order of one atomic percent. This indicates, that most polymer was etched away before annealing. However, possible small contamination, that is not detectable by EDX, seems not to affect the NW growth as the crystal quality is high, as seen from the high resolution TEM images and the fast Fourier transform (FFT) pattern in figures 5(c) and (d). The FFT pattern can be indexed by a single-phase zincblende GaAs structure, and indicates that the NW grows along the [111] direction. Infrequent crystal twinning was observed in some NWs. The main limitations of the presented technique relate to the minimum period of the mask array in the LIL process, and the adjustment of the seed diameter. To obtain smaller periods smaller wavelengths must be used, as the incidence angle of the exposure can not be increased much over 70°. This causes a problem as the absorbance of pDR1a peaks approximately at 450 nm and at 400 nm the absorbance is down over 50% [33], leading to a weaker movement of the polymer. More blue absorbing substances could be used, but pDR1a was chosen as pDR1a has been used for making SRGs for over 20 years [27], and it has been used succesfully before in nanofabrication [17]. Regarding the adjustment of the seed diameter, when the array period is changed then the resulting oxygen etching and argon milling times in RIE change significantly. Obtaining a specific diameter with a fixed period requires trial-and-error optimization for obtaining the right etching times. The minimum achievable feature size (the seed diameter) is determined by both the oxygen etching and the argon milling steps. In principle very small diameter polymer mask arrays can be made with oxygen plasma in RIE, but these would be very thin, and not very effective at protecting the underlaying metal. To compensate, less metal can be evaporated in the beginning. The authors speculate that although smaller diameter NWs are possible with the presented technique, dramatic (one order of magnitude) reduction in diameter is not feasible. Larger diameter NWs should be easier to achieve. Although only GaAs NWs were grown in this study, in principle, the presented technique is suitable for any material available for Au seeded NW growth. The sample sizes were limited to 2×2 cm, as larger semiconductor pieces could not be fit into the small reactor
Figure 4. (a) Finished NW array with the period of 255 nm. The
sample is tilted 10° from the surface normal in this SEM micrograph. (b) NW array with the 1000 nm period. This sample is tilted 30.4° from the surface normal. (c) The same sample as in figure 4(a) imaged with another magnification.
NW bottom diameter is significantly larger than the seed size. figure 4 confirms this further, as the longer period NWs appear to be tapered, while the shorter period NWs are not. The tapering of the 1000 nm period NWs can be explained by the fact that the availability of Ga adatoms for each NW is sixteenfold in the 1000 nm period compared to the 255 nm period. This increase is due to the sixteenfold NW density in the 255 nm period compared to the 1000 nm period. The length of the GaAs NWs is proportional to the growth rate, and the GaAs NW growth rate increases linearly at high V/III ratios (low TMG flow), but saturates to a sublinear increase at low V/III ratios (higher TMG flows) [30, 31]. If the growth rate of the NWs would depend linearly on the availability of Ga the 1000 nm period NWs would be roughly 4
Nanotechnology 27 (2016) 135601
C Kauppinen et al
Table 1. Mean seed and NW dimensions together with the standard deviations and the number of NWs or seeds measured (N) in the two
samples with different periods. Period Seed diameter NW diameter NW length
d=255 nm 46.3 ± 2.8 nm, 49.8 ± 2.7 nm, 1841 ± 24 nm,
d=1000 nm N=132 N=646 N=13
37.3 ± 11.9 nm, 83.2 ± 5.7 nm, 2662 ± 102 nm,
N=25 N=58 N=11
well ordered one-dimensional nanostructures without using time consuming or expensive lithography techniques. Futhermore, as MOVPE allows the creation of quantum wells in the NW and the doping of the NW to be controlled, this enables parallel fabrication of nanodevices with an areal density of around 109 cm−2.
Acknowledgments The authors thank the Academy of Finland (project 13251864), the Moppi project of the Aalto Energy Efficiency Program and the Aalto ELEC Doctoral School for funding. The authors also thank Dr Marco Mattila for inspiring discussions.
References Figure 5. TEM images, EDX spectra and electron diffraction pattern
of the NWs. (a) A TEM image of a single NW. (b) EDX spectra from two points on the wire. The lines in figure 5(a) point the EDX measurement areas on the NW. (c) A high resolution TEM image of a grown NW. (d) A magnified image of the boxed area in figure 5(c). The inset shows the FFT of the magnified area.
[1] Garnett E, Brongersma M, Cui Y and McGehee M 2011 Annu. Rev. Mater. Res. 41 269–95 [2] Cui Y, Zhong Z, Wang D, Wang W U and Lieber C M 2003 Nano Lett. 3 149–52 [3] Minot E D, Kelkensberg F, van Kouwen M, van Dam J A, Kouwenhoven L P, Zwiller V, Borgström M T, Wunnicke O, Verheijen M A and Bakkers E P A M 2007 Nano Lett. 7 367–71 [4] Tchernycheva M, Lavenus P, Zhang H, Babichev A V, Jacopin G, Shahmohammadi M, Julien F H, Ciechonski R, Vescovi G and Kryliouk O 2014 Nano Lett. 14 2456–65 [5] Duan X, Huang Y, Agarwal R and Lieber C M 2003 Nature 421 241–5 [6] Diedenhofen S L, Vecchi G, Algra R E, Hartsuiker A, Muskens O L, Immink G, Bakkers E P A M, Vos W L and Rivas J G 2009 Adv. Mater. 21 973–8 [7] Offermans P, Crego-Calama M and Brongersma S H 2010 Nano Lett. 10 2412–5 [8] Zheng G, Patolsky F, Cui Y, Wang W U and Lieber C M 2005 Nat. Biotechnol. 23 1294–301 [9] Yang X, Wolcott A, Wang G, Sobo A, Fitzmorris R C, Qian F, Zhang J Z and Li Y 2009 Nano Lett. 9 2331–6 [10] Schuster F, Hetzl M, Weiszer S, Garrido J A, de la Mata M, Magen C, Arbiol J and Stutzmann M 2015 Nano Lett. 15 1773–9 [11] Munshi A M et al 2014 Nano Lett. 14 960–6 [12] Wagner R S and Ellis W C 1964 Appl. Phys. Lett. 4 89–90 [13] Pierret A, Hocevar M, Diedenhofen S L, Algra R E, Vlieg E, Timmering E C, Verschuuren M A, Immink G W G, Verheijen M A and Bakkers E P A M 2010 Nanotechnology 21 065305 [14] Wei Y, Wu W, Guo R, Yuan D, Das S and Wang Z L 2010 Nano Lett. 10 3414–9 [15] Mårtensson T, Carlberg P, Borgström M, Montelius L, Seifert W and Samuelson L 2004 Nano Lett. 4 699–702
chamber of the MOVPE system. In principle, there are no physical limitations that would inhibit patterning for example whole 2 inch wafers with the presented technique. The authors speculate, that the leap from the presented size to a whole 2 inch wafer should be fairly straightforward.
Conclusions To conclude, we have demonstrated a scalable and high throughput technique for fabricating large-area position-controlled GaAs NW arrays using LIL and MOVPE, where the diameter and spacing of the NWs is well defined. The method can be applied to any material that can be grown with Au VLS. Laser interference patterns on azopolymer thin films and dry etching were used to control the period and the wire diameters. Analysis of the SEM images revealed the largescale ordering of the wires, and TEM images indicated that the NWs were single-phase zincblende. Using MOVPE to control the growth of the NWs and the ability to determine the period and diameter of the NWs on large areas opens up a new avenue for creating large-scale uniform NW arrays at high troughput. This provides an exciting ability to fabricate 5
Nanotechnology 27 (2016) 135601
C Kauppinen et al
[25] Hersee Stephen D, Sun X and Wang X 2006 Nano Lett. 6 1808–11 [26] Vogel A T, de Boor J, Becker M, Wittemann J V, Mensah S L, Werner P and Schmidt V 2011 Nanotechnology 22 015605 [27] Rochon P, Batalla E and Natansohn A 1995 Appl. Phys. Lett. 66 136–8 [28] Kim D, Tripathy S, Li L and Kumar J 1995 Appl. Phys. Lett. 66 1166–8 [29] Schneider C A, Rasband W S and Eliceiri K W 2012 Nat. Methods 9 671–5 [30] Soci C, Bao X Y, Aplin D P and Wang D 2008 Nano Lett. 8 4275–82 [31] Verheijen M A, Immink G, de Smet T, Borgström M T and Bakkers E P A M 2006 J. Am. Chem. Soc. 128 1353–9 [32] Dick K A, Deppert K, Samuelson L and Seifert W 2007 J. Crys. Growth 298 631–4 [33] Brown D, Natansohn A and Rochon P 1995 Macromolecules 28 6116–23
[16] Kim D S et al 2007 Small 3 76–80 [17] Kravchenko A, Shevchenko A, Grahn P, Ovchinnikov V and Kaivola M 2013 Thin Solid Films 540 162–7 [18] Bechelany M, Brodard P, Elias J, Brioude A, Michler J and Philippe L 2010 Langmuir 26 14364–71 [19] Bechelany M, Maeder X, Riesterer J, Hankache J, Lerose D, Christiansen S, Michler J and Philippe L 2010 Crys. Growth Des. 10 587–96 [20] Fan H J, Fuhrmann B, Scholz R, Syrowatka F, Dadgar A, Krost A and Zacharias M 2006 J. Cryst. Growth 287 34–8 [21] Madaria A R, Yao M, Chi C, Huang N, Lin C, Li R, Povinelli M L, Dapkus P D and Zhou C 2012 Nano Lett. 12 2839–45 [22] Kravchenko A, Shevchenko A, Ovchinnikov V, Priimagi A and Kaivola M 2011 Adv. Mater. 23 4174–7 [23] Greve M and Holst B 2013 J. Vac. Sci. Technol. B 31 043202 [24] de Boor J, Geyer N, Wittemann J V, Gösele U and Schmidt V 2010 Nanotechnology 21 095302
6
