Supporting Info: Synthesis, morphological and electro-optical ...
Supporting Information for: Synthesis, Morphological and Electro-optical Characterizations of Metal/semiconductor Nanowire Heterostructures Markus Glaser†, Andreas Kitzler†, Andreas Johannes‡, Slawomir Prucnal§, Heidi Potts║, Sonia Conesa-Boj║, Lidija Filipovic┴, Hans Kosina┴, Wolfgang Skorupa§, Emmerich Bertagnolli†, Carsten Ronning‡, Anna Fontcuberta i Morral║, Alois Lugstein†,*
†
Institute of Solid State Electronics, TU Wien, Floragasse 7, 1040 Wien, Austria
‡
Institute for Solid State Physics, Friedrich-Schiller-University Jena, Max-Wien-Platz 1, 07743 Jena, Germany
§
Institute of Ion Beam Physics and Materials Research, Helmholtz-Zentrum DresdenRossendorf, Bautzner Landstraße 400, 01328 Dresden, Germany
║
Laboratoire des Matériaux Semiconducteurs, Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland
┴
*
Institute for Microelectronics, TU Wien, Gußhausstraße 25-29, 1040 Wien, Austria
To whom correspondence should be addressed, E-mail: [email protected]
1
Contents Supporting Information 1. NW response on implantation and FLA. ..................................... 3 Supporting Information 2. Example of a NW heterostructure with thin crystallite sections................................................................................................. 6 Supporting Information 3. Raman investigations. ................................................................ 7 Supporting Information 4. Detailed elemental analysis of a Si/GaAs/Ga/Si NW heterostructure. .......................................................................... 10 Supporting information 5. Example of a twin defect in the GaAs/Si NW heterostructure. .. 12 Supporting information 6. Determining the doping type of Si NWs by thermoelectric measurements. .......................................................................................................... 13 Supporting information 7. Barrier height simulations of observed Si/GaAs/Si heterostructures. ..................................................................................... 15 Supporting Information 8. Hole concentration and impact ionization simulations of a Ga/GaAs/Si heterostructure............................................................. 16 References ........................................................................................................................ 18
2
Supporting Information 1. NW response on implantation and FLA. In Figure S1, tilted scanning electron microscope (SEM) images of particular Si nanowires (NWs) are shown, visualizing the effect of high fluence ion implantation and further flash lamp annealing (FLA). The left panel shows the detailed view of a group of as-grown Si NWs oriented perpendicular to the substrate after deposition of 15 nm SiO2 by plasma enhanced chemical vapor deposition (PECVD). In the center panel the same NW group is shown after room temperature In & As ion implantation with the sample mounted on a 45° tilted, rotating sample stage. Thereby NWs perpendicular to the substrate surface retain their orientation, whereas NWs with oblique growth directions or kinks are bent towards the ion beam and straightened1. Remarkably, all NWs are shortened considerably during ion implantation which is ascribed to plastic deformation under ion irradiation2. Finally, after FLA (right panel) no further change of geometry can be observed, however small sections of InAs have formed inside the NW core similar to the GaAs implanted and annealed samples described in the main text.
Figure S2 shows a cross sectional energy filtered TEM (EFTEM) image of the Si/SiO 2 core/shell NW after ion implantation and the respective diffraction pattern. The sample was prepared by focused ion beam (FIB) cutting immediately after ion implantation. Bright contrast correlates with occurrence of the investigated material. As can be seen, the about 20 nm thick SiO2 shell deposited on the VLS grown Si NW is still preserved after high fluence ion implantation. During implantation the NW cross section changed from hexagonal for pristine NWs 3 to circular due to reinforced sputtering at oblique edges. The diffraction pattern in Figure S2b clearly proves full amorphization of the Si NW due to high fluence Ga and As implantation.
3
Flash lamp annealing
Ion implantation
1 µm
Figure S1. Tilted SEM view of a Si NW sample before processing (left panel), after ion implantation (center panel) and after FLA (right panel). Si NWs were grown on a Si(111) substrate by Au-catalyzed VLS growth3 and 15 nm SiO2 was deposited by PECVD after etching of the Au catalyst. Room temperature ion implantation was done with samples mounted on a, with respect to the ion beam, 45° tilted and rotating sample stage. In and As was implanted alternatingly with an energy of 120 keV and 90 keV, respectively, and a fluence of 3 1016 ions/cm² each. Thereby a nominal concentration of approx. 7.5 at% In & As inside the Si NW core was achieved, according to simulations with the 3D Monte Carlo simulation tool iradina4. FLA was done in Ar atmosphere with a 3 min. preheating @600°C and a subsequent 20 ms flash with an energy density of 54.2 J/cm².
4
Figure S2. Cross section of an implanted NW: TEM investigations. (a) EFTEM images with plasmon scattering maps for SiO2 (left panel) and Si (right panel) of a Si/SiO2 core/shell NW after implantation of 1.904 1016 ions/cm² Ga and As. (b) Diffraction pattern of the whole NW cross section, exhibiting an amorphous material composition. The implanted NW was deposited on an SiO2 substrate and a TEM lamella was prepared by cutting a slice out of the NW center with an FEI Dual-Beam Ga FIB. Platinum was deposited on the sample surface prior to FIB cutting.
5
Supporting Information 2. Example of a NW heterostructure with thin crystallite sections.
Figure S3. NW heterostructure with thin crystallite sections. SEM image of a selected NW heterostructure with thin GaAs nanocrystallites formed after ion implantation and FLA. The NW was deposited on a Si substrate by drop casting. Prior to processing a 20 nm SiO2 shell was deposited by PECVD. Afterwards 1.904 1016 ions/cm² Ga and As were implanted and FLA was performed with a 20 ms flash at an energy density of 54.2 J/cm².
6
Supporting Information 3. Raman investigations. To circumvent any Raman scattering from the Si substrate and detect Raman signals from the NW only, a metal layer was deposited on the clean substrate by magnetron sputtering followed by an insulating 100 nm thick Al2O3 by layer atomic layer deposition. For Raman analysis of the Si/GaAs NW heterostructures, NWs were deposited on the substrate by drop casting of extracted NWs immersed in isopropanol. In a typical Raman spectrum as shown in Figure 1b of the main text, five distinct Raman features were identified as follows.
Peak I: The Raman peak at 272 cm-1 was assigned to the transverse optical (TO) phonon mode of GaAs. The slight shift with respect to bulk GaAs (268.6 cm-1)5 was ascribed to minor compressive strain of less than 0.4% along the NW axis6. Such compressive strain was observed for most of the GaAs nanocrystallites embedded in the Si NW heterostructure. The absence of the GaAs longitudinal optical (LO) phonon mode (291.9 cm-1)5 was due to an 111 crystal orientation along the nanowire axis, observed in TEM investigations (cf. Figure 1c in the main text), and the excitation laser's polarization parallel to the substrate surface. In this configuration, the TO phonon mode dominates due to the Raman selection rule in backscattering geometry7. The GaAs crystallites investigated are assumed to be zinc-blende phase only, as no wurtzite-related TO mode (254 cm-1)8 was observed.
Peak II: Next to the TO mode a prominent peak at about 287 cm-1 was identified as GaAs surface phonon mode. In Figure S4, a detailed view of the µ-Raman spectra taken at five positions along a conical GaAs crystallite is depicted, showing the GaAs TO as well as the surface phonon related mode. This surface phonon mode was dependent on the GaAs crystallite diameter and shifts to larger relative wavenumbers for increasing diameters9.
7
Figure S4. Verification of Raman surface phonon peak. µ-Raman spectra (left panel) taken at five positions along a GaAs crystallite as shown in the SEM image (right panel). The diameter of the GaAs nanocrystallite increases from the red arrow towards the yellow arrow resulting in a shift of the surface phonon mode related Raman peak. Gray lines in the left panel are a guide to the eye clarifying the constant position of the GaAs TO mode at about 270 cm -1 whereas the surface phonon related peak shifts from about 284 to 287 cm-1 for increasing crystallite diameters which is consistent to ref. 9.
Peak III: Important information about impurity sites in the crystal lattice can be obtained from investigating local vibrational modes (LVM). As Si is the main material present in the molten NW core during FLA, the resulting GaAs nanocrystallites were heavily doped with Si atoms occupying Ga and As lattice sites as well as forming compensating defect complexes. A Si atom, occupying a GaAs lattice site, is lighter than the atoms of the host lattice giving rise to spatially localized vibrational modes with frequencies higher than the ones of the GaAs modes10. For example, SiGa donors having four As neighbors give rise to a sharp LVM in the Raman spectrum at. 384 cm−1. The LVM of compensating SiAs acceptors is found at 399 cm−1. When both SiAs acceptors and SiGa donors 8
are present in high concentrations, nearest-neighbor SiGa–SiAs pairs may form with LVMs at 393 cm−1 and 464 cm−1. Other defects like for example the so called Si-X center at 369 cm−1 that is related to a SiAs–Ga vacancy complex may also be detected in the Raman spectrum. Thus we assigned the broad Raman feature around 390 cm-1 to LVM scattering at such defect sites what from a maximum Si impurity concentration of about 5 1018 cm-3 can be deduced11,12. Unfortunately, with the laser excitation energy used in our Raman experiments (2.33 eV) the LVM for SiGa was suppressed and no information about the doping type or level could be deduced from Raman experiments13. However, it is well known that GaAs epitaxially grown by LPE from a Si-rich melt at high temperatures leads to n-type GaAs14, which was also the case here, as temperatures of NW samples during FLA were estimated to be above15 1000°C.
Peak IV, V: When the excitation laser was focused on GaAs nanocrystallites, also Raman scattering from the adjacent Si core was detected because of the 1/e² laser spot size of approx. 610 nm. Thus, the Raman peak at 517.9 cm-1 (peak V) was assigned to the adjacent Si NW and was identified as the Brillouin zone center Raman peak of strained crystalline Si, slightly shifted compared to bulk Si (520.5 cm-1). In contrast to the GaAs nanocrystallites, the Si NWs appeared to be tensile strained (0.5–1%)16 along the NW axis. A small broad feature around 487.5 cm-1 (peak IV) was ascribed to nano-crystalline (nc) Si with diameters less than17 7 nm. In separate TEM investigations we observed these Si nanocrystallites within the SiO2 shell after FLA (see TEM image in Figure S5).
9
Supporting Information 4. Detailed elemental analysis of a Si/GaAs/Ga/Si NW heterostructure. An energy dispersive X-ray spectroscopy (EDX) mapping with the respective scanning TEM (STEM) image and thereof acquired material compositions of five designated areas are shown in Figure S5 for a NW heterostructure with an SiO2 shell. The similar Ga and As quantities inside the GaAs nanocrystallite (considering an EDX uncertainty of approx. 1 at%) account for a stoichiometric GaAs crystallite formation (area 1). For the Si NW (area 2, 5) a considerable amount of As was detected, which is in accordance with the n-type behavior due to As doping of the Si core, observed in electrical characterizations (see main text for details). Inside the SiO 2 shell (area 3) an increased amount of Si can be seen which is caused by nc-Si precipitations within the shell, as seen in the Raman spectrum (cf. Supporting Information 3). For the sample shown in Figure S5, an amorphous Ga section was found next to the GaAs nanocrystallite (area 4) which is typical for GaAs nanocrystallites with axial lengths of more than approx. 75 nm. Such metallic sections usually have an axial length of some tens of nm, up to 150 nm.
10
Figure S5. STEM EDX quantitative analysis of a Si NW heterostructure with a Ga/GaAs nanocrystallite. A composite EDX map for the elements As, Ga, Si and O in at% (top panel) and related STEM image (center panel), visualizing a detailed view of a GaAs/Ga nanocrystallite integrated in a Si NWs is shown. A 20 nm SiO2 shell is spanning the whole NW. Si, O, Ga and As quantities (with 1 at% uncertainty) acquired with EDX from the five regions marked in the STEM image are tabulated in the bottom panel.
11
Supporting Information 5. Example of a twin defect in the GaAs/Si NW heterostructure.
Figure S6. High resolution TEM image of a twin boundary (red arrows) observed across the interface of a Si/GaAs NW heterostructure, where the twin plane separates two different zincblende orientations of the Si core (see labeled {111} lattice planes). The Si crystal structure, including the twin plane, was inherited by the GaAs NC during NW core recrystallization after FLA.
12
Supporting information 6. Determining the doping type of Si NWs by thermoelectric measurements. To specify the doping type of the Si core in NW heterostructures, thermoelectric measurements were performed on single NWs as shown in Figure S7. For this, Si NWs without GaAs nanocrystallites from ion implanted and annealed NW samples were contacted (see Figure S7a). A separate heating structure next to one NW contact was utilized to induce a heat gradient along the NW due to Joule heating of this contact. Owing to the thermoelectric effect in semiconductors, majority charge carriers will then diffuse towards the cold contact resulting in a potential VT, known as the thermoelectric voltage. By sweeping the NW current IS, the I/V characteristic without heating yields the expected linear behavior as shown in Figure S7b for heating current IHEAT = 0 mA. As IHEAT is increased, the I/V characteristic is shifting to positive voltages due to a superimposed positive VT. This unambiguously proves the Si NW to be n-type as electrons, being the majority charge carriers, diffuse to the cold contact, causing the hot contact to be on a more positive potential than the cold one.
13
Figure S7. Thermoelectric measurements for determining the Si NW doping type. (a) SEM image with zoomed view of a contacted Si NW without GaAs nanocrystallites, deposited on an insulating substrate. A heating structure was placed next to the right-hand contact. (b) I/V characteristics obtained by a current sweep of the NW current IS for different heating currents IHEAT. A positive thermoelectric voltage VT at the heated contact was observed for elevated temperatures (IHEAT 0) and is obtained from I/V characteristics at IS = 0 mA. Thermoelectric measurements were performed on a Keithley 4200 semiconductor characterization system connected to a Summit 11000 AP probe station.
14
Supporting information 7. Barrier height simulations of observed Si/GaAs/Si heterostructures. Simulations of Si/GaAs/Si heterostructures without Ga nanocrystallites as observed for ion implanted and FLA Si NWs were performed for material parameters described in the main text. In Figure S8, the conduction band minimum as well as the Fermi energy of such a structure with a 124 nm long GaAs nanocrystallite is shown, yielding a barrier of qB = 67.8 meV for conduction band electrons. The low net n-type doping level of the GaAs nanocrystallite leads to large depletion zone lengths resulting in an overlap of the depletion zones inside the GaAs nanocrystallite which causes qB to be dependent on the GaAs nanocrystallite length. An increase of qB from approx. 47 meV to 80 meV for GaAs nanocrystallite lengths from 50 nm to 300 nm was observed in simulations.
Figure S8. Calculated barrier height of a Si/GaAs/Si heterostructure simulation. Conduction band minimum EC and Fermi energy EF for the Si/GaAs/Si heterostructure described in the main text are depicted. A barrier of qB = 67.8 meV was acquired for a GaAs nanocrystallite length of 124 nm. Simulations were performed with the semi-classical device simulator Minimos-NT from Global TCAD Solutions.
15
Supporting Information 8. Hole concentration and impact ionization simulations of a Ga/GaAs/Si heterostructure. To confirm the assumed hot carrier electroluminescence (EL) observed in NW heterostructures with Ga/GaAs nanocrystallites, simulations of hole concentrations and impact ionization rates were done for heterostructures as shown in Figure S9. Impact ionization generation rates could be obtained for the GaAs crystallite whereas no impact ionization is observed inside the Si NWs. By looking at the hole concentrations an obvious correlation between impact ionization and hole concentration can be seen for reverse biases. This is in agreement with the high band-to-band EL observed for NW heterostructures under reverse bias (cf. main text Figure 4). If high reverse biases are applied to the NW, electrons from the Ga segment can tunnel through the Schottky barrier and are accelerated into the GaAs (see main text Figure 3). Due to impact ionization, holes are generated resulting in hot carrier EL as electron/hole pairs recombine radiatively. Under forward bias, impact ionization is also present close to the Schottky barrier, but no hole concentration is obtained from simulations, hence no EL due to interband recombinations is possible in this case. Only a weaker luminescence due to Bremsstrahlung at high electric fields is observed as discussed in the main text. When investigating the spatial origin of EL by optical microscope, the luminescence center was found to be located at the Ga/GaAs interface for small biases. For higher reverse biases, EL was observed to expand further into the GaAs nanocrystallite, finally spanning nearly the whole crystallite. This is confirmed by the simulations shown in Figure S9, where impact ionization rates as well as hole concentrations inside the GaAs nanocrystallite toward the adjacent Si NW increased for larger reverse biases, hence raising probabilities for radiative recombinations.
16
Figure S9. Hole concentration and impact ionization in a Ga/GaAs/Si hetero-structure. Hole concentrations (left panel) and impact ionization generation rates (right panel) for a Ga/GaAs/Si NW heterostructure under different bias conditions are shown. Material parameters are similar to energy band simulations in the main text (cf. main text Figure 3). Positive and negative VSD values correspond to reverse and forward bias conditions, respectively. Simulations were performed with the semi-classical device simulator Minimos-NT from Global TCAD Solutions.
17
References (1)
Borschel, C.; Niepelt, R.; Geburt, S.; Gutsche, C.; Regolin, I.; Prost, W.; Tegude, F.; Stichtenoth, D.; Schwen, D.; Ronning, C. Small 2009, 5 (22), 2576–2580.
(2)
Johannes, A.; Noack, S.; Wesch, W.; Glaser, M.; Lugstein, A.; Ronning, C. Nano Lett. 2015, 15 (6), 3800–3807.
(3)
Wang, N.; Cai, Y.; Zhang, R. Q. Mater. Sci. Eng. R Rep. 2008, 60 (1-6), 1–51.
(4)
Borschel, C.; Ronning, C. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. At. 2011, 269 (19), 2133–2138.
(5)
In Group IV Elements, IV-IV and III-V Compounds. Part b - Electronic, Transport, Optical and Other Properties; Madelung, O., Rössler, U., Schulz, M., Eds.; Landolt-Börnstein Group III Condensed Matter; Springer-Verlag: Berlin/Heidelberg, 2002; Vol. 41A1b, pp 1–4.
(6)
Montazeri, M.; Fickenscher, M.; Smith, L. M.; Jackson, H. E.; Yarrison-Rice, J.; Kang, J. H.; Gao, Q.; Tan, H. H.; Jagadish, C.; Guo, Y.; Zou, J.; Pistol, M.-E.; Pryor, C. E. Nano Lett. 2010, 10 (3), 880–886.
(7)
Burns, G.; Dacol, F. H.; Wie, C. R.; Burstein, E.; Cardona, M. Solid State Commun. 1987, 62 (7), 449–454.
(8)
Zardo, I.; Conesa-Boj, S.; Peiro, F.; Morante, J. R.; Arbiol, J.; Uccelli, E.; Abstreiter, G.; Fontcuberta i Morral, A. Phys. Rev. B 2009, 80 (24), 245324.
(9)
Watt, M.; Sotomayor Torres, C. M.; Arnot, H. E. G.; Beaumont, S. P. Semicond. Sci. Technol. 1990, 5 (4), 285–290.
(10)
Newman, R. C. Semicond. Sci. Technol. 1994, 9 (10), 1749–1762.
(11)
Newman, R. C. Mater. Sci. Eng. B 1999, 66 (1–3), 39–45.
(12)
Ketterer, B.; Mikheev, E.; Uccelli, E.; Fontcuberta i Morral, A. Appl. Phys. Lett. 2010, 97 (22), 223103.
(13)
Ramsteiner, M.; Wagner, J.; Ennen, H.; Maier, M. Phys. Rev. B 1988, 38 (15), 10669– 10676.
(14)
Ahn, B. H.; Shurtz, R. R.; Trussell, C. W. J. Appl. Phys. 1971, 42 (11), 4512–4513.
(15)
Prucnal, S.; Glaser, M.; Lugstein, A.; Bertagnolli, E.; Stöger-Pollach, M.; Zhou, S.; Helm, M.; Reichel, D.; Rebohle, L.; Turek, M.; Zuk, J.; Skorupa, W. Nano Res. 2014, 7 (12), 1769–1776.
(16)
Ureña, F.; Olsen, S. H.; Raskin, J.-P. J. Appl. Phys. 2013, 114 (14), 144507.
(17)
Tanino, H.; Kawanami, H.; Matsuhata, H. Appl. Phys. Lett. 1992, 60 (16), 1978–1980.
18
†
Institute of Solid State Electronics, TU Wien, Floragasse 7, 1040 Wien, Austria
‡
Institute for Solid State Physics, Friedrich-Schiller-University Jena, Max-Wien-Platz 1, 07743 Jena, Germany
§
Institute of Ion Beam Physics and Materials Research, Helmholtz-Zentrum DresdenRossendorf, Bautzner Landstraße 400, 01328 Dresden, Germany
║
Laboratoire des Matériaux Semiconducteurs, Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland
┴
*
Institute for Microelectronics, TU Wien, Gußhausstraße 25-29, 1040 Wien, Austria
To whom correspondence should be addressed, E-mail: [email protected]
1
Contents Supporting Information 1. NW response on implantation and FLA. ..................................... 3 Supporting Information 2. Example of a NW heterostructure with thin crystallite sections................................................................................................. 6 Supporting Information 3. Raman investigations. ................................................................ 7 Supporting Information 4. Detailed elemental analysis of a Si/GaAs/Ga/Si NW heterostructure. .......................................................................... 10 Supporting information 5. Example of a twin defect in the GaAs/Si NW heterostructure. .. 12 Supporting information 6. Determining the doping type of Si NWs by thermoelectric measurements. .......................................................................................................... 13 Supporting information 7. Barrier height simulations of observed Si/GaAs/Si heterostructures. ..................................................................................... 15 Supporting Information 8. Hole concentration and impact ionization simulations of a Ga/GaAs/Si heterostructure............................................................. 16 References ........................................................................................................................ 18
2
Supporting Information 1. NW response on implantation and FLA. In Figure S1, tilted scanning electron microscope (SEM) images of particular Si nanowires (NWs) are shown, visualizing the effect of high fluence ion implantation and further flash lamp annealing (FLA). The left panel shows the detailed view of a group of as-grown Si NWs oriented perpendicular to the substrate after deposition of 15 nm SiO2 by plasma enhanced chemical vapor deposition (PECVD). In the center panel the same NW group is shown after room temperature In & As ion implantation with the sample mounted on a 45° tilted, rotating sample stage. Thereby NWs perpendicular to the substrate surface retain their orientation, whereas NWs with oblique growth directions or kinks are bent towards the ion beam and straightened1. Remarkably, all NWs are shortened considerably during ion implantation which is ascribed to plastic deformation under ion irradiation2. Finally, after FLA (right panel) no further change of geometry can be observed, however small sections of InAs have formed inside the NW core similar to the GaAs implanted and annealed samples described in the main text.
Figure S2 shows a cross sectional energy filtered TEM (EFTEM) image of the Si/SiO 2 core/shell NW after ion implantation and the respective diffraction pattern. The sample was prepared by focused ion beam (FIB) cutting immediately after ion implantation. Bright contrast correlates with occurrence of the investigated material. As can be seen, the about 20 nm thick SiO2 shell deposited on the VLS grown Si NW is still preserved after high fluence ion implantation. During implantation the NW cross section changed from hexagonal for pristine NWs 3 to circular due to reinforced sputtering at oblique edges. The diffraction pattern in Figure S2b clearly proves full amorphization of the Si NW due to high fluence Ga and As implantation.
3
Flash lamp annealing
Ion implantation
1 µm
Figure S1. Tilted SEM view of a Si NW sample before processing (left panel), after ion implantation (center panel) and after FLA (right panel). Si NWs were grown on a Si(111) substrate by Au-catalyzed VLS growth3 and 15 nm SiO2 was deposited by PECVD after etching of the Au catalyst. Room temperature ion implantation was done with samples mounted on a, with respect to the ion beam, 45° tilted and rotating sample stage. In and As was implanted alternatingly with an energy of 120 keV and 90 keV, respectively, and a fluence of 3 1016 ions/cm² each. Thereby a nominal concentration of approx. 7.5 at% In & As inside the Si NW core was achieved, according to simulations with the 3D Monte Carlo simulation tool iradina4. FLA was done in Ar atmosphere with a 3 min. preheating @600°C and a subsequent 20 ms flash with an energy density of 54.2 J/cm².
4
Figure S2. Cross section of an implanted NW: TEM investigations. (a) EFTEM images with plasmon scattering maps for SiO2 (left panel) and Si (right panel) of a Si/SiO2 core/shell NW after implantation of 1.904 1016 ions/cm² Ga and As. (b) Diffraction pattern of the whole NW cross section, exhibiting an amorphous material composition. The implanted NW was deposited on an SiO2 substrate and a TEM lamella was prepared by cutting a slice out of the NW center with an FEI Dual-Beam Ga FIB. Platinum was deposited on the sample surface prior to FIB cutting.
5
Supporting Information 2. Example of a NW heterostructure with thin crystallite sections.
Figure S3. NW heterostructure with thin crystallite sections. SEM image of a selected NW heterostructure with thin GaAs nanocrystallites formed after ion implantation and FLA. The NW was deposited on a Si substrate by drop casting. Prior to processing a 20 nm SiO2 shell was deposited by PECVD. Afterwards 1.904 1016 ions/cm² Ga and As were implanted and FLA was performed with a 20 ms flash at an energy density of 54.2 J/cm².
6
Supporting Information 3. Raman investigations. To circumvent any Raman scattering from the Si substrate and detect Raman signals from the NW only, a metal layer was deposited on the clean substrate by magnetron sputtering followed by an insulating 100 nm thick Al2O3 by layer atomic layer deposition. For Raman analysis of the Si/GaAs NW heterostructures, NWs were deposited on the substrate by drop casting of extracted NWs immersed in isopropanol. In a typical Raman spectrum as shown in Figure 1b of the main text, five distinct Raman features were identified as follows.
Peak I: The Raman peak at 272 cm-1 was assigned to the transverse optical (TO) phonon mode of GaAs. The slight shift with respect to bulk GaAs (268.6 cm-1)5 was ascribed to minor compressive strain of less than 0.4% along the NW axis6. Such compressive strain was observed for most of the GaAs nanocrystallites embedded in the Si NW heterostructure. The absence of the GaAs longitudinal optical (LO) phonon mode (291.9 cm-1)5 was due to an 111 crystal orientation along the nanowire axis, observed in TEM investigations (cf. Figure 1c in the main text), and the excitation laser's polarization parallel to the substrate surface. In this configuration, the TO phonon mode dominates due to the Raman selection rule in backscattering geometry7. The GaAs crystallites investigated are assumed to be zinc-blende phase only, as no wurtzite-related TO mode (254 cm-1)8 was observed.
Peak II: Next to the TO mode a prominent peak at about 287 cm-1 was identified as GaAs surface phonon mode. In Figure S4, a detailed view of the µ-Raman spectra taken at five positions along a conical GaAs crystallite is depicted, showing the GaAs TO as well as the surface phonon related mode. This surface phonon mode was dependent on the GaAs crystallite diameter and shifts to larger relative wavenumbers for increasing diameters9.
7
Figure S4. Verification of Raman surface phonon peak. µ-Raman spectra (left panel) taken at five positions along a GaAs crystallite as shown in the SEM image (right panel). The diameter of the GaAs nanocrystallite increases from the red arrow towards the yellow arrow resulting in a shift of the surface phonon mode related Raman peak. Gray lines in the left panel are a guide to the eye clarifying the constant position of the GaAs TO mode at about 270 cm -1 whereas the surface phonon related peak shifts from about 284 to 287 cm-1 for increasing crystallite diameters which is consistent to ref. 9.
Peak III: Important information about impurity sites in the crystal lattice can be obtained from investigating local vibrational modes (LVM). As Si is the main material present in the molten NW core during FLA, the resulting GaAs nanocrystallites were heavily doped with Si atoms occupying Ga and As lattice sites as well as forming compensating defect complexes. A Si atom, occupying a GaAs lattice site, is lighter than the atoms of the host lattice giving rise to spatially localized vibrational modes with frequencies higher than the ones of the GaAs modes10. For example, SiGa donors having four As neighbors give rise to a sharp LVM in the Raman spectrum at. 384 cm−1. The LVM of compensating SiAs acceptors is found at 399 cm−1. When both SiAs acceptors and SiGa donors 8
are present in high concentrations, nearest-neighbor SiGa–SiAs pairs may form with LVMs at 393 cm−1 and 464 cm−1. Other defects like for example the so called Si-X center at 369 cm−1 that is related to a SiAs–Ga vacancy complex may also be detected in the Raman spectrum. Thus we assigned the broad Raman feature around 390 cm-1 to LVM scattering at such defect sites what from a maximum Si impurity concentration of about 5 1018 cm-3 can be deduced11,12. Unfortunately, with the laser excitation energy used in our Raman experiments (2.33 eV) the LVM for SiGa was suppressed and no information about the doping type or level could be deduced from Raman experiments13. However, it is well known that GaAs epitaxially grown by LPE from a Si-rich melt at high temperatures leads to n-type GaAs14, which was also the case here, as temperatures of NW samples during FLA were estimated to be above15 1000°C.
Peak IV, V: When the excitation laser was focused on GaAs nanocrystallites, also Raman scattering from the adjacent Si core was detected because of the 1/e² laser spot size of approx. 610 nm. Thus, the Raman peak at 517.9 cm-1 (peak V) was assigned to the adjacent Si NW and was identified as the Brillouin zone center Raman peak of strained crystalline Si, slightly shifted compared to bulk Si (520.5 cm-1). In contrast to the GaAs nanocrystallites, the Si NWs appeared to be tensile strained (0.5–1%)16 along the NW axis. A small broad feature around 487.5 cm-1 (peak IV) was ascribed to nano-crystalline (nc) Si with diameters less than17 7 nm. In separate TEM investigations we observed these Si nanocrystallites within the SiO2 shell after FLA (see TEM image in Figure S5).
9
Supporting Information 4. Detailed elemental analysis of a Si/GaAs/Ga/Si NW heterostructure. An energy dispersive X-ray spectroscopy (EDX) mapping with the respective scanning TEM (STEM) image and thereof acquired material compositions of five designated areas are shown in Figure S5 for a NW heterostructure with an SiO2 shell. The similar Ga and As quantities inside the GaAs nanocrystallite (considering an EDX uncertainty of approx. 1 at%) account for a stoichiometric GaAs crystallite formation (area 1). For the Si NW (area 2, 5) a considerable amount of As was detected, which is in accordance with the n-type behavior due to As doping of the Si core, observed in electrical characterizations (see main text for details). Inside the SiO 2 shell (area 3) an increased amount of Si can be seen which is caused by nc-Si precipitations within the shell, as seen in the Raman spectrum (cf. Supporting Information 3). For the sample shown in Figure S5, an amorphous Ga section was found next to the GaAs nanocrystallite (area 4) which is typical for GaAs nanocrystallites with axial lengths of more than approx. 75 nm. Such metallic sections usually have an axial length of some tens of nm, up to 150 nm.
10
Figure S5. STEM EDX quantitative analysis of a Si NW heterostructure with a Ga/GaAs nanocrystallite. A composite EDX map for the elements As, Ga, Si and O in at% (top panel) and related STEM image (center panel), visualizing a detailed view of a GaAs/Ga nanocrystallite integrated in a Si NWs is shown. A 20 nm SiO2 shell is spanning the whole NW. Si, O, Ga and As quantities (with 1 at% uncertainty) acquired with EDX from the five regions marked in the STEM image are tabulated in the bottom panel.
11
Supporting Information 5. Example of a twin defect in the GaAs/Si NW heterostructure.
Figure S6. High resolution TEM image of a twin boundary (red arrows) observed across the interface of a Si/GaAs NW heterostructure, where the twin plane separates two different zincblende orientations of the Si core (see labeled {111} lattice planes). The Si crystal structure, including the twin plane, was inherited by the GaAs NC during NW core recrystallization after FLA.
12
Supporting information 6. Determining the doping type of Si NWs by thermoelectric measurements. To specify the doping type of the Si core in NW heterostructures, thermoelectric measurements were performed on single NWs as shown in Figure S7. For this, Si NWs without GaAs nanocrystallites from ion implanted and annealed NW samples were contacted (see Figure S7a). A separate heating structure next to one NW contact was utilized to induce a heat gradient along the NW due to Joule heating of this contact. Owing to the thermoelectric effect in semiconductors, majority charge carriers will then diffuse towards the cold contact resulting in a potential VT, known as the thermoelectric voltage. By sweeping the NW current IS, the I/V characteristic without heating yields the expected linear behavior as shown in Figure S7b for heating current IHEAT = 0 mA. As IHEAT is increased, the I/V characteristic is shifting to positive voltages due to a superimposed positive VT. This unambiguously proves the Si NW to be n-type as electrons, being the majority charge carriers, diffuse to the cold contact, causing the hot contact to be on a more positive potential than the cold one.
13
Figure S7. Thermoelectric measurements for determining the Si NW doping type. (a) SEM image with zoomed view of a contacted Si NW without GaAs nanocrystallites, deposited on an insulating substrate. A heating structure was placed next to the right-hand contact. (b) I/V characteristics obtained by a current sweep of the NW current IS for different heating currents IHEAT. A positive thermoelectric voltage VT at the heated contact was observed for elevated temperatures (IHEAT 0) and is obtained from I/V characteristics at IS = 0 mA. Thermoelectric measurements were performed on a Keithley 4200 semiconductor characterization system connected to a Summit 11000 AP probe station.
14
Supporting information 7. Barrier height simulations of observed Si/GaAs/Si heterostructures. Simulations of Si/GaAs/Si heterostructures without Ga nanocrystallites as observed for ion implanted and FLA Si NWs were performed for material parameters described in the main text. In Figure S8, the conduction band minimum as well as the Fermi energy of such a structure with a 124 nm long GaAs nanocrystallite is shown, yielding a barrier of qB = 67.8 meV for conduction band electrons. The low net n-type doping level of the GaAs nanocrystallite leads to large depletion zone lengths resulting in an overlap of the depletion zones inside the GaAs nanocrystallite which causes qB to be dependent on the GaAs nanocrystallite length. An increase of qB from approx. 47 meV to 80 meV for GaAs nanocrystallite lengths from 50 nm to 300 nm was observed in simulations.
Figure S8. Calculated barrier height of a Si/GaAs/Si heterostructure simulation. Conduction band minimum EC and Fermi energy EF for the Si/GaAs/Si heterostructure described in the main text are depicted. A barrier of qB = 67.8 meV was acquired for a GaAs nanocrystallite length of 124 nm. Simulations were performed with the semi-classical device simulator Minimos-NT from Global TCAD Solutions.
15
Supporting Information 8. Hole concentration and impact ionization simulations of a Ga/GaAs/Si heterostructure. To confirm the assumed hot carrier electroluminescence (EL) observed in NW heterostructures with Ga/GaAs nanocrystallites, simulations of hole concentrations and impact ionization rates were done for heterostructures as shown in Figure S9. Impact ionization generation rates could be obtained for the GaAs crystallite whereas no impact ionization is observed inside the Si NWs. By looking at the hole concentrations an obvious correlation between impact ionization and hole concentration can be seen for reverse biases. This is in agreement with the high band-to-band EL observed for NW heterostructures under reverse bias (cf. main text Figure 4). If high reverse biases are applied to the NW, electrons from the Ga segment can tunnel through the Schottky barrier and are accelerated into the GaAs (see main text Figure 3). Due to impact ionization, holes are generated resulting in hot carrier EL as electron/hole pairs recombine radiatively. Under forward bias, impact ionization is also present close to the Schottky barrier, but no hole concentration is obtained from simulations, hence no EL due to interband recombinations is possible in this case. Only a weaker luminescence due to Bremsstrahlung at high electric fields is observed as discussed in the main text. When investigating the spatial origin of EL by optical microscope, the luminescence center was found to be located at the Ga/GaAs interface for small biases. For higher reverse biases, EL was observed to expand further into the GaAs nanocrystallite, finally spanning nearly the whole crystallite. This is confirmed by the simulations shown in Figure S9, where impact ionization rates as well as hole concentrations inside the GaAs nanocrystallite toward the adjacent Si NW increased for larger reverse biases, hence raising probabilities for radiative recombinations.
16
Figure S9. Hole concentration and impact ionization in a Ga/GaAs/Si hetero-structure. Hole concentrations (left panel) and impact ionization generation rates (right panel) for a Ga/GaAs/Si NW heterostructure under different bias conditions are shown. Material parameters are similar to energy band simulations in the main text (cf. main text Figure 3). Positive and negative VSD values correspond to reverse and forward bias conditions, respectively. Simulations were performed with the semi-classical device simulator Minimos-NT from Global TCAD Solutions.
17
References (1)
Borschel, C.; Niepelt, R.; Geburt, S.; Gutsche, C.; Regolin, I.; Prost, W.; Tegude, F.; Stichtenoth, D.; Schwen, D.; Ronning, C. Small 2009, 5 (22), 2576–2580.
(2)
Johannes, A.; Noack, S.; Wesch, W.; Glaser, M.; Lugstein, A.; Ronning, C. Nano Lett. 2015, 15 (6), 3800–3807.
(3)
Wang, N.; Cai, Y.; Zhang, R. Q. Mater. Sci. Eng. R Rep. 2008, 60 (1-6), 1–51.
(4)
Borschel, C.; Ronning, C. Nucl. Instrum. Methods Phys. Res. Sect. B Beam Interact. Mater. At. 2011, 269 (19), 2133–2138.
(5)
In Group IV Elements, IV-IV and III-V Compounds. Part b - Electronic, Transport, Optical and Other Properties; Madelung, O., Rössler, U., Schulz, M., Eds.; Landolt-Börnstein Group III Condensed Matter; Springer-Verlag: Berlin/Heidelberg, 2002; Vol. 41A1b, pp 1–4.
(6)
Montazeri, M.; Fickenscher, M.; Smith, L. M.; Jackson, H. E.; Yarrison-Rice, J.; Kang, J. H.; Gao, Q.; Tan, H. H.; Jagadish, C.; Guo, Y.; Zou, J.; Pistol, M.-E.; Pryor, C. E. Nano Lett. 2010, 10 (3), 880–886.
(7)
Burns, G.; Dacol, F. H.; Wie, C. R.; Burstein, E.; Cardona, M. Solid State Commun. 1987, 62 (7), 449–454.
(8)
Zardo, I.; Conesa-Boj, S.; Peiro, F.; Morante, J. R.; Arbiol, J.; Uccelli, E.; Abstreiter, G.; Fontcuberta i Morral, A. Phys. Rev. B 2009, 80 (24), 245324.
(9)
Watt, M.; Sotomayor Torres, C. M.; Arnot, H. E. G.; Beaumont, S. P. Semicond. Sci. Technol. 1990, 5 (4), 285–290.
(10)
Newman, R. C. Semicond. Sci. Technol. 1994, 9 (10), 1749–1762.
(11)
Newman, R. C. Mater. Sci. Eng. B 1999, 66 (1–3), 39–45.
(12)
Ketterer, B.; Mikheev, E.; Uccelli, E.; Fontcuberta i Morral, A. Appl. Phys. Lett. 2010, 97 (22), 223103.
(13)
Ramsteiner, M.; Wagner, J.; Ennen, H.; Maier, M. Phys. Rev. B 1988, 38 (15), 10669– 10676.
(14)
Ahn, B. H.; Shurtz, R. R.; Trussell, C. W. J. Appl. Phys. 1971, 42 (11), 4512–4513.
(15)
Prucnal, S.; Glaser, M.; Lugstein, A.; Bertagnolli, E.; Stöger-Pollach, M.; Zhou, S.; Helm, M.; Reichel, D.; Rebohle, L.; Turek, M.; Zuk, J.; Skorupa, W. Nano Res. 2014, 7 (12), 1769–1776.
(16)
Ureña, F.; Olsen, S. H.; Raskin, J.-P. J. Appl. Phys. 2013, 114 (14), 144507.
(17)
Tanino, H.; Kawanami, H.; Matsuhata, H. Appl. Phys. Lett. 1992, 60 (16), 1978–1980.
18
