Polarity Control of Heteroepitaxial GaN Nanowires on Diamond

May 23, 2017 - Group III-nitride materials such as GaN nanowires are characterized by a spontaneous polarization within the crystal. The sign of the r...
1 downloads 12 Views 9MB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Communication

Polarity Control of Heteroepitaxial GaN Nanowires on Diamond Martin Hetzl, Max Kraut, Theresa Hoffmann, and Martin Stutzmann Nano Lett., Just Accepted Manuscript • Publication Date (Web): 23 May 2017 Downloaded from http://pubs.acs.org on May 25, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Nano Letters is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Polarity Control of Heteroepitaxial GaN Nanowires on Diamond Martin Hetzl, ∗ Max Kraut, Theresa Homann, and Martin Stutzmann



Walter Schottky Institut and Physics Department, Technische Universität München, 85748 Garching, Germany

E-mail: [email protected]; [email protected]

Abstract Group III-nitride materials such as GaN nanowires are characterized by a spontaneous polarization within the crystal. The sign of the resulting sheet charge at the top and bottom facet of a GaN nanowire is determined by the orientation of the wurtzite bilayer of the dierent atomic species, called N and Ga polarity. We investigate the polarity distribution of heteroepitaxial GaN nanowires on dierent substrates and demonstrate polarity control of GaN nanowires on diamond. Kelvin Probe Force Microscopy is used to determine the polarity of individual selective area-grown and self-assembled nanowires over a large scale. At standard growth conditions, mixed polarity occurs for selective GaN nanowires on various substrates, namely on silicon, on sapphire and on diamond. In order to obtain control over the growth orientation on diamond, the substrate surface is modied by nitrogen and oxygen plasma exposure prior to growth and the growth parameters are adjusted simultaneously. We nd that the surface chemistry and the substrate temperature are the decisive factors to obtain control of up to 93% for both polarity types, whereas the growth mode, namely selective area or self-assembled growth, does not inuence the polarity distribution signicantly. The experimental results are discussed by a model based on the interfacial bonds between 1

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the GaN nanowires, the termination layer and the substrate.

Keywords GaN, nanowires, polarity, Kelvin Probe Force Microscopy, heteroepitaxy, diamond

Introduction An important parameter for free-standing GaN nanowires (NWs) is the orientation of the wurtzite crystal. GaN NWs preferentially grow along their polar c-axis, which results in N or Ga polarity, depending on the top atom in the wurtzite bilayer. This inuences the interface charge at the top and the bottom facet of the NW and, thus, the band alignment of the heterointerface, which is of fundamental importance, e.g., in NW-based high-electron mobility transistors, tandem solar cells or heterodiodes. 14 In the case of homoepitaxial NW growth, i.e. growth on an intermediate group III-nitride buer layer between the NWs and the substrate, the polarity can be controlled since the NWs adopt the orientation of the polar substrate beneath. 5,6 However, for heteroepitaxy on non-polar substrates, such as sapphire, Si or diamond, the polarity assignments in the literature often are inconsistent with each other or not comparable due to dierent deposition techniques, growth parameters or growth modes. 7 Thus, the polarity characteristics of GaN NWs and the parameters inuencing it are still under debate. The main challenges for investigating the polarity of GaN NWs are: (i) The controlled growth of NWs directly on non-polar substrates and (ii) the determination of the polarity for a statistically signicant number of NWs. An important requirement to address individual NWs is a uniform growth environment with a large enough distance of the NWs to each other. This can be achieved by the so-called selective area growth (SAG), 8,9 which has recently been established in our group also for heteroepitaxy on various substrates, namely on Si, 2

ACS Paragon Plus Environment

Page 2 of 27

Page 3 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

on diamond and on sapphire, without the use of a buer layer. 1012 In order to measure the polarity of a single NW, transmission electron diraction techniques can be used to directly assess the atomic ordering of the GaN bilayers. 10,1315 These techniques are well-suited to unambiguously determine the polarity in a small area of a crystal, but are not practical for statistical investigations. Alternatively, the polarity-specic macroscopic properties of a NW crystal can be used to determine its polarity. In particular, wet chemical etching and electrostatic potential measurements, e.g., via Kelvin Probe Force Microscopy (KPFM) have turned out to be suitable methods to address NWs over a large scale and quantity. 1619 In this letter, we present the large scale determination of the polarity of heteroepitaxial GaN NWs on dierent substrates by wet chemical etching and KPFM. Moreover, the inuence of the GaN NW growth parameters and the substrate pretreatment are investigated in the case of diamond (111) substrates. Due to the high thermal, physical and chemical stability of diamond and the atomically sharp interface to GaN, this substrate is best suited for studies of the GaN nucleation process. 11,15 In addition, the experimental results are discussed by means of a nucleation model based on the surface chemistry of the substrate. By intentionally manipulating the surface environment of the substrate and adjusting the growth parameters, this allows us to control the polarity up to 93% of the GaN NWs in both directions.

Experimental The GaN NWs in this work have been grown by plasma-assisted molecular beam epitaxy (MBE) in the SAG and the self-assembled growth modes. In contrast to the statistical and random nucleation behavior of the self-assembled NW growth, SAG allows the control of the position of individual NWs on numerous substrates, e.g. on Si (111), diamond (111), c-plane sapphire and GaN templates. To achieve SAG, a Ti lm with a thickness of 10 nm has been

3

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

evaporated on the respective substrate which subsequently has been nano-structured by ebeam lithography with arrays of holes exposing the plain substrate surface. By adjusting the hole size and distance, dierent NW diameters and periods can be implemented. For this work, all SAG GaN NWs have been fabricated with periods of 300 and 500 nm. A detailed description of the nano-mask fabrication process can be found in previous publications. 10,11 In the case of the diamond substrates, O termination prior to Ti evaporation and again prior to NW growth has been performed via an RF O plasma at 200 W for 300 s to achieve a better sticking of the Ti lm and to guarantee a uniform surface termination of the diamond during GaN NW growth. In the case of SAG, all substrates have been exposed to N plasma within the MBE chamber at a substrate temperature of T sub =400◦ C for 10 min and at Tsub =800◦ C for 5 min to convert the Ti mask into thermally more stable TiN. 9,10,20 In order to investigate the inuence of the substrate species, namely diamond, Si, sapphire and GaN templates, on the GaN NW polarity, SAG GaN NWs have been grown at optimum growth conditions, i.e. in the N-rich growth regime with a N ux of 0.363 sccm and a power of 425 W of the RF plasma source (Oxford Applied Research HD25 ion source). According to GaN reference layers under metal-rich growth conditions, this corresponds to an equivalent growth rate of 18.3 nm/min. The Ga ux has been supplied by a Knudsen eusion cell and has been set to 4.5·10−7 mbar beam equivalent pressure (BEP), measured with a BayertAlpert ion gauge in the direct vicinity of the sample surface. A layer-equivalent growth rate of 5.6 nm/min has been obtained for this Ga ux under N-rich conditions. However, it must be noted that the elevated T sub for NW growth compared to layer growth leads to an increased desorption rate of Ga atoms on the substrate surface. Thus, the given Ga uxes are expected to overestimate the real values. All GaN NW samples have been intentionally n-type doped with a concentration of ≈ 5·1018 cm−3 Si to improve electric measurements. Note that the doping concentration refers to secondary ion mass spectrometry measurements on Si-doped self-assembled NWs on diamond. 21 Depending on the substrate used, dierent Tsub between 850 and 990 ◦ C have been applied to achieve optimum NW growth. All values

4

ACS Paragon Plus Environment

Page 4 of 27

Page 5 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

associated with Tsub refer to thermocouple measurements at the sample heater. Important to mention is that the NWs are directly grown on the respective substrates, with no intermediate buer layer within the mask holes. In order to investigate the inuence of the growth modes, the growth parameters and the surface pretreatment on the polarity distribution of GaN NWs on a specic substrate, diamond (111) substrates have been used. The corresponding sample parameters including surface pretreatment of the diamond (111) substrates are listed in Table 1. Table 1: Growth parameters and pretreatment of the GaN NW samples grown on diamond (111) substrates. The self-assembled growth mode is denoted as "self-ass.". Ga and N uxes are given in terms of layer-equivalent growth rates for N- and Ga-rich growth conditions, respectively. Sample

growth mode

1 2 3 4 5 6 7 8 9 10 11

SAG SAG SAG SAG SAG self-ass. self-ass. self-ass. SAG SAG SAG

Tsub [◦ C] 870 890 900 910 920 920 920 880 920 920 920

Ga ux [nm/min] 5.6 5.6 5.6 5.6 5.6 3.5 3.5 1.2 5.6 5.6 9.3

N ux [nm/min] 18.3 18.3 18.3 18.3 18.3 25.2 25.2 25.2 18.3 36.6 18.3

last substrate plasma treatment N N N N N O N N O N N

To clarify the inuence of T sub on the polarity of individual GaN NWs, ve samples with equivalent III/V ux ratio but dierent T sub between 870 and 920 ◦ C have been grown, including a N plasma exposure at lower T sub equivalent to the procedure mentioned above (Tab. 1, Samples 1-5). In order to demonstrate reproducibility, two samples have been fabricated twice, at 870 ◦ C and at 910◦ C (Tab. 1, Samples 1, 4). To elucidate the inuence of the growth mode, self-assembled GaN NWs have been grown at standard growth conditions without and with prior N exposure at lower T sub (Tab. 1, Samples 6, 7). Another N plasmaexposed self-assembled NW sample has been fabricated for a lower T sub =880◦ C and reduced Ga ux (Tab. 1, Sample 8). Another SAG sample has been fabricated with an O-terminated 5

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

diamond surface (Tab. 1, Sample 9). For that, the sample has been rst nitridated within the MBE chamber to stabilize the Ti mask and subsequently transferred out. Then, another O plasma step has been performed to destabilize the expected N termination of the diamond and to restore O termination. After that, the sample has been transferred back into the MBE system and SAG GaN NWs have been grown at T sub =920◦ C without further pretreatment. In addition, the III/V ux ratio has been varied on two SAG samples, with increased N and Ga ux, respectively (Tab. 1, Samples 10, 11).

The polarity of the GaN NW crystals has been measured by two dierent approaches: wet chemical etching and surface potential measurements. Polarity-selective wet chemical etching has been performed with 5-molar potassium hydroxide (KOH) solution for 15' at room temperature. 22 The corresponding scanning electron microscopy (SEM) images have been performed in 45 ◦ -tilted view at an acceleration voltage of 5 kV. Potential measurements have been performed by a Bruker MultiMode 8 atomic force microscope (AFM) by means of the PeakForce frequency-modulated KPFM technique in a dual pass mode. 23 The AFM probe (Bruker "PFQNE-AL") consists of a SiN cantilever and a Si tip with a triangular shape and a spring constant of 0.8 N/m. All measurements have been performed in the dark and under ambient conditions. First, a one-dimensional topography scan is measured in PeakForce mode, i.e. the cantilever is oscillated at 40 kHz well below its resonance frequency by ramping the z-stage upwards and downwards while performing a line scan in lateral direction. Next, the same line scan is repeated with the tip lifted at a distance of 45 nm with respect to the sample surface and oscillated at resonance frequency

ω (typically 100-300 kHz). ω is determined by both the spring constant of the cantilever and the electric force gradient F el ' arising from the contact potential dierence V CP D between the tip and the sample, i.e. the dierence between the respective work functions. By applying an AC-voltage with lower frequency of ωAC =2 kHz a frequency modulation occurs because of the Fel ' contribution, with specic sidebands at ω ± ωAC and ω ± 2ωAC . 23 The amplitude

6

ACS Paragon Plus Environment

Page 6 of 27

Page 7 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

of the ω ± ωAC branch is determined by a term proportional to (V DC -VCP D ), with VDC an externally applied voltage between the tip and the sample. Thus, by using the ω ± ωAC modulation band as feedback for KPFM, a suppression of the signal by adjusting V DC results in the desired VCP D . Note that no additional tip bias has been applied during measurements and that the sample is electrically grounded in this setup. To obtain a two-dimensional KPFM map of the sample surface, the setup is performing a set of line scans with a distinct oset to each other. In the case of GaN, the CPD signal shows a strong polarity dependence and therefore can be used for the local polarity assignment of individual NWs. This is further discussed below.

Results and discussion Selective GaN nanowire growth on dierent substrates

Figure 1: Tilted view SEM images (45 ◦ ) of SAG GaN NWs grown on (a) diamond (111), (b) Si (111), (c) c-plane sapphire and (d) Ga-polar GaN thin lm substrates. In Figure 1, 45◦ -tilted view SEM images of SAG GaN NWs grown for 90 min on dierent substrates are displayed. The NWs have been grown heteroepitaxially on the respective substrates at optimum growth parameters, i.e. at an equivalent Ga and N ux (5.6 and 7

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

18.3 nm/min layer-equivalent growth rates) and a substrate-specic T sub , namely 900◦ C on diamond (111), 850 ◦ C on Si (111), 990◦ C on sapphire and 910 ◦ C on Ga-polar GaN thin lms. In the case of the diamond substrates (Fig. 1a), the SAG GaN NWs are homogeneous in height, diameter and orientation with almost no parasitic nucleation on the mask surface. However, the NW diameter of ≈110 nm exceeds the hole diameter of the mask by 60 nm, which indicates a signicant lateral growth rate for SAG GaN NWs on diamond. 10,11 For SAG GaN NWs on Si (Fig. 1b), both the height and the diameter are similar for all NWs. However, some NWs show a tilt of their growth direction with respect to the substrate normal. This has been attributed to an insucient epitaxial relationship of the GaN NWs and the Si substrate due to an intermediate amorphous Si x Ny layer of 3-5 nm thickness arising from the N plasma exposure at lower T sub . 24,25 Moreover, lateral NW growth is strongly suppressed for SAG GaN NWs on Si compared to NWs on diamond. On c-plane sapphire the SAG GaN NWs exhibit a similar shape as NWs on Si (111). In addition, the NWs on sapphire are highlighted by a uniform orientation parallel to the substrate normal, indicating a well-dened epitaxial relationship with respect to the substrate. However, parasitic nucleation on the mask could not be completely suppressed. As a reference sample, SAG GaN NWs have been grown homoepitaxially on a Ga-polar GaN thin lm substrate (Fig. 1d). These NWs are completely uniform in shape and orientation, with similar diameters as on diamond (Fig. 1a). Consequently, a signicant lateral growth rate occurs also in this case.

GaN nanowire polarity determination In order to determine the polarity of SAG GaN NWs, wet chemical etching by KOH has been performed. As known for GaN layers, N-polar facets as well as the m-planes can be etched by KOH solution, whereas Ga-polar GaN stays unaected. 26,27 For this study, a SAG GaN NW sample with a T sub =890◦ C has been chosen (Tab. 1, Sample 2). In Figures 2a and b, 45 ◦ -tilted view SEM images of SAG GaN NWs on diamond before 8

ACS Paragon Plus Environment

Page 8 of 27

Page 9 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Figure 2: Tilted view SEM images (45 ◦ ) of SAG GaN NWs grown on diamond (111) (a) before and (b,c) after 15' KOH exposure. The assigned polarities of the respective NWs is inserted in (b) as blue 'N' for N-polar and red 'Ga' for Ga-polar NWs. NWs assigned to multiple polarities are marked with a green ellipse in (c). and after 15' KOH bath are displayed, respectively. Note that the same NWs are shown in both gures. Two dierent etching behaviors have been identied: (i) NWs which show a distinct decrease in height and diameter (Fig. 2b, blue 'N') and (ii) NWs which preserve their height (Fig. 2b, red 'Ga'). In the latter case, a conical shape with a decrease of the diameter towards the NW bottom has been observed for some of these NWs. The appearance of two types of etched NWs can be explained as follows: For N-polar GaN NWs, etching is only possible at the top facet and, with a reduced etch rate, at the sidewalls (m-planes). Thus, a distinct decrease of the NW height can be expected. 22 However, for Ga-polar NWs, the situation is more complex: The SAG GaN NWs on diamond are signicantly broadened in their diameter compared to, e.g., NWs on Si substrate (Figs. 1a and b). This lateral growth extends also over the mask hole, where nucleation initially has started. 10 However, in most cases, the GaN laterally overgrown over the TiN mask is not in direct contact to the mask, leaving a small gap of several nanometers between the GaN NW and the mask. A corresponding scanning transmission electron microscopy image showing this gap can be found in Ref. 10. For Ga polarity, the N-polar facets at the bottom of the NWs are locally revealed within the lateral overgrowth region and are exposed to the etching solution. Thus, a signicant etch rate of the outer GaN NW regions can occur from the bottom to the top,

9

ACS Paragon Plus Environment

Nano Letters

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

which results in the observed conical shape. Consequently, after KOH exposure, NWs with a distinctively reduced height can be assigned to N polarity, whereas NWs with an almost unaltered height and a cone shape are Ga-polar. Note that the tilted top facets of some NWs (Fig. 2a) are a result of coalescence of multiple nucleated crystals in the same mask hole. These crystals can have either the same or opposing polarity. In the latter case, this leads to an asymmetrical etching of these NWs (Fig. 2b, right NW and Fig. 2c, green ellipses). In order to obtain a statistically signicant distribution of the polarity of the GaN NWs on this sample, a large quantity of around 200 NWs has been investigated by SEM. An exemplary image of a dense NW array (300 nm period) is shown in Figure 2c. As a result, 16% of the NWs show a distinct reduction of their height and diameter, or are completely etched away. According to the considerations above, this NW fraction can be assigned either to N polarity or multiple polarity with a predominant fraction of N-polar crystallites, whereas 84% of the NWs reveal Ga polarity. To conclude, polarity-selective etching is a simple and fast technique to obtain the polarity distribution of GaN NWs. However, due to uncertainties associated with the wet etching process and multiple polarities within some NWs, large error bars have to be assumed for this method. In addition, this technique causes irreversible damage to the sample.

As an alternative, non-invasive method to evaluate the polarity distribution of individual GaN NWs, KPFM measurements have been performed. 19,28,29 Minj et al. have shown for self-assembled GaN NWs grown on an AlN buer layer that KPFM is a suitable technique to determine the polarity of NWs. 19 In particular, a characteristic variation of V CP D for dierent NWs has been assigned to the existence of dierent GaN polarities. In addition, they have demonstrated by spectroscopic imaging that the opposing sheet charges at the top facet of the NW for dierent GaN polarities modify the local band bending at the surface and are, thus, responsible for the polarity-sensitive V CP D . 7,19 However, they have also mentioned that the absolute and the relative values for N- and Ga-polar GaN vary strongly in

10

ACS Paragon Plus Environment

Page 10 of 27

Page 11 of 27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

the literature, even in their sign. 7,19 In order to reliably assign the dierence in V CP D to the particular polarity type for this work, planar n-type doped GaN hydride vapor phase epitaxy (HVPE) samples, purchased from Kyma Technologies, with known crystal orientation and similar doping concentrations as the GaN NWs have been measured by KPFM from both sides. We nd that the N-polar side always shows a lower V CP D than the Ga-polar side, i.e. V CP D,Ga >VCP D,N , whereas the absolute values of V CP D as well as the relative potential dierence ∆VCP D =VCP D,Ga -VCP D,N between Ga and N polarity can vary between +200 and +700 mV for dierent AFM tips and measurement cycles, e.g. due to humidity variations. Consequently, in a direct comparison of NWs with dierent CPD signal, a higher V CP D corresponds to Ga polarity for the present study. Worth mentioning is that the tip-to-sample distance during CPD measurements has a strong impact on the absolute V CP D values. 19,30 For all measurements, this distance has been kept constant at 45 nm. To reduce artifacts during the measurements, the NWs have been grown shorter for this study (45 min), resulting in a NW height