Nanowires possess unique strain relieving properties making them compatible with a wide variety of substrates ranging from single crystalline semiconductors, amorphous ceramics, and polycrystalline metals. Flexible metallic foils are particularly interesting substrates for nanowires for both flexible optoelectronics and high throughput manufacturing techniques. However, nanowires grown on polycrystalline metals exhibit grain-dependent morphologies. As an alternative route, the authors demonstrate the growth of highly uniform III-Nitride nanowires on bulk metallic glass (amorphous metal) and nanocrystalline Pt metal films using molecular beam epitaxy. Nanowire arrays on metallic glass substrates show uniformity over length scales >100 μm. The quality of these nanowires is explored by photoluminescence spectroscopy. The electrical characteristics of individual nanowires are measured via conductive atomic force microscopy, and mesoscale light-emitting diodes (LEDs) are fabricated. Nanowires grown on nanocrystalline Pt films showed an increase in output power by a factor of up to 32, and an increase in the overall LED efficiency by up to 13× compared with simultaneously grown nanowire LEDs on bare Si.

Nitride-based nanowire light emitting diode (LED) are being developed for a wide variety of applications, including LEDs and lasers.1–6 Of particular interest is the AlGaN system for ultraviolet (UV) applications, including germicidal disinfection, epoxy curing, and photolithography. Pushing deeper into the UV requires high Al%; however, typical thin-film LEDs suffer from strain-related efficiency reduction associated with increased Al incorporation, such as formation of dislocations and a switch from TE to TM polarization.7,8 In contrast to their thin-film counterparts, nanowires efficiently strain relax without the creation of extended defects9–11 enabling their growth on a wide variety of substrates. Nanowire devices are typically grown on Si wafers. Due to the inherently confined structure, the large differences in index of refraction between the nanowire and surrounding material (air, spin-on glass, etc.) provide large gains in the light extraction efficiency and even lead to lasing.12–15 AlGaN still presents difficulties in nanowires, specifically with p-type conduction due to high Mg activation energy.16–18 Polarization doping aids p-type doping of high Al% AlGaN.19–21 However, higher Al content makes p-type contacts difficult, partly because of worsening Mg activation, but also because of the high electron affinity and widening bandgap of AlGaN.22–26 While thick GaN layers enable more efficient electrical injection into the heterostructure, especially above the active region, they absorb deep UV light resulting in efficiency loss. The active region should be clad in high Al composition AlGaN; thus, the bottom (top) of the nanowires should be graded from GaN to AlN (AlN to GaN) which would yield p-type (n-type) polarization doping in primarily N-face nanowires.27 If the p-type side of the junction makes contact to the p-Si substrate, there is a ∼2 eV valence band offset at the p-Si/p-GaN interface, which greatly inhibits hole injection. This can be averted through the integration of an InGaN tunnel junction (TJ) in these nanowires,28 increasing device performance by 3.5×. However, the introduction of an InGaN layer, and complex TJ heterostructure, poses problems with nanowire uniformity which eventually limit device performance.29–32 

An exciting recent development might circumvent the challenges described above, namely, the demonstration of high quality growth of nitride nanowires and LEDs on metal films that act as improved back contacts and heat sinks.33–36 More recently, nanowires and LEDs were grown on flexible bulk metal foils.37–40 However, large variations in nanowire densities and growth axis orientation were observed. Although the intragrain uniformity of nanowire arrays is acceptable, the intergrain uniformity is poor. The polycrystalline nature of these foils presents very large discontinuities in nanowire characteristics at grain boundaries and between grains of the underlying foil. Here, we explore the impact of substrate microstructure (or lack thereof) by growing nanowire arrays on amorphous metal substrates and nanocrystalline thin films.

If the orientation of the nanowires depends on the underlying grain and the microstructure is much larger than the diameter of the nanowires, then one observes preferential orientation of the nanowires [Fig. 1(a)]. If the grain size is much smaller than the diameter of the nanowire [Fig. 1(b)], then the orientation preference due to the underlying microstructure disappears and the nanowires begin to grow along the c-axis, due to low c-plane surface energy.41 This was previously observed for nanowires grown on cold e-beam deposited metals, but delamination can occur due to the large thermal cycles occurring during the growth of deep ultraviolet (DUV) nanowire LEDs that require high substrate temperatures.42 As the grain size of the substrate approaches zero, as in the case of amorphous metals (metallic glasses), the microstructure disappears and nanowire uniformity should be restored [Fig. 1(c)]. Previously, GaN nanowires were shown to grow preferentially along the [0001¯] axis on SiOx,43 but insulating substrates are not desired for electronic devices, and silicates are not flexible. Here, we explore the use of nanocrystalline Pt films and bulk metallic glass for improving the uniformity of nanowire ensembles grown on metals. Additionally, we demonstrate the improvement of DUV LED performance using a Pt/p-GaN bottom contact.

Fig. 1.

Schematics of nanowires grown (a) on polycrystalline materials with grains much larger than nanowire diameter and (b) grains smaller than nanowire diameter. (c) Nanowire arrays on amorphous (glass) substrates.

Fig. 1.

Schematics of nanowires grown (a) on polycrystalline materials with grains much larger than nanowire diameter and (b) grains smaller than nanowire diameter. (c) Nanowire arrays on amorphous (glass) substrates.

Close modal

Thin (<5 nm) Pt films on p-Si are first examined to determine their impact on electrical injection into p-GaN. The work function of Pt is ∼5.65–6.1 eV,44,45 which should make efficient contact to p-GaN. Also, the use of metal at the interface can provide an efficient charge reservoir for improved polarization-hole doping.46 The film thickness is chosen to avoid delamination resulting from high temperature thermal cycling during growth of a DUV LED structure. Prior to Pt deposition, an oxide removal of the p-Si wafers is performed using a ten-minute piranha etch followed by a ten-minute dip in dilute HF. They are immediately loaded into an e-beam deposition chamber, where they are coated with 1, 2, or 5 nm of Pt at a rate of 0.1 Å s−1.

Multiple samples of different metallic glass foils are tested as nanowire substrates [metallic brazing foils (MBFs), MetGlas, USA]. Only metallic glasses with solidus temperatures greater than the active region growth temperature of 850 °C are investigated (Table I). The metallic glasses used here are primarily Ni-based alloys, which can be used as a contact to p-GaN.24–26 

Table I.

Metallic glass substrates investigated, solidus temperatures, alloy compositions, and resulting nanowire characteristics.

FoilSolidus temp.
(°C)
NiCrFeCoNbBSiPNanowire formation
MBF-20 969 82    Barely 
MBF-30 984 93      Barely 
MBF-50 1052 72 19     Lumps 
MBF-51 1030 77 15     No 
MBF-62 984 66 20   Lumps 
MBF-90 966 76   20    Yes 
MBF-91 1065 69 14    Barely 
MBF-92 1064 69 14     Barely 
FoilSolidus temp.
(°C)
NiCrFeCoNbBSiPNanowire formation
MBF-20 969 82    Barely 
MBF-30 984 93      Barely 
MBF-50 1052 72 19     Lumps 
MBF-51 1030 77 15     No 
MBF-62 984 66 20   Lumps 
MBF-90 966 76   20    Yes 
MBF-91 1065 69 14    Barely 
MBF-92 1064 69 14     Barely 

The as-received metallic glasses and Pt-coated Si are cut into 2 × 2 cm2 and thoroughly cleaned with acetone before being put in sequential ultrasonic baths of acetone, methanol, and isopropyl alcohol. The samples are then loaded into a multiple temperature zone (MTZ) substrate holder that can hold four substrates for simultaneous growth at separate temperatures.47 The MTZ holder is used without any substrate backing plates (isothermal) in order to hold the metallic glass substrates (which cannot be indium bonded) under uniform growth conditions. Three samples of different Pt thicknesses, or different composition metallic glasses, are loaded along with an uncoated piece of p-Si serving as a control sample.

AlGaN nanowires are grown using a Veeco Gen930 molecular beam epitaxy (MBE) chamber. Prior to introduction into the main chamber, the samples are baked in a UHV buffer chamber for 1 h, with the Pt samples baked at 800 °C, and the metallic glasses at 600 °C, to ensure they are well below the solidus temperatures. A Ga beam equivalent pressure (BEP) of 6.20 × 10−8 Torr is used, along with a Mg BEP of 3.7 × 10−9 Torr. A nitrogen plasma power of 400 W and a flow rate of 4.75 sccm are used, resulting in a growth chamber background pressure of 2.61 × 10−5 Torr. The growth temperature is calibrated using a pyrometer on the uncoated Si substrate present in the MTZ holder. To set the nanowire density, a 5 min p-GaN nucleation step is performed at colder temperatures, 740 °C on Pt, but range from 740 to 760 °C for the metallic glasses. After the nucleation, p-GaN is deposited for 1 h at 790 °C.

An LED structure is grown on a p-Si wafer that is half coated with 2 nm of Pt using the methods described above. The primary Al and Ga BEPs are 4.10 × 10−8 and 6.20 × 10−8 Torr, respectively. First, p-GaN nanowires are nucleated for 5 min at 730 °C, after which the substrate temperature is increased to 770 °C, followed by deposition of 100 nm of linearly composition graded AlxGa1−xN (x = 0–1) co-doped with Mg. The temperature is then increased further to 850 °C for the active region, consisting of five periods of nominally 3 nm AlGaN quantum wells separated by 3 nm AlN barriers. After completion of the active region, the shutters for Si and both Al and a separate Ga source (BEP 3.0 × 10−8 Torr) are immediately opened and the substrate temperature cooled to 790 °C at a rate of 25 °C/min. As the temperature cools, the amount of Ga incorporation increases as well, giving a composition gradient from pure AlN to AlGaN of a composition dictated by the relative Ga/Al fluxes and temperature. Growth of 190 nm of n-AlGaN is performed before steeply grading fully back to GaN over the course of 10 nm. LED structures are fabricated using photolithography to pattern semitransparent 10 nm Ti/20 nm Au top contacts and etch mesa devices. Backside electrical contact is made by mechanical removal of the nanowires and In soldering to the Si wafer. Calibrated output power is measured using a Thorlabs PM100D power-meter and S120VC calibrated Si photodiode.

Plan-view scanning electron microscopy (SEM) images of GaN nanowires on p-Si and different Pt thicknesses are shown in Fig. 2. The addition of 1 nm of Pt reduces the apparent large-scale density of nanowires, growing in clumps that are similarly dense. This could be an effect of incomplete coverage of Pt, or ultrasonication of the substrates could have damaged the Pt thin film during cleaning. The 2 nm film shows structure remarkably similar to that of the bare Si wafer. However, if the thickness of the Pt is increased to 5 nm, GaN nanowire growth does not proceed at the same conditions. This could be for two reasons, both involving inhibited nucleation of GaN. First, a thicker layer of Pt will absorb more heat from the infrared radiation passing through the Si from the substrate heater, creating a hotter surface. This is critical during the temperature-sensitive nucleation step. If the substrate is too hot, nucleation will be inhibited, and there will be no existing GaN nuclei for vertical growth to proceed.48 Second, as evidenced in previous reports,37 nanowire growth does not proceed at grain boundaries as readily as bulk grains. Thus, if the 5 nm Pt film exhibits finer grain size, it would similarly suppress nanowire nucleation. If one desires to grow on thicker nanocrystalline films, the problem could be addressed by nucleating at lower substrate temperatures.

Fig. 2.

Plan view SEM images of GaN nanowires on [(a) and (b)] uncoated p-Si, [(c) and (d)] 1 nm Pt, [(e) and (f)] 2 nm of Pt, and [(g) and (h)] 5 nm of Pt on p-Si (all scale bars are 1 μm for the top row and 250 nm for the bottom row).

Fig. 2.

Plan view SEM images of GaN nanowires on [(a) and (b)] uncoated p-Si, [(c) and (d)] 1 nm Pt, [(e) and (f)] 2 nm of Pt, and [(g) and (h)] 5 nm of Pt on p-Si (all scale bars are 1 μm for the top row and 250 nm for the bottom row).

Close modal

Selected representative SEM images for GaN nanowires grown on metallic glass foils are shown in Fig. 3. The metallic glass substrates resulted in a wide variety of morphologies (summarized in Table I) varying from virtually no growth (MBF-51/92), to columnar lumps (MBF-50/62) of unknown composition/structure, to pockets of growth between very rough material (MBF-20/30/91/92), to extremely uniform nanowire arrays over large areas (MBF-90). The origin of this large variation in surface morphology is not yet understood. Figure 3(d) shows a large area plan-view image. The defect in the picture is the only sizeable defect seen in a 100 × 100 μm2 image and, as the uniformity rivals that of Si, was chosen to remain in the image for contrast. Notably, nanowire uniformity persists on length scales much larger than the size of typical devices.

Fig. 3.

Plan-view SEM images of GaN nanowires grown on different metallic glass substrates: (a) MBF-92 nanowire circled, (b) MBF-20 (inset shows the tilt view), and (c) MBF-90 (inset shows the tilt view). All scale bars are all 1 μm. (d) Large area plan-view image of GaN nanowires on MBF-90 showing large-scale uniformity of nanowires (inset shows the cross section view).

Fig. 3.

Plan-view SEM images of GaN nanowires grown on different metallic glass substrates: (a) MBF-92 nanowire circled, (b) MBF-20 (inset shows the tilt view), and (c) MBF-90 (inset shows the tilt view). All scale bars are all 1 μm. (d) Large area plan-view image of GaN nanowires on MBF-90 showing large-scale uniformity of nanowires (inset shows the cross section view).

Close modal

Room temperature photoluminescence measurements are carried out using an excitation wavelength of 232 nm (Fig. 4). Band-to-band recombination is observed in samples grown both on Pt and on MBF-90. There is a small but broad range of defect luminescence from ∼475 to 625 nm.49 Unfortunately, several of the metallic glasses undergo embrittlement during the growth and become fragile (inset, Fig. 4) and unable to withstand the handling involved in spin-on of photoresist and device patterning. However, some of the metallic glasses tested retained their flexibility after growth; thus, future work will focus on developing nanowire devices on the more thermally resilient metallic glasses.

Fig. 4.

Photoluminescence of GaN grown on 2 nm of Pt on p-Si (dash) and metallic glass (MBF-90) (solid) (inset shows foil embrittlement postgrowth).

Fig. 4.

Photoluminescence of GaN grown on 2 nm of Pt on p-Si (dash) and metallic glass (MBF-90) (solid) (inset shows foil embrittlement postgrowth).

Close modal

Conductive atomic force microscopy (cAFM) is used to probe the electrical characteristics of individual p-GaN nanowires grown on different substrates. Note that the cAFM tip/nanowire interface adds a Schottky barrier to the substrate/nanowire circuit. Figure 5(a) shows the current–voltage (I–V) curves from individual p-GaN nanowires grown directly on p-Si (solid) and on 2 nm Pt (dash). The distribution in I–V characteristics is represented by the spread in color. Nanowires on p-Si exhibit a threshold voltage between 5 and 9 V and no notable current in reverse bias. However, upon the addition of a Pt interlayer, the current appears more ohmic, with some beginning to pass current <4 V and also appearing more symmetric about 0 V, passing current in reverse bias. These results are consistent with the predicted reduction of the large Schottky barrier at the p-GaN/p-Si interface for p-GaN/Pt.

Fig. 5.

I–V curves taken by (a) cAFM on individual p-GaN nanowires and (b) mesoscopic ensemble devices on p-Si (solid) and 2 nm Pt films (dashed) (inset displays a semilog scale).

Fig. 5.

I–V curves taken by (a) cAFM on individual p-GaN nanowires and (b) mesoscopic ensemble devices on p-Si (solid) and 2 nm Pt films (dashed) (inset displays a semilog scale).

Close modal

In addition to cAFM measurements, mesoscopic devices are fabricated to test the ensemble averaged effect of the Pt substrate. A p-type top contact consisting of 20/30/80 nm of Pt/Ni/Au (Ref. 22) is deposited using standard photolithography and liftoff techniques. Figure 5(b) shows direct current I–V curves of mesoscopic ensemble devices on the two different substrates. These measurements show that Pt is indeed more conductive at low bias. However, the devices on Pt have a lower differential conductivity, and at biases greater than ∼ ± 1.5 V, the bare Si starts to be more conductive. As these devices are ∼300 × 300 μm2 in size, this corresponds to injection currents of only ∼2 mA. The discrepancy between the reverse bias leakage current in the cAFM curves, especially for Si, compared to the ensemble devices is likely due to shunt resistances introduced in the processing of mesoscopic devices.

The deep UV reflectivity spectra of the nanowire LED heterostructures are plotted in Fig. 6(a) to explore the photonic impact of Pt films. The deep UV reflectivity for bare p-Si substrates is larger than that of Pt/p-Si, but they become similar around 300 nm. The measurements are in agreement with the published optical properties,50,51 from which one expects reflectivity in the range 200–300 nm varying from 63% to 68% and 36% to 61% for Si and Pt, respectively. Next, AlGaN nanowire devices are examined. The overall reflectivity is greatly reduced and exhibits Fabrey–Perot (FP) oscillations consistent with the presence of a thick AlGaN layer that is optically transparent until the wavelength is shorter than the bandgap. The disappearance of the FP oscillations coincides with the onset of strong absorption in the AlGaN layer at the bandgap, ∼220 nm (5.64 eV). Interestingly, the AlGaN/Pt reflectivity is higher than that of AlGaN/p-Si over certain deep UV wavelengths. This highlights the importance of distinguishing between the air/Pt and AlGaN/Pt reflectivity spectra in evaluating the spectral suitability of a substrate for deep UV application.

Fig. 6.

(a) Reflectance spectra of bare substrates and of these substrates with the AlGaN LED heterostructure deposited. (b) Tilt-view SEM images of the AlGaN LEDs on thin Pt films and STEM-XEDS compositional maps of the LED structures on Pt films (c) showing Pt, Al, Ga, N, and Si.

Fig. 6.

(a) Reflectance spectra of bare substrates and of these substrates with the AlGaN LED heterostructure deposited. (b) Tilt-view SEM images of the AlGaN LEDs on thin Pt films and STEM-XEDS compositional maps of the LED structures on Pt films (c) showing Pt, Al, Ga, N, and Si.

Close modal

A tilt-view SEM image of the nanowire LED structure on Pt films is shown in Fig. 6(b). The wires are sufficiently dense for deposition of a top contact and of fairly uniform height. Compositional maps from a cross section of the LED structure were obtained using x-ray energy dispersive spectroscopy (XEDS) in scanning transmission electron microscopy (STEM). Nanowires nucleate and grow directly on the Pt film, verifying that it serves as a virtual substrate for the nanowires. The nucleation diameter of the nanowires (∼10 nm) is consistent with previous studies of self-catalyzed GaN nanowires grown by MBE (Ref. 48), indicating that the nucleation and growth mechanisms are not altered by Pt, i.e., it is not acting as a catalyst. GaN nanowires formed by the vapor–liquid–solid mechanism involving Pt catalyst exhibit 10× larger diameters and do not taper.52 The apparently slight thickness increase of the Pt substrate layer is attributed to delocalization of the x-ray emission at the substrate interface. Also apparent in Fig. 6(c) is the as-designed composition gradient (GaN → AlN) within the first 100 nm of the nanowire heterostructures. Since these thinned STEM specimens contain multiple nanowires, the composition gradients are inhomogeneously broadened making it difficult to identify finer features, such as the AlGaN multiple quantum well active region. However, these features were previously identified from single nanowire STEM mapping on identically prepared nanowires.53 

Pulsed voltage I–V curves are taken at 100 Hz and 4% duty cycle [Fig. 7(a)] on 250 × 250 μm2 devices. The devices on Si (solid) show an ∼2 V lower Vth than the devices on Pt (dash), and 10× lower leakage current at −10 V. The electroluminescence (EL) spectra of these LEDs are shown in Figs. 7(b) and 7(c). The devices on Pt exhibit 3× brighter EL than the devices on p-Si at high injection currents. The EL wavelength blueshifts with increasing injection current by 2.5 and 1 nm, for devices on Si and Pt, respectively.

Fig. 7.

(a) I–V curves of LED devices on Si (solid) and on Pt (dashed). EL spectra of 250 × 250 μm2 devices with semitransparent Ti/Au fully conformal contacts on (b) p-Si and (c) 2 nm Pt on p-Si, both at injection currents ranging from 1 to 28 mA.

Fig. 7.

(a) I–V curves of LED devices on Si (solid) and on Pt (dashed). EL spectra of 250 × 250 μm2 devices with semitransparent Ti/Au fully conformal contacts on (b) p-Si and (c) 2 nm Pt on p-Si, both at injection currents ranging from 1 to 28 mA.

Close modal

Extracted LED characteristics shown in Fig. 8(a) show that the device on Pt is approximately 29× brighter at the point of max intensity than the identical device on p-Si. Operating voltages at different injection currents during acquisition of EL spectra are shown in Fig. 8(b). Similar to what is observed in the pulsed I–V curves, the device on Pt begins to turn on at higher voltages than the device on Si. It also begins to increase quickly but level off around 10 mA, before increasing more quickly again prior to failure. The device on Si has a lower operating voltage, which is relatively more constant, increasing ∼1.5 V over the first 20 mA, compared to 5.9 V for Pt. The main difference in the early operating voltage is likely due to a difference in series resistances between the two devices. The increased operating voltage for Pt could also result in increased joule heating; however, the light output of the device seems unaffected.

Fig. 8.

Device characteristics: (a) peak EL, (b) operating voltage, and (c) efficiency as a function of drive current (inset shows output power vs input power) where each graph contains data for samples directly on p-Si (circles) and on 2 nm Pt on p-Si (squares).

Fig. 8.

Device characteristics: (a) peak EL, (b) operating voltage, and (c) efficiency as a function of drive current (inset shows output power vs input power) where each graph contains data for samples directly on p-Si (circles) and on 2 nm Pt on p-Si (squares).

Close modal

By measuring the output power, the efficiency of the devices is calculated [Fig. 8(c)]. The output power of devices on Pt is 21× higher than Si at an input power of 310 mW and 32× higher at lower powers. The efficiency of the device on Pt starts at ∼0.18 m% but quickly decreases at higher injection currents, leveling off around 0.05 m%, due to the quick increase in operating voltages, but still remains 13× higher than the same device on Si. The efficiency of the devices on Si is lower, peaking around 0.02 m%.

AlGaN nanowires were grown via molecular beam epitaxy on amorphous metals and nanocrystalline thin films. This was shown to greatly increase the uniformity over large areas compared to growth on polycrystalline bulk foils. However, more work remains to find suitable nucleation and growth conditions for nanowires on amorphous metals allowing them to maintain their flexibility without embrittlement. The growth of nanowires on thin Pt films also showed uniform nanowire growth until the Pt layer became too thick, resulting in a surface temperature too hot to nucleate nanowires. The addition of the high work function Pt layer improves the current injection at low bias, but results in a larger overall resistance at higher bias. However, the Pt layer increases the reflection of the substrate at the emission wavelength, and improvements of up to 35× in the resulting output power and increases in efficiency of 13× were observed.

This work was supported by the Ohio Development Services Agency through the Ohio Third Frontier. B.W., B.D.E., and D.W.M. acknowledge support from the Center for Emergent Materials at the Ohio State University, a National Science Foundation Materials Research Science and Engineering Center (Grant No. DMR-1420451).

1.
S.
Li
and
A.
Waag
,
J. Appl. Phys.
111
,
071101
(
2012
).
2.
H. P. T.
Nguyen
,
S.
Zhang
,
A. T.
Connie
,
M. G.
Kibria
,
Q.
Wang
,
I.
Shih
, and
Z.
Mi
,
Nano Lett.
13
,
5437
(
2013
).
3.
A. T. M. G.
Sarwar
,
B. J.
May
,
M. F.
Chisholm
,
G. J.
Duscher
, and
R. C.
Myers
,
Nanoscale
8
,
8024
(
2016
).
4.
S. M.
Sadaf
,
Y. H.
Ra
,
T.
Szkopek
, and
Z.
Mi
,
Nano Lett.
16
,
1076
(
2016
).
5.
T.
Frost
,
S.
Jahangir
,
E.
Stark
,
S.
Deshpande
,
A.
Hazari
,
C.
Zhao
,
B. S.
Ooi
, and
P.
Bhattacharya
,
Nano Lett.
14
,
4535
(
2014
).
6.
P.
Bhattacharya
,
A.
Hazari
,
S.
Jahangir
,
W.
Guo
, and
T.
Frost
,
Semicond. Semimetals
96
, 385 (
2017
).
7.
K. B.
Nam
,
J.
Li
,
M. L.
Nakarmi
,
J. Y.
Lin
, and
H. X.
Jiang
,
Appl. Phys. Lett.
84
,
5264
(
2004
).
8.
P.
Zhao
,
L.
Han
,
M. R.
McGoogan
, and
H.
Zhao
,
Opt. Mater. Express
2
,
1397
(
2012
).
9.
F.
Glas
,
Phys. Rev. B
74
,
121302(R)
(
2006
).
10.
S. D.
Hersee
,
A. K.
Rishinaramangalam
, and
M. N.
Fairchild
,
J. Mater. Chem. C
26
,
2293
(
2011
).
11.
C. V.
Falub
 et al,
Science
335
,
1330
(
2012
).
12.
S.
Zhao
,
S. Y.
Woo
,
M.
Bugnet
,
X.
Liu
,
J.
Kang
,
G. A.
Botton
, and
Z.
Mi
,
Nano Lett.
15
,
7801
(
2015
).
13.
K. H.
Li
,
X.
Liu
,
Q.
Wang
,
S.
Zhao
, and
Z.
Mi
,
Nat. Nanotechnol.
10
,
140
(
2015
).
14.
S.
Zhao
 et al,
Sci. Rep.
5
,
8332
(
2015
).
15.
S.
Zhao
,
X.
Liu
,
Y.
Wu
, and
Z.
Mi
,
Appl. Phys. Lett.
109
,
191106
(
2016
).
16.
P.
Pampili
and
P. J.
Parbrook
,
Mater. Sci. Semicond. Process.
62
,
180
(
2016
).
17.
F.
Bernardini
,
V.
Fiorentini
, and
D.
Vanderbilt
,
Phys. Rev. B
56
,
R10024
(
1997
).
18.
T.
Tanaka
,
A.
Watanabe
,
H.
Amano
,
Y.
Kobayashi
,
I.
Akasaki
,
S.
Yamazaki
, and
M.
Koike
,
Appl. Phys. Lett.
65
,
593
(
1994
).
19.
D.
Jena
 et al,
Appl. Phys. Lett.
81
,
4395
(
2002
).
20.
J.
Simon
,
V.
Protasenko
,
C.
Lian
,
H. G.
Xing
, and
D.
Jena
,
Science
327
,
60
(
2010
).
21.
S.
Li
,
T.
Zhang
,
J.
Wu
,
Y.
Yang
,
Z.
Wang
,
Z.
Wu
,
Z.
Chen
, and
Y.
Jiang
,
Appl. Phys. Lett.
102
,
062108
(
2013
).
22.
J. S.
Jang
,
I. S.
Chang
,
H. K.
Kim
,
T. Y.
Seong
,
S.
Lee
, and
S. J.
Park
,
Appl. Phys. Lett.
74
,
70
(
1999
).
23.
R.
France
,
T.
Xu
,
P.
Chen
,
R.
Chandrasekaran
, and
T. D.
Moustakas
,
Appl. Phys. Lett.
90
,
062115
(
2007
).
24.
J. K.
Ho
,
C. S.
Jong
,
C. C.
Chiu
,
C. N.
Huang
,
K. K.
Shih
,
L. C.
Chen
,
F. R.
Chen
, and
J. J.
Kai
,
J. Appl. Phys.
86
,
4491
(
1999
).
25.
R. H.
Horng
,
D. S.
Wuu
,
Y. C.
Lien
, and
W. H.
Lan
,
Appl. Phys. Lett.
79
,
2925
(
2001
).
26.
J. K.
Sheu
,
Y. K.
Su
,
G. C.
Chi
,
P. L.
Koh
,
M. J.
Jou
,
C. M.
Chang
,
C. C.
Liu
, and
W. C.
Hung
,
Appl. Phys. Lett.
74
,
2340
(
1999
).
27.
S. D.
Carnevale
,
T. F.
Kent
,
P. J.
Phillips
,
A. T. M. G.
Sarwar
,
C.
Selcu
,
R. F.
Klie
, and
R. C.
Myers
,
Nano Lett.
13
,
3029
(
2013
).
28.
A. T. M. G.
Sarwar
,
B. J.
May
,
J. I.
Deitz
,
T. J.
Grassman
,
D. W.
McComb
, and
R. C.
Myers
,
Appl. Phys. Lett.
107
,
101103
(
2015
).
29.
A. L.
Bavencove
 et al,
Nanotechnology
22
,
345705
(
2011
).
30.
F.
Limbach
 et al,
Nanotechnology
23
,
465301
(
2012
).
31.
B. J.
May
,
C. M.
Selcu
,
A. T. M. G.
Sarwar
, and
R. C.
Myers
,
Appl. Phys. Lett.
112
,
093107
(
2018
).
32.
B. J.
May
,
M. R.
Belz
,
A.
Ahamed
,
A. T. M. G.
Sarwar
,
C. M.
Selcu
, and
R. C.
Myers
,
ACS Nano
12
,
3551
(
2018
).
33.
A. T. M. G.
Sarwar
,
S. D.
Carnevale
,
F.
Yang
,
T. F.
Kent
,
J. J.
Jamison
,
D. W.
Mccomb
, and
R. C.
Myers
,
Small
11
,
5402
(
2015
).
34.
C.
Zhao
,
T. K.
Ng
,
N.
Wei
,
A.
Prabaswara
,
M. S.
Alias
,
B.
Janjua
,
C.
Shen
, and
B. S.
Ooi
,
Nano Lett.
16
,
1056
(
2016
).
35.
B.
Janjua
 et al,
Opt. Express
25
,
1381
(
2017
).
36.
A.
Prabaswara
,
J. W.
Min
,
C.
Zhao
,
B.
Janjua
,
D.
Zhang
,
A. M.
Albadri
,
A. Y.
Alyamani
,
T. K.
Ng
, and
B. S.
Ooi
,
Nanoscale Res. Lett.
13
,
1
(
2018
).
37.
B. J.
May
,
A. T. M. G.
Sarwar
, and
R. C.
Myers
,
Appl. Phys. Lett.
108
,
141103
(
2016
).
38.
M.
Wolz
,
C.
Hauswald
,
T.
Flissikowski
,
T.
Gotschke
,
S.
Fernandez-Garrido
,
O.
Brand
,
H. T.
Grahn
,
L.
Geelhaar
, and
H.
Riechert
,
Nano Lett.
15
,
3743
(
2015
).
39.
H.
Li
,
G.
Zhao
,
L.
Wang
,
Z.
Chen
, and
S.
Yang
,
Nanomaterials
6
,
195
(
2016
).
40.
G.
Calabrese
,
P.
Corfdir
,
G.
Gao
,
C.
Pfüller
,
A.
Trampert
,
O.
Brandt
,
L.
Geelhaar
,
S.
Fernández-Garrido
, and
A.
Trampert
,
Appl. Phys. Lett.
108
,
202101
(
2016
).
41.
L.
Geelhaar
 et al,
IEEE J. Sel. Top. Quantum Electron.
17
,
878
(
2011
).
42.
T. F.
Kent
,
S. D.
Carnevale
,
A. T. M.
Sarwar
,
P. J.
Phillips
,
R. F.
Klie
, and
R. C.
Myers
,
Nanotechnology
25
,
455201
(
2014
).
43.
S.
Zhao
,
M. G.
Kibria
,
Q.
Wang
,
H. P. T.
Nguyen
, and
Z.
Mi
,
Nanoscale
5
,
5283
(
2013
).
44.
H. B.
Michaelson
,
J. Appl. Phys.
48
,
4729
(
2013
).
45.
G. N.
Derry
and
Z.
Ji-zhong
,
Phys. Rev. B
39
,
1940
(
1989
).
46.
S.
Li
 et al,
J. Appl. Phys.
112
,
053711
(
2012
).
47.
B. J.
May
and
R. C.
Myers
,
J. Vac. Sci. Technol. B
36
,
011203
(
2018
).
48.
S. D.
Carnevale
,
J.
Yang
,
P. J.
Phillips
,
M. J.
Mills
, and
R. C.
Myers
,
Nano Lett.
11
,
866
(
2011
).
49.
M. A.
Reshchikov
,
H.
Morkoç
,
B.
Nemeth
,
J.
Nause
,
J.
Xie
,
B.
Hertog
, and
A.
Osinsky
,
J. Appl. Phys.
97
,
061301
(
2005
).
50.
W. S. M.
Werner
,
K.
Glantschnig
, and
C.
Ambrosch-Draxl
,
J. Phys. Chem. Ref. Data
38
,
1013
(
2009
).
51.
D. E.
Aspnes
and
A. A.
Studna
,
Phys. Rev. B
27
,
985
(
1983
).
52.
E.
Park
,
S.
Shim
,
R.
Ha
,
E.
Oh
,
B.
Woo
, and
H.
Choi
,
Mater. Lett.
65
,
2458
(
2011
).
53.
P. J.
Phillips
,
S. D.
Carnevale
,
R.
Kumar
,
R. C.
Myers
, and
R. F.
Klie
,
ACS Nano
7
,
5045
(
2013
).