Diode Characteristics Approaching Bulk Limits in ... - ACS Publications

Mar 23, 2017 - California NanoSystems Institute, University of California at Los Angeles, Los Angeles,. California 90095, United States. §. School of...
0 downloads 12 Views 1MB Size
Nano Letters is Diode published by the American Chemical characteristics Society. 1155 Sixteenth Street N.W., approaching bulk Washington, DC 20036 limits inPublished GaAs by American Chemical Society. nanowire array Copyright © American

Pradeep Nuwan Nano Letters is Senanayake, xiao published by the Chemical meng, NickAmerican Y. Hsieh, Society. 1155 and Diana L. Huffaker Sixteenth Street N.W., Washington, DC

Nano Lett., Just Accepted 20036 Manuscript Published • Publication by American Society. Date (Web): Chemical 23 Mar 2017 Copyright © American

Nano Letters is published by the American Chemical Society. 1155 Street N.W., Just AcceptedSixteenth Washington, DC 20036 by American “Just Accepted”Published manuscripts have been Chemical Society. online prior to technical editing, Copyright © American formatti

appear in full in PDF format accompanie Nano Letters is fully peer reviewed, but should not be co published by the readers and citable byChemical the Digital Object American Society. 1155 to authors. Therefore, the “Just Accept Sixteenth Street N.W., in the journal. After a manuscript is tec Washington, DC Accepted” Web20036 site and published as a by American changes to thePublished manuscript text and/or Chemical Society. and ethical guidelines apply to th Copyright © that American

Page 1 of 21

1 2 3

Nano Letters

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

Diode characteristics approaching bulk limits in GaAs nanowire array photodetectors

Alan C. Farrell1, Pradeep Senanayake1, Xiao Meng3, Nick Y. Hsieh1, and Diana L. Huffaker1, 2, 3

1

Electrical Engineering Department, University of California at Los Angeles, Los Angeles, CA 90095, USA 2 California NanoSystems Institute, University of California at Los Angeles, Los Angeles, CA 90095, USA 3 School of Physics and Astronomy, Cardiff University, Cardiff, Wales, UK

1 ACS Paragon Plus Environment

Page 2 of 21

Page 3 of 21

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

Abstract We present the electrical properties of p-n junction photodetectors comprised of vertically oriented p-GaAs nanowire arrays on n-GaAs substrate. We measure an ideality factor as low as n = 1.0 and a rectification ratio > 108 across all devices, with some > 109, comparable to the best GaAs thin film photodetectors. Analysis of the Arrhenius plot of the saturation current yields an activation energy of 690 meV – approximately half the bandgap of GaAs – indicating generation-recombination current from mid-gap states is the primary contributor to the leakage current at low bias. Using fully 3-dimensional electrical simulations, we explain the lack of a recombination current dominated regime at low forward bias, as well as some of the issues related to analysis of the capacitance-voltage characteristics of nanowire devices. This work demonstrates that through proper design and fabrication, nanowire-based devices can perform as well as their bulk device counterparts.

Keywords Nanowires, GaAs, photodetectors, surface passivation 2 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

Semiconductor nanowires are frequently touted as promising building blocks for a variety of nanoscale electronic and optoelectronic devices. The primary criticism of nanowire devices is the large surface-to-volume ratio and the deleterious effect of surface states on electrical properties. Indeed, for thin film devices involving mesa etching, the etched surface presents a problem as the diameter of the mesa becomes small, eventually dominating the carrier transport in the device1,2. As a result, ex-situ passivation of the etched sidewalls is necessary to minimize these effects3,4. Nanowire growth and fabrication is intrinsically different in that there is no postgrowth etching required, and therefore no ex-situ passivation is required; all sidewalls are passivated through in-situ growth of a high-bandgap shell. To date however, all demonstrations of nanowire devices have been inferior to their thin film counterparts in most performance metrics, including ideality factor and dark current density. The dark current at room temperature in p-n diodes employing III-V semiconductors with a short carrier lifetime and/or small intrinsic carrier concentration is dominated by bulk generation current through recombination centers at the midgap in the depletion region5–7. Hence, even the best thin film GaAs diodes have an ideality factor of n = 1.1 – 1.55,8,9. However, the vast majority of nanowire-based p-n junctions reported to date have an ideality factor of  > 2, higher than expected for a GaAs diode dominated by bulk generation or surface leakage current. This is likely due to either improper device design, poor bulk material quality, or inadequate surface passivation10,11. The best reported ideality factor of  ≈ 1.6 and rectification ratio of 105 was achieved with vertically-aligned array core-shell p-n junction diode solar cells12 and photodetectos13 with an InGaP passivation shell. In both cases, nanowire arrays are embedded in benzocyclobutene (BCB), which has been shown to effectively passivate GaAs

3 ACS Paragon Plus Environment

Page 4 of 21

Page 5 of 21

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

surfaces12,13. It is difficult, however, to determine the effectiveness of the BCB passivation because for these core-shell structures a fraction of the junction area exists in air above the BCB. In this work, we present a p-GaAs nanowire array photodetector on n-GaAs substrate grown by selective-area metal-organic vapor deposition (SA-MOCVD) operating as p-n diode with the junction at the nanowire-substrate interface. The purpose of this approach is to ensure the depletion region remains below the surface of the BCB in order to assess the effectiveness of InGaP + BCB passivation. The standard perimeter-to-area analysis is not suitable for this study as it is not easy to control or define the nanowire diameter due to unavoidable lateral overgrowth and the variation in nanowire height with mask-hole diameter. Therefore, the effectiveness of the passivation is inferred from ideality factor and dark current density. Fully 3-dimensional simulations are employed in order to analyze the electrical characteristics. The results are compared to the standard 1-dimensional analysis and the range of applicability is discussed. Fig. 1a shows the device layout, including BCB, via, substrate, and nanowire electrical contacts. Note that the ground-signal-ground contacts are not necessary for this study, but are merely a convenience allowing the use of existing photolithography masks. The close-up view in Fig. 1b shows the partial gold shell and self-aligned hole resulting from tilted metal deposition. The purpose of this design is discussed in detail elsewhere14–16, but for our purposes in this study it simply serves as electrical contact to the nanowires. The schematic in Fig. 1c shows the nanowire device structure. A p-GaAs nanowire is surrounded by an InGaP shell acting as a passivation layer. Because the doping in the substrate (~3 × 10 cm ) is much higher than in the nanowire, the result is a one-sided p-n diode, with the depletion region extending primarily through the nanowire. Therefore, the electrical characteristics reflect the structure and properties of the nanowire component of the device. The current-voltage (I-V) relationship of an 4 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

Page 6 of 21

Figure 1. (a) False-color SEM image of fabricated nanowire array photodetector. Scale bar is 50 µm. (b) Close-up view of active area showing nanowire array. Scale bar is 500 nm. (C) Schematic of device structure. 80 μm diameter nanowire array (~5,026 nanowires) is shown in Fig. 2a along with the calculated I-V using the Shockley equation     

   #$

!"

%  1&

5 ACS Paragon Plus Environment

(1)

Page 7 of 21

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. (a) Measured (symbols) I-V of a typical photodetector at room temperature, and the calculated (line) I-V using the Shockley equation. (b) Temperature dependence of the ideality factor. Inset: Bias dependence of the ideality factor at room temperature. (c) Arrhenius plot of the saturation current. with   4.8 × 10 ( A,   1.15, and

!

 1.8 Ω as fitting parameters. The ideality factor

reaches the minimum of n = 1.15 above 250 K (Fig. 2b). The bias dependence of the ideality factor is calculated by numerical differentiation of the forward bias current with respect to voltage. As seen in the inset to Fig. 2b, the ideality factor is independent of bias up to about 0.4 V, where series resistance effects decrease the slope of the current and hence increase the ideality factor. Since series resistance is not taken into account when extracting the ideality factor from

6 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

Page 8 of 21

the slope of the current, the ideality factor at high forward bias is not the real, but an effective ideality factor17. In Fig. 2c, an Arrhenius plot of the saturation current is used to extract the activation energy, assuming ./  ∝  - 0 #$

(2)

giving ./  690 meV at room temperature and ./  21 meV at low temperature. At room temperature, the activation energy is roughly half the bandgap of GaAs, indicative of midgap states. The low temperature activation energy of 21 meV is likely due to zinc dopants18, but carbon impurities may also contribute19. The maximum rectification ratio of this particular device is 2.1 × 104 (at 1 Vfb), the highest reported to date for a nanowire p-n diode of any material, and comparable to high quality bulk devices. The series resistance for a single nanowire photodetector can be calculated by multiplying the total series resistance by the number of nanowires in the array, giving

!



9 kΩ. By calculating the area of the nanowire contacted with metal, the specific contact resistance is estimated to be 67  1.3 × 108 Ω ∙ cm( . Similarly, the dark current per nanowire at

1 Vrb is estimated to be about 0.8 fA. Single nanowire diodes and 3 × 3 nanowire photodetectors were also fabricated and characterized (Fig. 3). The plot shows two representative I-V curves of each array size, with very little variation between devices. Note that it is not possible to determine the rectification ratio for these devices as the dark current is below the sensitivity of the measurement. However, for the single nanowire device it is at least 10 and for the 3 × 3

nanowire photodetector it is at least 104 . The series resistance for a single nanowire photodetector was found by fitting to the Shockley equation and gave 3 × 3 nanowire photodetector we find

!

!

 9.4 kΩ, and for the

 12.1 kΩ per nanowire. Interestingly, the ideality 7

ACS Paragon Plus Environment

Page 9 of 21

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

factor for these nanowire photodetectors is   1.0, as opposed to   1.15 for the larger

nanowire array. The reverse bias current for both single and 3 × 3 nanowire photodetectors is less than the noise floor of the measurement setup (~50 fA), supporting the 0.8 fA/nanowire estimate from the larger array.

Figure 3.

Dark current at room temperature for two sets of identical photodetectors

composed of single nanowires and 3 × 3 arrays of nanowires. Inset: Close-up SEM of single nanowire photodetector. Scale bar is 500 nm. For bulk photodetectors, the dark current density is given by :  /??  :>?? . If only ?? and :>?? are given, we might incorrectly assume that these devices are equivalent, when in fact device 1 is superior since the same dark current is achieved with a greater number of nanowires. This simple example illustrates the importance of ?? , and := (or equivalently, ??

Conditions -1 V, 300 K -1 V, 300 K -1 V, 633 nm

Min. 6 × 10E 6 × 104

Typical 7 × 10C 8 × 10 0.1

Max. 1 × 108 2 × 10E

Units A/cm2 A/cm2 A/W

Table 1. Dark current density and responsivity calculated using both the total junction area as well as the effective device area.

fiber-pigtailed laser diode with 100 µW incident power. Although the responsivity is over a factor of 4 lower than a bulk GaAs photodetector, the effective dark current density is over an order of magnitude lower than bulk InGaAs, resulting in a sensitivity that is equal or higher than most commercial detectors. Detailed studies of the physics of absorption through surface plasmon resonances are out of the scope of this work and can be found elsewhere14–16. We now provide an analysis of the main electrical properties of the nanowire photodetectors. Recombination current in the region where the junction space-charge region intersects with the surface is known to produce a current with   2 at a forward bias as high as 1V20. Bulk current behaves similarly at low forward bias, where recombination in the junction space-charge region dominates, but drops to   1 at high forward bias where diffusion current

becomes very large21. The absence of an   2 region for the GaAs nanowire p-n photodetector

at low forward bias is surprising, and merits discussion. For an D -p junction at a forward bias greater than a few #$/, the total current can be approximated by 10 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

:FG  :H + :J>

KL M(  SH M  ≈ exp - 0 + exp 0 NL OP #$ 2TL 2#$

Page 12 of 21

(3)

where :H and :J> are the diffusion and bulk recombination current components, KL is the electron diffusion coefficient, TL is the electron lifetime, OP is the acceptor doping level, and SH is the

depletion region width. The ratio of the diffusion to recombination current is thus :H 2M KL TL   exp 0 :J> OP NL SH 2#$

(4)

With the parameters for GaAs given in the Supplementary Information, we get :H 10  cm ≈ :J> OP

(5)

at 300 mV. It is clear that even at background doping levels (~1015 cm-3) the diffusion current in GaAs is several orders of magnitude smaller than the recombination current and we expect  

2. At a bias greater than 1 V (with OP  10 C cm ), the diffusion current is greater than the

recombination current and   1. To understand why this behavior is not observed in the GaAs

nanowire photodetector, we note that since the nanowires are less than 1 μm long, then NL ≈

3 μm ≫ SVW , where SVW is the width of the quasi-neutral region in the nanowire. Thus, this is a narrow-base diode and the diffusion current component should be much larger than suggested by the previous analysis. In order to quantify the increase in diffusion current and determine whether it can account for the absence of a recombination dominated current at low forward bias, we perform fully 3dimensional (no rotational symmetry) electrical simulations. The simulation domain is schematically shown in Fig. 4a. The simulations include the effects of Shockley-Read-Hall recombination, including surface recombination. The dimensions of the nanowire were defined based on SEM measurements of the actual nanowires. Every surface was treated as non-ideal 11 ACS Paragon Plus Environment

Page 13 of 21

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

with a surface recombination velocity. All dimensions and material/surface parameters can be found in the Supplementary Information. It is important to note that the contact to the nanowire is defined not only on the top facet, but on three of the sidewalls above the BCB, as in the actual device. As we will see shortly, this has a dramatic effect on the device characteristics.

12 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

Page 14 of 21

Figure 4. (a) Electric field distribution in 3-dimensional space. The segment of the nanowire exposed to air is kept fixed at 400 nm, while the portion buried in BCB is varied between 300 and 600 nm. (b) 2D cuts of the electric field distribution at 0 Vrb and 5 Vrb, and 1D cuts of the electric field through the center of the nanowire. (c) Simulated depletion region width for several p-doping concentrations (symbols) and calculated depletion region width with the exponential dependence on the bias voltage used as a fitting parameter. At the top of Fig. 4b, 2-dimensional views of the electric field distribution are shown at 0 V and 5 V reverse bias, along with 1-dimensional cuts through the center of the nanowire below. It is immediately apparent that the electric field distribution within this nanowire device structure is markedly different than a thin film device. Initially, the electric field is confined within the

13 ACS Paragon Plus Environment

Page 15 of 21

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

BCB layer, and resembles the standard triangular profile of a 1-dimensional device. As the bias is increased, the depletion region edge approaches the portion of the electrical contact at the surface of the BCB. As the bias is increased further, the depletion region edge is pinned at the surface of the BCB and produces a second peak in the electric field. We next investigate the bias dependence of SH . We define the edge of the depletion region as the location where the electric

field drops below 10 kV/cm. Fig. 4c shows SH as a function of JX + YXM for nanowire doping

concentrations of 2 × 10 C cm to 5 × 10 E cm. It is well known that for 1-dimensional

devices with abrupt junctions, SH ∝ JX + YXM "Z , with [  0.5 and [  1/3 for graded

junctions. For the nanowire device, we find [ ≈ 0.5 for higher doping and gradually increases

to [ ≈ 1 as the doping decreases. When the depletion region edge is pinned to the BCB surface, we find [  0.2, regardless of doping concentration.

In order to estimate the doping concentration in the nanowire photodetector, we measured the capacitance-voltage (C-V) characteristics and assume \ ∝ SH to calculate the capacitance using the previous simulations. In Fig. 5a, the capacitance per nanowire (symbols) is plotted along with the capacitance of an ideal diode assuming \  \ /YXM + ".8 and the calculated

capacitance using \  \ /SH , where \ is a proportionality factor adjusted to fit the data. The

best fit to the data was achieved by using SH calculated for a doping concentration of 5 ×

10 C cm . The fact that the measured C-V exhibits an [  0.2 region is clear evidence that the

depletion region edge is pinned at the BCB surface for reverse bias greater than 1 V. This means that the depletion region edge is very close to the BCB surface at 0 V, and supports the narrowbase diode argument put forward earlier. It is worth noting the standard 1/C2 analysis of the measured C-V data gives YXM  4.1 V and OP  1 × 10( cm, which is clearly nonsensical.

14 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

Page 16 of 21

Figure 5. (a) Measured (symbols) capacitance per nanowire and calculated capacitance assuming the standard m = 0.5 bias dependence for an ideal diode (blue solid line) and the calculated capacitance using the simulated depletion region width (red solid line). (b) Simulated forward bias current (symbols) for a short (700 nm) and long (1000 nm) nanowire, and the calculated forward bias current using the Shockley equation using a single diode model for the short nanowire and a 2-diode model for the long nanowire. (c) Carrier concentration through the center of the p-GaAs nanowire. The position of the contact is This exemplifies the need for careful attention to the validity of the equations used to analyze nanowire devices. The question naturally arises as to whether the 1/C2 analysis yields accurate results if the measured capacitance is proportional to YXM + ".8. We found that 15 ACS Paragon Plus Environment

Page 17 of 21

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

some devices on the same sample do, in fact, have a capacitance bias dependence with [  0.5. The 1/C2 analysis was performed on one of these devices (Supplementary Info) and we found YXM  1.74 V and OP  6 × 10 C cm . Although the built-in voltage does not agree well with the simulated built-in voltage, YXM  1.33 V, the doping concentration is very close to what was found through 3D simulations. The forward bias current was simulated for nanowires with heights of 700 and 1000 nm, shown in Fig. 5b. The long nanowire exhibits a two-diode behavior with   2 region at low forward bias and an   1.15 region at high forward bias, typical of thin film GaAs

photodetectors. The short nanowire, on the other hand, has a single diode ideality factor of   1.3. The fact that   1.15 for the experimental devices is irrelevant, as this analysis is simply intended to elucidate the factors that lead to non-standard characteristics in nanowire photodetectors, i.e., [ ≠ 0.5 and single diode behavior. To understand the difference between the long and short diodes, we look at the electron concentration within the p-type nanowire at 0.3 V forward bias, shown in Fig. 5c. Since the excess electrons recombine at the contact, the electron concentration at the axial position of the contact is drawn with dotted lines for both long and short diodes. Note that the electron concentration at the contact is much higher in the short diode than in the long diode. This explains the increase in diffusion current responsible for eliminating the   2 region at low bias. We have demonstrated vertically-oriented nanowire array GaAs p-n junction photodetectors with ideality factors as low as   1, rectification ratios > 108 (with a maximum of 2 × 104 ), and dark current density comparable to bulk. It was shown through detailed 3-

dimensional simulations that the ideality factor is affected by an increase in the diffusion current resulting from the narrow-base diode structure. The diffusion current thus exceeds the 16 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

Page 18 of 21

recombination current and eliminates the low-bias   2 region typical of GaAs thin film photodetectors. In addition, the capacitance of a nanowire photodetector was shown to exhibit atypical behavior in the rate at which the capacitance decreases as a function of reverse bias, initially increasing more rapidly, then less rapidly than expected for thin film photodetectors. This was attributed to depletion region pinning at the BCB surface as the electric field approaches the electrical contact. This work demonstrates that it is entirely possible to achieve bulk-like performance from nanowire devices if proper design considerations are implemented. In addition, we showed that electrical characterization of nanowire devices must be done with care and the standard 1-dimensional 1/C2 analysis does not yield accurate results for the doping concentration unless [  0.5. Supporting Information Additional figures related to device uniformity, photoresponse, and capacitance. This material is available free of charge via the Internet at http://pubs.acs.org. Corresponding Author E-mail: [email protected] Notes The authors declare no competing financial interests. Acknowledgements The authors wish to acknowledge the generous financial support of this research by NSF (through ECCS-1202591, DMR-1007051, and STTR- 1346335), and by DARPA (W911NF-13-

17 ACS Paragon Plus Environment

Page 19 of 21

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

1-0188). We acknowledge the use of the Integrated Systems Nanofabrication Cleanroom at the California NanoSystems Institute.

References (1)

Bhagwat, V.; Xiao, Y.; Bhat, I.; Dutta, P.; Refaat, T. F.; Abedin, M. N.; Kumar, V. J. Electron. Mater. 2006, 35 (8), 1613–1617.

(2)

Prineas, J. P.; Maiorov, M.; Cao, X. C.; Olesberg, J. T.; Flatté, M. E.; Reddy, M.; Coretsopoulos, C.; Itzler, M. Spie 2006, 6119, 611904-611904–611912.

(3)

Kim, H. S.; Plis, E.; Gautam, N.; Myers, S.; Sharma, Y.; Dawson, L. R.; Krishna, S. Appl. Phys. Lett. 2010, 97 (14), 18–21.

(4)

Ma, Y.; Zhang, Y.; Gu, Y.; Chen, X.; Shi, Y.; Ji, W.; Xi, S.; Du, B.; Li, X.; Tang, H.; Li, Y.; Fang, J. Opt. Express 2016, 24 (7), 7823.

(5)

Farag, A. A. M.; Fadel, M.; Yahia, I. S. Curr. Appl. Phys. 2012, 12 (6), 1436–1444.

(6)

Forrest, S. R.; Leheny, R. F.; Nahory, R. E.; Pollack, M. A. Appl. Phys. Lett. 1980, 37 (3), 322–325.

(7)

Buchali, F.; Behrendt, R.; Heymann, G. Electron. Lett. 1991, 27 (3), 235–237.

(8)

Mauk, M. G.; Xu, S.; Arent, D. J.; Mertens, R. P.; Borghs, G. Appl. Phys. Lett. 1989, 54 (3), 213–215.

(9)

Dodd, P. E.; Stellwag, T. B.; Melloch, M. R.; Lundstrom, M. S. IEEE Trans. Electron Devices 1991, 38 (6), 1253–1261.

18 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

(10)

Chang, C.-C.; Chi, C.-Y.; Yao, M.; Huang, N.; Chen, C.-C.; Theiss, J.; Bushmaker, A. W.; Lalumondiere, S.; Yeh, T.-W.; Povinelli, M. L.; Zhou, C.; Dapkus, P. D.; Cronin, S. B. Nano Lett. 2012, 12 (9), 4484–4489.

(11)

Holm, J. V; Jørgensen, H. I.; Krogstrup, P.; Nygard, J.; Liu, H.; Aagesen, M.; Nygård, J.; Liu, H.; Aagesen, M. Nat. Commun. 2013, 4, 1498.

(12)

Chiu, H. C.; Hwu, M. J.; Yang, S. C.; Chan, Y. J. IEEE Electron Device Lett. 2002, 23 (5), 243–245.

(13)

Chiu, W. Y.; Huang, F. H.; Wu, Y. S.; Lin, D. M.; Chan, Y. J.; Chen, S. H.; Chyi, J. I.; Shi, J. W. Japanese J. Appl. Physics, Part 1 Regul. Pap. Short Notes Rev. Pap. 2005, 44 (4 B), 2586–2587.

(14)

Farrell, A. C.; Senanayake, P.; Hung, C.; Currie, M.; Huffaker, D. L. In 72nd Annual Device Research Conference; 2014; pp 175–176.

(15)

Senanayake, P.; Hung, C.-H.; Shapiro, J.; Lin, A.; Liang, B.; Williams, B. S.; Huffaker, D. L. Nano Lett. 2011, 11 (12), 5279–5283.

(16)

Senanayake, P.; Hung, C.-H.; Shapiro, J.; Scofield, A.; Lin, A.; Williams, B. S.; Huffaker, D. L. Opt. Express 2012, 20 (23), 25489.

(17)

Mikhelashvili, V.; Eisenstein, G.; Uzdin, R. Solid. State. Electron. 2001, 45, 143–148.

(18)

Hill, D. E. J. Appl. Phys. 1970, 41 (4), 1815–1818.

(19)

Scolfaro, L. M. R.; Pintanel, R.; Gomes, V. M. S.; Leite, J. R.; Chaves, A. S. Phys. Rev. B 1986, 34 (10).

(20)

Aberg, I.; Vescovi, G.; Asoli, D.; Naseem, U.; Gilboy, J. P.; Sundvall, C.; Dahlgren, A.; 19 ACS Paragon Plus Environment

Page 20 of 21

Page 21 of 21

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

Svensson, K. E.; Anttu, N.; Bjork, M. T.; Samuelson, L. IEEE J. Photovoltaics 2016, 6 (1), 185–190. (21)

Sze, S. M.; Ng, K. K. Physics of Semiconductor Devices, 3rd ed.; John Wiley & Sons, 2006.

20 ACS Paragon Plus Environment