Surface Emitting, High Efficiency Near-Vacuum Ultraviolet Light

Sep 16, 2015 - Molecular beam epitaxy growth of Al-rich AlGaN nanowires for deep ultraviolet optoelectronics. S. Zhao , S. Y. Woo , S. M. Sadaf , Y. W...
0 downloads 8 Views 2MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Surface Emitting, High Efficiency Near-Vacuum Ultraviolet Light Source with Aluminum Nitride Nanowires Monolithically Grown on Silicon S. Zhao, M. Djavid, and Z. Mi* Department of Electrical and Computer Engineering, McGill University, 3480 University Street, Montreal H3A 0E9, Canada S Supporting Information *

ABSTRACT: To date, it has remained challenging to realize electrically injected light sources in the vacuum ultraviolet wavelength range (∼200 nm or shorter), which are important for a broad range of applications, including sensing, surface treatment, and photochemical analysis. In this Letter, we have demonstrated such a light source with molecular beam epitaxially grown aluminum nitride (AlN) nanowires on low cost, large area Si substrate. Detailed angle dependent electroluminescence studies suggest that, albeit the light is TM polarized, the dominant light emission direction is from the nanowire top surface, that is, along the c axis, due to the strong light scattering effect. Such an efficient surface emitting device was not previously possible using conventional c-plane AlN planar structures. The AlN nanowire LEDs exhibit an extremely large electrical efficiency (>85%), which is nearly ten times higher than the previously reported AlN planar devices. Our detailed studies further suggest that the performance of AlN nanowire LEDs is predominantly limited by electron overflow. This study provides important insight on the fundamental emission characteristics of AlN nanowire LEDs and also offers a viable path to realize an efficient surface emitting near-vacuum ultraviolet light source through direct electrical injection. KEYWORDS: AlN, LED, vacuum UV, electrical injection, surface emission, TM polarization

A

light emission of AlN. In the wurtzite (WZ) form, AlN has a negative crystal field splitting energy (ECEF = −219 meV), due to the small c/a ratio (1.601) and large u (0.3819) parameter.15 The negative crystal field splitting, which is opposite to GaN (ECEF = +38 meV), implies that the light emission is predominantly TM polarized, with E∥c. As a consequence, the emitted light can only propagate in-plane and cannot be extracted from the top surface, illustrated in Figure 1a, leading to extremely low light extraction efficiency for surface emitting

n efficient light source that can operate in the wavelength range ∼200 nm, or shorter, is in demand for a broad range of applications including sterilization, medical treatment, surface cleaning, and industrial manufacturing to laboratory research. The current approaches, including mercury and excimer lamps, have extremely low efficiency, very short lifetime, poor stability, and also involve the use of toxic gas. To date, it has remained challenging to realize an electrically injected, all-semiconductor based high efficiency surface emitting light source in such a wavelength range.1−11 Among the various large bandgap semiconductor materials, aluminum nitride (AlN) exhibits a direct bandgap energy of ∼6.2 eV (∼200 nm in wavelength) and can be substitutionally doped nand p-type. With suitable quantum confinement, AlN-based nanostructures can exhibit emission wavelengths 20 V),5 with an electrical efficiency of ∼10% for a moderate injection current density of 20 A/cm2. Such a poor electrical efficiency is directly related to the presence of large densities of defects and dislocations, and the extremely inefficient p-type dopant (Mg) due to the large ionization energy (∼600 meV).5,12 Additionally, the performance of conventional c-plane AlN LED structures has been limited by the extremely low light extraction efficiency,1,2,5,13,14 due to the unique TM polarized © 2015 American Chemical Society

Figure 1. (a) Schematic of TM polarized light propagation in (a) conventional c-plane AlN planar LEDs, and (b) AlN nanowire LEDs. Received: July 31, 2015 Revised: September 7, 2015 Published: September 16, 2015 7006

DOI: 10.1021/acs.nanolett.5b03040 Nano Lett. 2015, 15, 7006−7009

Letter

Nano Letters devices.5 Various techniques, including the use of a-plane AlN13 and photonic crystals16 have been explored to enhance the light extraction efficiency, but with limited success. In this Letter, we show that such critical challenges can be addressed by using catalyst-free AlN nanowire structures grown under nitrogen rich conditions. Atlhough surface emission is largely prohibited in conventional c-plane AlN LEDs, the strong light scattering effect in nanowire structures can lead to significantly enhanced surface emission for TM polarized photons, illustrated in Figure 1b. Detailed angle-resolved electroluminescence (EL) emission measurements confirm that the surface emission is dominant for AlN nanowire LEDs. The correlation between light extraction efficiency and nanowire size and spacing is further discussed. Moreover, by exploiting the hole hopping conduction in the impurity band of Mg-doped AlN nanowire structures,17 an electrical efficiency of >85% at 20 A/cm2 was realized in such large area AlN nanowire LEDs for the first time. Through detailed current-dependent studies of the EL emission characteristics at various temperatures, we have further elucidated the role of surface recombination and electron overflow on the performance of AlN nanowire LEDs. The AlN nanowire LEDs were grown on low resistivity ntype Si substrate by radio frequency (RF) plasma-assisted molecular beam epitaxy (PAMBE) under nitrogen-rich conditions without using any external metal catalyst.6,17,18 Under these conditions, the formation of nitrogen vacancies and related defects, which often limit the achievement of p-type AlN, can be largely eliminated.6,18 The device structure consists of 90 nm n-GaN bottom contact layer, 90 nm n-AlN, 60 nm nondoped AlN, 15 nm p-AlN, and 3 nm p-AlGaN contact layer, illustrated in Figure 1b. In this experiment, a Si-doped GaN nanowire template was first grown to promote the formation of AlN nanowires.18 The growth conditions for GaN nanowire template included a substrate temperature of 790 °C, a Ga beam equivalent pressure (BEP) of 6 × 10−8 Torr, a nitrogen flow rate of 1.0 sccm, and a RF plasma forward power of 350 W. The AlN nanowire growth temperature was 800 °C, and Al BEP was 2 × 10−8 Torr. Under optimum growth conditions, the coalescence of AlN nanowires can be minimized or eliminated. The Si and Mg dopant concentrations for n-AlN and p-AlN sections, calibrated from Si- and Mg-doped thin films grown under similar conditions, were estimated to be 5 × 1017 cm−3 and 1 × 1021 cm−3, respectively.6 However, it is noted that the Mg dopant concentration in p-AlN nanowires could be much lower due to the large Mg surface desorption rate at the growth temperature.19 The corresponding SEM image, taken at a 45° angle, is shown in Figure 2a. It is seen that vertically aligned and

uniform AlN nanowire structures were formed. The average nanowire diameter is ∼60 nm, and average spacing is ∼30 nm. Previous studies have suggested that MBE-grown AlN nanowires have a nitrogen polarity.6,20 The room-temperature photoluminescence spectrum of AlN nanowires, which were excited by a 193 nm ArF2 excimer laser, is shown in Figure 2b. It is seen that strong band edge emission at ∼207 nm was measured with no defect-related emission in the deep ultraviolet (UV) spectral range, suggesting an excellent material quality. Such AlN nanowires can exhibit an internal quantum efficiency (IQE) of up to 80%,6 which is attributed to the low defect density and is naturally expected given the large exciton binding energy (∼60 meV) of AlN.21 The electrically injected devices were fabricated subsequently. In contrast to the previously reported III-nitride nanowire LEDs wherein polymer fillings were often used to planarize nanowire structures, no polymer was used in this work in order to eliminate any absorption of the deep UV light. Instead, a direct metal deposition, consisting of 15 nm Ni/10 nm Au, was employed to form p-metal contact on the nanowire top surface, which can enable efficient current injection into the nanowire structures, due to the highly dense and uniform nanowires. Shown in Figure 3 is the room temperature I−V

Figure 3. I−V characteristics of an AlN nanowire LED device with a size of 300 μm × 300 μm measured at different temperatures, with the inset showing the semilogarithmic plot of the room temperature I−V curve.

characteristics (solid squares), with the corresponding semilogarithmic scale plot shown in the inset. It is seen that excellent p−n junction, with a rectification factor of ∼108 at ±8 V, was formed. The device has a turn on voltage of ∼5.5 V, which agrees reasonably well with the bandgap energy of AlN. At a current density of 20 A/cm2, the forward voltage is ∼7 V, corresponding to an electrical efficiency ∼85%, which is nearly an order of magnitude higher than the previously reported cplane AlN LEDs.5 The low-temperature I−V characteristics were further measured at 260 K (solid circles) and 77 K (triangles), shown in Figure 3. It is seen that the forward bias resistance, indicated by the slope near 10 V, is nearly invariant with the temperature. In contrast to conventional AlN epilayer based devices, efficient electron and hole injection into AlN nanowires can be achieved even at low temperatures, which is directly related to the much more efficient dopant incorporation into nanowire structures, due to the significantly reduced formation energy of Al-substitutional Mg-dopant incorporation in the near-surface region of nanowires.6 The resulting hopping conduction in the impurity band explains the nearly temperature independent resistance.17 The performance characteristics of AlN nanowire LEDs operating at room temperature were first investigated. In this

Figure 2. (a) SEM image of AlN nanowire LED structures grown on Si taken with a 45° angle. (b) PL spectrum measured at room temperature with a 193 nm excitation source. 7007

DOI: 10.1021/acs.nanolett.5b03040 Nano Lett. 2015, 15, 7006−7009

Letter

Nano Letters

surface. Illustrated in Figure 5a, the EL emission peak is centered around 204 nm. Compared to the room temperature

experiment, the EL spectra were collected by a deep UV compatible optical fiber, and detected by a liquid nitrogen cooled CCD camera through a high-resolution spectrometer. The EL emission from different angles can be measured by changing the optical fiber position, schematically shown in Figure 4a. The 90° detection angle (with respective to the y axis

Figure 5. Injection current dependent EL emission characteristics of AlN nanowire LEDs measured at 77 K. (a) Injection current density dependent spectra from 0.5 to 10 A/cm2 (device size: 1 mm × 1 mm). (b) Light output vs the injection current density (device size: 300 μm × 300 μm). (c) Derived relative EQE as a function of the injection current density for AlN nanowire LEDs with areal sizes of 1 mm × 1 mm (blue circle) and 300 μm × 300 μm (red square). The solid blue and red curves are the simulated IQE using the ABC model.

Figure 4. Room temperature EL emission characteristics of AlN nanowire LEDs. (a) Schematic of the measurement setup. The angle (θ) is with respect to the y axis. (b) Injection current density dependent EL spectra measured from the nanowire top surface (θ = 90°) under pulsed biasing conditions. The dashed line is a guide to the eye. (c) Variations of the light intensity vs the injection current density for θ = 90°. (d) Angle dependent EL spectra measured at 36 A/cm2 under CW operation.

EL spectra, the wavelength is blue-shifted due to the increased bandgap energy.17 In addition, it is important to note that this peak was also measured when the bias voltage was below 6 V, that is, below the bandgap energy of AlN. The direct measurement of EL emission under biasing conditions below the energy bandgap of AlN suggests the extremely efficient current injection into the device active region. The peak wavelength is nearly independent of the injection current, confirming a stable exciton emission. The integrated EL intensity as a function of the injection current density is shown in Figure 5b. It is seen that as the injection current increases, the light output intensity first increases rapidly, and then shows saturation, followed by a drop. The derived relative external quantum efficiency (EQE, defined by the light output divided by the current density) is shown in Figure 5c. It is seen that the EQE exhibits a sharp increase as the injection current increases (up to 3 A/cm2), and then decreases rapidly, which can be mainly attributed to the lack of carrier confinement in the active region, and the resulting large electron overflow. To further understand the EQE trend, we considered the ABC model,

in the schematic) means that the fiber is positioned directly on the top surface of the device. We first measured the EL emission from the device top surface (θ = 90°). Figure 4b shows the EL spectra measured from 28 A/cm2 to 110 A/cm2 under a pulsed electrical injection with a duty cycle of 5% for a device size of 500 μm × 500 μm. A strong EL emission peak at ∼207 nm was measured, which agrees well with the emission spectrum measured under optical pumping (Figure 2b). The peak position is nearly independent of the injection current, indicating a negligible heating effect and highly stable exciton emission. The integrated EL intensity as a function of the injection current is shown in Figure 4c, and it is seen that the intensity increases nearly linearly. We have subsequently investigated the angle dependent EL emission by changing the fiber position while biasing the device under continuous wave (CW) conditions. Illustrated in Figure 4d, as the detection angle reduces, that is, as the fiber is moved from the top surface to the side of the device, the EL intensity is reduced significantly, suggesting the dominant emission is from the nanowire top surface. At a detection angle θ = 30°, the light intensity is 20% of that measured directly from the top surface. The light intensity diminishes to negligible values with further reducing the detection angle. For comparison, due to the predominant TM polarization for AlN, the emitted photons cannot be extracted efficiently from the top surface for conventional c plane AlN LEDs.5,13 To understand the underlying mechanism of the preferred surface emission of AlN nanowire LEDs, we have performed detailed simulation of TM polarized light propagation in randomly distributed nanowire arrays (see Supporting Information). It was found that in certain nanowire diameter and spacing ranges, due to the multiple light coupling and scattering process among nanowires, the dominant light emission can be from the nanowire top surface, that is, along the c axis, albeit the light is TM polarized. These studies provided a new approach to efficiently extract TM polarized light from AlN LEDs. We have further performed detailed measurements of the LED performance at low temperature (77 K) for various device sizes. The EL emission was measured from the nanowire top

ηi =

BN 2 AN + BN 2 + CN3

(1)

where ηi is the IQE, N is the carrier density in the device active region, and A, B, and C are the Shockley−Read−Hall nonradiative recombination coefficient, radiative recombination coefficient, and Auger recombination coefficient and/or related to electron overflow, respectively. The simulated IQE using the above ABC model is shown by the solid curve in Figure 5c. It is seen that the measured relative EQE can be well simulated using this model. The A, B, and C are estimated to be 5.2 × 106 s−1, 3.7 × 10−9 cm3 s−1, and 6.4 × 10−26 cm6 s−1 from the EQE curve of a device with a size of 1 mm × 1 mm, and 4 × 106 s−1, 3.7 × 10−9 cm3 s−1, and 4.5 × 10−26 cm6 s−1 from the EQE curve of a device with a size of 0.3 mm × 0.3 mm, respectively, illustrated in Figure 5c. Compared to the previously reported InGaN/GaN nanowire LEDs, wherein A is in the range of 107− 108 s−1,22,23 the derived A value for AlN nanowire LEDs is much smaller, which is consistent with the smaller surface recombination velocity of AlN compared to InGaN. In contrast, the extremely large C value derived here can be mainly 7008

DOI: 10.1021/acs.nanolett.5b03040 Nano Lett. 2015, 15, 7006−7009

Letter

Nano Letters

(9) Carnevale, S. D.; Kent, T. F.; Phillips, P. J.; Mills, M. J.; Rajan, S.; Myers, R. C. Nano Lett. 2012, 12, 915. (10) Wang, Q.; Nguyen, H. P. T.; Cui, K.; Mi, Z. Appl. Phys. Lett. 2012, 101, 043115. (11) Lee, Y.-B.; Wang, T.; Liu, Y.-H.; Ao, J.-P.; Izumi, Y.; Lacroix, Y.; Li, H.-D.; Bai, J.; Naoi, Y.; Sakai, S. Jpn. J. Appl. Phys. 2002, 41, 4450. (12) Bhattacharyya, A.; Moustakas, T. D.; Zhou, L.; Smith, D. J.; Hug, W. Appl. Phys. Lett. 2009, 94, 181907. (13) Taniyasu, Y.; Kasu, M. Appl. Phys. Lett. 2010, 96, 221110. (14) Hirayama, H.; Noguchi, N.; Fujikawa, S.; Norimatsu, J.; Kamata, N.; Takano, T.; Tsubaki, K.; Litton, C. W.; Chyi, J.-I.; Nanishi, Y.; Piprek, J.; Yoon, E. Proc. SPIE 2009, 7216, 721621. (15) Li, J.; Nam, K. B.; Nakarmi, M. L.; Lin, J. Y.; Jiang, H. X.; Carrier, P.; Wei, S.-H. Appl. Phys. Lett. 2003, 83, 5163. (16) Nepal, N.; Shakya, J.; Nakarmi, M. L.; Lin, J. Y.; Jiang, H. X. Appl. Phys. Lett. 2006, 88, 133113. (17) Connie, A. T.; Zhao, S.; Sadaf, S. M.; Shih, I.; Mi, Z.; Du, X.; Lin, J.; Jiang, H. Appl. Phys. Lett. 2015, 106, 213105. (18) Wang, Q.; Zhao, S.; Connie, A. T.; Shih, I.; Mi, Z.; Gonzalez, T.; Andrews, M. P.; Du, X. Z.; Lin, J. Y.; Jiang, H. X. Appl. Phys. Lett. 2014, 104, 223107. (19) Zhao, S.; Le, B. H.; Liu, D. P.; Liu, X. D.; Kibria, M. G.; Szkopek, T.; Guo, H.; Mi, Z. Nano Lett. 2013, 13, 5509. (20) de la Mata, M.; Magen, C.; Gazquez, J.; Utama, M. I.; Heiss, M.; Lopatin, S.; Furtmayr, F.; Fernandez-Rojas, C. J.; Peng, B.; Morante, J. R.; Rurali, R.; Eickhoff, M.; Fontcuberta i Morral, A.; Xiong, Q.; Arbiol, J. Nano Lett. 2012, 12, 2579. (21) Wu, J. J. Appl. Phys. 2009, 106, 011101. (22) Nguyen, H. P.; Djavid, M.; Cui, K.; Mi, Z. Nanotechnology 2012, 23, 194012. (23) Guo, W.; Zhang, M.; Bhattacharya, P.; Heo, J. Nano Lett. 2011, 11, 1434. (24) Nguyen, H. P.; Zhang, S.; Connie, A. T.; Kibria, M. G.; Wang, Q.; Shih, I.; Mi, Z. Nano Lett. 2013, 13, 5437. (25) Wang, Q.; Connie, A. T.; Nguyen, H. P.; Kibria, M. G.; Zhao, S.; Sharif, S.; Shih, I.; Mi, Z. Nanotechnology 2013, 24, 345201.

attributed to the electron overflow in the presented AlN LEDs with a homojunction structure. Although carrier confinement and surface passivation can be introduced by employing large bandgap materials in previously reported InGaN visible and AlGaN UV nanowire LEDs,24,25 it has remained challenging to realize effective carrier confinement in AlN LEDs due to the large bandgap of AlN. Boron nitride (BN) could be one possible solution to introduce carrier confinement into AlN LEDs, which is worth of further investigation. In conclusion, we have performed detailed studies of the optical and electrical performance of AlN nanowire LEDs. It was found that efficient light extraction could be achieved from the device top surface, albeit the light is TM polarized. This could provide a viable path to realize electrically injected surface emitting vacuum UV light sources that were not previously possible. In addition, AlN nanowire LEDs with an extremely large electrical efficiency (>85%) was demonstrated, for the first time. Our detailed studies further suggest that the performance of AlN nanowire LEDs is predominantly limited by electron overflow. Future work could include exploring large bandgap materials (>6 eV) to improve the carrier confinement in AlN LEDs, and using patterned substrate to realize precise control of the nanowire diameter and spacing to achieve more efficient light emission.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.5b03040. Simulation of the nanowire size-dependent light extraction efficiency of the TM polarized light. (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and U.S. Army Research Office under Grant W911NF-15-1-0168. Part of the work was performed in the McGill University Micro Fabrication Facility. The authors would like to thank Mr. Xianhe Liu for the help on Figure 1.

■

REFERENCES

(1) Khan, A.; Balakrishnan, K.; Katona, T. Nat. Photonics 2008, 2, 77. (2) Shur, M. S.; Gaska, R. IEEE Trans. Electron Devices 2010, 57, 12. (3) Hirayama, H.; Maeda, N.; Fujikawa, S.; Toyoda, S.; Kamata, N. Jpn. J. Appl. Phys. 2014, 53, 100209. (4) Muramoto, Y.; Kimura, M.; Nouda, S. Semicond. Sci. Technol. 2014, 29, 084004. (5) Taniyasu, Y.; Kasu, M.; Makimoto, T. Nature 2006, 441, 325. (6) Zhao, S.; Connie, A. T.; Dastjerdi, M. H.; Kong, X. H.; Wang, Q.; Djavid, M.; Sadaf, S.; Liu, X. D.; Shih, I.; Guo, H.; Mi, Z. Sci. Rep. 2015, 5, 8332. (7) Yanagihara, M.; Yusop, M. Z.; Tanemura, M.; Ono, S.; Nagami, T.; Fukuda, K.; Suyama, T.; Yokota, Y.; Yanagida, T.; Yoshikawa, A. APL Mater. 2014, 2, 046110. (8) Sekiguchi, H.; Kishino, K.; Kikuchi, A. Electron. Lett. 2008, 44, 151. 7009

DOI: 10.1021/acs.nanolett.5b03040 Nano Lett. 2015, 15, 7006−7009