This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.
Article http://pubs.acs.org/journal/acsodf
Ultraviolet Wavelength Identification Using Energy Distribution of Hot Electrons Zhiguo Yu,† Lei Liu,†,‡ Kaiyou Wang,‡,§ Junxi Wang,†,‡ Jinmin Li,†,‡ and Lixia Zhao*,†,‡ †
Semiconductor Lighting Research and Development Center, Institute of Semiconductors, Chinese Academy of Sciences, A35 Qinghua East Road, Haidian District, Beijing 100083, P. R. China ‡ College of Materials Sciences and Opto-Electronic Technology, University of Chinese Academy of Sciences, A19 Yuquan Road, Shijingshan District, Beijing 100049, P. R. China § SKLSM, Institute of Semiconductors, Chinese Academy of Sciences, P.O. Box 912, Beijing 100083, P. R. China S Supporting Information *
ABSTRACT: Light wavelength identification is essential for many optical and optoelectronic applications. Here, we report a novel wavelength identification photodetector based on the energy distribution of hot electrons at the metal/insulator interface. The information of the light wavelength can be stored in the energy distribution of the hot electrons, which can then be readout in the form of the current−voltage characteristics. On the basis of this principle, the highreliability wavelength identification of the monochromatic light has been realized with a simple Al/SiO2/Si structure. The device has an excellent stability with dark current below 1 × 10−7 A/ m2. Moreover, the wavelength of the monochromatic light in the deep ultraviolet range can be identified. This new principle will pave a new solution to design high-performance single-chip wavelength identification photodetectors and integrated miniaturized wavelength identification systems.
■
INTRODUCTION Light wavelength identification is critical for a wide range of applications, such as intelligent color recognition,1,2 automatic sorting system, tunable diode lasers,3 missile threat warning,4 or optical interconnects and plasmonic nanocircuits.5,6 Because of limitations of the carrier transport mechanism, conventional semiconductor photodetectors do not have the capability to identify light wavelength. To achieve wavelength identification, they need the aid of many kinds of auxiliary components, including grating,7,8 filter,9 macrobending single-mode fiber,10 distributed Bragg reflector,11 and resonant cavity.12 Recently, hot-electron photodetectors have been demonstrated to directly identify the wavelength of monochromatic light with a simple metal/insulator/metal (MIM) structure,13,14 which significantly reduces the cost and complication of the system and can be compatible with high-compact photonic integration. In these MIM junctions, the hot electrons photoexcited in the top metal layer could emit into the bottom metal layer over the conduction band of the middle insulation layer, inducing a detectable open circuit voltage (Voc). In the case that the hot electrons are not scattered in the insulation layer, the Voc can correspond one-to-one to the wavelength of the monochromatic light.13 On the basis of this principle, wavelength identification of monochromatic light in the infrared and visible range has been realized. However, the thickness of the middle insulation layer in these photodetectors must be ultrathin, typically less than 10 nm,13,15−17 to make sure that the hot electrons pass through the insulation layer © 2017 American Chemical Society
without scattering and energy loss. Such an ultrathin insulation layer would cause an intolerable leakage current, leading to device instability in practical applications.18 In addition, because the mean free path of the excited hot electrons in the ultraviolet range is much shorter than that in the visible or infrared range,15,19,20 the hot electrons may suffer strong scattering in the insulation layer as the wavelength decreases to the ultraviolet range, which will restrict the wavelength identification to be further extended to the ultraviolet or even shorter region. Here, we report on a novel wavelength identification photodetector based on the energy distribution of the hot electrons at the metal/insulator interface. Monochromatic light with a particular wavelength can excite its unique energy distribution of the hot electrons and lead to a corresponding unique current−voltage (I−V) characteristic. This wavelength identification is determined by the properties of the hot electrons just located at the metal/insulator interface and less restricted by transport behaviors (such as hot-electron scattering) in the insulation layer. Thus, the thickness of the insulator can be up to several tens of nanometers, which is about 1 order of magnitude larger than that of the mean free path, making for an excellent device performance and a very low dark current (∼1 × 10−7 A/m2). Meanwhile, the Received: April 12, 2017 Accepted: July 6, 2017 Published: July 18, 2017 3710
DOI: 10.1021/acsomega.7b00441 ACS Omega 2017, 2, 3710−3715
ACS Omega
Article
Figure 1. (a) Schematic diagram and cross-sectional SEM images of the Al/SiO2/Si structure. (b) The I−V curves of the hot-electron photodetector with a 40 nm SiO2 insulation layer under illumination conditions of 290 nm with 60 μW light power and dark, the dark current is also shown in the inset.
Figure 2. (a) Normalized I−V curves over a wavelength range of 260−360 nm. With the incident wavelength increasing, the I−V curves gradually change from the convex to the concave curve. Normalized I−V curves under illumination at 290 and 320 nm with different powers. When the illuminated light intensity is changed from 1 to 60 μW, the normalized I−V curves overlap with each other in this power range. (b) The ratio of current at any voltage to the current at 4 V for light wavelength from 260 to 360 nm, Ix/I4, where Ix and I4 are the photocurrents at voltages of x and 4 V, respectively. (c) A practical wavelength identification setup. (d) The input voltage and output current signal under illumination wavelengths of 290, 320, and 350 nm. (e) The measured I2/I4 values at 290, 320, and 350 nm, which are precisely consistent with the current ratios obtained from the normalized I−V curves.
layer, with an area of 1 mm2 as the hot-electron emission layer, chosen for the relatively low Al/SiO2 interface barrier height (3.2 eV)21 to get a wide ultraviolet wavelength detection range. Ti/Au layers were then evaporated onto the Si substrate and Al layer for the positive and negative ohmic electrodes, respectively. When monochromatic light (e.g., 290 nm 60 μW ultraviolet light) is illuminated on the Al/SiO2/Si junction form above, photocurrent is generated due to the hot-electron internal photoemission process, as shown in Figure 1b. Under positive bias, the hot electrons photoexcited in the Al layer will cross the Al/SiO2 interface barrier to form a forward photocurrent of 2 × 10−9 A at 4 V, corresponding to a responsivity beyond 3 × 10−5 A/W; under negative bias, the hot electrons photoexcited in the Si substrate will cross the Si/SiO2 interface barrier, contributing a reverse photocurrent. In this Al/SiO2/Si junction, the forward photocurrent is much greater than the reverse one, which is due to the lower interface electron barrier (3.2 eV for Al/SiO2, 4.2
photodetector has the capability to identify deep ultraviolet monochromatic light with a resolution of ∼1 nm. To our knowledge, this is the first report in which electrons’ energy distribution can be effectively used to identify the light wavelength. Furthermore, this is the first photodetector that has wavelength identification capability in the ultraviolet range. Our finding will pave a new solution to design highperformance single-chip wavelength detectors and integrated miniaturized wavelength identification systems.
■
RESULTS AND DISCUSSION A silicon-technology-compatible Al/SiO2/Si structure was fabricated to explore wavelength identification based on the energy distribution of hot electrons. The schematic diagram and the cross-sectional scanning electron microcopy (SEM) images are shown in Figure 1a. A 40 nm thick SiO2 was formed on the bottom p-type Si substrate by thermal oxidation, followed by an electron-beam evaporation of a 15 nm-thick Al 3711
DOI: 10.1021/acsomega.7b00441 ACS Omega 2017, 2, 3710−3715
ACS Omega
Article
Figure 3. Normalized I−V curves of the hot-electron devices with SiO2 thicknesses of 5, 10, 20, 40, and 60 nm under different light wavelengths. The electric field is calculated by V/d, where V and d are the applied voltage and SiO2 thickness, respectively.
modern optical spectrum analyzers of 0.1 nm) can be achieved when the device operates at a lower temperature. In addition, this wavelength identification is power-independent. With the illuminated light power increasing from 1 to 60 μW, the photocurrent increases linearly (see the Supporting Information S4), but the normalized I−V curves overlap with each other in this power range, as shown in Figure 2a. These powerindependent properties will facilitate the practical application without considering the variable received light intensity. Although the normalized I−V curves can be used to identify the wavelength of the incident light, in practical application, it is not needed to measure the full I−V curves. The current ratio at any two voltages can already determine the corresponding I−V curve and the incident light wavelength. Figure 2b shows the ratio of current at any voltage to the current at 4 V for light wavelengths from 260 to 360 nm, Ix/I4, where Ix and I4 are the photocurrent at voltage of x and 4 V, respectively. Ix/I4 decreases monotonically with the wavelength increasing. Taking I2/I4 as an example, with the wavelength increasing from 260 to 360 nm, I2/I4 decreases monotonically from 0.5 to 0.06. In addition, similar to the case of the normalized I−V curves, the current ratio is also power-independent (see the Supporting Information S5). Therefore, a monochromatic light wavelength can be simply identified by using the Ix/I4 value. It needs to be noted that the denominator of the current ratio is not limited to be 4 V, any other voltage can also serve as the standard (see the Supporting Information S6). Figure 2c illustrates the practical setup for light wavelength identification. A square wave signal is applied to the electrodes of the hot-electron device, with the high and low levels to be 4 and 2 V, respectively. When an unknown monochromatic light is illuminated on the top Al layer of the device, by reading the output photocurrent and calculating the current ratio, the incident wavelength can be identified. Figure 2d shows the input square wave signal and the output photocurrent at wavelengths of 290, 320, and 350 nm, respectively. The cycle of the input square wave signal is chosen to be 30 s by considering the relatively slow device response time of about 100 ms (see the Supporting Information S7). The output photocurrent and the current ratio I2/I4 are substantially constant during each cycle. The mean values of I2/I4 are 0.37, 0.24, and 0.08, respectively, for these three wavelengths, as shown in Figure 2e. On the basis of the relationship between wavelength and Ix/I4 in Figure 2b, the wavelengths can be identified to be 290, 320, and 350 nm respectively, which agree well with the actual wavelengths. Therefore, using this test system, monochromatic
eV for p-Si/SiO2) and higher light absorptivity of aluminum than those of the bottom silicon. It needs to be noted that hot holes can also be generated when the device is illuminated; however, because of the considerably high interface holes barrier (5.8 eV for Al/SiO2, 4.8 eV for p-Si/SiO2), the photocurrent induced by hot holes could be neglected. Without the illumination, the dark current of this wavelength detector is below 1 × 10−13 A (corresponding to current density of 1 × 10−7 A/m2) over the entire applied voltage range, which even reaches the noise level of the test system. This ultralow dark current is attributed to the high-quality thick SiO2 insulation layer, which is 1 order of magnitude higher than that of previously reported hot-electron photodetectors,13,16,17,22 making for an excellent device stability in practical applications. The signal-to-noise ratio of the Al/SiO2(40 nm)/Si junction at +4 V can be estimated to be larger than 2 × 104 by calculating the ratio of photocurrent and dark current (here, the noise level of the test system is regarded as the maximum possible dark current). It needs to be noted that only the photocurrent can be precisely measured in a wide range of insulator thicknesses (see the Supporting Information S1). The dark current is so small that it is beyond the accuracy of the test system. Therefore, we took the accuracy value of the test system as the noise level of the device, and an underestimated signal-to-noise ratio was obtained. According to the Poole−Frenkel tunneling model, the dark current of the Al/SiO2(40 nm)/Si junction can be theoretically predicted to be lower than 5 × 10−15 A (see the Supporting Information S2). In the future, the higher-accuracy test system with lower noise should be developed to precisely measure the dark current and signal-to-noise ratio of these wavelength detectors. By changing the wavelength of the incident monochromatic light, the Al/SiO2/Si junction exhibits a very sensitive wavelength-dependent characteristic, as shown in Figure 2a. With the wavelength increasing from 260 to 360 nm, the normalized I−V curves gradually transform from a convex to a concave curve. Each wavelength corresponds to its unique normalized I−V curve. Therefore, wavelength identification can be achieved by distinguishing the shape of the normalized I−V curves, with a limiting wavelength-identification resolution of about 1 nm (see the Supporting Information S3). The resolution of this kind of device may be determined by thermal fluctuation of room-temperature electrons near the Fermi level (with an energy of kBT), which disturbs the energy distribution of the hot electrons in the initial stage of the hot electrons generation. A higher resolution (approaching or exceeding the 3712
DOI: 10.1021/acsomega.7b00441 ACS Omega 2017, 2, 3710−3715
ACS Omega
Article
Figure 4. Energy diagram of the metal/insulator interface under illumination with different wavelengths, which corresponds to the photon energy of (a) ℏω1, (b) ℏω2, and (c) ℏω3, respectively. Monochromatic light excites electrons, with energy distribution N(E, ℏω) corresponding to its wavelength, the hot electrons with energy higher than the metal/insulator barrier Φ(V) may eject into the conduction band of the SiO2 layer and produce current Ia. With the applied negative voltage decreasing, Φ(V) decreases from Φ(Va) to Φ(Vb), causing an increased photocurrent Ib. When the illuminated wavelength increases, the current ratio Ia/Ib decreases; this characteristic can be used to identify the wavelength of the monochromatic light.
band of the insulator, contributing to a detectable photocurrent Ia. With the applied negative voltage decreasing, Φ(V) decreases at a rate proportional to the square root of the applied voltage due to the image force lowering effect,25 causing more hot electrons to emit into the conduction band of the insulator and an improved photocurrent Ib. For different wavelengths of the incident light, the current ratio Ia/Ib is different, as shown in Figure 4a−c. With the wavelength increasing, the upper limit of the N(E, ℏω) moves closer to the interface barrier and the initial photocurrent becomes smaller, leading to a decreased Ia/Ib, which agrees with the experimental trends of the current ratios shown in Figure 2b. This means that the longer wavelength could excite the smaller Ia/Ib and the steeper current increases with voltage increasing. Thus, wavelength identification can be realized by distinguishing the current ratio or I−V characteristics of devices with the metal/ insulator interface. On the basis of this principle, the wavelength identification is determined by the properties of the hot electrons just located at the metal/insulator interface and less restricted by the transport behaviors (such as hot electrons scattering) in the insulation layer. Thus, a wavelength identification photodetector with a thick middle insulation layer that permits excellent device stability has been successfully realized. In addition, arbitrary wavelengths, such as visible or infrared, can also be easily identified by choosing the suitable metal/insulator barrier height with proper work functions.
light wavelengths can be easily identified by the input and output alternating current square wave signal. Even for a wide range of SiO2 thicknesses, the photodetectors still show an excellent wavelength identification capability, as shown in Figure 3. At a thickness of 5 nm, the I− V curves show a crescent-like shape and turn on at zero electric field. This zero turn-on electric field may be attributed to the tunneling or defect-assisted leakage current, which usually influences the performance of optoelectronic devices with a thin dielectric layer.23 With the thickness increasing from 5 to 20 nm, the shape of the I−V curves gradually transforms to an olive-like shape. As the thickness further increases to 60 nm, the shape of I−V curves becomes almost unchanged. This trend indicates that our wavelength identification principle is dominated by the properties of electron energy distribution just near the Al/SiO2 interface and less influenced by the transport behavior (such as scattering) in the whole SiO2 layer. In addition, the open circuit voltage is always zero, even for the thickness down to 5 nm. This is because the hot-electron scattering in insulator is very strong and they cannot pass through the SiO2 layer in the absence of the applied voltage, suggesting that it is difficult to identify the wavelength on the basis of the previously reported principle by just reading out the open circuit voltage when the wavelength is too short.13 Figure 4 shows the energy diagram of the metal/insulator interface under illumination with different wavelengths, which can help to illustrate how the wavelength can be identified by using the hot-electron energy distribution. When the metal is photoexcited, the hot electrons with energy distribution N(E, ℏω) will be generated in a wide range from Ef to Ef + ℏω,24 where Ef is Fermi energy and ℏω is incident photon energy. Note that the hot electrons would be thermalized before they reach the Al/SiO2 interface because the mean free path of the hot electrons excited by ultraviolet light is about 2 nm,15 which is far less than the thickness of the Al layer; the hot electrons will redistribute their energy with other lower energy electrons before they reach the Al/SiO2 interface; then, a high-effectivetemperature Fermi−Dirac electron distribution is formed. When a negative voltage is applied to the metal layer, the hot electrons that have energies higher than that of the metal/ insulator interface barrier Φ(V) may emit into the conduction
■
CONCLUSIONS We report a novel wavelength identification photodetector based on the energy distribution of hot electrons at the metal/ insulator interface, which established the unique and powerindependent relationship between wavelength of monochromatic light and I−V characteristics. A deep ultraviolet wavelength identification ranging from 260 to 360 nm has been successfully demonstrated. Meanwhile, the thickness of the middle insulation layer can be up to several tens of nanometers, resulting in an excellent device stability and a very low dark current. In addition, we study the influence of the insulator thickness on the I−V curves of Al/SiO2/Si junctions, confirming that our wavelength identification is determined by 3713
DOI: 10.1021/acsomega.7b00441 ACS Omega 2017, 2, 3710−3715
ACS Omega
■
the hot electrons just located at the metal/insulator interface and not limited by the thickness of the insulation layer. Our findings could pave a way for designing the single-chip wavelength identification, with a wide detection range and an excellent reliability simultaneously.
EXPERIMENTAL SECTION Sample Preparation. Al/SiO2/Si devices were fabricated on p-type (100) silicon substrates (doping density: 1−5 × 1019 cm−3, resistivity: 2−8 × 10−3 Ω cm). The substrates were cleaned by the standard silicon cleaning recipe. SiO2 was formed on p-Si substrates by thermal oxidation in oxygen atmosphere at 950 °C, the thickness was controlled by thermal oxidation time, and SiO2 layers with thicknesses of 5, 10, 20, 40, and 60 nm were formed on p-Si, respectively. Fifteen nanometers of Al layers were deposited onto the SiO2-coated substrates using electron-beam evaporation with a shadow mask. Then, a layer of Ti/Au was deposited on the substrates in the selected region where the thermal-oxidized SiO2 has been removed by buffered oxide etch and the p-Si could directly come in contact with Ti/Au to form bottom electrodes. On the other hand, another Ti/Au layer was directly deposited on the Al layer to form upper electrodes using a shadow mask. Finally, the devices were wire-bonded to a chip carrier for electrical and optical measurement. Experimental Setup. Current−voltage characteristics were measured using an Agilent B1500A. A xenon lamp provided illumination from 260 to 1100 nm, and a monochromatic light was achieved when the white light passed through a monochromator. The light spot on the devices was controlled by adjusting the lens on the optical path. ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.7b00441. Photocurrent and dark current with different insulator thicknesses; theoretically predicted dark current; wavelength identification with 1 nm resolution; linear-powerdependent photocurrent; power-independent current ratio I2/I4; normalized I−V curves and current ratios with 3 and 8 V as the standard, respectively; response time (PDF)
■
REFERENCES
(1) Yang, M.-H.; Kriegman, D. J.; Ahuja, N. Detecting faces in images: A survey. IEEE Trans. Pattern Anal. Mach. Intell. 2002, 24, 34− 58. (2) Hjelmås, E.; Low, B. K. Face detection: A survey. Comput. Vis. Image Und. 2001, 83, 236−274. (3) Coldren, L. A. Monolithic tunable diode lasers. IEEE J. Sel. Top. Quantum Electron. 2000, 6, 988−999. (4) Razeghi, M. Short-wavelength solar-blind detectors - Status, prospects, and markets. Proc. IEEE 2002, 90, 1006−1014. (5) Assefa, S.; Xia, F. N. A.; Vlasov, Y. A. Reinventing germanium avalanche photodetector for nanophotonic on-chip optical interconnects. Nature 2010, 464, 80−84. (6) Lee, S. J.; Ku, Z. Y.; Barve, A.; Montoya, J.; Jang, W. Y.; Brueck, S. R. J.; Sundaram, M.; Reisinger, A.; Krishna, S.; Noh, S. K. A monolithically integrated plasmonic infrared quantum dot camera. Nat. Commun. 2011, 2, No. 286. (7) Chen, E.; Chou, S. Y. Wavelength detector using a pair of metalsemiconductor-metal photodetectors with subwavelength finger spacings. Electron. Lett. 1996, 32, 1510−1511. (8) Kersey, A. D.; Davis, M. A.; Patrick, H. J.; LeBlanc, M.; Koo, K. P.; Askins, C. G.; Putnam, M. A.; Friebele, E. J. Fiber grating sensors. J. Lightwave Technol. 1997, 15, 1442−1463. (9) Bao, J.; Bawendi, M. G. A colloidal quantum dot spectrometer. Nature 2015, 523, 67−70. (10) Wang, Q.; Farrell, G.; Freir, T.; Rajan, G.; Wang, P. F. Low-cost wavelength measurement based on a macrobending single-mode fiber. Opt. Lett. 2006, 31, 1785−1787. (11) Nabiev, R.; Chang-Hasnain, C. J.; Eng, L. E.; Lau, K. Y. In Spectrodetector − Novel Monolithic Wavelength Reader and Photodetector, LEOS ’95 − IEEE Lasers and Electro-Optics Society 1995 Annual Meeting − 8th Annual Meeting Conference Proceedings, 1995; Vols. 1−2, pp A21−A22. (12) Kung, H. L.; Miller, D. A. B.; Atanackovic, P.; Lin, C. C.; Harris, J. S.; Carraresi, L.; Cunningham, J. E.; Jan, W. Y. Wavelength monitor based on two single-quantum-well absorbers sampling a standing wave pattern. Appl. Phys. Lett. 2000, 76, 3185−3187. (13) Wang, F.; Melosh, N. A. Power-independent wavelength determination by hot carrier collection in metal-insulator-metal devices. Nat. Commun. 2013, 4, No. 1711. (14) Gong, T.; Munday, J. N. Aluminum-based hot carrier plasmonics. Appl. Phys. Lett. 2017, 110, No. 021117. (15) Brown, A. M.; Sundararaman, R.; Narang, P.; Goddard, W. A., 3rd; Atwater, H. A. Nonradiative Plasmon Decay and Hot Carrier Dynamics: Effects of Phonons, Surfaces, and Geometry. ACS Nano 2016, 10, 957−66. (16) Chalabi, H.; Schoen, D.; Brongersma, M. L. Hot-Electron Photodetection with a Plasmonic Nanostripe Antenna. Nano Lett. 2014, 14, 1374−1380. (17) Gong, T.; Munday, J. N. Angle-Independent Hot Carrier Generation and Collection Using Transparent Conducting Oxides. Nano Lett. 2015, 15, 147−152. (18) Koh, M.; Iwamoto, K.; Mizubayashi, W.; Murakami, H.; Ono, T.; Tsuno, M.; Mihara, T.; Shibahara, K.; Yokoyama, S.; Miyazaki, S.; Miura, M. M.; Hirose, M. In Threshold Voltage Fluctuation Induced by Direct Tunnel Leakage Current Through 1.2−2.8 nm Thick Gate Oxides for Scaled MOSFETs, International Electron Devices Meeting 1998 − Technical Digest, 1998; pp 919−922. (19) Bernardi, M.; Mustafa, J.; Neaton, J. B.; Louie, S. G. Theory and computation of hot carriers generated by surface plasmon polaritons in noble metals. Nat. Commun. 2015, 6, No. 7044. (20) Berglund, C. N.; Powell, R. J. Photoinjection into SiO2 electron scattering in image force potential well. J. Appl. Phys. 1971, 42, 573−579. (21) Mönch, W. On the band structure lineup at interfaces of SiO2, Si3N4, and high-kappa dielectrics. Appl. Phys. Lett. 2005, 86, No. 122101. (22) Harutyunyan, H.; Martinson, A. B. F.; Rosenmann, D.; Khorashad, L. K.; Besteiro, L. V.; Govorov, A. O.; Wiederrecht, G.
■
■
Article
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Zhiguo Yu: 0000-0001-8921-2277 Lei Liu: 0000-0001-8157-0088 Lixia Zhao: 0000-0002-0466-247X Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China (61504132 and 11574306) and the National Basic Research and High Technology Program of China (2017YFB0403601, 2015AA03A101, and 2015AA033303). 3714
DOI: 10.1021/acsomega.7b00441 ACS Omega 2017, 2, 3710−3715
ACS Omega
Article
P. Anomalous ultrafast dynamics of hot plasmonic electrons in nanostructures with hot spots. Nat. Nanotechnol. 2015, 10, 770−774. (23) Hiroki, A.; Odanaka, S. Gate-oxide thickness dependence of hotcarrier-induced degradation in buried p-MOSFETs. IEEE Trans. Electron Devices 1992, 39, 1223−1228. (24) Brongersma, M. L.; Halas, N. J.; Nordlander, P. Plasmoninduced hot carrier science and technology. Nat. Nanotechnol. 2015, 10, 25−34. (25) Mead, C. A.; Snow, E. H.; Deal, B. E. Barrier Lowering and Field Penetration at Metal-Dielectric Interfaces. Appl. Phys. Lett. 1966, 9, 53.
3715
DOI: 10.1021/acsomega.7b00441 ACS Omega 2017, 2, 3710−3715
