Evaluating the size-dependent quantum efficiency loss in a SiO2-Y2O3 hybrid gated type-II InAs/GaSb long-infrared photodetector array G Chen, A M Hoang, and M Razeghi Citation: Applied Physics Letters 104, 103509 (2014); doi: 10.1063/1.4868486 View online: http://dx.doi.org/10.1063/1.4868486 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/104/10?ver=pdfcov Published by the AIP Publishing This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions Downloaded to IP: 108.68.169.180 On: Fri, 14 Mar 2014 18:30:59 APPLIED PHYSICS LETTERS 104, 103509 (2014) Evaluating the size-dependent quantum efficiency loss in a SiO2-Y2O3 hybrid gated type-II InAs/GaSb long-infrared photodetector array G Chen, A M Hoang, and M Razeghia) Center for Quantum Devices, Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, Illinois 60208, USA (Received January 2014; accepted 25 February 2014; published online 14 March 2014) Growing Y2O3 on 20 nm SiO2 to passivate a 11 lm 50% cut-off wavelength long-wavelength infrared type-II superlattice gated photodetector array reduces its saturated gate bias (VG,sat) to À7 V Size-dependent quantum efficiency (QE) losses are evaluated from 400 lm to 57 lm size gated photodiode Evolution of QE of the 57 lm gated photodiode with gate bias and diode operation bias reveals different surface recombination mechanisms At 77 K and VG,sat, the 57 lm gated photodiode exhibits QE enhancement from 53% to 63%, and it has 1.2  10À5 A/cm2 dark current C 2014 AIP Publishing LLC density at À200 mV, and a specific detectivity of 2.3  1012 Jones V [http://dx.doi.org/10.1063/1.4868486] InAs/GaSb type-II superlattice (T2SL) has been making rapid improvements,1–6 but its performance has not reached its theoretical limits7 because of surface leakage current, which is more severe in the long-wavelength-infrared-regime (LWIR) Although different passivation techniques have been attempted,8–12 surface leakage still limits its performance Recently, gating technique demonstrated its capability of eliminating surface leakage current,13,14 restoring quantum efficiency (QE), and focal plane array (FPA) application.15 To suppress the high saturated gate bias (VG,sat), reducing dielectric layer thickness and applying high-k dielectric, yttrium sesquioxide (Y2O3), were both attempted but ultrathin dielectric layer is impractical for FPA application Therefore, high-k dielectric becomes the only option Thanks to its high dielectric constant, 600 nm thick Y2O3 passivated 100 lm pitch gated array was realized with times lower VG,sat than the one passivated with SiO2.15 However, thinning Y2O3 below 600 nm still resulted in high gate leakage current and reduced the fabrication yield, which made realization of small pitch, low VG,sat gated array difficult Therefore, improving the quality of Y2O3 will be important for further suppressing the VG,sat and scaling down pixel size Most importantly, the QE loss with different mesa sizes has not been evaluated and the evolution of the QE with gate bias (VG) and diode operation bias (VOP) can reveal different surface recombination mechanisms, which are very important information for further reducing pixel size for higher resolution FPA In this Letter, we reported a method of improving Y2O3 quality to increase the fabrication yield, evaluated the mesa size-dependent QE loss from LWIR T2SL Pỵ-p-M-Nỵ gated photodetector and revealed the evolution of surface recombination with VG and VOP Same structure material as Ref 15 was processed into one die containing unpassivated diode (UPD) for reference, 17 dies of single gated photodetectors and one die of gated photodetector arrays The 17 single gated photodetector samples contained both gated diode (GD) and ungated diode (UGD), and differed from each other by passivation methods and passivation layer thicknesses Sample S1 to S6 were passivated with SiO2 by plasma-enhanced chemical vapor deposition (PECVD), sample Y1 to Y6 were passivated with Y2O3 by ion-beam sputtering deposition (IBD), and their thickness were reported in Table II Sample H1 to H5 used SiO2-Y2O3 hybrid passivation by first depositing 20 nm thick SiO2 using PECVD and then depositing additional 50 nm, 100 nm, 200 nm, 300 nm, and 600 nm of Y2O3 using IBD The gated photodetector arrays contained square detector arrays with pixel size of 400 lm, 320 lm, 250 lm, 150 lm, 120 lm, 57 lm, and 37 lm 220 nm thick SiO2-Y2O3 hybrid passivation layer was used for array fabrication The processing detail was reported in Ref 15 As shown in Figure 1(a) and Table I, the dark current densities of UGD with SiO2, Y2O3, and SiO2-Y2O3 hybrid passivation are similar and all better than UPDs The VG,sat of each sample is plotted in Figure 1(b) and reported in Table II Samples passivated with Y2O3 (Y1 to Y6) show $2.7 times lower VG,sat than those using SiO2 (S1 to S6) Since the dielectric constant of SiO2 is $3.9 and the interface charge density between the SiO2/T2SL and Y2O3/T2SL are similar,14 the dielectric constant of Y2O3 grown on etched T2SL sidewall surface is $11, which is around the lowest value reported.16 The capacitance of the sample with SiO2-Y2O3 hybrid passivation can be expressed by Eq (1), where eSiO2 and eHybrid Y2 O3 are dielectric constants of SiO2 and Y2O3 grown on thin SiO2, dSiO2 and dY2 O3 are the thicknesses of SiO2 and Y2O3 dielectric layers, and A is the sidewall surface area Since samples with SiO2-Y2O3 hybrid passivation and the SiO2 passivation have the same SiO2/T2SL interface, their interface charge densities are the same and the dielectric constant of Y2O3 grown on thin SiO2 (eHybrid Y2 O3 ) can be calHybrid culated from Eq (2), where VSiO are the G;sat and VG;sat saturated gate biases of SiO2 and SiO2-Y2O3 hybrid passivation gated sample eHybrid Y2 O3 is estimated to be 20.8–22.6, which is around the highest value reported16 CHybrid ¼ a) Email: razeghi@eecs.northwestern.edu 0003-6951/2014/104(10)/103509/4/$30.00 104, 103509-1 eHybrid Y2 O3 eSiO2 eSiO2 dY2 O3 ỵ eHybrid Y2 O3 dSiO2 A; (1) C 2014 AIP Publishing LLC V This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions Downloaded to IP: 108.68.169.180 On: Fri, 14 Mar 2014 18:30:59 103509-2 Chen, Hoang, and Razeghi Appl Phys Lett 104, 103509 (2014) FIG (a) Comparison of dark current density between UPD and UGD different passivations (b) Correlation between saturated gate bias and passivation layer thickness of samples with different passivations (c) The yield of single gated photodetector with different passivations SiO e2SiO dY2 O3 VG;sat2 eHybrid Y2 O3 ¼ dSiO2 ỵdY2 O3 VHybrid G;sat e SiO d SiO2 Y2 O3 eSiO2 dSiO ỵdY O 2 VG;sat2 : (2) VHybrid G;sat The fabrication yield of 220 nm thick SiO2-Y2O3 hybrid passivated gated photodetector reaches 90% and is higher than those with SiO2 and Y2O3 passivation because of Y2O3 quality improvement (Figure 1(c)) Therefore, SiO2-Y2O3 hybrid gated array with 220 nm thick passivation layer with small pitch can be realized with VG,sat ¼ À7 V The average I–V characteristics of the 37 lm GDs at VG ¼ and À7 V are shown in Figure At À200mV operation bias, where the QE is saturated, the GDs exhibit four orders of magnitude lower dark current density than the one at zero gate bias and reach 1.2  10À5 A/cm2 The differential resistance area product at À200 mV (RAÀ200 mV) is 2.8  104 X cm2 The dark current density of the SiO2-Y2O3 hybrid gated array at VG,sat is similar as the ones with SiO2 and Y2O3 passivated single gated photodetector (Table I) Since the optical response of 37 lm pixel was close to the measurement system noise level, 57 lm pixel is used to represent the optical property of the gating array The spectral QE curves of the 57 lm GD array at VOP ¼ À200 mV and at VG ¼ and À7 V are shown in Figure At VG ¼ V, the QE at peak responsivity (QEpeak À200mV , k ¼ 8.6 lm) is 53% At VG ¼ À7 V, the QEpeak À200mV increases to 63%, which is 18.9% increment This increment is due to the suppression of surface recombination.15 As predicted, the surface assisted QE loss should have mesa size-dependent behavior, which is illustrated in Figure The QE at VG ¼ decreases linearly with P/A, where A and P are area and perimeter of the pixel, respectively At VG ¼ À7 V, due to minimizing of surface recombination, diodes with different sizes exhibit the same level of QE, QEpeak À200mV ¼ 63% For the large size mesa, the loss of QE is negligible The QEpeak À200mV of 400 lm pixel increases from 61.7% at VG ¼ to 63% at VG ¼ À7 V As shown in Eq (3), the measured photocurrent density (Jphoton) consists of one or more of the following four components: photocurrent density generated in the bulk (Jbulk), the recombination current density within the depletion region of the p-n junction (Jp-n), the recombination current density within the surface depletion region (JSDR), and the surface recombination current density (JS) (Jbulk À Jp-n) reaches maximum value at VOP ¼ À200 mV because the material has condition band misalignment between the M-barrier barrier and p-region Since Jbulk and Jp-n are independent of the diode size, the following analysis will be focused on JSDR and JS When the sidewall surface is accumulated or in flat band condition, JSDR and JS not contribute to the recombination of the photocurrent.17 When the sidewall surface is depleted, both JSDR and JS contribute to the recombination of the photocurrent The magnitude of JSDR depends on VG because the surface depletion width changes with VG, and the magnitude of JS is strongly related to the SiO2/T2SL interface trap density When the surface is inverted, JS not contribute to the recombination of the photocurrent anymore because all SiO2/T2SL interface traps are filled, which means they are not activated and can not acted as recombination-generation centers In this case, the recombination in the surface depletion region (JSDR) becomes maximum because the surface depletion width is in maximum, xdmax, whose value depends on the VOP.17 The expressions of each component are shown in Eqs (3)–(7),17 where ni is the intrinsic carrier concentration, which is around 5.9  1014 cmÀ3 as reported in Ref 5, W and xdmax are the bulk depletion width and the maximum surface depletion width, s0 pÀn and s0 SDR are the carrier life time in the bulk and surface depletion region, d is the mesa depth, which is 5.5 lm, s0 is surface recombination velocity, KS is the superlattice dielectric constant, which is 15.4,5 e0 is the vacuum permittivity, q is the electron charge, NA is the acceptor concentration, which is $1016 cmÀ3, and /Fp is the quasi-Fermi potential, which is 23.3 mV TABLE I Dark current density at À200 mV (JÀ200 mV) of UPD, UGD, and GD at VG,sat UPD UGD GD Passivation N/A SiO2 Y2O3 Hybrid SiO2 Y2O3 Hybrid JÀ200mV (A/cm2) 0.27 8.1  10À2 7.1  10À2 5.5  10À2 2.1  10À5 1.5  10À5 1.2  10À5 This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions Downloaded to IP: 108.68.169.180 On: Fri, 14 Mar 2014 18:30:59 103509-3 Chen, Hoang, and Razeghi Appl Phys Lett 104, 103509 (2014) TABLE II Saturated gate bias of different types of passivation and passivation layer thicknesses Sample Passivation thickness (nm) VG,sat (V) SiO2 Y2O3 SiO2-Y2O3 hybrid S1/Y1 S2/Y2/H1 S3/Y3/H2 S4/Y4/H3 S5/Y5/H4 S6/Y6/H5 20 2.2 N/A 70 16 120 22 8 220 37 14 320 53 20 11 620 95 34 20 Jphoton ¼ Jbulk À JpÀn À JSDR À JS ; ni q W; s0 pÀn ni d P JSDR ¼ q xdmax s0 SDR A P JS ¼ qni s0 d ; A sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  Ãffi 2Ks e0 VOP ỵ 2/Fp xdmax ẳ : qNA Jpn ẳ (3) (4) (5) (6) (7) As shown in Figure 5, the QE is evaluated with both VG and VOP First, we fix the VG at V and À7 V, and gradually change VOP from to À500 mV At VG ¼ V (magenta curve), the sidewall surface is inverted,14 and JS does not contribute to the recombination of the photocurrent JSDR is maximum and its maximum value increases with VOP increases Therefore, the QE reaches its maximum value of 53% at VOP ¼ À200 mV, where the (Jbulk À Jp-n) is maximum, and then the QE starts decreasing At VG ¼ V and VOP ¼ À200 mV, xdmax is estimated to be around 0.2 lm; and in this case, JSDR is the only term causes the P/A dependent QE behavior thus s0 SDR can be estimated from the slope of the fitting line in Figure and is estimated to be around 5.3 ps At VG ¼ À7 V (blue curve), the surface is under accumulation, the surface depletion region vanishes and both JSDR and JS not contribute to the photocurrent Therefore, the QE reaches maximum value of 63% at VOP ¼ À200 mV and stays in the same level up to VOP ¼ À500 mV Second, we fix the VOP at V and À200 mV, and gradually change VG from to À12 V The evolutions of the QE curves of VOP ¼ (red curve) and À200 mV (green curve) are similar except that the QE at VOP ¼ À200 mV is higher than VOP ¼ mV because of the bias-dependent QE behavior When Ϲ VG Ϲ À1 V, the QE does not change with VG because the sidewall surface is still inverted, the contribution of JSDR does not change, and there is no contribution from JS When À1 Ϲ VG Ϲ À3 V, the sidewall surface is depleted, and the SiO2/T2SL interface traps become active, which means both JS and JSDR cause the reduction of the QE When VG Ϲ À3 V, the JSDR keeps decreasing because the surface depletion width decreases, resulting in the increasing of the QE At VG ¼ À7 V, the surface is accumulated, therefore, and the QE reaches maximum value of 63% and stays the same as VG increases Therefore, there are different surface recombination mechanisms associated with the QE loss and their influences depend on the surface condition FIG Comparison of the dark current density and differential resistancearea product of 37 lm pixel at VG ¼ and À7 V FIG Saturated spectral QE of the 57 lm SiO2-Y2O3 hybrid passivated GDs at Vop ¼ À200 mV and at VG ¼ and À7 V FIG QE at peak responsivity (k ¼ 8.6 lm) and at Vop ¼ À200 mV of different sizes pixels at VG ¼ and VG ¼ À7 V This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions Downloaded to IP: 108.68.169.180 On: Fri, 14 Mar 2014 18:30:59 103509-4 Chen, Hoang, and Razeghi Appl Phys Lett 104, 103509 (2014) In summary, we showed that the SiO2-Y2O3 hybrid passivation improved the quality of Y2O3, resulting in much higher dielectric constant and improvement of fabrication yield, which realized the 37 lm pitch gated array and further suppressed the VG,sat to À7 V More importantly, the evolution of QE with different pixel sizes, VG, and VOP revealed different surface recombination mechanisms and carrier life time in the surface depletion region Thanks to the gating technique to suppress the surface recombination, at 77 K, the QE of the 57 lm size detectors was improved by 18.9% At VG,sat (VG ¼ À7 V), the 57 lm gated diode array exhibits JÀ200 mV of 1.2  10À5 A/cm2, RAÀ200 mV of 2.8  104 X cm2, a 63% quantum efficiency, and a detectivity of 2.3  1012 Jones FIG QE mapping of 57 lm pixel in the SiO2-Y2O3 hybrid gating array with VG and VOP Top inset: The evolution of QE with VOP at VG ¼ and À7 V Right inset: The evolution of QE with VG at VOP ¼ and À200 mV The authors acknowledge the support, interest, and scientific discussion of Dr Fenner Milton, Dr Meimei Tidrow, Dr Joseph Pellegrino, and Dr Sumith Bandara from the U.S Army Night Vision Laboratory; Dr William Clark, Dr Priyalal Wijewarnasuriya, and Dr Eric DeCuir, Jr from U.S Army Research Laboratory; Dr Nibir Dhar from DARPA; and Dr Murzy Jhabvala from NASA Goddard Space Flight Center A M Hoang, G Chen, A Haddadi, S Abdollahi Pour, and M Razeghi, Appl Phys Lett 100, 211101 (2012) B.-M Nguyen, G Chen, A M Hoang, S A Pour, S Bogdanov, and M Razeghi, Appl Phys Lett 99, 033501 (2011) S A Pour, E K.-w Huang, G Chen, A Haddadi, B.-M Nguyen, and M Razeghi, Appl Phys Lett 98, 143501 (2011) D Z.-Y Ting, C J Hill, A Soibel, S A Keo, J M Mumolo, J Nguyen, and S D Gunapala, Appl Phys Lett 95, 023508 (2009) G Chen, A M Hoang, S Bogdanov, A Haddadi, P R Bijjam, B.-M Nguyen, and M Razeghi, Appl Phys Lett 103, 033512 (2013) B M Nguyen, G Chen, M A Hoang, and M Razeghi, IEEE J Quantum Electron 47, 686 (2011) D L Smith and C Mailhiot, J Appl Phys 62, 2545 (1987) A Gin, Y Wei, J Bae, A Hood, J Nah, and M Razeghi, Thin Solid Films 447–448, 489 (2004) O Salihoglu, A Muti, K Kutluer, T Tansel, R Turan, C Kocabas, and A Aydinli, J Appl Phys 111, 074509 (2012) 10 H S Kim, E Plis, A Khoshakhlagh, S Mayer, N Gautam, Y D Sharma, L R Dawson, S Krishna, S J Lee, and S K Noh, Appl Phys Lett 96, 033502 (2010) 11 A Gin, Y Wei, A Hood, A Bajowala, V Yazdanpanah, M Razeghi, and M Z Tidrow, Appl Phys Lett 84, 2037 (2004) 12 E Plis, J B Rodriguez, S J Lee, and S Krishna, Electron Lett 42, 1248 (2006) 13 G Chen, B.-M Nguyen, A M Hoang, E K Huang, S R Darvish, and M Razeghi, Appl Phys Lett 99, 183503 (2011) 14 G Chen, E K Huang, A M Hoang, S Bogdanov, S R Darvish, and M Razeghi, Appl Phys Lett 101, 213501 (2012) 15 G Chen, A M Hoang, S Bogdanov, A Haddadi, S R Darvish, and M Razeghi, Appl Phys Lett 103, 223501 (2013) 16 L Manchanda and M Gurvitch, IEEE Electron Device Lett 9, 180 (1988) 17 A S Grove and D J Fitzgerald, Solid-State Electron 9, 783 (1966) FIG Specific detectivity mapping of 57 lm pixel in the SiO2-Y2O3 hybrid gating array with VG and VOP Figure exhibits the evolution of specific detectivity (D*) of the 57 lm pixel with VG and VOP The high D* region (D* > 1012 Jones) mainly falls in the high VG and low VOP region That is because the increment of VG can suppress surface leakage current, resulting in lower dark current density and higher value of differential RA product; and the increment of VOP, on the contrary, results in higher dark current and reduction of the RA value Moreover, as VG increases, the high D* region can extend to higher VOP, which means the high performance operation region of the gated pixel increases That is because by eliminating the surface leakage current, the dark current density of the pixel become less sensitive with VOP The maximum D* value is 2.3  1012 Jones at VOP ¼ À100 mV and VG ¼ À7 V This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions Downloaded to IP: 108.68.169.180 On: Fri, 14 Mar 2014 18:30:59