pubs.acs.org/NanoLett
Intersublevel Infrared Photodetector with Strain-Free GaAs Quantum Dot Pairs Grown by High-Temperature Droplet Epitaxy Jiang Wu,† Dali Shao,† Vitaliy G. Dorogan,‡ Alvason Z. Li,‡ Shibin Li,† Eric A. DeCuir, Jr.,† M. Omar Manasreh,†,‡ Zhiming M. Wang,*,‡ Yuriy I. Mazur,‡ and Gregory J. Salamo‡ †
Department of Electrical Engineering and ‡ Arkansas Institute for Nanoscale Materials Science and Engineering, University of Arkansas, Fayetteville, Arkansas 72701 ABSTRACT Normal incident photodetection at mid infrared spectral region is achieved using the intersublevel transitions from strainfree GaAs quantum dot pairs in Al0.3Ga0.7As matrix. The GaAs quantum dot pairs are fabricated by high temperature droplet epitaxy, through which zero strain quantum dot pairs are obtained from lattice matched materials. Photoluminescence, photoluminescence excitation optical spectroscopy, and visible-near-infrared photoconductivity measurement are carried out to study the electronic structure of the photodetector. Due to the intersublevel transitions from GaAs quantum dot pairs, a broadband photoresponse spectrum is observed from 3 to 8 µm with a full width at half-maximum of ∼2.0 µm. KEYWORDS Droplet epitaxy, infrared photodetector, quantum dot pair, photoluminescence
F
taxy technique. An 0.5 µm n-type GaAs contact layer with Si doped to 2 × 108 cm-3 is grown at 580 °C following a 0.5 µm undoped GaAs buffer layer. An Al0.3Ga0.7As layer of 50 nm is deposited as a spacer. Subsequently, As-valve is fully closed while the substrate is cooled down to 550 °C and Gallium coverage of 12 ML (monolayer) based on an equivalent amount of GaAs on (100) orientation is supplied to form Gallium droplets. These droplets are then crystallized into GaAs QDPs by supplying As4 flux of ∼1.2 × 10-5 Torr at the same temperature for 2 min. The GaAs QDPs are doped with Silicon equivalent to 5 × 1017 cm-3. The substrate temperature is ramped back to 580 °C and capped with another 50 nm Al0.3Ga0.7As spacer layer. Ten stacked layers of QDPs are grown to form the active region of the photodetector. After the active region, a 300 nm n-type Al0.3Ga0.7As window layer with Si doped to 3 × 1018 cm-3 is grown followed by 5 nm Si doped GaAs for making the top ohmic contact. A separated uncapped QDPs sample is grown under same conditions and imaged by Atomic force microscopy (AFM). Figure 1 shows the AFM images of the uncapped sample. It is well-known that the crystallization of Gallium droplet is anisotropic; the Gallium adatoms favor diffusion along the [01-1] direction compared with [011]. In particular, this anisotropic crystallization is enhanced at a crystallization temperature of 550 °C. As a result, traditional ring-shaped nanostructures evolve as quantum dot pairs. The density of QDPs is estimated to be 1.3 × 108 cm-2 from Figure 1a. Unlike the symmetrical quantum rings obtained by lowtemperature droplet epitaxy, the QDPs consist of a small QD and a big QD. The average height of QDPs is about 9 nm and the base diameters of QDPs are about 200 and 150 nm, respectively, as shown Figure 1b.
or over one decade, infrared photodetectors have made remarkable progress driven by urgent need in numerous applications.1-3 Self-assembled quantum dots (QDs) have attracted much attention due to their potential in normal incidence and high-temperature operation.4,5 However, one of the major disadvantages of quantum dots (QDs) fabricated by Stranski-Krastanov (SK) growth mode is the defects introduced by accumulated strain in the device, which has greatly undermined the performance of SK QDs based devices.6 At the same time, droplet epitaxy has been undergoing striking development.7-15 Compared with SK growth mode, droplet epitaxy is suitable for fabrication of novel nanostructures from lattice-mismatched materials as well as lattice-matched materials.16-19 Recently, lattice-matched GaAs/AlGaAs quantum dot laser and quantum ring photodetector have been reported by employing droplet epitaxy.20,21 Although droplet epitaxy has great potential for optoelectronic devices, the conventional lowtemperature growth makes it unfavorable for practical applications, such as point defects introduced by excess arsenic during low-temperature growth.22 Fortunately, novel quantum dot pairs (QDPs) are reported by using droplet epitaxy at temperature as high as 550 °C, which provide new possibilities for high-performance optoelectronic devices.23 In this work, therefore, we report a novel infrared photodetector with lattice-matched GaAs/Al0.3Ga0.7As QDPs fabricated by high temperature droplet epitaxy. The GaAs quantum dot pairs sample is grown on a (100) semi-insulating GaAs substrate by the molecular beam epi* To whom correspondence should be addressed. E-mail:
[email protected]. Received for review: 01/21/2010 Published on Web: 03/31/2010 © 2010 American Chemical Society
1512
DOI: 10.1021/nl100217k | Nano Lett. 2010, 10, 1512–1516
FIGURE 2. Continuous wave PL spectra measured with various laser excitations at 10 K. The spectral lines from the lowest intensity to highest intensity correspond to excitation intensities I0 ) 1.2 mW/ cm2, 4 × I0, 16 × I0, 60 × I0, 250 × I0, 1.1 × 103 × I0, 3.6 × 103 × I0, 1.0 × 104 × I0, 7.8 × 104 × I0, 1 × 106 × I0, 2.6 × 106 × I0, and 1.3 × 107 × I0. GS1 and GS2 indicate ground-state transitions from big and small dots, respectively.
FIGURE 1. (a) A 5 µm × 5 µm three-dimensional AFM image of uncapped QDPs grown at 550 °C by using high temperature droplet epitaxy. (b) Cross-sectional profile (left) and a two-dimensional AFM image of a single quantum dot pair (right). The cross-sectional profile is taken along [1-10] as indicated in the AFM image.
Continuous wave photoluminescence (PL) and timeresolved (PL) of the QDP photodetector sample are investigated to gain an insight into the optical properties as well as electronic structure of QDPs. The continuous wave PL measurements are performed using the 532 nm excitation from a Nd:YAG laser with a spot diameter at the sample of 20 µm. The time-resolved PL measurements are performed using 2 ps pulses at excitation of 750 nm from a modelocked Ti:sapphire laser producing an optical pulse train at 76 MHz and a Hamamatsu synchroscan streak camera C5680 with an infrared enhanced S1 cathode is used for signal detection. The overall time resolution of the system used in the time-resolved PL measurements is 15 ps. All the PL measurements are undertaken at ∼10 K. Figure 2 demonstrates the excitation intensity dependence of the continuous wave PL spectra from I0 ) 1.2 mW/cm2 to 1.3 × 107 × I0. At low excitation intensity I0, the PL spectrum has two peaks at the 1.53 and 1.71 eV. As increasing excitation intensity, the PL spectra undergo significant changes. At excitation intensity 16 × I0, three peaks at 1.50, 1.58, and 1.90 eV start to appear. In addition, a significant blue shift is observed with increasing excitation intensity. As the excitation intensity further increases to 106 × I0, another peak appears at 1.67 eV. The peak at 1.50 and 1.90 eV can be assigned to GaAs substrate and Al0.3Ga0.7As barriers, respectively. The peaks at 1.53 and 1.58 eV are attributed to ground states e1fhh1 transition from the two GaAs QDs within each QDP. The peaks at 1.67 and 1.71 eV correspond to excited states in the GaAs QDPs. The very broad emission peak around 1.71 eV may consist of multiple interband © 2010 American Chemical Society
FIGURE 3. PL decay transients measured with various laser excitations at 10 K. The red peak is from the laser. The average excitation power is varied from 0.3 (cyan) to 95 W/cm2 (blue). The dark lines are biexponential fitting curves. The inset is fast and slow decay time as a function of excitation intensity.
transitions in QDs and the blue shift (from 1.71 to 1.80 eV) of this peak position with increasing excitation intensity is attributed to the filling of higher excited energy states of QDs. Figure 3 shows the PL decay transients measured for detection wavelengths λdet ) 810 nm (1.53 eV) under various excitation intensities with excitation wavelength λexc ) 750 nm (1.65 eV). The time-resolved PL demonstrates bicomponent exponential electron relaxation in natural log scale, which may be due to the processes involving the darkexciton state.24 As the QDs of each pair are laterally ∼100 nm apart, the coupling of the two QDs may be ruled out. Biexponential fitting, y ) y0 + As exp[-(t - t0)/ts] + Af exp[-(t - t0)/tf], is used to obtain the exciton lifetime. In the fitting equation, y0, As, and Af are the fitting parameters, while ts and tf are the slow and fast exciton decay time, respectively. As shown in the inset of Figure 3, for different excitation intensities the slow decay time is from ∼0.3 to ∼0.7 ns while the fast decay time is around ∼100 ps. The large error at low-excitation intensity is introduced by the 1513
DOI: 10.1021/nl100217k | Nano Lett. 2010, 10, 1512-–1516
are well matched to the broad PL peak at ∼1.75 eV, which verifies that this broad peak consists of multiple excited states. The interband photoconductivity spectra under bias voltages of 0.0 and -1.0 V are shown in Figure 4b. Compared with PL spectra in Figure 2, the broadband peak at ∼1.52 eV under zero bias may be attributed to transitions from the ground states of two GaAs QDs within each QDP and low order excited states. Peaks around ∼1.62 and ∼1.70 eV under zero bias correspond to the PL peaks at ∼1.67 eV and the broadband peak at 1.71-1.80 eV. However, with increasing electron injection the interband transitions from high energy excited states become dominate due to occupancy of low energy levels by electrons. These observed excited states from PLE and photoconductivity are in agreement with PL measurement. Figure 5a shows the photodetector dark-current density measured at 80 and 300 K. The dark-current density is measured by using a Keithley 4200-SCS Semiconductor Parameter Analyzer. Positive bias corresponds to electron collection from the Au0.88Ge0.12/Ni/Au top contact and injection by the bottom contact, as shown in device schematic in Figure 5b. The dark current curves show asymmetric characteristics at both 80 and 300 K, which is most likely caused by the asymmetric device structure of the QDP photodetector. At a bias voltage of 1 V, the dark current densities are measured to 5.6 × 10-8 A/cm2 at 80 K and 5.76 × 10-5 A/cm2 at 300 K, respectively. The relatively low dark current may be due to thick Al0.3Ga0.7As barriers, which also affect photocurrent. For In(Ga)As QDIPs, a thick GaAs spacer (normally 50 nm) is needed for subsequent QD growth. Without the restriction of strain, the performance of QDP photodetector can be optimized by properly choosing thinner barriers. The mid-infrared (MIR) photoresponse spectra are measured using the FTIR spectrometer with a MIR source. All photoresponse spectra are measured in the normal incidence configuration. Broadband photoresponse spectra are observed in the mid infrared spectral range covering 3.0-8.0 µm as shown in Figure 5c. The spectra are measured at 80 K for bias voltages between 0.4 and -1.4 V. This MIR photoresponse is a result of the intersublevel transitions within the conduction band of GaAs/Al0.3Ga0.7As QDPs. Under positive bias, with increasing applied voltage, the photoresponse intensity increases because of field-assisted tunneling of photoexcited electrons. The main photoresponse intensity peak is measured at λp ) 5.5 µm (225 meV) with a large full width at half-maximum (fwhm). For example, the fwhm ∆λ is ∼2.1 µm at Vbias ) 0.4 V, which leads to ∆λ/λp ) 38%. Under negative bias, the response is found insensitive to applied voltages and a second peak appears at around 3.5 µm (354 meV) at negative bias voltages. In addition, the transition wavelength corresponds to a larger energy separation than the energy transition levels measured by photoluminescence and interband conductivity measurements. Therefore, such a large spectral width and
FIGURE 4. (a) Low-intensity PL spectrum and PLE spectrum measured at 10 K. (b) Visible-near IR photoconductivity spectra measured under bias voltages of 0.0 and -1.0 V at 300 K. The arrows indicate the transition peaks.
relatively low signal amplitudes. Since biexponential decay is observed for all excitation powers, the filling of excited states can also be excluded. Even though the ground state of the small QDs can also be excited, the possibility of transferring excitons from small QDs to large QDs is small due to the large spatial separation. Therefore, we conclude that the fast exciton decay time is due to the recombination of the bright state while the slow decay time is due to the dark state. For further understanding of the nature of the electronic energy levels in the QDPs, photoluminescence excitation (PLE) and visible-near IR photoconductivity studies are carried out. A tunable Ti:sapphire laser is used for the PLE measurement and a Bruker IFS 125HR Fourier-transform infrared (FTIR) spectrometer with a visible-near IR source is used for photoconductivity measurement. For photoconductivity measurement, the QDPs sample is fabricated into normal incidence configuration photodetector with pixels of 500 × 500 µm2 by using standard photolithography techniques. Ohmic contacts are formed by evaporating and rapid thermal annealing of Au0.88Ge0.12/Ni/Au metals. Figure 4a shows low-excitation PL and PLE spectra obtained for detection at 812 nm. Because of restrictions by the experimental setup, the PLE is only able to resolve spectral range between 1.62 to 1.80 eV. Three resolved peaks are observed at ∼1.66, ∼1.71, and ∼1.76 eV, which are about 130, 180, and 230 meV above the ground-state transition. These PLE peaks © 2010 American Chemical Society
1514
DOI: 10.1021/nl100217k | Nano Lett. 2010, 10, 1512-–1516
FIGURE 5. (a) Dark current density versus voltage characteristics measured at both 80 and 300 K. (b) A schematic of a single pixel photodetector. Positive bias corresponds to collection of electrons from the top contact. (c) Mid-infrared photoresponse spectra measured at 80 K for different bias voltages.
large energy separation may be due to bound-to-continuum transitions. The mid-infrared photoresponse spectra disappear when the temperature is raised above 80 K. The relatively low temperature operation may be due to the low density of QDPs, which, in this letter, is about two orders lower than the density of QDs grown by SK mode. The performance, such as operation temperature, of the QDPs photodetector at mid-infrared band may be further improved by increasing QDP density. One way to improve QDP density is modification of substrate surface energy. For example, high-density Ga droplets (2 × 1010 cm-2 at 300 °C and 2.3 × 109 cm-2 at 400 °C) and GaAs quantum dots (∼1.6 × 1011 cm-2 at 200 °C and 3.0 × 109 cm-2 at 500 °C) have been reported using high index GaAs substrates.25,26 Additionally, as a result of zero strain in the QDPs, multiple depositions or multiple layers can be grown in order to increase density of QDPs. In conclusion, strain-free GaAs QDs have been grown by high-temperature droplet epitaxy. PL technique, PLE, and visible-near-infrared photoconductivity have revealed several confined energy states in the QDPs. Excited electron lifetimes have been measured by time-resolved photoluminescence under various excitation powers. A prototype photodetector with ten stacked layers of QDPs is fabricated into normal incident configuration. Mid-infrared photoresponse and intersublevel transitions are observed from the QDPs photodetector. The mid-infrared spectrum covers 3 to 8 µm at 80 K. As a result of the formation of threedimensional nanostructures, normal incidence operation is realized in the quantum dot pairs detector compared with conventional quantum well infrared detectors. Moreover, the © 2010 American Chemical Society
present method of employing strain-free quantum dots is applicable to long wavelength infrared detector, high-efficiency intermediate band solar cell, as well as light emission devices. Acknowledgment. This work was funded by the Air Force Office of Scientific Research DEPSCoR Grant FA9550-06-10457, the Arkansas Science and Technology Authority Grant AR/ASTA/07-ARMF-02, and MRSEC Program of NSF Grant (DMR-0520550). REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
1515
Jiang, L.; Li, S. S.; Yeh, N.; Chyi, J.; Ross, C. E.; Jones, K. S. Appl. Phys. Lett. 2003, 82, 1986. Passmore, B. S.; Wu, J.; Manasreh, M. O.; Salamo, G. J. Appl. Phys. Lett. 2007, 91, 233508. Pan, D.; Towe, E.; Kennerly, S. Appl. Phys. Lett. 1998, 73, 1937. Liu, H.; Duboz, J.; Dudek, R.; Wasilewski, Z.; Farard, S.; Finnie, P. Physica E 2003, 17, 631–633. Lim, H.; Tsao, S.; Zhang, W.; Razeghi, M. Appl. Phys. Lett. 2007, 90, 131112. Martiı`, A.; Loı`pez, N.; Antoliı`n, E.; Caı`novas, E.; Luque, A.; Stanley, C. R.; Farmer, C. D.; Diı`az, P. Appl. Phys. Lett. 2007, 90, 233510. Kumah, D. P.; Shusterman, S.; Paltiel, Y.; Yacoby, Y.; Clarke, R. Nat. Nanotechnol. 2009, 4, 835–838. Somaschini, C.; Bietti, S.; Koguchi, N.; Sanguinetti, S. Nano Lett. 2009, 9, 3419–3424. Sablon, K. A. Nanoscale Res. Lett. 2008, 3, 191–192. Li, S.; Xia, J. Appl. Phys. Lett. 2007, 90, No. 092119. Hanke, M.; Schmidbauer, M.; Grigoriev, D.; Schaf¨er, P.; Koh¨ler, R.; Metzger, T. H.; Wang, Z. M.; Mazur, Y. I.; Salamo, G. J. Appl. Phys. Lett. 2006, 89, 053116. Lee, J. H.; Wang, Z. M.; Strom, N. W.; Mazur, Y. I.; Salamo, G. J. Appl. Phys. Lett. 2006, 89, 202101. Tong, C. Z.; Yoon, S. F. Nanotechnology 2008, 19, 365604. Pankaow, N.; Panyakeow, S.; Ratanathammaphan, S. J. Cryst. Growth 2009, 311, 1832–1835. DOI: 10.1021/nl100217k | Nano Lett. 2010, 10, 1512-–1516
(15) Schramm, A.; Tukiainen, A.; Aho, A.; Pessa, M. J. Cryst. Growth 2009, 311, 2317–2320. (16) Wang, Z. M.; Holmes, K.; Shultz, J. L.; Salamo, G. J. Physica Status Solidi A 2005, 202, 1339–1339. (17) Mano, T.; Kuroda, T.; Sanguinetti, S.; Ochiai, T.; Tateno, T.; Kim, J.; Noda, T.; Kawabe, M.; Sakoda, K.; Kido, G.; Koguchi, N. Nano Lett. 2005, 5, 425–428. (18) Heyn, C.; Stemmann, A.; Kop¨pen, T.; Strelow, C.; Kipp, T.; Grave, M.; Mendach, S.; Hansen, W. Appl. Phys. Lett. 2009, 94, 183113. (19) Li, A. Z.; Wang, Z. M.; Wu, J.; Xie, Y.; Sablon, K. A.; Salamo, G. J. Cryst. Growth Des. 2009, 9, 2941–2943. (20) Mano, T.; Kuroda, T.; Mitsuishi, K.; Yamagiwa, M.; Guo, X.; Furuya, K.; Sakoda, K.; Koguchi, N. J. Cryst. Growth 2007, 301302, 740–743.
© 2010 American Chemical Society
(21) Wu, J.; Li, Z.; Shao, D.; Manasreh, M. O.; Kunets, V. P.; Wang, Z. M.; Salamo, G. J.; Weaver, B. D. Appl. Phys. Lett. 2009, 94, 171102. (22) Melloch, M. R.; Woodall, J. M.; Harmon, E. S.; Otsuka, N.; Pollak, F. H.; Nolte, D. D.; Feenstra, R. M.; Lutz, M. A. Annu. Rev. Mater. Sci. 1995, 25, 547–600. (23) Wang, Z. M.; Holmes, K.; Mazur, Y. I.; Ramsey, K. A.; Salamo, G. J. Nanoscale Res. Lett. 2006, 1, 57–61. (24) Patton, B.; Langbein, W.; Woggon, U. Phys. Rev. B 2003, 68, 125316. (25) AbuWaar, Z. Y.; Wang, Z. M.; Lee, J. H.; Salamo, G. J. Nanotechnology 2006, 17, 4037–4040. (26) Kim, J. S.; Jeong, M. S.; Byeon, C. C.; Ko, D.; Lee, J.; Kim, J. S.; Kim, I.; Koguchi, N. Appl. Phys. Lett. 2006, 88, 241911.
1516
DOI: 10.1021/nl100217k | Nano Lett. 2010, 10, 1512-–1516
