Mid-IR Resonant Cavity Detectors

Transcription

1 Mid-IR Resonant Cavity Detectors Running title: Mid-IR Resonant Cavity Detectors Running Authors: O Loughlin et. al T.A. O Loughlin a) The Institute of Optics, University of Rochester, 275 Hutchison Rd., Rochester, NY G.R. Savich Air Force Research Laboratory, Kirtland AFB, NM D.E. Sidor, B.T. Marozas The Institute of Optics, University of Rochester, 275 Hutchison Rd., Rochester, NY T.D. Golding, K.D. Jamison, L. Fredin, B. Fowler, W. Priyantha Amethyst Research, Inc., 123 Case Cir., Ardmore, OK G.W. Wicks The Institute of Optics, University of Rocehster, 275 Hutchison Rd., Rochester, NY a) Electronic mail: toloughl@ur.rochester.edu Resonant cavity detectors based on III-V materials have been designed, grown entirely by MBE, fabricated and tested. They offer a low noise (dark current densities of 0.4 ma/cm 2 were measured at 298K, close to the predicted value of 0.31 ma/cm 2 ), narrow response detector (full width at half maximum of 57 nm in GaSb and 45 nm in InAs) in the mid-infrared region, with future applications in spectroscopy, gas sensing, and optical communications. 1

2 I. INTRODUCTION Standard III-V infrared (IR) detectors are commonly limited by two main noise sources: background current and dark current. Background current originates from objects other than the target source radiating (generally broadband) light in the detectors field of view. Dark current arises from thermal excitation across the small bandgap of the detector. Operating conditions determine which of these is the main source of the noise. If these noise sources can be decreased, without decreasing photoresponse, the detector performance will increase. The resonant cavity detector offers this performance improvement, resulting from the cavity enhancement of light-matter interaction. Small absorber volumes still maintain good photoresponse and enable substantial decreases in dark current. Light emitting and light detecting devices, enhanced by a resonant cavity, have been previously reported, mainly for near IR and shortwave IR 1. Vertical cavity surface emitting lasers (VCSEL s) are a common example 2. In the midwave IR, II-VI based systems are the main workhorse and resonant cavity designs have shown promise there 3. IV-VI systems allow for silicon integration of these devices, as well as showing sharp resonances 4,5. Previous III-V work includes quantum dot and quantum well detectors, as well as p-i-n architecture 6-8. The main thrust of this work is to examine a resonant cavity detector, designed to operate in the midwave IR and entirely grown entirely by molecular beam epitaxy (MBE) with III-V based materials, with a unipolar barrier architecture structure in mind 9. Unipolar barrier architecture allows for diffusion limited performance, therefore these resonant-cavity-enhanced midwave-ir detectors are expected to exhibit narrow spectral response, reduced dark current and background 2

3 current noise, and high speed. This approach has advantageous applications in areas such as spectroscopy, gas sensing, and optical communications. The device is qualitatively similar to the VCSEL in fabrication. The resonant cavity IR detector structure consists of a thin IR absorbing layer (detector layer) inserted into an optical cavity formed by transparent layers between two mirrors all made of MBEgrown epitaxial layers. The mirrors are quarter-wave semiconductor stacks, similar to the mirror used in VCSEL s. The high reflectivity mirrors cause the light to recirculate inside the optical cavity, passing through the absorber layer many times. The multiple passes through the absorber allow for much thinner absorber layers than in conventional IR detectors, while maintaining high quantum efficiency. The thinner absorber decreases dark current and increases speed. II. MBE GROWTH The devices reported on here were grown by solid source MBE utilizing elemental group III materials (Al, Ga, In) and valved crackers for the group V sources (As, Sb). They were grown on unintentionally doped n-type InAs and intentionally doped p-type GaSb substrates. Growth temperature and growth rate of the layers were calibrated utililizing known reflection high-energy electron diffraction (RHEED) transitions 10. Compositions were calibrated with X-ray diffraction measurements and photoluminescence measurements. The epitaxial growth of the mirrors allows for slight temperature tuning, as the resonance of the structure will shift with temperature. This allows the detector to be used over a small range of wavelengths, all with low noise. Sidor 11 presents a good overview on InAs growth. III. MODELING 3

4 High reflectivity mirror stacks are required for high performance resonant cavity devices. The high reflectivity allows light to propagate through the absorber many times, allowing for the thinner layer. The highest reflectivity is achieved by choosing material pairs with large refractive index contrast, subject to the constraint of lattice matching the epitaxial layers to the substrate. According to Adachi 12, a refractive index difference of Δn 0.5 is achievable for III-V based mirror layers. This index contrast allows 6 pairs of layers to produce >80% mirror reflectivity, while >90% reflectivity requires pairs of mirror layers. 80% and 90% mirror reflectivities have been achieved on both InAs and GaSb substrates, shown later in section IV. Higher reflectivity mirrors will produce narrower response peaks but require tighter tolerances on layer thickness and composition. Layer thickness and refractive index (and by extension, composition) help define the optical mode of the cavity as well as the mirror resonance. As such, they can be manipulated to change these resonances and change the wavelength at which the device will operate. Though small strains in composition may be tolerated, it is advisable to change the thickness of the layers instead. Examples of the full resonant cavity structures are shown in Fig. 1, for both GaSb and InAs based systems. The layer nearest the substrate should be the highest refractive index layer, however, this is impossible in the GaSb design as GaSb is the higher index. 4

5 FIG. 1. (Color Online) Base design for the resonant cavity structure on GaSb (a) and InAs (b). The number of pairs of high and low index layers can be varied. Higher number of pairs will lead to a smaller linewidth. The thicknesses given are in optical thickness. The designs in Fig. 1 were optically modeled with a characteristic matrix approach, as described by Macleod13. This shows that only a narrow band of recirculating wavelengths interfere constructively with themselves, which produces the resonant mode of the cavity. This resonance occurs because two beams interfere destructively. One beam is the incident light reflecting off the top surface and the other beam is the portion of the incident light that enters the cavity, reflects off the back mirror, and escapes from the cavity through the partially reflecting top mirror. The resonant mode of the cavity is observable as a large dip in the wafer s reflectivity and a large peak in the detector s absorptivity, as shown in Fig. 2. This reflectivity dip can be made to approach zero by 5

6 adjusting the reflectivities of the two mirrors, which in turn is accomplished by adjusting the number of quarter-wave layers in the mirror stacks 1. FIG. 2. (Color Online) Full reflectivity and absorption model of a resonant cavity grown on a GaSb substrate. The small absorption peak helps to give the device a narrow spectral response and a narrow quantum efficiency. The epitaxial structure of the cavity region is engineered to allow for unobstructed flow of the photogenerated carriers out of the cavity, while at the same time not impeding the circulation of light in the cavity. The absorbing layers under consideration are InAs and InAsSb, either lattice matched to GaSb or InAs. The device compatibility with unipolar barrier detector architectures, which produce diffusion limited performance, also enables the extraction of these photo-generated carriers while blocking the majority carrier dark current 9. In midwave-ir III-V devices, the dominant noise is often from fluctuations of dark current exacerbated by the small bandgap of the materials. Advanced detector designs using unipolar barrier detector architectures are limited by this diffusioninduced dark current, which scales directly with the absorber thickness 14. Therefore, decreasing this detecting region will decrease the dark current, as modeled in Fig.3. In addition to the dark current suppression, the angular dependence of acceptance into the 6

7 optical cavity and the natural wavelength selection of the resonant cavity suppress the detection of wavelengths away from cavity resonance, producing a suppression of background noise. FIG. 3. (Color Online) Modeled dark current density against absorber length at multiple temperatures for an InAs based device. Modeling was performed by starting with dark currents of conventional InAs unipolar barrier detectors and scaling the results by the absorber thickness. Modeling results predict a spectral response as narrow as a few nanometers for high reflectivity mirrors, reduction of dark and background currents by up to (which will increase signal-to-noise parameters, such as D* and noise equivalent power, by a factor of ), and speed in the sub-nanosecond range 12. IV. EXPERIMENTAL The mid-ir designs have been successfully implemented on both InAs and GaSb based devices. Measurements of the reflectivities agree well with the wafer reflectivity 7

8 simulations, demonstrating the creation of high quality III-V based mirrors and cavities (Fig. 4-6). The devices were grown on different substrates and different wavelengths were targeted for the growths, resulting in different resonances. Analysis of the devices show a full width at half maximum of 57 nm on the GaSb based sample (Fig. 4, grown to be resonant with 4.69 μm light) and 45 nm on the InAs based sample (Fig. 5, grown to be resonant with 2.91 μm light). In addition, a detector was fabricated on InAs. As modeled, near the resonant wavelength of the cavity, the response exhibits a peak and the reflectivity exhibits a minimum (Fig. 5). The dark current density of the device was measured to be 0.4 ma/cm 2 at 298K, close to the predicted value of 0.31 ma/cm 2 for a 100 nm thick absorber and around 35 times smaller than the modeled, thick absorbers and other unipolar barrier InAs designs 14. FIG. 4. (Color Online) Measurement and simulation of an 8 pair mirror and resonant cavity on GaSb. Inset: Zoom in on the dip, showing the 57 nm FWHM of the measured data as well as the simulation. 8

9 FIG. 5. (Color Online) Experimental results of an MBE grown InAs-based detector. Reflectivity shows the absolute reflectivity of the fully grown structure. Response shows the normalized response of the detector. Minimizing the reflectivity at the resonant wavelength of the full resonant cavity structure maximizes the detector s quantum efficiency. Cavity resonant wavelength reflectivities of just a few percent have been obtained, as shown in Figure 6 (grown to be resonant with 3.4 μm light on InAs). 9

10 FIG. 6. (Color Online) Reflectivity of a 6 pair mirror and a resonant cavity on an InAs substrate. Notice that the peak of the mirror stack lines up with the dip in reflectivity of the cavity. The different resonances in Figures 4-6 originate from the structures having different thicknesses and, to a lesser extent, the dispersion of the materials changing the refractive index with wavelength. Additional experiments are ongoing to electrically characterize a detector based on GaSb, as well as improve upon the InAs based device. V. SUMMARY AND CONCLUSIONS Resonant cavity detectors offer a high speed, low noise way to detect light in the mid-ir region. Currently these devices have been fabricated with around a 50 nm wide band of detection. This small absorption band, along with their small volume, help to decrease their limiting noise factors, while the mirrors work to keep their signal detection high. The authors envision this linewidth shrinking as growth and processing techniques are refined. 10

11 ACKNOWLEDGMENTS The authors would like to acknowledge the support of the United States Air Force Research Laboratory, the Army Research Office, and a NASA Space Technology Research Fellowship. 1 M. Selim Ünlü and Samuel Strite, J. Appl. Phys. 78, 607 (1995) 2 L. Cerutti, A. Ducanchez, G.Narcy, P.Grech, G.Boissier, A.Garnache, E.Tournié, F.Genty, J. Cryst. Growth 311, 1912 (2009) 3 J. G. A Wehner, C. A. Musca, R. H. Sewell, J. M. Dell, L. Faraone, Appl. Phys. Lett., 87, (2005) 4 M. Arnold, D. Zimin, H. Zogg, Appl. Phys. Lett., 87, (2005) 5 J. Wang, J. Hu, P. Becla, A. M. Agarwal, L. C. Kimerling, Opt. Express, 18, (2010) 6 T. Asano, C. Hu, Y. Zhang, M. Liu, J. C. Campbell, A. Madhukar, IEEE J. Quant. Electron, 46, 1484 (2010) 7 A. Shen, H. C. Liu, M. Gao, E. Dupont, M. Buchanan, J. Ehret, G. J. Brown, F. Szmulowicz, Appl. Phys. Lett., 77, 2400 (2000) 8 A. M. Green, D. G. Gevaux, C. Roberts, P. N. Stavrinou, C. C. Phillips, Semicond. Sci. Technol., 18, 964 (2003) 9 S. Maimon and G. W. Wicks, Appl. Phys. Lett. 89, (2006) 10 A. Bracker, M. Yang, B. Bennett, J. Culbertson and W. Moore, J. Cryst. Growth 220 (4), (2000). 11

12 11 D.E. Sidor, B.T. Marozas, X. Du, W.D. Hughes, G.W. Wicks, G.R. Savich, J. Vac. Sci. Technol. B, pending review 12 S. Adachi, J. Appl. Phys. 61, 4869 (1987) 13 H. Angus Macleod, Thin Film Optical Filters, edited by W. T Welford (Institute of Physics Publishing, Bristol, UK, 2001), 3 rd ed., pp G.R. Savich, D.E. Sidor, X. Du, C.P. Morath, V.M. Cowan, G.W. Wicks, Appl. Phys. Lett. 106, (2015) 12