Noise and photoconductive gain in InAs quantum-dot infrared photodetectors

Zhengmao Ye, Joe C. Campbell, Zhonghui Chen, Eui-Tae Kim, Anupam Madhukar
2003 Applied Physics Letters  
We report noise characteristics, carrier capture probability, and photoconductive gain of InAs quantum-dot infrared photodetectors with unintentionally doped active regions. At 77 K, a photoconductive gain of 750 was observed at a bias of 0.7 V. The high gain is a result of the low carrier capture probability: pϭ0.0012. Infrared detection ͑midand long-wavelength͒ has been a subject of extensive research due to its key roles in commercial, defense, and space applications. HgCdTe photodetectors
more » ... Te photodetectors continue to be performance leaders for many infrared detection applications. However, materials growth issues limit the spatial uniformity and yield, which, in turn, increase the cost of HgCdTe focal plane arrays. In the past decade, quantum-well infrared photodetectors ͑QWIPs͒, which operate on intraband transitions, emerged as an alternative for infrared detection. 1 Advantages to the QWIP approach include material uniformity and the ability to design complex multilayer, high-performance structures using advanced molecular-beam epitaxy ͑MBE͒ and metalorganic chemical vapor deposition growth technologies. A drawback to QWIPs, on the other hand, is the requirement for special coupling techniques since they do not operate in the normalincidence mode. Another promising type of infrared photodetector, which is also based on intraband transitions, is the quantum-dot infrared photodetector ͑QDIP͒. Unlike QWIPs, QDIPs are intrinsically sensitive to normal-incident infrared radiation, owing to the three-dimensional confinement of the electrons in the QDs. Furthermore, QDIPs have longer carrier lifetimes, which creates the potential for higher photoconductive gain and higher operating temperatures. 2 One figure of merit that is used to evaluate the performance of infrared photodetectors is the specific detectivity D*, which is the signal-to-noise ratio normalized to the wavelength. Recently, QDIPs with single or dual peak photoresponse at 3ϳ14 m 3-9 have been reported, among which the highest D* is 10 10 cmHz 1/2 /W at 77 K. 3 Although theoretical calculations suggest that QDIPs should achieve higher D* than QWIPs, 10 to date, this has not been experimentally confirmed. Current values of D* for QDIPs are at least an order of magnitude lower than that of QWIPs. Improvements in D* for QDIPs can be achieved by increasing the responsivity or lowering the noise. In this letter, we present a study of the noise in InAs QDIPs with unintentionally doped active regions. The InAs QDIP studied in this work belongs to the class of n -i -n structure QDIPs under examination by us. 3, 5, 6, 9 The sample was grown on semi-insulating GaAs ͑001͒ substrates by solid-source MBE. Five layers of nominally 3.0monolayer ͑ML͒ InAs QDs were inserted between highly Si-doped top and bottom GaAs contact layers. Thirtymonolayer GaAs cap layers were grown via migration enhanced epitaxy at ϳ350°C as the QD cap layers, followed by 121-ML GaAs grown at 500°C for a total of 151-ML GaAs spacer layers. The GaAs layers between the contact layers and the nearest QD layer had a thickness of 220 ϳ240 ML. Devices were fabricated from pieces of the same asgrown sample as reported in Ref. 9 and followed standard procedure: photolithography, wet chemical etching, metal deposition and lift-off, and rapid thermal annealing. Photolithography and wet chemical etching were used to define the device mesas, each of which had a diameter of 250 m and a height of ϳ1.4 m. Electron-beam deposition of AuGe/ Ni/Au and lift-off were then performed to form the top and bottom contacts. This was followed by a rapid thermal anneal at 430°C for 20 s in nitrogen. In the following discussion, "positive" bias means that a positive voltage was applied to the top contact. The QDIP exhibited a peak photoresponse at 7.2 m ͑not shown͒, and a spectral width of ⌬/ϳ14%, which is characteristic of an intraband transition within the conduction band. The dark noise current i n was characterized with very low noise current preamplifiers and a fast Fourier transform spectrum analyzer. The sample was mounted on a cold finger inside a cryostat and surrounded with a cold shield ͑dark environment͒. Figure 1 shows the noise current density versus frequency at a bias V b of 0.3 V, and temperatures of 77, 90, 105, and 150 K. At low frequency ( f Ͻ2 Hz), the dominant source of the noise current appears to be 1/f noise, which is characterized by an approximate dependence upon the reciprocal of the frequency and the square of the current. For f Ͼ2 Hz, the noise current is essentially independent of frequency. This is similar to the generation-recombination ͑GR͒ noise in bulk photoconductors and QWIPs, which leads a͒ Electronic
doi:10.1063/1.1597987 fatcat:bqywjqsxa5c27ivguwfxd7dfvq