Abstract-This paper gives an overview on the design, fabrication, and characterization of quantum cascade detectors. They are tailorable infrared photodetectors based on intersubband transitions in semiconductor quantum wells that do not require an external bias voltage due to their asymmetric conduction band profile. They thus profit from favorable noise behavior, reduced thermal load, and simpler readout circuits. This was demonstrated at wavelengths from the near infrared at 2 m to THz radiation at 87 m using different semiconductor material systems. In the NIR, fast intraband semiconductor photodetectors are only available for wavelengths up to about 1.6 m. On the other tail of optical frequencies, namely for detection of THz radiation, bolometers are widely used; however, they are not well suited for high-speed applications. For fast light detection at wavelengths above 1.6 m, ISB photodetectors are very promising candidates. As unipolar devices, their fundamental F. R. Giorgetta was with the University of Neuchatel, 2000 Neuchatel, Switzerland. He is now with the National

We demonstrated a GaAs/AlGaAs-based far-infrared quantum well infrared photodetector at a wavelength of ϭ84 m. The relevant intersubband transition is slightly diagonal with a dipole matrix element of 3.0 nm. At 10 K, a responsivity of 8.6 mA/W and a detectivity of 5ϫ10 7 cm ͱHz/W have been achieved; and successful detection up to a device temperature of 50 K has been observed. Being designed for zero bias operation, this device profits from a relatively low dark current and a good noise behavior.In recent years, there has been an increasing interest in the fabrication of so-called terahertz ͑THz͒ emitters and detectors. While electronic devices like Gunn diodes or Schottky diode frequency multipliers try to reach this range from the low frequency end, optical devices like gas or semiconductor lasers are quickly moving into the THz range from the high frequency side. [1][2][3] Since the THz region is traditionally defined as 0.1-3 THz, the currently available quantum cascade laser ͑QCL͒ sources with ϭ87 m can already be regarded as THz sources. At low temperatures, they emit several milliwatts of continuous wave output power. On the electronics side, Gunn diodes and frequency mixers have also achieved several milliwatts of radiated power at frequencies on the order of hundreds of gigahertz. 4 Once the entire THz frequency range is fully accessible by convenient radiation sources, it is obvious that the next important step towards applications is the development of suitable detectors. Like any other type of electromagnetic radiation, THz waves or pulses can be detected by coherent or incoherent means. Most coherent detection schemes utilize frequency conversion, whereas incoherent methods are based on the heat production of absorbed radiation. Typical examples of heat detectors include Si-bolometers or pyroelectric crystals like deuterated triglycerine-sulfate ͑DTGS͒; on the other hand, Schottky diode mixers, 5 nonlinear optical crystals like ͕110͖ ZnTe, 6 and gated photoconductive antennas are typical coherent detection schemes. 7 As a further incoherent solution, semiconductor-based quantum-type approaches like biased superlattices have attracted some attention.8 Bolometers are in general highly sensitive, but like all heat-based detection schemes, they are intrinsically slow and built for very low temperature operation only.9 DTGS detectors and pyroelectric crystals offer the advantage of faster detection at the prize of reduced sensitivity. Finally, extrinsic photoconductors such as doped Ge detectors are fast and sensitive, but they must be cooled to 4 K. Quite generally, coherent techniques profit from a good sensitivity, but they are experimentally more sophisticated than incoherent ones.10 Although semiconductor quantum devices might not be highly sensitive, their potential for mass fabrication and integration by means of semiconductor device technology is very appealing. This has been proven by different types of quantum well infrared photodetectors ͑QWIPs͒ which work in a variety of different wavelength...

We present a room temperature operated 5.35μm quantum cascade detector which was tested at high frequencies using an optical heterodyne experiment. Two slightly detuned continuous wave distributed feedback single mode quantum cascade lasers were used to generate a beating signal. The maximum frequency at which the resulting microwave signal could be detected was 23GHz. The cutoff behavior of our device was modeled with a simple RLC circuit and showed excellent agreement with the experimental data.

We present two InP-based quantum cascade detectors ͑QCDs͒ in the mid-infrared wavelength range. Their narrow band detection spectra are centered at 5.3 and 9 m. A vertical intersubband transition followed by a carefully designed extraction cascade, which is adapted to the LO-phonon energy, leads to 10 K responsivities R of 3.2 and 9.0 mA/ W and background limited detectivities D BLIP * of 2 ϫ 10 8 and 3 ϫ 10 9 Jones, for the 5.3 and the 9 m devices, respectively. Detection has been observed up to device temperatures of 300 K ͑RT͒, albeit reasonable performance is restricted to temperatures below 150 K ͑5.3 m͒ and 70 K ͑9 m͒. Designed for zero bias operation, QCDs do not produce any dark current and therefore do not suffer from dark current noise and capacitance saturation at long integration times, making them ideal devices for large focal plane arrays.Semiconductor-based mid-infrared detectors have a wide range of potential applications in sensing, security, and defense. The probably most advanced technology using these materials is the quantum well infrared photodetector ͑QWIP͒.1,2 This device utilizes bound-to-continuum intersubband transitions in quantum wells, generally operates in a photoconductive mode, and is typically held at cryogenic temperatures. Photovoltaic QWIPs ͑PV-QWIPs͒, having a built-in asymmetry which allows biasless operation, 3 have been shown to work up to room temperature. 4 Other promising solutions include interband mercury-cadmium-telluride detectors 5 and silicon microbolometers. 6 The former devices are quite rapid when used at 300 K, but are not very good in terms of signal-to-noise ratio. By cooling to 77 K, one can gain several orders of magnitude of signal-to-noise ratio, but the detectors then become slow. Microbolometers, finally, have made astonishing progress in the last couple of years and are now used in infrared camera systems. Recently, quantum cascade structures have been used for the detection of infrared radiation. [7][8][9] Compared to photoconductive QWIPs, this concept shares the advantage of a photovoltaic detection scheme with PV-QWIPs: as no bias voltage is applied, no dark current noise and no integration-time limitation due to capacitance saturation in the readout circuit occur. By using InP-based instead of GaAs-based materials, one can gain nearly a factor of 2 in absorption efficiency because of the lighter effective mass. Furthermore, the wells can be grown somewhat wider leading to less linewidth broadening due to interfacial roughness.10 Additional performance improvements are expected because of the bound-tobound transition, which leads to narrower detection peaks and thus a better noise figure. 11 Keeping in mind the fabrication technology, it is finally clear that in a QWIP, both conduction band discontinuity and well width are crucial parameters for high performance. In a QCD, these two parameters are not directly coupled, so that growth uncertainties in composition or thickness, leading to shifted energies of the states, will only slightly shift the...

The authors report on short-wavelength In 0.53 Ga 0.47 As/ AlAs 0.56 Sb 0.44 quantum cascade detectors ͑QCDs͒. At room temperature, one device detects at 505 meV ͑2.46 m͒ with a responsivity of 2.57 mA/ W, while a second QCD is sensitive at 580 meV ͑2.14 m͒ with a responsivity of 0.32 mA/ W.With the recent development and commercialization of quantum cascade lasers, fast and sensitive semiconductorbased detectors for the mid-to far-infrared wavelength range are rapidly becoming key components for future optical sensor systems. The photoconductive quantum well infrared photodetector ͑PC-QWIP͒ is certainly the most mature device for these wavelengths. Almost in parallel to the development of the PC-QWIP, photovoltaic QWIPs ͑PV-QWIPs͒ have been proposed and fabricated. The pioneering works of Schneider et al. 1 and Levine et al. 2 have paved the way for a thorough understanding of this device. Yet another version of the PV-QWIP, namely the quantum cascade detector 3 ͑QCD͒, has recently experienced substantial progress. Similar to the PV-QWIPs, these devices have the advantage that no external bias voltage is necessary for operation, leading to zero dark current and a favorable noise behavior. QCDs were reported at 5, 9, 17, and 84 m ͑Refs. 4-7͒ and have been fabricated using either InGaAs/ InAlAs or AlGaAs/ GaAs. Shorter operation wavelengths are however difficult to obtain with those semiconductor materials due to their relatively low conduction band offsets ͑CBOs͒ of 0.5 eV ͑InGaAs/ InAlAs͒ and 1 eV, respectively, ͑AlAs/ GaAs͒. An alternative system making the wavelength range between 2 and 3 m available for QCDs is the InGaAs/ AlAsSb lattice matched to InP; this combination offers a CBO of 1.6 eV, 8 which is theoretically sufficient to fabricate QCDs down to 1.55 m. Its potential has been demonstrated by the fabrication of short-wavelength quantum cascade lasers. 9 We therefore take advantage of this material and present here three InGaAs/ AlAsSb QCDs detecting down to 2.14 m ͑0.58 eV͒ at room temperature.The basic idea of the QCD is to achieve-via illumination-vertical electron transport along the growth axis. The design used for these QCDs is based on the approach of Graf et al.,5 and is illustrated by the schematic conduction band diagram shown in Fig. 1. The thick quantum well ͑QW͒ A is degenerately n-doped in order to populate its ground state A 1 . By absorption of a photon with energy A 1 → A 2 , these electrons are lifted into the excited state A 2 . From there, they either fall back to A 1 or tunnel to the adjacent state B 1 and "cascade" through C 1 , etc., to the ground state A 1 Ј of the following period, thereby generating the desired vertical transport. Using resonant tunneling between A 2 and B 1 allows us to separate the two active QWs by a relatively thick tunnel barrier; this should ideally result in a high resistivity without degrading the electron extraction efficiency from A 2 to A 1 Ј.For the design, self-consistent Schrödinger-Poisson calculations were performed using the material paramete...