Voltage-controlled wavelength-selective photodetector

The present invention relates to a voltage-controlled wavelength-selective photodetector and a method for it's use.

Voltage-controlled wavelength-selective photodetectors can be used in optoelectronic converters for signal processing and for logical switching networks.

The article by Friedmann et al. in Compound Semiconductor, p. 27, November/December 1996, and the article by Fang et al. in IEEE Electron Device Letters Vol. 12, No. 4, Apr. 1, 1991, pp. 172-174 describe the use of photodetectors for detection of light having different wavelengths. European Patent Application No. 682 375 A1 also describes a photodetector that makes it possible to detect light having different wavelengths in the visible range using two detectors that are arranged one on top of the other and that are sensitive to different wavelengths. In addition, in German Patent Application No. (AZ 197 14 054 A1), filed prior to the date of the present application, but published after that date, describes a photodetector which is made of at least two photodiodes of silicon/silicon-germanium (Si/SiGe) semiconductor material arranged one on top of the other. These detectors are used, for example, in solar cells with the purpose of generating as high a photoelectric current as possible by adding the charges produced in the individual detectors. The use of different semiconductor materials contributes to achieving a high degree of spectral sensitivity in addition to a high degree of efficiency. The individual photoelectric currents produced in the detectors always overlap additively. Thus, while the use of various photodetectors to detect light signals is customary, the method of processing signals by subtraction of the photoelectric currents is not.

In order to separate a useful signal S from noise N and to improve the signal/noise ratio (S/N), expensive and complex signal decoding arrangements involving a plurality of detectors, beam splitters, filters, and additional electronics, are used.

An object of the present invention is therefore to provide a cost-effective, monolithically highly integratable photodetector that opens up possibilities of signal conversion in a broad spectrum through wavelength selectivity and can be used, in particular, to improve the signal/noise ratio of an optical signal.

The photodetector according to the present invention is made of a double diode composed of a Si Schottky diode and a SiGe PIN diode of opposite polarities. The short-wavelength component (λ<0.9 μm) of light entering through a window in the detector preferably produces electron-hole pairs in the Si Schottky detector, while the long-wavelength component (1 μm<λ<2 μm) passes through the substrate and is preferably absorbed in the epitaxially deposited SiGe superlattice or the quantum well diode. The photoelectric currents of the two detectors flow in physically opposite directions and are subtracted from one another, so that a photoelectric current with a plus or minus sign, depending on the wavelength, results. The bias voltage applied decides whether the spectrum is defined by the photoelectric signal of the Si Schottky diode or of the Si/Ge PIN diode.

The present invention is elucidated below with reference to with reference to the drawings, in which the

FIG. 1 shows the design of a photodetector;

FIG. 2 shows an equivalent diagram of the dual diode of a photodetector;

FIG. 3 shows the photoelectric current of a detector at a higher positive and negative bias voltage as a function of wavelength λ;

FIG. 4 shows the photoelectric current of a detector during operation at a lower bias voltage between −0.3 and 0.4 V;

FIG. 5 shows the signal/noise ratio of a decoded signal.

In the exemplary embodiment shown in FIG. 1, the layer structure illustrated is deposited on a slightly p-doped Si semiconductor substrate 1 , for example, by molecular beam epitaxy. The layer structure contains

a graduated SiGe buffer 2 , having a thickness of approximately 650 nm, and a final Ge concentration of 60 atomic %, for example;

a constant SiGe buffer 3 , having a thickness of approximately 500 nm, preferably with a final Ge concentration of graduated buffer 2 of 60 atomic %;

an n-doped SiGe superlattice 4 with a thickness of approximately 200 nm having a periodically repeating layer sequence of, for example, six Si monolayers and four Ge monolayers, having a dopant concentration (Sb) of 10 17 per cm 3 ;

a highly n-doped (10 19 per cm 3 ) Si cover layer 5 having a thickness of approximately 2 nm.

The detector regions are insulated from the surrounding semiconductor material using the usual process steps of integrated semiconductor technology, for example, with silicon dioxide or groove etching. The metal plating, acting as a Schottky contact, of substrate-side first contact 6 , including window 8 , and of second metal contact 7 , is also structured on the epitaxial layer using standard semiconductor technology. Second metal contact 7 forms an ohmic contact due to the high degree of doping of thin cover layer 5 . In an alternative exemplary embodiment, the second detector is provided with an integrated resonator. In another exemplary embodiment, a Bragg reflector is positioned over the second detector.

The equivalent diagram corresponding to the layer structure of dual diode A and B is shown in FIG. 2 .

The mode of operation of the diodes having opposite polarities shows the curve of the photoelectric current as a function of the wavelength of the incident light. Depending on the bias voltage of the detector, either a photoelectric current 10 results from the silicon diode or a photoelectric current 11 results from the PIN SiGe diode in the event that the voltage is selected to be so high, for example, ±1 V, that one of the two diodes is always conductive and the other remains blocked.

The dynamic response of the detector that is used in operation is set for bias voltages between plus and minus 1 V. The curves are shown in FIG. 4 for some intermediate values. For a suitably selected bias voltage, for example, 0.2 V, the curve has both positive and negative values depending on the wavelength. This can be used, for example, in an application for decoding signals as follows:

For an incident optical signal having two carrier frequencies λ 1 and λ 2 , responsivity R of the detector is given by the equation

R=A 1 ( S+N )+ A 2 N,

where A 1 and A 2 represent the voltage-dependent responsivity of the individual diodes for wavelengths λ 1 and λ 2 , S represents the useful signal and N represents the noise. By applying a suitable bias voltage, the condition A 1 =−A 2 can be fulfilled in that equalization is performed so that the photoelectric currents in both diodes have opposite signs, wherefrom

R=A

1

S

Thus useful signal S can be separated from the noise level in a very simple manner. The signal/noise ratio (S/N) of a useful signal S, processed by decoding, is shown in FIG. 5 . The curve shows in this example that this method considerably increases the signal/noise ratio by overlapping the signals in the range of the selected wavelengths λ 1 and λ 2 .