Voltage-tunable photodetector.

<b>FIELD OF THE INVENTION</b>

The present invention generally relates to monolithic integrated circuit technology and electromagnetic communication technology and, more particularly, to a tunable monolithic integrated photodetector for detecting light at a selected wavelength within a wavelength range.

<b>BACKGROUND OF THE INVENTION</b>

When light strikes a photosensitive material, light energy, or photons, are absorbed by the material, and an electrical current, often called a photocurrent because it results from the absorption of photons, is generated in the material. Photocurrent refers to the flow of electrons or holes within the material. Moreover, a peak in the photocurrent spectrum of the material which results from the absorption of photons is oftentimes referred to as an exciton.

In the prior art, it has been shown that a multiple quantum well (MQW) diode can be used as a photodetector whose voltage of maximum photocurrent, <i>i.e</i>., a large exciton, is dependent upon the wavelength of incident light. In this regard, see P.H. Wood <i>et al</i>., "Wavelength-Selective Voltage-Tunable Photodetector Made from Multiple Quantum Wells," <i>Appl.</i><i>Phys. Lett.,</i> vol. 47, no. 3, pp. 190-192, August 1, 1985. An MQW material is essentially a narrow bandgap material which has a thickness substantially less than the diameter of an exciton and which is sandwiched between wider bandgap materials so as to enhance the exciton via confinement of the energy potential well. The voltage of maximum photocurrent within the MQW diode can be located and related to the wavelength of the incident light, thereby allowing for measurements of the wavelength. Identifying the wavelength of incident light is beneficial in imaging and spectroscopy applications.

However, the light incident on a photosensitive material can create multiple peaks in the photocurrent spectrum, that is, more than one exciton. The photocurrent exhibits a relatively large exciton, called a heavy hole (hh) exciton, at the particular wavelength of incident light, as well as any number of smaller excitons, called light hole (lh) excitons, at other wavelengths of the photocurrent spectrum. Needless to say, the occurrence of multiple excitons causes potential ambiguity when one attempts to determine the wavelength of incident light.

More specifically, at room temperature (i.e., approximately 300 degrees Kelvin), the selectivity of an MQW diode is step-like, producing photocurrent at all photon wavelengths at and above the exciton wavelength. If no electrical bias is applied to the MQW diode, the peak-to-valley ratio of the hh exciton to the lh exciton is typically about 1.3 to 1 and reduces considerably with applied bias. See A.M. Fox et <i>al.,</i> "Quantum Well Carrier Sweep Out; Relation to Electroabsorption and Exciton Saturation," <i>IEEE</i><i>Journal of Quantum Electronics,</i> vol. 27, p. 2281 (1991). As the foregoing ratio is reduced, the lh excitons become indistinguishable from the hh excitons. With a 10 V//µm bias, the photocurrent peak at the lh exciton is actually larger than the peak at the hh exciton and occurs at the same wavelength as the hh exciton at 0 V/µm bias.

Thus, a heretofore unaddressed need exists in the industry for a tunable monolithic photodetector for detecting light at a selected wavelength within a wavelength range and which does not suffer from the problems and deficiencies of the prior art, as noted in the foregoing.

<b>SUMMARY OF THE INVENTION</b>

Briefly described, the present invention is a tunable monolithic integrated photodetector for detecting light at a selected wavelength within a wavelength range. The photodetector comprises a filter means, a detector means, and a bias means. Both the filter means and the detector means are formed from multiple quantum well (MQW) materials which are monolithically integrated atop a substrate. The filter means (first MQW layer) receives and filters incident light, and the detector means (second MQW layer) receives light from the filter means and detects the selected wavelength. The bias means applies a fixed bias and a variable bias. The fixed bias is applied between the filter means and the detector means for the purpose of causing the filter means to absorb light wavelengths surrounding the selected wavelength, thereby enhancing the detectability of the selected wavelength by the detector means. The variable bias is applied proportionally to both the filter means and the detector means for specifying, in effect, the selected wavelength within the tunable range.

A fixed filter may also be implemented for filtering out wavelengths of incident light beyond the tunable range of the photodetector. Optimal selectivity is achieved utilizing the fixed filter in combination with the photodetector.

In accordance with another embodiment of the present invention, a plurality of photodetectors formed in accordance with the present invention may be situated in a plane to form a two-dimensional array. The array of photodetectors may be used for many purposes, including imaging applications.

Other objections, features, and advantages of the present invention will become apparent to one of skill in the art upon examination of the following drawings and detailed description.

<b>BRIEF DESCRIPTION OF THE DRAWINGS</b>

The present invention can be better understood with reference to the following drawings. The drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating principles of the present invention. <ul id="ul0001" list-style="none"><li>Fig. 1 is a schematic view of a monolithic integrated photodetector in accordance with the present invention;</li><li>Fig. 2 is a graph of the photodetector output (preamp output voltage) of Fig. 1 versus incident light wavelength when the fixed voltage bias V_▲ = 6 volts and the variable voltage bias V_p = 8, 10 volts;</li><li>Fig. 3 is a graph of the photodetector output of Fig. 1 versus incident light wavelength when the fixed voltage bias V_▲<b>=</b> 4.5 volts and the variable voltage bias V_p = 4, 6, 8, 10, 12 volts; and</li><li>Fig. 4 is a graph of the photodetector output of Fig. 1 versus incident light wavelength when the fixed voltage bias V_▲ = 4.5 volts, the variable voltage bias V_p = 4, 6, 8, 10, 12 volts, and the incident light is filtered via a conventional fixed filter.</li></ul>

<b>DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT</b>

With reference to Fig. 1, a monolithic integrated photodetector 10 residing on a substrate 11 can determine the wavelength of incident light 12. The monolithic integrated photodetector 10 is preferably a n-i(MQW)-p-i(MQW)-n device with independent contacts to each of the p and n doped layers. Specifically, referring to Fig. 1, the photodetector 10 comprises the following successive layers: an n-doped layer 13, an intrinsic MQW layer 14 serving as a filter means, a p-doped layer 15, an intrinsic second MQW layer 16 serving as a detector means for detecting the wavelength of the incident light 12, and the n-doped substrate 11.

A fixed voltage bias V_▲ is applied between the MQW filter means 14 and the MQW detector means 16 via a fixed voltage source 17 of Fig. 1. As illustrated in Fig. 1, the voltage source 17 is connected between a contact 21 on the top surface of the n-doped layer 13 and a buffer 24, which will be described in further detail hereinafter. The fixed voltage bias V_▲ causes the MQW filter means 14 to absorb light wavelengths substantially surrounding the selected wavelength to thereby enhance the detectability of the selected wavelength. By offsetting the bias V_▲ of the MQW filter means 14 with respect to the MQW detector 16, the photocurrent spectrum of the MQW detector means 16 is slightly shifted compared to that of the MQW filter means 14. This predicament creates a narrow wavelength region where detectability is maximized, thus closely approximating a single wavelength, voltage tunable photodetector.

A variable voltage bias V_p is proportionally applied by a variable voltage source 18 across both the MQW filter means 14 and MQW detector means 16. As illustrated in Fig. 1, the voltage source 18 is connected between a contact 22 and the buffer 24. The contact 22 is situated in a well through the n-doped layer 13 and MQW filter means 14 to the p-doped layer 15. As the voltage bias V_p is varied, the photodetector 10 is tuned. In other words, as the voltage bias V_p is varied, the wavelength at which the MQW detector means 16 will exhibit a heavy hole exciton from the incident light 12 is varied.

The buffer 24, preferably a current preamplifier having resistor R and operational amplifier OPAMP, is connected to the positive terminals of the voltage sources 17, 18 and to the substrate 11 via contact 25 in order to provide a photodetector output 26. Photocurrent generated in the MQW filter means 14 does not flow through the buffer 24, but rather is shunted through the voltage sources 17, 18.

In the preferred embodiment, the photodetector 10 is formed on an n-doped GaAs substrate 11 using GaAs-source molecular beam epitaxy. First, 0.5 µm of n-doped Al_0.3Ga_0.7As is formed on the GaAs substrate 11. Next, the MQW detector means 16 is formed on the foregoing layer. The MQW detector 16 preferably comprises 75 undoped periods of 95 <b>Å</b> GaAs wells and 45 <b>Å</b> Al_0.3Ga_0.7As barriers. The MQW detector means 16 is then clad with 200 <b>Å</b> Al_0.3Ga_0.7As spacers on both sides. A 0.8 ·m p-doped Al_0.3Ga_0.7As layer 15 is formed on the MQW detector means 16. Next, the MQW filter means 14 is formed on the doped layer 15. The MQW filter means 14 preferably comprises 71 undoped periods of 95 <b>Å</b> GaAs wells and 45 <b>Å</b> Al_0.3Ga_0.7As barriers. The MQW filter means 14 is clad with 200 A Al_0.3Ga_0.7As undoped spacers. Finally, a 0.6 µm Al_0.3Ga_0.7As n-doped layer 13 is formed on the MQW filter means 14, followed by a 50 <b>Å</b> n+doped GaAs cap layer. The MQW filter means 14 has less periods than the MQW detector means 16 so that the MQW filter means 14 red shifts faster with voltage and thus tends to maintain the difference of the red shifts of the filter means 14 and the detector means 16 constant if the fixed voltage bias V_▲ is kept fixed. Otherwise, the difference would not increase proportionally because the red shift is quadratic with respect to field strength. A "red shift" in the context of this document means a shift in the photocurrent spectrum of a material, and thus, a change in the photon energy for generating excitons. With the preceding configuration, tunability is achieved by varying only the bias V_p supplied by voltage source 18.

It should be noted that the fixed voltage source 17 for supplying the fixed voltage bias V_▲ could be replaced by a variable voltage source. In this event, the MQW filter means 14 would not need to have less periods than the MQW detector means 16. However, more than one voltage source would need to be adjusted in order to tune the photodetector 10, thus making this circuit configuration undesirable. Moreover, it should be noted that with the preceding configuration, there is still a fixed bias between the filter means 14 and the detector means 16 and that tuning is accomplished while still maintaining this relative fixed bias.

Next, a AuGe layer is placed photolithographically on the n-doped layer 13 to form the contact 21. Further, a self-aligned technique is used to place the contact 22 within the well 23 on the p-doped layer 15. The reason for using a self-aligned technique is that care must be taken to avoid exposing regions of the MQW detector means 16 to light which is unfiltered by the MQW filter means 14. In the self-aligned technique, openings are defined in a layer of photoresist, and the sample is etched through the photoresist layer to the p-doped layer 15. An AuZn layer is then deposited onto the p-doped layer 15 using the same photoresist mask for the lift-off in order to form the contact 22. Thus, the entire etched region is coated with metal, as illustrated in Fig. 1. Because the contact 22 is deposited using the foregoing procedure, the photodetector 10 functions without having to focus the incident light 12 on any particular area of the n-doped layer 13. Finally, a mesa configuration for the photodetector 10 is formed by etching the structure down to the bottom n-doped substrate 11.

Fig. 2 is a graph showing the photodetector output 26, <i>i.e</i>., preamp output voltage, versus wavelength of incident light 12 with the fixed voltage bias V_▲ set at about 6 volts and the variable voltage bias V_p set at about 8 volts and at about 10 volts during separate trials. The spectrum of Fig. 2 was generated by applying an unfocused light source over the mesa configuration. The heavy hole (hh) and light hole (lh) excitons of the MQW detector means 16 show up as peaks in the spectrum, while the heavy hole and light hole excitons of the MQW filter means 14 show up as dips. Thus, as can be seen from the graph of Fig. 2, the peak-to-valley ratio of the detector peaks compared to the remainder of the spectrum is enhanced dramatically.

Fig. 3 is a graph of the photodetector output 26 versus wavelength of incident light 12 with the fixed voltage bias V_▲ set at about 4.5 volts and the variable voltage bias V_p set at about 4, 6, 8, 10 and 12 volts on separate trials. The difference between the red shifts of the detector means 16 and the filter means 14 are smaller than in Fig. 2, giving a more distinguished single peak for wavelength. At this setting, the heavy hole exciton of the MQW filter means 14 lies approximately on the light hole exciton of the MQW detector means 16, thus having maximum effect at removing signals at wavelengths other than at the wavelength of the heavy hole exciton of the MQW detector means 16. With these settings, a tuning range of at least approximately 14 nm is achieved.

To further enhance the single peak response, a fixed filter 27, denoted by phantom lines in Fig. 1, may be employed to filter the incident light 12. Preferably, the fixed filter 27 eliminates all wavelengths of light outside of the tunable range of the photodetector 10. For this purpose, a suitable fixed filter 27 is a model RG.850 filter manufactured by and commercially available from Newport, Inc., California, U.S.A. The RG.850 filter exhibits greater than 95% transmission for light with wavelengths longer than 900 nm, and decreases to less than about 0.1% transmission for light with wavelengths less than 800 nm with a smooth transition thereinbetween.

Fig. 4 shows the photodetector output 26 versus wavelength of incident light 12 with the fixed voltage bias V_▲ set at about 4.5 volts, the variable voltage bias V_p set at about 4, 6, 8, 10, and 12 volts during separate trials, and the fixed filter 27. The fixed filter 27 achieves very good selectivity, and the maximum voltage response is fairly constant as a function of the variable voltage bias V_p. The ratio of the maximum response for V_p at 869 nm in Fig. 4 to the response in the spurious peak for the photodetector output 26 at 857 nm is about 1.9 to 1. Because this spurious peak is about the maximum response position of the V_p equal 4 volt curve, this ratio is a good measure of the wavelength selectivity of the photodetector 10.

It is envisioned that a plurality of the photodetectors 10 of Fig. 1 may be situated in a two dimensional array for spectroscopy and imaging applications. In this application, each of the photodetectors 10 would detect a particular element of an image exposed to the array.