WO1997009743A1

WO1997009743A1 - Pin photodiode having elliptical detector area

Info

Publication number: WO1997009743A1
Application number: PCT/US1996/014366
Authority: WO
Inventors: Randall Brian Wilson
Original assignee: The Whitaker Corporation
Priority date: 1995-09-07
Filing date: 1996-09-09
Publication date: 1997-03-13

Abstract

The present invention is a PIN photodiode having a substantially elliptical cross section active area. To this end in order to effect the transmission and reception of high data rate transmissions, the switching speeds of digital signals themselves must be great. As can be readily understood, the faster the switching speed of the digital pulse that is required, the less tolerance there is for the ill-effects of RC transience that are often intrinsic properties of semiconductor detectors. One major component of the transient tails is the intrinsic capacitance of the device. Accordingly, it is important to high data rate transmission (fast rise and fast fall times) to reduce as greatly as possible the intrinsic capacitance of the device. Accordingly, what is needed is a maximization of the active area of the device. What this means is a geometric match of the active area that most closely replicates the cross-sectional geometry of the optical beam impingent on the detector. In many applications in fiber optic communications, light is reflected off of a surface at an angle and accordingly light of circular cross section is reflected and emerges with elliptical cross section. What has been done in the past is to make sure that the cross section of the active area of the detector, usually circular, has been great enough to enable light of elliptical cross section reflected from a surface to be impingent upon the circular area of the detector. This means usually that there is 'unused' area in the active area of the detector. Because a PIN device is intrinsically a capacitor this 'unused' area merely increases the intrinsic capacitance of the detector, an unacceptable result in high frequency data transmission.

Description

PIN Photodiode Having Elliptical Detector Area

The invention relates to a P-I-N photodetector for use in optical communications.

Optical frequency communications have evolved greatly over the past couple of decades, but there is a need to increase the capabilities of data transmission. To this end, for example in the arena of digital signal transmissions, in order to effect the transmission and reception of high data rate transmissions, the switching speeds of the digital signals themselves must be great. As can be readily understood, the faster the switching speed of the digital pulse required, the less the tolerance for the ill-effects of RC transients that are often intrinsic properties of the semiconductor transmitters and detectors. One major component of the transient "tails" is the intrinsic capacitance of the device. As is obvious, it is important to high data rate transmission (fast rise and fall times) to reduce as greatly as is possible the intrinsic capacitance of a device in the transmission link. Examples of attempts to curb the ill-effects of parasitic capacitance can be found in related U.S. patents 5,003,358; 5,100,833; 5,194,399 and 5,275,968 to Takahashi, et al. incorporated herein by reference.

It is important to recognize that due to the absence of a gain mechanism in the PIN diode, the gain- bandwidth product is nearly equal to the bandwidth itself, the intrinsic bandwidth determined by the transit time of electron-hole pairs, and accordingly by the thickness of the intrinsic layer. Accordingly, the thickness of the intrinsic layer's effect on absorption efficiency must be balanced against its ill-effect on time of transit. However, the practical bandwidth of the PIN detector and first input stage, which comprise the optical receiver, can be limited by the RC time constant. The RC time constant is a known quantity to the skilled artisan, and C is the capacitance of the PIN detector and R is the effective input impedance of the pre-amplifier. It is desirable to make C as low a value as possible to maximize the RC limited bandwidth, while retaining design flexibility in the input impedance parameter R. As can be readily appreciated from a review of either the devices above described, the PIN detector is intrinsically a capacitor.

The capacitance C arises primarily from the junction capacitance, which is proportional to the area of the active region. In addition, parasitic capacitance can be the result of additional bonding pads located away from the active region. Parasitic capacitance from offset bonding pads can be significant, noticeably reducing the receiver bandwidth. The intrinsic RC transients affect the performance of the device in its switching speed. In a digital signal, this is manifested as an exponential tail in the time domain, and as a reduction in the bandwidth in the frequency domain. Accordingly, these effects result in reduction of signal capacity at higher frequencies, as well as a reduction in square wave definition. The latter is a result of the fact that square wave definition is greater the greater the number of Fourier components that define the square wave. Since the higher frequency components are lost due to intrinsic capacitance, the definition in the square wave is lost as well.

One approach to reducing the detector capacitance is to reduce the area of the detector, since the capacitance, and thus the deleterious effects of the RC time constant, is proportional to the detector area and will be accordingly reduced. However, this approach is limited by the finite width intensity distribution of the optical beam impinging on the photodetector. When the area of the incident optical beam cross-section is significantly larger than the area of the active region of the detector, there are adverse affects. These are as follows. The photons of the incident beam which are absorbed in the undepleted region adjacent to the active region result in lateral diffusion limited currents which are especially deleterious to the detector bandwidth as explained above.

Accordingly, the present invention increases the bandwidth of the detector by reducing the capacitance through reducing the detector area as well as by minimizing the area of the bonding pads, but also curbs the ill-effects on bandwidth of the diffusion current in the undepleted region by matching very closely the beam geometry and area to the detector geometry and area. To be clear, the area of the detector is reduced to reduce capacitance, but is not reduced so much that there is a significant amount of light impingent on the undepleted region outside the active region.

It is an object of the present invention to have a detector with a detector geometry and area substantially matching the cross-sectional geometry and area of the incident light.

It is a feature of the present invention that the intrinsic capacitance of the detector is greatly reduced by reducing the area of the detector active region.

It is a further feature that bonding pad geometry and area matches the wedge bond geometry and area. It is a further feature that a new p-contact geometry which protects the active region from wire bonding draws and provides good adhesion is described. It is a feature of the present invention that the bonding pad shape and area substantially matches the pad shape and area of the wedge wirebond which is used to attach the external circuit to the device.

It is a further feature of the present invention that a new p-contact structure that substantially protects the active region of from wirebonding damage and provides goods adhesion of the p-contact metallization.

It is an advantage of the present invention that the reception bandwidth of the detector is increased by the reduction of the intrinsic device capacitance. Invention will now be described by way of example with reference to the accompanying drawings in which:

Figure 1 is a cross-sectional view of the present invention where light is received at the rear surface of the detector.

Figure 2 is a top view of the present invention showing the elliptical annular contact.

Figure 3 is a top view of a conventional detector showing the bonding pad arrangement.

Figure 4 is a top view of the detector of the present invention showing the bonding pad structure with wire attached.

Figure 5 is the superposition of an ellipse on a circle where the major axis of the ellipse equals the diameter of the circle. The shaded region represents the "unused" detector area of a detector having a conventional circular detector area with the typical elliptical incident beam cross-section superposed thereon.

Figure 6 is the superposition of an ellipse on a circle where the circle represents the active area of a conventional detector and the ellipse represents the elliptical optical beam. The shaded areas are where the lateral diffusion current is produced.

Figure 7 is the superposition of an ellipse on a square where the major axis of the ellipse equals the length of the side of the square. The shaded region represents the "unused" detector area of a detector having a conventional square detector area with the typical elliptical incident beam cross-section superposed thereon.

Figure 8 is the superposition of an ellipse on a square where the square represents the active area of a conventional detector and the ellipse represents the elliptical optical beam. The shaded areas are where the lateral diffusion current is produced.

Figure 9 shows a typical system in which a substantially circular wave front impinges on a detector resulting in an elliptical wavefront due to the angle of incidence of the wavefront.

Figure 10 shows a typical system in which a wavefront is impingent on the detector having elliptical cross-section due to refractive affects of a cylindrical optical surface.

Figure 11 shows the refraction of the light from a fiber due to the cylindrical surface of the fiber.

Turning to Figures 1,2, we see the cross sectional and corresponding top views of the instant invention. Figure 4 shows a detector that receives light at its back surface. For purposes of discussion, the fabrication is the same for each, unless described otherwise. The doping types and levels of dopants described herein are exemplary and are in no way intended to be limiting, but rather merely descriptive. Accordingly, variations of the dopant types and doping levels that are within the purview of the artisan of ordinary skill are considered within the purview of the present invention. The base substrate of n InP 101 is grown by standard LEC techniques having been doped to a

18 —3 19 ~\ level in the range of 3 x 10 cm to 3 x 10 cm . Next a layer of n-doped InP 102 having a doping level in the range 2 x IO¹⁵ cm^"3 to 1 x IO¹⁶ cm^"3 is grown epitaxially. The epitaxial growth of this layer and subsequent layers is performed preferably by standard liquid-phase-epitaxy (LPE) techniques, but also by metal-organic-vapor-phase- epitaxy (MOVPE) . This n-type layer has grown thereon a layer of intrinsic (undoped) InGaAs 103, and functions as the photosensitive layer of the detector.

Thereafter a layer of

104 is grown. This layer is known as the cap layer to the artisan of ordinary skill. The p⁺ region 105 is formed by selective diffusion preferably of zinc through a patterned mask 106 of material such as SiN_x or Si0₂ or a combination thereof as described before. Thereafter an anti¬ reflective layer 107 of suitable material preferably of Si₃N₄ is deposited, and has a portion 109 in the window of the diffusion mask, the advantages of which are described herein. The elliptical p contact 108 is fabricated using standard photolithographic lift-off techniques and metal evaporation techniques, as are well known, and an n contact 110 is formed similarly. The p- side contact is substantially elliptical as shown in Figure 5. The p-side contact 108 is formed first by standard Ti-Pt-Au electron beam deposition techniques. The n-side contact 110 is formed by GeNiAu followed by Au plating for suitable solder bonding.

A unique feature of the structure of the embodiment of Figures 1 and 2 is that the p-side metallization is deposited over an elliptical Si₃N₄ pad region 109 which forms the inner part of the ring contact structure. By retaining the elliptical Si₃N₄ pad 109, the adherence of the gold contact pad is greatly improved since gold adheres better to the Si₃N₄ than it does to semiconductor. More importantly, this elliptical pad 109 protects the underlying active region from damage induced by the wire bonding process used to make the electrical connection to the anode of the device. Furthermore, the elliptical shape of the contact metallization matches well the elongated shape of the wire bond attachment created by the wedge bonding process. All of these factors allow the elimination of an additional wire bond pad, used in conventional devices and which adds parasitic capacitance as discussed previously.

The main feature of the structure of the present invention is that the elliptically shaped active region allows a minimization of the capacitance of the device. Junction capacitance reduces the effective bandwidth of the detector since transient signals reduce the effective switching capabilities of the device and the frequency bandwidth of the signal. This minimization of capacitance increases the operation speed and switching speed of the device. The means by which the capacitance reduction is accomplished is as follows. As can be appreciated to the artisan of ordinary skill, a P-I-N diode is intrinsically a capacitor, having a layer of one type of charge (p-doped layer, for example) , a layer of intrinsic semiconductor, which is in essence a functional insulator, and a layer of material of opposite type charge to that of the first layer (the n- doped layer, for example) . Because the area of the layers of the charge are directly proportional to the capacitance, the greater the area of the p-doped and n- doped layers, the greater the capacitance. Conversely, one way to reduce the intrinsic capacitance of the diode would be to reduce the area of the "charged" layers. To this end, the present invention reduces the capacitance by reducing the area of the capacitor made up of the P-I-N detector.

Figures 5 and 7 are representative of conventional detectors in which not all of the active region is utilized in the detection of the optical signal. To this end, Figure 5 shows an overlap of a geometric ellipse 501 with a major axis 502 overlapping and equal to the diameter of the geometric circle 503. The area in which there is no overlap of the circle 503 and the ellipse 501 is shown as the shaded section 504. In many modern optical systems, the shape of the l/e intensity level of the optical signal is that of an ellipse. For example, in some systems, the elliptical cross-section is a result of refraction of light of nearly circular cross-section through an asymmetric refractive element. Additionally, the elliptical cross-section can result from off-axis coupling of a nearly circular beam and a photodetector in which the beam impinges on the photodetector at an angle which is not orthonormal to the surface of the photodetector. Furthermore, as are encountered in many typical applications, the characteristic cross-section of output of a laser or other asymmetric optical beams is elliptical. In contradistinction to the above, many conventional optical systems had beam front cross-sections that were more circular. Accordingly in many conventional applications, the P-I-N detector has a photosensitive area which is circular or square. When such a detector is used in a system having an elliptical shaped beam, a disadvantage results. The area of the detector is like that of the circle 503 and the beam front has a cross- section like the ellipse 501. Clearly, the shaded region of the circle 504 represents that portion of the detector which is "unused" in the sense that very little, if any optical signal is impingent there, and yet this portion adds to the junction capacitance of the device. Thus, there is unused area of the P-I-N photosensitive region having a circular photosensitive area. To the extent that this unused area can be eliminated, the intrinsic capacitance of the detector can be reduced. Again, this is because the capacitance is directly proportional to the area of the charge, in this case the photosensitive area of the detector. This is the essence of the invention. By maximizing the area utilized by matching the area of beam cross-section of the optical signal, the intrinsic capacitance of the device is kept to a minimum. This feature as explained in detail above results in a detector with greater switching speed and thus greater bandwidth. Figure 7 shows the overlap of a geometric ellipse 701 having a major axis of a length equal to the length of a side of a square 702. In a conventional detector having a square detector, the shaded area 703 represents the "unused" detector area and like the circular detector described above, this area results in an increased detector intrinsic capacitance relative to an elliptical detector area of the present invention. In contrast, however, Figures 6 and 8 are representative of conventional detectors in which the detector area is not sufficient to capture the expanse. The result here is the ill-effect of lateral diffusion currents created in the undepleted region adjacent to the active region. In Figures 6 and 8, the detector active region is shown as the circle 601 and square 801, respectively. The incident light beam is shown as the ellipses 602 and 802. As stated above, one approach to reducing the detector capacitance is to reduce the area of the detector, since the capacitance, and thus the deleterious effects of the RC time constant, is proportional to the detector area and will be accordingly reduced. However, this approach is limited by the finite intensity distribution of the optical beam impinging on the photodetector. When the area of the incident optical beam cross-section is significantly larger than the area of the active region of the detector, there are adverse effects. These are as follows. The photons of the incident beam which are absorbed in the undepleted region adjacent to the active region result in lateral diffusion currents which are especially deleterious to the detector bandwidth. As represented in Figures 6 and 8, the shaded regions 603 and 803 represent the light impingent on the undepleted regions of the photodetector.

The present invention is drawn to curbing the ill- effects of both a detector having "unused" detector area as well as the ill-effects of having light impingent outside the active area of the detector. To this end, the present invention increases the bandwidth of the detector by reducing the capacitance through reducing the detector area as well as by minimizing the area of the bonding pads, but also curbs the ill-effects on bandwidth of the diffusion current in the undepleted region by matching very closely the beam geometry and area to the detector geometry and area. To be clear, the area of the detector is reduced to reduce capacitance, but is not reduced so much that there is a significant amount of light impingent on the undepleted region outside the active region.

Examples of an application of the present invention which has significant advantages over conventional detectors are shown in Figures 9 and 10. Figure 9 shows an optical fiber 901 in which light is reflected from a reflective surface 902. A symmetrical beam of circular cross-section emanates from single mode fiber and is deflected from the reflective surface. Because the reflective surface deflection is greater than 90°, the intensity distribution at the active region of the detector deviates from the circular cross-section and is approximately elliptically shaped, allowing application of the present invention. Figure 13 shows a typical application in which the reflective surface is at the endface of the fiber, and refraction at the cylindrical surface of the fiber transforms a circular cross-section beam to an elliptical one. The refraction is shown in Figure 11, where the line A-A' is along cylindrical surface of the fiber, which thereby acts as a cylindrical lens. In summary, while other applications are within the purview of the artisan of ordinary skill, four uses of the present invention are envisioned. In the first case, light is incident to the detector at other than parallel to the normal to the detector surface. If the light is substantially circular, angular incidence will result in a beam with an approximately elliptical cross-section at the detector surface. In the second case, the light will be impingent on an asymmetrical optical element, such as a diffractive or refractive element to transform a circular cross-section light beam to an elliptical one. In the third case, certain devices such as lasers emit beams that are characteristically elliptical in cross- section. Finally, the detector of the present invention is intended for use in a bidirectional module using an asymmetrical holographic optical element. Such a device is a disclosed in U.S. Patent Provisional Application No. 60/004,504, the disclosure of which is incorporated by reference.

The preferred embodiments having been described in detail, it will be appreciated that various changes and modifications can be made therein without departing from the theme and spirit of an optical detector that reduces intrinsic capacitance by closely matching the shape of the photosensitive area to the shape of the optical beam cross-section. This maximization of the use of the photosensitive area thus minimizing the intrinsic capacitance of the photodetector is the crucial feature of the invention. Modifications of this feature as would be obvious to the artisan of ordinary skill are intended to be within the scope of the invention.

Claims

WE CLAIM:

1. A photodetector having a substrate, a layer of semiconductor material of a first doping type disposed on said substrate, a layer of intrinsic semiconductor material forming a photosensitive region disposed on said layer of semiconductor material of said first doping type, and a region of semiconductor material of a second doping type disposed on said intrinsic semiconductor material, characterized in that: said region of semiconductor material of said second doping type has a substantially elliptical cross section.

2. A photodetector as recited in claim 1 further characterized in that: said substrate is InP.

3. A photodetector as recited in claim 1 further characterized in that: said layer of semiconductor material of said first doping type is n-doped InP.

4. A photodetector as recited in claim 1 further characterized in that: said intrinsic material is In^Ga_xAs or In₁._xGa_xAs_yP₁._y.

5. A photodetector as recited in claim 1 further characterized in that : said region of semiconductor material of a second doping type is diffusion doped In-_L.j-Ga_xAS_yP-_L._y.

6. A photodetector as recited in claim 1 further characterized in that: said photodetector is a rear illuminated photodetector.

7. A photodetector as recited in claim 6 further characterized in that: an elliptical contact pad forms an electrical contact for said semiconductor material of said second doping type.

8. A photodetector as recited in claim 7 further characterized in that: a layer of Si₃N₄ is disposed between said elliptical contact pad and a cap layer.