WO2001028001A1

WO2001028001A1 - Optical interconnect using multifunctional optical elements

Info

Publication number: WO2001028001A1
Application number: PCT/US2000/041160
Authority: WO
Inventors: John Trezza
Original assignee: Lockheed Corp; Lockheed Martin Corp
Current assignee: Lockheed Martin Corp
Priority date: 1999-10-13
Filing date: 2000-10-13
Publication date: 2001-04-19
Anticipated expiration: 2002-04-13
Also published as: AU1821001A; WO2001028001A9

Abstract

An optical-electronic interface device (10) includes a two-dimensional array (12) of emitter/detector elements (14), wherein each emitter/detector element (14) includes a semiconductor structure adapted for operation as a light emitting diode and as a diode detector in response to biasing thereof; and an integrated circuit (16) connected to the array (12) of emitter/detectors (14) and having a separate biasing circuit (18) connected to each emitter/detector element (14), wherein each biasing circuit (18) is adapted to bias its respective emitter/detector element (14) as either an emitter or a detector. The emitter/detector elements (14) for such a device (10) could include a PIN structure (40) with an intrinsic region (44) having a multiplicity of quantum wells adapted to either emit or detect light depending upon the biasing thereof.

Description

OPTICAL INTERCONNECT USING MULTIFUNCTIONAL OPTICAL ELEMENTS

Related Applications

The present application claims priority from US provisional patent application no. 60/159,167 filed October 13, 1999, entitled METHOD AND APPARATUS FOR IMPLEMENTING AN OPTICAL INTERCONNECT USING MODULATED DETECTORS.

Background of the Invention

Field of the Invention

The present invention generally relates to optical data communications systems, and particularly to such systems having two-dimensional, opto-electronic arrays of emitters and/or detectors for interfacing an optical medium and an electronic medium.

Statement of the Prior Art

The technology associated with digital communications has evolved extremely rapidly over recent years. Computers and related peripheral equipment, satellite and communication systems are becoming ever more sophisticated and powerful, demanding constantly higher rates of data communications. Unfortunately, data transfer remains a gating capability. This challenge holds for data transfer within an integrated circuit, from one chip to another, from hybrid circuit to hybrid circuit, from integrated circuit board to another integrated circuit board, and from system to system.

Data transfer rates have typically been increased by increasing both the data rate per data channel and by increasing the number of data channels. Data rates per channel ha\e been increased by the use of photonic or light energy. The numbers of data channels have been increased, even in photonic communications, by increasing the number of optical media channels, such as optical fibers.

Lately, significant effort has been invested into the use of frequency or wavelength modulation to create multiple data channels in the same medium. Recent approaches have also used integrated-circuit, opto-electronic components to drive and detect the optical data signals and thereby provide a higher channel density within a given amount of circuit space. Such high-density arrays would typically include both emitters and detectors for respectively generating and reading photonic signals. In order to achieve high density, such emitters and detectors would be flip-chip bump-bonded onto ASIC circuitry having the necessary respective drive and biasing circuits.

The emitters and detectors may be arranged on the ASIC in any pattern suitable for the intended application. For example, duplex communications could be provided by emitters and detectors being mounted to the same ASIC, which ASIC would have a corresponding arrangement of drive and sensing circuits constructed therein, thereby allowing use of a single bundle of optical fibers to create a single duplex link. This arrangement supports a network node which receives data from one fiber optic cable and transmits it through another. The most popular emitter for such applications is the Vertical Cavity Surface Emitting Laser, VCSEL, which can be driven at ver> high switching speeds to once again increase the data transmission rate. The detector used would typically be a PIN diode.

It would therefore be advantageous to have a massively parallel fiber optic interconnect transceiver with enhanced adaptability to different applications and also with lower fabrication costs than currently contemplated approaches. Summary of the Invention

Accordingly, an optical-electronic interface device includes a two-dimensional array of emitter/detector elements, wherein each emitter/detector element includes a semiconductor structure adapted for operation as a light emitting diode and as a diode detector in response to biasing thereof; and an integrated circuit connected to the array of emitter/detector elements and having a separate biasing circuit connected to each separate emitter/detector element, wherein each biasing circuit is adapted to bias its respective emitter/detector element as either an emitter or a detector.

The emitter/detector elements for such a device could include a PIN structure with an intrinsic region having a multiplicity of quantum wells adapted to either emit or detect light depending upon the biasing thereof.

The present invention could thereby provide an opto-electronic transceiver that is less complicated and less expensive than VCSEL based circuits because all of the emitters and detectors are identical. Identical emitter/detector elements could easily be fabricated on the same substrate and then flip-chip bump bonded to the drive ASIC without the use of complicated processes, such as the separate and different fabrication of emitters and detectors followed by separate bump-bonding and processing of those elements to a single ASIC.

Brief Description of the Drawings

The present invention is illustratively described in reference to the appended drawings in which:

Fig. 1(a) is a representational side view of an opto-electronic interface device constructed in accordance with one embodiment of the present invention;

Fig. 1(b) is a representational top view of the device of Fig. 1(a);

Fig. 2 is a representational, cross-sectional view of an emitter/detector element constructed in accordance with an embodiment of the present invention;

Fig. 3 is a schematic diagram of a bias circuit for forward biasing the device of Fig. 2;

Fig. 4 is a schematic diagram of a bias circuit for reverse biasing the device of Fig. 2; and

Fig. 5 is an diagram of electrical characteristics of a light emitting diode and a VCSEL.

Detailed Description of the Drawings

An opto-electronic interface device 10 is shown in Figs. 1(a) and 1(b) in which a two dimensional array 12 of emitter/detector elements 14 is constructed in combination with an ASIC CMOS 16. Each of the emitter/detector elements 14 may be biased to function as either an emitter or a detector. The ASIC CMOS 16 includes circuitry 18 integrally constructed in the upper surface 20 of a substrate 22. Circuitry 18 includes an individual biasing circuit for each of the emitter/detector elements 14, the construction of which biasing circuit is described below.

Electrical connections 26 are representationally shown on the bottom of substrate 22 for accessing the circuitry 18. In practice, connections 26 may take the form of plated vias passing through substrate 22 or of any other suitable electrical connections, located in any suitable area on substrate 22, for providing access to the circuitry 18. Connections 26 provide data to and from individual elements 14 as well as power to all of circuitry 18.

The array 12 is shown surrounded by a peripheral epoxy standoff 30 which is used to mount an optical medium 32 to pass photonic energy to and from array 12. Standoff 30 protects the array from contact with optical medium and also isolates the array 12 from contamination once the optical medium is attached. Optical medium 32 may include any suitable waveguide means, such as a light collimator or lens or a fiber optic bundle. Using a light collimator or lens is helpful in removably coupling array 12 to a fiber optic bundle. In this manner, removal of the bundle will not result in the exposure of array 12 to atmospheric contaminants.

An individual emitter/detector element 14 is representationally shown in Fig. 2 for the purpose of explaining the different material layers used to construct those elements. Emitter/detector element 14 generally includes a PIN semiconductor structure 40 having a P-doped layer 42, an intrinsic layer 44 and an N-doped layer 46. PIN structure 40 is enclosed on top and bottom by mirrors 48, 50 respectively.

Intrinsic layer 44 may be made of bulk undoped material, or alternatively of a multiplicity of quantum wells. P-doped layer 42 and N-doped layer 46 are respectively doped with material to create semiconductors with respective excesses of holes and electrons.

Mirrors 48, 50 typically include multiple alternating layers of two different semiconductor materials which are thermally and lattice matched, but which have different indexes of refraction. The different refractive indexes cause a certain percentage of light to be reflected at each transition from one layer to another. In addition, bottom mirror 50 has a reflectivity of 100 percent for practical purposes while top mirror 48 has a much lower reflectivity in the range of 50 to 75 percent. This difference in reflectivity is caused by different numbers of alternating layers and allows light to partially reflect internally and to partially pass through top mirror 48 in the manner described below. In one form, such mirrors are well known and referred to as distributed Bragg reflectors. Lastly, top mirror 48 is P-doped to create a conduction path to layer 42, and bottom mirror 50 is N-doped to create a conduction path to N-doped layer 46.

PIN structure 40 may be constructed by any suitable process. One example is the use of molecular beam epitaxy. Beginning with mirror 50, alternating layers of indium gallium arsenide, InGaAs and indium phosphide, InP, are deposited on a substrate thus forming the distributed reflector. This bottom layer is heavily N-doped, so obtaining an ohmic contact is straightforward. Then, in the same reactor, an N-layer 46, such as N-doped InGaAs is deposited. This is followed by the deposition of the undoped intrinsic layer 44, which may again be InGaAs. The intrinsic layer 44 is followed by depositing a P-layer 42, such as P-doped InGaAs. Then, another sequence of alternating layers of InGaAs and InP are deposited to form a second mirror 48. Finally, conductive leads are attached to the second mirror 48, and the device is functional. The use of similar materials for each of the layers provides lattice matching between layers and prevents thermal stress. The InP layers are likewise lattice matched to InGaAs.

In operation, a voltage bias is applied across the PIN structure 40 through mirrors 48, 50 to cause the intrinsic layer 44 to either emit light or absorb light depending upon the polarity of the bias voltage. A forward bias, applying a relative positive voltage to the P-doped layer 42, causes excess electrons and holes to enter the intrinsic layer 44 and combine, releasing energy in the form of light. This light is fully reflected by bottom mirror 50, partially reflected by top mirror 48 and partially emitted through top mirror 48. It is well known that the wavelength of the light so generated is dependent upon the material and construction of intrinsic layer 44.

In the case of reverse bias, the voltage applied to P-layer 42 is negative with respect to the voltage of N-layer 46. Reverse bias causes any excess holes in P-doped layer 42 and intrinsic layer 44 to be attracted to top mirror 48 and any excess electrons in N-doped layer 46 and intrinsic layer 44 to be attracted to bottom mirror 50. Thus, reverse bias removes substantially all current carriers from the PIN structure. When light of an approximate wavelength passes through top mirror 48 and enters intrinsic layer 44, it is absorbed thereby and causes the separation of electrons and holes in the intrinsic layer 44. The wavelength at which such light is absorbed is dependent upon the material and construction of the intrinsic layer 44 in the same relation as the light generated under forward bias. The electrons and holes which are separated by absorbed light are similarly attracted to the respective N-doped and P-doped layers by the reverse bias voltage. The movement of these electrons and holes may be measured as current flowing through PIN structure 40.

PIN structure 40 has a further operative condition in which no bias is applied thereto, thereby causing neither light emission nor light absorption, but rather reflection thereof. As a result, a substantial portion of the light entering through mirror 48 reflects off of mirror 50 and exits through mirror 48. Thus, applying substantially zero bias to the element 14 causes it to be a reflector. Substantially zero bias means any voltage between zero and approximately 0.7 volts forward bias, wherein PIN structure 40 neither emits nor absorbs light.

Various modifications may be made to the PIN structure just described while maintaining the intended functionality. PIN structure 40 may be constructed in the form of a resonant cavity by forming the layers of mirrors 48, 50 to have a thickness equivalent to one-quarter wavelength of the peak emission/absorption wavelength of intrinsic layer 44 in each material. This causes reflections from the separate mirror layers to be coherent upon reflection by the mirrors. Further, the total thickness of layers 42, 44 & 46 could be similarly constructed to have a thickness equal to an odd multiple of the half wavelength of the emission/detection wavelength of the intrinsic material. This arrangement causes reflections from the mirrors 48, 50 to be coherent and create standing waves within the PIN structure for a higher output power. The use of odd multiples of the half wavelength causes the standing waves to have a signal peak or anti-node in the intrinsic material thereby improving the power generated as well as detection sensitivity.

Intrinsic layer 44 may be constructed with quantum wells fabricated in either a superlattice or miniband. Quantum wells have a peak absorption (and emission) wavelength dependent upon the energy difference between their quantized valence and conduction energy levels. This energy difference will vary according to the construction characteristics of the quantum wells, such as thickness and composition, and it will also vary in accordance with the energy potential profile across the intrinsic layers. The use of such quantum well superlattice and miniband structures improves the detection sensitivity of the intrinsic layer 44 and their construction processes are well know in the art as exemplified by US Patent 5,166,946.

As mentioned, the peak absorption wavelength will vary with the voltage or energy profile across the intrinsic layer 44. Thus, varying the reverse bias potential will vary the peak absorption wavelength and thereby vary the amount of light absorbed at a given wavelength by the intrinsic layer 44 in accordance with the quantum Stark effect. Inasmuch as light is reflected from PIN structure 40 if that light is not absorbed, varying the absorption through the reverse bias voltage will vary the level of reflected light and thereby cause the reflected light to be modulated. Such quantum wells could be constructed in the same process that is described above for bulk material layers. Intrinsic layer 44 would be constructed with alternating layers of InGaAs and InP having thickness on the order of 0.1 run., as is known for the fabrication of such devices. A suitable quantum well structure could be fabricated for the present embodiment with anywhere from a few to approximately 20 such alternating layers.

The emission wavelength is determined by the intrinsic materials used, as mentioned above. One set of materials leads to a device that operates at a wavelength of 700 nm to 1.3 um. The most widely used wavelength is in the 845 nm to 865 nm range and is achievable with a GaAs/AlGaAs material system. The same functionality, though at different wavelengths, can be obtained by those skilled in the art using other materials such as indium phosphide, or ternary or quaternary combinations of semiconductor materials.

Fig. 3 shows a biasing circuit for creating a forward bias across the emitter/detector element 14 of Fig. 2. Element 14 is represented as diode 60 to which is applied a positive voltage for forward biasing. For this purpose, a positive V_DD is connected to the positive terminal of diode 60 and the negative terminal is connected to a simple gate, switch or operational amplifier 62. Operational amplifier 62 alternatively connects a lower voltage, such as ground or V_ss, verses V_DD, to the negative terminal of diode 60 in response to a data stream presented on input terminals 64 of operational amplifier 62. Light is thereby emitted by diode 60 whenever V_ss is applied to its negative terminal.

Fig. 4 shows a biasing circuit for creating a reverse bias across the emitter/detector element 14 and for sensing light absorbed in the detector. Element 14 is again represented as diode 60 to which a relatively negative V_ss is applied to the positive terminal of diode 60. The negative terminal of diode 60 is series connected through a pair of resistors 66, 68 to a relatively positive voltage V_DD and the inputs of a trans-impedance amplifier 70 are connected across the resistor 66 which is connected directly to diode 60. In this manner, diode 60 is reverse biased, and light at the proper wavelength, which light releases current in diode 60, will be detected by amplifier 70.

Because the emitter/detector element 14 may be either forward or reverse biased, it may be biased by a circuit capable of controllably biasing it in either direction. Such an arrangement makes emitter/detector element 14 more useful by reducing the number of required emitter/detector elements and their corresponding footprint.

The performance of the quantum well LED structures described in the present invention is different from that obtained when using VCSEL structures that are designed to solely produce laser light, which difference is depicted in Fig. 5. A laser is a threshold device, meaning that it is either "on" or "off. Furthermore, when the bias voltage is raised above a threshold value, the light generated exceeds the internal losses of the device and there is an abrupt increase in output light. When the bias is subsequently decreased, the light being emitted decreases to zero. Consequently, one can use a laser as a switching device by switching between a bias condition below the threshold value and a second bias condition above the threshold value. These two bias conditions are relatively close to one another compared to a similar structure that is emitting in a generally incoherent mode such as the present invention.

In a quantum LED structure operating in a light-emitting mode, the output power (light) increases slowly from zero, and then rises more rapidly as the PIN reaches its threshold value. Because the change in output light intensity is much more gradual for the PIN structure than for a laser structure, the PIN structure must be switched between bias ranges that are much greater than those for lasers. To increase the output light by an amount ΔI. corresponding to a factor of 10 increase in output power, a VCSEL requires a shift in bias of ΔV _Laser compared with a much larger amount of bias change, ΔV_LED, required by the LED.

Conclusion

The present invention provides an opto-electronic transceiver that is less complicated and less expensive than VCSEL based circuits, which include separate types of elements for emitting and detecting. The construction process of the similar devices of the present invention significantly reduces the cost of the devices. Because the elements of the present invention are alike they may be constructed in arrays and connected as such to the ASIC CMOS. More expensive construction methods, which mix different emitters and detectors in the same array, can be avoided. Also, the fabrication process for LED devices has fewer control parameters than for VCSELs, providing higher yield and device consistency. Lastly, the ability to change the operating mode of each device allows the devices to be programmable and much more adaptable for different applications.

Lastly, the consistency of the elements and their arrays provides a much more universal or fungible product with a broader range of applications. Such arrays will more quickly become commodities in the same manner as the individual elements used to construct them.

The embodiments described above are intended to be taken in an illustrative and not a limiting sense. Various modifications and changes may be made to the above embodiments by persons skilled in the art without departing from the scope of the present invention as defined in the appended claims.

Claims

WHAT IS CLAIMED IS:

1. An optical-electronic interface device, comprising: a two-dimensional array of emitter/detector elements, wherein each emitter/detector element includes a semiconductor structure adapted for operation as a light emitting diode and as a diode detector in response to biasing thereof; and an integrated circuit connected to the array of emitter/detector elements and having a separate biasing circuit connected to each separate emitter/detector element, wherein each biasing circuit is adapted to bias its respective emitter/detector element as either an emitter or a detector.

2. The device of claim 1, wherein each of the emitter/detector elements includes a PIN structure and upper and lower mirrors located around the PIN structure, with the lower mirror having substantially complete reflectivity and being located adjacent to the integrated circuit, and the upper mirror having a substantially less than complete reflectivity to allow the passing of light both into and out of its respective emitter/detector element.

3. The device of claim 2, wherein each mirror is adapted to coherently reflect light at a first predetermined wavelength and further wherein the PIN structure is adapted to form a resonant cavity at the first predetermined wavelength.

4. The device of claim 3, wherein each mirror is a distributed Bragg reflector.

5. The device of claim 3, wherein the PIN structure is adapted to form a resonant cavity at an odd multiple of a half wavelength of the first predetermined wavelength.

6. The device of claim 2, wherein the PIN structure has a central intrinsic region, without semiconductor doping, which intrinsic region includes a multiplicity of quantum wells adapted to cause the emission or detection of light depending upon biasing of the PIN structure.

7. The device of claim 5, wherein the multiplicity of quantum wells are adapted to emit and detect light at a second predetermined wavelength.

8. The device of claim 7, wherein each mirror is adapted to coherently reflect light at the second predetermined wavelength and further wherein the PIN structure is adapted to form a resonant cavity at the second predetermined wavelength.

9. The device of claim 8, wherein the PIN structure is adapted to form a resonant cavity at an odd multiple of a half wavelength of the second predetermined wavelength.

10. The device of claim 2, wherein the PIN structure has a central intrinsic region made of bulk undoped material.