Flip-chip heterogeneously integrated n-p-n InGaP/GaAs heterojunction bipolar transistors (HBTs) with integrated thermal management on wide-bandgap AlN substrates followed by GaAs substrate removal are demonstrated. Without thermal management, substrate removal after integration significantly aggravates self-heating effects, causing poor $I$-$V$ characteristics due to excessive device self-heating. An electrothermal codesign scheme is demonstrated that involves simulation (design), thermal characterization, fabrication, and evaluation. Thermoreflectance thermal imaging, electrical-temperature sensitive parameter-based thermometry, and infrared thermography were utilized to assess the junction temperature rise in HBTs under diverse configurations. In order to reduce the thermal resistance of integrated devices, passive cooling schemes assisted by structural modification, i.e., positioning indium bump heat sinks between the devices and the carrier, were employed. By implementing thermal heat sinks in close proximity to the active region of flip-chip integrated HBTs, the junction-to-baseplate thermal resistance was reduced over a factor of two, as revealed by junction temperature measurements and improvement of electrical performance. The suggested heterogeneous integration method accounts for not only electrical but also thermal requirements providing insight into realization of advanced and robust III-V/Si heterogeneously integrated electronics.
We report on the development of single-frequency VCSELs (vertical-cavity surface-emitting lasers) for sensing the position of a moving MEMS (micro-electro-mechanical system) object with resolution much less than 1nm. Position measurement is the basis of many different types of MEMS sensors, including accelerometers, gyroscopes, and pressure sensors. Typically, by switching from a traditional capacitive electronic readout to an interferometric optical readout, the resolution can be improved by an order of magnitude with a corresponding improvement in MEMS sensor performance. Because the VCSEL wavelength determines the scale of the position measurement, laser wavelength (frequency) stability is desirable. This paper discusses the impact of VCSEL amplitude and frequency noise on the position measurement.
Vertical-cavity surface-emitting lasers (VCSELs) are well suited for emerging photonic microsystems due to their low power consumption, ease of integration with other optical components, and single frequency operation. However, the typical VCSEL linewidth of 100 MHz is approximately ten times wider than the natural linewidth of atoms used in atomic beam clocks and trapped atom research, which degrades or completely destroys performance in those systems. This report documents our efforts to reduce VCSEL linewidths below 10 MHz to meet the needs of advanced sub-Doppler atomic microsystems, such as cold-atom traps. We have investigated two complementary approaches to reduce VCSEL linewidth: (A) increasing the laser-cavity quality factor, and (B) decreasing the linewidth enhancement factor (alpha) of the optical gain medium. We have developed two new VCSEL devices that achieved increased cavity quality factors: (1) all-semiconductor extended-cavity VCSELs, and (2) micro-external-cavity surface-emitting lasers (MECSELs). These new VCSEL devices have demonstrated linewidths below 10 MHz, and linewidths below 1 MHz seem feasible with further optimization.
Advanced optically-activated solid-state electrical switch development at Sandia has demonstrated multi-kA/kV switching and the path for scalability to even higher current/power. Realization of this potential requires development of new optical sources/switches based on key Sandia photonic device technologies: vertical-cavity surface-emitting lasers (VCSELs) and photoconductive semiconductor switch (PCSS) devices. The key to increasing the switching capacity of PCSS devices to 5kV/5kA and higher is to distribute the current in multiple parallel line filaments triggered by an array of high-brightness line-shaped illuminators. Commercial mechanically-stacked edge-emitting lasers have been used to trigger multiple filaments, but they are difficult to scale and manufacture with the required uniformity. In VCSEL arrays, adjacent lasers utilize identical semiconductor material and are lithographically patterned to the required dimensions. We have demonstrated multiple-line filament triggering using VCSEL arrays to approximate line generation. These arrays of uncoupled circular-aperture VCSELs have fill factors ranging from 2% to 30%. Using these arrays, we have developed a better understanding of the illumination requirements for stable triggering of multiple-filament PCSS devices. Photoconductive semiconductor switch (PCSS) devices offer advantages of high voltage operation (multi-kV), optical isolation, triggering with laser pulses that cannot occur accidentally in nature, low cost, high speed, small size, and radiation hardness. PCSS devices are candidates for an assortment of potential applications that require multi-kA switching of current. The key to increasing the switching capacity of PCSS devices to 5kV/5kA and higher is to distribute the current in multiple parallel line filaments triggered by an array of high-brightness line-shaped illuminators. Commercial mechanically-stacked edge-emitting lasers have been demonstrated to trigger multiple filaments, but they are difficult to scale and manufacture with the required uniformity. As a promising alternative to multiple discrete edge-emitting lasers, a single wafer of vertical-cavity surface-emitting lasers (VCSELs) can be lithographically patterned to achieve the desired layout of parallel line-shaped emitters, in which adjacent lasers utilize identical semiconductor material and thereby achieve a degree of intrinsic optical uniformity. Under this LDRD project, we have fabricated arrays of uncoupled circular-aperture VCSELs to approximate a line-shaped illumination pattern, achieving optical fill factors ranging from 2% to 30%. We have applied these VCSEL arrays to demonstrate single and dual parallel line-filament triggering of PCSS devices. Moreover, we have developed a better understanding of the illumination requirements for stable triggering of multiple-filament PCSS devices using VCSEL arrays. We have found that reliable triggering of multiple filaments requires matching of the turn-on time of adjacent VCSEL line-shaped-arrays to within approximately 1 ns. Additionally, we discovered that reliable triggering of PCSS devices at low voltages requires more optical power than we obtained with our first generation of VCSEL arrays. A second generation of higher-power VCSEL arrays was designed and fabricated at the end of this LDRD project, and testing with PCSS devices is currently underway (as of September 2008).
This report summarizes a 3-month program that explored the potential areas of impact for electronic/photonic integration technologies, as applied to next-generation data processing systems operating within 100+ Gb/s optical networks. The study included a technology review that targeted three key functions of data processing systems, namely receive/demultiplexing/clock recovery, data processing, and transmit/multiplexing. Various technical approaches were described and evaluated. In addition, we initiated the development of high-speed photodetectors and hybrid integration processes, two key elements of an ultrafast data processor. Relevant experimental results are described herein.
A previous LDRD studying radiation hardened optoelectronic components for space-based applications led to the result that increased neutron irradiation from a fast-burst reactor caused increased responsivity in GaAs photodiodes up to a total fluence of 4.4 x 10{sup 13} neutrons/cm{sup 2} (1 MeV Eq., Si). The silicon photodiodes experienced significant degradation. Scientific literature shows that neutrons can both cause defects as well as potentially remove defects in an annealing-like process in GaAs. Though there has been some modeling that suggests how fabrication and radiation-induced defects can migrate to surfaces and interfaces in GaAs and lead to an ordering effect, it is important to consider how these processes affect the performance of devices, such as the basic GaAs p-i-n photodiode. In this LDRD, we manufactured GaAs photodiodes at the MESA facility, irradiated them with electrons and neutrons at the White Sands Missile Range Linac and Fast Burst Reactor, and performed measurements to show the effect of irradiation on dark current, responsivity and high-speed bandwidth.
The authors have developed a chip-scale atomic clock (CSAC) for applications requiring atomic timing accuracy in portable battery-powered applications. At PTTI/FCS 2005, they reported on the demonstration of a prototype CSAC, with an overall size of 10 cm{sup 3}, power consumption > 150 mW, and short-term stability sy(t) < 1 x 10-9t-1/2. Since that report, they have completed the development of the CSAC, including provision for autonomous lock acquisition and a calibrated output at 10.0 MHz, in addition to modifications to the physics package and system architecture to improve performance and manufacturability.
Sandia National Laboratories and Mytek, LLC have collaborated to develop a monolithically-integrated vertical-cavity surface-emitting laser (VCSEL) assembly with controllable polarization states suitable for use in chip-scale atomic clocks. During the course of this work, a robust technique to provide polarization control was modeled and demonstrated. The technique uses deeply-etched surface gratings oriented at several different rotational angles to provide VCSEL polarization stability. A rigorous coupled-wave analysis (RCWA) model was used to optimize the design for high polarization selectivity and fabrication tolerance. The new approach to VCSEL polarization control may be useful in a number of defense and commercial applications, including chip-scale atomic clocks and other low-power atomic sensors.
The key need that this project has addressed is a short-wave infrared light detector for ranging (LIDAR) imaging at temperatures greater than 100K, as desired by nonproliferation and work for other customers. Several novel device structures to improve avalanche photodiodes (APDs) were fabricated to achieve the desired APD performance. A primary challenge to achieving high sensitivity APDs at 1550 nm is that the small band-gap materials (e.g., InGaAs or Ge) necessary to detect low-energy photons exhibit higher dark counts and higher multiplication noise compared to materials like silicon. To overcome these historical problems APDs were designed and fabricated using separate absorption and multiplication (SAM) regions. The absorption regions used (InGaAs or Ge) to leverage these materials 1550 nm sensitivity. Geiger mode detection was chosen to circumvent gain noise issues in the III-V and Ge multiplication regions, while a novel Ge/Si device was built to examine the utility of transferring photoelectrons in a silicon multiplication region. Silicon is known to have very good analog and GM multiplication properties. The proposed devices represented a high-risk for high-reward approach. Therefore one primary goal of this work was to experimentally resolve uncertainty about the novel APD structures. This work specifically examined three different designs. An InGaAs/InAlAs Geiger mode (GM) structure was proposed for the superior multiplication properties of the InAlAs. The hypothesis to be tested in this structure was whether InAlAs really presented an advantage in GM. A Ge/Si SAM was proposed representing the best possible multiplication material (i.e., silicon), however, significant uncertainty existed about both the Ge material quality and the ability to transfer photoelectrons across the Ge/Si interface. Finally a third pure germanium GM structure was proposed because bulk germanium has been reported to have better dark count properties. However, significant uncertainty existed about the quantum efficiency at 1550 nm the necessary operating temperature. This project has resulted in several conclusions after fabrication and measurement of the proposed structures. We have successfully demonstrated the Ge/Si proof-of-concept in producing high analog gain in a silicon region while absorbing in a Ge region. This has included significant Ge processing infrastructure development at Sandia. However, sensitivity is limited at low temperatures due to high dark currents that we ascribe to tunneling. This leaves remaining uncertainty about whether this structure can achieve the desired performance with further development. GM detection in InGaAs/InAlAs, Ge/Si, Si and pure Ge devices fabricated at Sandia was shown to overcome gain noise challenges, which represents critical learning that will enable Sandia to respond to future single photon detection needs. However, challenges to the operation of these devices in GM remain. The InAlAs multiplication region was not found to be significantly superior to current InP regions for GM, however, improved multiplication region design of InGaAs/InP APDs has been highlighted. For Ge GM detectors it still remains unclear whether an optimal trade-off of parameters can achieve the necessary sensitivity at 1550 nm. To further examine these remaining questions, as well as other application spaces for these technologies, funding for an Intelligence Community post-doc was awarded this year.
Optical lime-domain reflectometry (OTDR) is an effeclive technique for locating faults in fiber communication links. The fact that most OTDR measurements are performed manually is a significant drawback, because it makes them too costly for use in many short-distance networks and too slow for use in military avionic platforms. Here we describe and demonstrate an automated, low-cost, real-time approach to fault monitoring that can be achieved by integrating OTDR functionality directly into VCSEL-based transceivers. This built-in test capability is straightforward to implement and relevant to both multimode and single mode networks. In-situ OTDR uses the transmitter VCSEL already present in data transceivers. Fault monitoring is performed by emitting a brief optical pulse into the fiber and then turning the VCSEL off. If a fault exists, a portion of the optical pulse returns to the transceiver after a time equal to the round-trip delay through the fiber. In multimode OTDR, the signal is detected by an integrated photodetector, while in single mode OTDR the VCSEL itself can be used as a detector. Modified driver electronics perform the measurement and analysis. We demonstrate that VCSEL-based OTDR has sufficient sensitivity to determine the location of most faults commonly seen in short-haul networks (i.e., the Fresnel reflections from improperly terminated fibers and scattering from raggedly-broken fibers). Results are described for single mode and multimode experiments, at both 850 nm and 1.3 μm. We discuss the resolution and sensitivity that have been achieved, as well as expected limitations for this novel approach to network monitoring.
This report describes the research accomplishments achieved under the LDRD Project ''Leaky-mode VCSELs for photonic logic circuits''. Leaky-mode vertical-cavity surface-emitting lasers (VCSELs) offer new possibilities for integration of microcavity lasers to create optical microsystems. A leaky-mode VCSEL output-couples light laterally, in the plane of the semiconductor wafer, which allows the light to interact with adjacent lasers, modulators, and detectors on the same wafer. The fabrication of leaky-mode VCSELs based on effective index modification was proposed and demonstrated at Sandia in 1999 but was not adequately developed for use in applications. The aim of this LDRD has been to advance the design and fabrication of leaky-mode VCSELs to the point where initial applications can be attempted. In the first and second years of this LDRD we concentrated on overcoming previous difficulties in the epitaxial growth and fabrication of these advanced VCSELs. In the third year, we focused on applications of leaky-mode VCSELs, such as all-optical processing circuits based on gain quenching.
Vertical-external-cavity surface-emitting lasers (VECSELs) combine high optical power and good beam quality in a device with surface-normal output. In this paper, we describe the design and operating characteristics of an electrically-pumped VECSEL that employs a wafer-scale fabrication process and operates at 850 nm. A curved micromirror output coupler is heterogeneously integrated with AlGaAs-based semiconductor material to form a compact and robust device. The structure relies on flip-chip bonding the processed epitaxial material to an aluminum nitride mount; this heatsink both dissipates thermal energy and permits high frequency modulation using coplanar traces that lead to the VECSEL mesa. Backside emission is employed, and laser operation at 850 nm is made possible by removing the entire GaAs substrate through selective wet etching. While substrate removal eliminates absorptive losses, it simultaneously compromises laser performance by increasing series resistance and degrading the spatial uniformity of current injection. Several aspects of the VECSEL design help to mitigate these issues, including the use of a novel current-spreading n-type distributed Bragg reflector (DBR). Additionally, VECSEL performance is improved through the use of a p-type DBR that is modified for low thermal resistance.
This paper describes the photonic component development, which exploits pioneering work and unique expertise at Sandia National Laboratories, ARDEC and the Army Research Laboratory by combining key optoelectronic technologies to design and demonstrate components for this fuzing application. The technologies under investigation for the optical fuze design covered in this paper are vertical cavity surface emitting lasers (VECSELs), integrated resonant cavity photodetectors (RCPD), and diffractive micro-optics. The culmination of this work will be low cost, robust, fully integrated, g-hardened components designed suitable for proximity fuzing applications. The use of advanced photonic components will enable replacement of costly assemblies that employ discrete lasers, photodetectors, and bulk optics. The integrated devices will be mass produced and impart huge savings for a variety of Army applications.
A new approach to optical time-domain reflectometry (OTDR) is proposed that will enable distributed fault monitoring in singlemode VCSEL-based networks. In situ OTDR uses the transmitter VCSEL already resident in data transceivers as both emitter and resonant-cavity photodiode for fault location measurements. Also valuable at longer wavelengths, the concept is demonstrated here using an 850 nm oxide-confined VCSEL and simple electronics. The dead times and sensitivity obtained are adequate to detect the majority of faults anticipated in local- and metropolitan-area networks.
This report describes the research accomplishments achieved under the LDRD Project 'Radiation Hardened Optoelectronic Components for Space-Based Applications.' The aim of this LDRD has been to investigate the radiation hardness of vertical-cavity surface-emitting lasers (VCSELs) and photodiodes by looking at both the effects of total dose and of single-event upsets on the electrical and optical characteristics of VCSELs and photodiodes. These investigations were intended to provide guidance for the eventual integration of radiation hardened VCSELs and photodiodes with rad-hard driver and receiver electronics from an external vendor for space applications. During this one-year project, we have fabricated GaAs-based VCSELs and photodiodes, investigated ionization-induced transient effects due to high-energy protons, and measured the degradation of performance from both high-energy protons and neutrons.
The wavelength splitting between the LP01 and LP11 modes of selectively oxidized, ion implanted, and hybrid ion implanted/selectively oxidized vertical-cavity surface-emitting lasers is studied by experiment and theory. Measured splittings at threshold show marked differences between the different laser structures due to the effects of index guiding and thermal lensing. Theoretical results were obtained using a vector optical mode solver and show good agreement with experimental results. The hybrid lasers exhibited behavior intermediate between the ion implanted and selectively oxidized lasers and could be optimized for high power single transverse mode emission.
This report describes the research accomplishments achieved under the LDRD Project ''High-Bandwidth Optical Data Interconnects for Satellite Applications.'' The goal of this LDRD has been to address the future needs of focal-plane-array (FPA) sensors by exploring the use of high-bandwidth fiber-optic interconnects to transmit FPA signals within a satellite. We have focused primarily on vertical-cavity surface-emitting laser (VCSEL) based transmitters, due to the previously demonstrated immunity of VCSELs to total radiation doses up to 1 Mrad. In addition, VCSELs offer high modulation bandwidth (roughly 10 GHz), low power consumption (roughly 5 mW), and high coupling efficiency (greater than -3dB) to optical fibers. In the first year of this LDRD, we concentrated on the task of transmitting analog signals from a cryogenic FPA to a remote analog-to-digital converter. In the second year, we considered the transmission of digital signals produced by the analog-to-digital converter to a remote computer on the satellite. Specifically, we considered the situation in which the FPA, analog-to-digital converter, and VCSEL-based transmitter were all cooled to cryogenic temperatures. This situation requires VCSELs that operate at cryogenic temperature, dissipate minimal heat, and meet the electrical drive requirements in terms of voltage, current, and bandwidth.
Selectively oxidized vertical cavity surface emitting lasers (VCSEL) typically operate in multiple transverse optical modes. High power VCSEL operation is desirable for many applications such as optical storage and printing, modulation spectroscopy, bar code scanning, and data communication over single mode optical fiber. The modal discrimination can be augmented by creating a central region of gain surrounded by a region of optical loss. A VCSEL fabricated via hybrid ion implantation and selective oxidation device structure is designed to demonstrate a single mode output of more than 5 mW for 850 nm.
A record high fundamental-mode power of 5.1 mW was achieved from coupled-resonator vertical-cavity lasers (CRVCLs). In conventional VCSELs, the extent to which the gain volume may be increased is limited by the onset of multi-mode operation. Results indicate that this limitation is circumvented in a coupled-resonator device allowing high power fundamental-mode operation.
The continuous wave operation of 1.3 μm vertical cavity surface emitting laser (VCSEL) grown on GaAs substrates is achieved up to 55 °C, as motivated by demands of emerging VCSEL network applications. These VCSELs employ the mature AlGaAs/GaAs distributed Bragg reflector mirror technology, including selective oxidation for efficient cavity designs. By incorporating a tunnel junction near the optical cavity, both mirrors are doped n-type, which provides the benefits of low optical loss.
The authors report a monolithic coupled-resonator vertical-cavity laser with an ion-implanted top cavity and a selectively oxidized bottom cavity which exhibits bistable behavior in the light output versus injection current. Large bistability regions over current ranges as wide as 18 mA have been observed with on/off contrast ratios of greater than 20 dB. The position and width of the bistability region can be varied by changing the bias to the top cavity. Switching between on and off states can be accomplished with changes as small as 250 {micro}W to the electrical power applied to the top cavity. Theoretical analysis suggests that the bistable behavior is the response of the nonlinear susceptibility in the top cavity to the changes in the bottom intracavity laser intensity as the bottom cavity reaches the thermal rollover point.
Selectively oxidized vertical cavity lasers emitting at 1294 nm using InGaAsN quantum wells are reported for the first time which operate continuous wave at and above room temperature. The lasers employ two n-type Al{sub 0.94}Ga{sub 0.06}As/GaAs distributed Bragg reflectors each with a selectively oxidized current aperture adjacent to the optical cavity, and the top output mirror contains a tunnel junction to inject holes into the active region. Continuous wave single mode lasing is observed up to 55 C. These lasers exhibit the longest wavelength reported to date for vertical cavity surface emitting lasers grown on GaAs substrates.
The authors have developed electrically-injected coupled-resonator vertical-cavity lasers and have studied their novel properties. These monolithically grown coupled-cavity structures have been fabricated with either one active and one passive cavity or with two active cavities. All devices use a selectively oxidized current aperture in the lower cavity, while a proton implant was used in the active-active structures to confine current in the top active cavity. They have demonstrated optical modulation from active-passive devices where the modulation arises from dynamic changes in the coupling between the active and passive cavities. The laser intensity can be modulated by either forward or reverse biasing the passive cavity. They have also observed Q-switched pulses from active-passive devices with pulses as short as 150 ps. A rate equation approach is used to model the Q-switched operation yielding good agreement between the experimental and theoretical pulseshape. They have designed and demonstrated the operation of active-active devices which la.se simultaneously at both longitudinal cavity resonances. Extremely large bistable regions have also been observed in the light-current curves for active-active coupled resonator devices. This bistability can be used for high contrast switching with contrast ratios as high as 100:1. Coupled-resonator vertical-cavity lasers have shown enhanced mode selectivity which has allowed devices to lase with fundamental-mode output powers as high as 5.2 mW.
Previously, an effective index optical model was introduced for the analysis of lateral waveguiding effects in vertical-cavity surface-emitting lasers. The authors show that the resultant transverse equation is almost identical to the one typically obtained in the analysis of dielectric waveguide problems, such as a step-index optical fiber. The solution to the transverse equation yields the lateral dependence of the optical field and, as is recognized in this paper, the discrete frequencies of the microcavity modes. As an example, they apply this technique to the analysis of vertical-cavity lasers that contain thin-oxide apertures. The model intuitively explains the experimental data and makes quantitative predictions in good agreement with a highly accurate numerical model.
The authors report Q-switched operation from an electrically-injected monolithic coupled-resonator structure which consists of an active cavity with InGaAs quantum wells optically coupled to a passive cavity. The passive cavity contains a bulk GaAs region which is reverse-biased to provide variable absorption at the lasing wavelength of 990 nm. Cavity coupling is utilized to effect large changes in output intensity with only very small changes in passive cavity absorption. The device is shown to produce pulses as short as 150 ps at repetition rates as high 4 GHz. A rate equation approach is used to model the Q-switched operation yielding good agreement between the experimental and theoretical pulse shape. Small-signal frequency response measurements also show a transition from a slower ({approximately} 300 MHZ) forward-biased modulation regime to a faster ({approximately} 2 GHz) modulation regime under reverse-bias operation.
The impressive performance improvements of laterally oxidized VCSELs come at the expense of increased fabrication complexity for 2-dimensional arrays. Since the epitaxial layers to be wet-thermally oxidized must be exposed, non-planarity can be an issue. This is particularly important in that electrical contact to both the anode and cathode of the diode must be brought out to a package. We have investigated four fabrication sequences suitable for the fabrication of 2-dimensional VCSEL arrays. These techniques include: mesa etched polymer planarized, mesa etched bridge contacted, mesa etched oxide isolated (where the electrical trace is isolated from the substrate during the oxidation) and oxide/implant isolation (oxidation through small via holes) all of which result in VCSELs with outstanding performance. The suitability of these processes for manufacturing are assessed relative to oxidation uniformity, device capacitance, and structural ruggedness for packaging.
Vertical cavity surface emitting lasers (VCSELs) which operate in multiple transverse optical modes have been rapidly adopted into present data communication applications which rely on multi-mode optical fiber. However, operation only in the fundamental mode is required for free space interconnects and numerous other emerging VCSEL applications. Two device design strategies for obtaining single mode lasing in VCSELs based on mode selective loss or mode selective gain are reviewed and compared. Mode discrimination is attained with the use of a thick tapered oxide aperture positioned at a longitudinal field null. Mode selective gain is achieved by defining a gain aperture within the VCSEL active region to preferentially support the fundamental mode. VCSELs which exhibit greater than 3 mW of single mode output power at 850 nm with mode suppression ratio greater than 30 dB are reported.