Introduction

In this chapter we review our recent work on the dynamics of coherent semiconductor laser arrays. While most of the published literature on laser arrays has focused on the spatial properties of these sources, it is becoming increasingly apparent that their temporal behavior can be marvelously rich and complex. This complexity should not be surprising. A laser is, after all, a nonlinear oscillator. When one creates an array of coupled nonlinear oscillators the resulting dynamical behavior of the system will range from synchronization and phase locking to instabilities and chaos. An understanding of the complex dynamics is essential for the design of stable, compact, high-power sources for applications such as intersatellite optical communications. Because the characteristic time scale of semiconductor laser array dynamics is less than a nanosecond, it took the pioneering streak camera measurements of DeFreez et al., to provide the first indication of temporal instabilities in these lasers. Since then, complex dynamical behavior has been observed in most types of semiconductor laser arrays. Theoretical work has proceeded apace, from the early time-dependent coupled-mode theories to more recent partial differential equation models which treat the array as a single entity. The models predict both coupling induced instabilities and saturable-absorber induced instabilities, depending on array geometry and material parameters.

Introduction

Semiconductor diode lasers emitting normal to the substrate plane, known as surface-emitting lasers, are extremely promising for addressing a range of applications from optical interconnects, optical communications and optical recording to remote sensing. The most promising aspect perhaps lies in the prospect of eliminating low yield laser fabrication steps, i.e. laser packaging processing including wafer lapping, cleaving and dicing, facet coatings and diode bonding. The possibility of being able to make any number of lasers anywhere on a wafer is also an increasingly important factor for applications such as optical interconnects. At present two completely different approaches are aimed at realizing surface-emitting lasers. The first represents an extension of the existing technology for semiconductor edge-emitting lasers that uses a 45° slanted mirror or a second-order grating to vertically couple the light out (Figure 9.1(1). (2)). The second, pioneered by K. Iga in 1979, uses highly reflective mirrors to clad the active region, resulting in a vertical cavity that produces an output beam propagating normal to the substrate surface (Figure 9.1(3)).

The vertical cavity design offers important advantages over other surface-emitting laser designs. The unique topology of a vertical cavity facilitates large-scale processing, on-wafer testing and pre-process screening. The small lateral dimensions allow for fabrication of large 2-D arrays with high packing density and integration with other optical and electronic devices.

The preceding chapters of this book have described diode laser arrays with a variety of geometries and coherence properties. Edge-emitting geometries were shown to be capable of high power and high efficiency, whereas surface-emitting geometries permitted fabrication of large two-dimensional arrays. In addition, many of these arrays were designed to contain various degrees of mutual coherence between lasing apertures. In the current chapter, we consider several of the systems aspects of these laser arrays. In particular, we will consider external optical components and systems that manipulate the laser output light to satisfy the requirements of specific applications. We will describe optical systems that act as interfaces between the laser array and other electrooptical components, as well as external laser cavities that improve the performance of the laser array itself.

The recent advances in microoptical components permit unprecedented control of light from laser arrays. Many optical functions are now possible that were previously difficult or impossible to perform with conventional optical components. Microlenses and microlens arrays can be used to collimate and expand individual lasers in an array, making it possible to change the fill factor of a given array. Astigmatism, spherical aberration, and other optical aberrations, can be removed with aspheric and anamorphic microlenses.

Introduction

In recent years there has been an evolution in interest in semiconductor laser technology from exploring the performance limits of single devices of inherently low power to exploiting the potential of large arrays of devices of much higher power. While the low-power semiconductor devices are a proven component in many commercial systems, it is clear that a number of applications also exist for high-power arrays and that semiconductor-laser researchers are now poised to tackle them.

The manufacture of large volumes of semiconductor laser arrays with high yield has enabled the fabrication of larger and larger arrays. Rather than mounting single devices on copper submounts, and thus limiting the modest amount of steady state waste heat which can be dissipated, the laser designer is now faced with the challenging task of fabricating increasing numbers of devices in increasingly smaller volumes and operating these units at increased duty cycle or average power. We are faced with the same evolution of technology which IC chip manufacturers have faced for years; however, the semiconductor lasers are far more temperature sensitive than silicon-based electronic circuits and the task is ever more challenging.

Depending on the application, the utility of any laser is governed by its ability to deliver photons to a remote location, which in turn depends on the beam quality.

Introduction

Phase-locked arrays of diode lasers have been studied extensively over the last 15 years. Such devices have been pursued in the quest to achieve high coherent powers (> 100 mW diffraction limited) for applications such as space communications, blue-light generation via frequency doubling, optical interconnects, parallel optical-signal processing, high-speed, high-resolution laser printing and end-pumping solid-state lasers. Conventional, narrow-stripe (3–4 μm wide), single-mode lasers provide, at most, 100 mW reliably, as limited by the optical power density at the laser facet. For reliable operation at watt-range power levels, large-aperture (≥100 µm) sources are necessary. Thus, the challenge has been to obtain single-spatial-mode operation from large-aperture devices, and maintain stable, diffraction-limited-beam behavior to high power levels (0.5–1.0 W).

By comparison with other types of high-power coherent sources (master oscillator power amplifier (MOPA), unstable resonators), phase-locked arrays have some unique advantages: graceful degradation; no need for internal or external isolators; no need for external optics to compensate for phasefront aberrations due to thermal- and/or carrier-induced variations in the dielectric constant; and, foremost, beam stability with drive level due to a strong, built-in, real-index profile. The consequence is that, in the long run, phase-locked arrays are bound to be more reliable than either MOPAs or unstable resonators.

Introduction

Diode lasers were first demonstrated in 1962. These structures were broad area homojunction devices which could only be operated at 77 K with short (< 100 ns) pulse widths. From those early lasers, much progress has been made in increasing the output power, reliability, operating temperature and efficiency of diode lasers and diode laser arrays. For example, at present, monolithic laser diode arrays have been operated at temperatures in excess of 100°C at output power levels in excess of 100 W cw and at power conversion efficiencies of greater than 50% with multiwatt output powers. This chapter will describe the history of the improvements made in high-power diode laser research and the failure mechanisms responsible for the limitations in output power of high-power diode lasers. We also present the results of reliability testing at high output power for a variety of semiconductor laser structures and review environmental testing for laser array products. The design considerations for long life operation at high power are also presented.

7.1.1 History of high-power diode laser arrays

Following the early introduction of homojunction diode lasers, single heterostructure and double heterostructure (DH) diode lasers were demonstrated in the late 1960s. The double heterostructure laser was the first laser to exhibit cw room temperature operation This DH laser offered two major improvements over the original homojunction device.

Introduction

Diode laser arrays were conceived and fabricated in the late 1970s as an attempt to overcome the inherent power limitations of single-aperture diode lasers. Increasing the output power by simply increasing the lateral aperture size from the usual ≈ 5 μm to values of 50–100 μm had already been tried with disappointing results. These so-called ‘broad-area’ devices exhibited such poor modal characteristics that the power focusable into a diffraction-limited spot barely increased at all. Arrays were thus seen as a means of increasing output power by phase-locking several diode lasers together so that they operated as a single (hopefully) diffraction-limited source. The last decade has seen a tremendous research effort directed towards the design and fabrication of high-power arrays exhibiting good mode control, with (as usual) mixed results. Today's multi-watt arrays certainly provide more focusable laser power than could be obtained by simply stacking together individual diode lasers. On the other hand, achieving the desired mode control and coherence from arrays has proved considerably more difficult than was originally envisioned. As a result, virtually all the high-power arrays commercially available at present emit their radiation into two broad far-field lobes instead of the desired single diffraction-limited lobe.

This situation has arisen because the first simple types of diode arrays that were easily understandable and relatively easy to fabricate have been shown to exhibit poor mode discrimination, so that even devices that lased in the desired in-phase mode near threshold became multi-mode at higher currents.

Introduction

The oldest, best developed, and still most sophisticated heterostructure materials system for semiconductor diode laser arrays is the AlGaAs—GaAs system. This system has a unique combination of practical physical, electrical, optical and chemical features that has withstood the test of time. These features include a tractable chemical system that lends itself equally well to basic epitaxial growth methods, such as liquid phase epitaxy (LPE), and to more sophisticated thin layer epitaxial growth methods, such as metalorganic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE). Controlled thin layer epitaxy has, in turn, allowed the development of AlGaAs—GaAs quantum well heterostructure lasers. Perhaps the single most important attractive feature of the AlGaAs—GaAs heterostructure system is the fortunate coincidence that AlAs and GaAs have, for all practical purposes, the same lattice constant (Δa/a < 0.12%). This allows the design of any structure, with any combination of layer compositions, without regard to lattice mismatch or the associated dislocation formation.

The range of wavelengths available from lattice matched AlGaAs—GaAs conventional double-heterostructure lasers and quantum well heterostructure lasers is from λ ≈ 0.88–0.65 μm. The long wavelength limit is defined by the band edge for GaAs, and AlGaAs active layers or the quantum size effect, or both, are used to shift the emission to shorter wavelengths.

Ion implantation is a superb method for modifying surface properties of materials since it offers accurate control of dopant composition and structural modification at any selected temperature. In the field of semiconductor technology there was a time lag of some 20 years from the initial development of ion implantation to its becoming a cornerstone of production technology. A similar delay in the acceptance time occurred for metal surface treatments. For insulating crystals and glasses, the use of ion beams to modify such crucial optical parameters as refractive index, reflectivity, colour centre content, and luminescence has now passed this 20-year apprenticeship, and the subject is expanding to include the valuable application phase. Appreciation of possible uses of ion implantation is gaining momentum, in part as a result of the ease with which one can fabricate optical waveguides and waveguide lasers and tailor electro-optic and non-linear properties of the key materials of modern optics. Our own experience with these ion implanted property changes, and potential applications, encompasses a diversity of examples, from lasers to studies of fundamental imperfections in insulators, to fabrication of car rearview mirrors.

Since Sussex has been among the pioneers in the study of work with optical materials, we have written a text which has perhaps presented a disproportionate number of examples using our own data. They do, however, typify many aspects of the subject. The topics cover the basic ion beam interactions in solids, followed by the optical effects of absorption and luminescence. We have then included a chapter on waveguide theory and analysis in order to lead into the very exciting examples of ion implanted lasers, second harmonic generation and nonlinear waveguide optics.

Luminescence processes

Luminescence transitions may occur within localised defect sites, for example to give the characteristic line emissions of rare earth ions, narrow emission bands of chromium in ruby, or they may produce broad bands from charge transfer between defects. Overall, emission bands may vary greatly in width, but nevertheless the luminescence spectra provide a measure of specific defect types, and even offer some quantitative measure of the changes in defect concentrations. Since the excitation energy for luminescence may be provided by many routes, ion implantation is no exception and it frequently produces strong luminescence from insulating targets. This feature is often used as a means of aligning the ion beam and it is common practice to have defining apertures with silica plates to check the ion beam position visually. Such intense luminescence can reveal a number of features relating to the changing defect state of the target. For example, in many of the materials used to form optical waveguides by ion implantation there is a decrease in luminescence intensity which approximately follows the amorphisation in the crystal. Hence, one has a visual estimate of the progress of the amorphisation. Quantitative recording of the wavelength dependence of the signal, in terms of luminescence efficiency and spectral changes, should provide details on not only the defects pre-existing in the material but also the ion beam induced changes. Consequently, a number of research groups have used the luminescence produced during implantation to follow such modifications.

Lattices modified by the ion beam will show changes in their subsequent luminescence performance and the effects of implantation have been recorded in photoluminescence, laser emission, cathodoluminescence and thermoluminescence.

Analysis methods using absorption, ESR and RBS

Optical methods of studying defects have the advantage that if each defect has characteristic energy levels which lie within the forbidden energy gap, then they show separable optical absorption and luminescence bands. Higher photon energy absorption generally monitors electronic transitions, whereas infra-red absorption records vibrational spectra. Many of the optical transitions which result from the presence of impurities have energies in the visible part of the spectrum and consequently the defects are referred to as colour centres. Examples of colour centres are widespread and include the impurities which give colouration to ruby and sapphire or stained glass. They are the basis for photographic and photochromic materials, and frequently involve a mixture of impurity and intrinsic defects. Whilst analysis of absorption bands may determine defect symmetry and inter-relationships of different colour centres, it is unusual to be able to confirm precise models of defect sites solely from the absorption data. In this respect the processes which involve hyperfine interactions such as Electron Spin Resonance (ESR), Electron Nuclear Double Resonance (ENDOR), Mossbauer spectroscopy or spin precession techniques provide more specific answers if they can be applied. In the ion implantation literature there are frequent presentations of data from Rutherford Back-scattering Spectrometry (RBS) to give the depth distributions of impurities or damage in the target material. In part, RBS appears to be popular for implantation analysis because it requires a high energy ion accelerator, which is normally a feature of an implantation laboratory. The information is useful but, like electron microscopy, it rarely gives precise details of individual defect arrangements.

Over the past decade ion implantation has been demonstrated as a suitable technique for the fabrication of waveguide structures in an ever increasing number of optical materials. The stage has now been reached where actual device-oriented structures are being designed and implemented using this technology. This chapter will deal with the progress which is being made in the development of useful devices, and how ion implantation is being used to achieve these ends.

Many authors have recognised the advantages of waveguide structures for signal processing, for coupling to optical fibres or for frequency conversion. Optical circuitry has been proposed which, at least initially, was purely hypothetical, but has advanced to include ideas of entirely solid state waveguide structures (sometimes termed holosteric systems). Forseeable objectives include multi-frequency, or tunable, compact lasers in which the pump power is provided by a semiconductor diode. This inherently waveguide power source could then be matched into other waveguides for a combination of frequency conversion, SHG, pumping a tunable laser waveguide such as alexandrite or Ti:sapphire, or driving an optical parametric oscillator (OPO). The conversion efficiencies in the various stages are often as high as 30%, hence even in a several-stage process a 100 W semiconductor array could result in a few watts of tunable laser power. Such systems could of course supersede the highly inefficient Ar or Kr gas laser sources for many applications. The concepts are clear, and this final chapter will indicate how close the components are to realisation when using ion implantation fabrication routes.

Characteristics of ion implanted waveguides

A waveguide is characterised by a region of high refractive index bounded by regions of lower index. The confinement of the light, as well as the spatial distribution of optical energy inside the guiding layer depends on the refractive index profile. There are several conventional techniques for fabricating optical waveguides. These techniques, including epitaxial growth, metal diffusion and ion exchange, increase the refractive index of the surface layer for a few microns (Figure 5.1), and this high index layer is surrounded by the low index of air and substrate to form an optical waveguide. Ion implantation, as a surface modification technique, can modify the optical properties of an insulator surface. However, when light ions are used, particularly when dealing with crystals, instead of changing the refractive index of the surface layer, a low index optical barrier is built up at the end of the ions' track due to elastic energy deposition from ions to the lattice. Therefore, the surface layer, ideally the same as the substrate, is surrounded by the low index of air and this optical barrier (Figure 5.1(c)). During the ion implantation, some point defects may be produced in the surface layer due to ionisation and excitation when the ions are travelling fast. These simple defects will change the properties of the material, and induce absorption and scattering loss. In practice, it has been found that post-implant annealing at a moderate temperature can either reduce or aggregate these defects depending on the material in question. In many materials, a low loss optical waveguide (∼0.5dB/cm) can be produced by ion implantation and subsequent annealing.

Development of ion implantation

The control of surface properties is of paramount importance for a wide range of materials applications, and craftsmen and technologists of all scientific disciplines have battled with problems of corrosion, surface hardness, friction and electrical and optical behaviour for many hundreds of years. Even for the simplest of articles, whether they be knives, bottles or non-stick frying pans, the manufacture of materials which have the desired surface properties is often incompatible with bulk performance, and so there is an emphasis on finding ways to modify surface layers. Processes such as thermal quenching prove effective for hardening steel and glass bottles but lack the finesse which is required for more sophisticated technology. Instead, these use more controllable treatments, including the deposition of surface coatings or diffusion of impurities into the surface layer and, of course, ion implantation.

Historically, ion implantation has generally been the last of the treatments to receive widespread acceptance. The reason for this is that, compared with coating and diffusion treatments, it appeared to require more complex and expensive equipment which was not readily available. Figure 1.1 indicates that implantation systems may come in several levels of complexity. There are those similar to sophisticated laboratory research machines, which have ion sources, pre-acceleration, mass analysis followed by additional acceleration and then the target region. Commercial applications with requirements of uniformity and a large sample throughput may result in sample handling and beam sweeping equipment as complex and expensive as the accelerator.