Invited Paper 488nm coherent emission by intracavity frequency doubling of extended cavity surface-emitting diode lasers A. V. Shchegrov, D. Lee, J. P. Watson, A. Umbrasas, E. M. Strzelecka, M. K. Liebman, C. A. Amsden, A. Lewis, V. V. Doan, B. D. Moran, J. G. McInerney, A. Mooradian Novalux, Inc., 117 Sonora Court, Sunnyvale, CA 9486, USA ABSTRACT We describe a novel blue-green laser platform, based on the intracavity frequency doubling of Novalux Extended Cavity Surface Emitting Lasers. We have demonstrated 5 to 4mW of single-ended, 488nm, single-longitudinal mode emission with beam quality M 2 <1.2. The optical quality of these lasers matches that of gas lasers; their compactness and efficiency exceed ion, DPSS, and OPSL platforms. These unique properties are designed to serve diverse instrumentation markets such as bio-medical, semiconductor inspection, reprographics, imaging, etc., and to enable new applications. We also present data on the reliability of this novel laser platform and its extensions to different wavelengths (in particular, 46nm and 532nm) and to next-generation, highly compact, monolithic intracavity-doubled lasers. Keywords: blue lasers, visible lasers, diode lasers, frequency doubling, surface-emitting lasers, single mode 1. INTRODUCTION Visible cw lasers in the blue-green frequency range have been attracting increasing interest in recent years due to their applications in many diverse industries: bio-medical instrumentation, semiconductor inspection and alignment, digital imaging and reprographics, optical recording and storage, displays, confocal microscopy, etc. The 488nm wavelength has a special role in these applications. First, many physical phenomena such as laser-induced fluorescence employed in the bio-medical analytical instrument industry have peak efficiency at or near this particular wavelength. Second, 488nm is one of the primary emitting wavelengths of the Argon-ion laser. The high-quality technical characteristics of Argonion lasers include a Gaussian beam, stable power, and stable beam pointing. These propertied offered enough incentive to design the 488nm wavelength and Argon-ion lasers into thousands of instruments and experiments. The standards set by the Argon-ion laser can be summarized briefly as follows 5-15mW of power in the majority of applications, and >15mW in some other cases High beam quality, i.e. TEM spatial profile with M 2 <1.2 Low noise Pointing stability Long-term power stability Low cost. However, the current trends in instrumentation development call for the same or improved technical performance in a more compact, reliable, efficient, and flexible product without any significant cost increase. A more complete set of requirements reflecting these demands would include all of the items summarized above in and the following Low power consumption (high efficiency) Compactness High reliability Narrow spectrum (single emission wavelength) Custom-designed wavelength (in the blue-green range) A laser meeting all the requirements summarized in these two lists would be an ideal fit for existing applications employing 488nm lasers and could enable many new applications. However, no blue-green laser platform could fully satisfy all of these demands until now. Argon-ion lasers, which have dominated the 488nm markets for several decades, Vertical-Cavity Surface-Emitting Lasers VII, Chun Lei, Sean P. Kilcoyne, Editors, Proceedings of SPIE Vol. 4994 (23) 23 SPIE 227-786X/3/$15. 197
cannot meet these demands because of their very high power consumption (~15W), large size (~12 cu. inches), multiple plasma spectral lines requiring filtering out unwanted wavelengths, and inferior reliability compared to that of diode lasers. A diode laser solution is desirable to meet the compactness, reliability, custom wavelength, and low cost demands. However, no commercially viable semiconductor gain material exists in the blue-green frequency range. Many of the items summarized above have been recently addressed by the introduction of frequency-doubled diodepumped solid state laser (DPSS) [1] and optically-pumped semiconductor laser (OPSL) [2] platforms. In both platforms, a high-power edge-emitting diode laser at the wavelength ~81nm is used to pump a gain medium (solid state in the DPSS platform and semicondutor in the OPSL platform) to generate another infrared wavelength, typically 976 or 164nm. The resulting IR beam is frequency doubled to a visible wavelength (blue or green). These laser platforms were designed to meet beam quality, narrow spectrum, low noise, and other specifications. The OPSL platform also allows designing the wavelength of semiconductor gain material. The power consumption and dimensions were reduced to the levels ~6W and 5 cu. inches, respectively, which are orders of magnitude lower than the numbers for typical Argonion lasers. However, the complexity of DPSS and OPSL platforms, requiring circularization of the 81nm pump beam, managing 3 different wavelengths and 2 pump lasers in a single product, and stabilization of the spectral mode in such a complex structure makes these lasers inherently more costly to build than gas lasers. In this paper, we introduce a new type of cw, 488nm visible laser, called Protera, based on the intracavity frequency doubling of the electrically pumped Novalux Extended Cavity Surface-Emitting Laser (NECSEL ). We believe that this platform addresses both sets of demands listed above, and its simple and efficient design can enable many existing and new applications for high-quality visible lasers. In the following sections we discuss the design and technical performance data for Protera lasers. We also demonstrate the flexibility of this laser platform by producing 46nm and 532nm wavelengths in the same laser design and by rescaling the laser dimensions to smaller or larger levels to meet particular design goals. External mirror n-dbr QWs p-dbr Thermal lens GaAs Substrate w w Cavity mode w BeO sub-mount Figure 1. NECSEL structure. Parameters w 1, w 2, and w describe the beam size at various points in the cavity the chip, the OC, and the waist, respectively. 2. NECSEL PLATFORM The design, performance, and manufacturing of NECSELs are described in other papers published by Novalux [3,4]. Here we briefly summarize some characteristics pertinent to the design of a visible laser by intracavity frequency doubling. The Novalux Extended Cavity Surface-Emitting Laser (NECSEL) is an electrically pumped diode laser that has a three-mirror, coupled cavity design as shown in Fig. 1. The substrate-emitting die includes a top-side GaAs/AlGaAs epitaxial high reflector which is doped p-type when the usual n-type GaAs substrates are used. The critical design parameters for the NECSEL cavity are the gain region diameter, the n-dbr and output coupler (OC) reflectivities, the OC radius of curvature, and the external cavity length. The power can be scaled up (down) by increasing (decreasing) the gain aperture diameter. The balance between n-dbr and OC reflectivities (typically, in the range 6% to 9% for one of these mirrors and 9% to 6% for the other) is worked out to minimize external cavity losses (primarily, in the substrate) and to maximize the beam power and quality. The OC radius of curvature and cavity length are designed to ensure the maximum overlap of the NECSEL eigenmode diameter 2w 1 at the chip with the gain diameter and to ensure best manufacturing tolerances to mechanical misalignment and to thermal lens variations. When designed for a maximum outcoupled TEM power, the NECSEL normally lases in a single longitudinal spectral mode, which is the highest-gain spectral eigenmode of the external cavity. 198 Proc. of SPIE Vol. 4994
We believe that the NECSEL platform combines the major advantages of semiconductor lasers and solid-state lasers. Some of the advantages NECSEL shares with the world of semiconductor lasers are High, scalable power levels (we demonstrated up to 1W multimode, and up to.5w TEM operation) High reliability (driven by Telcordia requirements) Custom IR wavelength - with the nominal wavelength of 976nm, we demonstrated equally reliable and efficient NECSEL devices in the range of 92nm to 164nm High efficiency we have consistently demonstrated NECSELs with ~2% wall-plug efficiency in TEM operation High manufacturability surface-emitting geometry allows wafer-level screening Some advantages common with solid-state lasers are Flexible cavity design and ability to control optical performance (mode quality, polarization, frequency, etc.) by introducing intracavity elements High beam quality we have measured beam quality M 2 <1.1 when diode current was varied from lasing power threshold to rollover Narrow spectral line controlled by the coupled-cavity design Needless to say, the full utilization of these advantages requires performing the sophisticated optimization task of optical, electrical, semiconductor, mechanical, and thermal designs of the NECSEL. This has been done by the multidisciplinary team of Novalux engineers and scientists as described in Refs.3 and 4. The set of NECSEL design features we have just summarized makes this platform well suited for intracavity frequency doubling to visible wavelengths. We will address the design optimization and trade-offs for visible light generation in the next section. 3. INTRACAVITY FREQUENCY-DOUBLING OF NECSEL: 488nm PROTERA LASER To achieve high nonlinear conversion and obtain a high-spatial-quality, stable optical beam via the second harmonic generation process with power levels on the order of ~1mW, it is necessary to satisfy several requirements for the fundamental beam [5,6]: high (intracavity) power, ~Watts narrow spectral line - typically, under.1nm high-quality TEM beam, M2<1.2 Conventional edge-emitting diode lasers can provide high power but lack narrow spectral linewidth and spatial beam quality. For this reason, one has to design a 2-step conversion process used in DPSS and OPSL platforms, e.g. conversion from 81nm to 976nm and from 976nm to 488nm in the OPSL platform. Another type of diode laser, the vertical-cavity surface-emitting lasers (VCSELs), are capable of producing high-quality, circular Gaussian beams but they do not reach power levels beyond a few milliwatts. The surface-emitting laser with an external cavity (NECSEL) can meet all three demands and use the space provided by the external cavity for a nonlinear crystal. Compared to the high-power 976nm NECSEL, we have to take several extra steps in the design of a frequency-doubled laser. The partially reflective output mirror maximizing the outcoupled IR power is replaced by a mirror with high reflectivity at 976nm and high transmission at 488nm. This allows us to maximize the circulating power at the fundamental wavelength and to minimize losses in the generated 488nm beam. Since the nonlinear frequency-doubling process is always polarization sensitive, we have to lock the polarization in the fundamental IR beam. Depending on the required discrimination level between the two polarizations, we can lock polarization either on the chip level (using techniques such as those employed with VCSELs [7]), or in the external cavity (e.g., with a Brewster plate). Another critical step is to design the wavelength stabilization mechanism. Using a high-reflectivity mirror at 976nm can allow lasing in more than one longitudinal mode. Regardless of the phase-matching spectral bandwidth of the selected nonlinear material, unstable longitudinal mode operation (mode hopping) will result in a noisy, unstable output, similar to the noisy mode hopping regime of DPSS green lasers [8]. Therefore, we made it a critical requirement to design Protera as a single-longitudinal mode laser. Figure 2 illustrates how this can be achieved. First, some wavelength stabilization is already provided by the coupled-cavity design (Fig.1). This is shown in Fig.2 which illustrates the spectral (longitudinal) eigenmodes of the compound NECSEL cavity. The external cavity defines the spacing of modes shown in the solid line while the internal cavity provides modulation shown in dotted line. By changing the parameters Proc. of SPIE Vol. 4994 199
of both cavities, one can achieve a regime when only a single longitudinal cavity mode can lase. Alternatively, one can again use the space in the external cavity to insert another optical filter (with transmission curve illustrated by the dashed line in Fig.2) to lock a single longitudinal mode.. 1.9.8.7.6.5.4.3.2 NECSEL Longitudinal Modes.1 Short (internal) Cavity Eigenmode Intracavity Filter Transmission 974 974.5 975 975.5 976 976.5 977 977.5 978 Wavelength, nm Figure 2. Illustration of the longitudinal mode spectrum in the composite NECSEL cavity and the wavelength stabilization mechanism Next, we optimize the nonlinear conversion process. A number of nonlinear materials, including periodically-poled materials such as PPLN, PPLT, and single-crystal materials such as KNbO 3 and LBO, are available for conversion from 976nm to 488nm. Most of these materials have been evaluated at Novalux and found to be usable for obtaining 5-4mW of 488nm blue light. Matching our theoretical design with experimental data, we found that the range of fundamental beam waist diameters (2w in Fig.1) of 5 m to 2 m has to be described by the theories developed for the nonlinear conversion of Gaussian beams [5,6] and not simplified plane-wave models with the well-known sinc 2 dependence of the second-harmonic power on the phase mismatch. In general, the power in the second-harmonic beam can be expressed in terms of the power P in the fundamental beam, the material-dependent nonlinear coefficient w P 2 K and a function of the fundamental beam waist radius, waist position in the crystal, walk-off angle (in the case of non-critical phase matching), and the crystal length L : 2 P (, 2 P K F w f,, L) The function F has to be computed numerically and has different forms for type-i and type-ii phase matching. This model proves to be consistent with experiment and helps us select the nonlinear crystal, its length, the fundamental beam size, power, and to optimize the second harmonic power and mechanical and temperature tolerances. After we have defined the design space for all the critical parameters (IR power, cavity stability and tolerances, nonlinear conversion, wavelength stabilization, polarization stabilization), we have to look at them as a whole and make necessary adjustments to ensure that they can co-exist in a real laser. For example, a nonlinear conversion optimization step can call for a smallest possible waist diameter, which is difficult to achieve in a real cavity. The approach we use is to start with a set of specifications e.g. we are currently targeting two separate power specifications of 5mW and 15mW, and to design the laser around these specifications. The result is shown in Fig.3, illustrating the look of both 5mW and 15mW Protera lasers. Both 5mW and 15mW Proteras share the same small package size ~2 cu. inches, and typically consume ~5W of power. f 2 Proc. of SPIE Vol. 4994
Figure 3. Protera laser shown without a housing cover. The block on the right side in the housing is the laser cavity, the block on the left side is the collimation optics, IR filter, and photodetector. The screw holes on the plate under the housing are separated by 1 inch and illustrate the scale of the picture. 4. TECHNICAL CHARACTERISTICS AND RELIABILITY OF 488nm PROTERA We have measured spectra of multiple Protera units. A typical result is shown in Fig.4, illustrating the spectral purity of Protera with side mode suppression of 4dB. The linewidth is below the resolution of.1nm of the optical spectrum analyzers we used. Like in all diode lasers, the Protera power output can be controlled by changing the drive current. Figure 5 illustrates a light-power (LI) curve of a 488nm Protera. This type of behavior allows us to control the stable power output by adjusting the drive current along the slope of the LI curve. Another advantage one can utilize in this platform is highspeed current modulation, not available in traditional 488nm Argon-ion platforms. We also make certain that the laser operates in the same stable, single-longitudinal mode over a range of ambient temperatures, and that it returns to the same power and wavelength after several turn off / turn on cycles simulating the laser operation in a real instrument. Figure 6 shows the performance of a 5mW Protera laser on a hot plate. The power level is stabilized by means of a conventional feedback loop using the integrated photodetector and modulating the diode current. In the first two time intervals ~2.2hrs each the plate is temperature cycled in the range of 2 o C to 45 o C, and in the third interval (only the first 6hrs of which are shown) the laser is operating at a normal room temperature. Two turnoff / turn-on tests are performed in the time interval shown. The hot plate temperature in non-operating windows was changed from 45 o C to 65 o C and back to 45 o C. The total power modulation in the operating windows is under 1% and the laser stays in the same single longitudinal mode, resulting in an extremely low intensity noise shown on the right axis, exceeding our specification of <.2% by orders of magnitude. As can be seen from Fig.6, the laser is stable with respect to temperature cycling, turn-on / turn-off tests, and normal operation. Proc. of SPIE Vol. 4994 21
1.E+ 1.E-1 Optical Spectrum [db] 1.E-2 1.E-3 1.E-4 1.E-5 1.E-6 ~ 4 db 1.E-7 1.E-8 482 483 484 485 486 487 488 489 49 491 492 Wavelength [nm] Figure 4. Spectrum of a 5mW Protera laser. 9 Blue laser output power (mw) 8 7 6 5 4 3 2 1 53 54 55 56 57 58 59 6 61 Drive Current (ma) Figure 5. Light-current (LI) curve of a 488nm Protera. 22 Proc. of SPIE Vol. 4994
1.2.25 1 turned off turned off.2 Normalized Power.8.6.4.2 Power (6mW) Noise.15.1.5 Noise, % (per cent) 1 2 3 4 5 6 7 8 9 1 11 12 Time (hours) Figure 6. Power and noise of a 488nm Protera laser on a hot plate. The plate temperature in first two windows of operation is cycled 2 o C to 45 o C, in the third window of operation it is not controlled (normal room temperature). A similar temperature-cycling setup was used to analyze the far-field pointing stability of 5mW and 15mW Proteras. The data we collected to date shows pointing changes under 5 rad per 1 o C temperature change with the beam pointing returning to the original point after the temperature returned to its original point. This exceeded the performance of the Argon-ion laser we tested. The beam quality tests are performed with an M 2 meter and have consistently showed M 2 <1.2. As mentioned in the previous sections, the key element in obtaining the high-quality 488nm beam is a high-quality fundamental, 976nm beam, which is controlled by the external cavity of the NECSEL. The reliability of NECSEL die, driven by the stringent Telcordia qualification, exceeds the requirements in the visible laser market by at least an order of magnitude. The interested reader is referred to Ref.4 for a more complete summary of accelerated life tests on the NECSEL die. A more challenging task is to define meaningful accelerated life tests for nonlinear materials and fully assembled lasers. The approach we are taking now is performing the life test on a population of Protera lasers, simulating their normal operating conditions, with a goal of achieving 2k failure-free hours per device. To date, we have accumulated >25k hours on the entire population with the oldest device running for over 5k hours without degradation. Other tests helping us to understand possible failure modes of Proteras within and beyond the specified range of operating conditions include temperature cycling, damp heat, shock and vibration, etc. 5. NECSEL PLATFORM FOR VISIBLE LASERS: BEYOND 488nm PROTERA In Ref.4 we describe the capabilities of the NECSEL platform in the near-infrared wavelength range. Here we briefly discuss the advantages of the frequency-doubled NECSEL platform in the visible wavelength range. By adjusting the composition of the semiconductor gain material, we have demonstrated efficient and reliable NECSEL devices at wavelengths from 92nm to 164nm. These dies were frequency doubled in the Protera configuration, with a Proc. of SPIE Vol. 4994 23
5mW single-longitudinal, single-transverse mode output at 46nm and 532nm, respectively. These two wavelengths, like 488nm, are commonly used in many applications of visible lasers. Other wavelengths can also be realized using the same architecture and die material system. Another demonstration of NECSEL capabilities is a highly compact, monolithic version of a frequency-doubled laser with the entire external cavity filled with a nonlinear material. Figure 7 shows the first laboratory demonstration built at Novalux. The 488nm beam has the spatial quality factor M 2 <1.5 and 2.3mW of power. These early units show the significant promise of the NECSEL platform for a variety of visible laser applications with the laser designed for a particular set of requirements such as power, compactness, cost, etc. Figure 7. A compact, monolithic, frequency-doubled NECSEL producing 2.3mW of blue light at 488nm 6. CONCLUSIONS In conclusion, we have demonstrated a novel 488nm laser, based on the intracavity frequency doubling of a diode surface-emitting laser (NECSEL). The optical performance of Protera (beam quality, low noise, and pointing stability) matches or exceeds the characteristics of Argon-ion lasers and blue-green DPSS and OPSL lasers. The direct frequency doubling of a diode laser allows simple design which results in the combination of low cost, small size, high reliability, and efficiency that cannot be achieved with other laser platforms. The key design features of the Protera laser are its external cavity that controls a high-quality transverse mode that is used for frequency doubling, and its singlelongitudinal mode design that provides low-noise, high stability operation. We also demonstrated the flexibility and future outlook of the NECSEL/Protera platform for visible lasers by desingning and building 46nm and 532nm lasers and a highly compact, monolithic version of a frequency-doubled NECSEL with the external cavity filled with a nonlinear material. It is our belief that visible lasers based on frequency doubling of NECSEL will serve a variety of existing and new applications requiring high-quality visible lasers in the blue-green wavelength range. 24 Proc. of SPIE Vol. 4994
ACKNOWLEDGEMENTS The authors would like to thank G. Carey, W. Ha, J. Harrington, S. Venkatkrishnan, B. Cantos, and W. Hitchens for their outstanding work on Protera and NECSEL engineering, and also K. Kennedy, R. Martinsen, D. Heald, R. Lujan, and H. Zhou for their contributions at earlier stages of NECSEL development. Many thanks are also due to W. Krupke, R. Waltonsmith, J. Cannon, and I. Jenks for the important role they played in bringing the NECSEL platform into the visible laser markets. REFERENCES 1. W. Krupke, Advanced diode-pumped solid state lasers (DPSSLs): near-term trends and future prospects, Proc. SPIE, vol.3888, p.21 (2). 2. J. L. A. Chilla, A. Caprara, E. Mao, L. Spinelli, W. Seelert, J. Rosperich, and A. Salokatve, Solid-state blue laser technology, IEEE Digest of LEOS Summer Topical Meeting 21, IEEE, p.2 (21). 3. A. Mooradian, High brightness, cavity-controlled, electrically pumped surface emitting GaInAs lasers operating at 98nm, Proc. Int. Conf on Opt. Fiber Commun. (OFC), postdeadline paper (21). 4. J. G. McInerney, A. Mooradian, A. Lewis, A. V. Shchegrov, E. M. Strzelecka, D. Lee, J. P. Watson, K. W. Kennedy, G. P. Carey, H. Zhou, B. D. Cantos, W. R. Hitchens, V. Doan, Novel 98nm light sources using vertical cavity lasers with extended optical cavities, Proc. SPIE, 4994-6 - this volume (23). 5. G. D. Boyd and D. A. Kleinman, Parametric interaction of focused gaussian beams, J. Appl. Phys., vol. 39, p.3597 (1968) 6. J.-J. Zondy, Comparative theory of walkoff-limited type-ii vs type-i second harmonic generation with gaussian beams, Opt. Commun., vol.81, p. 427 (1991). 7. T. Yoshikawa, T. Kawakami, H. Saito, H. Kosaka, M. Kajita, K. Kurihara, Y. Sugimoto, and K. Kasahara, Polarization-controlled single-mode VCSELs, IEEE J. of Quant. Electr., vol. 34, p. 19 (1998). 8. D. W. Anthon, D. L. Sipes, T. J. Pier, and M. R. Ressl, Intracavity doubling of cw diode-pumped Nd:YAG lasers with KTP, IEEE J. of Quant. Electr., vol.28, p.1148 (1992). Proc. of SPIE Vol. 4994 25