High reflectivity, electrothermal and electrostatic MEMS (Micro-Electro-Mechanical Systems) micromirrors were used as a control element within a Nd-doped laser cavity. Stable continuous-wave oscillation of a 3-mirror Nd:YLF laser at a maximum output power of 200mW was limited by thermally-induced surface deformation of the micromirror. An electrostatic micromirror was used to induce Q-switching, resulting in pulse durations of 220ns - 2μs over a repetition frequency range of 6kHz - 40kHz.
© 2011 OSA
Micro-electro-mechanical systems (MEMS) have seen increasing integration in photonic systems and applications [1,2]. The incorporation of MEMS devices offers opportunities in reducing the cost, size, weight and electrical power consumption of such systems. When configured within laser resonators, MEMS devices can introduce new concepts and functionalities: for instance, temporal control of the laser output (ranging from low-speed modulation to the use as a Q-switching element [3–7]), real-time misalignment correction, miniaturisation and active spectral control . In addition, as arrays of MEMS devices can be easily fabricated, 1- and 2-D array lasers with active temporal and spectral control become realistic prospects.
Recently, Fabert et al. demonstrated the use of a cantilever-type MEMS micromirror as an active Q-switching element in a fibre-based laser system obtaining pulse durations as short as 8ns . In their work , the dimensions of the largest MEMS devices were 150μm x 350μm and the micromirror reflectivity was R = 80%. In this paper, we present a control method with the potential to be applied to a wider range of laser systems. The use of a dielectrically-coated MEMS micromirror enables a significant reduction of the intra-cavity loss. In this paper, we describe the temporal control of an Nd:based laser using an intra-cavity micromirror and discuss the limitations imposed by the MEMS devices in the context of these laser systems.
In section 2, a full description of the MEMS devices used in this work will be given. Section 3 then describes the continuous-wave performance of an Nd:YLF laser featuring an electrothermal MEMS micromirror. Section 4 describes the Q-switching of a Nd:YLF laser using an intra-cavity electrostatic micromirror and section 5 provides a discussion.
2. Electrothermal and electrostatic MEMS micromirrors
The development of high optical quality MEMS micromirrors has been the subject of great interest [9–12]. The silicon micromirrors used in this work were fabricated by the silicon-on-insulator foundry process (SOIMUMPs) at MEMSCAP Inc . These 10μm-thick micromirrors  were designed with either one or two-axis actuation enabling precise angular positional control – the mirror dimensions ranged from 0.5mm to 3mm and they can have a circular or square shape. Two families of micromirrors have been developed: electrothermal actuation, and electrostatic actuation devices. The current-driven, electrothermal variant, shown in Fig. 1a , provides a wide range of angular adjustment whereas the resonant voltage-driven, electrostatic micromirror (see Fig. 1b) produces rapid x-y scanning (10s of kHz) but has no appreciable DC response. In this project, a square-shaped, 1mm x 1mm, electrothermal micromirror and a circular-shaped (diameter = 1mm) electrostatic micromirror were utilized. Both types of micromirrors were coated with eight, quarter-wave thick, pairs of SiO2/Nb2O5 layers to provide a reflectivity measured to be >99% at 1047nm enabling their use as low-loss intra-cavity mirrors. Initially, the surface of the coated electrothermal micromirrors had a convex shape with a radius of curvature (ROC) of −8cm, whereas the electrostatic devices featured a concave shape with a ROC of 25cm. The MEMS chip was mounted on a PCB backplane to provide the necessary electrical connections. Each test-chip could feature up to 5 electrothermal micromirrors having various actuation schemes revealing the potential for fabricating densely packed arrays of micromirrors, which could form the basis of a new type of compact laser array.
2.1. Assessment of the electrothermal micromirror
The electrothermal micromirror has a square shape (1mm x 1mm) and provides tip/tilt correction where the value of the deflection angle is dependent on the voltage applied to the actuator .
In order to measure the beam deflection angle resulting from the micromirror tilt, a collimated HeNe laser light (λ = 633nm, beam diameter~1mm) was incident at the centre of the micromirror and the reflected beam was observed on a screen. The beam deflection angle was measured as a function of the voltage applied to the actuator with results shown in Fig. 2 . Significant heat-induced surface distortion was observed when voltage was applied to the actuator. It was evident from the experiments that the deformation was predominantly spherical. Therefore, by measuring the dimensions of the reflected beam incident at the screen, the radius of curvature of the micromirror surface was calculated and the resulting focusing power of the mirror is shown as a function of the applied voltage in Fig. 3 . The thermally-induced increase in focal power was found to be linear up to 100D with a voltage of 15V. Such a deformation will have significant impact on the intra-cavity use of these micromirrors. While the initial curvature must be accounted for when designing the laser cavity, thermal deformation on this order can also prevent laser oscillation, adding another element of output beam control. However, although initially a limitation, this thermal deformation could be exploited providing the micromirror the potential to be used as an intra-cavity adaptive optics element to increase the laser brightness  or reduce the laser warm-up time .
2.2 Assessment of the electrostatic micromirror
The electrostatic micromirror is a resonant device constantly scanning at discrete frequencies. The micromirror used in this experiment has a circular shape with a diameter of 1mm. When a sinusoidal voltage oscillating at a frequency near one of the micromirror resonant frequencies (or its harmonics) is applied to one of its actuators, angular deflection along the main axis of the actuator or along an axis perpendicular to the main axis can be obtained. At all times, a positive DC voltage bias equal to the voltage amplitude is added to ensure that the frequency of the electrostatic force applied to the MEMS mirror is equal to the frequency of the applied electrical signal.
The two main oscillation modes of the micromirror are the torsional mode (T-mode), shown in Fig. 4a , whose resonance frequency was measured to be ~6.5kHz and the flexing mode (F-mode), shown in Fig. 4b, whose resonance frequency was ~10.3kHz. Table 1 lists the most significant modes obtained from the electrostatic micromirror used in this paper (the amplitude of the unipolar voltage signal was set to a maximum of 200V peak-to-peak). The value of the beam angular deflection depended on the value of the amplitude and frequency of the voltage signal applied. Using a 200V peak-to-peak unipolar drive signal, the maximum half angle beam deflection values were measured to be 8.7° and 9.3° for the T-mode at f = 6.5kHz and F-mode at f = 10.3kHz respectively. As expected, unlike in the electrothermal mirror, no significant surface deformation was observed when voltage was applied to the actuators. Using a vibrometer, the frequencies of the resonating oscillations were measured and, in all cases, matched the frequencies of the applied unipolar sinusoidal voltage.
Given that, with the maximum voltage value applied (200V peak-to-peak), the electrostatic MEMS mirror can reach 9.3° of half angle beam deflection at 10.3kHz, the maximum angular speed of the micromirror can be measured by applying a linear fit to the sinusoidal wave corresponding to the oscillation of the mode as seen in Fig. 5 . The resultant angular speed is 3.105 deg/s. Typical commercial acousto-optic Q-switches have rise times of the order of 100ns. Therefore, the corresponding misalignment sensitivity required for the laser to experience a similar rise time using the MEMS micromirror is 0.03deg (or 0.5mrad). In the case of the F-mode observed at 41.6kHz (maximum half angle beam deflection = 7.7°), the angular speed becomes 1.106 deg.s−1 with the resulting required misalignment sensitivity being 0.1deg (or 1.75mrad). In the case of the F-mode operating at 55.1kHz, the required misalignment sensitity is 0.11deg. Such angular misalignment sensitivities (corresponding to a ~1/10 screw turn of a one-inch mirror mount with 100turns per inch resolution) are a common occurrence in laser systems; therefore, the electrostatic micromirror operating at these two frequencies (41.6kHz and 55.1kHz) has the potential to provide a Q-switching as efficient as that provided by acousto-optic Q-switches.
3. Continuous-wave MEMS laser investigation
In this paper, both types of MEMS micromirrors were inserted inside three different solid-state laser cavities operating at the CW regime in order to assess the temporal modulation of the laser output.
3.1 2-mirror Nd:YLF laser cavity
The 100mm long, 2-mirror laser cavity (see Fig. 6 ) was built around a 63mm long, 3mm diameter Nd:YLF rod. An electrothermal micromirror was used as an end mirror along with a T = 20% output coupler with a radius of curvature of 1m. The gain medium was a pump/rod module obtained commercially from Cutting Edge Optronics  and was side-pumped evenly with 3 sets of 3 diodes. The pump power could be modulated reaching a maximum value of 180W. The radius of the laser fundamental transverse mode incident to the micromirror was calculated at ~250μm.
Stable CW laser oscillation was obtained for total side-pump powers of up to ~18W (3W above laser threshold) leading to a maximum output power of 20mW. However, above this pump power value, the laser output turned on and off repetitively in the manner shown in Fig. 7 . Using a 1mW HeNe probe beam (λ = 633nm), an examination of the micromirror surface behavior was performed, using the same technique described in section 2.1, while the Nd:YLF laser oscillated. Evident concave deformation of the micromirror surface was observed. This fluctuation is due to the heat-induced deformation of the surface of the MEMS. The heat deposited by the small proportion of the intra-cavity light not reflected by the coating (a few mW in this case) cannot be dissipated efficiently and is therefore absorbed by the silicon-based MEMS chip. In addition, unquantifiable residual pump light (at λ = 808nm) was incident to the micromirror surface and fully absorbed by the silicon MEMS. A bimorph effect occurred resulting in an inwards bulge of the micromirror surface. With further increase of the intra-cavity laser power, the surface deformation of the MEMS micromirror is so significant (ROC~1.5cm) that the laser cavity becomes optically unstable and therefore results in the total loss of laser oscillation. Consequently, the micromirror surface cools down leading a significant reduction of the heat-induced deformation. The laser cavity becomes optically stable again restoring laser oscillation. This oscillation leads to further heat-induced deformation of the MEMS surface. This phenomenon is repeated several times leading to a continuously interrupted laser oscillation with a period of the order of a few hundreds of ms (see Fig. 7). Figure 8 displays the intensity of the Nd:YLF laser output as a function of the time during which the pump power is linearly increased over time. Initially CW laser oscillation occurs; at higher pump powers (~18W), the laser emission is reduced; the output beam then becomes rapidly modulated – here, the surface distortion due to heating dominates the laser dynamics and limits performance. At pump power values above 24W, the heat-induced deformation mainly fuelled by the pump light absorption, is so severe that laser oscillation cannot be established at all.
At pump power levels resulting in the unstable laser oscillation described above (~20W) with an average output power of ~20mW, the laser output could be modulated by applying a voltage to the micromirror actuator (Fig. 9 ). Although the beam deflection resulting from the electrothermal MEMS is ~4.2° with the maximum voltage applied (15V), only 1V was required (resulting in a beam deflection angle of 0.2°) to misalign the laser cavity. Therefore, temporal modulation of the laser output (with a period superior to 1s) was obtained using an intra-cavity electrothermal micromirror.
3.2 3-mirror Nd:YLF laser cavity
In order to reduce the impact of the thermally-induced surface deformation on the laser cavity stability, the Nd:YLF laser cavity was modified so that the laser mode size incident at the electrothermal micromirror was minimal. To achieve this, the micromirror was inserted at the focus of a 3-mirror laser cavity (Fig. 10a ) where the incident fundamental laser mode was estimated at 80μm radius (compared to a radius of 250μm in the previous laser configuration). Since the curved mirror was anti-reflection coated at λ = 808nm, negligible pump light was incident at the micromirror surface, therefore reducing thermally-induced micromirror surface deformation. Stable CW oscillation with a maximum output power of 200mW was obtained with ~20W of side-pump power. The resulting power density incident to the micromirror was ~5kW/cm2. At higher pump powers, the micromirror experienced significant heat-induced surface distortion rendering the laser cavity unstable. Using this configuration, about 10V (corresponding to approximately a beam deflection half angle of 3°) were required to temporally modulate the laser cavity. In this way, a ten-fold power scaling was reached by reducing the incident laser mode size on the micromirror; however, this was obtained at the expense of the angular misalignment sensitivity of the laser cavity.
3.3 Investigation of the heat-induced surface deformation in a Nd:glass laser cavity
In order to fully assess the physical power handling of MEMS micromirrors, an electrothermal device was inserted in the Nd:glass laser cavity as shown in Fig. 10b. This laser cavity was built around an end-pumped 25mm-long Nd:glass rod. In the middle of the rod, the pump beam had a diameter of 100μm with an optical pump power of 460mW at a wavelength of 808nm. The micromirror was placed at the focus of the laser cavity (radius of the beam on the micromirror ~60μm). In this case, the laser operated in the CW regime and delivered an output power of 40mW giving an incident power density of about 20kW/cm2. This experiment demonstrates that this micromirror has the potential to sustain the intra-cavity power densities encountered in higher power lasers with output powers typically >10W level. However in this case, the maximum deflection available (4.2°) was necessary to misalign the laser cavity.
4. Q-switched MEMS laser investigations
To investigate Q-switching, an electrostatic micromirror was inserted in the Nd:YLF laser cavity described in Fig. 10a effectively replacing the electrothermal micromirror. The size of the fundamental laser mode incident to the micromirror remained at 80μm radius and was initially located at the center of the micromirror. Without any driving voltage signal applied to the micromirror, CW laser oscillation was observed with a maximum output power of 300mW for ~20W of pump power (~3W above laser threshold). Similar to the electrothermal micromirror, the output power was limited by thermal deformation of the reflective surface. However, no thermally-induced variation of the resonance frequency could be observed within the accuracy of the driving signal frequency (+/− 0.01%). Subsequently, a voltage signal consisting of a sinusoidal wave with the maximum amplitude (200Vp-p) and a frequency varying discretely from 6kHz to 41kHz (i.e.: 6, 10, 30 and 41kHz corresponding to the frequencies where significant angular deflections were observed, see section 2.2) was applied to the micromirror. Again, a positive bias equal to the voltage amplitude was added to this electrical signal. As a result, the laser operated in the Q-switch regime with a pulse repetition frequency equal to the double of the applied voltage signal frequency to the micromirror. The pulse duration ranged from 430ns (Fig. 11a ) to 2μs with average powers ranging from 10 to 60mW. The micromirror was then manually tilted so that the laser oscillation angular range was pushed towards the edge of the micromirror angular range. As seen in Fig. 12 , this resulted in the alteration of the period between two laser pulses since the laser oscillation angular range is not equally distributed in time and tends to come in pairs (Fig. 12b). The time interval between the two pulses is not sufficiently long for maximum energy pulse generation. In the case of the MEMS being aligned to the edge of its scanning range (Fig. 12c), only one active state occurs within a full tilting cycle resulting in a pulse repetition frequency equal to the MEMS resonant frequency.
The micromirror could also be horizontally translated so that the incident laser beam was located on the edge of the mirrored surfaced. In this way, the intra-cavity laser beam was rapidly apertured during the mirror scan. Hence, the shortest pulse recorded (220ns) in this configuration was obtained at a frequency of 10.3kHz with an average power of 20mW (see Fig. 11b). The resulting peak power was ~13.5W and peak power density 330kW/cm2. The addition of this aperturing effect (induced by the proximity of the laser beam to the mirror edge) to the tilting effect described in the previous paragraph contributed to the pulse shortening from 430ns to 220ns. Placing the laser beam as far as possible from the mirror’s rotation axis has the potential to reduce the switching time. Unfortunately, larger circular micromirrors where this edge effect can be achieved will inherently possess a lower resonance frequency. Hence micromirrors with a cantilever shape or with an oval shape present significant scope and will be the subject of future investigations.
It has been shown that the minimum pulse duration τmin that can be obtained from a Q-switched laser is :Fig. 10a was calculated to be 4.1ns. Since the pump power (20W) was close to the pump power required to reach the laser threshold (17W), the round trip gain can be estimated to be slightly above 20%. Using Eq. (1), the minimum possible pulse duration is therefore expected to lie in the range of 160-180ns indicating that the MEMS micromirror provided near-optimal Q-switching performance. In the present configuration, the pulse duration was mainly limited by the cavity length and the small signal gain induced. Modeling suggests that a reduction in the cavity length to 100mm would result in pulse duration of ~25ns. Further significant pulse width reduction (via increased small signal gain) would require the misalignment sensitivity of the resonator to be increased. Larger spot sizes on the micromirror are needed to obtain shorter gate times. However, this would significantly enhance the perceived heating effects described in section 3.1. For this reason, we believe that the thermal-induced surface deformation must be reduced in an optimized system.
The heat-induced deformation of the micromirror surface represents the main challenge to overcome when using MEMS inside laser cavity at the power levels used in our experiments. In the laser system described here, incident radiation (with λ = 1047nm) on the MEMS is mainly reflected by the dielectric coating. However, the small proportion not reflected (<1%) is completely absorbed by the silicon leading to a bimorph effect taking place and therefore to the deformation of the reflective surface of the MEMS.
Power scaling of intra-cavity MEMS lasers can therefore be obtained with the use of MEMS at wavelengths where silicon is transparent (i.e. above 1.2µm). Alternatively, a careful design of MEMS can also significantly reduce the heat-induced distortion (i.e. the development of ‘athermal’ MEMS micromirrors). Since the deformation is due to the difference in thermal expansion between the dielectric coating and the silicon, it is anticipated that the application of a similar dielectric coating on the rear surface of the micromirror could compensate for this effect. In addition, the bimorph effect can be reduced by reducing the difference in coefficient of thermal expansion between the silicon-based chip and the dielectric coating. It is anticipated that the use of hybrid coatings (i.e. coatings made of a combination of gold and dielectric materials) could reduce this mismatch and therefore significantly limit the surface curvature.
Finally, using a material transparent to the laser light in building athermal MEMS micromirrors is another option. For instance, silicon carbide MEMS devices developed for applications in harsh environments , have the potential to power scale MEMS lasers operating at λ~1μm due to the wide transparency range above ~400nm and the high thermal conductivity (4.9W/cm/K) of this material .
A MEMS-based laser system has been demonstrated with temporal output modulation using an intra-cavity micromirror. Slow (period ~few s) temporal modulation of the laser output was obtained with a micromirror based on thermal actuation while laser Q-switching was obtained with a micromirror based on electrostatic actuation. Significant heat-induced deformation of the surface of the electrothermal micromirror particularly was observed when power densities greater than 50W/cm2 were incident to the MEMS mirror, eventually leading to the complete loss of laser oscillation (the laser cavity became optically unstable). Such a deformation is due to the strong absorption by the silicon of the non-reflected laser light, this in turn limited the CW laser maximum output power to 200mW and the average Q-switched power to 60mW. The shortest pulse duration (220ns) recorded was consistent with the theoretical minimum obtainable pulse duration. The surface deformation then represents the primary limitation in power scaling of lasers featuring MEMS components – this is the subject of on-going investigations. Not withstanding this issue, this type of device has the potential to Q-switch solid-state laser systems and opens up opportunities for independently programmable arrays of lasers with a compact footprint and using batch-manufactured MEMS.
References and links
1. O. Solgaard, Photonic Microsystems: Micro and Nanotechnology Applied to Optical Devices and Systems, (Springer, New York, 2009), Chap. 7.
2. A. Q. Liu, X. Zhang, J. Li, S. H. G. Teo, F. Lewis, and B. Borovic, Photonic MEMS Devices: Design, Fabrication and Control, (CRC Press, Boca Ranton, USA, 2009).
3. Y.-A. Peter, H. P. Herzig, E. Rochat, R. Dändliker, C. Marxer, and N. F. de Rooij, “Pulsed fiber laser using micro-electro-mechanical mirrors,” Opt. Eng. 38(4), 636–640 (1999). [CrossRef]
4. D. Bouyge, A. Crunteanu, D. Sabourdy, P. Blondy, V. Couderc, J. Lhermite, L. Grossard, and A. Barthélémy, “Integration of micro-electro-mechanical deformable mirrors in doped fiber amplifiers,” Microsyst. Technol. 13(11-12), 1607–1613 (2007). [CrossRef]
5. A. Inoue, T. Komikado, K. Kinoshita, J. Hayashi, and S. Umegaki, “Deformable Mirror for Mechanical Q-Switching of Laser-Diode-Pumped Microchip Laser,” Jpn. J. Appl. Phys. 46(42), L1016–L1018 (2007). [CrossRef]
6. M. Fabert, A. Desfarges-Berthelemot, V. Kermène, A. Crunteanu, D. Bouyge, and P. Blondy, “Ytterbium-doped fibre laser Q-switched by a cantilever-type micro-mirror,” Opt. Express 16(26), 22064–22071 (2008). [CrossRef] [PubMed]
7. M. Fabert, A. Crunteanu, V. Kermène, A. Desfarges-Berthelemot, D. Bouyge, and P. Blondy, “8ns pulses from a compact fibre-laser Q-switched by MOEMS,” in Conference on Laser and Electro-Optics 2009, Technical Digest (CD) (Optical Society of America, 2009), paper CFB6.
8. S. Schilt, K. Zogal, B. Kögel, P. Meissner, M. Maute, R. Protasio, and M.-C. Amman, “Spectral and modulation properties of a largely tunable MEMS-VCSEL in view of gas phase spectroscopy applications,” Appl. Phys. B 100(2), 321–329 (2010). [CrossRef]
9. S. Hsu, T. Klose, C. Drabe, and H. Shenk, “Fabrication and characterization of a dynamically flat high resolution micro-scanner,” J. Opt. A, Pure Appl. Opt. 10(4), 044005 (2008). [CrossRef]
10. D. Hah, P. R. Patterson, H. D. Nguyen, H. Toshiyoshi, and M. C. Wu, “Theory and experiments of angular vertical comb-drive actuators for scanning micromirrors,” IEEE J. Sel. Top. Quantum Electron. 10(3), 505–513 (2004). [CrossRef]
11. A. Jain, A. Kopa, Y. T. Pan, G. K. Fedder, and H. K. Xie, “A two-axis electrothermal micromirror for endoscopic optical coherence tomography,” IEEE J. Sel. Top. Quantum Electron. 10(3), 636–642 (2004). [CrossRef]
12. T. Sandner, J. U. Schmidt, H. Schenk, H. Lakner, M. Yang, A. Gatto, N. Kaiser, S. Braun, T. Foltyn, and A. Leson, “Highly reflective optical coatings for high power applications of micro scanning mirrors in the UV-VIS-NIR spectral region,” Proc. SPIE 6114, H1140–H1140 (2006).
13. MEMSCAP Inc, 12 Alexander Drive, Building 100, Research Triangle Park, NC 27709, USA, www.memscap.com.
14. L. Li, M. Begbie, G. Brown, and D. Uttamchandani, “Design, simulation and characterization of a MEMS optical scanner,” J. Micromech. Microeng. 17(9), 1781–1787 (2007). [CrossRef]
16. W. Lubeigt, M. Griffith, L. Laycock, and D. Burns, “Reduction of the time-to-full-brightness in solid-state lasers using intra-cavity adaptive optics,” Opt. Express 17(14), 12057–12069 (2009). [CrossRef] [PubMed]
17. Cutting Edge Optronics, 20 Point West Boulevard, St. Charles, MO 63301, USA, http://www.st.northropgrumman.com/ceolaser/.
18. J. Zayhowski, “Microchip lasers,” Opt. Mater. 11(2-3), 255–267 (1999). [CrossRef]
19. R. Cheung, Silicon Carbide Microelectromechanical systems for harsh environments (Imperial College Press, London, UK, 2006).
20. M. E. Levinshtein, S. L. Rumyantsev, and M. S. Shur, Properties of advanced semiconductor materials: GaN, AIn, InN, BN, SiC, SiGe (John Wiley & Sons, New York, USA, 2001), Chap. 5.