SEMICONDUCTOR lasers play a very important role for

Transcription

1 116 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 43, NO. 2, FEBRUARY 2007 Characterization of the Main Semiconductor Laser Static and Dynamic Working Parameters From CW Optical Spectrum Measurements Asier Villafranca, Javier Lasobras, José A. Lázaro, Member, IEEE, and Ignacio Garcés, Member, IEEE Abstract We present a complete characterization of the work parameters of several types of semiconductor lasers. Static parameters as: power, linewidth and linewidth enhancement factor and also dynamic parameters such as: relaxation oscillations, relative intensity noise and damping rates are calculated using measurements of the optical spectrum of the lasers operated in continuous-wave mode. Methods for the calculation of these parameters are described and applied to the lasers under test by means of a single general setup and a single set of measurements. Index Terms Distributed feedback lasers (DFB), laser measurements, linewidth enhancement factor, multiquantum-well lasers (MQW), relative intensity noise (RIN), semiconductor lasers, vertical-cavity surface-emitting lasers (VCSELs). I. INTRODUCTION SEMICONDUCTOR lasers play a very important role for practical high-speed optical transmission systems and also as a basic research topic. A precise characterization of a semiconductor laser is a hard task that normally involves multiple setups and measurements [1]. Most of the experiments to be carried out for the characterization of a laser diode require the laser to be modulated. Although this does not represent a problem for final packaged devices, it is a drawback for in-chip devices that are common in research and initial stages of fabrication. The techniques we propose and study in this paper have the advantage of being performed with the laser emitting in continuous-wave (CW) mode, thus not needing modulation to measure dynamic parameters. In the present work, we will describe a process for the complete experimental evaluation of the main working parameters for semiconductor lasers, and we will apply it to several types of market-available semiconductor lasers. When referring to working parameters, we include the power response, emission wavelength, linewidth, linewidth enhancement factor, relative intensity noise (RIN), relaxation oscillations peak frequency and damping rate. The main advantages of our proposal are its simplicity, both theoretical and experimental, the ease of implementation and automation, and, as has been pointed out, the fact that it does not require the laser to be modulated. Additionally, we will perform a systematic comparison of the parameters values for several types of semiconductor lasers applying the same methods and instruments which has not, to our knowledge, been done previously. This paper is structured at follows. In Section II, we will present the theoretical background, deriving the expressions for the optical CW spectrum from the laser rate equations. An experimental approach is favored in this paper in order to provide a whole view of the characterization method. Consequently, a reduced version of the calculations is presented as the underlying semiconductor laser theory can be found in given references. In Section III, we will present the methodology for the measurements, with the setup and measurement process explained in Section IV. Results will be given and discussed in Section V, and the conclusions will be addressed in Section VI. II. THEORETICAL BACKGROUND Semiconductor lasers can be modeled as non-ideal oscillators that have a narrow but finite width caused by noise-driven modulation [2]. Carrier noise produces a direct optical intensity noise, and also induces a refractive index fluctuation that leads to frequency or phase noise [3]. We can analyze the effect of these two sources of noise on the optical spectrum of a single mode semiconductor laser operating in CW far from threshold. The expression for the power spectral density (PSD) in these conditions, shifted to zero frequency, is [4] Manuscript received August 3, 2006; revised October 4, This work was supported in part by the Diputación General de Aragón Pproject PIP108/2005, in part by the Ramón y Cajal Program from M.E.C. and in part by the the i3a Research Laboratories,, Cuarte, Spain. A. Villafranca, J. Lasobras, and J. I. Garcés are with TOYBA Laboratory, Aragon Institute of Engineering Research, Cuarte 22197, Spain ( asiervv@unizar.es; lasobras@unizar.es; ngarces@unizar.es). J. A. Lázaro is with the Signal Theory and Communications Department, Universitat Politècnica de Catalunya, Barcelona 08034, Spain ( jose.lazaro@tsc.upc.edu). Digital Object Identifier /JQE where is the full-width at half-maximum of the lasing mode, given by the phase noise, is the relaxation oscillation peak frequency, is the damping rate of the relaxation oscillations, and is the linewidth enhancement factor. We can separate two components in (1). The first term is a Lorentzian curve centered at the lasing wavelength, related to (1) /$ IEEE

2 VILLAFRANCA et al.: CHARACTERIZATION OF THE MAIN SEMICONDUCTOR LASER STATIC AND DYNAMIC WORKING PARAMETERS 117 the phase noise [5] that gives the main contribution to the total power and linewidth. The second term arises a few gigahertz from the peak and has an asymmetric Lorentzian shape, given by the interaction between the relaxation oscillations and the intensity noise [6], which form a pair of shoulders at both sides of the main peak. This asymmetry is influenced by a and has been reported as a measurement method for the linewidth enhancement factor [4]. However, the precision of this method has been regarded as poor compared with other reported methods [7] [9]. Therefore, this second term in (1) can be considered as the output of the laser modulated with a noise source [6]. The oscillation relaxation peak frequency and damping rate can be obtained directly from fitting to (1), but also the RIN can be calculated from the optical CW spectrum. The RIN is defined as the ratio of the electrical power spectral density at a given frequency to the average power. It is normally measured in the electrical domain, so that the power spectral density is the sum of the detected intensity noise and the contribution of the shot noise. To calculate the RIN from the electrical spectrum after the photodiode the following expression is used [6]: where is the intensity detected by a photodiode and is the shot noise. The optical spectrum is not affected by shot noise, but it has contributions of the phase noise that will not produce intensity in a photodetector. However, from (1) we know that the phase noise has a Lorentzian distribution that we can measure and subtract from the optical spectrum so we can calculate the electrical spectrum of the signal without the contribution of the shot noise where is the part of the optical power spectral density corresponding to the phase noise, given by the first term of (1). When the laser is operated far from saturation, a simpler form can be used With a precise knowledge of the linewidth enhancement factor, the laser spectrum can be fitted to (1) for most of the operating range of the laser, obtaining the values for the most often used working parameters for the use of semiconductor lasers in telecommunications. For the measurement of, in this paper we use the modified linewidth method, which has been shown to have a high accuracy by comparison with other methods [9]. This method extracts a from the relation of the linewidth with emitted power below and above threshold. In the case of semiconductor lasers biased below threshold, this relation follows the well-known Schawlow Townes equation for gas lasers [10] (2) (3) (4) (5) where is the total spontaneous emission rate and is the optical power in the mode. During the threshold transition this relation is far more complex [11], but it tends rapidly to the asymptotic behaviour predicted by the Henry law [12] The relation between (5) and (6) depends only on the linewidth enhancement factor, so can be calculated as This equation provides a value of for the laser diode, which is constant as long as there is no power saturation [13]. III. METHODOLOGY The CW spectrum of a semiconductor laser contains all the information needed to characterize its working parameters, but is not normally not exploited because measuring it with enough resolution or dynamic range, and specially with a good combination of both, is hard with traditional optical spectrum analyzers (OSA) based on gratings and other spectrometry techniques such as tunable Fabry Perot filtering or heterodyne detection. However a new generation of analyzers has recently become available, based on the stimulated Brillouin scattering nonlinear effect [14]. Using such analyzers it is possible to obtain a resolution of 10 MHz along with a dynamic range of around 80 db. When measuring the optical spectrum with this scope we can see details that cannot be measured by other means, and are normally not resolved either in frequency or in dynamic range. The measurements performed here are the closest to the ideal optical spectrum with theoretical infinite resolution to our knowledge. However, the optical spectra we present are well known from literature and optical communications simulators, providing accurate information from which a very precise analysis of the measurement can be made. We can distinguish two sets of measurements: a sweep for the whole range of operating currents of the laser, and a detailed sweep at shorter steps in the threshold region, which has a particular interest for power and linewidth. When passing the laser threshold, an important narrowing of the laser linewidth is produced. While below threshold no correction due to the analyzer filter is needed, as the spontaneous emission has linewidth values on the order of the gigahertz, above threshold the linewidth gets closer to the filter bandwidth of the analyzer and it must be taken into account. We can take advantage of the Gaussian profile of this filter [15] and the Lorentzian lineshape of the phase noise spectrum of the semiconductor lasers and use the well known Voigt function with and where is the central wavelength of the laser, is the width of the analysis filter assuming Gaussian profile and (6) (7) (8)

3 118 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 43, NO. 2, FEBRUARY 2007 Fig. 1. Measured spectra for the lasers under test, measured around threshold region: (a) Laser A (c) Laser B (e) Laser C; and far above threshold: (b) Laser A (d) Laser B (f) Laser C. Bias current is normalized to I =I 0 1. is the linewidth of the laser. This fact is only really important in the proximities of the central emission wavelength, which is where the width of the laser shape can be of the order of the analysis filter. As we move towards lower power levels, the laser emission shape is wider, so this is no longer relevant for the functional form of the measurements and only a magnitude correction must be made if the power spectral density (PSD, in dbm/hz) is measured. In this case, the effective width of the analysis filter has to be considered, i.e., the width of an equivalent square filter that would give the same power measurement. For the Gaussian shape, this effective width is 1.06 times the measured width at 3 db, resulting in a relation of db from the power measured with the instrument, dbm through a filter having a FWHM of 10 MHz, and the PSD of the spectrum. This is also useful to measure the full emission power of the laser, which can be calculated as the integral of the measured PSD, and thus avoiding the contribution of the suppressed side modes that may distort the measurements close to threshold when using a photodetector. IV. MEASUREMENTS For this paper we have characterized three commercially available lasers: a distributed feedback (DFB) laser for CW (which we will denote by laser A from now on), a low-chirp multiquantum well (MQW) DFB laser (laser B), and a long-wavelength vertical-cavity surface-emitting laser (VCSEL) (laser C). We operate the lasers under CW conditions using a laser mount and a bias and temperature controller LDC2000 from ThorLabs Inc. The output of the lasers is directly inserted into a high resolution optical analyzer BOSA-C

4 VILLAFRANCA et al.: CHARACTERIZATION OF THE MAIN SEMICONDUCTOR LASER STATIC AND DYNAMIC WORKING PARAMETERS 119 Fig. 2. Sample measured and fitted spectra of Laser A biased at 50 ma. from Aragon Photonics. An additional isolator is used with the VCSEL due to its strong sensitivity to backreflections. All the data needed for our analysis is a series of measurements of the laser under test in CW emission mode under different biasing conditions. The bias current is swept from the lowest the analyzer can measure, which is typically around 5% 10% below threshold for tested devices, to some arbitrary current in the operating range given by the laser manufacturer. The region around threshold must be measured in very small steps, as changes are very fast in this region for most of the properties of the laser. Measured spectra for the lasers around the threshold region can be seen in Fig. 1(a) (c). The lower curves correspond to spontaneous emission below threshold and they have a typical Gaussian lineshape, with a width on the order of gigahertz. As bias current is increased, stimulated emission becomes dominant and the width of the spectrum decreases, especially when approximating to the threshold current. Also the lineshape suffers some changes, resulting in a very well fitted Lorentzian function around threshold. At this point, only the phase noise of the laser is considered. If we continue increasing the bias current, an additional phenomenon can be observed as a certain deformation in the Lorentzian shape of the spectrum, as can be seen in the upper curves in Fig. 1(a) (c). At higher bias currents, this effect becomes quite more recognizable, as its behavior follows clearly the theory of laser intensity noise, as can be seen in Fig. 1(d) (f), where a higher range of bias current values is represented. The laser intensity noise becomes manifested as a soft increase of the power spectral density at frequencies far from the peak power where the spectral density due to phase noise has fallen to very low values. The sum of phase noise and intensity noise forms a certain plateau that continues until a maximum is reached and then decreases sharply. In the case of the VCSEL in Fig. 1(f), mode instability is observed [16], resulting in mode hopping that does not allow a good measurement of the laser profile above 20 dbm. However, this frequency hopping is only critical to the peak of power, as it is very narrow, but does not represent a big problem in the measurement of the PSD of the intensity noise, which is quite broader. Fig. 3. Power response of the three lasers under test and fitting to (9). The frequency separation between the peak of power and the maximum of the intensity noise varies with the value of the bias current, following the well-known linear relationship between the squared frequency and the bias current [17]. The other consequence of the increasing bias current is reflected in the decreasing of the spectral density of the intensity noise, which is consistent with the typical behavior of the RIN [18]. A fit of the measured optical spectrum to a Voigt function is shown Fig. 2. Linewidth values can be obtained from this fit. The remaining power spectral density after subtracting this profile is basically the intensity noise spectrum, which can be used to calculate both the relaxation oscillation peak frequency and the RIN. A. Power Response V. RESULTS AND DISCUSSION The optical power contained in the lasing mode is calculated as the integral of the PSD in the wavelength region of the lasing mode. The first order approximation of the power response is for with the threshold current and the inverse of the power to bias current ratio in linear regime, given in A/W. However, (9) is quite inaccurate in the threshold region, which is critical for the measurement of. A more complex relation between the bias current and the emitted power is used for fitting [1] (9) (10) where is the spontaneous emission term. As can be seen in Fig. 3, measurements fit very well to the theoretical behavior given by (10). Measuring this with a photodetector usually leads to an overestimation of the mode power, as below threshold all modes contribute to the power, and (10) does not give a good fit. In the case when the direct measurement of a standard OSA is used, the filter bandwidth must be carefully selected as a broad filter will have a certain power offset due

5 120 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 43, NO. 2, FEBRUARY 2007 Fig. 4. Measured dependence of the linewidth with emitted optical power for the lasers under test. TABLE I SUMMARY OF MEASURED PARAMETERS Fig. 5 Mesaured dependence of the relaxation oscillations peak frequency (squared) with emitted power for lasers under test. applying (7). Calculated parameters for the three lasers are given in Table I. These measurements have been confirmed using a network analyzer and using the transfer function after propagation through some hundreds of kilometers of fiber [10]. Measurements obtained with this technique give a very good agreement. C. Relaxation Oscillations to noise, while for a narrow filter some power may be outside the filter. Fitted values for the threshold current are given in Table I. B. Linewidth and Linewidth Enhancement Factor The linewidth of the DFB laser can be calculated quite precisely if we take (8) into account. The Voigt function has been extensively studied and we can use an easier expression to calculate laser linewidth [19] (11) where is the measured linewidth. From (10) we can deduce that the overestimation on the measured linewidth value is only important when the laser linewidth is comparable to the bandwidth of the analysis filter. Linewidth values corrected with (11) are shown in Fig. 4. The dependence of the linewidth with emitted power is quite complex [11]. We can see how linewidth is much greater below threshold, as the contribution from spontaneous emission is dominant. The threshold region can be identified as a local maximum of the linewidth, and the linewidth continues decreasing as power increases. In Fig. 4, the asymptotic behaviors of the linewidth versus the inverse optical power are shown, following the Schawlo Townes (5) and Henry (6) laws. The linewidth enhancement factor can be calculated from those two laws by From (1) we can notice that the optical spectrum of the lasers should exhibit a maximum coinciding with the relaxation oscillation peak frequency. This maximum is observed in the second row of measurements in Fig. 1, and can be easily measured in the optical spectrum. The common measurement method involves measuring the laser transfer function with a network analyzer [20]. In Fig. 5, the values of the relaxation oscillation frequency measured from the optical spectra versus the current are shown, presenting the usual square relation with bias current above threshold. The relaxation oscillations peak frequency gives information about the modulation bandwidth by the relation [21]. D. RIN The measurement of RIN from the high-resolution optical spectrum is also very direct for frequencies far enough from the peak of emission, normally above 1 GHz, as for lower frequencies the fitting error to the Voigt function is comparable or even higher than the intensity noise itself. Applying (4), we obtain the desired RIN curves as function of frequency, equivalent to those that we could measure in the electrical domain after detection, but with the advantage that shot noise is not present. These curves, along with a fit to the second term on (1) are shown in Fig. 6. The peak RIN value as a function of bias current is given in Fig. 7. Some typical values for the oscillation relaxation frequency and RIN are also given in Table I. The usual method for the measurement of laser RIN is to perform a direct detection of the laser output with a calibrated fast photodiode and use an electrical spectrum analyzer. Shot noise must be measured separately using a light-emitting diode

6 VILLAFRANCA et al.: CHARACTERIZATION OF THE MAIN SEMICONDUCTOR LASER STATIC AND DYNAMIC WORKING PARAMETERS 121 Fig. 8. Damping rate as a function of bias current for measured lasers. we avoid the need of doing these separate measurements and we obtain very accurate results for a wide range of values. It is worth noting that RIN measurements for the VCSEL (Laser C) are only valid over a small range of currents, due to the instability in the measured optical spectrum. Nevertheless, measurements made in this range are good enough for extrapolating RIN values at higher currents. The damping rate can also be obtained after fitting the RIN curves, as contained in (1). Fitted values for this parameter are given in Fig. 8. Very linear behavior with bias current is obtained for this parameter. E. Summary and Discussion Fig. 6. Relative intensity noise as a function of frequency from the peak of emission for several bias current for Laser A (upper graph), Laser B (middle graph), and Laser C (bottom graph). Values of the measured parameters for all lasers under test are summarized in Table I. The measurements we have performed cover a wide range of values for the work parameters of communication grade semiconductor lasers. Measured parameters show typical values and good behavior compared to theory and to these values provided by the manufacturers. Measurement with other methods have shown relative differences of less than 5%. The value of a for the VCSEL is higher than other reported values [22], but it has been checked with other methods [23] confirming the reported value. This may be caused by the internal optical feedback [16]. VI. CONCLUSION Fig. 7. Peak RIN as a function of bias current for lasers under test. Curves follow the expected behavior. (LED) and subtracted from the measured spectrum of the semiconductor laser [18]. Flat responsivity of the photodiode is normally assumed in this calibration. With our proposed method We have presented a methodology for the characterization of the main working parameters of semiconductor lasers emitting in CW. Compared with traditional methods used to perform these same measurements, we think our method has some advantages: it does not require modulation, so it can be performed over chips under development at early stages, and it requires the use of only one setup for all the measurements, which makes the process much faster and easier to automate. We have performed the measurements over several types of semiconductor lasers with very different characteristics, showing the robustness and accuracy of the method. Even for low power VCSELs it can be used to extract very relevant

7 122 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 43, NO. 2, FEBRUARY 2007 information of the devices, although results may be more inaccurate as the power decreases. We believe this same methodology can be applied to other types of single-mode semiconductor lasers without further revisions. ACKNOWLEDGMENT The authors wish to thank useful discussions within the European COST action 288. REFERENCES [1] L. Bjerkan, A. Røyset, L. Hafskjaer, and D. Myhre, Measurement of laser parameters for simulation of high-speed fiberoptic systems, J. Lightw. Technol., vol. 14, no. 5, pp , May [2] D. E. McCumber, Intensity fluctuations in the output of cw laser oscillators. I, Phys. Rev., vol. 141, no. 1, pp , [3] Y. Yamamoto, AM and FM quantum noise in semiconductor lasers Part I: theoretical analysis, IEEE J. Quantum Electron., vol. QE-19, no. 1, pp , Jan [4] M. P. van Exter, W. A. Hamel, J. P. Woerdman, and B. R. P. Zeijlmans, Spectral signature of relaxation oscillations in semiconductor lasers, IEEE J. Quantum Electronics, vol. 28, no. 6, pp , Jun [5] R. D. Hempstead and M. Lax, Classical noise VI: noise in self-sustained oscillators near threshold, Phys. Rev., vol. 161, no. 2, pp , [6] R. Schimpe, Intensity noise associated with the lasing mode of a GaAlAs diode laser, IEEE J. Quantum Electron., vol. QE-19, no. 6, pp , Jun [7] F. Devaux, Y. Sorel, and J. F. Kerdiles, Simple measurement of fiber dispersion and of chirp parameter of intensity modulated light emitter, J. Lightw. Technol., vol. 11, no. 12, pp , Dec [8] Y. Yu, G. Giuliani, and S. Donati, Measurement of the linewidth enhancement factor of semiconductor lasers based on the optical feedback self-mixing effect, IEEE Photon. Technol. Lett., vol. 16, no. 4, pp , Apr [9] A. Villafranca, J. A. Lázaro, I. Salinas, and I. Garcés, Measurement of the linewidth enhancement factor in DFB lasers using a high-resolution optical spectrum analyzer, IEEE Photon. Technol. Lett., vol. 17, no. 11, pp , Nov [10] A. L. Schawlow and C. H. Townes, Infrared and optical masers, Phys. Rev., vol. 112, no. 6, pp , [11] Z. Toffano, Investigation of threshold transition in semiconductor lasers, J. Sel. Topics Quantum Electron., vol. 3, no. 2, pp , Apr [12] C. H. Henry, Theory of the linewidth of semiconductor lasers, J. Quantum Electron., vol. QE-18, no. 2, pp , Feb [13] G. P. Agrawal, Intensity dependence of the linewidth enhancement factor and its implications for semiconductor lasers, IEEE Photon. Technol. Lett., vol. 1, no. 8, pp , Aug [14] J. Subías, J. Pelayo, F. Villuendas, C. D. Heras, and E. Pellejer, Very high resolution spectroscopy by stimulated Brillouin scattering, IEEE Photon. Technol. Lett., vol. 17, no. 4, pp , Apr [15] A. Villafranca, J. A. Lázaro, I. Salinas, and I. Garcés, Stimulated Brillouin scattering gain profile characterization by interaction between two narrow-linewidth optical sources, Opt. Exp., vol. 13, pp , [16] J. Y. Law and G. P. Agrawal, Effects of optical feedback on static and dyamic characteristics of vertical-cavity surface-emitting lasers, IEEE J. Sel. Top. Quantum Electron., vol. 3, no. 2, pp , Apr [17] K. Peterman, Laser Diode Modulation and Noise. Norwell, MA: Kluwer, [18] I. Joindoint, Measurement of relative intensity noise (RIN) in semiconductor lasers, J. Phys. III, vol. 2, pp , [19] J. J. Olivero and R. L. Longbothum, Empirical fits to the voigt linewidth: a brief review, J. Quantum Spectrosc. Radiat. Transfer, vol. 17, p. 233, [20] S. Lindgren, H. Ahlfeldt, and L. Backlin et al., 24-GHz modulation bandwidth and passive alignment of flip-chip mounted DFB laser diodes, IEEE Photon. Technol. Lett., vol. 9, no. 3, pp , Mar [21] G. P. Agrawal, Fiber-Optic Communication Systems. New York: Wiley, [22] P. Signoret, F. Marin, S. Viciani, G. Belleville, M. Myara, J. P. Tourrenc, B. Orsal, A. Plais, F. Gaborit, and J. Jacquet, 3.6 MHz linewidth 1.55-m monomode vertical-cavity surface-emitting laser, IEEE Photonics Technol. Lett., vol. 13, no. 4, pp , Apr [23] R. C. Srinivasan and J. C. Cartledge, On using fiber transfer functions to characterize laser chirp and fiber dispersion, IEEE Photon. Technol. Lett., vol. 7, no. 11, pp , Nov Asier Villafranca received the M.Sc. degree in communications engineering from University of Zaragoza, Zaragoza, Spain, in 2004, where he is currently working towards the Ph.D. degree in signal theory and communications. Since 2005, he has been with the Laboratory of Optical Transmission and Broadband Technologies (TOYBA Lab.), Aragon Institute of Engineering Research (I3A), Cuarte, Spain. His research interests include measurement techniques for electro-optical devices, nonlinear effects and passive optical networks. Javier Lasobras received the M.Sc. degree in communications engineering from University of Zaragoza, Zaragoza, Spain, in Since 2006, he has been with the Laboratory of Optical Transmission and Broadband Technologies (TOYBA Lab.), Aragon Institute of Engineering Research (I3A), Cuarte, Spain. His research interests include optical spectrum measurement techniques and applications. José A. Lázaro (M 01) received the Ph.D. degree in physics from University of Zaragoza, Zaragoza, Spain, in 1999 for his research work in erbium-doped waveguide amplifiers. He was as Assistant Professor in the University of Zaragoza teaching physics, as an R&D Engineer in Alcatel at the Department of Passive Optical Components ( ), working mainly in design and characterization of Athermal AWGs, and the Deptartment of Optical Transmission Systems ( ) in R&D of new modulation formats and 40 Gb/s subsystems. He was Senior Researcher at the Optical Transmission and Broadband Technologies Laboratory, Aragon Institute for Engineering Research (I3A), Cuarte, Spain ( ). Currently, he is a Ramon-y-Cajal Researcher at the Optical Communications Group, Polytechnic University of Catalonia (UPC), Barcelona, Spain. Juan Ignacio Garcés (M 96) received the M.S. and Ph.D. degrees in physics from the University of Zaragoza, Zaragoza, Spain, in 1989 and 1996, respectively. After working at Cables de Comuniaciones for three years, he joined University of Zaragoza, where he is now an Associate Professor at the Departamento de Ingeniería Electrónica y Comunicaciones. Since 2004, he has been the Coordinator of the Optical Transmission and Broadband Technologies Laboratory (TOYBA Lab.). His research interests include optical networks and applications of high resolution spectroscopy. Dr. Garcés is currently Vice-President of the Optoelectronics Comitee of the Spanish Optical Society.