Open Access
Add to BibSonomyAbstract
We report the fabrication of single mode quantum cascade lasers using a shallow-etched distributed Bragg reflector as frequency selective element. Quasi-continuous single mode tuning over 15 cm−1 at room temperature and 25 cm−1 via temperature tuning at Peltier temperatures is demonstrated. The behavior of both electro-optic and spectral characteristics under variation of the segment currents is analyzed, showing a maximum peak output power at room temperature of 600 mW. Thermal crosstalk between the laser segments is investigated. The spectral resolution of a gas absorption experiment is determined to be better than 0.0078 cm−1.
©2012 Optical Society of America
1. Introduction
Most laser spectroscopy experiments and applications in the mid-infrared (MIR) regime rely on sources that emit light with a single (and preferably tunable) wavelength. Multiple ways to achieve single mode emission of quantum cascade lasers (QCLs) for gas sensing in the MIR have been demonstrated. The most commonly used devices are distributed feedback (DFB) QCLs [1–3]. However, they are usually tuned by changing the operation temperature, which fundamentally limits their tuning range on standard thermoelectric coolers to about 5 cm−1 for QCLs emitting in the 700-1000 cm−1 range [4,5]. Different solutions to extend the tuning range have been realized: External cavity QCLs [6] offer single mode tuning over most of the available gain width (up to > 400 cm−1) [7], but require external optics and therefore a larger setup. If a highly miniaturized and rugged solution is required, monolithic approaches such as DFB-QCL arrays [8] or coupled cavity QCLs [9–11] are preferred choices. However, the former require additional optics to combine the individual beams and suffer from considerable performance spread of the individual lasers if edge emitters are used [12]. The latter can be operated only at limited output powers, since undesired multiple resonances are excited at high injection currents if the lengths of the segments are chosen such that quasi-continuous tuning is possible. QCLs with deeply etched and highly reflective distributed Bragg reflectors (DBR) have previously been demonstrated, however they did not offer an extended tuning range [13] in comparison to DFB-QCLs. Another approach yielded multimode emission that could only be tuned once by an irreversible change of the refractive index of a chalcogenide glass which was deposited on a shallow-etched DBR-section [14].
To overcome these drawbacks, our approach is based on monolithic devices comprising a DBR-section with a shallow-etched grating and a gain section (see Fig. 1(a) ) with lengths LDBR and Lgain, respectively. These two sections are electrically separated at the top contact and share a common bottom contact. The sequence of current pulses that are applied to the individual segments is depicted in Fig. 1(b). First, a current IDBR is injected into the DBR-segment, heating it to the desired temperature before the gain current Igain is applied and laser emission is excited. As confirmed by a thermal simulation of the device and experimental data presented in section 3.3, no significant heat transfer from the DBR- to the gain-segment occurs before the gain pulse is applied. Therefore the photon generation in the gain segment, which remains at the heat sink temperature Tsink, is not affected by the extensive heating of the DBR-mirror if the pulsed driving scheme described above is chosen. It should be mentioned that the tuning behavior does not change essentially if the current pulses in the gain- and the DBR-segment do not overlap and subsequently no gain in the DBR-segment is present during laser emission. But since overlapping pulses lead to lower threshold currents Ith,gain and higher output power due to reduced absorption in the DBR-segment all presented data was recorded with τDBR ≥ Δt + τgain. For phase matching of the DBR-mirror to the front facet a constant sub-threshold current Iphase can be injected into the gain segment.
Fig. 1 (a) Schematic drawing of the DBR-QCL with two segments and related currents. (b) Sequence of pulsed currents IDBR (black line) and Igain (red line).
2. Device fabrication and experimental setup
The DBR structure of the devices emitting at 10.3 µm (grating depth: 1.4 µm, period: 1.594 µm) was dry etched [4] into the cap of an epitaxial structure based on the gain design published in [5]. Furthermore, devices emitting at 13.5 µm were fabricated from the gain material described in [4] and etched to a grating depth of 1.7 µm (period: 2.097 µm). Both layer structures are based on the InGaAs/InAlAs/InP material system and grown lattice matched to InP by gas source molecular beam epitaxy. Double channel ridges were formed through wet chemical etching and insulated with SiO2. A contact window was opened, Ti/Pt/Au was deposited and removed from the boundary surface between the two laser segments via a lift-off step. Thick gold was electroplated to improve heat dissipation from the ridge. Finally the substrate was thinned to 150 µm and a Ti/Pt/Au back contact was evaporated. Lasers were soldered on c-mounts with facets left as-cleaved.
Tsink was controlled with a thermoelectric cooler. The pulses from two laser drivers (Directed Energy, model PCX 7410) which were triggered with a time delay of Δt were delivered via microprobes. The electric current was measured using inductive current probes. The peak output power was collected from the facet of the gain segment and was determined with a HgCdTe detector after calibration with a thermopile detector. Spectra were collected with a Fourier transform infrared (FTIR) spectrometer with a resolution of 0.12 cm−1.
3. Device characterization
3.1 Spectral properties
The tuning behaviour of devices under variation of IDBR is shown in Fig. 2 . The corresponding spectra were recorded with an automated measurement setup using a step width of ΔIDBR = 10 mA. A tuning range of 15 cm−1 was observed for devices emitting at 10.3 µm at constant Tsink = 20°C. Additional temperature tuning through variation of Tsink enabled a spectral coverage of 25 cm−1 (see Fig. 2(a)). A staircase shape is observed as the reflectivity of the DBR is pushed to longer wavelengths and the emission switches between neighbouring Fabry-Pérot (FP) modes. Clearly a jump over several FP modes can be observed at IDBR = 1.4 A. Data from this device and several other DBR-QCLs indicates that this is due to an interference of the light reflected by the back facet with the light reflected by the DBR. In Fig. 2(b) spectral tuning of 8 cm−1 at constant Tsink = 20°C is depicted using a device emitting at 13.5 µm. The side-mode suppression ratio for all examined DBR-QCLs was noise limited at 30 dB for high pulse repetition frequencies (PRF) (see inset of Fig. 2(a)) and 20 dB for a low PRF of 1 kHz (inset of Fig. 2(b)) which allows to choose a long pulse duration τDBR and to consequently access the entire tuning range. The temperature tuning coefficient dν/dT for the lasers emitting at 10.3 µm (13.5 µm) was found to be −0.077 cm−1/K (−0.057 cm−1/K) which indicates that the temperature offset of the DBR-segment with respect to Tsink at maximum applied IDBR amounts to 200°C (135°C). In order to fill the gaps between mode jumps an adjustment of Tsink or Iphase can be used which allows to shift the position of the mode jumps. Since only electric currents need to be changed to acquire a quasi-continuous tuning range extremely fast switching of the wavelength from one pulse-cycle to another is possible.
Fig. 2 (a) Tuning of a DBR-QCL emitting around 10.3 µm with LDBR = 2 mm, Lgain = 3 mm and a ridge width of 25 µm. Temperature tuning (triangles) between −35°C and 120°C is applied to enhance the tuning range and fill the gap at IDBR = 1.4A. (b) Pure current tuning of a DBR-QCL emitting around 13.5 µm with LDBR = 2 mm, Lgain = 2 mm and a ridge width of 28 µm.
3.2 Electro-optic properties
Electro-optic characteristics for the device emitting at 10.3 µm at constant Tsink = 20°C are shown in Fig. 3(a) . An automated measurement of electro-optic characteristic curves in dependence of IDBR with ΔIDBR = 10 mA was conducted. Values for Ith,gain and the slope efficiency were extracted and found to be periodically modulated as IDBR is increased because of the jumps between FP modes (see Fig. 3(b)). On average, an increase of IDBR leads to an increased slope efficiency and a reduction of Ith,gain because the net gain in the DBR-segment is gradually rising for IDBR <2.2 A. For greater values of IDBR this trend is reversed since high temperatures in the DBR-segment lead to increased absorption and declining gain. For the tuning range and the driving parameters given in Fig. 2(b) the device emitting at 13.5 µm exhibited peak output powers ranging from 26 to 104 mW with slope efficiencies between 124 and 247 mW/A. Threshold current densities Jth,gain = Ith,gain/(W·Lgain) were between 5.1 and 6.1 kA/cm2.
Fig. 3 (a) Characteristic electro-optic curves of the DBR-QCL emitting at 10.3 µm for five equidistant values of IDBR. (b) Slope efficiency and threshold current density plotted over IDBR. The driving parameters are the same as given in the main graph of Fig. 2(a).
In order to compare the performance of DBR-QCLs to 2 mm long DFB-QCLs (with a highly reflective coating on the back facet and the front facet left as-cleaved) of identical width and processed from the same gain materials, devices with Lgain = 2 mm and LDBR = 2 mm were chosen and driven with τDBR = 8 µs, τgain = 50 ns and Δt = 7.8 µs to acquire the full tuning range. The DBR-QCL at 10.3 µm showed a relative slope efficiency between 48% and 90% (depending on IDBR) and a relative threshold current density between 88% and 147%. At 13.5 µm the relative slope efficiency ranged from 46% to 92% and the relative threshold current density from 106% and 127%. The reduction in output power is mainly due to the absorption in the DBR-segment and the lower reflectivity of the back facet.
3.3 Thermal crosstalk
The spectral position of a single cavity mode in a DBR-laser depends on the effective refractive index [15] and therefore on the average temperature Tcavity in the entire lasing cavity of length Lcavity. Lcavity of the DBR-QCL emitting at 13.5 µm can be extracted from the spacing of two neighbouring longitudinal FP modes νn-ν(n + 1) = 0.37 cm−1 in Fig. 2(b). This yields Lcavity = 1/(2·ng·(νn-v(n + 1))) = 3.9 mm using an effective group index ng = 3.47 which was determined from a simple FP QCL of known length. The effective length of the DBR segment LDBR,eff is then given by Lcavity-Lgain. The spectral position of the Bragg wavelength, however, is solely determined by the temperature in the DBR-segment TDBR. In case of no thermal crosstalk (dTcavity/dIDBR)/(dTDBR/dIDBR) should therefore be equal to LDBR,eff/Lcavity = 0.49.
For the presented DBR-QCL emitting at 13.5 µm the temperature of the DBR-segment TDBR(IDBR) was estimated by translating the change in emission wavenumber Δν(ΤDBR) with respect to ν(Τsink) into a temperature rise ΔT(IDBR) = Δν(ΤDBR)·(dν/dΤ)−1. The temperature tuning coefficient dν/dΤ given in section 3.1 can be assumed as constant in the relevant temperature range. The power dissipated in the DBR section is a quadratic function of the drive current, hence the temperature rise can be fitted with a parabola. In Fig. 4(a) the heating of the DBR-segment with increasing IDBR is plotted for different values of τDBR. From that the time resolved temperature evolution for different values of IDBR as depicted in Fig. 4(b) can be extracted. For τDBR>8 µs the heating and also the tuning begins to saturate as the section is approaching a stationary thermal condition. The analysis of the thermal crosstalk was therefore carried out for τDBR = 8 µs: For that purpose the parabola ΔTDBR(IDBR) and the derivative dTDBR(IDBR)/dIDBR (blue straight line) are plotted in Fig. 5(a) . The red straight line depicts the expected behavior of dTcavity(IDBR)/dIDBR = LDBR,eff/Lcavity·d(TDBR(IDBR))/dIDBR in case of negligible thermal crosstalk. Discrete values for ΔTcavity/ΔIDBR can also be extracted from the slope of the individual steps (green squares) which are in good agreement with dTcavity(IDBR)/dIDBR. Therefore no significant heat transfer from the DBR-segment to the gain segment is evident as expected from a thermal simulation which suggests that the heat front originating from the DBR-segment does not reach the gain segment for Δt<20 µs.
Fig. 4 (a) Tuning (continuous lines) and temperature evolution of DBR-segment (dashed lines) vs. IDBR for different values of τDBR. (b) Time resolved evolution of the temperature of the DBR-segment for different values of IDBR.
Fig. 5 (a) Temperature rise of the DBR-segment in dependence of IDBR (continuous black line) determined from tuning via IDBR (dashed line). Current heating rates dTDBR/dIDBR (blue straight line) and dTcavity/dIDBR (red straight line) are plotted as well as the values for ΔTDBR/ΔIDBR (green squares) extracted from the tuning of single longitudinal cavity modes for every step of the staircase. (b) Emission wavenumber vs. Iphase at constant IDBR = 1370 mA and other parameters as given in Fig. 2(b).
In order to estimate the thermal crosstalk caused by the constant current Iphase, the tuning values ΔνDBR/ΔΙphase and Δνcavity/ΔΙphase were determined via the spectral tuning caused by an increase in Iphase as depicted in Fig. 5(b).
For typical maximum values of Iphase of a few 100 mA which is sufficient to shift the position of a mode jump over 0.5 cm−1, ΔΤDBR/ΔΙphase and ΔΤcavity/ΔΙphase were calculated using dν/dT given in section 3.1. Since the spectral position of a cavity mode depends on the average temperature in the entire lasing cavity Tcavity = (TDBR·LDBR,eff + Tgain·Lgain)/Lcavity a value for ΔΤgain/ΔΙphase can be calculated yielding 126 K/A. ΔΤDBR/ΔΙphase and ΔΤgain/ΔΙphase therefore relate like 126:13 resulting in a moderate thermal crosstalk between the laser segments in case of injection of a constant current.
3.4 Intra-pulse tuning
Due to the injected current Igain and the concomitant temperature rise in the gain section the emission wavelength is red-shifted with the chirp-rate (given in cm−1/100 ns) during the gain pulse (intra-pulse tuning) which can be exploited for gas sensing experiments [4,16]. As expected from the tuning behavior shown in Fig. 2 for certain values of IDBR mode-hops can be observed during the intra-pulse tuning. As is evident from Fig. 6 the switching between modes occurs abruptly - which means that two neighbouring cavity modes do not lase simultaneously – and therefore mode-hops can be pushed out of the intra-pulse tuning range via a change of IDBR or Iphase. Consequently single-mode intra-pulse tuning over a spectral range of 0.4 cm−1 (for τgain = 250 ns) is possible anywhere in the accessible tuning range.
Fig. 6 Time dependent signals from the gain current probe (dotted line) and HgCdTe detector (continuous line) are plotted in the left column next to corresponding FTIR spectra for different values of IDBR in the right column. The fringe spacing in the detector signal corresponds to the free spectral range of 0.049 cm−1 of a Ge-etalon which was inserted in the beam path. Arrows indicate the position of mode jumps. Due to the spectrum acquisition time of a few seconds the FTIR spectra exhibit the integrated signal of both modes for IDBR = 2260 and 2280 mA.
An absorption experiment using a 15 cm gas cell filled with 50 mbar C2H4 was conducted in order to investigate the spectral resolution that can be achieved. In Fig. 7(a) the transmission of C2H4 around 10.3 µm according to the HITRAN database [17] is shown. The wavelength range that is accessible with pure DBR-current tuning at constant Tsink of 18°C is marked in light gray.The driving parameters were chosen to be τgain = 160 ns, τDBR = 8.3 µs, Δt = 8.1 µs, Igain = 2.5 A, IDBR = 289 mA, Iphase = 0 mA, PRF = 1 kHz. A magnification of the selected absorption features for the experiment is given in Fig. 7(b). The detector signal (continuous line) in Fig. 7(c) corresponds very well to the theoretical absorption features and allows for an estimate of the spectral resolution of this experiment. For that purpose a magnification of the detector signal is depicted in Fig. 7(d) along with the theoretical transmission curve.
Fig. 7 (a) Absorption features of C2H4 around 10.3 µm taken from the HITRAN database [17]. (b) Magnification of the absorption features chosen for the gas sensing experiment. (c) Time dependent detector signals of a 160 ns laser pulse recorded as single-shots without averaging: The raw pulse shape, a signal modulated with the transmission of an etalon and the signal acquired with a C2H4 gas cell in the beam path is plotted. The fringe spacing in the dashed curve corresponds to the free spectral range of 0.049 cm−1 of a Ge-etalon which was inserted in the beam path and indicates a chirp-rate of 0.19 cm−1/100ns. (d) Magnification of two closely spaced absorption peaks.
Since the peaks can be well separated the resolution is 7.8·10−3 cm−1 or better and currently limited by the sampling rate of the oscilloscope (2 Gs/s) and the bandwidth of the detector (200 MHz). To allow for averaging of the absorption signal low jitter of Δt of less than 1 ns and excellent current stability of IDBR is mandatory. Since our bench-top laser drivers do not exhibit sufficient accuracy for averaging, adapted driver electronics are currently being developed in order to apply single mode DBR-QCL to trace gas detection.
4. Summary
In conclusion, we have demonstrated monolithic QCLs with distributed Bragg reflectors emitting in single mode with a quasi-continuous tuning range of 25 cm−1 at Peltier temperatures. The pulse scheme and tuning behavior were described as well as the electro-optical characteristics in dependence of the current through the DBR-section featuring a minimum threshold current density of 2.2 kA/cm2 and a maximum peak output power of 600 mW at room temperature. Thermal crosstalk between the laser segments was investigated and found to be non-existent for zero phase current and to have only a minor effect even for ample phase currents. Continuous wavelength scanning in intra-pulse mode was used for a gas absorption experiment. The spectral resolution was determined to be better than 7.8·10−3 cm−1 and limited by the bandwidth of the signal acquisition electronics. Fast wavelength scanning based on a modulation of currents rather than temperature should allow for monitoring of dynamic reaction processes using widely spaced absorption lines.
Acknowledgments
We would like to thank Silke Kuhn, Daniela Lummel and Evelyn Wimmer for expert processing as well as the INP Greifswald for providing the gas cell used in the absorption experiment. We gratefully acknowledge support from the BMBF in the framework of the project QUIP (Grant No. 13N9335 and 13N9336).
References and links
1. J. Faist, C. Gmachl, F. Capasso, C. Sirtori, D. L. Sivco, J. N. Baillargeon, and A. Y. Cho, “Distributed feedback quantum cascade lasers,” Appl. Phys. Lett. 70(20), 2670–2672 (1997). [CrossRef]
2. C. Gmachl, J. Faist, J. N. Bailargeon, F. Capasso, C. Sirtori, D. L. Sivco, S. N. G. Chu, and A. Y. Cho, “Complex-coupled quantum cascade distributed-feedback laser,” IEEE Photon. Technol. Lett. 9(8), 1090–1092 (1997). [CrossRef]
3. Q. Y. Lu, Y. Bai, N. Bandyopadhyay, S. Slivken, and M. Razeghi, “2.4 W room temperature continuous wave operation of distributed feedback quantum cascade lasers,” Appl. Phys. Lett. 98(18), 181106 (2011). [CrossRef]
4. P. Fuchs, J. Semmel, J. Friedl, S. Höfling, J. Koeth, L. Worschech, and A. Forchel, “Distributed feedback quantum cascade lasers at 13.8 µm,” Appl. Phys. Lett. 98(21), 211118 (2011). [CrossRef]
5. R. Maulini, M. Beck, J. Faist, and E. Gini, “Broadband tuning of external cavity bound-to-continuum quantum-cascade lasers,” Appl. Phys. Lett. 84(10), 1659–1661 (2004). [CrossRef]
6. A. Hugi, R. Maulini, and J. Faist, “External cavity quantum cascade laser,” Semicond. Sci. Technol. 25(8), 083001 (2010). [CrossRef]
7. Y. Yao, X. Wang, J.-Y. Fan, and C. F. Gmachl, “High performance ‘continuum-to-continuum’ quantum cascade lasers with a broad gain bandwidth of over 400 cm−1,” Appl. Phys. Lett. 97(8), 081115 (2010). [CrossRef]
8. E. Mujagić, C. Schwarzer, Y. Yao, J. Chen, C. Gmachl, and G. Strasser, “Two-dimensional broadband distributed-feedback quantum cascade laser arrays,” Appl. Phys. Lett. 98(14), 141101 (2011). [CrossRef]
9. L. Hvozdara, A. Lugstein, S. Gianordoli, W. Schrenk, G. Strasser, K. Unterrainer, E. Bertagnolli, and E. Gornik, “Self-aligned coupled cavity GaAs/AlGaAs midinfrared quantum-cascade laser,” Appl. Phys. Lett. 77(8), 1077–1079 (2000). [CrossRef]
10. P. Fuchs, J. Seufert, J. Koeth, J. Semmel, S. Höfling, L. Worschech, and A. Forchel, “Widely tunable quantum cascade lasers with coupled cavities for gas detection,” Appl. Phys. Lett. 97(18), 181111 (2010). [CrossRef]
11. Y. Wakayama, S. Iwamoto, and Y. Arakawa, “Switching operation of lasing wavelength in mid-infrared ridge-waveguide quantum cascade lasers coupled with microcylindrical cavity,” Appl. Phys. Lett. 96(17), 171104 (2010). [CrossRef]
12. B. G. Lee, M. A. Belkin, R. Audet, J. MacArthur, L. Diehl, C. Pflügl, F. Capasso, D. C. Oakley, D. Chapman, A. Napoleone, D. Bour, S. Corzine, G. Höfler, and J. Faist, “Widely tunable single-mode quantum cascade laser source for mid-infrared spectroscopy,” Appl. Phys. Lett. 91(23), 231101 (2007). [CrossRef]
13. J. Semmel, L. Nähle, S. Höfling, and A. Forchel, “Edge emitting quantum cascade microlasers on InP with deeply etched one-dimensional photonic crystals,” Appl. Phys. Lett. 91(7), 071104 (2007). [CrossRef]
14. S. Song, S. S. Howard, Z. Liu, A. O. Dirisu, C. F. Gmachl, and C. B. Arnold, “Mode tuning of quantum cascade lasers through optical processing of chalcogenide glass claddings,” Appl. Phys. Lett. 89(4), 041115 (2006). [CrossRef]
15. L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated Circuits (Wiley-Interscience, 1995), Chap. 3.
16. T. Beyer, M. Braun, and A. Lambrecht, “Fast gas spectroscopy using pulsed quantum cascade lasers,” J. Appl. Phys. 93(6), 3158–3160 (2003). [CrossRef]
17. The HITRAN database, http://www.cfa.hitran.com/
