Electronic tuning of integrated blue-violet GaN tunable coupled-cavity laser

Electronic tuning of integrated blue-violet GaN tunable coupled-cavity laser O. Guziy,1,a S. Grzanka,2,3 M. Leszczyński,2,3 P. Perlin,2,3 M. Schemmann,4 and H. W. M. Salemink1 1Kavli Institute of Nanoscience, Delft University of Technology, Lorentzweg 1, 2628 CJ Delft, The Netherlands 2Institute of High Pressure Physics, Sokolowska 29/37, 01-142 Warsaw, Poland 3TopGaN Ltd., Sokolowska 29/37, 01-142 Warsaw, Poland 4FOCE Technology Intl BV, The Netherlands


I. INTRODUCTION
Tunable blue-violet laser diodes are of increasing interest for a wide range of applications, including optical data storage, chemical or biological hazard detection, laser-induced fluorescence spectroscopy, photobiology and medical diagnostics. 1,2 urrently, nonlinear optical conversion of tunable infrared lasers is widely used to obtain tunable blue lasers.Optical parametric oscillation 3 or sum-frequency mixing of dye lasers 4 and Ti-sapphire lasers 5 are examples of this approach.Although external-cavity tunable semiconductor lasers have been reported for the blue range, 6,7 their main disadvantage for applications is that they require precise mechanics and alignment.The complexity of the approaches mentioned above leads to bulky and expensive systems, whereas low cost and portability are desirable features for most applications. 8By introducing the integrated coupled-cavity design to GaN semiconductor lasers, one can achieve a practical tuning range and higher reliability while keeping device size, cost and complexity low.GaN-coupled cavity lasers open the way to inexpensive blue-violet tunable lasers because they lend themselves to mass-production fabrication techniques.
Coupled-cavity lasers consist of two Fabry-Perot cavities separated by a sub-wavelength air gap (see Fig. 1).The coupled-cavity concept can be used to tune the laser emission over the wavelength range where substantial gain is available. 9In this paper we present a multimode index-guided InGaN laser.

II. FABRICATION
Lasers were fabricated from an AlGaN/GaN/InGaN multiple quantum well laser structure emitting at 409 nm and grown on bulk gallium-nitride crystals obtained with the high-nitrogenpressure solution method.The GaN substrate was grown at very high nitrogen pressure and therefore has a low dislocation density (below 10 5 cm -2 ). 10 The laser structure ( The laser ridges are defined by optical lithography and Cl 2 /Ar inductively coupled plasma etching.The sidewalls of the ridges were passivated with 200 nm of SiO 2 and an opening was etched into the top of the ridge to provide contact.Thin Ni/Au electron-beam-evaporated contacts were patterned by optical lithography and liftoff.Subsequently, 3-μm-thick gold contacts were electroplated on top of the evaporated contacts, providing both current injection and heat removal from the laser ridge.After thinning the substrate, a Ni/Au contact was evaporated on the n-type side of the substrate.The wafer was cleaved into dual-section lasers.A 26-μm-wide gap was left between the two contacts to provide electrical insulation of each section and to facilitate milling of a 340-nm-wide and 1.2-μm-deep slot in the laser ridge by a focused ion beam (FIB).For this purpose a dual-beam system FEI Strata DB 235 with Ga+ ions at 30 kV acceleration voltage and 50 nA beam current was used.Milling of the air slot resulted in long front and short back Fabry-Perot cavities with lengths of 701 and 689 μm, respectively.High-reflection coatings were applied on both front and back cleaved facets.The intensity reflection coefficients are estimated to be R front = R back = 90%.

III. EXPERIMENTAL CHARACTERIZATION
GaN lasers were mounted on a copper heat sink onto a thermoelectric cooler with temperature controller.Optical spectra were measured as a function of currents in the short and long cavities.For this purpose a computer-controlled, dual-channel, pulse-driving unit was used to drive the laser.To minimize thermal heating of the device at high currents, 200-ns pulses with a repetition rate of 50 kHz (duty cycle of 1%) were used.Two calibrated coils were placed around the contact wires to monitor the current with minimal distortion to the circuit.Laser emission was collected from the front facet and collimated by an AR-coated aspheric lens with a focal length of 2.8 mm (Newport 5721-H-A).Spectral measurements were taken by an Ocean Optics HR4000 spectrometer with 0.045-nm resolution.All measurements discussed in this paper were taken with the same laser device.

IV. SIMULATIONS
Semiconductor lasers can be tuned by varying the laser driving current.Changing the refractive index by varying the current shifts the lasing wavelength, a process known as the electronic tuning of semiconductor lasers.The main physical mechanisms that link changes in the current to changes in the refractive index are (i) the free-carrier plasma effect, (ii) the quantum-confined Stark effect (QCSE) as well as (iii) the temperature 11 and (iv) gain dependence of the refractive index.In this work, the temperature was kept constant (iii) and the QCSE (ii) was not included into our simulation model.Refractive index changes due to carrier injection into the semiconductor can be estimated by: 12 where λ is the photon wavelength, m e and m h are the effective masses of the injected electrons (0.18 m 0 ) and holes (0.95 m 0 ), 13 respectively, n is the refractive index of the semiconductor, E is the photon energy in eV, and N and P are the concentration of electrons and holes, respectively, in cm -3 .Note that the refractive index decreases with increasing current.As the refractive index reduction is inhomogeneous, we estimated the effective refractive index changes by the formula: 14 where E(x) is the amplitude of the electric field in the vertical direction.The free-carrier concentration was simulated using the commercial software SiLENSe (Fig. 2(a)).McConville et al. 15 and Asryan et al. 16 suggest that non-pinning of the free-carrier concentration above the threshold current might occur due to violations of charge neutrality in the active region.Therefore we have made calculations for both cases, including pinning and nonpinning of the free-carrier concentration.For an increase in current from 540 to 720 mA, the effective refractive index change n eff of −1.9 • 10 -4 for the pinning case and −3.6 • 10 -4 for the non-pinning case (Fig. 2(b)) was estimated using Eqs.( 1) and ( 2).These simulations show that, even in the pinning case, n eff will change due to an increased free-carrier concentration outside the active region, mainly in the electron blocking and waveguide layer (Fig. 2(b)).The gain in the active layers of the laser structure can be represented by an imaginary part of the refractive index.The change in the real part of the refractive index due to the change in the imaginary part is connected by the Kramers-Kronig relationship: 17 As can be seen in Fig. 3 the refractive index change due to gain does not exceed +4 • 10 -3 for the current increase from 540 to 720 mA.Taking into account that refractive index change occurs only in the active layers with optical confinement factor of 2.1%, the effective refractive index change is estimated by n eff = • n = +0.8• 10 -4 .In combination with the free-carrier effect, this results in the total calculated effective refractive index change of −1.1 • 10 -4 for the free-carrier pinning case and −2.8 • 10 -4 for the non-pinning case.The change of the effective refractive index of the laser will translate into a change in the operation wavelength, which for a single-cavity Fabry-Perot laser is defined by: 18 where λ 0 is the center lasing wavelength, n eff is the change in the effective refractive index and n g,eff = 3.388 is the group effective refractive index.However, the continuous tuning range is limited by the possible effective index change.For the refractive index changes mentioned above, a tuning range of only 0.03 nm is available according to Eq. ( 4).However, by using the Vernier tuning mechanism in the coupled cavity laser, the tuning range can be increased significantly compared to a conventional single-cavity laser. 19With Vernier tuning the lasing mode jumps from one longitudinal AIP Advances 2, 032130 (2012) mode to another.It is worth mentioning that, if the refractive index decreases in the short cavity, the lasing mode shifts to a longer wavelength (see Fig. 4).At the same time, if the refractive index decreases in the long cavity, the lasing mode shifts in the opposite direction to a shorter wavelength.

V. EXPERIMENTAL RESULTS
The spectral tuning characteristics of the laser are shown in Fig. 5.The experimental wavelength map in Fig. 5(c) shows the lasing wavelength as a function of the short and long cavity currents.The sections of this map (see Figs. 5(a) and 5(b)) visualize the spectral behavior when one of the currents is kept constant while another is increased.The experimental wavelength map is in good qualitative agreement with predictions obtained with the Vernier theory.
An increase in current decreases the refractive index as a result of the free-carrier plasma effect.A decrease in the refractive index in the long cavity causes a blue shift (Fig. 5(a)), whereas the same decrease in the short cavity causes a red shift (Fig. 5(b)) of the laser emission.Both effects combined result in a diagonally striped wavelength map (Fig. 5(c)).The main mechanism of the tuning is Vernier effect, meaning Fabry-Perot resonances coincidence in the short and long cavities.Multiple neighboring resonances can lase at the same time taking into account mode spacing of 35 pm.Small refractive index variations in one of the cavity shifts its Fabry-Perot resonances and makes other Fabry-Perot resonances to coincide, thus providing tuning. 18If resonances in one of the cavities is shifted more than spacing between the resonances in another cavity the jump in tuning will happen (Fig. 5(a)) and cycle repeat itself slightly shifting in wavelength.The laser threshold was found to be around 650 mA per cavity, which corresponds to a current density of 11.9 kA/cm 2 .This value is relatively high compared to that of state-of-the-art InGaN/GaN lasers. 20We believe this can be improved by further optimizing the laser diode efficiency as well as the sidewall roughness and angle of the air slot fabricated by FIB.Laser emission is multimode, which is expected for 7.8-μm-wide laser ridges with an average full width at half maximum (FWHM) of 0.3 nm.

FIG. 2 .
FIG. 2. (a) Simulation of the difference in electron density N between 540 and 720 mA injection currents (current density 9.9 and 13.1 kA/cm 2 , respectively).Free-carrier pinning above the threshold is indicated by the dotted line.(b) Effective refractive index change with free-carrier pinning (circles) and non-pinning (triangles) as a function of injected current.

FIG. 3 .
FIG. 3. (a) Simulated gain curves and (b) calculation of the refractive index change due to the gain difference for minimum (540 mA) and maximum (720 mA) currents used during tuning measurements.

FIG. 5 .
FIG. 5. Experimental lasing spectra (color) as a function of increasing current in the (5 a) long and (5 b) short cavity.The Fabry-Perot resonances, that experience low optical losses (shown in black) were simulated: n g .=3.388 and refractive index decrease due to carrier injection according to free carrier pinning case (Fig. 2(b)) being from -(35.0 to 36.0) • 10 -4 for the long cavity scan(a) and from -(34.5 to 36.0) 10 -4 for the short cavity scan (b).All spectra shifted vertically for clarity.(c) Experimental and (d) simulated wavelength map.Color represents the lasing wavelength as a function of current in the long cavity (horizontal axis) and short cavity (vertical axis).

Table I )
consists of 360 and FIG. 1. Fabry-Perot coupled-cavity laser with sub-wavelength air gap can be tuned by varying currents in the short and long cavities.

TABLE I .
Laser diode structure details.