Compact tunable lasers are an important enabling technology for a broad range of applications in optical communications and integrated optical data processing. Photonic crystals provide an ideal platform for implementing such lasers due to their ability to integrate a large number of optical components in a small chip-sized device [1–3]. These structures can be seeded by active materials such as quantum wells that offer high optical gain at room temperature [4] and quantum dots which exhibit high carrier confinement, low transparency carrier density, and small nonradiative decay rates [5–7]. The engineering of photonic crystal (PhC) cavities with high quality factors (Q) and small mode volumes seeded with a variety of active materials has resulted in the development of low threshold lasers with high emission efficiencies [8–17]. In addition, the strong optical confinement of photonic crystals enables lasing with a single QD emitter [18].

One of the primary limitations of photonic crystal lasers is that they are very difficult to tune locally after fabrication. Such local tuning is essential for integrated optical structures to ensure that individual optical components are resonantly excited. In addition, local tunability could enable the engineering of reconfigurable optical devices whose functionality can be modified post-fabrication. Photonic crystal structures can be tuned via temperature [19], gas deposition methods [20], but these methods are non-local and will tune all optical structures on the fabricated device simultaneously. Improved temperature tuning methods based on local heating pads have been demonstrated [21] and provide more localization but still require large spacing between optical components. Other techniques based on chalcogenide glass film deposition provide highly localized tunability but are not reversible [22]. Electrical tuning of photonic crystal laser with nematic liquid crystal infiltration has also been demonstrated [23]. More recently a novel laser based on a tapered fiber evanescently coupled to a photonic crystal nanobeam cavity has been shown to be reversibly tunable by up to 7 nm [13]. This technique relies on locally positioning a tapered fiber on a photonic crystal structure, making the extension to tuning of multiple closely packed devices challenging.

Recently, an alternate method for local reversible tuning of photonic crystal devices has been demonstrated using photochromic thin-films [24]. These films exhibit a reversible change in their index of refraction when irradiated by UV and visible light, enabling a highly controllable all-optical tuning of the cavity mode at near-infrared (NIR) wavelengths. By depositing these films on a photonic crystal cavity coupled to a low density of QD spontaneous emitters, reversible cavity-frequency tuning was observed through the fluorescence of the device. Here, we demonstrate a method for using these photochromic thin films to create a reversibly tunable photonic crystal laser. The laser consists of a cavity coupled to an active high gain medium composed of three high density quantum dot layers embedded in the device. The tuning method we present is both local and reversible, enabling the tuning of multiple closely spaced lasers on an integrated optical chip. We show that by controlling film thickness we can maintain a high bare cavity Q to support lasing while simultaneously achieving a sufficiently large modification of the effective index to enable a reversible tuning range of up to 2.68 nm.

A schematic of the device used to realize the tunable laser is shown in Fig. 1a . The initial wafer was comprised of a 160-nm gallium arsenide (GaAs) membrane with three layers of InAs QDs grown at the center (corresponding to a QD density of ~500 µm⁻²), on a 1-µm thick sacrificial layer of aluminum gallium arsenide (Al_0.78Ga_0.22As). Photonic crystals were defined on the GaAs membrane using electron-beam lithography and chlorine-based inductively coupled plasma dry etching, followed by a selective wet etch to remove the sacrificial AlGaAs layer, resulting in a free-standing GaAs membrane. The cavity design used in this experiment was a three-hole linear defect (L3) cavity with three-hole tuning [25] as shown in the fabricated device in Fig. 1b. The parameter a was set to 240 nm and the hole diameter was varied between 130 nm and 160 nm to achieve cavity resonances across the QD gain bandwidth. The three holes at the edge of the cavity, labeled A, B, and C in the figure, were shifted by 0.176a, 0.024a and 0.176a respectively.

 

Fig. 1 a) Schematic showing the cross section of photonic crystal cavity laser with 3 QD layers embedded at the center of the GaAs slab. After fabrication, the photochromic thin-film is spun on the surface. b) SEM image of cavity with side holes A, B, C shifted. Scale bar: 1μm. c) Cavity emission spectrum of a typical device at 80 K recorded for increasing excitation powers using the 780 nm pump laser.

Download Full Size | PDF

After fabricating the devices, the photochromic material was prepared as outlined in Ref [23]. by mixing 5 wt% 1,3,3-Trimethylindolinonaphthospirooxazine (TCI America) and 0.5 wt % 950 PMMA A4 dissolved in anisol. The polymer mixture was spun on the sample at a spinning rate of 3500 RPM, resulting in a film of 50 nm thickness. This thickness was found to be small enough to minimally affect the cavity quality factors while providing a sufficiently high index change to reversibly tune the resonances of the nanocavities. For device characterization, the fabricated structures were placed in a continuous-flow liquid He cryostat and cooled to a temperature ranging between 20K and 80 K. In order to observe lasing, the QDs in the cavity region were excited above-band by a continuous wave titanium sapphire laser tuned to 780 nm wavelength. Emission was collected by a confocal microscope setup using a 0.7 NA objective lens and measured by a grating spectrometer with a wavelength resolution of 0.02 nm.

Upon exciting the cavity with the 780 nm pump laser, a bright narrowband emission is observed from the cavity, as shown in Fig. 1c. This figure shows several representative spectra for the cavity emission with increasing pump powers at 80K. As the pumping power is increased, there is a visible change in the output emission spectrum from a broad emission to a sharp narrowband lasing emission. The small shift of 0.3 nm in the cavity mode emission with increasing power is due to thermal effects caused by above band pumping [12].

In order to verify that the bright emission from the cavity mode is due to lasing, we investigate both the output power and linewidth of the cavity emission as a function of pump power. The measurement results are shown in Fig. 2 . Figure 2a plots the cavity output power (red circles) and linewidth (green diamonds) as a function of input pump power at 20 K temperature where the QD linewidth is minimally perturbed by phonon broadening. The cavity output power curve, commonly referred to as the light-in light-out (L-L) curve, exhibits a clear threshold behavior in that the emitted light power rapidly increases when the pump power exceeds a critical value. Well above threshold the output power is linearly increasing with pump power, as expected from standard laser theory. To estimate the pumping threshold we extrapolate the linear region of the L-L curve to determine the x-intercept. Using this method, we estimate the lasing threshold to be 80µW at 20K.

 

Fig. 2 (a) Laser output intensity (red circles) and linewidth (green diamonds) as function of input power at 20K. The blue line represent the theoretical fit to the cavity intensity using Eq. (1). (b) Cavity resonance as a function of input power at 20K (c) Laser output intensity (red circles) and linewidth (green diamonds) as function of input power at 80K. The blue line represents the theoretical fit to the cavity intensity using Eq. (1). (d) Cavity resonance as a function of input power at 80K

Download Full Size | PDF

An important signature of lasing is the dependence of linewidth of the laser emission on the pump power. In a microcavity laser, the linewidth is expected to initially decrease due to absorption saturation, then increase near threshold due to gain-refractive index coupling, and finally decrease again due to onset of stimulated emission [17,26–28]. For microcavity lasers with high spontaneous emission coupling factors the linewidth broadening near threshold is small and is usually observed as a plateau rather than an increase [29,30]. The emission linewidth of the laser, shown in Fig. 2a, is calculated by fitting the spectrum at each pumping power to a Lorentzian. The linewidth exhibits an initial drop from the absorption limited linewidth of 0.73 nm (Q = 1300) to a cavity linewidth of 0.31 nm followed by a plateau right at threshold. From the linewidths at the threshold power we estimate the bare cavity Q to be 3100, which corresponds to a bare cavity decay rate of κ = 0.31nm. The experimentally measured bare cavity Q is limited by a combination of fabrication errors and degradation in Q due to the photochromic film. The linewidth then continues to decrease beyond the threshold region, as expected from standard microcavity laser theory.

An important figure of merit for a microcavity laser is the spontaneous emission coupling factor β, defined as the percentage of spontaneous emission that couples to the lasing mode [31]. This parameter can be extracted from the L-L curve. Using a simple rate equation model, the relation between the input power $P_{in}$ and output power $P_{out}$is given by [32]:

In Fig. 2b we plot the resonance wavelength of the cavity mode as a function of pump power. As the pump power is increased, the cavity resonance blue shifts as threshold is approached, then reaches a stable value which is shifted by 0.25 nm from the initial low pump power regime. This blue shift may be attributed to the change in refractive index of the cavity medium due to the injection of free carriers [33].

Figure 2c shows the L-L curve, linewidth, and cavity center wavelength for a different device that was maintained at 80 K. At this temperature the lasing threshold is measured to be 920 μW, which is higher than the threshold for the device measured at 20 K. The higher lasing threshold is expected due to increased dephasing and non-radiative decay of the QDs [34].The linewidth of the cavity decreases from its absorption limited initial value of 0.5 nm to 0.21 nm (Q ~4500), followed by a plateau at threshold. The linewidth then continues to decrease beyond the threshold and reaches its minimal value of 0.1 nm at a pump power of 2.75mw. At pumping powers beyond 2.75mw the linewidth begins to increase again, and at the same point the center wavelength of the cavity begins to red-shift. The increase in linewidth and simultaneous red-shift of the cavity wavelength is attributed to cavity heating and has been observed in previous works [13,35]. In contrast, the device measured at 20 K exhibited no such heating due to the lower threshold which enabled us to pump it at lower pump powers.

The tunability of our photonic crystal laser is enabled by the changing index of refraction of the photochromic thin film with UV or visible light irradiation [36]. The properties of the photochromic film have been reported in previous work [24]. We investigate the tunability of the laser by first exposing the cavity structure to 375 nm UV light generated from the second harmonic of a pulsed Ti:Sapphire laser. The cavity was irradiated by a UV intensity of 3 kW/cm² for a period of 50 s, while the emission spectrum was monitored by the spectrometer. The intensity of the UV light was set to be sufficiently low so that the laser was tuned slowly over its entire range during the 50 s exposure (the shifting rate can be significantly increased by increasing the UV intensity [37]).The spectrometer acquired and saved the entire emission spectrum every 500 ms during the exposure period. Figure 3a shows several snapshots of the laser emission spectrum from the cavity taken at different times during this exposure. UV irradiation at the cavity increased the index of refraction resulting in a continuous red shift over the 50 s exposure period. At any time during the tuning process, blocking the UV laser maintained the laser emission at that wavelength.

 

Fig. 3 (a) Cavity emission spectra (solid circles) recorded as a function of photochromic tuning from initial resonance using UV radiation, with Lorentzian fits (solid lines). (b) Linewidth (blue circles) and intensity (green squares) of the photochromic laser as a function of tuning from initial resonance, derived from the same scan as (a).

Download Full Size | PDF

Each spectrum acquired over the 50 s tuning window was fit to a Lorentzian in order to extract the linewidth, center frequency and intensity. The Lorentzian fits for the different snapshots shown in Fig. 3a are plotted as solid lines. Fig, 3b plots both the linewidth and integrated lasing emission intensity as a function of emission wavelength shift over a 2.5 nm tuning range. We observe that the power output of the laser is stable over the photochromic tuning process with fluctuations of less than 10%. The cavity mode linewidth also stays near the threshold value of 0.10 during the tuning process. The variations in the linewidth and intensity for the first few steps of photochromic tuning are likely caused by slight drifts in the cavity position due to vibrations in the optical setup. Because our pump laser was highly focused, drift in the cavity caused it to be pumped less hard resulting in an increase in linewidth and a simultaneous decrease in output intensity. In general, the lasing mode spectrum was found to be highly sensitive to the excitation spot of the 780 nm laser.

Having demonstrated the ability to continuously red-shift the cavity resonance, we next investigate the reversibility of the tuning processes. Reversibility is accomplished by irradiating the photonic crystal cavity by green light from a frequency-doubled Nd:YAG laser with an emission wavelength of 532 nm and at an intensity of 3 kW/cm². In Fig. 4a and b , we show the reversibility of the photochromic tuning process at 20 K and 80 K. The cavity mode spectrum is shown at 6 (8) second- intervals at 20 (80) K while the cavity was exposed to UV or green radiation. Diamonds represent regions when the UV was turned on while the green laser was turned off. Circles represent regions when the green laser was turned on while the UV laser was turned off. After the initial UV exposure, the cavity mode red-shifted with a maximum red-tuning range of 1 nm at 20K and 3.4 nm at 80K. The tuning range at 20K is lower than the tuning range at 80K due to the lower conversion efficiency from the spiropyran to the merocyanine states at low temperatures [38,39]. After UV exposure the green laser was turned on in order to blue-shift the cavity mode. At the end of the first green exposure, we observe that the cavity shifted back for an overall detuning of −0.23 nm ( + 0.72 nm) from the initial resonance at 20K (80K). This shift corresponds to a maximum reversible tuning range of 1.23 nm for 20 K operation and 2.68 nm for 80 K operation.

 

Fig. 4 Tunability of the photonic crystal quantum dot laser at (a) 20 K and (b) 80K. Regions corresponding to UV exposure are shown by diamonds. Green exposure is shown by gray circles. The average pump intensity for UV and green are 3kW/cm² except for the region shown by arrows where 0.5 kW/cm² was used.

Download Full Size | PDF

One limitation of spiropyran-based photochromic films is that the reversibility fatigues over repeated cycles, reducing the range over which the device can be tuned. This fatigue can be readily observed both at 20 K and 80 K through a decrease in the tunability with successive exposures. After four consecutive cycles of red and blue shifting, the reversible tuning range, defined as the wavelength range over which the laser is reversibly tunable in each UV-green cycle, is reduced from its maximum value of 1.23 nm to 0.38 nm at 20 K, while at 80 K it is reduced from 2.68 nm to 0.83 nm. Both cases represent a 70% reduction of the maximum tuning range on the fourth cycle. Once the film has fatigued, it can be removed with acetone and a new film can be spun without destroying the device itself.

In conclusion, we have demonstrated a locally and reversibly tunable near-infrared photonic crystal laser with high beta factors (0.14-0.41). Tunability was achieved using a photochromic thin-film that was spun onto the sample surface. The resulting bare cavity Qs were measured to be as high as 4500, while lasing threshold was achieved at pump powers of 80μW at 20 K and 920μW at 80K. The reversible tuning range of the device was measured to be 1.23 nm at 20 K and 2.68 nm at 80 K, while the maximum tuning range was measured to be 3.4 nm. The photochromic tuning process was found to maintain a stable output power emission linewidth over the entire tuning range of the device. Larger tuning could be potentially achieved at higher temperatures using a different gain medium such as quantum wells that can support room temperature operation. Further device improvements in reversibility and tuning range could be attained by using other photochromic materials that exhibit better reversibility over hundreds of tuning cycles [40,41]. Furthermore, the method we demonstrate for reversible tuning is not restricted to photonic crystals, and could be applied to a broad range of other photonic structures such as microdisks and distributed feedback structures in a straightforward way. The combination of photochromics with integrated devices could open up the possibility for integration of tunable lasers, filters, beamsplitters, and other optical components to create highly reconfigurable photonic systems, with applications in optical communications and optical computation.