1. Introduction

Flash light sources, which are capable of illumination over a wide field of view (FOV), are indispensable in various applications including light-detection and ranging (LiDAR) for autonomous vehicles and robots [1–5], optical wireless communications [6], and adaptive lighting [7]. In the specific application of non-mechanical flash LiDAR [8–10], a flash light source is utilized for ranging by emitting short pulses of light over a wide FOV, then measuring the time taken for these pulses to be reflected or scattered from objects and detected by a time-of-flight (ToF) camera. So far, flash illumination in such applications has been realized by utilizing semiconductor lasers combined with complicated external optical elements, such as arrayed vertical cavity surface emitting lasers (VCSELs) combined with wide-angle-range diffusers [11,12]; systems combining semiconductor lasers with multiple lenses [13] and/or DOEs (diffractive optical elements) [14,15] have also been reported. Such combinations increase the cost of the light source and complicate the system by requiring fine alignment between the optics and the light source, which are at risk of becoming misaligned due to physical shocks to the device. In addition, when lenses and diffusers are used, the beam profile becomes, in principle, a Lambertian distribution, for which the beam intensity degrades at oblique angles from the surface normal.

In order to solve these problems, it is desirable to develop a semiconductor laser which enables on-chip flash irradiation without optical elements. Doing so requires diffraction functionality – here, the ability to modify the emission direction, divergence angle, and shape of the laser beam – to be embedded directly into the internal structure of the laser device, which is fundamentally difficult using conventional semiconductor lasers such as VCSELs and Fabry-Perot laser diodes. On the other hand, photonic crystal surface-emitting lasers (PCSELs) [16–19], which operate coherently over a large two-dimensional area by virtue of band-edge resonance, show great potential to achieve on-chip diffraction functionality. A PCSEL capable of emitting a laser beam with high quality (or low divergence) and high output power has already been achieved [20,21], which has led to the application of PCSELs in mechanical-scanning-type LiDAR without lenses. Furthermore, on-chip control of the beam emission direction has been demonstrated using modulated photonic crystals [22,23], into which diffraction functionality is embedded via lattice-point modulations; in particular, the emission of high-output-power, high-quality beams in arbitrary directions has been realized using dually modulated photonic-crystal surface-emitting lasers (DM-PCSELs) [24], in which the sizes and the positions of the air holes are modulated simultaneously. An array of multiple DM-PCSELs has even been fabricated to demonstrate lensless non-mechanical beam scanning. However, in this case, the dual modulation was evenly applied to the entire photonic crystal, so the light was diffracted and emitted as a narrow beam in a single direction over the entire photonic-crystal area. Whether the divergence and shape of the beam from a PCSEL can be arbitrarily broadened to achieve flash illumination has yet to be investigated [24].

In this study, we advance the concept of DM-PCSELs to investigate the possibility of achieving wide-FOV flash illumination without external optical elements. First, we propose and design a dually modulated photonic crystal, into which we embed the information of a diffraction vector that gradually varies over the photonic-crystal area, in order to artificially broaden the beam divergence angle. We then experimentally demonstrate broadening of the beam divergence angle (to up to 8°) based on this concept. Finally, by arraying 100 such PCSELs, each with a beam divergence angle of ∼8° and a distinct central emission angle, and driving all of these lasers simultaneously, we achieve on-chip flash illumination with an output power of ∼4 W and a wide FOV of 30° × 30° without using external optics for the first time. This type of on-chip flash light source is expected to contribute to realizing a compact, stable, non-mechanical LiDAR sensing system, and can even be combined with an on-chip beam-scanning light source to achieve systems with even higher functionalities.

2. Design of dually modulated photonic crystals for beam expansion

Before describing the design of dually modulated PCSELs with arbitrary beam divergence angles, we first review the principle of conventional dually modulated PCSELs with fixed emission angles. In a conventional dually modulated PCSEL, light resonating at the M point is diffracted into the light cone by the diffraction vector k, and this light is emitted at arbitrary emission angles $({{\theta_x},{\theta_y}} )$ as shown in Figs. 1(a) and 1(b). If we let the position of each air hole in the unmodulated state be ${\mathbf{r}}_{m,n}^0 = ({ma,na} )$ (where a is the lattice constant and m and n are integers), each air hole is modulated with a phase $\varPsi ({{\mathbf{r}}_{m,n}^0} )$ corresponding to k:

 

Fig. 1. (a) Photonic band structure with a single diffraction vector k introduced via modulation. (b) Schematic diagram of a photonic crystal designed to have a single emission angle over the entire active area. (c) Photonic band structure with a spatially dependent diffraction vector k(x, y) introduced via modulation. (d) Schematic diagram of a photonic crystal designed to have a spatially dependent emission angle. The color gradient represents the change of k(x, y) within the photonic crystal.

Download Full Size | PDF

Note that Eq. (1) satisfies Eq. (4) in the special case where k is independent of r. When solving Eq. (4), care should be taken to prevent destructive interference of the light within the beam. To accomplish this, we have the beam emission angle vary monotonically and continuously (specifically, by having ${\theta _x}$ and ${\theta _y}$ linearly increase along the x and y directions, respectively). By doing so, we obtain the following form of $\varPsi ({\mathbf{r}} )$:

Note that the first-order spatial derivatives ${\rm{d}}{\theta _x}/{\rm{d}}x$ ($= \theta _x^{\prime}$) and ${\rm{d}}{\theta _y}/{\rm{d}}y$ ($= \theta _y^{\prime}$) are finite, but all other first- and higher-order spatial derivatives are zero. Equations (5) and (6) indicate that the emission angle ${\theta _x}$ changes by $({{\theta_{x2}} - {\theta_{x1}}} )$ over a distance of ${L_x}$ in the x direction and that the emission angle ${\theta _y}$ changes by $({{\theta_{y2}} - {\theta_{y1}}} )$ over a distance of ${L_y}$ in the y direction, where ${\theta _{j0}}$, ${\theta _{j1}}$, ${\theta _{j2}}$, and ${L_j}$ ($j \in \{{x,y} \}$) are parameters. Assuming that a circular p-side electrode with a diameter of ${L_{{\rm{electrode}}}}$ defines the size of the active area and that the center of this electrode is the origin of x and y, Eq. (6) can be rewritten as

Here, $({{\rm{\varDelta }}{\theta_x},{\rm{\varDelta }}{\theta_y}} )$ represents the broadening of the beam divergence angle about the central emission angle $({{\theta_{x0}},\;{\theta_{y0}}} )$. Also, note that Eq. (5) diverges when ${\rm{\varDelta }}{\theta _x}$ or ${\rm{\varDelta }}{\theta _y}$ is 0$^\circ $; in this case, the spatial dependence of the emission angle disappears, so Eqs. (1) and (2) should be used instead of Eq. (5).

In this work, we aim to broaden the beam isotopically, so we set ${\rm{\varDelta }}{\theta _x} = {\rm{\varDelta }}{\theta _y} \equiv {\rm{\varDelta }}\theta $. In our forthcoming calculations based on nanoantenna theory [24], we specifically set $({{\theta_{x0}},\;{\theta_{y0}}} )= ({ \pm 10^\circ ,0^\circ } )$, ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $, and ${L_{{\rm{electrode}}}} = 100\;{\rm{\mu m}}$. In these cases, Eq. (7) can be rewritten as

The modulation phase $\varPsi ({\mathbf{r}} )$ is then obtained using Eqs. (5) and (8). We note that the current distribution is assumed to extend laterally outward in all directions from the electrode boundary by 10-20 µm, and that the emission angle continues to change at a rate of per unit distance beyond the electrode boundary.

Next, the position and size of each lattice point were simultaneously modulated in order to facilitate the efficient emission of radiation from all band edges [24]; specifically, the following modulations were applied:

Respectively, ${{\mathbf{\bar r}}_{m,n}}$ and ${\bar S_{m,n}}$ are the position and the area of the (m,n)th air hole after modulation, ${\mathbf{r}}_{m,n}^0$ and $S_{m,n}^0$ are the position and the area of the (m,n)th air hole before modulation, and ${\rm{\varDelta }}{\mathbf{d}}$ and ${\rm{\varDelta S}}$ are the amplitudes of the position and size modulations. The position modulation is applied in the y direction in this study. The modulation phase $\varPsi ({{\mathbf{r}}_{m,n}^0} )$ is found using Eqs. (5) and (8). In our calculations, we used $|{{\rm{\varDelta }}{\mathbf{d}}} |= 0.08a,\;S_{m,n}^0 \equiv {S^0} = 0.08{a^2},{\rm{\;}}$ and ${\rm{\varDelta }}S = 0.03{a^2}$. These variables are not designed to target a specific band edge for oscillation; however, a band edge can be targeted by adjusting the radiation coefficients at band ends A and B via changing ${\rm{\varDelta }}{\mathbf{d}}$, and by adjusting the radiation coefficients at band ends C and D via changing ${\rm{\varDelta S}}$.

The beam pattern of a PCSEL with the above modulations was analyzed using nano-antenna theory [24]. In nano-antenna theory, the electric field radiated from each air hole is represented by a single electric field vector ${{\mathbf{E}}_{{\rm{rad}}}}$ obtained by integrating the in-plane electric-field over the area of the air hole. The magnitude of ${{\mathbf{E}}_{{\rm{rad}}}}$ is proportional to the in-plane electric field at the center of the air hole, which depends on the band edge and the area of the air hole. ${{\mathbf{E}}_{{\rm{rad}}}}$ is given by

Here, $C$ is a proportionality constant, and ${({ - 1} )^{m + n}}$ indicates that the in-plane electric fields at adjacent air holes are 180° out of phase with each other at the M point. The electric field is represented by a sine function at band edge A/B, where the air hole is located at a node of the electric field, and by a cosine function at band edge C/D, where the air hole is located at an anti-node of the electric field. $\xi ({\mathbf{r}} )$ is the in-plane electric-field envelope function determined by the electrodes; in this work, we use a sigmoid envelope function, corresponding to the excitation profile observed in experiments for a circular p-side electrode with a diameter of 100 µm. ${{\mathbf{e}}_{{\rm{rad}}}}$ represents the unit vector of the radiated electric field. In a finite-sized device, the in-plane wavevector is slightly detuned from the M point due to the influence of the boundary conditions. Consequently, the positions of the electric-field (anti-)nodes slightly drift with respect to the air holes over the active area, which causes ${{\mathbf{e}}_{{\rm{rad}}}}$ to become a function of ${\mathbf{r}}_{m,n}^0$. Nevertheless, for the 100-µm-diameter circular electrode considered in this work, this spatial dependence is negligible, so we maintain that ${{\mathbf{e}}_{{\rm{rad}}}}$ is spatially independent; specifically, we fix ${{\mathbf{e}}_{{\rm{rad}}}}$ to be parallel to the x axis for the modes of band edges A and C and to the y axis for the modes of band edges B and D. Also, when the position modulation amplitude is small (${\rm{\varDelta }}{\mathbf{d}} \ll a$), the shift of the position of the radiated electric field vector following modulation can be neglected, so we further set ${{\mathbf{\bar r}}_{m,n}} \to {\mathbf{r}}_{m,n}^0$. As a result, Eq. (10) is simplified to

Equation (11) describes the individual vector at point $({m,n} )$ of an evenly spaced, discrete, grid-like array of points, corresponding to the individual contribution of each air hole to the total radiation. Here, since the radiated electric field vectors are assumed to extend from the center of each air hole, ${{\mathbf{E}}_{{\rm{rad}}}}({\mathbf{r}} )= 0$ when ${\mathbf{r}} \ne {\mathbf{r}}_{m,n}^0$. By performing a Fourier transform of the electric field distribution described by Eq. (11), we obtain the electric field distribution ${{\mathbf{E}}_{{\rm{far}}}}({\mathbf{K}} )$ in the far field, where ${\mathbf{K}}$ represents the wavenumber in the far field.

For the design of the emission angle, we set $({{\theta_{x0}},\;{\theta_{y0}}} )= ({ \pm 10^\circ ,0^\circ } )$ and ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $ as defined in Eq. (8). The intensity distribution of the radiated electric field vector ${{\mathbf{E}}_{{\rm{rad}}}}({\mathbf{r}} )$ at each lattice point of the photonic crystal calculated using Eq. (11) and the electric field ${{\mathbf{E}}_{{\rm{far}}}}({\mathbf{K}} )$ in the far field calculated using Eq. (12) are shown in Fig. 2. In the intensity distribution of the photonic crystal with ${\rm{\varDelta }}\theta = 0^\circ $ [Fig. 2(a)], we can see that there are parallel and evenly spaced interference fringes in the y direction; these interference fringes correspond to an emission angle of $({{\theta_{x0}},\;{\theta_{y0}}} )= ({ \pm 10^\circ ,0^\circ } )$. Since the beam emission angle is fixed over the whole area (${\rm{\varDelta }}\theta = 0^\circ $), the spacing between the interference fringes is constant (in general, we expect the spacing between the fringes to become narrower when the emission angle becomes larger, and vice versa). Meanwhile, the intensity distribution in the far field (i.e., the beam pattern) [Fig. 2(b)] exhibits a beam divergence angle close to the diffraction limit (< 0.8°). Next, for the photonic crystals with ${\rm{\varDelta }}\theta $ values of 4° and 8°, we see in the intensity distributions of the radiated electric field [Figs. 2(c) and 2(e)] that the spacing between the fringes become smaller along the positive x direction, indicating that the x component of the emission angle is broadened about the central angles of ${\pm} 10^\circ $. Meanwhile, the bending of the fringes in the y direction is due to a commensurate broadening of the y component of the emission angle about the central angle of 0°. Also, in the beam patterns [Figs. 2(d) and 2(f))], we see that the beam divergence angles broaden by an amount corresponding to ${\rm{\varDelta }}\theta $. Here, we note that the intensity at the center of the beam is slightly lower than elsewhere, which is due to the sigmoid shape of $\xi ({\mathbf{r}} )$. The precise shape of the beam profile can be controlled via appropriate design of $\xi ({\mathbf{r}} ).$ For example, a Gaussian-shaped beam profile can be achieved by designing a device structure for which $\xi ({\mathbf{r}} )$ is Gaussian.

 

Fig. 2. (a, c, e) Calculated intensity distribution of the radiated electric field of photonic crystals with designed ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $, respectively using Eq. (10). In (a), the evenly spaced vertical interference fringes correspond to an emission angle of (θ_x0, θ_y0) = (±10°, 0°), and in (c, e) the narrowing and bending of these interference fringes correspond to the isotropic broadening of the beam divergence angle by Δθ. (b, d, f) Intensity distribution in the far field with designed ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $, respectively, calculated using Eq. (11). The beam divergence angles are broadened to the designed amounts ${\rm{\varDelta }}\theta $.

Download Full Size | PDF

3. Demonstration of beam broadening with dually modulated photonic crystal lasers

We fabricated dually modulated PCSELs to demonstrate beam broadening. These PCSELs were designed to have emission angles $({{\theta_{x0}},\;{\theta_{y0}}} )$ of $({0^\circ ,0^\circ } )$ or $({ \pm 10^\circ ,0^\circ } )$ and beam broadening angles ${\rm{\varDelta }}\theta $ of $0^\circ $, $4^\circ $, or $8^\circ $. The parameters of the photonic crystal modulations were $|{{\rm{\varDelta }}{\mathbf{d}}} |= 0.08a$, ${S^0}\sim 0.10{a^2}$, and ${\rm{\varDelta }}S\sim 0.028{a^2}$, and the diameter of the electrode was ${L_{{\rm{electrode}}}} = 100\;{\rm{\mu m}}$ as used in the above calculations. The photonic band structures of the fabricated lasers were measured from their angle-resolved spontaneous emission spectra to verify the effect of continuously varying the diffraction vector. To obtain an accurate measurement of the photonic band structure, we measured the emission spectrum of the PCSELs for which $({{\theta_{x0}},\;{\theta_{y0}}} )= ({0^\circ ,0^\circ } )$ and ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $, as shown in Figs. 3(a-c). In the band structure of Fig. 3(a), where ${\rm{\varDelta }}\theta = 0^\circ $, a band edge exists only at the M point, and the resonant light is diffracted only from this point. On the other hand, in the band structure of Fig. 3(b) [Fig. 3(c)], where ${\rm{\varDelta }}\theta = 4^\circ $ ($8^\circ $), the band edge is spread around the M point over an angular range of $4^\circ $ (8$^\circ $), and the resonant light is diffracted throughout this range. These measurements confirm that introducing a spatially-dependent diffraction vector has the effect of broadening the resonant band edge.

 

Fig. 3. (a, b, c) Measured photonic band structures of devices designed with $({{\theta_{x0}},\;{\theta_{y0}}} )= ({0^\circ ,0^\circ } )$ and ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $. It is seen that the introduction of a spatially-dependent diffraction vector has the effect of broadening the resonant band edge by an amount commensurate with ${\rm{\varDelta }}\theta $. (d, e, f) Measured far-field patterns of devices designed with $({{\theta_{x0}},\;{\theta_{y0}}} )= ({ \pm 10^\circ ,0^\circ } )$ and ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $ under pulsed operation. The beam divergence angles were succesfully broadened by around ${\rm{\varDelta }}\theta $ as designed. (g) Measured light-current (L-I) characteristics of devices designed with $({{\theta_{x0}},\;{\theta_{y0}}} )= ({ \pm 10^\circ ,0^\circ } )$ and ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $. The slope efficiencies were confirmed to be 0.3∼0.4 W/A for all ${\rm{\varDelta }}\theta $.

Download Full Size | PDF

Next, we measured the far-field pattern of the PCSELs for which $({{\theta_{x0}},\;{\theta_{y0}}} )= ({ \pm 10^\circ ,0^\circ } )$ and ${\rm{\varDelta }}\theta = 0^\circ ,\;4^\circ ,\;8^\circ $. The beam divergence angle was less than $1^\circ $ when ${\rm{\varDelta }}\theta = 0^\circ $ [Fig. 3(d)], whereas it was approximately $4^\circ $ when ${\rm{\varDelta }}\theta = 4^\circ $ [Fig. 3(e)] and approximately $8^\circ $ when ${\rm{\varDelta }}\theta = 8^\circ $ [Fig. 3(f)]. These measurements confirm that the beams were broadened as designed. We also measured the light-current (L-I) characteristic of each of these lasers. From these L-I characteristics, which are plotted in Fig. 3(g), the slope efficiencies were determined to be 0.3∼0.4 W/A for all ${\rm{\varDelta }}\theta $. These measurements confirm that the beams were successfully broadened as designed without degrading the slope efficiency.

4. Realization of flash-type dually-modulated photonic-crystal laser arrays with wide field of view

Above, we have demonstrated beam broadening of individual PCSELs. Next, we designed and fabricated multiple beam-broadened lasers in an on-chip two-dimensional array to demonstrate flash-type illumination over a wide FOV. The FOV can be flexibly controlled via proper selection of the number of array elements, the beam broadening angle ${\rm{\varDelta }}\theta $ of each element, and the range of center emission angles $({{\theta_{x0}},\;{\theta_{y0}}} )$ among all elements. In this work, we designed a laser array for illumination over a FOV of $30^\circ $ × $30^\circ $ at a central emission angle of $({{\theta_x},\;{\theta_y}} )= ({0^\circ ,25^\circ } )$; specifically, we designed a 10×10 array of one hundred lasers, each of which had a beam broadening angle of ${\rm{\varDelta }}\theta = \;8^\circ $ and a center emission angle $({{\theta_{x0}},\;{\theta_{y0}}} )$ that differed by $3^\circ $ with respect to its neighbors. A schematic diagram of the arrayed device structure is shown in Fig. 4(a). Concerning the electrode structure, ten lines of ten elements each were connected in parallel, and the ten elements of each line were connected in series. This electrode structure was used to ensure that current is injected into each of the one hundred elements as uniformly as possible.

 

Fig. 4. (a) Schematic diagram of the arrayed device structure for flash illumination with any FOV. (b) Circuit diagram of the electrode structure of the arrayed device in which ten lines are connected in parallel and ten lasers are connected in series in each line. (c) Schematic top-view (left) and cross-sectional (right) diagrams of the device structure. The p and n electrodes are connected in series by line electrodes. As illustrated in the right panel, an insulating layer is inserted between the n electrodes and the overlapping line electrodes in order to avoid the formation of a short circuit between these electrodes.

Download Full Size | PDF

The device was fabricated as follows. First, a n-doped GaAs contact layer, an AlGaAs n-cladding layer, an InGaAs/AlGaAs multiple-quantum-well (MQW) layer, an AlGaAs carrier-blocking layer, and a p-GaAs layer were deposited on a semi-insulating GaAs substrate, and then the photonic crystal pattern was etched into the p-GaAs layer by electron-beam lithography and reactive-ion etching. Next, the air holes were embedded and a p-cladding AlGaAs layer was deposited by MOVPE regrowth. Afterward, the electrode structure was fabricated on the p-side of the device to enable current to be injected into every element at once without obstructing the beam to be emitted from the n-side. Mesas were first formed by dry etching down to the n-doped GaAs contact layer, and then grooves were etched into the GaAs substrate to electrically isolate each element. Next, p-type and n-type electrodes were deposited on each element, and an insulating layer was deposited over the entire device surface. Parts of this insulating layer were etched, and then line electrodes were deposited to connect the electrodes of adjacent elements. A schematic cross-sectional diagram of the device structure formed up to this point in the fabrication process is shown in Fig. 4(c). Next, the insulating layer was redeposited to cover the line electrodes, then re-etched for the deposition of Ti/Au pad electrodes. Ti/Au was also deposited over the insulating layer, directly above the elements, to facilitate heat dissipation to the sub-mount. The completed laser chip, whose total area was 3.5 × 3.5 mm², is shown in Figs. 5(a) and 5(b); Fig. 5(a) shows a microscope image of the electrodes on the backside, and Fig. 5(b) shows a microscope image of the AR-coated emitting surface on the front side. The emitting surface was completely free of electrode shadow loss. This laser chip was attached to a sub-mount, then mounted onto a package. A photograph of the mounted chip is shown in Fig. 5(c).

 

Fig. 5. (a) A microscope image of the electrode structure on the backside of the fabricated arrayed dually modulated PCSEL device, consisting of ten lines of ten elements each connected in parallel, and the ten elements of each line connected in series. (b) A microscope image of the AR-coated emitting surface of the fabricated arrayed device. The emitting surface is completely free of electrode shadow loss. (c) Photograph of the fabricated chip mounted on a package. In the center is the laser chip, which is assembled into the package after being mounted on a sub-mount. Multiple wires are used to inject current uniformly into the electrodes.

Download Full Size | PDF

To evaluate the far-field beam pattern of the fabricated dually modulated PCSEL for flash illumination, the device was tilted and one of its two beams was irradiated onto a screen, as shown in Fig. 6(a). Then, an image of the irradiated screen was taken from the opposite side using a CCD camera. This image, shown in Fig. 6(b), confirms that the device is capable of flash illumination over a wide $30^\circ $ × $30^\circ $ FOV. (Note that Fig. 6(b) shows only one of the two point-symmetric patterns.) Judging from the spatially uniform intensity over the illuminated area, current was uniformly injected into all one hundred laser elements, owing to the parallel-connected lines of series-connected lasers. Figure 6(c) shows the measured L-I characteristic of the arrayed device under pulsed operation at room temperature. Since ten elements were connected in series in each line, the slope efficiency of the device was ∼3 W/A, which is around ten times larger than that of a single element, and the maximum measured output power was 4 W. The slope efficiency is expected to be doubled by introducing a backside reflector to the device [21]. Measurements of the lasing spectrum under pulsed operation, shown in Fig. 6(d), confirmed that every element supported laser oscillation in a single mode with approximately the same wavelength.

 

Fig. 6. (a) Schematic diagram of the experimental setup for measuring the far-field beam pattern. One of the flash patterns emitted from the device is positioned to be incident on the center of the screen; this pattern is measured with a CCD camera from behind the screen. (b) Measured far-field beam pattern for wide-area ($30^\circ $×$30^\circ $) flash illumination by arrayed dually modulated PCSELs with a 10-series 10-parallel combinatorial electrode structure. (c) Measured L-I characteristic of the arrayed device. Under pulsed operation at room temperature, the slope efficiency of the device is ∼3 W/A, and the maximum measured output power is 4 W. (d) Measured lasing spectrum of the arrayed device under pulsed operation at room temperature. The 100 arrayed elements supported laser oscillation in a single mode with approximately the same wavelength.

Download Full Size | PDF

On the other hand, due to the 10-series electrode structure, the arrayed device suffered from a high threshold voltage of ∼10 V. To lower this voltage, which is crucial for various applications, we devised an alternative electrode structure consisting of twenty parallel-connected lines of five series-connected elements each. Figure 7(a) shows the circuit diagram of the electrode structure of an arrayed device with a 5-series 20-parallel combinatorial electrode structure. We achieved this by directly wiring together every second element in each of the ten original lines, as shown in the inset of Fig. 7(a). An optical microscope image of the electrode structure on the backside of the fabricated laser is shown in Fig. 7(b). As shown in the inset of Fig. 7(b), the thick line electrode extending horizontally across the center connects the p electrodes of the two elements in parallel, and the two thin line electrodes above and below connect the n electrodes of the two elements in parallel; here, the formation of a short circuit between the thick line electrode and the n electrode is avoided by inserting an insulating layer between these electrodes. The measured current-voltage (I-V) characteristic of this device, shown in Fig. 7(c), confirms that the threshold voltage of the device was successfully lowered from ∼10 V to ∼5 V by virtue of the 5-series circuit. The measured beam pattern shown in Fig. 7(d), which was obtained by the same method as above, confirms that this device, too, is capable of flash illumination over a wide $30^\circ $ × $30^\circ $ FOV. These results show that various serial and parallel combinational electrode structures can be applied to meet system-specific requirements without compromising the performance of the device.

 

Fig. 7. (a) Circuit diagram of the electrode structure of an arrayed device with a 5-series 20-parallel combinatorial electrode structure. The inset shows a magnified view of two elements connected in parallel via line bypasses. These pairs of parallelly connected electrodes are connected in series and parallel, resulting in an overall 5-series 20-parallel electrode configuration. (b) A microscope image of the electrode structure on the backside of the arrayed device. The inset shows a thick line electrode extending horizontally across the center, which connects the p electrodes of the two elements in parallel, as well as two thin line electrodes above and below, which connect the n electrodes of the two elements in parallel. (c) Measured current-voltage (I-V) characteristic of 5-series 20-parallel and 10-series 10-parallel arrayed devices. The threshold voltage of the 10-series 10-parallel arrayed device was ∼10 V (around ten times that of a single element) and that of 5-series 20-parallel and 10-series 10-parallel arrayed devices was ∼5 V (around five times that of a single element). (d) Measured flash-beam pattern using the 5-series 20-parallel arrayed device.

Download Full Size | PDF

In general, the outer edge of the emitted flash pattern is blurred, which may be an issue in certain applications. This issue may be circumvented by simply designing the system to irradiate a wider FOV than is required. Alternatively, we expect that this issue can be eliminated by designing the modulation phase of each lattice point of the dually modulated PCSEL based on the inverse Fourier transformation of a far-field beam pattern with sharp edges; this design technique [22] is expected to be applicable even to the realization of various other illumination patterns.

5. Conclusion

We have proposed a dually modulated PCSEL that can emit an isotropically broadened beam in an arbitrary direction via the introduction of spatially-dependent diffraction wavevectors in the plane of the photonic crystal. Beams with divergence angles closely approximating their designed values were obtained in the calculations. Informed by these calculations, we fabricated dually modulated PCSELs that emit beams with divergence angles of up to $8^\circ $. Measurements of the photonic band structure showed that the lasing band edge spread out from the M point across angular ranges commensurate with the designed divergence angles, and measurements of the far-field beam patterns confirmed that the emitted beam divergence angles closely approximated their designed values. Next, we arrayed one hundred laser elements, each with a beam divergence angle of $8^\circ $ and a unique beam emission angle, in a 10-series 10-parallel configuration, with which we achieved flash illumination over a wide FOV of $30^\circ $ ×$30^\circ $. Under pulsed operation, this arrayed device was measured to have a slope efficiency of 3 W/A, an output power of 4 W, and a threshold voltage of ∼10 V. In the future, the slope efficiency is expected to be doubled by introducing a backside reflector to the device. In order to reduce the threshold voltage, we devised an alternative, 5-series 20-parallel electrode structure, with which we realized an arrayed device with a lower threshold voltage of ∼5 V and with no degradation of the device performance. The performance of the devices in this paper eclipse those of conventional flash light sources in the respect that the present devices are single-chip flash-type light sources that do not require external optical elements such as lenses, DOEs, and diffusers, and they can be configured to realize flash illumination over an arbitrarily wide FOV. The emitted flash pattern had blurred edges, which we expect can be eliminated in the future by designing the modulation phase of each lattice point of the dually modulated PCSELs based on the inverse Fourier transformation of a far-field pattern with sharp edges. As compact, low-cost, alignment-free light sources, our devices are expected to find use in various applications, including not only flash LiDAR, but also sensing, optical wireless communications, and adaptive illumination.