Teaching Light New Tricks

Photonic crystals are materials in which the refractive index changes periodically, creating a bandgap at optical frequencies. By using photonic crystals and artificially introduced defect states in photonic crystals, scientists and engineers can control spontaneous emission,¹ bend light very sharply,² trap photons,³ and make devices that provide other important advances in optical technology.

Photonic crystals can be thought of as extensions of diffraction gratings or as extensions of the naturally occurring crystals used in x-ray crystallography. Gratings diffract light of wavelengths related to the grating period; in effect, they act as wavelength-selective mirrors. Distributed Bragg gratings that act as cavity mirrors in vertical-cavity surface-emitting lasers (VCSELs) are one application of this effect. Imagine a 3-D, rather than linear, optical grating structure. It would be a periodic structure that diffracts light of specific wavelengths and does not allow that light to propagate through the structure. This is a 3-D photonic crystal.

To put it another way, consider crystal diffraction. X-rays exit the periodic structure of a crystal at a specific Bragg angle, creating diffraction patterns that provide clues about the structure. If the structure could be made larger than the atomic arrangements in naturally occurring crystals so that the period was submicron, then instead of diffracting x-rays, the material could be made to diffract light with longer wavelengths. This is the principle behind photonic crystals.

Using microfabrication techniques, various groups of researchers have made photonic crystals that block electromagnetic radiation ranging from millimeter waves to the near infrared (IR) wavelengths used for telecommunications. These crystals have varying dielectric constants. Just as electrons in a semiconductor organize into bands of various energies, light waves propagating in a structure with a periodically modulated dielectric constant organize into photonic bands. If the contrast ratio between the dielectric constants is high enough, then these photonic bands are separated by photonic gaps in which propagating states are forbidden. Researchers have made photonic crystals from a variety of materials such as gallium arsenide (GaAs), indium phosphide (InP), silicon, silicon germanium (SiGe), and so on.

Research has focused on both 3-D and 2-D photonic crystals. As the name suggests, in 3-D photonic crystals, the structural variation occurs in three dimensions, and the light is controlled in a volume. In 2-D, the variation occurs in two dimensions. These "slab" photonic crystals only forbid transmission in a plane, which, nevertheless, can provide useful devices, as described below.

Figure 1. One goal of photonic crystals: a compact optical chip is built into a photonic crystal by introducing appropriate defects. Photonic crystal technology enables arrays of nano-ampere lasers with different oscillation frequencies, an optical modulator, wavelength selectors, very sharp bend waveguides, and surface output couplers.

One of the goals of photonic crystals is to create ultrasmall optical and optoelectronic integrated circuits (see figure 1).⁴  Such devices integrate nano-ampere laser arrays with different oscillation frequencies, waveguides that incorporate very sharp bends, optical modulators, wavelength selectors, and so on, all in an area less than 100 x 100 µm. The photonic crystal structure confines the light tightly within the components. The optical devices are created by introducing appropriate artificial defects and light emitters in the crystal. To realize such integrated circuits with photonic crystals, designers must construct a 3-D photonic crystal with a complete photonic band gap in optical wavelength region and introduce an arbitrary defect and/or light-emitting element at an arbitrary position. The device must be made using an electronically conductive crystal that is desirable for the actual device application.

Keeping these requirements in mind, our group at Kyoto University developed a method to construct a 3-D crystal with a woodpile (aka diamond) structure based on wafer-fusion of III-V semiconductor stripes using a laser-beam diffraction pattern to aid alignment, which satisfied all of the requirements described above.^5,6

Figure 2. An SEM image of a 3-D photonic crystal at optical communication wavelengths. The structure is perfectly regular over a wide area. The regularity, which is vital for photonic crystals, extends well below the 700-nm stripe widths in the section with increased magnification.

Using metal organic chemical vapor deposition (MOCVD), we produced a crystal with a stripe period of 0.70 µm, a stripe width of 0.19 µm, and a stripe depth of 0.20 µm, which yielded a structure with a bandgap at optical communication wavelengths (see figure 2).To build the photonic crystal structure, we first stacked a pair of striped wafers in a face-to-face configuration and fused them together. Next, we etched off one of the substrates, then cut the wafer sandwich into two pieces. Finally, we stacked the resultant pair of wafers and wafer-fused them together to yield a four-layer structure. By iterating this cut-stack-fuse process, we can produce eight-layer structures and higher. We constructed a very uniform structure with an accuracy of 30 nm. Testing the transmission spectra for various incident angles, we demonstrated that the structure has complete photonic bandgaps at 1.3 and 1.55 µm. A two-unit crystal composed of eight stacked-stripe layers created the maximum bandgap effect of more than 40 dB attenuation, which corresponds to 99.99% reflection.

Figure 3. An SEM image of a 3-D sharp (90°) bend waveguide. In this structure, the stripe period is 4 µm

We also fabricated a sharp (90°) bend in a 3-D waveguide.⁵ To do this, we fabricated a stripe pattern that was missing one stripe. Using wafer fusion, we then stacked the incomplete layer with another stripe layer that was complete. We created a pair of these flawed/unflawed stacks and bonded them together in a crossed configuration so that the missing stripes were orthogonal to each other (see figure 3). Finally, we sandwiched the central four layers between a pair of four-layer striped structures that did not include defects. In total, the structure has twelve layers, including the two defect layers in the middle of the structure. Based on these results, we have developed various ultra-small optical circuits based on the 3-D photonic crystal.

Although the control of light in a 2-D photonic crystal is not as complete as in a 3-D crystal, it is possible to construct a variety of important functional devices. Two possibilities include a novel laser with multidirectional distributed feedback (DFB) and any number of devices that depend on photon trapping in defects.

DFB diode lasers generate output with increased stability and at more precisely controlled wavelengths than regular diode lasers. Typical DFB lasers incorporate a grating into the design so that feedback from the grating causes interference that promotes gain at a wavelength related to the period of the grating. Usually, however, the grating (and the feedback) occurs in only one dimension. Photonic crystals allow us to make multidirectional DFB lasers, which could provide high-power surface-emitting lasers with narrow wavelength bandwidths.

Figure 4. The structure of a 2-D photonic crystal laser (top) provides multidirectionally distributed feedback via the effect of a triangular lattice (bottom).

We made such a device with a 2-D photonic crystal, in which we integrated two wafers (A and B) using a wafer-fusion technique (see figure 4).⁴ Wafer A incorporates an indium gallium arsenide phosphide/indium phosphide (InGaAsP/InP) multiple quantum well active layer ( = 1.3 µm) grown on a p-type in InP substrate. Wafer B features a triangular lattice structure on an n-type InP substrate. The period of the triangular lattice in the -X direction is designed to coincide with the wavelength of the light emitted from the active layer; thus, a second-order distributed feedback effect occurs in the -X direction.

Distributed feedback from a triangular lattice creates a number of interesting effects. The triangular lattice structure features six equivalent -X directions, so the lightwaves propagating in individual -X directions are diffracted not only backward but also to four other -X directions (±60^o, ±120^o). In other words, six waves couple with each other. As a result, a coherent coupling occurs among six lightwaves propagating to equivalent -X directions, and 2-D lasing oscillation occurs. The light is also coupled out of the plane of the lattice normal to the substrate surface through the upward diffraction since the Bragg condition is also satisfied for this direction. The near-field pattern and the spectra observed at individual positions reveal that the device actually oscillates two-dimensionally and coherently. The 2-D photonic crystal controls not only the longitudinal mode, like linear DFB lasers, but also the lateral mode. Thus, we expected that the device can work as a high-output-power surface-emitting laser that can oscillate in a large 2-D area. Recently, we have succeeded in controlling the polarization mode of a 2-D photonic crystal laser by unit cell structure design.⁷

Above, we described how stripe defects created tightly confining waveguides in a 3-D optical crystal. In 2-D optical crystals, line defects can guide light in a similar way. Moreover, isolated defects can act as traps or output couplers for light in 2-D so that combinations of defects allow us to build unique devices with useful applications.³ Consider a 2-D photonic crystal slab with a triangular lattice. The 2-D photonic bandgap effect confines light in the in-plane direction, and the large refractive index contrast confines light in the vertical direction.

We designed such a slab with a line-shaped defect and two isolated defects (labeled i and j). The line-shaped defect forms a straight waveguide; with an appropriate design, it provides lossless transmission.⁸  Each isolated defect acts as a photon trap, in the same way that a defect in a semiconductor traps electrons or holes. The wavelength of photons that can be trapped by the defect can be tuned by the defect size. The defect constructs a very small cavity with each resonant frequency (or wavelength). In other words, if we change the size of the defect, the resonant wavelength of the defect can be changed. There is no nonlinear effect here. The trapped photons are emitted to relatively unconstricted regions; in this case, they are emitted up out of the slab into free space.

We fabricated this device and obtained experimental results that demonstrate how the defects trap photons at wavelength _i and _j and emit them normal to the slab surface. Using a lensed optical fiber, we launched light into the waveguide from the right edge of the slab. The light source was a tunable semiconductor laser operating in the 1.55 µm region.  We used an infrared camera to monitor the photon trapping emission. When the laser was tuned to 1.545 µm, the defect j emitted the light strongly, but the defect i did not emit light. On the other hand, at 1.566 µm, the emission from defect j disappeared, but the strong emission from defect i appeared.

With this principle, it is possible to create a simple add/drop device just by setting up an optical fiber above the photonic crystal to catch the emitted photons. The device is compact: The width of the waveguide and isolated defects need only be about 10 µm, and the distance between point defects about 4 µm, so a device for 100 wavelengths would only take about 10 x 400 µm² of space.

The example is only one of many possible devices that can be made utilizing the strong localization of photons at a defect; others include enhancing nonlinear optical phenomena and trapping nanoparticles. Researchers around the world are investigating photonic crystals for applications in numerous areas, including telecom, biomedical engineering, and scientific research. oe

References

1. E. Yablonovitch, J. Opt. Soc. Am. B 10, p. 283 (1993).

2. J. D. Joannopoulos, P. R. Villeneuve, and S. Fan, Nature 386, p. 143 (1997).

3. S. Noda, A. Chutinan, and M. Imada, Nature 407, p. 608 (2000).

4. S. Noda, N. Yamamoto, et al., IEEE J. Lightwave Technol. 17, p. 1948 (1999).

5. S. Noda, K. Tomoda, et al., Science 289, p. 604 (2000).

6. K. M. Ho, C. T. Chan, et al., Solid State Commun. 89, p. 413 (1994).

7. S. Noda, M. Yokoyama et al., Science (10 August 2001).

8. A.Chutinan and S. Noda, Phys. Rev. B 62, p. 4488 (2000).

Susumu Noda began his work with photonic bandgap (PBG) materials, popularly known as photonic crystals, at Kyoto University (Kyoto, Japan) in 1994. "We decided that as long as we were studying 3-D devices, we might as well include 2-D PBG materials at the same time," Noda says. And that led to a serendipitous discovery.

"We first designed a 3-D sharp-bend waveguide, and then, incidentally, a 2-D one. The results showed the 3-D waveguide superior to the 2-D, because the 2-D slab structure leaked photons," Noda explains. But he wondered if there were not some way to control and utilize this leakage, so he introduced defects or holes into the slab. "Our experiments then showed that photons escaped defects. Due to the nature of the Bose particle, defects in PBG materials can trap all the photons possible. So a single defect can control many photons. Furthermore, we found we could tune the size of the defects so they would emit only photons of specific wavelengths," he says.

Corporations are showing interest in the work that Noda and his team have done. "For example, medical companies are looking at our vertical-emitting PBG slabs for use in very compact gas analyzers," he says. "We've already demonstrated the viability of 2-D slabs as waveguides. Products are probably three to five years away." Practical optical chips made from 3-D PBG devices, on the other hand, will take at least another decade.

From the beginning, Noda concentrated on finding technology that could be applied to telecommunications. "But I never expected results so quickly," he says. "It happens so many times in research. You stumble across something you're not even looking for. And that turns out to be a breakthrough."

—Charles Whipple