Using 3-D photonic crystals (PCs) with point defects, Susumu Noda and colleagues at Kyoto University (Kyoto, Japan) demonstrated PCs that suppress light in the photonic bandgap (PBG) but allow light to exit the point defects, which are tuned to the 1.45- to 1.6-µm wavelength range. "People have postulated the possibility ever since the concept of PCs came in 1987," says Noda. "Now we have achieved spontaneous emission control with PCs."
The results should be useful not only for present communication systems but also for quantum communications, which requires efficient single photon light sources. "According to our results," says Noda, "the excited carriers (electrons and holes) are very effectively recombined and converted to photons." He sees the technology ultimately resulting in photonic chips, zero threshold current lasers, memory devices, and nonlinear devices.
Noda's team constructed "woodpile" PCs featuring gallium arsenide (GaAs) substrates with up to nine stacked layers of 0.7-µm-period striped GaAs, which were laser aligned for precision and then fused together. The addition of a light-emitting layer was a challenge. The active region contained an indium gallium arsenide phosphide (InGaAsP) multiple quantum well (MQW) layer grown on an InP substrate, which has a different thermal-expansion coefficient than GaAs. "When we first attempted to fuse the two, the GaAs layer and the InP layer separated during the cooling process," Noda explains. "We were able to overcome these thermal expansion issues by developing a multistep wafer-fusion technique that involves heating for longer periods at lower temperatures. We did this in combination with thinning the InP substrate and conducting high-temperature heating."
Plots show emission when the defect wavelengths are outside the bandgap (left) and when the defect is at 1.55 µm, or within the bandgap (right).
The team then fused a second PC on top of the light-emitting layer. They made two PC samples, one with five layers and another with nine layers. The light-emitting layer in the samples was also stripe-layered to preserve the periodicity of the PC. The period length, width, and thickness of the stripe layers, including the light-emitting active MQWs, was 0.7 µm, 0.2 µm, and 0.2 µm, respectively. This placed the bandgap in the optical communications wavelength region at 1.55 µm.
Chips in hand, Noda's team set out to test the effect of the PBG on suppression of spontaneous emission. They excited the devices with a continuous-wave titanium-aluminum oxide laser operating at 900 nm. "We took care to see that only the MQW light-emitting layer was excited," says Noda. The team set the excitation power density at about 10 µW/µm2, at which point the MQW layer became transparent to the 1.55 µm emission wavelength of MQWs, then took the results of both five-layer and nine-layer crystals. These were compared with a reference sample and the results enabled the team to say that suppression of light emission was the result of the PBG effect.
Next, the team set about introducing several different sizes of artificial point defects into separate samples: (i) 3.77 µm * 3.60 µm, (ii) 2.39 µm x 2.52 µm, (iii) 1.75 µm x 1.77 µm, (iv) 1.02 µm x 1.03 µm, (v) 0.76 µm x 1.46 µm, (vi) 0.76 µm x 0.65 µm, and (vii) 0.44 µm x 0.60 µm. Photoluminescence (PL) spectra were recorded for these defects in the Γ-X' direction, then compared with measurements taken from a full PBG without defects (see figure on p. 9). Noda's group found that the largest defect (i) shows strong emission in a broad range of wavelengths contrasting with that of the full PBG region, indicating that defect-cavity modes are created. The number of defect-cavity modes is so large that emission from individual modes overlap. Emission in the wavelength region 1.45 to 1.6 µm is especially strong, suggesting that the full PBG results in enhancement of defect-mode emission by suppressing other leakage paths. When the defect size is steadily reduced from (i) to (vii), the broad multimode emission narrows until a single emission peak is observed for the smallest defect (vii), which the group believes to be the first clear emission from a nanocavity in the 3-D PC. Mathematical modeling supported the data. "Our results are direct evidence that light emission is suppressed in the PC region with emission coming only from the artificial defect," says Noda.
"It's the first time that spontaneous emission has been significantly suppressed and controlled in all three dimensions," says PC pioneer Eli Yablonovitch. "The suppression factor Noda achieved is about 100, but it could have been even larger with more layers in the photonic crystal. Noda's 3-D accomplishment stands as a tour de force. He is ahead of everybody in this field."