In the past, vertical-cavity surface-emitting lasers (VCSELs) with quantum dots (QDs) in their active regions were limited to lasing from indium arsenide (InAs) QDs at 1 µm, because the gallium arsenide (GaAs) substrate's compressive strain changed the energy bandgap of the InAs. However, a research team at the National Institute of Information and Communications Technology (NICT; Tokyo, Japan) used denser indium gallium antimony (InGaSb) QDs discovered while building side-emitting lasers to construct VCSELs that exhibit continuous wave (CW) operation at 1.34 µm at room temperature.
"We grew the VCSEL structure on a (001) oriented n-GaAs substrate using solid-source molecular-beam epitaxy," explains team leader Naokatsu Yamamoto. "We then laid down a buffer layer of n-GaAs, and grew a bottom distributed Bragg reflector (DBR) as multiple layers of 20-period silicon-doped AlGaAs/GaAs on the buffer layer at 540°C."
Atop the DBR, Yamamoto's team placed a guiding layer of silicon-doped n-GaAs at 400°C. "The next step was the most important one in this process," says Yamamoto. "We used an irradiating antimony-flux to irradiate silicon atoms onto the GaAs guiding layer immediately before fabricating the quantum dots. This enabled us to create a high-density active layer of 8-period InGaSb QDs. While it is not a difficult process, it is vital to achieving laser emissions at 1.3 µm."
The team made chips with and without irradiation, and checked the QD density of both with atomic force microscopy. The microscopy proved that QD density was dramatically enhanced by irradiation from the silicon atoms. "In fact," says Yamamoto, "the highest density of InGaSb QDs is achieved around 4.4 x 109/cm2, which is nearly 100 times higher than without silicon atom irradiation."
The high-density QD layer then received a 15-nm-thick cap layer of undoped GaAs, followed by a beryllium-doped GaAs guide layer. A top DBR mirror was then fabricated of 14-period beryllium-doped AlGaAs/GaAs multilayers laid at 540°C. The team then used a vapor deposition/lift-off technique to deposit and pattern electrode metals (gold-zinc and gold-germanium). Finally, the sample was wet-etched into 400-µm square mesas.
When the sample was completed, Yamamoto's team applied various levels of DC current. The researchers then used the output figures to estimate the threshold current at about 500 mA. "When we used currents under the threshold amperage," says Yamamoto, "we observed an electro-luminescence emanating from the VCSEL. But when the current surpassed the threshold value, we got a clear emission peak, that is, lasing, at about 1335-nm frequency. And, the stronger the current we applied, the more intense the peak became."
Laser intensity spectra for different grating periods (curves A-G) and photoluminescence spectra for the bulk material (dashed line), both normalized to unity.
Compared to extant quantum well (QW) lasers that emit at 1.3 µm, Yamamoto says his team's laser has several potential advantages. QD lasers have a lower threshold current than QW lasers, which means less power consumption. QD lasers are considered temperature-independent, so they do not need a cooling system. Yamamoto says QW lasers conventionally require expensive indium-phosporus (InP) substrates. While InGaAs material on an InP substrate emit long wavelengths at 1.3 or 1.5 µm, it is very difficult to make DBR mirrors on InP. On the other hand, DBR mirrors are easily fabricated on GaAs.
VCSEL pioneer Kenichi Iga says, "The formation of Q-dot looks interesting for specialists in this field. However, the laser device performance leaves much to be desired. It is difficult to judge that the device reached the lasing threshold, based upon the spectrum and current level. Nevertheless, the wave size is huge and I did not know of surface emitting laser oscillation in this dimension, other than in carefully designed high-powered devices."
Yamamoto adds that although this test performance was lacking in some ways, the team plans to improve performance by increasing the reflectance of the DBR mirrors and decreasing the size of the mesas. Yamamoto says those steps should also decrease the threshold current. In addition, he says it is difficult to fabricate materials on a GaAs substrate that emit around 1.3 µm, which is why this experiment is important.