Connecting the dots
As broadband optical network systems continue to grow, there is demand for high-performance optical devices. One of the most promising devices for the future is the quantum-dot (QD) laser. The concept of the QD was first proposed in 1982 by Hiroyuki Sakaki and me,1 together with quantum wire lasers for the three-dimensional nanoscale structures of artificial atoms. In our first paper, we called the structure a three-dimensional quantum well, but a year later named it a quantum box. Since the early 1990s, the term quantum dot has been more popular, probably because it is extremely difficult to fabricate a small box-like structure.
Figure 1. Schematic illustration and density states of QDs.
When electrons are confined in 10-nm-scale, 3-D semiconductor heterostructures, the electron motion is fully quantized, which creates artificial atomic states in semiconductors. With the decrease of dimensionality of freedom for electron motion, the electronic states gain unique features, such as atomic-like discrete states with a -function density of states (see figure 1). A semiconductor laser with a QD active region promises ultralow and temperature-independent threshold current,2-4 high-frequency modulation with negligible chirping effect,5,6 and nonlinear gain effect.7
In the 1980s, it was almost impossible to fabricate these lasers, however. The first demonstration of QD effects in semiconductor lasers was achieved by placing quantum-well (QW) lasers in high magnetic fields in which the Lorentzian force eliminated the 2-D freedom of electron motion. Semiconductor double heterostructure lasers and quantum-well lasers placed in magnetic fields as high as 30 T demonstrated reduced temperature dependence, narrower spectral line width, and enhanced modulation frequency.8-10 These experiments convinced us that QD lasers should be very useful even though we thought, frankly, that the QD would be a structure not realized until the 21st century.Growing pains
In spite of difficulty in fabricating the QD structures, there were various trials in the late 1980s and early 1990s. The most straightforward technique to produce QDs was to fabricate suitably sized mesa-etched quantum wells grown by metal-organic chemical vapor deposition (MOCVD) or molecular beam epitaxy. The nonradiative defects produced during etching led to material degradation, however, which made the structures unsuitable for lasers. Despite these challenges, by 1994 researchers at the Tokyo Institute of Technology had reported lasing operation from an etched QD laser. They claimed they achieved lasing action in gallium indium arsenide/gallium indium arsenide phosphide/indium phosphide (Ga0.67In0.33As/GaInAsP/InP) tensile-strained, quantum-box lasers.
Figure 2. Self-assembled QDs and selective grown QDs on a SiO2 patterned substrate.
During the 1990s, researchers developed both selective growth and self-assembled growth techniques that could avoid nonradiative defects (see figure 2). In particular, the Stranski-Krastanow growth mode is very successful for indium gallium arsenide/gallium arsenide (InGaAs/GaAs) systems. Self-assembled QD (SAQD) islands form with the strain-driven process when the misfit is larger than 1.7%. In 1985 a group at the Centre Nationales d'Etudes Telecommunications (Brittany, France) found that the growth of InAs on a lattice mismatched GaAs layer led to 3-D nanoscale islands on top of a thin InAs wetting layer.11
By 1995 growth of the SAQDs was well understood,12-15 and groups even reported natural alignment of the SAQDs without any pre-patterning.16 The area density can be controlled from 109 /cm2 to 1012 /cm2 by varying total deposition of InGaAs on the GaAs epitaxial layer. The size of the QDs strongly depends on growth temperature and other conditions, including the III/V ratio. The average size of the QDs is in the range between 10 and 50 nm, with a size fluctuation of 10% to 30% for InGaAs structures. This size range produces a large inhomogeneous broadening in the photoluminescence spectrum. Multiple layers of vertically stacked QDs separated by barrier layers with suitable thickness also can be grown in order to increase the area density effectively.
This self-assembled growth for making nanoscale islands led to a breakthrough in QD devices. In 1994 researchers at the Technical University of Berlin (TUB) and Ioffe Research Institute (St. Petersburg, Russia) reported the first self-assembled InAs/InGaAs QD laser, showing reduced temperature dependence of the threshold current.17 Since then, InGaAs/GaAs QD lasers have successfully demonstrated high-performance lasing characteristics,18-24 including low threshold current density of 21 A/cm2, high T0 up to 385 K, and high differential gain at room temperature. Furthermore, QD devices have lased at 1.3 µm, a necessary attribute for access network communication systems that use GaAs substrates. The technology has also been used to produce vertical-cavity surface-emitting lasers (VCSELs). These results are very promising for light sources in access and home/business local area network systems.
Recently, our group achieved a wavelength of emission from the ground state of over 1.4 µm by modifying the QD structure.25 We formed InAs QDs, then grew a 5-nm strained InxGa1-xAs QW capped by GaAs. Peak wavelength of InAs QDs, plotted as a function of the percent of indium content x, shows that device output shifts toward a longer wavelength when the indium composition x of the strained InGaAs QW is increased. Indeed 1.52-µm emissions were obtained from InAs SAQDs in In0.45Ga0.55As QW. The higher subbands are also clearly observable, showing the quantized states formed in the QDs.Benefits for blue
Short-wavelength QD lasers have particular promise for highly dense optical data-storage applications. The reduction of threshold current provided by QD technology is even more pronounced in wider-bandgap semiconductors such as gallium nitride (GaN), for example, yielding a more practical device.
Generally speaking, the threshold current of a QW laser increases as the effective mass of electrons mc or the ratio of the effective mass of holes mv to mc (mv/mc) grows. In GaN-based devices and other wide bandgap semiconductors, however, these values are larger than those for GaAs-based semiconductors. Threshold current Ith is directly proportional to carrier density ntr at the condition under which the material becomes optically transparent; ntr depends on mv/mc. Calculations show that ntr increases monotonically with the increase of mv/mc, and that ntr also grows when mc becomes larger. This difference in ntr leads to difference in the minimal threshold current density Jthin GaN-based QW lasers, Jth is about 1 kA/cm2, while the Jth of GaAs-based QW lasers is approximately 1A/cm2.
Now consider a device using QDs in the active region. If the QDs are small enough, we can ignore the population of carriers in the higher subband. In this case, the achievable threshold current Ith in both GaAs-based diode lasers and GaN-based diode lasers is almost the same: about 1A/cm2 to 1 µA. In other words, the use of QD technology improves the threshold current by a factor of 100 in GaN-based lasers compared to GaAs-based lasers.
Recently, we have succeeded in growing InGaN QDs on a GaN epitaxial layer and operated an InGaN/GaN QD laser at room temperature.26 We looked at the relationship between the excitation energy per pulse and the emission intensity polarized in the transverse electric (TE) or transverse magnetic (TM) mode and observed a clear threshold in the dependence of the TE polarized emission intensity on the excitation energy. The spectra of the emitted light from the laser structure, below and above the threshold, also confirm the lasing action in the InGaN QD laser. Light emission from the InGaN quantum dots was evidenced by single dot spectroscopy.27Challenges
Despite the potential benefits of QD technology, the size variation of QD arrays grown by present methods prevents us from achieving a true zero-dimensional quantum effect, which is indispensable for obtaining high-speed modulation dynamics and no-chirping effects. The inhomogeneous broadening of typical QD arrays due to size variation is about 20 to 30 meV. If we can reduce the size variation by 80%, the inhomogeneous broadening will be smaller than homogeneous broadening due to carrier dephasing at room temperature. Such size consistency will require overcoming various challenges in the development of the technology, however.
Researchers have for some time discussed the so-called phonon bottleneck, which degrades the luminescence properties of QD lasers. This issue is related to the inhibition of carrier relaxation due to the selection rule of interaction between electrons and phonons in discrete energy level systems. Various mechanisms can almost suppress such bottlenecks. Researchers have used Auger interaction, breaking of the selection rule in a strong interaction regime, or Coulomb interaction at room temperature to produce QD lasers with an extremely low threshold current.
The development of 40 GHz, directly modulated QD lasers operating at 1.3 to 1.5 µm with low chirping effects would provide QD technology with an entry point in the telecom market. To achieve this device performance by 2010, we must reduce inhomogeneous broadening due to the size fluctuation and create high-density QDs. On the other hand, by taking advantage of the inhomogeneous broadening in the gain spectrum, we can develop ultrawide-band amplifiers and nonlinear optical switching devices for broadband dense-wavelength-division-multiplexing (DWDM) systems. Such QD devices can be integrated into the photonic crystals that will be used as active optical router circuits. Moreover, as mentioned before, 1.3-µm QD lasers on GaAs substrates look promising for low-cost access network devices in five years.
Once more uniform QDs are developed, the technology will be open to various future high-performance devices, not only as telecommunication light sources but also for quantum computing devices through ultra-precise control of the shape and position of QDs. The use of intersubband in the QDs is also attractive for future light sources, as well as infrared detectors. Other promising applications include cascade QD lasers for light sources, and single-photon QD emitters with 3-D photonic crystal microcavities. As the world moves toward ubiquitous broadband networks, QD lasers will be indispensable. oe
1. Y. Arakawa and H. Sakaki, Appl. Phys. Lett., vol. 40, pp. 939-941(1982).
2. M. Asada, Y. Miyamoto, et al., IEEE J. Quantum Elect., 22, 19151921, (1986).
3. A. Yariv, Appl. Phys. Lett., 53, 1033 (1988).
4. T. Takahahsi and Arakawa, Optoelectronics-Devices and Technologies, 3, 155 (1988).
5. Y. Arakawa, K. Vahala, et al., Appl. Phys. Lett. 45, 950 (1984) .
6. Y. Arakawa and A. Yariv, IEEE J. of Quantum electron., QE-22, 1887 (1986).
7. Y. Arakawa, T. Takahashi, Elect. Lett., Vol.25, pp.169 (1989).
8. Y. Arakawa, et al., Jpn. J. of Appl. Phys., 22, L804 (1983).
9. Y. Arakawa , K. Vahala, et al., Appl. Phys. Lett., 47, 1142 (1985).
10. Y. Arakawa , K. Vahala, et al., Appl. Phys. Lett., 48, 384 (1986).
11. L. Goldstein, F. Glas, et al., Appl. Phys. Lett. 47, 1099 (1985).
12. D. Leonard, K. Pond, et al., Phys. Rev. B, vol. 50, 11687 ( 1994).
13. P. Chen, Q. Xie, A. Madhukar, et al., J. Vac. Sci. Tech. B, 12, 2568 (1994).
14. J.Oshinowo, M.Nishioka, et al., Appl. Phys. Lett., 65, 1421 (1994).
15. Q. Xie, A. Madhukar, et al., Phys. Rev. Lett., 75, 2542 (1995).
16. M. Kitamura, M. Nishioka, et al., Appl. Phys. Lett. 17, 1675 (1995).
17. D. Bimberg, N. Kirxtaedter, et al., IEEE Select. Topics in Quantum Electron., 3, 196 (1997).
18. H. Shoji, Y. Nakata, et al., Appl. Phys. Lett., 71 79 (1997).
19. Y. Arakawa, Nishioka, et al., IEICE, E-79-C-11, 487 (1996).
20. K. Nishi, H. Saito, et al., Appl. Phys. Lett. 73(1998) 526.
two decades on the dot
In 1981 Yasuhiko Arakawa became the first researcher to propose the concept of three-dimensional confinement of carriers and quantum-dot lasers, predicting significant reductions in the temperature dependence of threshold current. "The only problem was that at the time, we didn't have the technology to fabricate the kind of multidimensional nanostructures necessary to prove my concept," Arakawa says, "so I did it by placing quantum-well lasers or DH lasers in a high magnetic field."
In 1981 Arakawa received his Ph.D. in electrical engineering from the University of Tokyo. He is now a full professor at the university's Research Center for Advanced Science and Technology. While at the California Institute of Technology (Caltech; Pasadena, CA) for two years beginning in 1986, Arakawa did seminal work in theorizing that low-dimensional carriers in quantum wires and dots are effective in enhancing modulation dynamics and spectral properties of semiconductor lasers. "Our discussions at that time had quite an impact on research in quantum well, wire, and dot lasers," Arakawa says. Still, it was 1992 before he was able to use metal-organic chemical vapor deposition (MOCVD) to grow gallium arsenide (GaAs) quantum wires and dots.
Other firsts came quickly: quantum wire vertical-cavity surface-emitting lasers (VCSELs) in 1994, and quantum-dot VCSELs and nitride-based quantum-dot lasers at room temperature in 1997. Arakawa's contributions also include discovery of continuum states in the energy range below the band gap energy of the wetting layer, which showed that self-assembled indium arsenide (InAs) quantum dots were different from ideal quantum dots. He also contributed to the theoretical understanding of the strong interaction between confined electrons and photons in quantum dots.
The year 2002 will see Arakawa head a bigger research team as the Japanese government increases funding for scientific research at major universities. "I must put together a 28-member team to delve into increasing optical memory sizes, nanotechnology, and cost reductions through technology," Arakawa says. -Charles Whipple
Yasuhiko Arakawa is a professor of engineering at the Research Center for Advanced Science and Technology, University of Tokyo.