Open Access
Add to BibSonomyAbstract
Different capping of quantum dot (QD) materials is known to produce different degrees of intermixing during a post-growth thermal annealing process. We report a study of the effect of different degrees of intermixing on modulation beryllium doped quantum dot superluminescent light emitting diodes (QD-SLEDs). The intermixed QD-SLEDs show high device performance whilst achieving a large central emission wavelength shift of ~100nm compared to the as-grown device. The evolution of the emission spectra and power with drive current suggest a transition from QD-like to QW-like behavior with increasing degree of intermixing. A selective area intermixed QD-SLED is demonstrated, and with optimized differential intermixing, such structures should allow ultra-broadband sources to be realized.
©2010 Optical Society of America
1. Introduction
Recently self-assembled quantum dot (QD) structures have attracted considerable interests for the realization of broadband superluminescent light emitting diodes (SLED) utilizing their inhomogeneously broad emission spectra and readily saturable states [1–8]. High power and broadband QD-SLEDs have been achieved by several research groups [2–7], which are very promising for a wide range of applications such as wavelength division multiplexing system testing, fibre optic gyroscopes and optical coherence tomography.
Modifying the emission properties of quantum well (QW) structures using a post-growth annealing process (PAP) is a very useful technique for the fabrication of integrated photonic circuits. Selective area PAP processes have been applied to realize active and passive portions of a device allowing, for example, non-absorbing mirror facets to realize high power single-mode lasers [9]. The modification of QD properties using PAP has also attracted significant interests. However, the study of PAP applied to QDs has so far been limited to the photoluminescence studies of materials and the realization of integrated active/passive devices, with only a few reports on intermixed active devices [10–12]. One main reason is that QD structures are highly sensitive to the annealing parameters, with high annealing temperature or long annealing time sharply degrading the devices’ performance. Recently, modulation carbon doped (C-doped) QD structure have been successfully employed as the active region for the fabrication of intermixed QD laser [13] and intermixed QD-SLEDs [14]. These devices exhibited small emission wavelength shift (~10nm blueshift of the ground state (GS) emission), a decrease in splitting between GS and excited state (ES), and a broadening of the inhomogeneous broadening of these states.
In this paper, we report on modulation-doped structures where beryllium (Be) replaces carbon as the p dopant in the QD active region for the fabrication of intermixed QD-SLEDs. Beryllium has a very different diffusion rate compared to carbon, providing an additional degree of freedom in the design of an intermixed device. The device characteristics of intermixed QD-SLEDs where SiO2 caps and GaAs proximity caps are used during the annealing process are reported. Devices using both types of cap produce high pulsed output powers, and the change of the emission spectrum following intermixing is discussed. Based on these results, the selective area annealing process, with different caps used during the same annealing procedure, for fabricating a selective area intermixed QD-SLED is investigated and discussed.
2. Experiments
A 6 layer InGaAs/InAs/InGaAs dot in well (DWELL) structure was grown in a molecular beam epitaxy (MBE) system on a Si-doped GaAs (100) substrate. Each QD layer consists of 2.9 monolayer (ML) InAs grown on a 2nm InGaAs layer and covered by a 6nm InGaAs layer. The 6 DWELL structures were separated by 45 nm GaAs barriers. Modulation p-doping with Be to a concentration of 30 acceptors per dot was located in a 6nm wide GaAs layer, 9nm beneath each DWELL [15]. The whole QD active region was sandwiched by lower n-Al0.25Ga0.75As and upper p-Al025Ga0.75As cladding layers. Annealing processes were performed in an N2 ambient at temperatures of 700°C and 750 °C for 5 min or 10min, by using GaAs proximity capping or a 200nm plasma-enhanced chemical-vapor deposition (PECVD) deposited SiO2 layer. Room temperature photoluminescence (RT-PL) spectra were obtained by exciting the sample with a HeNe laser emitting at 633nm. PL samples were prepared by etching the p-AlGaAs cladding of the sample. The signal was detected with an InGaAs detector. The PL system used was a commercial PL mapper (RPM2000). Selected samples were processed into 5µm wide ridge waveguide structures by dry etch through the p-side AlGaAs cladding layers, stopping ~200nm above the QD active region. Au/Zn/Au and InGeAu provided the p-type and n-type Ohmic contacts, respectively. The waveguide structures were oriented at ~7° from the facet normal to suppress lasing. 6mm long bars were mounted on gold plated copper tiles without antireflection coating on the facets. Device characterisation was performed at room temperature under pulsed operations (5 µs pulse width, 1% duty cycle) to eliminate the effects of self-heating.
3. Results
Figure 1 shows RT-PL spectra of the samples under identical excitation power of 9.5mW, as a function of annealing temperature and time using a GaAs proximity cap and a SiO2 cap. Data for the as-grown sample is also plotted. For the as-grown sample the GS emission peak is at 1296 nm, and ES peak at 1214 nm. The full width at half maximum (FWHM) of the GS and ES are ~70 nm and ~90 nm, respectively. The PL linewidth of the 700 °C annealed sample with GaAs proximity cap are significantly broadened with a GS emission peak blue-shifted to 1234 nm (FWHM ~145 nm) and an ES peak to 1194 nm (FWHM ~140 nm). This increase in linewidth is attributed to increased interface fluctuations between QD and the surrounding matrix. In the as-grown case the inhomogeneous linewidth is governed by QD size and composition variations. The inhomogeneity may increase when different intermixing effects due to differing composition gradients driving the diffusion process [8,12].
Fig. 1 Room temperature PL spectra of the as-grown QD sample and annealed QD samples by using different annealing temperature, different time, and different caps during annealing process.
As PL samples were prepared by etching away the p-AlGaAs cladding, we do not rule out the possibility of errors in the comparison of PL intensity. However, a trend is observed in the reduction of PL intensity with increased PAP temperature and time. For the sample annealed for 5 minutes at 700°C the peak intensity is observed to reduce to around half that of the as-grown sample. However, the integrated intensity is very similar between these two samples due to the significant increase in line-width of the states. This suggests the optical quality of the QDs has been preserved at this annealing condition. The large increase in linewidth of the QD states for this annealed sample is very promising for achieving a broadband QD-SLED. The sample annealed at 700 °C for 5 minutes with a SiO2 cap exhibits a larger GS emission peak blue-shift of ~126nm (from 1296 to 1170nm) with around a factor of six reduction in PL intensity compared to the as-grown sample. The observation of an enhancement to interdiffusion for samples with SiO2 caps is in agreement with other reports in the literature [16]. The difference in the effect of interdiffusion between the above two 700 °C annealed QD samples may be due to two factors. In the annealed sample with a GaAs proximity cap, As will be desorbed from the GaAs surface resulting in a Ga rich surface and subsequent diffusion of Ga into the sample increasing the concentration of group-III interstitials which will reduce interdiffusion compared to the SiO2 capped sample. In the annealed sample with a SiO2 cap, the SiO2 layers deposited using PECVD are porous structures, which enhance the diffusion of Ga atoms through the SiO2 matrix, and hence the density of Ga vacancies will be increased, which will enhance the interdiffusion [16].
Samples annealed at 750 °C for 5 minutes and 10 minutes with GaAs proximity caps also exhibit large GS emission peak blue-shifts to ~1155nm and 1140nm, respectively with large PL intensity reductions compared to the as-grown sample. These observations are in agreement with a stronger interdiffusion under these annealing conditions at higher temperatures and longer times than those discussed previously. Based on the observations from PL test samples, the as-grown sample, and samples annealed at 700°C with GaAs proximity cap and SiO2 cap were fabricated into SLED devices.
The light-injection current (L-I) characteristics of the as-grown QD-SLED, QD-SLED with GaAs proximity cap and QD-SLED with SiO2 cap (both annealed at 700C for 5 minutes) are shown in Figure. 2. Inspection of the emission spectrum (discussed below) indicates that lasing is successfully suppressed in these devices, and that the large optical power is due to amplified spontaneous emission. Applied current was pulsed-source limited to 3A, and we cannot rule out the complete illumination of self-heating effects at these high currents. At 3A a maximum output power of ~120mW was achieved by the as-grown QD-SLED, while for the other two intermixed QD-SLEDs, the maximum power at 3A are higher than the as-grown device. The intermixed QD-SLED with GaAs proximity cap exhibits a maximum higher power of around 190mW, whilst the SiO2 cap intermixed QD-SLED exhibits ~140mW at maximal current.
Figure 3(a) shows the electroluminescence (EL) spectra as a function of drive current for the as-grown QD-SLED. Under low drive currents, the emission is mainly from the GS of QD with central emission peak at ~1316nm. With increasing injection current, the ES gradually dominates the emission spectra, with peak emission at around 1242nm. This is due to the saturation of the GS and increased population of the ES due to Pauli-blocking. The onset of the dominance of the ES in the emission spectrum occurs at >1A, corresponding to the strong increase in emission power observed in Fig. 2 . The higher degeneracy of the ES results in a two-fold increase in both spontaneous emission and saturated gain compared to the GS, resulting in the dominance of the ES in the emission spectrum. The large energy splitting of the ES and GS in the as-grown sample results in a large spectral dip occurring in the EL spectrum. For single contact QD-SLEDs the length of the device dictates the power at maximum bandwidth, corresponding to the balance of GS and ES power. The presence of large spectral dips at this condition of maximum bandwidth is undesirable.
Fig. 3 Electroluminescence spectra as a function of drive currents of as-grown QD-SLED, 700 °C intermixed QD-SLED with GaAs proximity cap and 700 °C intermixed QD-SLED with SiO2 cap during annealing process.
Fig. 2 Light-injection current curves of as-grown QD-SLED, 700 °C intermixed QD-SLED with GaAs proximity cap and 700 °C intermixed QD-SLED with SiO2 cap during annealing process.
The EL spectra as a function of injection current for the device fabricated from 700°C annealed material with GaAs proximity cap is shown in Fig. 3(b). At low currents the emission is dominated by the QD GS emission which has been shifted to ~1212nm by the intermixing process. As current is increased, the ES emission (peak emission ~1176nm) increases in power to provide an equal contribution to total output power at ~3A. Due to the significantly reduced energy separation of QD states in this intermixed device (40nm compared to 82nm in the as-grown material from PL) and an increase in inhomogeneous linewidth compared to that in the as-grown material from PL as shown in Fig. 1, a very flat topped (spectra dip <0.8 dB) emission spectrum was achieved. A small spectral modulation is highly desirable for high resolution optical coherence tomography image application [17]. A bandwidth of 78nm from 1155nm to 1233nm with output power of 190mW is achieved at 3A drive current. In contrast to the as-grown device, GS saturation occurs at higher current densities in the device. The ES power is observed to emerge at ~1.5A, corresponding to an increase in differential efficiency (7.6% at ~1100 mA and 8.5% at ~1750 mA) observed in Fig. 2.
The EL spectra as a function of injection current for the device fabricated from 700°C SiO2 cap annealed material is shown in Fig. 3(c). A single peak is observed at all powers centered at ~1190nm with ~30nm bandwidth. The emission spectra show neither appreciable change in bandwidth nor strong shift in the emission peak with increasing injection current. At the highest drive current of 3A, some evidence for gain saturation and increase in power in a shoulder to the main peak at 1150nm is observed. A shoulder at ~1220nm is also observed which saturates at 1500 mA drive current.
The change and evolution with current of the emission spectra for different annealing conditions is noteworthy. For the as-grown device, state saturation and the evolution of the emission spectrum is well documented and well described by QD state filling effects. For the intermixed samples, we expect a transition from a QD-like system to a QW-like system in the case of extreme annealing (i.e. high temperature, long duration). In the case of the SLED realized from material annealed with a SiO2 cap at 700°C for 5 minutes there is no strong evidence for QD-like state filling from the emission spectra. We therefore consider that the confinement of carriers is more QW-like than QD-like, and assume the resultant hetero-structure to be like that of an inhomogeneous QW. The shoulder which saturates at ~1220nm is consistent with state filling in an inhomogeneous 2D system. For the SLED device fabricated from 700°C annealed material with GaAs proximity cap, clear evidence for state filling is observed, yet this occurs at significantly higher currents than for the as-grown sample. For a pure QD system, this would be explained by a reduction in the average occupancy of the QDs at a given current. This could change if the carrier lifetime were reduced, dot density increased, or relaxation processes were inhibited. The observation of high output powers from the intermixed device is not consistent with increased non-radiative recombination, which would give rise to a reduced carrier lifetime. In addition, an increase in QD density cannot be expected as a consequence of the annealing process. We have no evidence to believe relaxation processes to be significantly different between the as-grown QD samples and the annealed QD sample with GaAs proximity cap. In both cases no significant GaAs band-edge emission is observed. We therefore consider the GaAs proximity capped sample annealed at 700°C for 5 minutes to be a QD structure strongly modified by inter-diffusion between the QDs and the surrounding matrix during the annealing process. This modification appears to manifest itself in an increase in the density of states to shorter wavelength consistent with band-tailing in an inhomogeneous 2D system. The difference between the GaAs capped and SiO2 capped devices therefore appears to be in the magnitude of this transition from a 0D to a 2D system. The minute study of this evolution/transformation of the QD structures and the dimensionality of carrier confinement as a consequence of annealing by e.g. transmission electron microscopy and magneto-spectroscopy is required to clarify this point. The change in dimensionality of carrier confinement, and the subsequent modification of the development of the gain with current may explain the observation of a low PL intensity of the SiO2 intermixed sample, yet a high emission power (albeit with narrow bandwidth).
In order to achieve an ultra-broad-band source, one important approach for future research is to vary the emission wavelength spatially within a device. This may be achieved by selective area epitaxy growth [18] or by selective area post-growth intermixing [19,20]. As most high quality QD-SLED structures are grown by MBE, the later is preferred. Based on the above results, as the effects of inter-diffusion are quite different between the intermixed device with GaAs proximity cap and the intermixed device with SiO2 cap, we have realized a QD-SLED comprising regions with different cap along it’s length. This was realized by combining the GaAs proximity cap with a sample selectively patterned with SiO2. The schematic for this device is shown in Fig. 4(a) . The device is identical to those discussed previously in terms of length (6mm), ridge width (5 µm) and 7 degree angle of the waveguide to the normal to the cleaved facet. It was fabricated simultaneously to the two intermixed devices discussed previously. This selective area intermixed QD-SLED device consists of a 4mm long QD-SLED A (capped by a SiO2 layer during annealing) and a 2mm long QD-SLED B (GaAs proximity cap during annealing process).
Fig. 4 (a). Schematic device structure of the selective area intermixed QD-SLED;(b). L-I curves of the selective area intermixed QD-SLED measured from front and back facet, respectively. Inset: the corresponding spectra of the device at 3A.
The L-I characteristics from the two different ends of this device are shown in Fig. 4(b). The associated emission spectrum from each end of the device, obtained at the maximum current of 3A is shown as an inset. As expected for such an asymmetric device, the power-current characteristics and emission spectrum are different for the two ends of the device. From facet A a maximum power of ~120mW is obtained and at this maximum in drive current a ~45nm 3dB band-width is measured centred at ~1185nm. From facet B a maximum power of ~79mW is obtained and at this maximum in drive current a ~60nm 3dB band-width is measured centered at 1170nm. The modeling of such a device is complex, requiring a time dependant travelling wave solution, incorporating a number of currently unknown material parameters for the two intermixed sections [21]. However, we are able to derive an expectation of the resultant device characteristics if we consider two SLEDs with different gain and ASE bandwidths acting as signal and amplifier. Due to the mismatch in bandwidths we can expect a lower power when compared to a single device of equivalent length. This is in agreement with what we observe from both facets of the device. Similarly, the narrower bandwidth source can have it’s emission broadened due to the additional ASE of the other device in the case of a strong spectral overlap. This is also observed from facet A. The spectral overlap of the current device is not optimal for obtaining ultra-broadband emission. However, we believe that if the spectral overlapping occurs at the edge of the two emission spectra, not overlapped centrally as in the present case as shown in Fig. 3, an ultra broad emission spectrum would be achieved. This may be possible via the use of different cap materials (e.g. sputtered SiO2 [22], TiO2 [23]) or by selective implantation [24] or laser annealing [19].
4. Conclusion
In summary, Be-doped DWELL structures has been demonstrated as a strong candidate to fabricate high performance intermixed QD-SLEDs. The intermixed QD-SLEDs exhibit good device performance whilst simultaneously achieving a large central emission wavelength shift of ~100nm compared to the as-grown device. The study of the emission spectra and power-current characteristics of the devices suggest a transition for QD-like to QW-like behavior with increasing degree of intermixing. A selective area intermixed QD-SLED is demonstrated, and with optimized differential intermixing, such structures would allow ultra-broadband sources to be realized.
Acknowledgements
This work was supported by EPSRC grant number EP/D04801x/1.
References and links:
1. Z. Sun, D. Ding, Q. Gong, W. Zhou, B. Xu, and Z. G. Wang, “Quantum-dot superluminescent diode: A proposal for an ultra-wide output spectrum,” Opt. Quantum Electron. 31(12), 1235–1246 (1999). [CrossRef]
2. Z. Y. Zhang, Z. G. Wang, B. Xu, P. Jin, Zh. Sun, and F. Q. Liu, “High performance quantum-dot superluminescent diodes,” IEEE Photon. Technol. Lett. 16(1), 27–29 (2004). [CrossRef]
3. M. Rossetti, L. H. Li, A. Markus, A. Fiore, L. Occhi, C. Velez, S. Mikhrin, I. Krestnikov, and A. Kovsh, “Characterization and modeling of broad spectrum InAs-GaAs quantum-dot superluminescent diodes emitting at 1.2-1.3 μm,” IEEE J. Quantum Electron. 43(8), 676–686 (2007). [CrossRef]
4. Y. C. Xin, A. Martinez, T. Saiz, A. J. Moscho, Y. Li, T. A. Nilsen, A. L. Gray, and L. F. Lester, “1.3-μm quantum-dot multisection superluminescent diodes with extremely broad bandwidth,” IEEE Photon. Technol. Lett. 19(7), 501–503 (2007). [CrossRef]
5. H. S. Djie, C. E. Dimas, D. N. Wang, B. S. Ooi, J. C. M. Hwang, G. T. Dang, and W. H. Chang, “InGaAs/GaAs quantum-dot superluminescent diode for optical sensor and imaging,” IEEE Sens. J. 7(2), 251–257 (2007). [CrossRef]
6. Y. C. Yoo, I. K. Han, and J. I. Lee, “High power broadband superluminescent diodes with chirped multiple quantum dots,” Electron. Lett. 43(19), 1045–1046 (2007). [CrossRef]
7. S. K. Ray, K. M. Groom, M. D. Beattie, H. Y. Liu, M. Hopkinson, and R. A. Hogg, “Broad-band superluminescent light-emitting diodes incorporating quantum dots in compositionally modulated quantum wells,” IEEE Photon. Technol. Lett. 18(1), 58–60 (2006). [CrossRef]
8. Z. Y. Zhang, R. A. Hogg, X. Q. Lv, and Z. G. Wang, “Self-assembled quantum dot superluminescent light emitting diodes,” Adv. Opt. Photon. 2(2), 201–228 (2010). [CrossRef]
9. C. L. Walker, A. C. Bryce, and J. H. Marsh, “Non absorbing mirror laser with improved catastrophic optical damage level”, Lasers and Electro-Optics Society, 2002. LEOS 2002. The 15th Annual Meeting of the IEEE, 2, 643 – 644, (2002)
10. S. Mokkapati, S. Du, M. Buda, L. Fu, H. H. Tan, and C. Jagadish, “Multiple wavelength InGaAs quantum dot lasers using ion implantation induced intermixing,” Nano. Res. Lett. 2(11), 550–553 (2007). [CrossRef]
11. H. S. Djie, Y. Wang, Y. Ding, D. Wang, J. C. M. Hwang, X. M. Fang, Y. Wu, J. M. Fastenau, A. W. K. Liu, G. T. Dang, W. H. Chang, and B. S. Ooi, “Quantum dash intermixing,” IEEE J. Sel. Top. Quantum Electron. 14(4), 1239–1249 (2008). [CrossRef]
12. Z. Y. Zhang, R. A. Hogg, B. Xu, P. Jin, and Z. G. Wang, “Realization of extremely broadband quantum-dot superluminescent light-emitting diodes by rapid thermal-annealing process,” Opt. Lett. 33(11), 1210–1212 (2008). [CrossRef] [PubMed]
13. Z. Y. Zhang, Q. Jiang, and R. A. Hogg, “Tunable Interband and Intersubbnad transitions in modulation C-doped InGaAs/GaAs quantum dot lasers by post-growth annealing process,” Appl. Phys. Lett. 93(7), 071111 (2008). [CrossRef]
14. Z. Y. Zhang, Q. Jiang, I. J. Luxmoore, and R. A. Hogg, “A p-type–doped quantum dot superluminescent LED with broadband and flat-topped emission spectra obtained by post-growth intermixing under a GaAs proximity cap,” Nanotechnology 20(5), 055204 (2009). [CrossRef] [PubMed]
15. Q. Jiang, Z. Y. Zhang, M. Hopkinson, and R.A. Hogg, “High performance intermixed p-doped quantum dot superluminescent diodes at 1.2μm,” Electron. Lett. 46(4), 295-296 (2010). [CrossRef]
16. B. S. Ooi, K. Mcilvaney, M. W. Street, A. S. Helmy, S. G. Ayling, A. C. Bryce, J. H. Mash, and J. S. Roberts, “Selective quantum-well intermixing in GaAs/AlGaAs structures using impurity-free vacancy diffusion,” IEEE J. Quantum Electron. 33(10), 1784–1793 (1997). [CrossRef]
17. W. Drexler, “Ultrahigh-resolution optical coherence tomography,” J. Biomed. Opt. 9(1), 47–74 (2004). [CrossRef] [PubMed]
18. J. H. Song, K. Kim, Y. A. Leem, and G. Kim, “High-Power Broadband Superluminescent Diode Using Selective Area Growth at 1.5 μ m Wavelength,” IEEE Photon. Technol. Lett. 19(19), 1415–1417 (2007). [CrossRef]
19. C. K. Chia, S. J. Chua, J. R. Dong, and S. L. Teo, “Ultra-wide band quantum dot light emitting device by post- fabrication laser annealing,” Appl. Phys. Lett. 90(6), 061101 (2007). [CrossRef]
20. T. K. Ong, M. Yin, Z. Yu, Y. C. Chan, and Y. L. Lam, “High performance quantum well intermixed superluminescent diodes,” Meas. Sci. Technol. 15(8), 1591–1595 (2004). [CrossRef]
21. P. Bardella, M. Rossetti, and I. Montrosset, “Modeling of Broadband Chirped Quantum-Dot Super-Luminescent Diodes,” IEEE J. Sel. Top. Quantum Electron. 15, 785–791 (2009). [CrossRef]
22. H. D. Sun, R. Macaluso, M. D. Dawson, F. Robert, A. C. Bryce, J. H. Marsh, and H. Riechert, “Characterization of selective quantum well intermixing in 1.3 μm GaInNAs/GaAs structures,” J. Appl. Phys. 94(3), 1550–1556 (2003). [CrossRef]
23. L. Fu, P. Lever, H. H. Tan, C. Jagadish, P. Reece, and M. Gal, “Suppression of interdiffusion in InGaAsÕGaAs quantum dots using dielectric layer of titanium dioxide,” Appl. Phys. Lett. 82(16), 2613–2615 (2003). [CrossRef]
