Estimation of relative defect densities in InGaN laser diodes by induced absorption of photoexcited carriers

Defects are one of the most important factors influencing the optical properties of groups III–V nitride semiconductor materials and thereby their applicability to light-emitting diodes. In this paper, we demonstrate that it is possible to estimate the presence of defects in InGaN laser diodes by performing pump-probe measurements and observing the induced absorptions. We have confirmed that the induced absorption originates from defects by performing experiments in which the pump intensity is varied. We believe that our method provides a powerful tool for evaluating the optical quality of InGaN materials before processing them into device fabrications. ©2010 Optical Society of America OCIS codes: (320.7130) Ultrafast processes in condensed matter, including semiconductors; (190.5970) Semiconductor nonlinear optics including MQW. References and links 1. S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matsushita, Y. Sugimoto, and H. Kiyoku, “Room-temperature continuous-wave operation of InGaN multi-quantum-well structure laser diodes with a lifetime of 27 hours,” Appl. Phys. Lett. 70(11), 1417 (1997). 2. S. F. Chichibu, A. Uedono, T. Onuma, B. A. Haskell, A. Chakraborty, T. Koyama, P. T. Fini, S. Keller, S. P. Denbaars, J. S. Speck, U. K. Mishra, S. Nakamura, S. Yamaguchi, S. Kamiyama, H. Amano, I. Akasaki, J. Han, and T. Sota, “Origin of defect-insensitive emission probability in In-containing (Al,In,Ga)N alloy semiconductors,” Nat. Mater. 5(10), 810–816 (2006). 3. D. Morita, M. Yamamoto, K. Akaishi, K. Matoba, K. Yasutomo, Y. Kasai, M. Sano, S.Nagahama, and T. Mukai, “Watt-Class High-Output-Power 365 nm Ultraviolet Light-Emitting Diodes,” Jpn. J. Appl. Phys. 43(No. 9A), 5945–5950 (2004). 4. K. Y. Zang, Y. D. Wang, H. F. Liu, and S. J. Chua, “Structural and optical properties of InGaN/GaN multiple quantum wells grown on nano-air-bridged GaN template,” Appl. Phys. Lett. 89(17), 171921 (2006). 5. D. M. Shin, J. Park, D. H. Nguyen, Y. D. Jang, K. J. Yee, D. Lee, Y. H. Choi, S. K. Jung, and M. S. Noh, “A critical factor affecting on the performance of blue-violet InGaN multiquantum well laser diodes: Nonradiative centers,” Appl. Phys. Lett. 97(7), 071910 (2010). 6. X. H. Wu, L. M. Brown, D. Kapolnek, S. Keller, B. Keller, S. P. DenBaars, and J. S. Speck, “Defect structure of metal‐organic chemical vapor deposition‐grown epitaxial (0001) GaN/Al2O3,” J. Appl. Phys. 80(6), 3228 (1996). 7. V. Potin, P. Ruterana, and G. Nouet, “The atomic structure of {1010} inversion domain boundaries in GaN/sapphire layers,” J. Appl. Phys. 82(5), 2176 (1997). 8. A. Hierro, S. A. Ringel, M. Hansen, J. S. Speck, U. K. Mishra, and S. P. DenBaars, “Hydrogen passivation of deep levels in n–GaN,” Appl. Phys. Lett. 77(10), 1499 (2000). 9. H. K. Cho, J. Y. Lee, G. M. Yang, and C. S. Kim, “Formation mechanism of V defects in the InGaN/GaN multiple quantum wells grown on GaN layers with low threading dislocation density,” Appl. Phys. Lett. 79(2),


Introduction
Group III-V nitride semiconductors have attracted considerable interest as potentially applicable materials in high-brightness and high-power light-emitting devices (LEDs) that encompass a wide spectral range from the near infrared to the ultraviolet [1].Efficient light emission is possible in indium-containing (Al, In, Ga)N films even with considerable defects, which results from short diffusion lengths of carriers due to carrier localization into a potential minima formed by the high concentrations of InN clusters [2].Nevertheless, defect density is a critical factor for determining the performance of InGaN optical devices [3][4][5].
When InGaN materials are grown on the popularly employed Al 2 O 3 substrates, structural defects such as threading dislocations and stacking faults are unavoidable due to the large lattice mismatch [6,7].Although deep level transient spectroscopy (DLTS) can be used to measure the energies and relative densities of defect states while transmission electron microscopy (TEM) can reveal the microscopic structures of the defects [8,9], it is highly desirable to develop alternative measurement methods, especially nondestructive methods, that are sensitive to the defect states influencing the optical properties.
In this paper, we demonstrate that optical pump-probe spectroscopy can perform sensitive measurements of the defects that adversely affect the performance of the optical processes in InGaN laser diode (LD) devices.The amount of the induced absorption in transient transmission measurements was found to be proportional to the defect density.The correlation observed between the threshold lasing current in LD devices and the signal strength of the induced absorption supports the validity of our method.

Experiments
LD structures were grown on a c-axis GaN substrate by metal organic chemical vapor deposition.The targeted operation region was a blue-violet region at around 395 nm, and the active region is comprised of three periods of a 3-nm-thick InGaN QW and a 7-nm-thick In 0.02 Ga 0.98 N barrier.A 22-nm-thick electron blocking layer (EBL) of Al 0.17 Ga 0.83 N was inserted after the active region, and 125-nm-thick GaN layers were grown before and after the active and EBL regions, to guide photons along the active layer.Four LD samples, named LD-A, LD-B, LD-C, and LD-D with different threshold currents for lasing, were chosen for our experiments.Note that the lasing characteristics depended primarily on the growth conditions, and the nature of the doping of the waveguide region, n-type or p-type doped, did not have a significant impact.
The transient absorption induced by photoexcited carriers was measured at room temperature using femtosecond pulses having a center wavelength of 395 nm.The pulses were obtained by frequency doubling near-infrared pulses from a mode-locked Ti:sapphire oscillator.In the degenerate pump-probe technique, where pump and probe photons have the same energy, a pump pulse excites the carriers in the InGaN quantum wells (QWs), and the dynamic relaxation of the carriers is studied by detecting transmission changes in the probe beam as a function of the time delay relative to the pump pulse.In our experiments, the photo-generated carrier density was varied by changing the pump intensity from 0.3 to 11 μJ/cm 2 while the probe intensity was fixed at a smaller value of 0.2 μJ/cm 2 .Pump beam was modulated by a chopper, and the induced changes of the probe transmission were measured by using a lock-in technique.To minimize the coherent artifact signals occurring from nonlinear interactions between pump and probe photons near zero time-delay, probe polarization was set to be perpendicular to pump polarization.

Results and discussion
Figure 1 shows the changes in transmission as a function of time delay for the sample that exhibited the worst performance (LD-A); LD-A did not show lasing behaviors up to an applied current of 110 mA.The excitation intensity of the pump beam was 1.1 μJ/cm 2 , which corresponds to a photoexcited carrier density of 3.3 × 10 9 cm 2 if an absorption probability of 0.0015 per quantum well is assumed [10].When the number of electrons in the conduction band increases by the pump pulse, the absorption coefficient decreases in most cases due to the band-filling effect [11]; this is consistent with the initial positive changes in the probe transmission.Interestingly, the signal became negative within 100 ps, and the reduced transmission saturated at around 700 ps.
In order to explain the gradual appearance of a negative transmission change after photoexcitation, we consider the process depicted in the inset of Fig. 1.Defects or impurities generally introduce transition levels within the band gap of semiconductors [12].An example of the carrier transition to defect states is the yellow luminescence in GaN materials, which is possibly related with defect states of Ga vacancies (V Ga ) [13].When carriers are excited by pump photons, those carriers can be trapped into such defect states before returning to the valence band.There is a probability that a probe photon is absorbed by a trapped electron with exciting it into upper conduction band states.This kind of induced absorptions could be observed in pump-probe measurements for low-temperature grown GaAs materials with large defect densities [14,15].As more carriers excited by pump photons transit from the conduction band to the defect states, the transmission change of the probe beam will gradually shift from positive values (due to band-filling effects) to negative values (due to the induced absorption by the trapped electrons).
The dynamics of carrier trapping into defect states were examined by analyzing the temporal evolution of the transmission changes.The decrease in the signal between 4 ps and 1000 ps was successfully fit by employing a stretched exponential function I(t) = I 0 exp[-(t/τ) β ], which is indicated by the dashed red line.The effective trapping time τ = 129 ps and the stretching parameter β = 0.71 were obtained.The stretched exponential function has been previously demonstrated to be useful in describing carrier dynamics in disordered systems in which recombination or scattering centers are randomly distributed [16,17].In InGaN materials, randomly distributed potential minima originate from non-uniform In concentrations.Because photoexcited electrons become localized at these potential minima before the carriers are trapped in the defects, the carrier trapping in the defects will occur while the electrons are drifting between localization centers.The negative transmission change decays slowly with a decay time longer than 2 ns, and this corresponds to the recombination dynamics of the trapped electrons either by emitting luminescences or nonradiative processes [18].The density of the available defects which can trap electrons will differ depending on sample quality.Since the probability of a carrier to be trapped into defects will be proportional with the number of available empty defect states, the trapping dynamics will vary with the occupation of the defect states by electrons.Thus, an excitation-density dependent trapping dynamics is expected, which has a similar origin as observed in other defect-related phenomena showing saturation behaviors at high excitation intensities [19].Figure 2 shows the dependence of the pump-probe signal on the pump intensity for the LD-A sample.The initial positive signal was proportional to the pump intensity, as is expected from the constant absorption coefficient in the low excitation regime.On the other hand, the strength of the maximum induced absorption shows saturation behavior at an excitation density of around 4.5 μJ/cm 2 , as shown in the inset of Fig. 2.
Since carrier trapping in defects occurs considerably faster than the recombination of band electrons, most electrons will be trapped in the defects if the number of photoexcited electrons is smaller than the capacity of the defect states.However, if the photo-generated carrier density is sufficiently large, the excess electrons that remain after the filling of the available defect states will occupy the states at the conduction band, until they recombine with holes or transit into empty defect states.As long as the defect states remain full of trapped electrons, the strength of the induced absorption will remain at a maximum value.The slow decrease in the positive signal observed at high excitation intensities resulted from a combination of the two contributions: positive contribution due to remaining band electrons and negative one due to the induced absorption.The slow transition to negative values at high intensities represents the longer residence time of electrons in the conduction band when the defect states are occupied.The maximum negative signals at high intensities will occur at the time delay for which most band electrons has just relaxed to defect states.The saturation of the maximum negative signal with pump intensity, thus, confirms that the negative signal originates from the defect states.Note that in optical devices, the density of the defects that participate in the carrier trapping processes is fairly important, and its values can be estimated from the excitation density that saturates the induced absorption.In addition to the LD-A sample, in which lasing was unsuccessful possibly due to an abundance of defects, we tested three other successful lasing LD samples with different threshold currents.Figure 3(a) shows the curves for the output power vs. current and it indicates that the threshold currents of the LD-B, LD-C, and LD-D samples were 89 mA, 53 mA, and 40 mA, respectively.The lasing spectrum of the LD-D sample, which was obtained at an applied current of 60 mA, is displayed in the inset.The abundance of defects in LD-A sample as compared to other samples could also be verified from the strong defect-related yellow luminescence at around 2.15 eV in electroluminescence measured at a very low applied current of 0.5 mA.As shown in Fig. 3(b), the pump-probe signals obtained at a pump intensity of 4.5 μJ/cm 2 differed considerably depending on the lasing characteristics.The induced absorption with negative transmission changes is clearly observable in the LD-A and LD-B layers with high threshold currents.In contrast, the transmission changes remain positive until 2000 ps in the LD-C and LD-D devices with low threshold currents, which indicate that the defect densities are considerably smaller.
A comparison between the device performance and the time-resolved pump-probe signals of the samples clearly demonstrates that the defect density is a critical factor for determining the quality of the InGaN optical devices; furthermore, the defect densities can be estimated using pump-probe measurements by analyzing the transient transmission.In this manner, the pump-probe method, which is sensitive to the defect densities influencing in the optical processes, can be used to examine InGaN wafers and determine if the layer will perform as designed when incorporated into LDs or LEDs.We note that the signal decay in timeresolved photoluminescence will also depend on defect densities, because of the non-radiative decay rate related with defect states.However, our pump-probe technique can be more definitive in that the transmission change shows different signs depending on whether the electrons are trapped at defect states or the electrons are at the conduction band.

Conclusion
In this study, negative transmission changes due induced absorptions by photoexcited carriers trapped in defect states were observed in pump-probe measurements performed for InGaNbased LD structures.The excited carriers in one LD sample (LD-A), which presumably had a relatively large defect density, were trapped in the defects with an effective lifetime of 129 ps.The strength of the induced absorption saturated as the excitation density increased, thereby confirming that induced absorption is related to the defects.The device performance of the LDs is closely related to the defect density, and we have shown that the performance can thus be deduced from the induced absorption signal.In this manner, our method can be a powerful tool for evaluating the optical quality of InGaN materials before processing them into devices.

Fig. 1 .
Fig. 1.Transmission changes as a function of time delay in an InGaN laser diode sample (LD-A), which did not successfully produce lasing operation.Pump excitation density was 1.1 μJ/cm 2 .The red line indicates a stretched exponential fit to the curve between 4 ps and 1000 ps.The inset shows a schematic of the absorption process of probe photons, which is induced by pump-excited carriers trapped in defect states.

Fig. 2 .
Fig. 2. Time-resolved transmission changes at several pump intensities (up to 11 μJ/cm 2 ).The inset shows the value of the negative maximum for each curve as a function of the pump intensity.

Fig. 3 .
Fig. 3. (a) Output power vs. applied current curves for several LD devices having different threshold currents.The inset shows the lasing spectrum for LD-D device having a threshold current of 40 mA.(b) Comparison of the time-resolved transmission changes for different LDs measured at a pump intensity of 1.1 μJ/cm 2 .