- Email: [email protected]
Ultrafast carrier trapping in Er-doped and Er,O-codoped GaAs revealed by pump and probe technique
ARTICLE IN PRESS
Physica B 401–402 (2007) 234–237 www.elsevier.com/locate/physb
Ultrafast carrier trapping in Er-doped and Er,O-codoped GaAs revealed by pump and probe technique Y. Fujiwaraa,, S. Takemotoa, K. Nakamuraa, K. Shimadaa, M. Suzukib, K. Hidakaa, Y. Teraia, M. Tonouchib a
Division of Materials and Manufacturing Science, Graduate School of Engineering, Osaka University 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan b Institute of Laser Engineering, Osaka University 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
Abstract Dynamics of photoexcited carriers in Er-doped GaAs (GaAs:Er) and Er,O-codoped GaAs (GaAs:Er,O) have been systematically investigated by means of a pump and probe technique. A characteristic ps-scale relaxation was clearly observed in transient reflectance (dR/R) and transmission (dT/T) curves. The relaxation was due to the trapping of photoexcited carriers by an Er-related trap, which was closely related to efficiency of Er3+ luminescence. r 2007 Published by Elsevier B.V. Keywords: GaAs; Erbium; Carrier trapping; Ultrafast phenomena; Pump and probe technique
1. Introduction Much attention has been paid to rare-earth (RE) doped semiconductors as a promising new class of materials that emit light from the RE 4f shell by means of electrical injection. The intra-4f shell transitions of RE ions give rise to sharp emission lines whose wavelengths are largely independent of both the host materials and temperature. This stability occurs because the filled outer 5s and 5p electron shells screen transitions within the inner 4f electron shell from the interaction with the host. The intra-4f shell transitions from the first excited state (4I13/2) to the ground state (4I15/2) of Er3+ ions at around 1.5 mm are of special interest because the wavelength matches the minimum loss region of silica fibers used in optical communications. We have intensively investigated organometallic vapor phase epitaxy (OMVPE) and luminescence properties of Er-doped III–V semiconductors [1]. In Er,O-codoped GaAs (GaAs:Er,O), the Er-related photoluminescence (PL) spectrum was dominated by sharp emission lines under hostexcited conditions at a low temperature [2]. The Er center has been identified as an Er atom located at the Ga sub-lattice Corresponding author. Tel.: +81 6 6879 7498; fax: +81 6 6879 7499.
E-mail address: [email protected] (Y. Fujiwara). 0921-4526/$ - see front matter r 2007 Published by Elsevier B.V. doi:10.1016/j.physb.2007.08.155
with two adjacent O atoms (hereafter referred as Er–2O) together with two As atoms [3]. Recently, we have fabricated GaAs:Er,O homostructure and GaInP/GaAs:Er,O/GaInP double heterostructure (DH) light emitting diodes (LEDs) by OMVPE and successfully observed 1.5 mm electroluminescence (EL) due to an Er-2O center under forward bias at room temperature [4–6]. The dependence of the EL intensity on the injection current density indicated extremely large excitation cross-section of Er ions by current injection (1–2 1015 cm2) in the LEDs [4]. However, the Er excitation cross-section decreased with increasing active layer thickness in the DH LEDs, suggesting reduced diffusion lengths of injected carriers in a GaAs:Er,O active layer [7]. In this article, we investigated nonequilibrium carrier dynamics in OMVPE-grown GaAs:Er,O as well as Erdoped GaAs (GaAs:Er), which was revealed by a pump and probe technique. 2. Experimental 2.1. Sample preparation The samples were grown on (0 0 1) semi-insulating GaAs substrates by low-pressure OMVPE. The details of sample preparation were previously described [4].
ARTICLE IN PRESS Y. Fujiwara et al. / Physica B 401–402 (2007) 234–237
For pump and probe transmission measurements, the samples with the thickness of 1 mm were grown on a 0.15 mm GaInP etch-stop layer, followed by chemical etching of the GaAs substrate with H3PO4:H2O2: H2O ¼ 3:4:4. The Er concentration of the samples was less than 1017 cm3, which was below the detection limit of secondary ion mass spectroscopy (SIMS) measurements. 2.2. Pump and probe measurements Dynamics of nonequilibrium carriers in GaAs:Er,O and GaAs:Er have been studied directly by pump and probe reflection and transmission measurements at room temperature. A mode-locked Ti:sapphire laser was used for fs pulses with a pulse width of about 100 fs, a center wavelength of about 840 nm, and a repetition rate of 82 MHz. Optical pulses from the laser were divided to two beams; the first beam excited the sample as a pump pulse, and the second beam was used as a probe pulse. The power of the pump and probe pulses were fixed at 30 and 1 mW, respectively. The pump and probe pulses polarized perpendicularly were focused with a diameter of about 10 mm on the sample surface. The transient of reflectivity and transmittance was monitored as a function of the relative time delay between pump and probe pulses. 3. Results and discussion 3.1. Time-resolved reflectivity Fig. 1 shows time-resolved reflectivity in two GaAs:Er,O samples (high-concentration sample and low-concentration sample). Er and O concentrations are 9 1018 and
Fig. 1. Time-resolved reflectivity for GaAs:Er,O and GaAs:Er. The result for undoped GaAs is also shown for comparison.
235
3 1018 cm3 in the high-concentration sample, and 8 1017 and 9 1017 cm3 in the low-concentration sample, respectively. The results in nominally undoped GaAs and GaAs:Er (Er concentration: 1 1018 cm3) are also shown for comparison. Time-resolved reflectivity in the undoped GaAs is governed by two relaxation components: a fast component (not shown in the figure) and a slow component. They correspond to the carrier transit lifetime from high-energy level to band edge, and the carrier recombination lifetime of approximately 1 ns, respectively. On the other hand, the reflectivity of GaAs:Er,O reveals an abrupt increase in amplitude, followed by a steep decrease down to negative in less than 1 ps and then a gradual increase in more than 100 ps [8]. The steep decrease is due to bandgap renormalization produced by a large number of nonequilibrium carriers. The recovery from the negative minimum in the time-resolved reflectivity depends strongly on Er concentration. In the high-concentration sample, a rapid recovery is clearly observed at the initial stage. We have investigated quantitatively the initial recovery observed in GaAs:Er,O. The experimental data were changed to be absolute values and subtracted by a slow background due to carrier recombination between conduction band and valence band. It has been found that there are two components, fast component tfast at the initial and then slow component tslow. The Er concentration dependence of tfast is shown by closed circles in Fig. 2. Experimental results for other samples not cited in this article are also plotted together in the figure. The recovery time is almost inversely proportional to Er concentration.
Fig. 2. Er concentration dependence of tfast in GaAs:Er,O. The results calculated using Eq. (1) are shown by solid lines.
ARTICLE IN PRESS 236
Y. Fujiwara et al. / Physica B 401–402 (2007) 234–237
Fig. 3. Mechanism of excitation and relaxation of Er3+ ions in GaAs.
In discussing this behavior, we should remind an energytransfer model recognized generally for excitation of RE ions doped in semiconductors. The model is schematically shown in Fig. 3. In the model, a doped RE atom produces a bound state like an isoelectronic trap in the bandgap. If the trap is a donor-like hole trap, it captures a photoexcited hole to be positively charged. Then the trap attracts an electron by Coulombic interaction and creates an electron–hole pair, or an exciton at a low temperature. The electron–hole pair recombines and part of the recombination energy is transferred to the Er 4f shell, resulting in a 4f shell excitation. In GaAs:Er,O, a trap formed by Er and O doping is assumed to behave as a hole trap and capture a photoexcited hole very quickly. The quick capture process is supported experimentally by the observation of negative reflectivity difference in less than 1 ps, which is assigned to bandgap renormalization. Capture time t of an electron by the positively charged trap can be described as follows: t ¼ ðsn vth N Er Þ1 ,
(1)
where vth is the thermal velocity of free electrons, given by ð3kT=me Þ1=2 , me is electron effective mass, and sn is the electron capture cross-section. NEr is the trap density, corresponding to Er concentration. In Fig. 2, solid lines are the calculated one using Eq. (1) as a parameter of sn. The best fitting is obtained when sn is 6 1016 cm2. In order to confirm this conclusion, we have investigated temperature dependence of time-resolved reflectivity in the high-concentration sample. tfast determined at each temperature is shown in Fig. 4. Solid lines also indicate the calculated ones using Eq. (1). sn of 6 1016 cm2 explains again relatively well experimental results, which supports our conclusion that the tfast corresponds to the capture time of electrons by positively charged trap induced by Er and O codoping. The origin of tslow is not clear at present. Considering multi-phonon mechanism proposed for the energy-transfer process from GaAs host to Er ions [9], the energy-transfer time is calculated to be 54 ps, which agrees well to experimentally determined tslow (30–60 ps). It suggests that
Fig. 4. Temperature dependence of tfast in GaAs:Er,O. The results calculated using Eq. (1) are shown by solid lines.
the tslow corresponds to the energy-transfer from the GaAs host to 4f shell of Er ions. This was also supported by the observation that the slow recovery process is quite ambiguous at low temperatures. In GaAs:Er, the behavior is quite different from undoped GaAs and GaAs:Er,O. As seen in GaAs:Er,O, the negative reflectivity difference due to the bandgap renormalization is also observed at the initial stage. The two recovery processes discussed in GaAs:Er,O, however, are obscure. This might be due to the coexistence of several Er traps, producing multiple paths for the excitation of Er3+ ions. 3.2. Time-resolved transmission In order to obtain further information, we have performed pump and probe transmission measurements
ARTICLE IN PRESS Y. Fujiwara et al. / Physica B 401–402 (2007) 234–237
237
the initial stage depended strongly on Er concentration. Exponential fitting to the initial increase revealed two recovery components, fast component at the first and successively slow component. The recovery time for the fast component was almost inversely proportional to Er concentration. This indicated that the initial recovery is due to the capture of nonequilibrium carriers by a trap induced by Er and O codoping. The recovery time for the slow component, on the other hand, was well comparable to the energy-transfer time from GaAs host to Er ions, which was calculated considering the multi-phonon mechanism. In GaAs:Er, the two recovery processes were obscure after the negative reflectivity difference due to the bandgap renormalization. In pump and probe transmission measurements, time-resolved transmission has been successfully obtained, which was initiated by the intra-band relaxation and followed by the inter-band relaxation. In GaAs:Er,O and GaAs:Er, an additional relaxation process was observed and assigned to the capture of photoexcited electrons by a positively charged Er-related trap. Fig. 5. Time-resolved transmission for GaAs:Er,O and GaAs:Er, which is compared with the result for undoped GaAs.
on GaAs:Er,O and GaAs:Er with extremely low Er concentration. Fig. 5 shows time-resolved transmission for the samples, which is compared with that for undoped GaAs. In all the samples, intra-band relaxation in sub-psscale is followed by inter-band relaxation. In GaAs:Er and GaAs:Er,O, however, an additional relaxation is clearly observed within 10 ps as shown in the inset of Fig. 5. In the inset, the transient transmission is subtracted by the component of the intra-band relaxation. The additional relaxation was not observed in undoped GaAs. The relaxation times are 4 ps in GaAs:Er,O and 11 ps in GaAs:Er. By considering the energy-transfer model shown in Fig. 3, the additional relaxation could be assigned to the capture of photoexcited electrons by a positively charged Er-related trap. We have also investigated the relationship between the relaxation time and the Er-related luminescence intensity. It has been found that the time decreases in the samples with increasing Er3+ luminescence, indicating that efficient carrier-trapping is quite important for excitation of Er3+ ions. 4. Conclusion We have investigated carrier dynamics in GaAs:Er,O and GaAs:Er by a pump and probe technique. Timeresolved reflectivity in GaAs:Er,O exhibited a characteristic dip; a steep decrease to negative in less than 1 ps and then an increase in more than 100 ps. The reflectivity increase at
Acknowledgments This work was supported in part by Grant-in-Aid for Creative Scientific Research No. 19GS1209 from the Japan Society for the Promotion of Science. This work was also supported in part by Grant-in-Aids for Exploratory Research No. 19656082 for Scientific Research of Priority Areas ‘‘Panoscopic Assembling and Highly Ordered Functions for Rare Earth Materials’’ Nos. 17042016 and 19018014, and the Global COE Program ‘‘Advanced Structural and Functional Materials Design’’ from the Ministry of Education, Culture, Sports, Science and Technology. References [1] For example, Y. Fujiwara, H. Ofuchi, M. Tabuchi, Y. Takeda, in: M.O. Manasreh (Ed.), InP and Related Compounds—Materials, Applications and Devices—Optoelectronic Properties of Semiconductors and Superlattices, vol. 9, Gordon & Breach Science Publisher, The Netherlands, 2000, pp. 251–311. [2] K. Takahei, A. Taguchi, J. Appl. Phys. 74 (1993) 1979. [3] K. Takahei, A. Taguchi, Y. Horikoshi, J. Nakata, J. Appl. Phys. 76 (1994) 4332. [4] A. Koizumi, Y. Fujiwara, K. Inoue, A. Urakami, T. Yoshikane, Y. Takeda, Jpn. J. Appl. Phys. 42 (2003) 2223. [5] A. Koizumi, Y. Fujiwara, A. Urakami, K. Inoue, T. Yoshikane, Y. Takeda, Appl. Phys. Lett. 83 (2003) 4521. [6] Y. Fujiwara, A. Koizumi, A. Urakami, T. Yoshikane, K. Inoue, Y. Takeda, Mater. Sci. Eng. B 105 (1–3) (2003) 57. [7] A. Koizumi, Y. Fujiwara, A. Urakami, K. Inoue, T. Yoshikane, Y. Takeda, Physica B 340–342 (2003) 309. [8] K. Nakamura, S. Takemoto, Y. Terai, M. Suzuki, A. Koizumi, Y. Takeda, M. Tonouchi, Y. Fujiwara, Physica B 376/377 (2006) 556. [9] A. Taguchi, K. Takahei, J. Appl. Phys. 79 (1996) 4330.
Physica B 401–402 (2007) 234–237 www.elsevier.com/locate/physb
Ultrafast carrier trapping in Er-doped and Er,O-codoped GaAs revealed by pump and probe technique Y. Fujiwaraa,, S. Takemotoa, K. Nakamuraa, K. Shimadaa, M. Suzukib, K. Hidakaa, Y. Teraia, M. Tonouchib a
Division of Materials and Manufacturing Science, Graduate School of Engineering, Osaka University 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan b Institute of Laser Engineering, Osaka University 2-1 Yamadaoka, Suita, Osaka 565-0871, Japan
Abstract Dynamics of photoexcited carriers in Er-doped GaAs (GaAs:Er) and Er,O-codoped GaAs (GaAs:Er,O) have been systematically investigated by means of a pump and probe technique. A characteristic ps-scale relaxation was clearly observed in transient reflectance (dR/R) and transmission (dT/T) curves. The relaxation was due to the trapping of photoexcited carriers by an Er-related trap, which was closely related to efficiency of Er3+ luminescence. r 2007 Published by Elsevier B.V. Keywords: GaAs; Erbium; Carrier trapping; Ultrafast phenomena; Pump and probe technique
1. Introduction Much attention has been paid to rare-earth (RE) doped semiconductors as a promising new class of materials that emit light from the RE 4f shell by means of electrical injection. The intra-4f shell transitions of RE ions give rise to sharp emission lines whose wavelengths are largely independent of both the host materials and temperature. This stability occurs because the filled outer 5s and 5p electron shells screen transitions within the inner 4f electron shell from the interaction with the host. The intra-4f shell transitions from the first excited state (4I13/2) to the ground state (4I15/2) of Er3+ ions at around 1.5 mm are of special interest because the wavelength matches the minimum loss region of silica fibers used in optical communications. We have intensively investigated organometallic vapor phase epitaxy (OMVPE) and luminescence properties of Er-doped III–V semiconductors [1]. In Er,O-codoped GaAs (GaAs:Er,O), the Er-related photoluminescence (PL) spectrum was dominated by sharp emission lines under hostexcited conditions at a low temperature [2]. The Er center has been identified as an Er atom located at the Ga sub-lattice Corresponding author. Tel.: +81 6 6879 7498; fax: +81 6 6879 7499.
E-mail address: [email protected] (Y. Fujiwara). 0921-4526/$ - see front matter r 2007 Published by Elsevier B.V. doi:10.1016/j.physb.2007.08.155
with two adjacent O atoms (hereafter referred as Er–2O) together with two As atoms [3]. Recently, we have fabricated GaAs:Er,O homostructure and GaInP/GaAs:Er,O/GaInP double heterostructure (DH) light emitting diodes (LEDs) by OMVPE and successfully observed 1.5 mm electroluminescence (EL) due to an Er-2O center under forward bias at room temperature [4–6]. The dependence of the EL intensity on the injection current density indicated extremely large excitation cross-section of Er ions by current injection (1–2 1015 cm2) in the LEDs [4]. However, the Er excitation cross-section decreased with increasing active layer thickness in the DH LEDs, suggesting reduced diffusion lengths of injected carriers in a GaAs:Er,O active layer [7]. In this article, we investigated nonequilibrium carrier dynamics in OMVPE-grown GaAs:Er,O as well as Erdoped GaAs (GaAs:Er), which was revealed by a pump and probe technique. 2. Experimental 2.1. Sample preparation The samples were grown on (0 0 1) semi-insulating GaAs substrates by low-pressure OMVPE. The details of sample preparation were previously described [4].
ARTICLE IN PRESS Y. Fujiwara et al. / Physica B 401–402 (2007) 234–237
For pump and probe transmission measurements, the samples with the thickness of 1 mm were grown on a 0.15 mm GaInP etch-stop layer, followed by chemical etching of the GaAs substrate with H3PO4:H2O2: H2O ¼ 3:4:4. The Er concentration of the samples was less than 1017 cm3, which was below the detection limit of secondary ion mass spectroscopy (SIMS) measurements. 2.2. Pump and probe measurements Dynamics of nonequilibrium carriers in GaAs:Er,O and GaAs:Er have been studied directly by pump and probe reflection and transmission measurements at room temperature. A mode-locked Ti:sapphire laser was used for fs pulses with a pulse width of about 100 fs, a center wavelength of about 840 nm, and a repetition rate of 82 MHz. Optical pulses from the laser were divided to two beams; the first beam excited the sample as a pump pulse, and the second beam was used as a probe pulse. The power of the pump and probe pulses were fixed at 30 and 1 mW, respectively. The pump and probe pulses polarized perpendicularly were focused with a diameter of about 10 mm on the sample surface. The transient of reflectivity and transmittance was monitored as a function of the relative time delay between pump and probe pulses. 3. Results and discussion 3.1. Time-resolved reflectivity Fig. 1 shows time-resolved reflectivity in two GaAs:Er,O samples (high-concentration sample and low-concentration sample). Er and O concentrations are 9 1018 and
Fig. 1. Time-resolved reflectivity for GaAs:Er,O and GaAs:Er. The result for undoped GaAs is also shown for comparison.
235
3 1018 cm3 in the high-concentration sample, and 8 1017 and 9 1017 cm3 in the low-concentration sample, respectively. The results in nominally undoped GaAs and GaAs:Er (Er concentration: 1 1018 cm3) are also shown for comparison. Time-resolved reflectivity in the undoped GaAs is governed by two relaxation components: a fast component (not shown in the figure) and a slow component. They correspond to the carrier transit lifetime from high-energy level to band edge, and the carrier recombination lifetime of approximately 1 ns, respectively. On the other hand, the reflectivity of GaAs:Er,O reveals an abrupt increase in amplitude, followed by a steep decrease down to negative in less than 1 ps and then a gradual increase in more than 100 ps [8]. The steep decrease is due to bandgap renormalization produced by a large number of nonequilibrium carriers. The recovery from the negative minimum in the time-resolved reflectivity depends strongly on Er concentration. In the high-concentration sample, a rapid recovery is clearly observed at the initial stage. We have investigated quantitatively the initial recovery observed in GaAs:Er,O. The experimental data were changed to be absolute values and subtracted by a slow background due to carrier recombination between conduction band and valence band. It has been found that there are two components, fast component tfast at the initial and then slow component tslow. The Er concentration dependence of tfast is shown by closed circles in Fig. 2. Experimental results for other samples not cited in this article are also plotted together in the figure. The recovery time is almost inversely proportional to Er concentration.
Fig. 2. Er concentration dependence of tfast in GaAs:Er,O. The results calculated using Eq. (1) are shown by solid lines.
ARTICLE IN PRESS 236
Y. Fujiwara et al. / Physica B 401–402 (2007) 234–237
Fig. 3. Mechanism of excitation and relaxation of Er3+ ions in GaAs.
In discussing this behavior, we should remind an energytransfer model recognized generally for excitation of RE ions doped in semiconductors. The model is schematically shown in Fig. 3. In the model, a doped RE atom produces a bound state like an isoelectronic trap in the bandgap. If the trap is a donor-like hole trap, it captures a photoexcited hole to be positively charged. Then the trap attracts an electron by Coulombic interaction and creates an electron–hole pair, or an exciton at a low temperature. The electron–hole pair recombines and part of the recombination energy is transferred to the Er 4f shell, resulting in a 4f shell excitation. In GaAs:Er,O, a trap formed by Er and O doping is assumed to behave as a hole trap and capture a photoexcited hole very quickly. The quick capture process is supported experimentally by the observation of negative reflectivity difference in less than 1 ps, which is assigned to bandgap renormalization. Capture time t of an electron by the positively charged trap can be described as follows: t ¼ ðsn vth N Er Þ1 ,
(1)
where vth is the thermal velocity of free electrons, given by ð3kT=me Þ1=2 , me is electron effective mass, and sn is the electron capture cross-section. NEr is the trap density, corresponding to Er concentration. In Fig. 2, solid lines are the calculated one using Eq. (1) as a parameter of sn. The best fitting is obtained when sn is 6 1016 cm2. In order to confirm this conclusion, we have investigated temperature dependence of time-resolved reflectivity in the high-concentration sample. tfast determined at each temperature is shown in Fig. 4. Solid lines also indicate the calculated ones using Eq. (1). sn of 6 1016 cm2 explains again relatively well experimental results, which supports our conclusion that the tfast corresponds to the capture time of electrons by positively charged trap induced by Er and O codoping. The origin of tslow is not clear at present. Considering multi-phonon mechanism proposed for the energy-transfer process from GaAs host to Er ions [9], the energy-transfer time is calculated to be 54 ps, which agrees well to experimentally determined tslow (30–60 ps). It suggests that
Fig. 4. Temperature dependence of tfast in GaAs:Er,O. The results calculated using Eq. (1) are shown by solid lines.
the tslow corresponds to the energy-transfer from the GaAs host to 4f shell of Er ions. This was also supported by the observation that the slow recovery process is quite ambiguous at low temperatures. In GaAs:Er, the behavior is quite different from undoped GaAs and GaAs:Er,O. As seen in GaAs:Er,O, the negative reflectivity difference due to the bandgap renormalization is also observed at the initial stage. The two recovery processes discussed in GaAs:Er,O, however, are obscure. This might be due to the coexistence of several Er traps, producing multiple paths for the excitation of Er3+ ions. 3.2. Time-resolved transmission In order to obtain further information, we have performed pump and probe transmission measurements
ARTICLE IN PRESS Y. Fujiwara et al. / Physica B 401–402 (2007) 234–237
237
the initial stage depended strongly on Er concentration. Exponential fitting to the initial increase revealed two recovery components, fast component at the first and successively slow component. The recovery time for the fast component was almost inversely proportional to Er concentration. This indicated that the initial recovery is due to the capture of nonequilibrium carriers by a trap induced by Er and O codoping. The recovery time for the slow component, on the other hand, was well comparable to the energy-transfer time from GaAs host to Er ions, which was calculated considering the multi-phonon mechanism. In GaAs:Er, the two recovery processes were obscure after the negative reflectivity difference due to the bandgap renormalization. In pump and probe transmission measurements, time-resolved transmission has been successfully obtained, which was initiated by the intra-band relaxation and followed by the inter-band relaxation. In GaAs:Er,O and GaAs:Er, an additional relaxation process was observed and assigned to the capture of photoexcited electrons by a positively charged Er-related trap. Fig. 5. Time-resolved transmission for GaAs:Er,O and GaAs:Er, which is compared with the result for undoped GaAs.
on GaAs:Er,O and GaAs:Er with extremely low Er concentration. Fig. 5 shows time-resolved transmission for the samples, which is compared with that for undoped GaAs. In all the samples, intra-band relaxation in sub-psscale is followed by inter-band relaxation. In GaAs:Er and GaAs:Er,O, however, an additional relaxation is clearly observed within 10 ps as shown in the inset of Fig. 5. In the inset, the transient transmission is subtracted by the component of the intra-band relaxation. The additional relaxation was not observed in undoped GaAs. The relaxation times are 4 ps in GaAs:Er,O and 11 ps in GaAs:Er. By considering the energy-transfer model shown in Fig. 3, the additional relaxation could be assigned to the capture of photoexcited electrons by a positively charged Er-related trap. We have also investigated the relationship between the relaxation time and the Er-related luminescence intensity. It has been found that the time decreases in the samples with increasing Er3+ luminescence, indicating that efficient carrier-trapping is quite important for excitation of Er3+ ions. 4. Conclusion We have investigated carrier dynamics in GaAs:Er,O and GaAs:Er by a pump and probe technique. Timeresolved reflectivity in GaAs:Er,O exhibited a characteristic dip; a steep decrease to negative in less than 1 ps and then an increase in more than 100 ps. The reflectivity increase at
Acknowledgments This work was supported in part by Grant-in-Aid for Creative Scientific Research No. 19GS1209 from the Japan Society for the Promotion of Science. This work was also supported in part by Grant-in-Aids for Exploratory Research No. 19656082 for Scientific Research of Priority Areas ‘‘Panoscopic Assembling and Highly Ordered Functions for Rare Earth Materials’’ Nos. 17042016 and 19018014, and the Global COE Program ‘‘Advanced Structural and Functional Materials Design’’ from the Ministry of Education, Culture, Sports, Science and Technology. References [1] For example, Y. Fujiwara, H. Ofuchi, M. Tabuchi, Y. Takeda, in: M.O. Manasreh (Ed.), InP and Related Compounds—Materials, Applications and Devices—Optoelectronic Properties of Semiconductors and Superlattices, vol. 9, Gordon & Breach Science Publisher, The Netherlands, 2000, pp. 251–311. [2] K. Takahei, A. Taguchi, J. Appl. Phys. 74 (1993) 1979. [3] K. Takahei, A. Taguchi, Y. Horikoshi, J. Nakata, J. Appl. Phys. 76 (1994) 4332. [4] A. Koizumi, Y. Fujiwara, K. Inoue, A. Urakami, T. Yoshikane, Y. Takeda, Jpn. J. Appl. Phys. 42 (2003) 2223. [5] A. Koizumi, Y. Fujiwara, A. Urakami, K. Inoue, T. Yoshikane, Y. Takeda, Appl. Phys. Lett. 83 (2003) 4521. [6] Y. Fujiwara, A. Koizumi, A. Urakami, T. Yoshikane, K. Inoue, Y. Takeda, Mater. Sci. Eng. B 105 (1–3) (2003) 57. [7] A. Koizumi, Y. Fujiwara, A. Urakami, K. Inoue, T. Yoshikane, Y. Takeda, Physica B 340–342 (2003) 309. [8] K. Nakamura, S. Takemoto, Y. Terai, M. Suzuki, A. Koizumi, Y. Takeda, M. Tonouchi, Y. Fujiwara, Physica B 376/377 (2006) 556. [9] A. Taguchi, K. Takahei, J. Appl. Phys. 79 (1996) 4330.