978
OPTICS LETTERS / Vol. 31, No. 7 / April 1, 2006
Terahertz wave generation and detection from a CdTe crystal characterized by different excitation wavelengths Xu Xie, Jingzhou Xu, and X.-C. Zhang Center for Terahertz Research, Rensselaer Polytechnic Institute, Troy, New York 12180 Received October 4, 2005; revised December 16, 2005; accepted December 20, 2005; posted January 4, 2006 (Doc. ID 65172) Terahertz (THz) wave generation and detection from a 具110典-oriented CdTe crystal via optical rectification and electro-optic sampling has been performed with laser wavelengths ranging from 710 to 970 nm. Three optical rectification regimes are studied with various excitation wavelengths: a resonance-enhanced regime above the bandgap, a highly phase-mismatched regime near the band gap, and a semi-phase-matched regime. A CdTe crystal generates more THz power than a ZnTe crystal at 970 nm. © 2006 Optical Society of America OCIS codes: 190.4400, 320.7110.
Recent advances in optical rectification and free space electro-optic (EO) sampling methods have boosted terahertz (THz) science and technology applied in various fields including material characterization, nondestructive evaluation, imaging, and pharmaceuticals.1 In order to obtain a high efficiency of optical rectification and an EO sampling for higher THz power generation or more sensitive detection, long interaction lengths are crucial. The phasematching condition in optical rectification and EO sampling processes is satisfied when the group velocity of optical laser pulses equals the phase velocity of THz pulse in the nonlinear crystal. Thus it is of importance to fully understand how the phase match– mismatch affects the THz wave generation or detection in an optical rectification or EO sampling process. A direct way is to study THz generation or detection in crystals at different excitation optical wavelengths. The phase-mismatch behavior of ZnTe at near 800 and 1000 nm has been studied.2,3 Currently most pulsed THz systems are based on Ti:sapphire laser systems with a central wavelength of 800 nm. However, the large size and low efficiency of Ti:sapphire lasers prevents them from being used in a compact and portable THz system. Study of THz emitters and detectors at about 1550 and 1000 nm, which are widely used Er- and Yb-doped fiber laser wavelengths to build compact THz systems, has become attractive. THz wave generation using a 1550 nm laser system has been realized with magneticinduced InAs and InSb4,5 and optical rectification through GaAs.6 CdTe was one of the first zinc-blende structure crystals studied for THz emission and detection at 800 nm,7 and its physical properties in THz region have also been studied.8,9 In this Letter we report the study of CdTe as a THz emitter and detector via optical rectification and EO sampling. The laser wavelengths range from 710 to 970 nm, which covers three different wavelength regimes: a resonance-enhanced regime above the bandgap, a highly phasemismatched regime near the bandgap, and a semiphase-match regime. The experimental results at a 970 nm excitation wavelength indicate that CdTe 0146-9592/06/070978-3/$15.00
crystals are a good candidate for THz wave emission and detection with Yb-doped fiber lasers. In the process of optical rectification to generate THz radiation, coherence length is defined as c lc =
THz兩nTHz − ng兩
,
共1兲
where THz is the frequency of the THz wave, nTHz is the refractive index of the emitter crystal in the THz region, and ng is the group index of optical pulses, which is expressed by ng = nopt − opt共nopt / opt兲. Equation (1) shows that the maximum coherence length is where the refractive index of THz radiation equals the group index of the laser pulses. Using 3.23 as the refractive index of CdTe at 1 THz,10 calculation shows that at 1050 nm CdTe crystals have maximized coherence length. This indicates that CdTe is suitable to be used to generate and detect THz waves with 1000 nm wavelength lasers. The experimental setup for THz generation is based on Michelson interferometer geometry as depicted in Fig. 1. In order to investigate only THz wave generation, the influence of the wavelength de-
Fig. 1. Layout of the experimental setup, which is based on a Michelson interferometer geometry. The ultrafast laser beam is split into two parts and the recombined beams are focused on the CdTe crystal. Two parabolic mirrors are used to collect THz radiation in the forward direction and focus it into a bolometer. Autocorrelation of the THz waveform is recorded when one arm of the interferometer is scanned. © 2006 Optical Society of America
April 1, 2006 / Vol. 31, No. 7 / OPTICS LETTERS
Fig. 2. Autocorrelation of THz waveforms from a CdTe crystal at different wavelengths. All waveforms are normalized to unity. Phase mismatching and group-velocity mismatching at 880 nm cause broadening of the THz waveform in the time domain compared with those at 710 and 970 nm.
pendence of THz wave detection must be eliminated. Thus a liquid-helium-cooled silicon bolometer (Infrared Labs, LN-6/C) was used to detect the THz radiation. A 0.96 mm thick 具110典-oriented CdTe crystal was used as the THz emitter. Normal optical incidence onto the CdTe surface was applied to eliminate the carrier transport effect when the excitation photon energy is above the energy gap of CdTe.9 The laser wavelength was tuned from 710 to 970 nm (Coherent Chameleon-XR, 90 MHz repetition rate; ⬍140 fs pulse duration). A pellicle film split the laser beam into the two arms of the Michelson interferometer. The recombined beams were focused onto the CdTe crystal. The emitted THz beams interfere with each other when one arm of the interferometer is scanned; thus the autocorrelation of the THz waveform is presented. The bandgap of the CdTe crystal is 1.47 eV at room temperature,10 corresponding to an excitation wavelength of 845 nm. The phase-matching condition changes significantly with the excitation wavelength, and it can be categorized into three regimes. When the excitation wavelength is shorter than 845 nm, the optical rectification is resonantly enhanced and non-phase-match is required due to the essentially short interaction length, which is limited by absorption occurring at the crystal surface. A significant phase mismatch happens when the excitation wavelength is near 845 nm due to the high dispersion. Further lengthening of the excitation wavelength will increase the coherent length. Therefore the efficiency of THz generation will increase. The observed autocorrelation of the THz wave from a CdTe emitter varying with laser wavelength follows the above discussion. Measurements were taken from 710 to 970 nm laser wavelength. Figure 2 shows autocorrelation of normalized THz waveforms at 710, 880, and 970 nm laser wavelengths. The THz waveform at 880 nm is broader in the time domain compared with those at 710 and 970 nm. Further discussion of this phenomenon follows. Figure 3(a) shows the peak intensity of THz radiation from the CdTe crystal as a function of the excitation wavelength; the THz intensity is normalized about the square of the laser power. The peak inten-
979
sity of THz wave at different laser wavelength from a 3 mm thick ZnTe crystal is used for comparison. For illustration purposes, the THz radiation generated from both ZnTe and CdTe crystals is normalized to their own maximum in the measured excitation wavelength range. The measurement shows that, for these two emitters, maximum THz power generated from the ZnTe (at 830 nm) crystal is five times stronger than the CdTe crystal (at 970 nm). However, when working at 970 nm, the CdTe crystal generates five times more THz power than the ZnTe crystal. The solid curve in the figure is calculated by considering the phase-matching condition of Eq. (1). When calculating the phase matching of CdTe, we started from 850 nm, which is away from its bandgap, and the optical index is available.10 It is noticeable that the relationship between THz power from the CdTe crystal and the optical laser wavelength is not monotonic compared with the calculated coherence length. One can see that the measured data have a large deviation near a laser wavelength of 900 nm. As laser photon energy is above the bandgap of CdTe, the signal amplitude increases as the laser wavelength decreases. This phenomenon was also observed when GaAs was used as a THz emitter.11 This can be explained as the dramatic change of nonlinear susceptibility near the bandgap, which is illustrated in Fig. 3(b). Below and near the bandgap of CdTe, the signal amplitude does not increase monotonically as the laser wavelength increases. In Fig. 3(a) one can see a fluctuation around 925 nm. Considering the autocorrelation at 880 nm laser wavelength in Fig. 2, the broadened THz waveform implies two autocorrelated waveforms in the time domain. This is interpreted as the phase mismatching and group-velocity mismatching of laser and THz pulses in the CdTe crystal. First, the coherence length for THz generation is relatively small at this wavelength. Optical rectification happens only at the surfaces of the CdTe crystal. After THz pulses are generated on the incident surface, the rest of the optical pulses keep traveling in the crystal and will generate THz pulses again on the second surface. Due to the different group velocities of THz and optical pulses in the CdTe crystal, the THz pulses generated on the two surfaces
Fig. 3. (a) Peak intensity of THz power normalized by the square of the laser power emitted from the CdTe and ZnTe crystals as a function of laser wavelength. The solid curve in (a) is a calculated fit obtained by considering only the phase-matching condition of the CdTe crystal. These two different materials have maximum coherence lengths at different laser wavelengths. (b) Power transmission of the CdTe crystal shows that the bandgap of CdTe is about 845 nm.
980
OPTICS LETTERS / Vol. 31, No. 7 / April 1, 2006
Fig. 4. Temporal separation of two THz pulses as a function of laser wavelength in the EO sampling process. Separation is defined as the time interval between the first peak and valley of these two THz waveforms, as marked in the inset. A 具100典 p-InAs wafer is used as the THz emitter. The solid curve was calculated from the measured group index of the CdTe crystal in this wavelength range. The inset shows THz waveforms at 870 and 950 nm measured by the EO sampling method.
will not overlap in the time domain. Since these two THz pulses have opposite polarity, interference between them will reduce emitted THz power. This is the reason for the fluctuation in the THz power at 925 nm. When the laser wavelength approaches 1050 nm, the coherence length increases and THz radiation is generated throughout the whole CdTe crystal. To investigate the highly phase-mismatched regime and the double-pulse phenomenon, the same CdTe crystal is used as an EO sampling detection crystal for time-domain THz measurement. A p-type 具100典-oriented InAs wafer with 1016 cm−3 doping was used as the THz emitter. The experimental setup is described in Ref. 12. EO sampling is the inverse process of optical rectification and abides by the same phase-matching conditions. We did the measurement between 860 and 975 nm in which CdTe is transparent and suitable for EO detection. According to our previously observed results, THz radiation emitted from an InAs wafer does not change in this wavelength range. The measured THz waveforms clearly prove that the double-pulse phenomenon happens in a specific wavelength range and this temporal separation changes with laser wavelength. Figure 4 is the measured separation between the two THz waveforms (as marked in the inset) as a function of laser wavelength. Using the measured group index of CdTe, we calculated the separation ⌬t = 共l / c兲兩nTHz − ng兩 with l as the thickness of the CdTe crystal and plotted it by the solid curve. The THz waveforms at 870 and 970 nm are plotted in the inset. When the
probe beam wavelength is 870 nm, there are two THz waveforms separated in the time domain, indicating that the EO process happens only at the surfaces of the CdTe crystal. When the probe beam is tuned to 970 nm, there is just one THz waveform. This has demonstrated the long coherence length for THz generation at long wavelengths. In conclusion, we calculated the coherence length of the CdTe crystal at 1 THz and measured THz radiation from the CdTe crystal at different excitation laser wavelengths. We demonstrated that, in the range of 710–970 nm, phase mismatching and groupvelocity mismatching due to susceptibility dispersion near the bandgap will affect the emitted THz power. As a result, the longer laser wavelength can give more powerful THz radiation in this wavelength range. This has made it possible to use Yb-doped fiber lasers with wavelength of about 1000 nm as a laser source for THz applications. We are grateful to Tao Yuan for useful discussion. This work is supported by the National Science Foundation and the Army Research Office. X.-C. Zhang’s e-mail address is
[email protected]. References 1. B. Ferguson and X.-C. Zhang, Nat. Mater. 1, 26 (2002). 2. J. Ahn, A. V. Efimov, R. D. Averitt, and A. J. Taylor, Opt. Express 11, 2486 (2003). 3. N. C. J. van der Valk, P. C. M. Planken, A. N. Buijserd, and H. J. Bakker, J. Opt. Soc. Am. B 22, 1714 (2005). 4. H. Ohtake, Y. Suzuki, N. Sarukura, S. Ono, T. Tsukamoto, A. Nakanishi, S. Nishizawa, M. L. Stock, M. Yoshida, and H. Endert, Jpn. J. Appl. Phys. Part 2 40, L1223 (2001). 5. H. Takahashi, Y. Suzuki, M. Sakai, S. Ono, N. Sarukura, T. Sugiura, T. Hirosumi, and M. Yoshida, Appl. Phys. Lett. 82, 2005 (2003). 6. M. Nagaia, K. Tanaka, H. Ohtake, T. Bessho, T. Sugiura, T. Hirosumi, and M. Yoshida, Appl. Phys. Lett. 85, 3974 (2004). 7. A. Rice, Y. Jin, X. F. Ma, X.-C. Zhang, D. Bliss, J. Larkin, and M. Alexander, Appl. Phys. Lett. 64, 1324 (1994). 8. M. Schall, M. Walther, and P. Uhd Jepsen, Phys. Rev. B 64, 094301 (2001). 9. M. Schall, H. Helm, and P. Uhd Jepsen, Presented at the International Terahertz Workshop, September 17–19, 2000, Sandbjerg Castle, Denmark. 10. E. D. Palik, ed., Handbook of Optical Constants of Solids (Academic 1985). 11. X.-C. Zhang, Y. Jin, K. Yang, and L. J. Schowalter, Phys. Rev. Lett. 69, 2303 (1992). 12. P. Y. Han, X. G. Huang, and X.-C. Zhang, Appl. Phys. Lett. 77, 2864 (2000).
