Terahertz generation with tandem seeded optical ... - OSA Publishing

10 downloads 383 Views 141KB Size Report
Mar 15, 2005 - Dayton, Ohio 45469-2314 .... This spectrum was made by overlapping 12 smaller ... Projects Agency Small Business Innovation Research.
640

OPTICS LETTERS / Vol. 30, No. 6 / March 15, 2005

Terahertz generation with tandem seeded optical parametric generators P. E. Powers, R. A. Alkuwari, and J. W. Haus Department of Physics and the Electro-Optics Program, University of Dayton, 300 College Park, Dayton, Ohio 45469-2314

K. Suizu and H. Ito Research Institute of Electrical Communication, Tohoku University, 2-1-1 Katahira, Sendai 980-8577, Japan Received August 2, 2004 A simple difference frequency generation (DFG) scheme based on two seeded optical parametric generators is presented as a tunable terahertz (THz) source. Using the nonlinear optical crystal 4-dimethylaminoN-methyl-4-stilbazolium-tosylate (DAST) as the DFG crystal, our system has demonstrated continuous and seamless tunable operation from 1.6 to 4.5 THz. The output bandwidth of the THz source is 2.4 GHz. The utility of the source over this spectral range is demonstrated by measuring a high-resolution transmission spectrum of water vapor in air. © 2005 Optical Society of America OCIS codes: 190.2620, 140.3070.

Recently several schemes for terahertz (THz) generation by means of difference frequency generation (DFG) were successfully demonstrated. The DFG scheme is attractive from the standpoint of simplicity, and, provided that one or both of the sources is tunable, it offers a pathway to the development of tunable narrowband THz frequencies. To achieve efficient THz outputs, high-peak-power pulsed lasers are used in conjunction with a crystal with a high nonlinear coefficient. Generating two wavelengths with a THz frequency separation has been accomplished with a variety of sources. These methods include mixing the output of a laser oscillating at two wavelengths,1 mixing the outputs of a dualwavelength optical parametric oscillator,2,3 and mixing a laser fundamental with the output of an optical parametric oscillator.4 These systems have shown that the DFG scheme can produce useful energies at THz frequencies for a variety of applications. Generating continuously tunable THz frequencies with these sources can be complicated because of difficulties in tuning the laser source. Effects such as mode hopping need to be overcome to obtain seamless tuning. This Letter presents a new approach to the pump sources: we take advantage of the simple tuning properties and high efficiency of pulsed optical parametric generators (OPGs) to provide two independently tunable sources as inputs to a DFG stage that are subsequently mixed to generate THz frequencies. In this Letter we report continuous and seamless tunable operation from 1.6 to 4.5 THz via DFG of OPGs with a narrow 2.4-GHz bandwidth by use of the nonlinear optical crystal 4-dimethylaminoN-methyl-4-stilbazolium-tosylate (DAST). The advantage of using the OPG as a pump source is that the OPG is a single-pass process in which a pump laser is spontaneously frequency converted to a signal and idler pair. With nanosecond pulses and with periodically poled lithium niobate (PPLN), this process typically has a conversion efficiency of the order of 0146-9592/05/060640-3/$15.00

10% to the signal and idler. However, since this is a spontaneous process, the output bandwidth is broad, of the order of 400 GHz. This bandwidth has been reduced to close to the pulse transform limit by introduction of a narrow-bandwidth seed coaligned with the pump and tuned within the bandwidth of the signal.5,6 Since the OPG is a single-pass spontaneous process it does not have resonator mode structure, hence seeding within the OPG bandwidth is simple and easily tuned. The output energy of an OPG based on PPLN is limited by the poling aperture size and by the ability to seed effectively at higher energies.7 The output energy is typically of the order of 150 ␮J in the signal. Compared with other DFG sources that generate energies of the order of several millijoules, the seeded OPG is a relatively low-energy source. However, this relatively low energy can still generate useful THz outputs by use of a DFG crystal with a large nonlinear susceptibility such as DAST. DAST has a nonlinear coefficient in excess of 200 pm/ V that can compensate for the lower input energies of the OPG sources. DAST has other favorable characteristics, such as a low dielectric constant and relatively low loss in the THz region.8 A schematic of the experiment showing the seeded OPGs and the THz generation stage is presented in Fig. 1. The two OPG stages are essentially identical with minor differences in beam focusing. Each OPG stage consists of a 5-cm-long by 1-mm-thick PPLN crystal with a multigrating pattern. For the work presented here, only one of the periods 共29.5 ␮m兲 was used, and tuning was accomplished by changing the crystal temperature. To maintain the crystal temperature, we placed the PPLN crystal in a temperature-controlled oven. A single-frequency Q-switched Nd:YAG laser operating at 30 Hz with 3.5-ns FWHM pulses served as the pump source for both OPGs. A half-wave plate and polarizer combination acted as a beam splitter and 2 mJ of pump energy was sent to both OPG stages. The pump laser was focused to 0.8 mm in one of the crystals and © 2005 Optical Society of America

March 15, 2005 / Vol. 30, No. 6 / OPTICS LETTERS

0.6 mm in the other (1 / e intensity diameters). The difference in spot sizes is due to slightly different focusing conditions for each stage that result from different path lengths from the pump laser to the two OPGs. The performance of the two OPGs was not appreciably different. Two tunable diode lasers with output powers of the order of 2 mW served as the seed sources for the OPG stages. The seeding effectiveness was not sensitive to the seed power, and powers greater than 1 mW were sufficient to seed each stage. The characteristics of the seeded OPG are similar to those reported in Ref. 5. Both diode lasers were continuously tunable over a large wavelength range 共1475– 1575 nm兲. In both OPG stages the output of a tunable diode laser was focused to overlap the pump laser in the PPLN crystal. Once seeded, the output energy of the stage was 150 ␮J in the signal. The bandwidth of the seeded OPG pulses was measured in a previous experiment to be of the order of 1.5 GHz,5 and this is consistent with the THz bandwidth reported below. To generate THz frequencies, the two seeded OPGs are tuned to two different wavelengths with a difference frequency that is the THz frequency of interest. The signal output of each OPG stage was collimated with a 10-cm focal-length lens, and the idler output was dumped by use of dichroic mirrors. All the interacting waves needed to be polarized along the a axis of the DAST crystal for maximum conversion efficiency. The output polarization of the OPG stages was vertical, so we oriented the a axis of the DAST crystal in this vertical direction. Because the polarizations were the same, a 50% beam splitter was used to combine the two OPG outputs. Both beams were then focused with a 50-mm focal-length lens. The focused spot sizes were measured to be 200 ␮m (1 / e intensity) with a knife-edge technique. The DAST crystal was placed at the focus. Several DAST crystals were used; the results presented here were from an as-grown sample with a 1.7-mm thickness. Tuning the OPG sources to the 1.5-␮m region resulted in THz frequencies in the 1 – 4-THz region.3 Tuning away from the phase-matched peak reduced the energy, but significant deviation from this phasematched frequency still yielded appreciable output. A plano–convex lens made from Zeonex with a 5.5-mm radius of curvature was used to collimate the THz output, which placed the lens close to the output face of the DAST crystal. A 40-cm-long beam tube with Teflon windows was placed between this lens

Fig. 1. Schematic of the DFG THz setup. Two PPLN OPGs serve as the DFG inputs to a DAST crystal. The THz output is collimated and transmitted through a beam tube. The beam tube has Teflon windows and can be evacuated to 1 mTorr. SLM, single longitudinal mode; R, reflectivity.

641

Fig. 2. Measured and calculated transmission through a 40-cm path length of air at 175 Torr containing 3-Torr partial pressure of water vapor.

and the entrance to the bolometer housing. This beam tube was connected to a vacuum pump that allowed us to evacuate the cell down to 1 mTorr. The short path lengths between the lens and the beam tube and the region inside the bolometer housing were purged with nitrogen. The bolometer housing consisted of a 13-cm focal-length off-axis parabolic reflector that focused the THz radiation onto the bolometer detector. The bolometer was cooled to 4.2 K with liquid helium. The THz frequencies were measured in two ways. The first was by scanning a low-finesse germanium Fabry–Perot etalon and looking at the transmission as a function of cavity length. The fringe spacing was confirmed to be the expected frequency based on measurements of the two input wavelengths. The second technique employed a transmission measurement of water vapor present in the beam tube. A swept-frequency scan was obtained by tuning one of the OPG’s wavelengths while keeping the other OPG wavelength fixed. The scanning was accomplished by tuning the diode seed laser within the bandwidth of the OPG as discussed above. A typical tuning range of 400 GHz was possible while maintaining a narrow seeded output. Figure 2 shows such a scan. For the measurement in this figure the beam tube was pumped down to 175 Torr with 3-Torr partial pressure of water vapor. The transmission spectrum was obtained by recording the bolometer signal while tuning one of the OPGs and then evacuating the beam tube and taking a background reference scan. Several scans with the cell evacuated were averaged together to give the background scan. Dividing the signal with the cell at 175 Torr by the background scan resulted in the transmission spectrum. By recording the seed laser wavelengths as the scan progressed, we were able to calculate the THz frequency for each transmission measurement. The absorption features present are due to water vapor as shown by plotting both the measured absorption and the predicted absorption of water vapor. The predicted absorption was calculated by use of the high-resolution transmission molecular absorption database (HITRAN). The overlap of the measured and calculated spectra is a clear indication that the signal we are detecting is the THz difference frequency.

642

OPTICS LETTERS / Vol. 30, No. 6 / March 15, 2005

A transmission scan without background correction over a large frequency range is shown in Fig. 3. This spectrum was made by overlapping 12 smaller continuous scans. Each of these scans was performed by temperature tuning one of the OPGs to increase or decrease its center frequency. Once at a new temperature, one of the seeded OPGs was tuned to obtain the swept-frequency output. Superimposed on this data set is the calculated phase-matching efficiency from the DAST crystal, normalized to the peak signal of the transmission scan. The variation in THz energy is largely due to phase mismatch and crystal absorption loss at the THz frequencies. The sharp dips in the spectrum are water-vapor absorptions. From Fig. 3 it is clear that tuning from 1.6 to 4.5 THz is achieved with this setup. Furthermore, it is clear that the THz source has sufficient energy for making spectroscopic measurements. A linewidth measurement of the THz output is possible by use of this same swept-frequency approach. The water-vapor absorption features are a function of pressure, and as the pressure is reduced in the beam tube, the absorption features approach the Dopplerlimited linewidth. The Doppler-limited width is much narrower than an ideally transform-limited THz pulse based on a duration of 3.5 ns. When one is making such a scan, the absorption feature can act as a delta function for measuring the bandwidth. The measured transmission spectrum is essentially a convolution of the THz bandwidth and the actual watervapor spectrum. Figure 4 shows such a measurement across a water-vapor feature at different pressures. Superimposed on this figure are calculated curves based on HITRAN. The HITRAN spectrum was convolved with a 2.4-GHz (FWHM) Gaussian profile that gave the best fit to the measured spectrum. Since the THz bandwidth is most likely not Gaussian, this number serves as an estimate. However, an upper bound on the bandwidth is 3 GHz as given by the FWHM of the measured transmission curve at low pressure. The bolometer is not calibrated for pulsed THz operation, and we estimate the THz energy based on the continuous-wave calibration to be of the order of 10 fJ/ pulse. Other sources are capable of generating more energy per pulse; however, Fig. 3 shows that detailed spectral information is attainable with moder-

Fig. 3. Transmission spectrum without background correction from 1.6 to 4.5 THz, made by overlapping 12 smaller continuous scans. The dips in the spectrum are water-vapor absorption features. Superimposed on the plot is the phase-matching efficiency, including estimated loss.

Fig. 4. Transmission measurements through a 40-cm path length for different pressures. The partial pressure of water vapor was 13 Torr before pumping down.

ate output energies. Moreover, the laser energy used to pump the two OPGs is low enough that scaling up in repetition rate with a different pump laser is possible. The same energy of 4 mJ/ pulse at 30 Hz used in this experiment can be replaced with a commercial system generating the same energy per pulse at kilohertz frequencies. In conclusion, the results of the measurements in this Letter show that tandem seeded OPGs are a feasible pump source for narrowband THz generation. The tunability of the OPGs combined with the high nonlinearity of the DAST crystal makes a THz source with sufficient tunability and energy output to perform spectroscopic measurements, as demonstrated by the high-resolution transmission spectrum shown in Figs. 2 and 3. Because of the overall simplicity of the system, this makes an attractive tunable narrowband THz source for spectroscopy in this THz region. The authors thank Jonathan Goldstein and Kenneth Schepler of the U.S. Air Force Research Laboratory for the loan of the bolometer and the PPLN crystals used in this experiment. This research was partially supported by Defense Advanced Research Projects Agency Small Business Innovation Research grant DAAH-03-C-R295. P. E. Powers’ e-mail address is [email protected] References 1. K. Kawase, M. Mizuno, S. Sohma, H. Takahashi, T. Taniuchi, Y. Urata, S. Wada, H. Tashiro, and H. Ito, Opt. Lett. 24, 1065 (1999). 2. T. Taniuchi, J. Shikata, and H. Ito, Electron. Lett. 36, 1414 (2000). 3. T. Taniuchi, S. Okada, and H. Nakanishi, J. Appl. Phys. 95, 5984 (2004). 4. W. Shi, Y. J. Ding, N. Fernelius, and K. Vodopyanov, Opt. Lett. 27, 1454 (2002). 5. P. E. Powers, K. A. Aniolek, T. J. Kulp, B. A. Richmann, and S. E. Bisson, Opt. Lett. 23, 1886 (1998). 6. M. Rahm, U. Bader, G. Anstett, J.-P. Meyn, R. Wallenstein, and A. Borsutzky, Appl. Phys. B Lasers Opt. 75, 47 (2002). 7. Y. Y. Guan, J. W. Haus, and P. E. Powers, J. Opt. Soc. Am. B 10, 443 (2004). 8. U. Meier, M. Bosch, Ch. Bosshard, F. Pan, and P. Gunter, J. Appl. Phys. 83, 3486 (1998).