Ultra-broadband terahertz pulses generated in the ... - OSA Publishing

3404

Letter

Vol. 40, No. 14 / July 15 2015 / Optics Letters

Ultra-broadband terahertz pulses generated in the organic crystal DSTMS CARMINE SOMMA, GIULIA FOLPINI, JYOTSANA GUPTA, KLAUS REIMANN,* MICHAEL WOERNER, AND THOMAS ELSAESSER Max-Born-Institut für Nichtlineare Optik und Kurzzeitspektroskopie, 12489 Berlin, Germany *Corresponding author: reimann@mbi‑berlin.de Received 14 May 2015; revised 19 June 2015; accepted 21 June 2015; posted 24 June 2015 (Doc. ID 240967); published 14 July 2015

Electric-field transients covering the extremely wide frequency range from 0.5 to 26 THz are generated in the organic nonlinear crystal 4-N,N-dimethylamino-4 0 -N 0 -methylstilbazolium 2,4,6-trimethylbenzenesulfonate (DSTMS). Parametric difference frequency mixing within the spectrum of 25-fs amplified pulses centered at 800 nm provides a highly stable broadband output with an electric-field amplitude of up to several hundred kilovolts/cm. The high stability of the terahertz pulse parameters allows for sensitive phase-resolved broadband spectroscopy of optically thick crystalline samples. © 2015 Optical Society of America OCIS codes: (320.7110) Ultrafast nonlinear optics; (190.4710) Optical nonlinearities in organic materials; (260.3090) Infrared, far; (300.6495) Spectroscopy, terahertz. http://dx.doi.org/10.1364/OL.40.003404

Ultrashort optical pulses with spectra spanning more than an octave in frequency have found widespread application in attosecond physics [1], ultrafast terahertz (THz) and mid-infrared (MIR) spectroscopy [2–4], and optical frequency combs [5,6]. In the THz and MIR spectral range, different schemes of broadband pulse generation have been demonstrated, such as accelerator-based sources [7,8], nonlinear frequency conversion in laser-driven gaseous plasmas [9], as well as differencefrequency mixing and optical rectification in crystals displaying a second-order optical nonlinearity [2,10]. Phase-matching of the incoming and generated waves in the nonlinear material has been exploited to generate high electric-field amplitudes and/or pulse energies. In such schemes, the spectral width of the generated pulses is limited by the phase-matching bandwidth, which is related to the group velocity mismatch of the pulses involved. Depending on the phase-matching angle and bandwidth, broadband pulses have mainly been generated either below or above 10 THz [2]. So far, parametric-difference frequency mixing has mainly exploited the second-order nonlinearities of inorganic crystalline materials, e.g., GaSe, GaP, AgGaS2 , or AgGaSe2. 0146-9592/15/143404-04$15/0$15.00 © 2015 Optical Society of America

Recently, an interesting class of organic nonlinear crystals has been introduced for nonlinear frequency conversion because of their very high second-order optical nonlinearities. A prototype material is the stilbazolium derivative 4-N,N-dimethylamino4 0 -N 0 -methylstilbazolium 2,4,6-trimethylbenzenesulfonate (DSTMS) [11–22] consisting of positively charged stilbazolium and negatively charged sulfonate ions. Short pump pulses with center wavelengths of 1.3, 1.5, or, recently [17], 0.8 μm have been used for generating THz and mid-infrared frequencies in DSTMS by difference-frequency mixing of spectral components within the broad pump spectrum. Up to now, the highest frequency obtained from DSTMS in this way was 12 THz [14], while difference-frequency mixing of narrowband nanosecond pulses at different wavelengths produced frequencies up to 30 THz [22]. In this Letter, we report the generation of ultrashort electricfield transients in DSTMS covering the ultra-broad spectral range from 0.5 to 26 THz continuously. We apply difference-frequency mixing within the spectrum of amplified 25-fs pulses centered at 800 nm and characterize the generated transients in amplitude and phase by electro-optic sampling. Electric-field amplitudes of up to several hundred kilovolts/cm are demonstrated. As a first spectroscopic application of the highly stable broadband pulses, we report linear transmission spectra of strongly absorbing crystalline materials. The spectral width of a broadband femtosecond pulse sets a principal limit for the maximum THz/MIR frequency generated by difference-frequency mixing within the pulse spectrum. If absorption in the nonlinear material and the frequency dependence of the nonlinear coefficient can be neglected, the THz/MIR spectrum is determined by the input spectrum and by the phase-matching bandwidth (Eq. (1) of [23]). As shown in the following, phase-matching plays a minor role in DSTMS, whereas absorption and the frequency dependence of the second-order nonlinearity are highly relevant. In our experiment, pump pulses with a center wavelength of 800 nm, a repetition rate of 1 kHz, and pulse energies of up to 1 mJ are generated by a Ti:sapphire oscillator–amplifier system. The spectral widths of these pulses are 22 THz (FWHM) and 40 THz (full width at 1/10 of the maximum). A built-in pulse

Letter shaper is used to adjust the spectral phases to obtain the highest electric-field amplitudes of the generated THz radiation. A parallel pump beam was sent onto a DSTMS crystal to generate the difference frequency of spectral components within the spectrum of a single 800-nm pulse. For the measurements on DSTMS, we restricted the pulse energy to 200 μJ, corresponding to a peak intensity of 100 GW∕cm2 , to prevent damage [17]. The DSTMS crystal with a thickness of 0.38 mm was purchased from Rainbow Photonics. The a-b crystal surface had a diameter of 3 mm, and the polarizations of the pump and the emitted THz pulses were parallel to the a axis, making d 111 the relevant nonlinear coefficient. The generated THz pulses are detected by free-space electrooptic sampling [24–27]. There are several factors limiting the detection bandwidth in this method. First, the frequency at which the sensitivity decreases by a factor of two is given by νmax 0.44∕T , where T is the duration of the sampling pulse [28]. Second, the properties of the electro-optic crystal also restrict the frequency range of detection. Of particular importance is the mismatch between the group velocity v g of the sampling pulse and the phase velocity v p of the THz transient to be characterized (wavelength λ). This mismatch restricts the useful thickness of the electro-optic crystal to the coherence length cλ∕2jv g − v p j [25]. In our setup, a 10-μm-thick ZnTe crystal and 12-fs pulses from the oscillator of the Ti:sapphire laser system as sampling pulses are used, resulting in a detection bandwidth of approximately 30 THz [29]. The (phonon) reststrahlen band of ZnTe is located between 4 and 6 THz, where the strong optical reflection prevents electro-optic sampling of the THz field. Moreover, the electro-optic coefficient can have contributions from optical phonons, leading to different values of the electrooptic coefficient for frequencies above and below the optical phonon frequencies. ZnTe has the advantage of having a very small phonon contribution, so that the electro-optic coefficient is essentially constant over a very wide frequency range [27]. The electric field of the generated pulse as a function of time is shown in Figs. 1(a)–1(c). Apart from the short initial spike with a maximum amplitude of 800 kV/cm [for an enlargement, see Fig. 1(c)], there is a component with an initial amplitude of 100–200 kV/cm that is damped on a time scale of approximately 4 ps. At late times, this transient shows a pronounced beating with a period of 0.25 ps [Fig. 1(b)]. Spectra derived by a Fourier transform of the electric-field transients are shown in Figs. 2(a) and 2(c) and cover an extremely broad range from 0.5 to 26 THz. The broad THz spectrum is superimposed by narrow lines at 16.7 and 20.7 THz, both of a width of 0.27 THz. Since the measurements with ZnTe cannot cover the range of the reststrahlen band between 4 and 6 THz, we performed additional measurements with a 100-μm-thick GaP crystal (reststrahlen band 7–9 THz) for electro-optic sampling. Such measurements show a smooth THz spectrum in the 4–6 THz range with spectral power comparable to the value at 3 THz [cf. Fig. 2(a)]. To assess the relevance of phase-matching in the broadband generation process, we first compare the spectra generated with the DSTMS crystal to data taken with a 100-μm-thick GaSe crystal under otherwise identical experimental conditions. In


3405

Fig. 1. (a)–(c) Electric-field transients emitted by the DSTMS crystal and measured by electro-optic sampling. (b) and (c) are enlargements of selected time ranges. (d) Transient after transmission through a 0.445-mm-thick DAST crystal with the electric field parallel to the a axis.

Fig. 2(c), we present spectral profiles generated in GaSe for two different angular orientations. The spectra shift with angle, a clear indication of phase-matched difference-frequency mixing. Moreover, the generated spectral width is in agreement with previous studies and consistent with the calculated phasematching bandwidth [23,27,31]. The profiles generated in GaSe are substantially narrower than the overall spectral width generated in DSTMS [Fig. 2(a)], the latter only weakly depending on the angular orientation. Such behavior points to a nonphase-matched generation process in DSTMS, which, thus, allows for the very large bandwidth generated. For an analysis, we calculated the parametric coherence length l eff π∕Δk, which is determined by the phase mismatch Δk k3 − k 2 − kTHz between the electric fields interacting in a collinear geometry. Here, ki ni ωi ∕c is the (collinear) wavevector of the field i with the frequency ωi and the index of refraction ni . The index of refraction in the near infrared is taken from [12], while the refractive index in the THz range is calculated from the Sellmeier equations in [13]. The latter predict an essentially constant value nTHz ≈ 2.22 for frequencies higher than 8 THz and neglect the index modulations Δn in the range of vibrational resonances. The analysis in [14] has shown that Δn ≤ 0.1, thus leading to minor modulations of the parametric coherence length. The calculated l eff , which is a measure for the interaction length in the DSTMS crystal

3406


Fig. 2. (a) Solid line, spectral power density of the pulse emitted from DSTMS for normal incidence, measured with a spectral resolution of 0.05 THz. In the frequency range between 4 and 6 THz, the data obtained with ZnTe for electro-optic sampling are not reliable because of the reststrahlen band (marked) of ZnTe. Dashed line, coherence length for difference-frequency generation calculated with the refractive index data of [12,13]. (b) Optical density of a 0.445-mmthick DAST crystal. Noise limits the measurable optical density to around five [30]. The observed absorption peaks are marked by vertical lines; see Table 1. (c) Spectral power density of the pulse emitted from DSTMS for normal (0°) incidence, for incidence under 45°, and for pulses from a 100-μm-thick GaSe crystal under 40° and 48°. The spectral resolution for these measurements is 0.5 THz. Note the different abscissa scales in (a)–(c).

over which THz and mid-infrared fields are generated, is plotted in Fig. 2(a) and has a value below the crystal thickness of 380 μm for frequencies larger than 0.7 THz. Above 2 THz, the parametric coherence length in DSTMS is substantially shorter than the phase-matched interaction length of 100 μm in the GaSe crystal. However, the shorter interaction length in DSTMS is compensated by the larger second-order nonlinearity in DSTMS (d 111 21420pm∕V [12,21]) compared to GaSe (86 17 pm∕V [32], see also

Letter [33,34]). As a result, the maximum electric-field amplitudes generated in the two materials are of the same order of magnitude. DMSTS and the closely related material 4-N,N-dimethylamino-4 0 -N 0 -methylstilbazolium tosylate (DAST) display a multitude of vibrational absorption bands in the spectral range covered by the generated broadband pulses. Such bands lead to reabsorption of the generated light, an effect that is particularly pronounced for the comparably long l eff in the range below 3 THz. Using the infrared absorption data from [13,14], one estimates reabsorption losses between more than 99% at 1 THz and 40% at 3 THz. For the shorter coherence lengths between 6 and 16 THz, reabsorption losses are on the order of 50% at the maxima of vibrational bands and negligible in between. As a result, one observes a rise in spectral power [Fig. 2(a)] with increasing frequency and a relatively smooth spectrum for frequencies above the ZnTe reststrahlen band. Strong modulations of spectral power occur in the range of the absorption bands at 16.7 and 20.7 THz, which originate from molecular vibrations of the system [11]. The two strong emission components give rise to the beating, which persists on a picosecond time scale [Fig. 1(b)] and points to a slow dephasing of the underlying polarizations. In principle, such polarizations could be caused by resonances in the second-order nonlinearity of DSTMS or represent coherent infrared emission on the v 0 → 1 transition of the respective vibration. Most vibrations are both Raman and infrared active because of the low symmetry of the DSTMS crystal structure. As a result, a Raman process within the broad spectrum of the 800-nm driving pulses could generate a v 1 population that gives rise to infrared emission via the vibrational transition dipole [35]. Further detailed studies are required to clarify this issue. It should be noted that the modulations of the refractive index in the range of vibrational resonances are limited, as shown in [14]. To demonstrate a first application of the novel broadband source, we performed transmission measurements on a 0.445-mm-thick crystal of DAST [Figs. 1(d) and 2(b)]. Although this thickness is too large to determine linewidths and oscillator strengths with high precision, we are able to determine the vibration frequencies (see Table 1) in the whole range from 1 to 25 THz from a single measurement. To conclude, the nonlinear organic crystal DSTMS allows the generation of ultra-broadband radiation in the THz and Table 1. Frequencies of Absorption Lines in DAST (in Terahertz) in the Frequency Range of 0–25 THz from This Work (Experimental Uncertainty 0.1 THz as Determined by the Vibrational Linewidths), Compared with Absorption Data from [14,36] This Work

[14]

This Work

[36]

This Work

[36]

1.1 3.1 7.1 8.5 11.0 12.2 12.8

1.1 3.1 7.0 8.4 11.0 12.3

14.0 14.7 15.1 16.0 16.4 16.9 17.7 18.9

14.1 14.5 15.0 16.1 16.5 17.0

20.7 21.6 23.5 23.9 24.5 24.8 25.2

20.5 21.7

18.7

24.5 24.8


Letter mid-infrared frequency range by difference-frequency generation within the spectrum of ultrashort Ti:sapphire laser pulses. The parameters of the generated pulses are highly stable and allow for precise transmission measurements with optically thick samples. We envisage a broad applicability of the broadband source in time-resolved nonlinear spectroscopy of molecular systems where a large variety of intra- and intermolecular vibrational modes are located in this frequency range. Funding. Deutsche (RE806/9-1).

Forschungsgemeinschaft

(DFG)

Acknowledgment. We acknowledge valuable discussions with Prof. Christos Flytzanis. REFERENCES AND NOTES 1. F. Krausz and M. Ivanov, Rev. Mod. Phys. 81, 163 (2009). 2. R. Ulbricht, E. Hendry, J. Shan, T. F. Heinz, and M. Bonn, Rev. Mod. Phys. 83, 543 (2011). 3. T. Elsaesser, K. Reimann, and M. Woerner, J. Chem. Phys. 142, 212301 (2015). 4. C. Calabrese, A. M. Stingel, L. Shen, and P. B. Petersen, Opt. Lett. 37, 2265 (2012). 5. S. A. Diddams, D. J. Jones, J. Ye, S. T. Cundiff, J. L. Hall, J. K. Ranka, R. S. Windeler, R. Holzwarth, T. Udem, and T. W. Hänsch, Phys. Rev. Lett. 84, 5102 (2000). 6. S. Koke, C. Grebing, H. Frei, A. Anderson, A. Assion, and G. Steinmeyer, Nat. Photonics 4, 462 (2010). 7. T. Nakazato, M. Oyamada, N. Niimura, S. Urasawa, O. Konno, A. Kagaya, R. Kato, T. Kamiyama, Y. Torizuka, T. Nanba, Y. Kondo, Y. Shibata, K. Ishi, T. Ohsaka, and M. Ikezawa, Phys. Rev. Lett. 63, 1245 (1989). 8. T. Nakazato, M. Oyamada, N. Niimura, S. Urasawa, O. Konno, A. Kagaya, R. Kato, T. Kamiyama, Y. Torizuka, T. Nanba, Y. Kondo, Y. Shibata, K. Ishi, T. Ohsaka, and M. Ikezawa, Phys. Rev. Lett. 63, 2433 (1989). 9. K. Y. Kim, A. J. Taylor, J. H. Glownia, and G. Rodriguez, Nat. Photonics 2, 605 (2008). 10. M. C. Hoffmann and J. A. Fülöp, J. Phys. D 44, 083001 (2011). 11. Z. Yang, L. Mutter, M. Stillhart, B. Ruiz, S. Aravazhi, M. Jazbinsek, A. Schneider, V. Gramlich, and P. Günter, Adv. Funct. Mater. 17, 2018 (2007). 12. L. Mutter, F. D. J. Brunner, Z. Yang, M. Jazbinšek, and P. Günter, J. Opt. Soc. Am. B 24, 2556 (2007).

3407

13. M. Stillhart, A. Schneider, and P. Günter, J. Opt. Soc. Am. B 25, 1914 (2008). 14. P. D. Cunningham and L. M. Hayden, Opt. Express 18, 23620 (2010). 15. P.-J. Kim, M. Jazbinsek, and O.-P. Kwon, Crystal Growth Design 11, 3060 (2011). 16. Y. Li, J. Zhang, G. Zhang, L. Wu, P. Fu, and Y. Wu, J. Cryst. Growth 327, 127 (2011). 17. B. Monoszlai, C. Vicario, M. Jazbinsek, and C. P. Hauri, Opt. Lett. 38, 5106 (2013). 18. C. Ruchert, C. Vicario, and C. P. Hauri, Phys. Rev. Lett. 110, 123902 (2013). 19. C. Vicario, B. Monoszlai, and C. P. Hauri, Phys. Rev. Lett. 112, 213901 (2014). 20. C. Vicario, A. V. Ovchinnikov, S. I. Ashitkov, M. B. Agranat, V. E. Fortov, and C. P. Hauri, Opt. Lett. 39, 6632 (2014). 21. T. Seidler, K. Stadnicka, and B. Champagne, J. Chem. Phys. 141, 104109 (2014). 22. P. Liu, D. Xu, Y. Li, X. Zhang, Y. Wang, J. Yao, and Y. Wu, Europhys. Lett. 106, 60001 (2014). 23. R. A. Kaindl, F. Eickemeyer, M. Woerner, and T. Elsaesser, Appl. Phys. Lett. 75, 1060 (1999). 24. Q. Wu and X.-C. Zhang, Appl. Phys. Lett. 67, 3523 (1995). 25. A. Nahata, A. S. Weling, and T. F. Heinz, Appl. Phys. Lett. 69, 2321 (1996). 26. Q. Wu and X.-C. Zhang, Appl. Phys. Lett. 71, 1285 (1997). 27. A. Leitenstorfer, S. Hunsche, J. Shah, M. C. Nuss, and W. H. Knox, Appl. Phys. Lett. 74, 1516 (1999). 28. K. Reimann, Rep. Prog. Phys. 70, 1597 (2007). 29. At this frequency, the finite pulse length of the sampling pulse leads to an amplitude of the electro-optic signal of 0.63, compared to an amplitude of 1 for a δ pulse. The coherence length for electro-optic sampling is 9 μm, resulting in an amplitude of 0.8 compared to the case of infinitely long coherence length. Both effects together reduce the amplitude to 0.5. 30. P. U. Jepsen and B. M. Fischer, Opt. Lett. 30, 29 (2005). 31. K. Reimann, R. P. Smith, A. M. Weiner, T. Elsaesser, and M. Woerner, Opt. Lett. 28, 471 (2003). 32. K. R. Allakhverdiev, M. Ö. Yetis, S. Özbek, T. K. Baykara, and E. Y. Salaev, Laser Phys. 19, 1092 (2009). 33. T. Dahinten, U. Plödereder, A. Seilmeier, K. L. Vodopyanov, K. R. Allakhverdiev, and Z. A. Ibragimov, IEEE J. Quantum Electron. 29, 2245 (1993). 34. N. C. Fernelius, Prog. Cryst. Growth Charact. Mater. 28, 275 (1994). 35. J. Savolainen, S. Ahmed, and P. Hamm, Proc. Natl. Acad. Sci. USA 110, 20402 (2013). 36. T. Vijayakumar, I. H. Joe, C. P. R. Nair, M. Jazbinsek, and V. S. Jayakumar, J. Raman Spectrosc. 40, 52 (2009).