1 Introduction

Terahertz (THz) wave has attracted great interests in various applications, especially in biology, chemistry, non-destructive evaluation, communications, and molecular analysis [1,2,3]. From the perspective of practical applications of THz technologies, high-energy output and high spectral resolution as well as wide tunability of THz sources are the particular concerns [4]. There are a number of methods for generating coherent terahertz waves, including electron accelerators [5], ultrashort pulse terahertz-wave generation [6,7,8], and nonlinear frequency conversion such as terahertz-wave parametric oscillation and difference frequency generation. For the narrowband THz source, electron accelerator is an effective method to obtain extremely intense THz output, but the availability and versatility are limited by the large equipment. Terahertz parametric oscillator (TPO) is a more accessible method to generate monochromatic THz radiation, but the tunability is restricted by the properties of inorganic crystals adopted in TPO [9]. Difference frequency generation (DFG), based on dual-wavelength near-infrared (IR) laser incidence into nonlinear optical (NLO) crystals under proper phase-matching conditions, has the inherent characteristics of ultra-wideband THz frequency tunability and narrow linewidth though with a disadvantage of lower conversion efficiency. The generated THz energy of nonlinear DFG scheme can be scalable with high power pump lasers and NLO crystals with sufficient physical dimensions. It is anticipated that DFG-THz sources with broadband tunability and high-energy output would satisfy the demands of high resolution spectroscopy and imaging applications.

Until now, considerable efforts have been made to improve the performance of DFG-THz sources based on dual-wavelength pump sources, nonlinear crystals and phase-matching (PM) configurations. In actual, a dual-wavelength pump source with high-energy output and wideband wavelength tunability is required simultaneously for optimizing THz performances. Two separate laser cavities or gain mediums can easily generate two different wavelengths as the pump sources of CW/monochromatic THz systems. But it has the disadvantage of complexity in mechanical alignment and the spatial overlap of laser transverse modes is poor [10]. A dual-wavelength fiber laser has been proposed for CW THz-wave generation while it suffers from the limited THz frequency tuning range and power enhancement [11]. It is well known that optical parametric oscillator (OPO) is an effective method for arbitrary dual-wavelength generation based on different crystals, such as β-BaB2O4 (BBO), PPLN and KTiOPO4 (KTP). Considering that PPLN crystal is vulnerable with a relatively low damage threshold, it is unsuitable for high power dual-wavelength output [12]. Although widely tunable dual-wavelength from BBO-OPO (800–1800 nm) has been readily achieved, the conversion efficiency is restricted by the pump wavelength of 355 nm and small nonlinear coefficient of BBO crystal [13]. Due to the relatively high damage threshold and large nonlinear coefficient of KTP crystal, KTP-OPO is a favorable approach as the dual-wavelength pump source for high-energy and ultra-wideband THz-wave generation. NLO crystals play the key role in DFG for THz generation. Many investigations have been conducted with some inorganic crystals [14,15,16,17] and organic crystals [18,19,20,21,22,23]. Recently, SiC crystal [24,25,26] has been proven as one promising material for THz generation without the structured spectra. Compared with some inorganic crystals, organic crystals, such as 4′-dimethylamino-N-methyl-4-stilbazolium tosylate(DAST), 4-N, N-dimethylamino-4′-N’-methyl-stilbazolium 2,4,6-trime-thylbenzenesulfonate (DSTMS), N-benzyl-2-methyl-4-nitroaniline (BNA), 2-(3-(4-hydroxystyryl)-5,5-dime-thylcyclohex-2-enylidene) malononitrile (OH1) and their derivatives, exhibit larger nonlinear coefficient and lower dielectric constant in THz region. These make contribution to higher THz output power, higher conversion efficiency, wider tuning range and more favorable type-0 phase-matching configuration. DAST crystal has been developed in the decades and extensively used for ultra-wideband tunable THz generation because of its large second-order nonlinearity (d 11 = 1010 ± 110 pm/V at λ = 1318 nm). In 2004, Adachi et al. and Taniuchi et al., respectively, reported a THz-DFG in DAST crystal under similar dual-wavelength pump intensity (~ 100 MW/cm2), and got the similar energy of THz output of about 100 nJ [27, 28]. In 2007, Suizu et al. utilized a conventional mid-infrared powermeter to detect the THz wave generated from THz-DFG in DAST crystal under the dual-wavelength pump intensity of 35 MW/cm2. However, such power meter had no sensitivity in the lower frequency range (below 20 THz), the THz-wave energy was estimated in reference to the output spectrum obtained using a DTGS detector [29]. Overall, most of the studies focused on the tunability characteristics; nevertheless, the input–output characteristics of DAST–DFG have not been clarified, especially under high dual-wavelength pump intensity.

In this paper, for the purpose of fully exploiting the output characteristics of DAST crystal under intense pump, we presented a high-energy, broadly tunable THz source based on DFG in our home-made DAST crystal. The high-energy dual-wavelength KTP–OPO was developed as pump source, where the effect of blue light was first observed accompanying with dual-wavelength pump light due to different nonlinear processes. The THz frequency tuning range was from 0.3 to 19.6 THz. The maximum THz output energy reached 870.4 nJ at 18.9 THz under the dual-wavelength pump intensity of 247 MW/cm2. The THz energy dips above 3 THz has been analyzed and mainly attributed to the resonance absorption induced by lattice vibration in DAST crystal. The THz output characteristic has been studied and the energy saturation phenomenon can be well explained by three photons absorption (3PA) and free carrier absorption (FCA) in DAST crystal theoretically. In addition, tests of transmission spectroscopy of four typical samples were demonstrated in ultra-wideband THz range.

2 Experimental setup

The schematic diagram of DFG system with DAST crystal is illustrated in Fig. 1. The pump beam from Nd:YAG laser (Spectra Physics, 1064 nm, 9 ns, 10 Hz) was shaped and collimated by a 2:1 telescope lens (TL) to 4 mm in diameter. A half wave plate (HWP) was used to adjust the polarization of pump beam to satisfy phase-matching condition of second harmonic generation (SHG) in KTP (7 × 7 × 10 mm3, θ = 90°, φ = 23.5°) crystal. A double-pass optical parametric oscillator, consisting of cavity mirrors (M2&M3), and KTP1 and KTP2 crystals (7 × 10 × 15 mm3, θ = 65°, φ = 0°) rotated by galvano-optical beam scanners (Cambridge Technology, 6230H), was pumped by the frequency doubled Nd:YAG (532 nm) laser. M1 and dichroic mirror (DM) with high-reflection at 532 nm and antireflection at 1.3–1.6 μm were utilized to reflect and block the 532 nm pump wave, respectively. The DAST crystal in the dimension of 5 × 5 × 0.3 mm3, shown in the inset of Fig. 1, was illuminated with the dual-wavelength pump wave focused by a convex lens (M4, f = 200 mm) to achieve the high intensity. Monochromatic THz wave was generated from the DAST crystal via type-0 phase-matched DFG and collected by two off-axis parabolic mirrors (OAP, f = 101.6 mm) into a helium-cooled Si bolometer (IR Labs Inc.) or Golay cell (TYDEX, Inc.: GC-1P). Black polyethylene sheets with different thicknesses were utilized to filter out the residual near-infrared signal. The THz frequency was calculated from the difference between the two measured frequencies of the dual-wavelength pump light and confirmed by the scanning Fabry–Perot etalon at 4THz.

Fig. 1
figure 1

Schematic diagram of the experimental setup for THz-DFG with DAST crystal

Full size image

3 Results and discussion

3.1 Dual-wavelength pump light from KTP-OPO

The tunability of the dual-wavelength pump light was realized by KTP1 crystal fixed and KTP2 crystal rotated in the OPO cavity, corresponding to idler wavelength λ 1 fixed at 1357.44 nm (the signal wavelength λ s1 = 875 nm) and idler wavelength λ 2 tuned from 1.35 to 1.5 μm (the signal wavelength λ s2 tuning from 878 to 797 nm). The tunability of our experiment could cover the frequency range of 0.3–20 THz according to the Bolometer. To obtain higher dual-wavelength pump energy, the shorter cavity length of 10 cm was first chosen. The maximum energy of dual-wavelength pump light was achieved to be 15.7 mJ/pulse under the 532 nm laser energy of 74.4 mJ with the OPO conversion efficiency of 21.1%. At the same time, a blue light was observed accompanied with the dual-wavelength output throughout the whole tuning process. The wavelengths of the blue light and dual-wavelength pump light from the OPO cavity were measured with two spectrometers (Ocean Optics HR4000 and Yokogawa AQ6375), respectively. Figure 2 shows the relationship between the blue light wavelength and the phase-match angle θ of KTP2, which was calculated based on the recorded λ 2. The intensity was normalized in the figure for simplicity. The focused beam distribution of blue light was depicted in the inset. It is clearly seen that the wavelengths of blue light were tuned simultaneously while rotating the KTP2 crystal to change the idler wavelength λ 2. When λ 2 was tuned to be θ = 63.33°(λ 2 = 1412 nm), θ = 62.75°(λ 2 = 1430 nm), θ = 62.37°(λ 2 = 1442 nm), θ = 61.72°(λ 2 = 1464 nm), θ = 61.30°(λ 2 = 1479 nm), and θ = 60.22°(λ 2 = 1511 nm), the accompanied blue light wavelength was 426.85, 423.5, 421.5, 427.67 and 417.84 nm, 426.32 and 415.42 nm, and 423.62 and 410.56 nm, respectively. Considering that there was tunable signal λ s2 oscillation in the cavity, the blue light of 426.85, 423.5, 421.5, 417.84, 415.42, and 410.56 nm can be deduced as the SHG of λ s2, whereas the other wavelengths of blue light of 427.67, 426.32, and 423.62 nm were generated as the sum-frequency generation (SFG) of λ s1 and λ s2. Actually, when θ is 63.33°, 62.75°, and 62.37°, the wavelengths of blue light of 432.09, 430.43, and 429.35 nm should be observed as the SFG of λ s1 and λ s2. However, the measurement was limited by the detectable wavelength range of the spectrometer (200–428 nm). Moreover, it should be mentioned there should exist the third wavelength of 437.53 nm, which is from the SHG of λ s1. In other words, the SHG and SFG processes of the two signal lights occurred in the cavity at the same time under high-energy intensity.

Fig. 2
figure 2

Wavelength of blue light changes as rotating KTP2 crystal (the inset shows the blue light beam)

Full size image

The appearance of generated blue light is unbeneficial, even possibly harmful for THz generation from the aspects of dual-wavelength pump light and DAST crystal. The effects of multi-nonlinear processes in the cavity and blue light induced infrared absorption (BLIIA) [30] in KTP crystals would attenuate the output energy of the dual-wavelength pump light. On the other hand, in addition, that the single-photon energy of blue light is higher than the bandgap of DAST crystal (2.33 eV), free carrier absorption (FCA) for the generated THz wave would be enhanced by the blue light incidence in DAST crystal [31]. Therefore, the blue light should be suppressed for efficient THz generation. In our experiment, the OPO cavity length was optimized to be 30 cm for reducing the energy intensity in the OPO cavity to restrict the nonlinear conversion processes of SHG and SFG of signal lights.

The output characteristics of KTP-OPO were measured under the cavity length of 30 cm. Figure 3a shows an example of dual-wavelength pump light at 1357.44 and 1369.40 nm with the almost identical energy for each wavelength. The full widths at half maximum (FWHM) for two wavelengths were both as narrow as 0.44 nm (about 70 GHz). The pulse width for dual-wavelength was about 7.7 ns measured by the photodiode (Thorlabs DET08C) as shown in inset. The output energy of dual-wavelength light varied little as the wavelength tuning. Figure 3b shows the output energy and conversion efficiency of KTP-OPO versus the incident 532 nm laser energy. It is indicated that the maximum output energy of dual-wavelength light was amount to 13.86 mJ/pulse under the 532 nm laser energy of 74.4 mJ, and the corresponded OPO conversion efficiency was 18.6%. Then, the output dual-wavelength pump light was focused by a convex lens M4 on the DAST crystal to be 0.75 mm in diameter, which is measured by knife-edge method. Figure 3c shows the relationship of the phase-match angle of KTP and the THz frequency based on the OPO and DFG calculation, which has the almost linear relationship.

Fig. 3
figure 3

Output characteristics of KTP–OPO; a spectrum of the dual-wavelength light of DFG (the inset shows the pulse width of the dual-wavelength); b output energy and conversion efficiency of KTP–OPO versus 532 nm input energy; c relationship of the phase-match angle of KTP and the THz frequency

Full size image

3.2 THz generation based on DFG in DAST crystal

Our home-made DAST crystals were grown by slope nucleation method (SNM) coupled with seed-crystal method (SCM) [32]. The crystal quality was improved by precisely controlling the synthesis and growth technique to avoid the effects of temperature fluctuation and supersaturation, which may lead to the inclusions and growth lines in the crystal.

The square solid line in Fig. 4 (top) shows the THz tuning characteristics generated with DAST crystal under a dual-wavelength pump energy of 8.4 mJ/pulse. A helium-cooled Si bolometer with the calibration of 2.89 × 105 V/W was used as the THz detector. A 0.55 mm-thick black polyethylene sheet was placed before the entrance window to block the residual infrared pump light completely. The tuning range of 0.3–19.6 THz was obtained. Based on the different loss of the 0.55 mm-thick black polyethylene sheet at different THz frequency, the output energy of THz wave can be deduced. The maximum output energy of THz wave was up to 870.4 nJ/pulse at 18.9 THz, corresponding to the energy conversion efficiency of 1.036 × 10− 4. The decay of the THz output and defect on the crystal have not been observed after the laser illumination under such pump intensity. To evaluate the stability of THz source, 1000 pulses at 4 THz were measured, where the direct measurement voltage value from Golay cell was the maximum. The root mean square (RMS) was calculated to be 4.21%. The fluctuation of the THz radiation could be attributed to the fluctuation of dual-wavelength light, whose RMS was tested to be 2.74% for 1000 pulses with powermeter (Newports). Furthermore, the normalized THz intensity can be theoretically calculated using the following DFG equation,

Fig. 4
figure 4

a THz tuning curve under the dual-wavelength energy of 8.4 mJ/pulse (top) and the Raman spectrum of DAST (bottom); b detected THz wavelength using the scanning Fabry–Perot etalon at 4 THz

Full size image
$${I_T}=\frac{{2{\omega _T}^{2}{d_{11}}^{2}{L^2}{I_{{\text{pump}}}}^{2}}}{{{\varepsilon _0}{c^3}{n_1}{n_2}{n_T}}}{T_1}{T_2}{T_T} \times {e^{ - {\alpha _T}L}} \times \frac{{1+{{\text{e}}^{ - \Delta \alpha L}} - 2{{\text{e}}^{ - {{\Delta \alpha L} \mathord{\left/ {\vphantom {{\Delta \alpha L} 2}} \right. \kern-0pt} 2}}}\cos \left( {\Delta kL} \right)}}{{{{\left( {\Delta kL} \right)}^2}+{{\left( {{{\Delta \alpha L} \mathord{\left/ {\vphantom {{\Delta \alpha L} 2}} \right. \kern-0pt} 2}} \right)}^2}}},$$
(1)

where I T is the THz intensity generated by the DFG, L is the interaction length in the crystal, ω T is the THz frequency, d 11 is the second-order nonlinear coefficient along a-axis of DAST crystal, I pump are the optical intensities of the two pump wavelengths with identical energy distributions, n T and n j (j = 1, 2) are the refractive indices of THz wave and two pump wavelengths, respectively, α j are the absorption coefficients, ∆α=|α 1 + α 2-α T |, the wave vector mismatch ∆k = k 1 – k 2 – k T , and k j  = n j ω j /c, T j  = 4n j /(1 + n j )2, and T T are the transmittances at the crystal surface considering the Fresnel reflection on the front and back surfaces of DAST crystal, respectively, which are related to the refractive index at different frequencies. Due to the related parameters of DAST crystals which are just valid in the range of 1–10 THz [33, 34], the theoretical fitting curve was plotted as the red line in Fig. 4 (top). It is clearly seen that calculation was well fitted with the experimental results. The minor deviation can be explained that the plane-wave approximation is taken for DFG in Eq. (1), whereas the particular Gaussian beam suffers from the diffraction effect, especially for the low THz frequency. At a certain THz frequency, the THz output intensity is closely related to the absorption of THz wave and the crystal length. As for the large absorption coefficient of THz wave in the DAST crystal, the optimum crystal length should be chosen to achieve the ultra-wideband tunability and the sufficient interaction of DFG. The crystal length was chosen to be 300 μm to optimize the THz generation in the tuning range of 0.3–20 THz.

Ultra-wideband THz output was not flat, which was mainly attributed to the crystal absorption compared with the phase mismatch in the DFG. The absorption of THz wave below 2 THz (e.g., at 1.1 THz) was attributed to the resonance of the transverse optical phonon in DAST crystal, and the absorption at 3 THz was caused by the non-resonance absorption of the crystal [32]. To figure out the origination of the absorption above 3 THz, natural Raman scattering was measured as a convincing method to study the mechanism. Figure 4a (bottom) shows the Raman spectrum of DAST crystal measured by confocal Raman microscope (Renishaw:inVia). There were some peaks at 5.1, 6.7, 12.35, 15.0, and 17.1 THz in Raman spectrum of DAST crystal. They are coincided with the THz output energy dips at 5.1, 6.7, 12.35, 15.0, and 17.1 THz. Therefore, we can infer that the resonance absorption induced by the lattice vibration of DAST crystal, which is characterized in Raman modes, is one of the main reasons resulting in the THz energy dips at 5.1, 6.7, 12.35, 15.0, and 17.1 THz. The THz output energy dips at 8.55 and 13.9 THz might be caused by other factors which cannot be indicated by the Raman spectrum of DAST crystal.

Furthermore, a scanning Fabry–Perot etalon consisting of two silicon plates was used to verify the correctness of the calculated THz frequency based on the law of energy conservation, as depicted in Fig. 4b. The THz wavelength of 75 μm was measured using scanning F-P with a scanning step of 2.5 μm, which was in good agreement with the calculated results based on the pump wavelength of 1357.44 and 1382.46 nm. Moreover, the spectral bandwidth of the THz wave can be obtained according to the equation. Here, the free spectral range (FSR) \({(\Delta \upsilon )_{{\text{THz}}}}=\frac{c}{{\bar {\lambda } - {{(\Delta \lambda )}_{{\text{S.R}}}}/2}} - \frac{c}{{\bar {\lambda }+{{(\Delta \lambda )}_{{\text{S.R}}}}/2}}\) \({(\Delta \lambda )_{{\text{S.R}}}}={{{{\bar {\lambda }}^2}} \mathord{\left/ {\vphantom {{{{\bar {\lambda }}^2}} {2h}}} \right. \kern-0pt} {2h}}\) can be experimentally acquired by observing no interference phenomenon of Fabry–Perot etalon when the distance h of two Si wafer was longer than1.875 mm at 4THz. Thus, the THz spectral resolution at 4 THz was 80 GHz.

Considering the intense absorption of black polyethylene sheet at 18.9 THz frequency, the THz input–output characteristics at 11.5 THz were measured, as shown as blue squares in Fig. 5. When the pump energy was below 2.75 mJ with the incident intensity of 80.84 MW/cm2, the THz intensity increased with the dual-wavelength pump energy increasing. The data trace can be well fitted by quadratic curve (black curve in Fig. 5) based on Eq. (1), as a typical DFG feature. However, THz output reached the saturation eventually at higher pump intensities. The nonlinearities in the DAST crystal are very complex, and many elements might make contribution to the saturation, such as the heat generated inside the crystal and the restorable structure change under intense pump. Except that, considering that the bandgap of DAST crystal is about 2.33 eV, the free carrier absorption which was enhanced by the three photons absorption of the intense dual-wavelength might also be one of the reasons resulted in the saturation of THz wave, as depicted with the red curve in Fig. 5 [22, 35].

Fig. 5
figure 5

Input–output characteristics of THz wave at 11.5 THz

Full size image

4 Spectroscopic applications of DAST THz-DFG source

Knowledge of the material transmittance in THz range is important not only for fundamental physics, but also for industrial applications. Especially, the characteristics in the ultra-wide frequency are necessary for material evaluation. Based on the ultra-wideband, tunable monochromatic THz–DFG source, shown in Fig. 1 as THz frequency-domain spectroscopy, we conducted transmission spectral measurements for solid and liquid materials. Figure 6 shows the transmittance of 1.2 mm-thick white-polyethylene (PE), 0.4 mm-thick SiC sheets and 0.4 mm-thick high-resistance Si sheet in the range of 1.29–19.6 THz, where the frequency scanning step was 0.1 THz. It was notable that both white PE and SiC sheets had descending trends of transmittance as the THz frequency increasing, and the white PE performed a relatively higher transmittance compared with SiC, though with a larger thickness. The transmittance of Si was observed to be quite stable (~ 45%) in the ultra-wide THz frequency range, illustrating that high-resistance Si could be an ideal material for the applications in the ultra-wideband THz frequency range. Considering the fabrication difficulty and cost, white PE can be employed as an economic substitute for THz applications in ultra-wide range, such as THz lens, window of THz detectors, and other components.

Fig. 6
figure 6

Transmittance of white PE, SiC, and high-resistance Si sheets

Full size image

Besides, Fig. 7 a, b shows the transmittance and absorbance of oleic acid in liquid state with the thickness of 0.8 mm measured with frequency scanning step of 0.1 THz. It is seen that the transmittance of oleic acid decreased quickly with the frequency increasing. Due to the strong absorption at high-frequency range, the measurement was limited up to 14 THz. The absorbance spectrum of oleic acid fitted pretty well with that of 0.5 mm-thick oleic acid measured by FTIR [36]. It should be mentioned that some observed peaks at 103.4, 177.2, and 272.7 cm− 1, could be attributed to the higher spectral resolution in our system, whereas the spectral resolution of FTIR was about 0.5 THz. The mismatch of absorbtance between two curves might be caused by different concentration and thickness of oleic acid.

Fig. 7
figure 7

a Transmittance and b absorbance of oleic acid measured with DAST–DFG and with FTIR [36]

Full size image

5 Conclusion

A high-energy and broadly tunable monochromatic THz-wave generation with home-made organic crystals DAST has been presented in this paper. The high-energy KTP–OPO has been constructed as the dual-wavelength pump source, where the effect of blue light was first observed accompanying with dual-wavelength pump light due to SHG and SFG of signal lights oscillating in the cavity. The THz frequency was tuned randomly in the range of 0.3–19.6 THz. The highest energy of 870 nJ/pulse was obtained at 18.9 THz under the intense pump intensity of 247 MW/cm2. The THz energy dips above 3 THz can be mainly attributed to the resonance absorption induced by lattice vibration in DAST crystal. Meanwhile, the dependence of THz output on the input energy has been studied, and THz output saturation was well explained based on 3 PA and FCA theoretically. Furthermore, tests of transmission spectroscopy of four typical samples were demonstrated with this ultra-wideband THz source. Our studies suggest that more comprehensive spectral information of matters, such as the biomedical tissue, chemicals, and the multiple mixtures, could be obtained with such high-energy, ultra-wideband THz source. Future optimization of the DAST-based THz source is foreseen to further enhance the performance of linewidth and output energy for THz spectroscopy applications.