INTRODUCTION

Currently, the generation of high-power terahertz (0.1–10 THz) radiation is actively studied in view of the development of nonlinear terahertz optics, where impact of terahertz pulses on a material is considered [1], and various applications, in which terahertz radiation is used as a channel of noninvasive diagnostics. For example, many molecular complexes have resonance absorption lines in the terahertz frequency range, which makes it possible to use terahertz radiation sources in time-resolved spectroscopy with multispectral visualization [2]. In contrast to optical and ultraviolet pulses, terahertz photons have a fairly low energy (0.4–124 meV), which allows for noninvasive diagnostics of biological tissues [3] and investigation of low-energy excitations of rotational and vibrational molecular levels, lattice vibrations, and excitations of bound electron–hole pairs [4]. The use of strong terahertz pulses (with a field strength of about a few megavolts per centimeter) provides new possibilities for studying interaction between terahertz radiation and a material, including harmonic generation, bleaching, and impact ionization [5, 6].

Various sources of high-power terahertz radiation and methods of its generation were proposed [7, 8] in view of the great variety of its applications, including optical rectification in nonlinear organic [9] and inorganic crystals [6] and filamentation in gases [11].

In the case of generation of terahertz pulses based on two-color filamentation in gases, the optical-to-terahertz conversion efficiency is quite low (about 0.01%) [12–14]. However, the energy of generated terahertz radiation can be increased by increasing the pump energy (specifically, using high-power laser systems [15, 16]), because gas media cannot be damaged, thus allowing one to use high-intensity laser radiation without limitations on the pump energy. However, a significant drawback of this method is a strong angular divergence and a conical structure of terahertz radiation [17]. A higher terahertz generation efficiency is observed for generation based on optical rectification in nonlinear inorganic crystals. The optical-to-terahertz conversion efficiency is about 0.1% in this case [18].

The efficient generation of few-cycle terahertz pulses with a field strength up to several tens of megavolts per centimeter and the optical-to-terahertz conversion efficiency up to 3% [19–21] have become available only recently, with the discovery of organic crystals such as DAST, DSTMS, OH1, and BNA [9]. These crystals provide high nonlinearity and are transparent to both optical pumping and terahertz fields, which makes the process of optical rectification the most efficient; however, the dispersion properties of these materials require the use of a near-infrared (1.2–1.5 µm) pump source [22]. Note that the conversion efficiency in organic crystals is limited by multiphoton absorption and, as a consequence, by the breakdown threshold of these crystals.

A BNA crystal [23] is a promising material for terahertz radiation generation because it has fairly large values of the breakdown threshold (~1 TW/cm2) [24] and the electro-optic coefficient r [25], which are comparable to the characteristics of a more popular DAST organic crystal. Currently, BNA crystals are widely applied in schematics with Ti:sapphire laser pumping at a wavelength of 0.8 µm [23], for which the group velocities of interacting terahertz and optical pump pulses are matched in the frequency range up to 15 THz [26]. However, the dispersion properties of these crystals allow one to use longer-wavelength pump radiation [27]. In this case, more intense pumping can be used owing to a decrease in the influence of two-photon absorption.

To date, the generation of terahertz pulses in BNA crystals with long-wavelength pumping has been performed using radiation of a parametric amplifier [27], where the spectral and temporal parameters of the generating radiation (which significantly affect the shape and width of the spectrum of generated terahertz pulses) cannot be controlled [28]. In this context, a solid-state Cr:forsterite laser system, which generates femtosecond radiation at a wavelength of 1.24 µm, allows chirping and shortening of generated pulses, and has a potential to increase the peak and average radiation power [15], is a promising source for BNA crystal pumping. Reduction of the pulse duration makes it possible to extend the terahertz spectrum to higher frequencies. Thus, the generation of terahertz radiation in a BNA crystal pumped by radiation from the Cr:forsterite laser system can be considered as an alternative to standard schemes based on combination of a BNA crystal and a Ti:sapphire laser system.

NUMERICAL MODEL

To describe the experimental data obtained on the generation of terahertz radiation in an organic BNA crystal using ultrashort Cr:forsterite laser pulses, we performed numerical simulation by solving Maxwell’s equations including the nonlinear response of the medium, in particular, three-wave interaction. The application of this approach is necessary because the short pulse duration (35 fs), which comprises about nine field cycles, hinders the application of the slowly varying amplitude approximation used in [28].

Processes of the generation of the sum and difference frequencies were considered within the expansion of nonlinear polarization of matter in the weak-field approximation up to third-order terms [29] and in the plane-wave approximation taking into account for material dispersion and absorption in an organic BNA crystal in the spectral range up to 10 THz. The model used consider second- and third-order nonlinear processes responsible for the generation of the difference frequency (optical rectification), including the cascade effect, and for the generation of the third optical harmonic falling in the crystal absorption band. The model also includes two- and one-photon absorption. Nonlinear self-action effects on third- and higher-order nonlinearities for the pump radiation were disregarded in view of the infignificance of their contribution to the interaction between the radiation and matter. Thus, the ratio of the crystal length to the nonlinear length was L/Lnl = 0.8 × 10–3/40 = 2 × 10–5 and L/Lnl = 0.8 × 10–3/10 = 8 × 10–5 for the pump pulse durations of 100 and 35 fs, respectively (assuming that n2 = 6.76 × 10–18 m2/W [30]). In addition, the model includes the mismatch and dispersion of the group velocities of interacting pulses and the linear absorption of optical and terahertz radiation in the crystal.

In the above approximations, the generation of terahertz radiation in the organic BNA crystal based on the effect of optical rectification is described by Maxwell’s equations:

$$\left\{ \begin{gathered} \frac{{\partial {\mathbf{D}}}}{{\partial t}} = \boldsymbol{\nabla} \times {\mathbf{H}} \hfill \\ \frac{{\partial {\mathbf{H}}}}{{\partial t}} = - \frac{1}{{{{\mu }_{0}}}}\boldsymbol{\nabla} \times {\mathbf{E}}{\kern 1pt} {\kern 1pt} . \hfill \\ \end{gathered} \right.$$
(1)

Here, \({\mathbf{D}}(\omega ) = {{\varepsilon }_{0}}\varepsilon _{r}^{*}(E,\omega ){\mathbf{E}}(\omega )\) is the electric displacement vector, where ε0 is the permittivity of free space, \(\varepsilon _{r}^{*}(E,\omega ) = {{\varepsilon }_{\infty }} + {{\chi }^{{(1)}}}(\omega ) + {{\chi }^{{(2)}}}(\omega )\)E(ω) + χ(3)(ω) × \(E(\omega )E{\text{*}}(\omega )\) is the relative permittivity of the medium, and \({{\chi }^{{(n)}}}\) is the nth order dielectric susceptibility of the medium; t is the time; \(\boldsymbol{\nabla} \) is the Hamilton operator; H is the magnetic field vector; E is the electric field vector; ω is the angular frequency; and μ0 is the permeability of free space.

To describe the linear response of the medium, we use the Lorentz oscillator model:

$${{\chi }^{{(1)}}}(\omega ) = \sum\limits_{l = 1}^L \frac{{\Delta {{\varepsilon }_{l}}\omega _{{0,l}}^{2}}}{{\omega _{{0,l}}^{2} + 2i\omega {{\nu }_{l}} - {{\omega }^{2}}}},$$
(2)

where L is the number of oscillators, i is the imaginary unit, and \(\Delta {{\varepsilon }_{l}}\), \({{\omega }_{{0,l}}}\), and \({{\nu }_{l}}\) are the force, eigenfrequency, and damping coefficient of the lth oscillator, respectively.

Using the formalism of the finite-element method with spacings Δx and Δt [31], the Newmark method [32], and the substitutions

$$\left\{ \begin{gathered} {\mathbf{\tilde {E}}} = \sqrt {\frac{{{{\varepsilon }_{0}}}}{{{{\mu }_{0}}}}} {\mathbf{E}} \hfill \\ {\mathbf{\tilde {D}}} = \sqrt {\frac{1}{{{{\mu }_{0}}{{\varepsilon }_{0}}}}} {\mathbf{D}} \hfill \\ {\mathbf{\tilde {P}}} = \sqrt {\frac{1}{{{{\chi }^{{(1)}}}}}} {\mathbf{E}} \hfill \\ \Delta t = \frac{{\Delta x}}{{2c}}, \hfill \\ \end{gathered} \right.$$
(3)

where \({\mathbf{\tilde {P}}}\) is the polarization vector of the medium, Δt is the temporal grid spacing, Δx is the spatial grid spacing, and c is the speed of light in vacuum, we obtained the following system of equations for the spatiotemporal difference grid:

$$\left\{ \begin{gathered} \widetilde D_{x}^{{n + \frac{1}{2}}}(k) = \widetilde D_{x}^{{n - \frac{1}{2}}}(k) + \frac{{H_{y}^{n}\left( {k + \frac{1}{2}} \right) - H_{y}^{n}\left( {k - \frac{1}{2}} \right)}}{2} \hfill \\ {{\varepsilon }_{\infty }}\tilde {E}_{x}^{{n + \frac{1}{2}}}(k) = \widetilde D_{x}^{{n + \frac{1}{2}}}(k) - \sum\limits_{l = 1}^L \tilde {P}_{{x,l}}^{{n - \frac{1}{2}}}(k) - {{\chi }^{{(2)}}}{{\left[ {\tilde {E}_{x}^{{n - \frac{1}{2}}}(k)} \right]}^{2}} - {{\chi }^{{(3)}}}\left\{ {3{{{\left[ {\tilde {E}_{x}^{{n - \frac{1}{2}}}(k)} \right]}}^{2}}\tilde {E}_{x}^{{n + \frac{1}{2}}}(k) - 2{{{\left[ {\tilde {E}_{x}^{{n - \frac{1}{2}}}(k)} \right]}}^{3}}} \right\} \hfill \\ \tilde {P}_{{x,l}}^{{n + \frac{1}{2}}}(k) = {{\omega }_{{1,l}}}\tilde {P}_{{x,l}}^{{n - \frac{1}{2}}}(k) + {{\omega }_{{2,l}}}\tilde {P}_{{x,l}}^{{n - \frac{3}{2}}}(k) + {{u}_{{0,l}}}\tilde {E}_{x}^{{n + \frac{1}{2}}}(k) + {{u}_{{1,l}}}\tilde {E}_{x}^{{n - \frac{1}{2}}}(k) + {{u}_{{2,l}}}\tilde {E}_{x}^{{n - \frac{3}{2}}}(k) \hfill \\ H_{y}^{{n + 1}}(k + 1{\text{/}}2) = H_{y}^{n}(k + 1{\text{/}}2) - \frac{{E_{x}^{{n + 1/2}}(k + 1) - E_{x}^{{n + 1/2}}(k)}}{2}. \hfill \\ \end{gathered} \right.$$
(4)

Here, coefficients \({{\omega }_{{n,l}}}\) and \({{u}_{{n,l}}}\) are calculated according to [32] for each Lorentz oscillator. Under the assumption that a Gaussian pulse is applied at the crystal input, the pump field can be set in the form

$$\begin{array}{*{20}{c}} {{{{\tilde {E}}}_{{{\text{op}}}}}(t) = \left[ {\frac{1}{2}{{E}_{{{\text{0op}}}}}{{e}^{{ - 2\ln 2{{{\left( {\frac{t}{{\Delta {{t}_{{{\text{op}}}}}}}} \right)}}^{2}}}}}{{e}^{{i\omega t}}} + {\text{c}}{\text{.c}}{\text{.}}} \right]\sqrt {\frac{{{{\varepsilon }_{0}}}}{{{{\mu }_{0}}}}} ,} \\ {\tilde {E}_{x}^{{n + \frac{1}{2}}}\left( {{{k}_{{{\text{source}}}}}} \right) = {{{\tilde {E}}}_{{{\text{op}}}}}\left( {\Delta t\left( {n + \frac{1}{2}} \right) - {{t}_{{{\text{shift}}}}}} \right).} \end{array}$$
(5)

The following initial and boundary conditions are also satisfied:

$$\begin{array}{*{20}{c}} {\tilde {E}_{x}^{{\frac{1}{2}}} = \widetilde D_{x}^{{\frac{1}{2}}} = \tilde {P}_{x}^{{\frac{1}{2}}} = H_{y}^{1} = 0,} \\ {\tilde {E}_{x}^{{n + 1/2}}(1) = \tilde {E}_{x}^{{n - 2 + 1/2}}(2),} \\ {\tilde {E}_{x}^{{n + 1/2}}(N) = \tilde {E}_{x}^{{n - 2 + 1/2}}(N - 1),} \end{array}$$
(6)

where Eop is the optical pump wave field and Δtop is the pulse duration. Numerical solution of the equations was performed in Python [33] using the fast-calculation Numexpr library.

The main characteristics of the BNA crystal for the simulation were calculated as follows. The permittivity in the entire frequency range was calculated according to the Lorentz oscillator model, where approximation in the optical range was performed using the Sellmeier formula [34] and the parameters in the terahertz range were chosen using the crystal absorption spectrum [26]. In addition, multiple absorption lines in the far-infrared region were replaced by one total line with absorption at a wavelength of 7 µm beyond the wavelength range under study. This replacement was made for two reasons. First, there are no data in the literature on these lines with a sampling rate sufficient for the calculations. Second, the addition of this group of lines increases the calculation time because calculations of the oscillator responses is an essential part of the calculations. This replacement is physically correct because the radiation under consideration is absent in this spectral range. It also yields smooth continuous functions of the refractive index and the absorption coefficient. Actually, this replacement is similar to the approach described in [35].

The effective nonlinear coefficient deff used in the simulation can be expressed in terms of the refractive index n and electro-optic coefficient reff as deff = 0.25n4reff and, in the case of 1.24-µm pump radiation from the Cr:forsterite laser, equals 116 pm/V. Values of the electro-optic coefficient responsible for optical rectification were obtained from [36]. It was taken into account that the crystal is oriented for the generation maximum; therefore, the electro-optic coefficient was assumed to be reff = r111. The permittivity of the material beyond the crystal was assumed to be equal to the permittivity of free space.

EXPERIMENTAL SETUP

A sketch of the experimental setup is presented in Fig. 1. To pump the organic BNA crystal, we used the Cr:forsterite laser system (wavelength is 1.24 µm, FWHM pulse duration is 100 fs, pulse energy up to 16 mJ (the experiment was carried out using 3.5 mJ), pulse repetition frequency is 10 Hz) [15]. To shorten the pump pulse, the radiation spectrum was extended due to the effect of self-phase modulation of the pulse passing through a cell filled with a gas and subsequent compensation of an arising chirp using chirping mirrors. This schematic with 50% energy transmission (the loss is due to Fresnel reflection from the cell windows and multiple reflections from the mirrors) made it possible to obtain 1.5-mJ pulses with a FWHM of 35 fs, a value close to the bandwidth-limited duration (the time–bandwidth product is 0.5; Fig. 1, inset).

Fig. 1.
figure 1

(Color online) Sketch of the experimental setup: (1) Cr:forsterite laser system (λ = 1.24 µm, τ = 100 fs (FWHM), and E = 3.5 mJ); (2) stage of compression of the laser pulse (pulse parameters at the output are τ = 35 fs (FWHM) and E = 1.5 mJ); and (3) Michelson interferometer.

Full size image

For the generation of terahertz radiation, a pump pulse was fed to the BNA crystal (6 × 0.8 mm, Swiss Terahertz) oriented for the maximum optical-to-terahertz conversion. The energy of the 100-fs generating pulse was varied using an assembly of a half-wave plate and a polarizer mounted in front of the compressor. A Tydex LPF23.4 filter was used to block the pump radiation and transmit the terahertz radiation. Fourier-transform spectroscopy characterization of the spectrum of the generated terahertz radiation was performed by means of detection of the first-order autocorrelation function (field ACF) in the Michelson interferometer geometry, where the beam splitter was a high-resistivity silicon plate (Tydex HRFZ-Si). The registration of terahertz radiation in the microjoule energy region was performed using a Gentec QE8SP-B-BC-D0 pyroelectric detector with a clear aperture size of about 8 × 8 mm2. To detect energies of few nanojoules, a Tydex GC-1P Golay detector with the clear aperture size of 6 mm was used. Focusing of terahertz radiation to a ~1-mm2 spot at the 1/e2 level during the detection ensured the measurement of the total terahertz pulse energy using both detectors.

RESULTS AND DISCUSSION

Figure 2 shows the dependences of the energy of the terahertz pulse and its generation efficiency on the fluence of the 100-fs pump pulse. As can be seen, there is no saturation of the generation efficiency in the pump fluence range of 1–15 mJ/cm2. Pumping of the BNA crystal by 35-fs pulses with an energy of 1.5 mJ and a fluence of 9.6 mJ/cm2 led to the generation of a terahertz pulse with an energy of 2.2 µJ. The generation efficiency in this case is higher compared to that in the case of pumping by 100-fs pulses with the same pump fluence, which can be explained by an increase in the pump pulse intensity due to the reduction of its duration.

Fig. 2.
figure 2

(Color online) Energy of the terahertz pulse and its generation efficiency versus the 100-fs pump pulse fluence.

Full size image

In the experiment, no breakdown of the BNA crystal was observed under pumping in the fluence range of 1–15 mJ/cm2. It was reported in [23] that the breakdown threshold of the BNA crystal pumped by 0.8-µm pulses with a duration of 50 fs and a repetition frequency of 100 Hz is 6 mJ/cm2, which limited the terahertz pulse energy by 0.54 µJ. In [37], the breakdown threshold was 4 and 2 mJ/cm2 in the case of pumping by 0.8-µm pulses with a duration of 100 fs and a repetition frequency of 0.5 and 1 kHz, respectively. In [27], it was reported that pumping by 1.2-µm pulses with a duration of about 35 fs and a repetition frequency of 1 kHz leads to the breakdown threshold of no less than 10 mJ/cm2, which is due to the decrease in the two-photon absorption coefficient with an increase in the pumping wavelength. Thus, the results of this study supplement these data: pumping by 1.24-µm pulses with durations of 100 and 35 fs and a repetition frequency of 10 Hz yields the breakdown threshold of no less than 16 and 9.6 mJ/cm2, respectively, which allows one to obtain terahertz pulses with an energy of above 2 µJ.

Figure 3 shows the terahertz radiation spectra obtained in the BNA crystal pumped by pulses with a duration of 100 and 35 fs. Since terahertz radiation is generated in the organic crystal according to the mechanism of difference-frequency generation for the spectral components of the pump pulse, the extension of its spectrum with a decrease in the duration induces higher-frequency components in the terahertz spectrum, which is observed experimentally (Fig. 3). Thus, a decrease in the pulse duration leads to a shift of the initial spectrum from 2–5 to 2.5–6.5 THz with the occurrence of an additional peak in the range of 9‒10.5 THz. The terahertz radiation spectrum under pumping by 100-fs pulses lies in the range below 5 THz and has a characteristic dip at 2 THz related to the absorption line, which is in agreement with the data in the literature where the generation was ensured by 1250-nm pump radiation obtained in the parametric amplification scheme [38]. It is important that nonlinear optical enrichment of the pump spectrum without compensation of the arising chirp does not lead to the generation of the higher-frequency component, because the spectral width of terahertz radiation is determined by the instantaneous time overlap of spectral components of the pump pulse in the nonlinear crystal, which is confirmed by the experimental results [28].

Fig. 3.
figure 3

(Color online) Spectrum of the terahertz radiation generated in the BNA crystal pumped by (top panel) 100‑fs pulses with a pump fluence of 15 mJ/cm2 and (bottom panel) 35-fs pulses with a pump fluence of 9.6 mJ/cm2: (solid lines) experimental spectra, (dashed lines) spectra obtained in the finite-difference time-domain method simulation, and (dotted lines) spectra calculated by Eq. (7). The black lines are the noise spectra measured in the absence of the pump pulse.

Full size image

As noted in [28], a terahertz pulse generated in the BNA crystal pumped by 100-fs laser pulses from the Cr:forsterite system has a bandwidth-limited duration of ~660 fs, which was determined from the measured spectral width. Since numerous experimental data obtained in electro-optical measurements of the temporal shape of the terahertz pulse [9, 38] indicate that a nearly bandwidth-limited terahertz pulse is formed in crystals during the generation of terahertz radiation through the optical rectification mechanism, the estimate ~660 fs for the width of the terahertz pulse envelope under pumping by 100-fs pulses is correct. In this case (see Fig. 3), under pumping of the BNA crystal by 30-fs pulses, the main part of the spectrum lying near 4.5 THz and determining to a great extent the temporal shape of the terahertz pulse has nearly the same spectral width as in the case of pumping by 100-fs pulses. In this case, the higher-frequency components in the spectrum (with a smaller spectral amplitude) yield high-frequency oscillations in the temporal pulse shape. Therefore, the terahertz pulse duration at the 30‑fs pump pulse can also be estimated as ~660 fs.

The occurrence of high-frequency components in the terahertz radiation spectrum can be qualitatively described by analytical expressions for the spectrum and temporal shape of the terahertz pulse. In particular, under phase-matching condition in the constant-field and nonabsorbing and nondispersive medium approximations, the spectrum and temporal shape of the terahertz pulse can be written as follows [39]:

$$\begin{array}{*{20}{c}} {{{E}_{{{\text{THz}}}}}(\Omega ) \propto \Omega \int\limits_0^\infty {{{A}_{{{\text{op}}}}}(\omega + \Omega )A_{{{\text{op}}}}^{*}(\omega )d\omega ,} } \\ {{{E}_{{{\text{THz}}}}}(t) \propto \frac{{\partial {{{\left| {{{A}_{{{\text{op}}}}}(t)} \right|}}^{2}}}}{{\partial t}},} \end{array}$$
(7)

where t is the time in the moving coordinate system; z is the propagation coordinate; ω and Ω are the angular frequencies of the pump and terahertz radiation, respectively; and Aop(ω) and Aop(t) are the pump pulse amplitudes in the spectral and temporal representations, respectively. According to Eq. (7), the terahertz pulse in the temporal representation has a shape of the first derivative of the pump field intensity. An analysis of the spectral shape obtained by Eq. (7) shows that the spectral width of the terahertz pulse is proportional to that of the optical pump pulse, while the generation maximum is observed at the frequency ν = \(0.44\sqrt 2 {\text{/}}\pi \tau \), where τ is the FWHM of the pump pulse. Thus, a decrease in the pump pulse duration by a factor of ~3 (from 100 to 35 fs) shifts the maximum of the terahertz radiation spectrum from 2 to 5.6 THz, which is illustrated in Fig. 3.

Differences between the analytical and experimental spectra are related to the dispersion properties of the BNA crystal, which were neglected within the analytical model given by Eq. (7). To take these properties into account, we numerically calculated the terahertz radiation spectrum solving Maxwell’s equations by the above-described finite-difference time-domain method. The calculation results are shown in Fig. 3. The used theoretical model describes the main spectral features of the generated terahertz radiation caused by the dispersion properties of the crystal and propagation effects; the rest differences in the high-frequency region can be explained by the influence of water vapor absorption in air [40] at a relative humidity of 35%. As a result, both the high-frequency shift of the spectrum with a decrease in the pump pulse duration and the manifestation of absorption bands of the BNA crystal in the terahertz radiation spectrum can be described.

The frequency ranges of terahertz radiation generated in the BNA crystal pumped by 35-fs pulses are 2.5–6.5 and 9–10.5 THz, which indicates that the radiation is higher-frequency, in comparison with radiation generated with pumping by 100-fs pulses at a wavelength of 0.8 µm from a Ti:sapphire laser system (0.1–5 THz) and with pumping by 100-fs pulses at a wavelength of 1250 nm based on parametric amplification (0.5–5 THz) [38]. Note that the spectrum obtained in this study contains an additional peak at frequencies of 9–10.5 THz, in contrast to the case of generation in the BNA crystal by 35-fs pulses at a wavelength of 1150 nm [27] produced by a parametric amplification schematic.

CONCLUSIONS

The generation of terahertz radiation in a BNA crystal by femtosecond radiation from a Cr:forsterite laser system has been performed for the first time. It has been found that no saturation of the generation efficiency is observed in the pump fluence range of 1‒15 mJ/cm2; the maximum generation efficiency was 0.1%. It has been shown that a decrease in the pump pulse duration from 100 to 35 fs shifts the terahertz radiation spectrum to higher frequencies and induces an additional spectral component at a frequency of 10 THz, which is indicative of the generation of broadband terahertz radiation at frequencies of 2.5–6.5 and 9–10.5 THz in the BNA crystal. It has been found that simulation of the generation process directly solving Maxwell’s equations by the finite-difference time-domain method taking into account the dispersion properties of the crystal makes it possible to correctly describe the spectrum of the generated terahertz pulse for both the long (100 fs, 24 field cycles) and short (35 fs, 9 field cycles) pump pulses, which cannot be achieved using the slowly varying amplitude approximation. The results obtained make it possible to consider a BNA crystal with pumping by a Cr:forsterite laser system as an alternative to sources based on a lithium niobate crystal or a BNA crystal pumped by Ti:sapphire laser systems.