1. Introduction

Terahertz radiation (THz) finds innumerable applications in spectroscopy and nonlinear control of matter [1,2]. The latter requires intense terahertz radiation that is most commonly generated using a tabletop laser source and a nonlinear medium which can be a plasma or a crystal [3]. Compared to plasma based sources, nonlinear crystals extend compact and simple experimental schemes. Furthermore, the laser to terahertz conversion efficiency is also higher due to their high nonlinear coefficients [3,4]. Organic and non organic crystals can be employed for this purpose [2,4]. Similar to other nonlinear frequency conversion processes, efficient THz generation also requires a large nonlinear coefficient and a small difference between the group refractive index of the pump laser and phase refractive index of the THz pulse [5,6]. Though crystals like LiNbO$_3$ offer large nonlinear coefficients, due to the huge difference in relevant refractive indices of the pump and the generated THz frequencies, fulfilling their phase matching inside the crystal for efficient THz generation can lead to complicated setups [7]. In this regard organic crystals are a viable alternative thanks to their large nonlinear coefficient and easy phase matching geometry [4,8–19]. The most commonly used organic crystals are DAST, DSTMS and OH1 [4]. These crystals typically have intrinsic peak efficiencies about $2\%$ [20] and with optimal focusing geometry peak electric fields up to $83\, \mathrm {MV/cm}$ can be reached [14]. Such intense terahertz radiation enabled the experimental demonstration of higher order nonlinear effects and induced damages in materials [14,21]. However, organic crystals come with their own technological challenges. The most commonly used crystals (DAST, DSTMS, OH1) require pump laser excitation at long wavelengths ($> 1.2\,\mu \mathrm {m}$) which are challenging to obtain at high energies and with good beam quality [4,12,22]. N-benzyl-2-methyl-4-nitroaniline (BNA) is a novel candidate that could overcome the pump wavelength challenge as it can be efficiently used with conventional Ti-Saph lasers [9,15,19]. However, the THz generation properties in BNA at high pump fluence are still unknown. In order to utilize the full potential of BNA crystals for commercial applications a systematic study at the commonly available laser wavelength is required. Here we carry out a meticulous study on terahertz radiation generation process by optical rectification (OR) in BNA crystals at $800\,\mathrm {nm}$ pump wavelength by analyzing the transmitted pump spectrum, generated THz yield, beam profile, pulse duration, spectrum and polarization. The saturation threshold of the incident pump laser fluence for different repetition rates is also determined. By analyzing the transmitted laser spectrum an estimate of the nonlinear refractive index of BNA at $800$ nm was also obtained.

2. Experimental setup

For our studies, two Ti-Saph laser systems at the Institute of Optics and Quantum Electronics at the University of Jena were employed. The first system (Laser1), delivered pulses with $100\,\mathrm {fs}$ duration and energy up to $0.6\,\mathrm {J}$ with a spectrum centered around $800\,\mathrm {nm}$. The repetition rate of the laser system could be varied from $0.5$ to $10$ Hz using a synchronized built-in shutter. The output energy could be varied by adjusting the Pockels cell delay in the amplification stage. Online beam diagnostics such as wizzler, spider, energy meter and far-field monitors are installed in the laser system to control the laser parameters. The second Ti-Saph laser (Laser2) is a commercial turnkey system which operates at $1 \,\mathrm {kHz}$ with a lower output energy of $3\,\mathrm {mJ}$ and $25\,\mathrm {fs}$ pulse duration. Here the repetition rate was varied by controlling the Pockels cell driver in the amplification stage. The energy was measured using a calibrated energy meter and the pulse duration using a third order autocorrelator. A schematic of the experimental setup for THz generation and detection is shown in Fig. 1. For THz yield measurements the BNA crystal mounted on a kinematic stage was placed in the focusing laser beam. The incident laser fluence was varied by moving the crystal along the focal axis of an off-axis parabola (OAP) with $3\, \mathrm {inch}$ diameter and $17\,\mathrm { cm}$ focal length. The crystal c-axis was rotated such that the laser polarization was aligned parallel to it giving rise to the maximum THz signal. A part of the transmitted pump laser pulse was then send to an optical spectrometer to record its spectrum. The pump radiation was first filtered out using a calibrated lowpass plastic filter followed by a $1\,\mathrm {mm}$ thick high resistive float zone (HRFZ) silicon wafer. The transmitted terahertz radiation was then collected and collimated using a $3\,\mathrm {inch}$ OAP with $10.5\,\mathrm {cm}$ focal length before focusing down onto the terahertz detector using another $3\, \mathrm {inch}$ OAP with $7.5\,\mathrm {cm}$ focal length.

 

Fig. 1. Schematic of the experimental setup for the generation and detection of THz pulses from a BNA crystal. Electro-optic diagnostic based on spatial encoding technique is employed to detect the THz radiation. QWP-quarter wave plate, LP-linear polarizer, ALP-linear polarizer as analyzer, ZnTe-Zinc Telluride crystal.

Download Full Size | PDF

Incoherent and coherent detection schemes were employed for the detection and characterization of the THz radiation [23]. The integrated THz yield, polarization and beam profile were measured using a Golay cell detector fitted with a $500\,\mu \mathrm {m}$ thick HRFZ silicon window. Based on these results a coherent detection scheme using the spatial encoding technique was developed to measure the THz pulse duration [23,24]. In the spatial encoding geometry, the THz pulse is incident normal to the $<110>$ plane of the electro-optic (EO) crystal and the linearly polarized optical probe pulse is incident at an angle with respect to the THz pulse. This leads to a different time of arrival for the probe pulse in the horizontal direction and thus enables spatial encoding of the temporal evolution of the THz pulse on the probe beam profile. The birefringence induced by the THz pulse causes phase retardation of the probe pulse which can be turned into spatial modulation of the intensity by using a prism polarizer as an analyzer and a $16$-bit Andor CCD camera. The total time window ($t$) of the setup is defined by the transverse width of the probe ($x$) and the angle ($\theta$) between the terahertz and the probe pulses ($t=\frac {x}{c}\mathrm {tan}(\theta )$), where $\mathrm {c}$ being the velocity of light in vacuum. The temporal resolution is defined by the pixel size of the CCD camera. The induced birefringence was measured using the balanced detection scheme [23]. The output signal of the balanced detection scheme and the phase retardation $\Gamma$ experienced by the probe beam are directly proportional to the THz electric field. The THz electric field can be calculated from the phase retardation $\Gamma$ using the following formula $\Gamma =\frac {n_{o}^{3}r_{41}E_{THz}\sqrt {1+3cos\alpha }\, \omega d}{2c}$. Here $n_{o}$ corresponds to the refractive index of the EO crystal at the optical probe frequency $\omega$ and $r_{41}$ is the electro-optic coefficient of the EO crystal with a thickness $d$. $E_{THz}$ is the amplitude of the THz electric field and $\alpha$ is the angle between the THz electric field and $[-1,1, 0]$ axis of the EO crystal.

In order to implement the EO detection scheme, the Golay cell was replaced with a $500\,\mu \mathrm {m}$ thick Zinc Telluride (ZnTe) crystal cut along the $<110>$ plane. About $1\%$ of the pump beam was coupled out using a leakage mirror to probe the THz pulse and was sent to the ZnTe crystal at an angle with respect to the THz pulse as shown in Fig. 1. The polarization of the probe beam was adjusted by a $\lambda /4$-plate (QWP) and a linear polarizer (LP) to be perpendicular to the pump laser and therefore also to the THz polarization. Temporal overlap between the optical probe and the THz pulse inside the ZnTe crystal was achieved using a high-precision kinematic delay stage with a step size of $5\,\mathrm {\mu m}$. Longer temporal scans lasting several tens of picoseconds were carried out to ensure that the recorded THz pulse is the main signal and reflections arising from optical components in the beam path were removed from the measurement time window. The polarization and beam profile measurements were carried out at a repetition rate of $10\,\mathrm {Hz}$ using a Golay cell and the electro-optic measurements were carried out at $1\,\mathrm {Hz}$ limited by the acquisition time of the $16$-bit CCD camera.

3. Results and discussion

Initially we investigated the influence of the pump laser fluence and repetition rate on the THz yield from a $600\,\mu \mathrm {m}$ thick, $5\times 5\,\mathrm {mm}^2$ BNA crystal at room temperature. The results presented in Fig. 2(a) (blue shaded area) show that for repetition rates $\le 10\,\mathrm {Hz}$ using Laser1 the terahertz yield increases quadratic as expected from an OR process and reaches saturation around $10.5\,\mathrm {mJ/cm}^2$. At this input fluence, the terahertz yield neither diminished nor the crystal showed any visible thermal damage for long measurement times, implying that the thermal dissipation time of the crystal is smaller than $100 \,\mathrm {ms}$. Further increase in the input pump fluence until $20\,\mathrm {mJ/cm}^2$ neither increased the THz yield nor caused any visible thermal degradation of the crystal. Thereafter we carried out measurements for repetition rates higher than $10\,\mathrm {Hz}$ using Laser2. The outcome of these studies also showed a similar trend until $100\,\mathrm {Hz}$ (pink shaded area of Fig. 2(a)), albeit at a faster growth rate reaching saturation around $8\,\mathrm {mJ/cm}^{2}$. Beyond $100\,\mathrm {Hz}$, the THz generation process is becoming more unstable which is more evident at $250\,\mathrm {Hz}$ (green curve) and at $500\,\mathrm {Hz}$ (not shown), where the crystal showed visible thermal degradation even at low input laser fluence ($\sim 4\,\mathrm {mJ/cm}^2$). Based on the data provided by the crystal supplier, organic crystals are known to have a thermal constant of $5-10\,\mathrm { ms}$, which clearly matches with our experimental data. From these measurements it can be concluded that the BNA crystal is safe to operate until $100\,\mathrm {Hz}$ without thermal damage for a pump laser fluence up to $10.5\,\mathrm {mJ/cm}^2$ at room temperature. Furthermore, it can also be said that at higher repetition rates, saturation of THz generation process is reached earlier and a decrease in the THz yield is observed due to increasing thermal load in the crystal. Hence it may be safe to deduce a thermal dissipation time of $\sim 10\,\mathrm {ms}$. However, a precise estimation requires detailed investigation on the thermal load and change in refractive index as a function of the temperature [18,25]. The results presented in Fig. 2 are normalized to the respective maximum THz yield for each repetition rate. Irrespective of the repetition rates an optical to THz conversion efficiency of $0.2 \,\%$ was estimated (Fig. 2(b)) and the maximum THz yield measured by the Golay cell is above $10\,\mu \mathrm {J}$, which was corrected for the losses in the beam path. The pump laser energy has been measured for each repetition rate and found to not vary significantly ($<\, 3\,\%$) for the Laser1. For higher repetition rates using Laser2, the optimal performance was at the factory setting of $1\,\mathrm {kHz}$ and at lower repetition rates the output pulse energy was reduced to $0.8\, \mathrm {mJ}$ while the energy fluctuation was below $1\,\%$. The input pump fluence was accurately estimated by measuring the laser energy and the spot size at the crystal position.

 

Fig. 2. a) THz yield for different pump laser fluences and repetition rates in the region $< 10\,\mathrm {Hz}$ using Laser1 (blue background) and $< 250\,\mathrm {Hz}$ using Laser2 (pink background). Each data set is normalized to its maximum THz yield. b) Normalized conversion efficiency for repetition rates until $10\,\mathrm {Hz}$.

Download Full Size | PDF

Next, the polarization of the THz radiation was measured by introducing a thin wire-grid (WG) polarizer in the collimated beam path and recording the transmitted signal for various rotation angles of the WG. The result presented in Fig. 3(a) suggests that the THz radiation is linearly polarized as expected from the OR process [5,6]. Here $90^\circ$ corresponds to s-polarization or the polarization of the incident laser pulse, which is almost parallel to the THz polarization. The small discrepancy in the polarization state between the pump laser and the generated THz radiation can be attributed to the measurement errors due to the low extinction ratio of the WG at optical wavelengths compared to the THz band [26]. Furthermore, a small ellipticity is evident, which might be caused by the non optimal alignment of the BNA crystal with respect to the incident laser polarization or due to the non uniformity of the crystal.

 

Fig. 3. a) Normalised THz yield with respect to the rotation angle of the wire-grid polarizer. b) THz beam profile measured using $2\mathrm {D}$ knife-edge scan.

Download Full Size | PDF

Thereafter, we measured the spatial profile of the THz beam close to the focus using the $2\mathrm {D}$ knife-edge technique [27]. The knife-edge placed on a high resolution kinematic stage intersected the THz beam in $\mathrm {x}$ and $\mathrm {y}$ directions perpendicular to the propagation direction (z-axis) of the beam. The Golay cell measured the transmitted signal of the unblocked beam as the knife edge moved through the beam in steps of $500\,\mu \mathrm {m}$. The raw data obtained from measurements taken by advancing the knife-edge in the $\mathrm {x}$ ($20$ points) and $\mathrm {y}$ ($11$ points) directions for a fixed $\mathrm {z-}$position is analyzed for both gaussian and non-gaussian beam profiles. The uncertainty arising from shot-to-shot fluctuations were taken into consideration by averaging over $500$ shots. The raw data of the transmitted signal (S) was interpolated and the derivative of the transmitted signals ($S_{1}(x) \propto \int _{-\infty }^{x} I_{1}(x')dx'$) in both $\mathrm {x}$ and $\mathrm {y}$ directions were taken. The outcome was then smoothed and the $\mathrm {x}$ and $\mathrm {y}$ scans were combined to obtain the $2\mathrm {D}$ intensity distribution ($I(x,y)=I_{1}(x)\times I_{2}(y)$). In Fig. 3(b) the outcome of this beam intensity profile analysis with the $2\mathrm {D}$ knife-edge scan is presented. It is clear that the beam diameter is comparable ($\sim 2.0\,\mathrm {mm}$) in both $\mathrm {x}$ and $\mathrm {y}$ directions implying that the focus is devoid of large aberrations caused by the focusing optics. This assumption also agrees well with the gaussian fits for $I_{1}$ and $I_{2}$.

This conclusion was further verified with an independent diagnostic. Single-shot non-collinear EO detection [24] was employed to measure the spot size and the pulse duration of the THz pulse. In this scheme the signal on the EO crystal and thereby on the CCD camera provides both spatial and temporal information simultaneously. The vertical direction on the CCD image corresponds to the beam size while the horizontal direction corresponds to the temporal duration of the signal. A raw image of the signal recorded on the CCD is shown in Fig. 4. At first, we extracted the temporal information of the THz pulse by taking a horizontal line-out of the CCD image. By careful calibration, the pixel value was converted to time and the intensity variation was converted to phase shift due to THz induced birefringence and subsequently to THz electric field. The outcome of this analysis is presented in the top inset of Fig. 4 (blue curve) and shows a single cycle pulse with a peak-to-peak duration of about $500\,\mathrm {fs}$. Taking the absolute value and applying a gaussian envelope, the FWHM pulse duration can be estimated to be around $850\,\mathrm {fs}$. Afterwards, the spectrum of the THz pulse was obtained by taking the Fourier transform (FT) of the time domain data and deconvoluting with the electro-optic function of the crystal (inset red curve). The spectrum extends up to $2.5\,\mathrm {THz}$ with its peak around $1.1\,\mathrm {THz}$ as observed in previous works [12,15]. Removing the lowpass plastic filter after the BNA extended the spectrum to $3\,\mathrm {THz}$, albeit, with a low spectral content. Similarly the dip at the lower end of the spectrum can also be attributed to the lowpass filter. However, it should also be mentioned that small variations in the spectral content beyond $2\,\mathrm {THz}$ were observable when crystals of different thickness were tested (not shown here). In particular, a noticeable reduction in the spectral components beyond $2.5\,\mathrm {THz}$ observed for thicker crystals. Thus, when employing large area crystals, it is important to consider the quality of the crystal in terms of uniformity over the whole area along with surface roughness. This may affect the generated THz yield, spectrum and its focusability.

 

Fig. 4. CCD image of the EO signal obtained using single-shot spatial encoding technique. The horizontal axis provides the temporal information while the vertical axis provides the spatial information. The temporal shape of generated THz pulse is obtained by taking a horizontal line-out (inset blue curve). The spectral content is obtained by Fourier transforming the temporal data. Purple and red curves present the spectra with and without the presence of the lowpass filter employed to remove the residual pump laser light.

Download Full Size | PDF

Next, we evaluated the focal spot size by taking a vertical line-out of the CCD image and estimating the $1/e^{2}$-width and a beam diameter of $1.9\pm \, 0.1\,\mathrm {mm}$ was obtained. With a calculation based on the parameters of the focusing optics ($f=7.5\,\mathrm {cm}$ and $d=1.5\,\mathrm {cm}$) and $\lambda =300\,\mu \mathrm {m}$ and a $M^2$ value of $1.75$ from the z-scan, the THz beam waist diameter after the focus is estimated to be about $1.9 \pm \,0.1\,\mathrm {mm}$, which is very close to the result obtained from knife-edge measurements.The evolution of the THz beam around focus can be obtained by moving the EO crystal in the z-direction. A montage of the evolution of the THz focus is shown in Fig. 5. It can be seen that the wavefront curvature changes smoothly as the pulse is detected at various positions along the propagation axis. Vertical line outs of the images provide the focal spot-size along the z-direction. The outcome of these analysis are presented in Fig. 6(a), which shows that the Rayleigh range ($z_R$) of the THz focus can be estimated to be $6.59\pm 0.17\,\mathrm {mm}$.

 

Fig. 5. A montage of the z-scan measurements showing the evolution of THz pulse through the focus recorded with the non-collinear EO diagnostic. The EO crystal was moved along z-direction with corresponding distances shown in millimeters for each image.

Download Full Size | PDF

 

Fig. 6. a) Change of beam diameter in the y-direction as a function of z-position obtained from EO detection. Blue dots are the experimental data where each data point is an average of $5$ shots and red line represents a non gaussian fit for an $M^2$ value of $1.75$. b) Temporal evolution of the THz pulse recorded by moving the EO crystal in the z-direction.

Download Full Size | PDF

Further analysis of the EO signal from the z-scan presented the convergence of different spectral components around the focus. It is known that THz pulses carrying multiple octaves suffer from temporal phase shift away from the focus resulting in longer temporal durations. Taking a horizontal line-out of the images in Fig. 5 provides the time domain data for each z-position which is presented in Fig. 6(b). Here care was taken to record the THz signal always at the same pixel position on the camera by adjusting the probe delay and correcting for the lateral shift in the image due to the non-collinear geometry of the detection scheme. The systematic temporal shift observed in the electric field measurements can be attributed to the non convergence of high frequency components on axis.

With the complete characterization of the THz pulse, we tried to gain further insight into the THz generation process. To this end, we recorded the spectra of the pump laser pulse before and after its interaction with the BNA crystal. Leakages of the pump beam before and after going through the BNA crystal were collected using fiber coupled photodiodes and the spectra were recorded using high resolution spectrometers. Figure 7(a) presents the normalized spectra of the laser pulse, where a visible decrease in the spectral components above $790\,\mathrm {nm}$ compared to the shorter wavelength side of the spectrum is observed. It is known that better THz yield is observed at pump wavelengths longer than $800\,\mathrm {nm}$ [15].Furthermore, a clear broadening of the transmitted laser spectrum for high THz yield corresponding to higher pump fluence (see Fig. 7 b) is also evident. Moreover the observed spectral broadening is comparatively symmetric to the central wavelength unlike previous observations where only a red shift of the pump spectrum was observed [15]. Thus a credible explanation could be the self-phase modulation of the laser pulse itself due to the large nonlinearity generated inside the BNA crystal by the pump pulse [4]. Subsequently, we estimated the amount of broadening for different THz yields. The central wavelength ($\lambda _c$) of the incoming laser spectrum was at $797.8\,\mathrm {nm}$ with a bandwidth of $50.1\,\mathrm {nm}$. At maximum THz yield, corresponding to maximum laser energy on the BNA, the resultant broadening of the pump pulse was $14.3\,\mathrm {nm}$ as seen in Fig. 7(b). A clear close to symmetric broadening of the transmitted laser spectrum on both sides of the central wavelength is also visible with increasing pump laser energy and thereby THz yield. However, the amount of broadening was less than the inverse of the pulse duration. It is also to be noted that a slight redshift of the central wavelength is observed with increasing THz yield or pump energy. Nevertheless, the broadening of the spectrum shows a linear dependence on the THz energy produced within the BNA crystal and therefore the induced nonlinearity leading to self-phase modulation. In other words, the leading and the trailing edge of the pulse are extended to each side with the centroid shifting towards longer wavelengths [28,29]. The above features are clear indications of self-phase modulation of the pump pulse inside the BNA crystal. Based on the measured spectral broadening we performed a simple estimate of the nonlinear refractive index ($\mathrm {n_{2}}$) [30]. For a crystal thickness of $600\,\mathrm {\mu m}$ and incident laser intensity, $I=1.1\times 10^{11}\,\mathrm {W/cm}^2$, the nonlinear refractive index is estimated to be $6.76 \times 10^{-14}\,\mathrm {cm^2/W}$. Considering the high value of the effective nonlinear coefficient of BNA, the estimated value of the nonlinear refractive index is reasonable and is comparable to other THz crystals like ZnTe and ZnSe [31].

 

Fig. 7. a) Relative change of laser spectrum before (dotted black line) and after (solid lines) the BNA crystal for different THz pulse energies. Each line is an average of $100$ shots. b) Spectral width and central wavelength shift of the transmitted spectra for different THz pulse energies. The central wavelength of the incoming laser spectrum was at $797.8\,\mathrm {nm}$ with a bandwidth of $50.1\,\mathrm {nm}$.

Download Full Size | PDF

Finally, we also changed the temporal duration of the pump pulse from $100\,-200\,\mathrm {fs}$ by altering the second order spectral phase using the Dazzler [32] setup installed at the front end of the laser system. However, no significant change was observed either on the THz yield or on the temporal duration of the generated THz pulse for small variations in the pulse width. However, when the thickness of the BNA crystal was changed a clear dependence on the THz yield was visible (see Fig. 8(a)). Here crystals with different thicknesses and similar area were exposed to a laser fluence of $8\,\mathrm {mJ/cm^2}$ at $10\,\mathrm {Hz}$. It is clear that for our laser parameters crystals of thickness between $500-600\,\mu \mathrm {m}$ are suitable candidates to obtain maximum THz yield. This value agrees well with the calculations reported in previous works [9]. To better understand the dependence of THz yield on crystal thickness and thereby to calculate the optimum crystal thickness for our laser system, we numerically calculated the refractive index, absorption and coherence length for our laser parameters [10,22,33]. The results are shown in Fig. 8(b)-(d). Figure 8(b) shows the real part of the refractive index for BNA at THz frequencies and group refractive index at our pump laser wavelength. It is also clear that excellent phase matching is possible for $800\,\mathrm {nm}$ pump wavelength and THz radiation below $2\,\mathrm {THz}$. Figure 8(c) presents the absorption coefficient of the BNA at THz frequencies, which shows a maximum of $220\,\mathrm {cm}^{-1}$ at $2.1\,\mathrm {THz}$. Furthermore, it can be seen that the absorption coefficient in the remaining spectral band is $\sim 150\,\mathrm {cm}^{-1}$ and thus using thinner crystals could lead to a broader THz spectrum [10,15,22,34]. By taking into consideration the above values of refractive index and absorption coefficient, Fig. 8(d) presents the relation between crystal thickness and normalized THz spectral yield and bandwidth. It shows that thinner crystals are more suitable for generating broader spectra. For the crystal employed in our measurements, the spectral peak of the THz radiation expected to be around $1.8\,\mathrm {THz}$ while the EO measurements (Fig. 4) shows a peak above $1.1\,\mathrm {THz}$. Though the spectral measurements agree with previous observations of Shalaby et al., [15], the discrepancy between the outcomes of calculation and experiments can be attributed to the crystal quality and on the accurate estimation of refractive index [10,18,22,35,36].

 

Fig. 8. a) Normalized conversion efficiency for crystals of different thickness each normalized by their squared thickness. b) Refractive index, c) Absorption and d) Numerical estimate of the normalized spectral yield for our laser parameters.

Download Full Size | PDF

4. Conclusion

In summary, we have investigated the THz generation process in BNA crystals at room temperature using $800\,\mathrm {nm}$ pump wavelength by characterizing the spatial, temporal and polarization properties of the THz pulse along with the transmitted pump laser spectrum. The dependence of THz yield on incident laser fluence is quadratic and reaches saturation for a maximum fluence of $10.5\,\mathrm {mJ/cm}^2$ for repetition rates below $10\,\mathrm {Hz}$. At repetition rates higher than $10\,\mathrm {Hz}$, saturation is observed earlier and above $100\,\mathrm {Hz}$, the THz generation process is rather unstable at high laser fluences and saturation and thermal damage are observed at $4\,\mathrm {mJ/cm^2}$. This observation is supported by the fact that organic crystals for THz generation are known to have a thermal time constant in the range of $5$ to $10\,\mathrm {ms}$, implying that the effect of heat due to the average power of the pump laser on the efficiency and damage threshold starts to appear above $100\,\mathrm {Hz}$. The optical to THz conversion efficiency is estimated to be $0.2\%$ neglecting the atmospheric losses and the peak-peak value of the electric field at the focus is $\sim 4\,\mathrm {MV/cm}$. The generated THz radiation has a spectral peak around $1.1\,\mathrm {THz}$ and a smooth beam profile with a quality factor of $M^2=1.75$ enabling focusing to $1.9\,\mathrm {mm}$ diameter ($1/e^2$) focal spot with a $f/1$ optics. The spectral content of the generated radiation is greatly influenced by the crystal thickness and quality. Finally, recording the transmitted pump laser spectra allowed us to calculate the nonlinear refractive index of BNA crystals at $800\,\mathrm {nm}$, providing a first estimate in the order of $6.76\times 10^{-14}\, \mathrm {cm^2/W}$. Unlike other nonlinear crystals for THz generation BNA crystals are less studied and it would be worthwhile to further investigate the thermo-optic effect of the crystal to increase the laser fluence at repetition rates higher than $100\,\mathrm {Hz}$, thereby making them a viable competitor to THz sources in commercial THz-TDS systems [37]. Despite this current limitation BNA crystals extend as a potential alternative to develop an ultra-compact system for spectroscopic studies due to the simple phase matching geometry compared to nonlinear crystals like LiNbO$_{3}$ [4,7,20].