1. Introduction

High power, ultrafast, high repetition rate coherent radiation in the deep ultraviolet (DUV) are of great demand due to their variety of applications including ultrafast time resolved measurements [1,2], laser ablation [3], photo-lithography [4] and biomedical applications [5]. Due to the unavailability of suitable gain medium lasing at DUV, this part of the electromagnetic spectrum is accessed efficiently through nonlinear frequency up-conversion of high power ultrafast solid-state lasers working in visible and near infrared (NIR). Owing to the advancement in the fibre based laser technology, compact, air cooled fibre laser systems are now commercially available delivering high power, high repetition rate, stable radiation in the NIR [6,7]. Single-pass fourth harmonic generation (FHG) of such femtosecond fibre lasers operating at NIR can be an excellent cost effective method in generating robust ultrafast DUV radiation with small footprints. The efficient generation of high power DUV radiation relies not only on the effective nonlinear coefficient and length of the medium, but also on some critical parameters like temporal, spatial walk-off [8] and the pump focusing condition [9]. Owing to the low UV transparency cut off at 190 nm and high second order nonlinearity, Beta barium borate (BBO) crystal has become a natural choice for this purpose [10–12], however, appreciable spatial walk-off have undesirable effect during the up-conversion process demanding optimum choice of crystal length and beam focusing condition. Unlike the continuous-wave and long pulse regime, the onset of temporal walk-off become increasingly significant in case of femtosecond wave-mixing processes. Therefore, it is imperative to have a proper selection of experimental parameters through the systematic studies for the successful generation of DUV radiation with substantial output power/energy as required for practical applications [1].

Efforts have been made to generate DUV power at 258 nm wavelength using single-pass FHG of an ultrafast fibre chirped-pulse amplification system providing 80 W at low repetition rate [13]. While the presence of high average power of the amplified laser system produces substantial output power even at a lower near-IR to UV conversion efficiency of 5.7% [13], however, one need appreciable single-pass FHG efficiency to generate DUV radiation from a high repetition rate laser. Here we report, to the best of our knowledge, the first experimental study on the FHG of fibre based high repetition rate femtosecond source centered at 1064 nm with 4.8 W of output power, generating DUV wavelengths at 266 nm with power as high as 616 mW and corresponding to a single-pass near-IR to DUV conversion efficiency of ∼ 12.8 % at a repetition rate of 78 MHz. We also present the systematic study on the crystal, and focusing parameters influencing the overall conversion efficiency and output characteristics of the DUV radiation. The experimental setup used for the generation of DUV is illustrated in Fig. 1. A Yb- fibre laser, producing output pulses of temporal width of ∼ 260 fs at a repetition rate of 78 MHz with a spectral width of 15 nm centered at 1064 nm, is used as the fundamental laser in the present experiment. The laser has a maximum average output power of 4.8 W in Gaussian profile with a beam waist radius of ∼ 1 mm. A combination of a half-wave plate, (λ/2), and a polarizing beam splitter cube, PBS, is used to control the power of fundamental laser to the experiment. Another λ/2 plate controls the polarization of the fundamental laser depending upon the crystal orientation for perfect phase-matching. A 20-mm long and 3 mm × 3 mm in aperture LBO crystal, C1, frequency doubles the fundamental laser into green under non-critical phase-matching. The lens, L1, of focal length, f = 50 mm focuses the fundamental beam at the center of the nonlinear crystal, C1, to a beam waist radius of, w_p ∼ 18 μm, corresponds to a beam focusing parameter, ξ = L/b = 6.6. Here, L is the length of the crystal, b, is the confocal parameter given by, $b=2\pi n{w}_p^2/\lambda$, n is the refractive index of the crystal at fundamental wavelength, λ = 1064 nm. The crystal is housed in an oven to maintain the phase-matching temperature with a stability of ±0.1°C. End faces of the crystal are antireflection coated (R < 0.1%) for both fundamental and green wavelengths. The green beam generated in crystal, C1, is extracted from the residual fundamental radiation using a pair (only one shown in figure) of dichroic mirrors, S1, having transmission, T > 99%, at 532 nm and reflection, R > 99% at 1064 nm, and collimated using the lens, L2, of focal length, f =100 mm. A λ/2 plate is used to control the polarization of the green beam with respect to the orientation of the nonlinear crystal for frequency doubling of green into DUV. Two beta barium borate (BBO) crystals, C2, of length, 2 mm and 5 mm, with cut angle, θ = 47.56° (ϕ = 90°) are used for type-I (o + o → e) frequency-doubling of green into DUV. Both the BBO crystals have an aperture of 3 mm × 4 mm and anti-reflection coating for green and DUV wavelengths. Different lenses, L3, of focal lengths, f = 25, 50, 100, 150 and 200 mm are used to focus the green beam at the center of the BBO crystals to study the optimum focusing condition for high DUV conversion efficiency. A set of four (only one shown in figure) dichroic mirror, S2, with a high reflection (R > 99%) at 266 nm and high transmission (T > 95%) at 1064 nm, and 532 nm is used to separate the DUV radiation from residual radiations.

 

Fig. 1 Schematic illustration of the experimental setup used for the generation of ultrafast DUV radiation. λ/2, half-wave plates at different wavelengths; PBS, polarizing beam splitter cube; L1–3, lenses; C1, LBO crystal in temperature oven; C2, BBO crystal; PM, power meter; S1–2, dichroic mirrors.

Download Full Size | PDF

First, we have studied the SHG characteristics of the LBO crystal to optimize the single-pass efficiency to achieve maximum green radiation. The results are shown in Fig. 2. Inset (a) of Fig. 2 shows the variation of SH power with the set oven temperature. While pumping with a fundamental laser power of 4.8 W, the LBO crystal was found to have an temperature acceptance bandwidth of 8.1°C centered at 149.5°C. Keeping the crystal temperature fixed at 149.5°C, we studied the variation of SHG power and efficiency as a function of the input fundamental power. As evident from Fig. 2, the SHG power increases linearly with the fundamental power resulting in a maximum SHG power of 2.4 W for the available fundamental power of 4.8 W at a single-pass SHG efficiency of ∼ 50%. While one can expect a quadratic dependence of SHG power to the fundamental power, the linear variation of SHG power to the fundamental power can be attributed to the saturation effect arising from high pump depletion, back conversion, and thermal phase-mismatch effects [14]. The saturation effect is evident from the variation of SHG efficiency with the fundamental power. The SHG efficiency increases from 26% to 50% with the increase of pump power from 0.52 to 3.1 W, however, with further increase in fundamental power, the SHG efficiency remains almost constant around 50%. We further characterized the SHG beam in terms of its characteristic parameters such as temporal, spectral and spatial profile important for FHG process. The far-field intensity profile of the SHG beam, as shown in the inset (b) of Fig. 2, has a TEM₀₀ spatial mode with M² < 1.2 due to the use of noncritical phase-matching of the LBO crystal. Using an intensity autocorrelator (Femtochrome FR-103XL), assuming a sech² pulse shape, the generated SHG radiation is found to have a temporal width (full width at half maximum, FWHM) of τ_sh = 190 fs. Similarly, using a CCD based spectrometer (Ocean Optics, HR 4000), we measured the spectrum of the SHG beam, as shown by the inset (c) of Fig. 2, of width (FWHM) ∼ 1.8 nm centered at ∼ 532 nm corresponding to a time-bandwidth product of 0.36.

 

Fig. 2 Power scaling characteristics of single-pass SHG of femtosecond fundamental laser in a non-critically phase matched 20 mm long LBO crystal. Insets, (a) Dependence of SH intensity on the crystal temperature (line is guide to eye). (b) Spatial intensity profile, and (c) (solid circles) spectrum of SH beam, (solid line) Gaussian fit to the measured spectral data point.

Download Full Size | PDF

Using the green beam at 532 nm, we have studied the FHG characteristics of BBO crystals of two different lengths, L = 2 mm and L = 5 mm, producing DUV radiation at 266 nm. In order to find the optimum focusing condition for highest single-pass FHG conversion efficiency, we have pumped the crystals at a constant green power of 1.5 W, using lenses of different focal lengths. We have measured the green beam waist radius of w_g ∼ 104, 78, 52, 26, and 12 μm while focused with the lens (L3) of focal lengths, f = 200, 150, 100, 50 and 25 mm corresponding to the focusing parameter, ξ = 0.15, 0.27, 0.6, 2.4, and 11.2 and ξ = 0.37, 0.67, 1.50, 5.99, and 28.1 for 2 mm and 5 mm long BBO crystals, respectively. The variation of DUV power with the focusing parameter for both crystals are shown in Fig. 3. As depicted in Fig. 3, the DUV power (open circle) varies from 358 mW at ξ = 0.15 to 199 mW at ξ = 11.2 with a maximum of 450 mW at ξ = 0.27 for 2 mm crystal. On the other hand, the 5 mm long crystal shows DUV power (solid circles) variation from 450 mW at ξ = 0.37 to 215 mW at ξ = 28.1 with a maximum power of 512 mW at ξ = 0.67. To gain further insight of the experimentally measured focusing dependent FHG power, we fitted our data points with the dimensionless function, h(A,B,ξ) [8]. Here, A and B represents the temporal and spatial walk-off parameters respectively. For the BBO crystal, the spatial walk-off angle is found to be ρ = 85 mrad [15]. On the other hand, for the measured pulse-width, τ_sh = 190 fs of the green beam, the group velocity mismatch (GVM) among the interacting waves in FHG process, is calculated to be, β = 573 fs/mm. The temporal walk-off length measured for BBO crystal is 0.3 mm. By incorporating the values of these parameters in the formula of Ref. [8], we have calculated the A and B parameters to be (A, B) = (7.1, 8.5) and (17.8, 13.4) for 2 mm and 5 mm long BBO crystals respectively. Here, A and B parameters represent the temporal and spatial walk-off effect of the crystal. As evident from Fig. 3, our experimental results (solid and open circles) showing variation of DUV power with focusing condition for 5 and 2 mm crystal, follow closely the respective theoretical fits (solid and dash lines). We note that the theoretically predicted optimum focusing condition for 2 and 5 mm long crystals are ξ = 1.6 and 1.8 respectively. Although we have measured highest DUV power at ξ = 0.27 and 0.67 from 2 mm and 5 mm long crystals, respectively, the theoretical study indicates the possibility of further increase in the DUV power with ξ = 1.6 − 1.8. It is important to note that, unlike our previous report [7], the present observation clearly indicates that in presence of high values of A and B parameters, the optimum focusing condition for frequency doubling of ultrafast lasers is influenced by both temporal and spatial walk-off effect of the nonlinear crystal.

 

Fig. 3 Dependence of DUV power to the pump beam focusing parameter. Solid and open circles represent experimental data for 2 mm and 5 mm long BBO crystals respectively. Lines (solid and dotted) are theoretical fit to the experimental results.

Download Full Size | PDF

Knowing the optimum focusing condition for both the crystal lengths and the generation of higher DUV power from a longer crystal, we focused the green pump beam to the 5 mm long BBO crystal using lens of focal length, f = 150 mm (ξ = 0.67) and studied the power scaling characteristics of the FHG process. It must be noted that out of the 2.4 W green power generated by the LBO crystal only 2.1 W is available for the BBO crystal for harmonic generation due to loses during routing of the beam. The results are shown in Fig. 4. As evident from Fig. 4, the DUV power (solid circles) increases with the green power providing DUV radiation in access of 0.6 W for the green power of 2.1 W. The single-pass conversion efficiency is measured to be ∼ 29.3% for green to DUV, and ∼ 12.8% for near-IR to DUV. Similarly, the conversion efficiency (open circles) increases from 9.5% to 24.4% with the increase of green power from 0.14 W to 0.9 W, however, with further increase in green power to 2.1 W, the conversion efficiency shows very small or no variation. On contrary to quadratic dependence of DUV power on the green power, here, the linear variation of DUV power with green pump power at a slope efficiency of 31.2% can be attributed to the saturation effect due to high parametric gain arising from the combination of high intensity of the SH pulses (pulse width ∼ 190 fs), long length (5 mm), and high nonlinear coefficient (d_eff ∼ 1.7 pm/V) of the BBO crystal. To ascertain the saturation effect in DUV conversion efficiency, we have recorded the variation of DUV power with the square of the green pump power with the results shown in the inset of Fig. 4. As evident from the inset of Fig. 4, the linear variation of DUV power with the square of the green power is only maintained at lower power (< 0.6 W). However, at higher green power, the DUV power deviates from the expected linear dependence, clearly confirming the saturation effect in single-pass FHG of ultrafast laser in BBO crystal.

 

Fig. 4 Variation of DUV power and single-pass conversion efficiency with the green pump power for 5 mm long BBO crystal. (Inset) Dependence of DUV power with the square of the green pump power. Lines are guide to eye.

Download Full Size | PDF

Further, we have studied the focusing dependent spatial profile of the DUV radiation and corresponding angular acceptance bandwidth of the 5 mm long BBO crystal. Pumping the crystal with green power of 1.5 W with lenses of different focal lengths, we have recorded the far-field transverse intensity profiles of DUV radiation > 1 m away from the source and calculated the beam ellipticity, ∊ = 1 − w_x/w_y, where, w_x and w_y are the beam waist radii along x- axis and y- axis, respectively. The results are shown in Fig. 5. As evident from Fig. 5, the percentage of beam ellipticity (solid circle) rapidly increases from 38% at ξ = 0.37 to 76% at ξ = 5.99, however, with further increase of focusing to ξ = 28.1 results the beam ellipticity to be constant ∼ 83%. To understand such focusing dependent beam ellipticity, we have calculated the variation of spatial walk-off length of BBO crystal with the focusing. Using the green beam waist radius, w_g ∼ 104, 78, 52, 26, and 12 μm corresponding to the focusing parameter, ξ = 0.37, 0.67, 1.50, 5.99, and 28.1, we calculate the spatial walk-off length, L_s = 2w_g/tan(ρ), to vary from 3.6, 2.7, 1.9, 0.9 and 0.4 mm for 5 mm long BBO crystal. Here, w_g is the beam waist radius of green beam estimated at the centre of the crystal and ρ is spatial walk-off angle of the BBO crystal. At loose focusing, ξ = 0.37, the spatial walk-off length is comparable to the crystal length, as a result the generated DUV has lower beam ellipticity. Given the constant angular phase-matching bandwidth of the nonlinear crystal, with tighter focusing to ξ = 5.99, the increase in the beam divergence and associated decrease in the walk-off length results increased DUV beam ellipticity. However, further increase in the focusing parameter up to ξ = 28.1, the spatial walk-off length is very small with respect to the crystal length. As a result, the majority of the pump beam falls outside the angular phase-matching bandwidth of the crystal and results negligible change in the beam ellipticity. Using a 2 mm long crystal we observe similar variation of the beam ellipticity with focusing. The beam ellipticity varies from 10% at ξ = 0.15 to 80% at ξ = 11.2. To get further insight, we have measured the change in the angular acceptance bandwidth of the 5 mm crystal with focusing. The results are shown in Fig. 5. As evident from Fig. 5, the angular acceptance bandwidth (open circle) of the crystal follows the similar trend as beam ellipticity with focusing varying from 14 mrad at ξ = 0.37 to 56.4 mrad at ξ = 28.1. Such correlation between the angular acceptance bandwidth and ellipticity are due to the direct consequence of variation of spatial walkoff due to beam focusing. It should be noted that, the beam ellipticity can be easily corrected using a pair of cylindrical lenses [11] and there are recent studies towards the compensation of spatial walkoff using crystal stack [16]. The typical variation of DUV intensity with pump incidence angle with respect to the crystal for pump focusing of ξ = 0.67 is shown in the inset of Fig. 5. The systems exhibits a angular acceptance bandwidth of 15.2 mrad. Owing to the restriction in the working spectral range of our autocorrelator (410–1800 nm), we could not characterize the temporal properties of the DUV source. The spectrum of the generated DUV was measured using the CCD based spectrometer. The generated fourth harmonic of the fundamental laser at 1064 nm has a spectral width (FWHM) of 1.5 nm centred at 266 nm. The DUV radiation has a repetition rate of ∼ 78 MHz same as that of the fundamental laser.

 

Fig. 5 Variation of DUV beam ellipticity and the angular phase-matching acceptance bandwidth with the green beam focusing parameter, ξ (The lines are guide to eye). Inset (a) (solid blue circles) Spectrum of the generated DUV radiation, (solid line) Gaussian fit to the measured spectrum, (b) (blue open circles) Dependence of DUV power with the pump incidence angle for a pump focusing parameter of ξ = 0.67, (Solid line) sinc² fit to the experimental data.

Download Full Size | PDF

We also have measured the power stability of our DUV source while focusing at optimum focusing condition, f = 150 mm (ξ = 0.67) with the results shown in Fig. 6. Since the power stability of the DUV output beam can be influenced by the power stability of the pump beam, we have measured the passive power stability of the fundamental, green in Fig. 6. The fundamental laser has a measured peak-to-peak power fluctuation 0.5% over 2.5 hours at the maximum power of 4.8 W. The green beam generated from temperature controlled LBO crystal has relatively higher peak-to-peak power fluctuation of ∼ 2% at the maximum output power of 2.4 W over the same period of time. The DUV power shows a peak-to-peak power fluctuation of ∼ 4% over 2.5 hours. Although the BBO crystals show long term power drop especially for high power DUV sources [11] due to the thermal effects arising from two photon absorption [17] and color centers [18], we neither observed any long term power drop nor any crystal damage even after long-term operation. Such observation clearly confirms the potential of our source for long term applications. The inset of Fig. 6 shows the far field intensity distribution at farfield of the DUV source for the pump focusing parameter, ξ = 0.67.

 

Fig. 6 Power stability of the fundamental laser, green beam and the DUV radiation over 2.5 hours. (Inset) Transverse intensity profile of the generated DUV > 1 m away from the source.

Download Full Size | PDF

In conclusion, we have demonstrated the fourth harmonic generation of a high power, high repetition rate femtosecond fibre laser at 1064 nm generating DUV radiation at 266 nm through single-pass two stage frequency doubling involving LBO and BBO crystals. Using a 20 mm long LBO crystal, we have generated green radiation of 2.4 W at a single-pass conversion efficiency of ∼ 50%. Again by single-pass frequency-doubling of the green in a 5 mm long BBO crystal, we have generated DUV power as high as 616 mW with a single-pass near-IR to DUV conversion efficiency of ∼ 12.8%. We have measured the peak-to-peak power fluctuation of the frequency-doubled green and DUV to be 4% over 2.5 hours without any degradation at even after long-term operation.