1 Introduction

Calcium isotopes have been useful in many applications. For example, Ca-48 is used as a raw material for the measurement of the neutrinoless double beta decay and as the ion beam source in the search for new heavy elements in accelerators [1, 2]. In addition, Ca-41 can be used as a tracer to study the metabolism of living things. Therefore, an efficient isotope analysis and isotope separation techniques have been previously studied, and in some cases, laser-based spectroscopy has been performed to obtain the corresponding spectroscopic coefficients for an efficient isotope analysis and isotope separation.

Recently, we proposed calcium isotope separation using the optical pumping method and considered the 4 s2 1S0–4s5p 1P1 transition of calcium as an optical pumping transition. This method was used to produce isotopically pure samples of Ca+ ions [3–5]. We also demonstrated isotope-selective optical pumping and ionization using the previously reported 272-nm beam [6]. To estimate the optical pumping efficiency, we need the definite value of the oscillator strength of the 4 s2 1S0–4s5p 1P1 transition of calcium. Therefore, we will measure the oscillator strength by observing the power-broadening effect. For this work, we need a tunable watt-class single-frequency 272-nm beam.

Typical methods for generating a 272-nm beam include frequency doubling of a dye laser and frequency tripling of a Ti:sapphire laser [7]. However, these methods have several disadvantages, such as the complexity of the system, frequent maintenance, and limited operating power because of the limited power of the fundamental beam [8]. Recently, frequency doubling and quadrupling based on fiber amplifiers or lasers were demonstrated at 1064 nm [9–12]. Because a DOFA has advantages from both the diode laser and the fiber amplifier, such as good frequency tunability, good beam quality, reliability, compactness, and the availability of high-power operation, it enables the realization of similarly advantageous visible and ultraviolet lasers when simple frequency conversion techniques are used.

In this work, we considered two frequency conversion stages of a DOFA: One was a single-pass frequency-doubling stage with a Mg-doped periodically poled stoichiometric LiTaO3 (PPSLT) crystal for the 544-nm beam, and the other was a cavity enhanced frequency-doubling stage with a bulk beta barium borate (BBO) crystal for the 272-nm beam. For the first frequency-doubling stage, we chose the single-pass second-harmonic generation configuration with the Mg-doped PPSLT crystal because of the simplicity of the system [9–12], even though the cavity enhancement technique achieves higher conversion efficiency and power [13–16]. For the second frequency-doubling stage (fourth-harmonic generation stage), we considered the cavity enhancement technique because no other simple frequency conversion technique for high-power ultraviolet generation currently exists.

In this paper, we briefly introduce the experimental setup for second- and fourth-harmonic generations in Sect. 2, and the experimental results and discussion are in Sect. 3. In addition, we also discuss the application to calcium spectroscopy in Sect. 4.

2 Experimental setup

In this experiment, we used a diode-oscillator fiber-amplifier system and two frequency-doubling stages to generate the 544- and 272-nm beams. Figure 1 shows a schematic diagram of a DOFA, the second-harmonic generation stage, and the fourth-harmonic generation stage. We used an external-cavity diode laser (ECDL) in a Littman configuration with a bandwidth of approximately 1 MHz as a seed laser. The seed beam of approximately 2 mW was delivered to a Yb-doped fiber amplifier (NuAmp, Nufern Inc.) through a polarization-maintaining single-mode fiber. The fiber amplifier consisted of three amplification stages and produced a 1088-nm beam with a power of approximately 30 W, a Gaussian profile, and an M 2 factor of less than 1.05. The wavelength of the seed laser was far from the gain center of the Yb-doped fiber amplifier, and thus we checked the strength of the amplified spontaneous emission (ASE) with an optical spectrum analyzer (AQ6315A, ANDO) and observed that the ratio of the laser signal peak to the ASE peak was more than 32 dB at approximately 30 W.

Fig. 1
figure 1

Experimental setup for the diode-oscillator fiber-amplifier and the second- and fourth-harmonic generations

Full size image

The infrared beam passed through an optical isolator and a three-lens set and focused on a Mg-doped PPSLT crystal for the second-harmonic generation in a single-pass configuration. The crystal had an approximately 8.5-μm poling period and was mounted on an oven for phase matching by tuning the temperature. We compared the conversion efficiencies of two crystals with different lengths of 20 mm (Oxide Inc.) and 30 mm (HC photonics Inc.). The 544-nm beam was separated by two dichroic mirrors after the PPSLT crystal and delivered to the fourth-harmonic generation stage.

The beam size and divergence of the 544-nm beam were controlled by a lens set to match the spatial mode of the green beam to that of the enhancement cavity. The beam was incident on the bow tie ring cavity employing a Brewster angle-cut BBO crystal with a length of 10 mm. The crystal was located between two concave mirrors with a focal length of approximately 50 mm, and the free spectral range of the cavity was approximately 460 MHz. The input coupling mirror had a transmittance of approximately 2.5 %. The longitudinal mode of the cavity was matched to the green beam frequency by the Hansch-Coullioud technique [17]. The ultraviolet beam power was measured by a power meter after the band-pass filter and the collimation optics.

3 Experimental results and discussion

We obtained frequency conversion efficiency with respect to the focusing conditions, crystal length, and position of the focus for several incident laser powers. We controlled the beam waist by varying the positions of three lenses, placed the crystal at the focus, and moved the crystal back and forth to optimize the power. The 20- and 30-mm-long crystals produced 6 and 9.2 W at 544 nm with 34- and 36-μm beam waists, respectively. Based on this crystal length dependence, we chose a 30-mm-long crystal for investigation of its beam characteristics. The full width at half maximum (FWHM) of the temperature tuning curve was approximately 2.4 °C at 0.8 W of the incident infrared beam, and the phase-matching temperature was approximately 62 °C, as shown in Fig. 2.

Fig. 2
figure 2

Temperature tuning curve for the 30-mm-long MgO-doped PPSLT crystal at an incident infrared power of about 0.8 W. The poling period was approximately 8.5 μm

Full size image

Figure 3 shows the 544-nm beam power versus the fundamental beam power at the optimum beam waists for the (a) 20-mm and (b) 30-mm-long crystals. The corresponding conversion efficiency was measured to be approximately 30 % for the 30-mm-long crystal. The dotted line is the quadratic function that represents the power characteristics of SHG in the low-power region, and the deviations from the dotted line are caused by the thermal dephasing effect, the frequency back conversion, and the pump depletion. A decrement of conversion efficiency above 25 W can be attributed to the thermal dephasing and frequency back conversion. The oven temperature was slightly changed to optimize the power, when the incident beam power was changed. We decreased the temperature by 0.7 °C for power optimization at the maximum incident power. The changes of the optimum temperature may originate from the phase mismatching arising from local heating due to the absorption of the fundamental and frequency-doubled waves in the crystal. But, we could not observe clear saturation for the 20-mm-long crystal. We recorded the data when the conversion power remained stable for a couple of minutes for each data point.

Fig. 3
figure 3

Second-harmonic power versus the fundamental beam power at the optimum beam waists for the a 30-mm- and b 20-mm-long crystals. Hollow triangles c represent the corresponding conversion efficiency for the 30-mm-long crystal. The dotted line represents the quadratic function, which gives the expected power characteristics in the low-power region

Full size image

Figure 4 shows the SHG powers versus the beam waist of the incident infrared beam for three incident power levels. In this experiment, we optimized both the oven temperature and the focus position for every incident power and beam waist. The optimum waist is generally expected to be approximately 29 μm according to the Boyd and Kleinman equation [18]. In our experiment, the optimum beam waist was measured to be larger than the expected one, and the SHG power was insensitive to the beam waist. The larger optimum beam waist may be analyzed by thermal lensing, but the insensitivity of the power to the beam waist around optimum condition could not be explained.

Fig. 4
figure 4

Second-harmonic generation power as a function of the beam waist at the focus for three incident laser powers

Full size image

The beam shape and beam quality factor of the 544-nm beam are important for mode matching the 544-nm beam to the enhancement cavity. Therefore, we measured the beam shape and beam quality factor of the frequency-doubled beam with respect to the green beam power. Figure 5 presents the spatial profile of the 544-nm beam at 5 W. The profile was measured using a photodiode and pinhole set moving along the transverse direction of the 544-nm beam, and the profile was fit well with a Gaussian function. But the beam had the elliptical shape which arose from the astigmatism, and it was partially compensated for by tilting the collimation lens.

Fig. 5
figure 5

Spatial profile of the 544-nm beam at a power of about 5 W. The black dots represent the values measured by a moving pinhole along the spatial direction, and the line represents the curve fit with a Gaussian function

Full size image

Figure 6a shows an example of the beam radii as a function of the position along the beam propagation direction around the beam waist when the green beam was focused by a lens. The power was approximately 1.5 W. The M 2 factor of the green beam was obtained by fitting data with the following propagation equation.

$$\omega(z) = {\omega_0}{\left[ {1 + {{\left( {\frac{{z{M^2}\lambda }}{\pi \omega_0^2}} \right)}^2}} \right]^{1/2}}$$
(1)

where ω(z) and ω 0 are the radii at z and z = 0 and λ is the wavelength.

Fig. 6
figure 6

544-nm beam radii along the horizontal axis (filled triangles) and along the vertical axis (filled circles) as a function of the location along the beam direction near the focus at the power of about 1.5 W (a); the M 2 factor with respect to the 544-nm beam power (b)

Full size image

Figure 6b shows the M 2 factors for the horizontal (M 2 x ) and vertical (M 2 y ) axes as a function of the second-harmonic beam power. The M 2 factors varied between 1.07 and 1.15 for the horizontal and vertical axes, respectively. Thus, the 544-nm beam is expected to be matched well to the enhancement cavity for the 272-nm beam generation. The beam was delivered to the enhancement cavity for the 272-nm beam after passing through the mode-matching lens set.

The enhancement factor of the cavity was approximately 50, and the power inside the cavity was estimated to be approximately 350 W using the produced UV power and input coupling efficiency. The beam waist of the 544-nm beam at the fourth-harmonic crystal was estimated to be approximately 22 μm from the resonant cavity geometry, and thus, the intensity at the focus was lower than the damage threshold (10–48 GW/cm2) of the BBO crystal [19, 20]. Figure 7 shows the UV power versus the 544-nm beam power. The frequency quadrupler produced approximately 1.5 W of the 272-nm beam at 7.3 W of the incident green beam, and thus, the conversion efficiency was approximately 20 %. The generated beam had an asymmetric shape due to a walk-off, and it was compensated by a cylindrical lens, as shown in the inset of Fig. 7. The M 2 factor of the 272-nm beam was measured to be 1.6(± 0.2).

Fig. 7
figure 7

272-nm beam power (solid circle) and the conversion efficiency (hollow triangle) versus the 544-nm beam power. The inset shows the beam profile of the 272-nm beam

Full size image

The cavity could also be stabilized over a span of several GHz at a 10-Hz repetition rate, and an example at an ultraviolet power of ~100 mW is shown in Fig. 8. In this experiment, the PZT of the ECDL was driven with a sawtooth signal from a function generator at 10 Hz. Figure 8a shows the photodiode signal of the ultraviolet beam, while the seed beam was scanned, and Fig. 8b shows the transmitted 544-nm beam through a free-running cavity with a free spectral range of approximately 460 MHz. Thus, the corresponding interval in the ultraviolet region was approximately 920 MHz. Figure 8c shows the transmitted infrared beam through a Fabry–Perot etalon with a free spectral range of approximately 500 MHz when the seed beam was scanned, and thus the interval between the transmittance peaks in the figure represents a 2-GHz interval in the ultraviolet region. T.P. denotes the turning point, and we observed more noise at the turning points, as shown in the figure. We are certain that the cavity locking during the seed beam frequency scanning is indirect proof of the tunability of the ultraviolet beam.

Fig. 8
figure 8

Ultraviolet beam intensity when the infrared beam was scanned at approximately 10 Hz for the locked cavity (a) and free-running cavity (b). The ultraviolet beam power was about 100 mW. The Fabry–Perot etalon signal of the fundamental beam was used to measure the frequency span of the ultraviolet beam. The free spectral range was 500 MHz in the infrared region, and the corresponding frequency is 2 GHz in the ultraviolet region. T.P. represents the turning point

Full size image

4 Applications to Ca spectroscopy

For the study of the isotope-selective and efficient optical pumping of the ground-state calcium, we considered the 4 s2 1S0–4s5p 1P1 transition of calcium for the optical pumping transition with a corresponding resonance wavelength of 272 nm. In the calcium isotope separation method, we provided a 272-nm beam to the calcium atomic beam to prepare the metastable calcium atoms in the 1D2 state by an optical pumping process [3]. Next, the metastable calcium atoms are ionized by stepwise excitations to the autoionization level using two broad-band pulsed lasers. In the previous work, we demonstrated the isotope-selective characteristics of the optical pumping process and introduced the technique of frequency stabilization of the optical pumping beam to the Ca-48 transition [21, 22]. In this demonstration, we used a strong 532-nm pulsed laser beam instead of two resonant pulsed lasers, and the generated ions were analyzed by a time-of-flight mass spectrometer. In this scheme, the atoms went through the optical pumping and the ionization regions sequentially, and the regions were spatially separated to avoid two-photon ionization involving the 532- and 272-nm beams. Therefore, the isotope selectivity could be largely obtained by the optical pumping process. When we tuned the laser wavelength to the calcium-48 resonance, we observed and compared the ion signals at the positions of each calcium isotope and obtained an isotope selectivity, \({\hbox{IS}}_{{\rm k}} = \left(\frac{X_{{\rm A}}^{{\rm f}}}{1-X_{{\rm A}}^{{\rm f}}}\right)/\left(\frac{X_{{\rm A}}^{{\rm i}}}{1 -X_{{\rm A}}^{{\rm i}}}\right)\), of more than 7000 (the measurement was limited by the noise and dynamic range of the detection system), where \(X_{{\rm A}}^{{\rm i}}\) and \(X_{{\rm A}}^{{\rm f}}\) denote the initial and final abundances of the target isotope A; however, a more accurate value could not be obtained because of the experimental limitations. But, we expect that a precise measurement could provide a higher isotope selectivity, as reported previously [23].

For efficient optical pumping of atoms in the gaseous state, we should consider the power-broadening effect to cover the Doppler-broadened atomic gases, while the isotope selectivity is maintained as required. The coverage by the power broadening depends on the gf value of the 4 s2 1S0–4s5p 1P1 transition; however, the gf values are scattered from 9.1 × 10−4 to 8 × 10−3, and this uncertainty hinders the estimation of an efficient laser power [24–26]. Therefore, we tried to directly measure the gf value by observing the power broadening. In the case of gf = 9.1 × 10−4, 1 W/cm2 at 272 nm induced power broadening of approximately 500 MHz. Thus, we will deform the spatial profile of the ultraviolet beam to be a top hat profile to induce uniform broadening at all spatial positions and determine the gf value of the 4 s2 1S0–4s5p 1P1 transition by measuring the broadened line.

5 Conclusion

We demonstrated and characterized a green beam with a power of approximately 6 W by using a 30-mm-long MgO-doped PPSLT crystal. The beam quality factor of the 544-nm beam was approximately 1.15 up to approximately 6 W. We provided the green beam to the enhancement cavity for the fourth-harmonic generation and achieved a power of approximately 1.5 W and an efficiency of approximately 20 %. The beam quality parameter was measured to be approximately 1.6(2). Therefore, the overall efficiency from the infrared laser was approximately 6 %. The ultraviolet beam power was improved eight times over that previously reported [8, 13].

Regarding the confocal parameter dependence of the 544-nm beam, we observed that the conversion power and efficiency were recovered by moving the focusing condition, and thus, they were insensitive to the confocal parameter for this crystal. In addition, we achieved a power of up to approximately 9 W at 544 nm after optimization, even though the beam has not yet been characterized. However, the beam shape did not appear to be distorted, and thus the 544-nm beam can be applied to the enhancement cavity to improve the fourth-harmonic generation power. To that end, we will characterize the green beam in the near future.

Regarding the frequency characteristics, we showed the scanning property with a scanning of the seed beam and observed that the ultraviolet beam can be scanned at 2–3 GHz at 10 Hz. The frequency span can be further expanded up to about 8 GHz if we scan the frequency more slowly. We expect that the scanning feature is sufficient to study the power-broadening and isotope-selective characteristics of the corresponding calcium transition. Finally, we discussed further work using a 272-nm beam. In the future, we will apply a watt-class 272-nm beam to an experiment for determination of the oscillator strength of the 4 s2 1S0–4s5p 1P1 transition.