1. Introduction

Recent progresses in high power alkali vapor lasers may open a new door to generate new laser wavelengths via the nonlinear frequency conversion. Such diode-pumped alkali vapor lasers have been recently demonstrated with very high output power, system compactness, excellent beam quality and very narrow emission linewidth at a handful of new wavelengths from infrared (IR) to the near IR range [1]—all important features that are required to realize an efficient frequency conversion. For example, the output power of an alkali laser can be easily scaled to tens to hundreds of watts in the continuous wave (CW) mode. The emitting wavelengths of Cs, Rb and K vapor lasers are 894.6 nm, 795.0 nm, and 770.1 nm, respectively, which correspond to blue to ultraviolet (UV) wavelengths of 447.3 nm, 397.5 nm and 385.0 nm, respectively, if using the second harmonic generation (SHG) scheme inside a high quality nonlinear optical crystal. A compact laser with such short wavelengths and with an output power level to a few tens of watts will be highly desirable for applications such as data recording, high definition laser projection, underwater communication, etc. [2] Furthermore, a possible extension of new laser wavelengths into the deep UV band [3] will be very important to many other potential applications including spectroscopic analysis, processing, biophysics, etc. For example, in some specific UV spectroscopic analysis an extremely narrow linewidth and high enough output power are both highly required, and their availability can hardly be compensated by using common lasing approaches including the multiple quantum well (MQW) technology fabricated in wide band gap semiconductors.

In frequency conversion, the use of quasi-phase matching (QPM) offers many unique features that are highly required for efficiently generating such short wavelengths [4, 5]. For example, the QPM SHG process between the interacting fundamental and harmonic waves can be realized by using a periodically poled (PP) ferroelectric domain structure [6] in which each domain section has a thickness equal to the corresponding coherence length (L_c). QPM technique provides a collinear interaction among all interacting waves, which diminishes the ‘walk-off’ effect commonly happening to traditional critical phase-matching (CPM) methods. The process also offers a wide achievable phase matching wavelength range and the ability to utilize the largest component in the optical material’s nonlinear coefficient tensor. Therefore availability of a good nonlinear PP crystal used in the blue to UV wavelength range will be very crucial to further enhance the conversion efficiency. Factors to evaluate such PP crystals are both intrinsic (a crystal’s transparency, nonlinearity, and laser damage threshold, etc.) and extrinsic (qualities related to those artificially poled domain structures).

Extrinsic factors include the poled domain structure’s period fluctuation, imbalanced duty cycle, incomplete poling across the crystal wafer’s thickness, domain wall scattering, etc [7]. Such extrinsic imperfectness related to electrical poling usually has a large impact on a PP crystal’s wavelength acceptance bandwidth, effective crystal length, and effective nonlinear coefficient, etc., which will reduce the frequency conversion efficiency. Normally, a real PP crystal can have a wavelength acceptance bandwidth as broad as 1 nm (case-dependent), and this is much broader than an atomic vapor laser’s emission linewidth (normally <0.1 Å). This bandwidth mismatch may significantly affect the conversion process, especially at high power levels where thermal effect via material absorption starting to take effect. Actually, very narrow linewidth of an alkali vapor laser’s emission can be an effective means to probe a PP crystal’s instant thermal response and to evaluate such PP crystals’ quality related to electrical poling. The latter can be effective for realizing quality assurance in the PP crystal production. In this research, a Cs laser operating at 894 nm was used to generate the blue 447 nm light via the second harmonic generation in a commercial periodically poled KTP (PPKTP). The crystal’s sensitivity over the light-introduced thermal effect was experimentally monitored by recording the temporal profile of both fundamental and harmonic pulses, and the observed slight deficiency in SHG efficiency was explained by possible bandwidth mismatch between the laser and the PPKTP crystal. In this experiment, PPKTP crystal was selected because of its large optical nonlinearity and good transparency in short wavelengths, high material damage threshold, the maturity in electrical poling for realizing high quality domain structures, and its easy availability from commercial vendors.

2. Experiment

The experimental setup for the measurement is presented in Fig. 1. The fundamental laser beam was generated by a high power diode laser array pumped Cs vapor laser operating at the near infrared wavelength of 894 nm and in the continuous wave (CW) mode [1]. The laser beam was first chopped and then focused into the 1×2 mm² aperture of a 30 mm long PPKTP crystal with a 100 mm focal length lens (L). Both pulse duration and pulse repetition rate can be adjusted by the computer-controlled mechanical chopper. The PPKTP crystal (Raicol Crystals Ltd.) was placed inside a temperature-controlled oven in order to keep the crystal at the best suitable temperature corresponding to the exact QPM condition when measuring the conversion efficiency. In order to separate the generated second harmonic wave from the residual fundamental laser radiation, a set of two dichroic mirrors were used. Each mirror had 99.8% reflectivity at 447 nm transmitted 99% of the 894 nm pump beam. The data were finally corrected according to the dichroic mirrors’ reflectivity and transmittance.

 

Fig. 1. Experimental setup for the second harmonic generation measurement.

Download Full Size | PDF

3. Results and Discussion

KTP crystal is a positive-biaxial crystal having a point group of 2 mm. The principle values of refractive indices n_x, n_y, and n_z of the crystal are functions of temperature and wavelength. The temperature derivatives of the refractive indices in KTP can be represented as [8]:

 

Fig. 2. The emission spectrum of the high power Cs laser obtained with an autocorrelation interferometer. The measured emission linewidth is about 6 pm.

Download Full Size | PDF

Assuming that without affection of those extrinsic factors discussed before (assuming no period fluctuation, period duty cycle is equal to 1, optically very sharp domain wall, etc.), the wavelength acceptance bandwidth of a PP KTP crystal will be simply a function of the crystal length, wavelength and the material’s dispersion, and can be predicted by the commonly used theoretical equation under the exact QPM condition [9]:

where Δλ is the full width at the half maxima (FWHM) from the generated harmonic power vs the fundamental wavelength curve (wavelength acceptance bandwidth), ‘L’ the crystal length, λ_ω the wavelength in vacuum of the fundamental light, n_ω and n_2ω the fundamental and SH indices of refraction of the crystal. The calculated wavelength acceptance for the used 30 mm long PPKTP crystal poled with a domain period 4.975 microns is about 80 pm. Figure 2 shows that the measured emission linewidth of the used Cs laser operating at 894 nm was about 6 pm that is nearly 13 times narrower than the crystal’s wavelength acceptance width at the ideal situation. In an actual PPKTP crystal, however, its wavelength acceptance bandwidth could be much broader because of the existence of possible extrinsic factors such as period fluctuation, imbalanced duty cycle, and domain wall fluctuation, etc.

 

Fig. 3. Dependence of the second harmonic on the fundamental power for pulse durations of 50 and 500 µs, respectively. The repetition rate is kept 150 Hz for both measurements.

Download Full Size | PDF

Figure 3 shows the measured second harmonic power vs the fundamental power under laser pulse durations of 50 and 500 microseconds, respectively. In both cases, the chopper operates at the same repetition rate of 150 Hz. At low fundamental power, the harmonic power grows proportional to the fundamental power squared for both pulse durations. As the fundamental power increases, this dependence changes—it deviates from the ideal quadratic dependence (the dotted line) and power deviation is becoming higher for the longer pulse duration. In general, from Fig. 3 one can see that: 1) the normalized SH conversion efficiency for the actual PPKTP crystal is about η ~4.4 %/W (at low fundamental power, when thermal effects are negligible), which is close to the theoretical normalized SHG efficiency of η ~4.6 %/W, when only considering using a single-frequency laser, using a perfect PPKTP crystal for the perfect QPM condition, and an assumed optical nonlinearity of ~15.8 pm/V for the KTP under the laser wavelength of 894 nm; and 2) the deviation of the measured SHG vs fundamental curves from their corresponding exponential fit is dependent on pulse duration, which indicates a possible existence of thermal effect that will be separately discussed in the session below. The cause for such a high SHG efficiency that is very close to the theoretical case can be actually from the similarity of the comparison made above: the theory assumes the use of a single frequency laser, and the actually used Cs laser indeed has a very narrow emission linewidth. The potential to maximize the efficiency can be made from optimizing the matching between the crystal’s wavelength acceptance and the fundamental laser’s emission linewdith. Precise and thorough theoretical expression on this optimization will be complicate, due to the involvement of multiple interacting wavelengths which will further result in many nonlinear processes including SHG, sum-frequency generation (SFG), etc, and involvement of many intrinsic and extrinsic factors that actually affect a PP crystal’s wavelength acceptance. Further following this theory, however, monitoring the SHG using this ‘single frequency-like’ Cs laser can be an actual quality assurance tool during the fabrication of such PP crystals. In order to optimize the SHG efficiency, an approach to optimize the bandwidth matching between the laser emission and the crystal acceptance was suggested [10], and in this paper we will not discuss this issue further.

An emerging issue in this research is the thermal effect, which is becoming more serious when increasing the operation power, and is becoming especially obvious (starting to occur at much lower fundamental powers) at the case when the incident laser’s emission linewidth is much narrower than the used crystal’s acceptance. According to above Fig. 3, the observed deviation to the quadratic fit increases when increasing the laser’s pulse duration. This result indicates a strong response from a possible thermal effect, and this response is much more sensitive than the case when we use a similar PPKTP crystal to frequency-double a high power optical fiber laser having a much broader emission linewidth [11]. In that case the thermal effect is only becoming obvious when the fundamental power exceeds ~3 W (for the 1064 nm light), and here in this case it starts to take effect at much lower power of ~1 W (see Fig. 3).

Figure 4 shows the comparison among those recorded pulse shapes for both fundamental and harmonic waves, under two different pulse durations: 50 µs and 500 µs, respectively, and the measurements were kept at the same repetition rate of 150 Hz and similar power ~1.5W. When using the short pulse duration, both fundamental and harmonic pulses are very similar in shape, which indicates synchronized responses between them with a negligible affection from thermal effect. However, when increasing the pulse duration, thermal effect takes effect significantly due to a longer irradiation inside the pulse duration time domain, and which depletes the SHG energy when the pulse evolutes over the time and over the crystal length.

 

Fig. 4. Recorded pulse shapes for both fundamental and harmonic radiations having the 50 µs (a) and 500 µs (b) pulse duration, respectively.

Download Full Size | PDF

When reducing the fundamental pulse’s repetition rate but keeping at the similar input powers, on the other hands, the tendency of a slight reducing of the second harmonic pulse’s energy depletion can be observed (Fig. 5). In Fig. 5 the comparison was given to two cases having repetition rates of 150 and 4 Hz, respectively. This result further indicates the existence of a possible thermal effect, even though the observed difference between the two cases is not large. One possible reason for this is that the internal thermal dissipation rate is still faster than the thermal building-up rate when a series of fundamental laser pulses pass, even at the high repetition rate of 150 Hz that we used in this experiment.

 

Fig. 5. Recorded pulse shapes for harmonic pulses having the 500 µs pulse duration and similar power ~1.5W, but different repetition rates of 150 (left) and 4 Hz (right), respectively.

Download Full Size | PDF

The power depletion inside a harmonic pulse can be explained by thermal detuning of the PPKTP crystal from the exact QPM condition [12], and the observed high thermal sensitivity (much lower fundamental power required for the thermal effect to occur, when comparing to the case of using high power fiber lasers) is arising from the narrow Cs vapor laser’s emission linewidth that largely mismatches in shape to the 30 mm long PPKTP crystal’s wavelength acceptance. Dependence of the exact QPM condition on temperature in a PPKTP crystal can be determined by its temperature tuning curve, defined by the following equation [9]:

where ΔT is the FWHM of the temperature tuning curve, and β is the KTP crystal’s thermal expansion coefficient. The phase matching temperature tuning curve of the used PPKTP crystal was determined experimentally (see Fig. 6) and the exact phase matching appeared to be at 38.65 °C. Observed ΔT for this crystal is then determined to be 1.2 °C, which is comparable to or slightly less than the simulated value of ~1.45 °C according to Eq. (5) for the 894.6 nm fundamental wavelength. This small difference can be explained case-dependently, according to the used crystal’s growth history, actual composition variation, heat treatment and electrical field treatment history, etc.

The result in Fig. 6 indicates that, in order to fully offset from the exact QPM condition (the second harmonic power becomes zero), the local crystal temperature must be able to change ~1.2 °C. A long laser pulse will be much easier to cause such a small temperature fluctuation locally along the laser pulse’s path due to material absorption (especially for generated blue 447.3 nm light, which is close to the material’s absorption edge), without affecting the overall crystal’s temperature. The latter was monitored by attaching a thermocouple to the crystal’s surface and thermally controlled during the experiments. This local temperature variation changes the exact QPM condition, shifts the peak of the SH power vs the wavelength curve, and then offsets this shifted peak from the laser’s emission wavelength which is fixed during the tests. Since that the laser’s emission linewidth is much narrower than the crystal wavelength acceptance, this offset will have much obvious affection on the SHG pulse’s power than that from the case using broad emission optical fiber lasers. Therefore, this local thermal detuning will clearly explain the observed power depletion inside those long harmonic pulses shown in above Fig. 4. If thinking about this technology differently, the observed pulse depletion in harmonic pulses could be used in other technology areas when precise and remote temperature sensing is required.

 

Fig. 6. Measured temperature dependence of SH power in a 30 mm long PPKTP.

Download Full Size | PDF

4. Conclusion

Very efficient blue 447.3 nm light generation was demonstrated using a PPKTP to frequency double a high power CW Cs vapor laser having a very narrow emission linewidth. Power depletion in long harmonic pulses was observed and was explained by local thermal detuning of the crystal away from the exact QPM condition. Band matching approach was suggested to further optimize the conversion efficiency and potential applications of the experimental concept into other technology areas such as sensing and electrical poling quality assurance were also discussed.