We report an efficient 1.485 μm external cavity diamond Raman laser operating on the 2nd Stokes shift of a 1.064 μm Nd:YAG pump laser. 1.63 W pulsed at 5 kHz is produced with a quantum conversion efficiency of 71% and excellent beam quality. Numerical modelling confirms that optimal operation is achieved with low output coupling reflectivity.
©2011 Optical Society of America
Pulsed lasers in the ‘eye-safe’ wavelength region in the vicinity of 1.5 μm comprise an important technology in applications such as rangefinding and lidar. Wavelengths longer than 1.4 μm are strongly absorbed in the eye prior to focussing on the retina and thus much higher intensities can be safely used in scenarios where inadvertent exposure is considered likely. Established laser designs are mainly based on Er doped laser materials or optical parametric oscillators (OPOs), but these have limitations in terms of their performance and practicality. Er:glass lasers suffer from poor average power capability in bulk form or when in fibre form have peak power restrictions. In-band pumped Er crystalline lasers are a recent hybrid approach that can achieve high peak and average powers at 1.645 μm or 1.617 μm [1, 2] however, bright and narrowband pump sources are required for efficient operation and these can add to the system complexity and cost. Methods that can leverage mature Nd pump laser technology are perhaps more commercially attractive. Although OPOs use this approach, phasematching requirements (e.g. temperature control) and operation near damage thresholds for many crystals result in problems for creating highly stable and robust systems.
Crystalline Raman lasers, when operating in the external cavity configuration, provide a compact and efficient method of wavelength shifting Nd lasers to a longer wavelength. Narrower free running linewidth compared to OPOs, freedom from phasematching constraints, pulse length shortening and beam quality improvement through Raman beam cleanup are attractive features. The typical Stokes shift of most Raman crystals (800-1100 cm−1) provides two main options for converting to the eye-safe region: 3rd Stokes shift of the 4F3/2 → 4I11/2 Nd laser line (e.g., 1.064 µm in YAG) and 1st Stokes shift of the 4F3/2 → 4I13/2 lines (e.g., 1.338 µm and 1.319 µm in YAG). Quantum conversion efficiencies to the eye-safe region greater than 60% have been achieved for both these Nd transitions using Raman crystals such as Ba(NO3)2 [3, 4] and BaWO4 . Although pumping at 1.3 µm simplifies the design of the Raman conversion stage, several important advantages are obtained by using the higher gain 1.064 µm laser line. As well as being more widely available, the design of efficient pump lasers with short pulsed q-switched output, which is crucial for many eye-safe sensing applications, is typically simpler as a result of the larger emission cross section at 1.064 µm. Moreover, Raman shifting from 1.064 µm transfers a larger proportion of the total thermal load from the laser crystal into the Raman crystal (when considering the change in photon energy from the diode/flashlamp to eye-safe output). Although this is normally a disadvantage as most Raman crystals have poorer thermal properties than the pump laser material, it becomes an advantage when using diamond as a Raman material due to its excellent thermal properties.
Synthetic diamond of good optical quality and low absorption loss is increasingly available commercially and has been demonstrated to enable Raman conversion of q-switched pulses in the visible and near IR with high efficiencies. 1st Stokes diamond Raman lasers operating at 0.573 µm and 1.240 µm have been demonstrated with quantum conversion efficiencies from 0.532 µm and 1.064 µm of 68%  and 71% , respectively. Diamond has several properties of interest for substantially extending the range of Raman laser performance. The thermal conductivity of diamond is two to three orders of magnitude higher than other Raman and laser host crystals and combined with its low coefficient for thermal expansion it should allow the creation of small Raman lasers of high average power. Recently a record 24.5 W of 1st Stokes power from a crystalline Raman laser was achieved using an 8 mm diamond crystal without any sign of thermal effects . Its high damage threshold and high Raman gain coefficient allows low thresholds from compact (<1 cm) devices. Finally diamond’s long Raman shift of 1333 cm−1 compared to <1100 cm−1 for other commonly used Raman crystals allows the eye-safe region to be reached from 1.064 μm in 2 shifts instead of 3, which reduces the complexity of mirror coatings. To our knowledge diamond Raman lasers operating efficiently in the eye-safe region near 1.5µm have not been reported.
In this paper we report a detailed characterization of a 1.064 μm pumped 2nd Stokes diamond Raman laser generating output at 1.485 μm. We have implemented a numerical model for the Raman laser which allows for differing waist sizes of the fundamental and Stokes fields, in order to optimise design parameters for efficient operation.
A 6.9 mm long, low birefringence, Type IIa single diamond crystal grown by chemical vapour deposition (Element Six, UK) was used for the experiment. The rectangular crystal was cut for propagation along the <110> axis with an AR coating on the end faces (reflectivity per surface of <0.4% at 1.240 μm and ~2% for 1.064 μm and 1.485 μm.)
The external Raman cavity was pumped by a linearly polarised Q-switched Nd:YAG laser operating at 1.064 µm with 0.9 mJ pulses at a repetition rate of 5 kHz (4.5 W average power). Power incident on the Raman cavity was varied using a waveplate before an isolator to ensure consistent pulse shape and beam quality. A singlet lens was used to focus the pump beam in the cavity with a measured spot diameter of 140 μm. The pump polarisation was aligned with the <111> axis of the crystal to maximize the Raman gain .
The Raman laser cavity input coupler mirror was highly transmitting at 1.064 μm (97%T), and highly reflective (>99%R) at the 1st and 2nd Stokes wavelengths. Laser performance was investigated for 4 output couplers that each had high reflectivity at the 1st Stokes wavelength, and 2nd Stokes output coupling fractions in the range 39-92%T (see Table 1 ). The output couplers were also selected to have low reflectivity at the 3rd Stokes to suppress loss to higher Stokes orders. All mirrors were flat except for the input coupler and M3 which had radii of curvature of 100 mm and 300 mm, respectively. The cavity mirrors were placed as close as possible to the crystal (separation approximately 1 mm) to achieve minimum laser threshold.
Power measurements were made using a Newport 818P-010-12 head and temporal traces were recorded with fast InGaAs detectors (100 ps rise time) and a 1 GHz oscilloscope. Longpass and shortpass filters were used to remove the residual pump and 1st Stokes wavelengths where required. The spot sizes of the pump and Stokes wavelengths within the cavity were measured by imaging the beams in the Raman cavity onto a camera with appropriate filters to isolate the desired wavelength.
The Raman laser model (Eq. (1)) is based upon the established coupled intensity equations used previously by several authors, e.g [5, 9, 10], where we have made the beam spot size in the crystal explicit in order to enable us to take into consideration the different spot sizes of the oscillating fields.Eq. (2). The approximation is made that the spot size for a given wavelength is constant along the length of the cavity as the cavity length is much shorter than the confocal parameter of the beams.
The Raman gain coefficient used was estimated to be 16.6 cm/GW for the <111> orientation at 1.064 μm as discussed previously in . The absorption coefficient within the crystal was measured using laser calorimetric tests to be less than 0.003 cm−1 for the pump and first two Stokes wavelengths. A spontaneous scattering factor of 10−13 cm−1 was estimated from , taking into account the longer wavelength, single polarisation and assuming a cone angle of 10−6 Sr. Also included in the model were diamond coating losses of 2%, 0.4% and 2% for the pump, 1st and 2nd Stokes wavelengths, respectively.
When calculating the beam areas, intracavity spot diameters of 140 μm, 230 μm and 120 μm were measured for the pump, 1st and 2nd Stokes wavelengths, respectively. The values were consistent ( ± 5%) across the different output couplers used and at different power levels (in particular changing from a plano to a 300 mm ROC output coupler was observed to have a minimal impact on the spot sizes). The pump beam under-fills the cavity mode and as such the generated 1st Stokes expands to fill the cavity mode as it resonates within the high Q cavity. In contrast the high output coupling fraction for the 2nd Stokes results in the observed waist existing within the high pump intensity region on axis. Using the model, we found that the threshold was most sensitive to the chosen pump beam spot size, whereas the slope efficiency was more influenced by the sizes of the Stokes beams.
The modelled pump pulse was approximated as Gaussian with a 0.7 GHz ripple to simulate the observed mode beating within the pump pulse. The relationship between the pump and Stokes powers was solved numerically by dividing the Raman cavity into 100 sections. The calculated Stokes pulse traces contained high frequency cavity round trip oscillations that were largely removed when passed through a low pass filter designed to simulate the response of the InGaAs detectors and oscilloscope.
3. Results and discussion
The highest 2nd Stokes average power output was obtained using output coupler M3 (R = 16% at 1.485 μm). 1.63 W was achieved at 3.2 W of pump power incident on the crystal, which was the maximum available from the pump laser. The output power increased linearly with pump power above the 0.25 W threshold as shown in Fig. 2a . The slope efficiency was 56% and at maximum pump power the quantum conversion efficiency was 71% (power conversion efficiency of 51%). Due to the low reflectivity of the cavity mirrors at 1.852 μm no significant 3rd Stokes power (< 1 mW) was measured. Also in Fig. 2a the modelled output power and conversion efficiency show good agreement with the experimental results and the improvement in agreement when considering the actual waist sizes is readily seen. The laser demonstrated excellent beam quality with a measured M2 < 1.1 as shown in Fig. 2b. In this case the pump laser exhibited good beam quality (M2 < 1.5) but Raman beam cleanup has been demonstrated to enhance the brightness of a low quality pump source .
The measured and modelled pulse shapes for 0.4 mJ input pump are shown in Figs. 3a and 3b and there is strong agreement in the key features of the pulse traces. Once above threshold the 2nd Stokes pulse largely follows the profile of the pump pulse. In contrast, the 1st Stokes pulse is relatively flat, after an initial spike to reach the 2nd Stokes threshold, with any additional power being converted to the 2nd Stokes. The presence of significant residual pump power during the peak of the input pulse (at ~10 ns) is a notable difference between the pulse shapes seen here and those for the more efficient 1st Stokes lasers presented in [6, 7]. This largely accounts for the shortfall of the measured slope efficiency below the quantum limit.
The output coupling value at the 2nd Stokes is an important design parameter for determining overall conversion efficiency. As a function of output coupler reflectivity in the range 16% to 61%, the laser threshold varied only 8% and the slope efficiency provided a simple indicator of the output coupler performance (note that a notably higher threshold is obtained using mirror M4 due to its lower reflectivity of the pump beam). The measured slope efficiency values for each output coupler are plotted in Fig. 4 along with a modelled curve for the reflectivity range of 1% to 70%. Both experimental and theoretical slope efficiencies show a trend favouring lower reflectivity values. The model peaks at a slope efficiency of 56% for reflectivity values around 8-16%.
The improved 2nd Stokes performance for low reflectivity output couplers has been theoretically explored previously  and can be inferred from the coupled Stokes equations. A balance needs to be achieved between maintaining sufficient 1st Stokes intensity to effectively deplete the pump while maintaining good power transfer from the 1st to 2nd Stokes. Increasing the 2nd Stokes output coupler reflectivity beyond the ideal value leads to increased 2nd Stokes intracavity intensity which will more thoroughly deplete the 1st Stokes pulse, as shown in Fig. 5b . The lower 1st Stokes field in turn results in reduced pump depletion as shown in Fig. 5a. For reflectivities less than the optimum value, the modelling shows that there is still an improvement in pump depletion but this is more than offset by reduced conversion from the 1st Stokes to the 2nd Stokes.
The 71% quantum conversion efficiency achieved here compares favourably with other eye-safe crystalline Raman and OPO conversion results. The highest previously reported quantum conversion efficiencies are 63% from 1.064 μm using two 70 mm Ba(NO3)2 crystals  and 68% from 1.338 μm using a 40 mm BaWO4 crystal . Similar conversion results have been obtained in efficient extracavity OPOs, for example 71% quantum conversion efficiency from 1.064 μm using a 20 mm KTP crystal . The main advantage of diamond, however, comes from its exceptional thermal properties. As expected, given the moderate pump powers used, no evidence was seen of any rollover in power output or beam quality degradation due to thermal effects within the crystal. Diamond should be able to sustain high conversion efficiencies at much higher average power outputs and in smaller crystals. In contrast the poor thermal properties of Ba(NO3)2 lead to strong thermal lensing for average powers beyond Watt level resulting in reduced beam quality  and optical conversion efficiency [15, 16]. Optical compensation can be applied to counter the effect of a thermal lens but this confines stable operation to a narrow range of operating parameters. BaWO4 has been shown to have an order of magnitude weaker thermal lens than Ba(NO3)2 under similar conditions  and based on the comparison of thermal material parameters such as thermal conductivity, thermal expansion coefficient and thermo-optic coefficient, we would expect the onset of the thermal effects in diamond to be 2-3 orders of magnitude higher again. Diamond also presents a thermal advantage over KTP OPOs since parasitic absorption at the long idler wavelengths in KTP can be a design consideration in these systems.
Although wavelengths longer than 1.4 μm are considered ‘retina safe’ the wavelength of 1.485 μm reported here is outside the 1.5 μm-1.6 μm band commonly targeted by eye-safe lasers. 1.485 μm is on the shoulder of the atmospheric transmission window and modelling using the MODTRAN software package indicates that the atmospheric transmission at 1.485 ± 0.001 μm varies between 10% and 75% per km compared to 96% per km around 1.55 μm (horizontal path at 10m height, rural aerosol model). For applications requiring longer range atmospheric propagation it may be an advantage to use pump lasers employing Nd:YAP, which has a dominant line at 1.080 μm and thus a 2nd Stokes wavelength of 1.517 μm (T~90% per km).
We have demonstrated efficient conversion of a 1.064 μm laser to eye-safe wavelengths using a diamond Raman laser. The 71% quantum conversion efficiency to the 2nd Stokes at 1.485 μm exceeds conversion efficiencies to eye-safe wavelengths demonstrated with alternate Raman crystals and matches efficiencies of OPO based systems. Analysis shows that high output couplings are needed to optimise conversion to the 2nd Stokes. The unmatched thermal properties of diamond provides excellent scope for further power scaling or for the use of small crystals with minimal thermal issues in microchip style lasers.
R.P. Mildren is grateful to sponsorship by the Australian Research Council Future Fellowship Scheme (FT0990622). A. Sabella wishes to thank J. Haub and O. Samardzic for their thoughtful input into the preparation of this manuscript.
References and links
1. J. W. Kim, D. Y. Shen, J. K. Sahu, and W. A. Clarkson, “Fiber-Laser-Pumped Er:YAG lasers,” IEEE J. Sel. Top. Quant. 15(2), 361–371 (2009). [CrossRef]
2. M. Eichhorn, “High-power resonantly diode-pumped CW Er3+:YAG laser,” Appl. Phys. B 93(4), 773–778 (2008). [CrossRef]
3. P. G. Zverev, T. T. Basiev, and A. M. Prokhorov, “Stimulated Raman scattering of laser radiation in Raman crystals,” Opt. Mater. 11(4), 335–352 (1999). [CrossRef]
4. V. A. Lisinetskii, A. S. Grabtchikov, I. A. Khodasevich, H. J. Eichler, and V. A. Orlovich, “Efficient high energy 1st, 2nd or 3rd Stokes Raman generation in IR region,” Opt. Commun. 272(2), 509–513 (2007). [CrossRef]
5. T. T. Basiev, M. N. Basieva, A. V. Gavrilov, M. N. Ershkov, L. I. Ivleva, V. V. Osiko, S. N. Smetanin, and A. V. Fedin, “Efficient conversion of Nd: YAG laser radiation to the eye-safe spectral region by stimulated Raman scattering in BaWO4 crystal,” Quantum Electron. 40(8), 710–715 (2010). [CrossRef]
9. S. Ding, X. Zhang, Q. Wang, P. Jia, C. Zhang, and B. Liu, “Numerical optimization of the extracavity Raman laser with barium nitrate crystal,” Opt. Commun. 267(2), 480–486 (2006). [CrossRef]
10. Y. R. Shen and N. Bloembergen, “Theory of Stimulated Brillouin and Raman Scattering,” Phys. Rev. 137(6A), A1787–A1805 (1965). [CrossRef]
11. A. M. Zaitsev, Optical Properties of Diamond (Springer-Verlag, Berlin, 2001), Chap. 4.
12. J. T. Murray, W. L. Austin, and R. C. Powell, “Intracavity Raman conversion and Raman beam cleanup,” Opt. Mater. 11(4), 353–371 (1999). [CrossRef]
13. L. R. Marshall and A. Kaz, “Eye-safe output from noncritically phase-matched parametric oscillators,” J. Opt. Soc. Am. B 10(9), 1730–1736 (1993). [CrossRef]
15. V. A. Lisinetskii, T. Riesbeck, H. Rhee, H. J. Eichler, and V. A. Orlovich, “High average power generation in barium nitrate Raman laser,” Appl. Phys. B 99(1-2), 127–134 (2010). [CrossRef]
16. N. Takei, S. Suzuki, and F. Kannari, “20-Hz operation of an eye-safe cascade Raman laser with a Ba(NO3)2 crystal,” Appl. Phys. B 74(6), 521–527 (2002). [CrossRef]
17. T. T. Basiev, A. V. Gavrilov, V. V. Osiko, S. N. Smetanin, and A. V. Fedin, “High-average-power SRS conversion of radiation in a BaWO4 crystal,” Quantum Electron. 34(7), 649–651 (2004). [CrossRef]