In recent years, Er:YAG lasers have been developed for eye-safe emission at $1645\mathrm{nm}$ for both cw and Q-switched applications, including remote sensing, range finding, spectroscopy, lidar, and wavelength conversion [1, 2, 3]. Certain applications, such as coherent lidar, require narrowband operation at modest to high power levels, which necessitates seeding with a stable, single-frequency source. The nonplanar ring oscillator (NPRO) has these qualities and has been demonstrated to operate at wavelengths near $1\mathrm{\mu m}$ (Nd:YAG) [4, 5], $2.01\mathrm{\mu m}$ (Tm:YAG) [6], and $2.09\mathrm{\mu m}$ (Ho:YAG) [7]. In this Letter, we report, to our knowledge, the first demonstration of an Er:YAG NPRO at $1645\mathrm{nm}$. The device is pumped with an Er fiber laser, as opposed to laser diodes, which enables optimum matching of the pump and resonator modes. This reduces spatial hole-burning and yields a more stable single-mode device without using an undoped section of material.

The design of our Er:YAG NPRO follows that described in [8], which relaxes the alignment tolerance yet still enables sufficient differential loss to be achieved between counter propagating waves in the presence of a magnetic field (Verdet effect) for unidirectional operation. A schematic of the crystal with the dimensions $14\mathrm{mm}(\text{width})\times 3\mathrm{mm}(\text{height})\times 13\mathrm{mm}(\text{length})$ is shown in Fig. 1. The crystal material is $2\text{\hspace{0.17em} at.}\%$ doping Er:YAG. As depicted in the figure, the optical path ABCD forms a nonplanar ring with an approximately $30\mathrm{mm}$ round trip.

The pump laser beam is incident at point A with a $45°$ angle of incidence. The ring laser output also exits at point A at an angle of $45°$ as shown. The relative tilt angles of the surfaces containing points A, B, C, and D are such that a $45°$ angle is achieved between the two optical planes defined by points A, B, and D and B, C, and D. Using $45°$ achieves a good balance between relaxing the alignment sensitivity and maximizing the differential loss, which minimizes the magnetic field required for unidirectional operation. The facets containing B, C, and D are uncoated and polished flat, allowing for total internal reflection, while the surface containing A, designated the input/output coupler (IC/OC), is flat for one crystal and $5\mathrm{cm}$ concave (internally) for a second. The IC/OC of both crystals was coated for high transmission of the $1532\mathrm{nm}$ p-polarized pump and a reflectivity (R) of 99% for the s-polarized $1645\mathrm{nm}$ laser output. The high R at $1645\mathrm{nm}$ was chosen to ensure a low threshold and narrow linewidth, rather than to optimize power and efficiency. Additionally, the IC/OC coating has a higher loss ($R<90\%$) for p-polarized light at $1645\mathrm{nm}$ to ensure a linearly polarized output.

The NPRO crystals were held in a copper mount that was thermoelectrically cooled to $15°\mathrm{C}$ during testing. In order to achieve the single directional oscillation, a permanent magnetic field, H, was applied to the NPRO in the direction shown in Fig. 1. NdFeB magnets (grade N42) with a permanent field of $0.4\mathrm{T}$ were stacked as needed. (The Verdet constant for the YAG crystal at $1645\mathrm{nm}$ is calculated to be $\sim 6.9\times {10}^{-5}\mathrm{rad}/\mathrm{mm}\text{-}\mathrm{T}$ [9], which is about 2.6 times smaller than at $1\mathrm{\mu m}$. ) As depicted in Fig. 2, the laser crystal was pumped with an IPG cw linearly polarized Er fiber laser at $1532.4\mathrm{nm}$, with a spectral linewidth of $0.2\mathrm{nm}$ (FWHM). The pump output was oriented to be p-polarized and mode-matched by a lens L1 into the NPRO crystal at $45°$. An optical isolator (OI) was used to prevent laser feedback into the pump. Laser output at $1645\mathrm{nm}$ was collimated by a lens L2 and separated from the reflected pump with a dichroic mirror (DM). An Ando AQ6319 optical spectral analyzer (OSA) with a $0.01\mathrm{nm}$ ($1.1\mathrm{GHz}$) resolution was used to confirm single longitudinal mode operation but was inadequate for determining the actual linewidth. Quantitative measurements were performed using a delayed self-heterodyne method [10], as described below.

Overall, the performance of the NPRO with the flat OC was slightly better than that with the curved OC, so only results for that former crystal will be presented. It was observed that a single $0.4\mathrm{T}$ magnet was sufficient to yield a stable unidirectional single frequency up to $0.1\mathrm{W}$, whereas four magnets were required for unidirectional operation at $0.5\mathrm{W}$. We further observed that stable single-frequency operation also necessitated that the pump mode diameter be smaller than $170\mathrm{\mu m}$. Optimum performance was achieved when the pump diameter was set to $120\mathrm{\mu m}$, yielding an optical-to-optical efficiency of 11% at $0.45\mathrm{W}$ of output. The performance of the crystal with this pumping spot is shown in Fig. 3. Figure 4 shows a comparison of the NPRO’s spectra for single-mode (blue solid curve) and multimode (red dashed curve) operation.

The $0.05\mathrm{nm}$ mode separation measured with the OSA agreed with the predicted value for the $30\mathrm{mm}$ round trip ring cavity. Fine tuning of the Er:YAG NPRO was achieved by changing the crystal temperature. When the crystal temperature was changed from $15°\mathrm{C}$ to $13°\mathrm{C}$, the laser wavelength was continuously tuned without mode hopping from 1645.258 to $1645.214\mathrm{nm}$ ($\sim 5\mathrm{GHz}$), as shown in Fig. 5. The wavelength tuning rate is $0.022\mathrm{nm}/°\mathrm{C}$ ($-2.5\mathrm{GHz}/°\mathrm{C}$). The NPRO output power increased linearly from $189\mathrm{mW}$ at $15°\mathrm{C}$ to $198\mathrm{mW}$ at $13°\mathrm{C}$.

The NPRO output beam quality was measured as a function of power with a Shack–Hartmann wavefront sensor manufactured by FLIR Systems. Figure 6 shows that beam quality $M^2$ was less than 1.2 over the entire range. An image of the NPRO far-field beam profile is shown as an inset of the Fig. 6.

As mentioned above, we used a delayed self- heterodyne method to measure NPRO output linewidth. The setup is shown in Fig. 7. The laser output was split by a $90/10$ beam splitter (BS). The reflected beam was coupled into a $25\mathrm{km}$ fiber for a time delay ${\tau }_d$ of about $125\mathrm{\mu s}$, which is sufficiently longer than the expected laser coherence time. The transmitted beam was frequency shifted by an acousto-optic modulator (AOM) operating at $27.13\mathrm{MHz}$. The two beams were recombined by a fiber coupler. The beat signal was detected by an amplified InGaAs photodetector and displayed on an RF spec trum analyzer. The resolution of the delayed self- heterodyne method can be estimated from the expression $[-0.4\log (\mathrm{\Delta }{f}_B{\tau }_d)+0.6]/{\tau }_d$, where $\mathrm{\Delta }{f}_B$ is the spectral analyzer bandwidth [9]. For our setup, ${\tau }_d=125\mathrm{\mu s}$ and $\mathrm{\Delta }{f}_B=0.5\mathrm{kHz}$, which yields a resolution of $8.6\mathrm{kHz}$.

An RF spectrum of the self-heterodyne signal is shown in Fig. 8 at the NPRO output power $100\mathrm{mW}$ over a $\mathrm{10}\mathrm{s}$ sweep time. This spectrum shows a central peak and relaxation oscillation sidebands at $\pm 146\mathrm{kHz}$. The amplitudes of both sidebands are $25\mathrm{dB}$ below that of the central feature. Relaxation oscillations are common in most solid-state lasers, occurring when the upper laser level lifetime is much longer than the cavity decay time. The central peak of $-3\mathrm{dB}$ bandwidth in Fig. 8 was measured to be $30\mathrm{kHz}$, which corresponds to a deconvolved FWHM linewidth of $21\mathrm{kHz}$, assuming a Gaussian lineshape. We also observed that the laser linewidth and relaxation oscillation frequency increased to $26\mathrm{kHz}$ and $173\mathrm{kHz}$, respectively, at a higher laser power level of $150\mathrm{mW}$. At present we are unable to perform the measurements on a shorter time scale, which could reveal an even narrower instantaneous linewidth.

In summary, we have demonstrated for the first time, to our knowledge, a $0.5\mathrm{W}$ single-frequency Er:YAG NPRO operating at $1645\mathrm{nm}$ that is resonantly pumped by a Er fiber laser at $1532.4\mathrm{nm}$. The device was tunable over a $5\mathrm{GHz}$ range without mode hopping and had a bandwidth of $21\mathrm{kHz}$ (over a $10\mathrm{s}$ interval). The wavelength tuning rate of $0.022\mathrm{nm}/°\mathrm{C}$ ($-2.5\mathrm{GHz}/°\mathrm{C}$) is obtained from the experiment. The beam quality of the device was observed to be near-diffraction-limited. As such, this device is a viable source for injection seeding, which is needed to obtain high-power narrowband eye-safe emission for various lidar and other remote sensing applications.

This work was supported under the Aerospace Corporation’s Mission Oriented Investigation and Experimentation program, funded by the U.S. Air Force Space and Missile Systems Center under Contract No. FA8802-04-C-0001.

 

Fig. 1 Schematic of Er:YAG NPRO crystal.

Download Full Size | PDF

 

Fig. 2 Experimental layout for the NPRO pumped by an Er fiber laser.

Download Full Size | PDF

 

Fig. 3 NPRO single-frequency power performance data comparison: pump beam diameter $120\mathrm{\mu m}$ (triangles with blue solid curve) versus $140\mathrm{\mu m}$ (squares with red dotted curve). Smaller pump beam diameter yields more efficient output and stable single-frequency operation.

Download Full Size | PDF

 

Fig. 4 NPRO output mode structures measured by Ando OSA: red dotted curve (multimode) versus blue solid curve (single mode).

Download Full Size | PDF

 

Fig. 5 NPRO single-mode cw tuning as a function of the crystal temperature. The tuning rate of $0.022\mathrm{nm}/°\mathrm{C}$ is measured.

Download Full Size | PDF

 

Fig. 6 NPRO output beam quality $M^2$ data versus output power. Inset: image of output far-field beam profile.

Download Full Size | PDF

 

Fig. 7 Experimental setup for NPRO delayed self-heterodyne linewidth measurement.

Download Full Size | PDF

 

Fig. 8 Delayed self-heterodyne signal recorded by an RF spectral analyzer at $100\mathrm{mW}$. NPRO single-frequency operation. Full width at $-3\mathrm{dB}$ was measured to be $30\mathrm{kHz}$.

Download Full Size | PDF

1. Y. Young, S. Setzler, K. Snell, P. Budni, T. Pollak, and E. Chicklis, Opt. Lett. 29, 1075 (2004). [CrossRef]   [PubMed]  

2. K. Spariosu, V. Leyva, R. Reeder, and M. Klotz, IEEE J. Quantum Electron. 42, 182 (2006). [CrossRef]  

3. D. Shen, J. Sahu, and W. Clarkson, Opt. Lett. 31, 754 (2006). [CrossRef]   [PubMed]  

4. T. J. Kane and R. L. Byer, Opt. Lett. 10, 65 (1985). [CrossRef]   [PubMed]  

5. E. J. Zang, J. P. Cao, Y. Li, T. Yang, and D. M. Hong, Opt. Lett. 32, 250 (2007). [CrossRef]  

6. C. Gao, M. Gao, Y. Zhang, Z. Lin, and L. Zhu, Opt. Lett. 34, 3029 (2009). [CrossRef]   [PubMed]  

7. B.-Q. Yao, X.-M. Duan, D. Fang, Y.-J. Zhang, L. Ke, Y.-L. Ju, Y.-Z. Wang, and G.-J. Zhao, Opt. Lett. 33, 2161 (2008). [CrossRef]   [PubMed]  

8. E. J. Zang, J. P. Cao, M. Zhong, C. Li, N. Shen, D. Hong, L. Cui, Z. Zhu, and A. Liu, Appl. Opt. 41, 7012 (2002). [CrossRef]   [PubMed]  

9. E. Munin, J. A. Roversi, and A. Balbin Villaverde, J. Phys. D 25, 1635 (1992). [CrossRef]  

10. H. Tsuchida, Opt. Lett. 15, 640 (1990). [CrossRef]   [PubMed]