1. Introduction

Optical parametric oscillators have proved to be flexible sources of tunable coherent radiation, particularly in the near and mid infrared spectral regions. Although such oscillators have been developed covering all time scales from femtosecond pulse to continuous-wave, it is the latter class of device that provides the highest level of spectral resolution [1]. The availability in recent years of the periodically-poled nonlinear materials in allowing enhanced flexibility in phase matching and, as a result access to higher nonlinear coefficients in crystals, resulted in substantial progress being made in the development of singly-resonant continuous-wave optical parametric oscillators [2–7]; a particular advantage of singly-resonant over doubly-resonant devices being greater ease of smooth and continuous tuning at the single frequency level. However, in spite of these developments the pump power required to reach oscillation threshold in singly-resonant oscillators still remained high (typically ~5W), so demanding substantial pump lasers. The development of the pump-enhanced parametric oscillator [8–11], a configuration in which the pump wave, as well as the signal wave, is resonant in the cavity containing the nonlinear crystal (hence resulting in the intensity of the pump wave being enhanced within the crystal), resulted in a substantial reduction in the (external) pump power required to reach threshold, and so allowed less-powerful and hence more compact pump sources to be used. The pump-enhanced optical parametric oscillator has more recently been demonstrated as a useful source in a number of spectroscopic applications [12–16].

In the present communication we report on the further refinement of the pump-enhanced optical parametric oscillator and its use as the source of coherent radiation for an optical imaging system of high spectral resolution in the mid-infrared suitable for imaging low concentrations of transparent (in the visible spectral region) gases though the use of the backscatter absorption gas imaging (BAGI) technique [17]. In particular we have applied the device to imaging low concentrations of methane gas under atmospheric pressure conditions. This technique may, of course, be readily extended to a wide range of other gases through appropriate tuning of the parametric oscillator.

Previous gas imaging schemes based on lasers have included the use of carbon dioxide lasers, dye lasers, and, more recently, lasers combined with nonlinear optical techniques as sources of the coherent radiation used for illumination [17–19]. Two particular previous schemes are relevant to the present work, one based on a continuous-wave singly-resonant optical parametric oscillator pumped by a fibre laser, the other based on a pulsed optical parametric generator pumped by a Q-switched diode-laser-pumped Nd laser [19]. We consider these further in relation to our own system in section 5 below.

2. Pump-enhanced optical parametric oscillator

Typically transitions in the mid infrared associated with a wide range of gases under atmospheric pressure conditions show linewidths of the order of 5GHz, due largely to pressure broadening. This means that an ability to hop reliably the frequency of the parametric oscillator over intervals of the order of 1GHz, say, is all that is required to satisfactorily access the chosen transition. In methane those transitions that exhibit the highest absorption occur in bands around 3000 cm^-1, and in particular the most strongly absorbing of these transitions is at 3057.7cm^-1 (3.27µm) [20]. Transitions in other bands (for example at around 1.5µm), exhibit absorption coefficients that are typically three orders of magnitude smaller, so there is considerable advantage in using a source at 3.27µm, as reported here.

 

Fig. 1. Pump Enhanced Optical Parametric Oscillator

Download Full Size | PDF

The pump-enhanced optical parametric oscillator, based upon a 10×3×20mm long periodically-poled RbTiOAsO₄ (PP-RTA) nonlinear crystal (Λ=39.6µm), is illustrated in Fig. 1, and is a further development of a device previously reported by us [21]; in particular to improve the frequency control for the idler wave. This is accomplished through extending the common cavity that resonates both the pump wave and the signal wave so as to introduce an arm able to accommodate an intracavity etalon, following a design proposed by Kovalchuk et al. [13]. The etalon E, mounted upon a galvanometer G for fine angular control, is fabricated in fused silica (1mm thick, 83GHz FSR) which is transparent at both the resonant signal and pump waves, and is coated for antireflection at the pump wavelength (1064nm), while exhibiting a reflectivity of 15% per surface at the signal wavelength (1577nm). The cavity arrangement employed is such that the etalon does not have to be transparent at the wavelength of the non-resonant idler wave since this wave exits the oscillator through mirror M2, before encountering the etalon (This mirror is coated on a calcium fluoride substrate, which is transparent at the idler wavelength). Since the etalon is antireflection coated at the pump wavelength it exhibits no frequency selection effects at this wavelength, and the change in optical path length that accompanies the tilting of the etalon (for mode selection of the signal wave) is readily compensated by a Pound-Drever-Hall servo arrangement that holds the cavity on resonance with the pump wave via the plane mirror M3 mounted upon a piezo actuator P. The fold mirror (M2), placed 9.5cm from NLC, has a radius of curvature of 20cm, and in association with the plane mirror M3 at the other end of the auxiliary arm of the cavity containing the etalon results in a large beam waist (~700µm calculated radius for the signal wave) at this mirror close to which the etalon is placed in order to minimise walk-off loss from the etalon and hence optimise frequency selection (M2–M3 separation is 8cm). The input mirror M1 (radius of curvature 2.5cm, placed 1.5cm from NLC), coated for 6% transmission at the pump wavelength and highly reflecting at the signal wavelength, also reflects the idler wave, which hence double passes the nonlinear crystal resulting in all of the idler output exiting through mirror M2. Frequency down conversion is facilitated via the PP-RTA non-linear crystal NLC which is enclosed within a servo-controlled oven (not shown) and maintained at a temperature of ~180°C in order to match the peak of the phase-matched bandwidth to that of the absorption feature of interest.

 

Fig. 2. OPO Tuning Behaviour. The shaded area represents the pressure-broadened absorption feature of methane at 3057.69cm-1

Download Full Size | PDF

Changing the crystal temperature facilitates coarse tuning of the signal and idler wavelengths over the range 1.53–1.60µm(signal) and 3.18–3.50µm(idler). We have also used multi-grating period PPLN in order to increase the idler power and extend the tuning range of the system to 3.70µm, limited by the bandwidth of the mirror optical coatings. The room temperature operation capability of PP-RTA makes it an attractive alternative where multiple, extended tuning ranges available from multi-grating PPLN are not required. The frequency of the idler was deduced by measuring the wavelengths of the pump and signal waves, leaking through mirror M2, with a precision wavemeter. The (fundamental) transverse mode of the OPO cavity is matched to that of the pump laser via the mode-matching lens L (focal length 100mm). The on-resonance pump beam radius at the focus within the nonlinear crystal is calculated to be 74µm×66µm (tangential × sagital). The slightly astigmatic nature of the mode is due to the (20°) off-axis mirror M2. With the above-described modifications to the cavity the OPO now exhibits a pump power threshold of 450mW, with a pump enhancement factor of ~9 implying that the circulating field within the cavity is of the order of 4W. When operated at a pump level of 900mW, some 55mW of idler can be extracted through mirror M2, corresponding to a down-conversion efficiency of 20%. Power conversion characteristics and other properties of this device are as reported by us in [21].

The improved tuning characteristics of the device are shown in Fig. 2. Coarse tuning of the signal and idler frequencies is achieved by varying the temperature of the nonlinear crystal; fine control by varying the angle of the etalon. Since the mode spacing of the signal wave in the present cavity is 670MHz (consistent with a cavity length of 21cm), fine-tuning (within a mode spacing) is not required in the present application, so the cavity can be maintained on resonance with the pump throughout as the signal frequency, and hence the idler frequency, is hopped from mode to mode through tilting the etalon. This process is illustrated in Fig. 2. Close examination of this figure shows that the hopping induced is not always to the nearest neighbour mode, but is erratic to the extent of possibly missing one to two intermediate modes on occasions, although the overall trend closely follows the frequency control expected of the etalon. We have considered this type of behaviour in the context of the superiority of ring resonators elsewhere [22], where we report how the cavity axial mode is systematically hopped from mode to nearest-neighbour adjacent mode, over several FSR’s of the etalon employed. However, in spite of the occasional lack of systematic hopping exhibited by the standing wave system under discussion here, the linewidths associated with pressure-broadened transitions in gases under atmospheric pressure conditions are such that the precision of tuning demonstrated here is perfectly adequate for the current purpose.

3. Imaging system

 

Fig. 3. Polygonal Scanner

Download Full Size | PDF

The geometry of the scanner used for imaging is shown in Fig. 3. Radiation from the optical parametric oscillator is directed along the optical axis of the arrangement by a small plane mirror m placed on-axis in front of the calcium fluoride collection lens L (focal length=14cm, f-number=2.8) and thence via the polygon scanner PS (Japan Aviation Electronics Industries Ltd., 10 facets each with dimensions (w×h) 20×30mm) and tilting mirror TM (GSI Lumonics Inc., model no 000-3008038, mirror area (w×h) 62×43mm) to the scene under surveillance. The back-scattered radiation returning from the scene is collected via the same tilting mirror and polygon scanner and is then focused by the collection lens onto the detector D (Hamamatsu thermoelectrically-cooled MCT detector, type no. P4631), located in the image plane of the collection lens. The area of the aforesaid collection lens (diameter=5cm) is sufficient such that the effective limiting collection aperture for the returned signal occurs at the polygon mirror facet. The above arrangement ensures that the detector always views that area of the scene currently being illuminated by the scanned radiation from the optical parametric oscillator (i.e., the viewing direction is scanned in spatial synchronism with the illuminating beam). The calcium fluoride lens LC (focal length 15cm) placed before the mirror m allows independent adjustment of the focusing of the illuminating radiation on the chosen target, in particular it allows the projection of a beam waist onto the target area so as to optimise the spatial resolution of the scanner in relation to the response time of the detector and the lateral extent of the area being scanned (see below). Since the MCT detector employed exhibits sensitivity over a broad range of wavelengths, a band pass filter F is placed in close proximity to the detector active area in order to reject stray infrared radiation from hot objects, laboratory lights and pump and signal fields that are leaked through OPO mirror M2. A Helium-Neon laser and detector TD are employed in order to trigger the image acquisition electronics at the correct point of the polygon rotation when scanning a horizontal line.

 

Fig. 4. Image of a methane plume acquired at an idler wavelength of 3.27µm (3057.7cm-1)

Download Full Size | PDF

The polygon scanner provides line scanning of the illuminating beam in a horizontal direction. Since it has 10 facets, then at the rotation rate employed here of 60 revolutions/sec, it scans the beam over an angle of ± 36° in a time of around 1.7ms. In the present arrangement the lateral dimension of the tilting mirror is insufficient to accommodate the full extent of the above horizontal scanning angle available from the polygon and in practice the externally available angular scan is limited to about ±14° as a result (This has the undesirable consequence that for about 60% of the scanning time the external scene is not being illuminated, and it is apparent that by increasing the lateral dimension of the tilting mirror the horizontal extent of the scanned area could be increased by a factor of about 2.5 without the need to increase optical power or scanning time). The tilting mirror provides scanning in the orthogonal (vertical) direction, and is set up so as to provide beam deflection over an angle of ±20° similar to that of the polygon scanner. The detector response time is of the order of 5µs, and the analogue signal from the detector is sampled, for the purpose of digitisation and storage, at 200kHz in accordance with this response time. This results in some 150 pixels being placed across the horizontal angular field (of ± 14°). The vertical angular field, as determined by the tilting mirror, is scanned with 100 lines, so resulting in a frame acquisition time of 170ms. In the present arrangement an additional time period of 150ms is required to up-load the acquired image data to a display computer (this period could be readily reduced by the use of a faster computer, improved interfacing or ultimately in-built display hardware), leading to an overall framing time of 320ms (3 frames/s).

Figure 4 shows a typical recovered image of a methane gas escape which is between 1.5 and 2m in front of the scanner, at which distance the angular scanning field (as described above) results in an area with dimensions of approximately 1m×1.5m (horizontal × vertical) being covered. The lens LC was used to set the projected beam waist within the above distance range and with a waist diameter of around 5mm so as to be consistent with the sampling time and the spatial extent of the line scan.

 

Fig. 5. Snaphot of accompanying video clips. The colour video (457kB) was shot under normal lighting conditions with a standard camera. The black and white clip (1MB) was captured with the gas imaging system at 3.35µm: the emerging methane plume is clearly visible. The video files are in Quick-time (*.MOV) format

Download Full Size | PDF

The accompanying video clips (running in real time) show the capability of the system for detecting the temporal evolution of a methane cloud. The effective CW illuminating power from the optical parametric oscillator was 30mW (although for the reason given above this power was only being used for about 40% of the time). To highlight the spectral flexibility of this system the optical parametric oscillator was tuned to a different absorption feature in methane to that used above, namely the line at 3.35µm (2988.7cm^-1), for which the absorption coefficient is similar. A sample frame from each video is shown in Fig. 5. The first (colour) video clip shows the author, holding a pipe through which methane can be released, standing within the target area of the polygon scanner and was recorded with a normal video camera. Throughout the clip methane is periodically allowed to escape from the pipe as it is moved about. As this clip was recorded under normal lighting conditions with a standard video camera, the leaking methane is not visible. The second (black and white) video clip, recorded synchronously with the colour video, shows the scene as captured by the polygonal scanner. The sequence appears somewhat strobed as the infrared scanning system has a reduced frame refresh rate (3fps) in comparison to the colour video camera. The leaking methane, released at a rate of approximately 70cm³s^-1, is clearly visible as it escapes from the pipe and is dispersed into the atmosphere. We have also successfully imaged the less demanding (in terms of spectral resolution [19]) case of butane gas.

Typically it is observed experimentally that for an idler illumination power of 30mW some 20–80nW of back-scattered radiation is collected by the collection lens L and imaged onto the detector in the case when no absorbing medium is present. The scattering background behind the methane cloud shown in Fig. 4 is at a distance of 3m from the scanner and the above is consistent with some 10–40% of the incident infrared radiation being (lambertian) back-scattered into 2π steradians. Since D*=8×10¹⁰cm. Hz^1/2/W for the above detector which has an area is 1×1mm, and for the employed sampling rate of 200kHz, the noise equivalent power is 0.6nW, comfortably small in comparison with the signal. The image at the detector of the projected spot has a linear dimension of 0.3mm, appropriately smaller than the size of the detector.

In order to ascertain the sensitivity of the system for the detection of methane, we employed a gas container built in the shape of a hollow wedge with polythene film for windows and which could be filled with methane gas in air at various methane partial pressures. On this basis we concluded that with straightforward direct viewing alone by the operator of the recovered image when the infrared radiation from the OPO was tuned on resonance with the methane absorption line, methane concentrations of the order of 30ppm-m could readily be visually detected. Implementing more advanced signal recovery techniques, such as on/off resonance comparisons (Fig. 6), would effect improvements in this detectability limit in addition to improving the image presented to the operator.

 

Fig. 6. Image enhancement by differential absorption. Figures 6(a) and 6(b) show images captured with the OPO idler frequency both on and off resonance with the methane absorption feature. The difference between these images (i.e., the methane plume) is shown in Fig. 6(c). Finally, false colour is applied to Fig. 6(c) which is then superimposed upon Fig. 6(b) to generate Fig. 6(d).

Download Full Size | PDF

5. Discussion

We discuss our own system in relation to two recently reported systems also based on parametric generation/oscillation as the source of the illuminating radiation.

Kulp et al. [19] have described a system based on a continuous-wave coherent source in which the output from a diode-laser-pumped single-frequency Nd laser is firstly amplified in a Yb-doped fibre amplifier, itself pumped by a diode laser, so as to provide some 4.2W of coherent radiation at 1064nm with which to pump a singly-resonant optical parametric oscillator based on PPLN and configured with a ring-resonator geometry. Some 250–300mW of tunable continuous-wave idler radiation was generated by this means for subsequent use in an imaging scheme based on the raster scanning technique. Fine frequency control of the idler wave was not implemented (in particular mode hopping was not suppressed), and subsequent applications of the source were demonstrated only in the context of gases having broad (and continuous) absorption features (typically 20–40cm^-1 wide), such as the higher hydrocarbons (e.g., propane).

The present scheme reported here, being based upon a pump-enhanced optical parametric oscillator and hence requiring <500mW of pump power in order to reach oscillation threshold (compared to 2.5W in the case of the singly-resonant oscillator just described), provides adequate idler power (~50mW) when pumped by a diode-laser-pumped Nd laser delivering only 1W, so substantially reducing the power demands placed upon the pump laser. This in turn reduces overall power requirements and the amount of waste heat generated, with the potential for a more compact device. It is noteworthy that the pump power required to reach oscillation threshold in the pump-enhanced optical parametric oscillator as quoted above refers to the case where PP-RTA is employed. This nonlinear material has a nonlinear coefficient roughly half that of PPLN, implying a threshold >4W in the case of the singlyresonant oscillator. The pump-enhanced approach, on the other hand, allows PP-RTA to be used efficiently even with a pump laser of modest power, such as the 1W system reported here. The considerable benefits of PP-RTA (and other periodically-poled arsenates and phosphates of the KTP type) over PPLN are (close to) room temperature operation (when an appropriate grating period is chosen in relation to the required wavelengths), unlike PPLN which must be operated above 120°C in order to avoid photorefractive damage, and greater immunity to thermally-induced focusing effects.

In the case of the pump-enhanced optical parametric oscillator employed in the present studies, the use of an etalon as a frequency selective element when placed within an appropriately designed cavity has enabled controllable hopping as a means of fine tuning (to ~1GHz precision) as well as ensuring that the selected frequency is retained during imaging. In this manner we have been able to image methane, where absorption features have linewidths of the order of 0.1cm-1.

A further scheme explored by Kulp et al. [19] employs a parametric generator based on PPLN, seeded by a laser diode and pumped by a Nd laser itself pumped by a 12W (pulsed diode-laser bar). The low repetition rate associated with this pulsed source restricts this approach to the use of line-by-line scanning/detection techniques hence requiring a multi-element linear array of detectors. On the other hand the fully continuous-wave approach based on the pump-enhanced optical parametric oscillator as described here employs only a single element detector leading to lower cost as well as increased flexibility in spectral coverage through ease of detector interchange.

At present our system is bench-top operated and further engineering including sub-system integration is required to make it fully portable.

6. Conclusions

We have demonstrated that a pump-enhanced optical parametric oscillator, based upon PP-RTA, is an effective source for gas imaging under atmospheric pressure conditions. With a pump laser of only modest output power (Nd:vanadate, 1W) the low threshold and efficient down-conversion of the pump-enhanced OPO allows sufficient output power to be generated (~50mW) so as to be able to scan significant target areas (typically 4m², at a distance of 3m) with sub second image acquisition times. Further, the combination of its broad spectral coverage (3.18–3.50µm) with its mode-hop tuning characteristics is such that with comparatively straightforward spectral control of the signal cavity mode, individual atmospheric pressure broadened lines with widths down to 0.1cm^-1 in gases over a wide spectral range may be routinely accessed. Absorption features of other molecular species out with this range could readily be imaged using this system by the substitution of a nonlinear crystal bearing an appropriate grating period in order to phase-match the down converted wave into the correct spectral range.

Scanning techniques are particularly appropriate for use with highly collimated, spatially coherent sources such as lasers and parametric oscillators. Without compromising sensitivity, the scanning approach allows single-element detectors as opposed to multi-element detectors in the form of cameras to be employed, leading to spectral flexibility since new spectral ranges (UV to far IR) can readily be accommodated simply by changing the detector technology involved.