Aug 15, 2004 - James B. Kelman and Douglas A. Greenhalgh. Cranfield University, Cranfield MK43 0AL, United Kingdom. Euan Ramsay, Dong Xiao, and ...
August 15, 2004 / Vol. 29, No. 16 / OPTICS LETTERS
1873
Flow imaging by use of femtosecond-laser-induced two-photon fluorescence James B. Kelman and Douglas A. Greenhalgh Cranfield University, Cranfield MK43 0AL, United Kingdom
Euan Ramsay, Dong Xiao, and Derryck T. Reid Ultrafast Optics Group, School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh E14 4AS, United Kingdom Received March 31, 2004 A novel technique is demonstrated for the imaging of turbulent f lows in which a single window to the f low is the only optical access required. A femtosecond laser is used to excite two-photon f luorescence in a disodium-f luorescein-seeded water jet. The f luorescence signal is generated at only the focal point of the laser because of the highly nonlinear nature of the two-photon absorption, and it is collected in a direction counterpropagating to the excitation beam. Tight focusing of the laser is used to limit the probe volume, and the two-dimensional mean and rms concentration images are collected by raster scanning the laser. © 2004 Optical Society of America OCIS codes: 110.2970, 140.7090, 280.2490, 280.7090, 300.2530.
The imaging of industrial f lows is often limited by the difficulties in obtaining good optical access to those f lows. Typically, at least two access windows are required.1 Although traditional planar laser-induced f luorescence (PLIF) imaging2 is powerful, in many engine conf igurations the extent of optical access required is rarely available without compromising the experimental configuration.3 The experimental method reported here demonstrates a technique for imaging the concentration in a f low tracer in which only a single access point is required. The single-pointaccess laser-imaging (SPALI) technique requires only a single window for both the laser access and the image collection. Although the technique relies on a scanned laser beam, creating a trade-off in temporal resolution versus the imaged region, it has many advantages for restricted engine geometries and increases the possibility of using industrial geometries for experimental measurements. Two-photon laser-induced f luorescence (LIF) measurements in combustion gases are not new4,5; however, these approaches have not exploited the potential for single-ended measurement offered by modern femtosecond sources. The SPALI technique relies on exploiting the high pulse peak powers available from femtosecond lasers6 to induce two-photon LIF in a disodiumf luorescein-seeded water f low. Although this f low is removed from engineering applications, it serves to demonstrate the technique at this stage. In our experiments the f low cell consisted of a single axial jet in a cof lowing stream with a turbulence grid upstream of the jet exit. The turbulence grid had 2-mm holes and a 32% blockage ratio, and the jet had an internal diameter of 2 mm. The high energy densities associated with femtosecond laser excitation allow the focusing optics and the two-photon f luorescence process to completely def ine the spatial imaging resolution in all three dimensions. (In traditional PLIF the spatial resolution of the image is def ined by the pixel separation on the camera for two 0146-9592/04/161873-03$15.00/0
dimensions and the thickness of the laser sheet in the third.) The highly nonlinear nature of two-photon absorption means that excitation of the dye occurred over only a limited focal region, which we measured to be between 50 and 100 mm in depth and approximately 50 mm in diameter, using optics with an f -number of 3. The f luorescence signal was emitted uniformly in all directions and could be collected from any direction without degrading the spatial resolution. The development of the SPALI technique was conducted in two parts. The f irst was performed with a frequency-doubled Nd:YAG laser (wavelength of 532 nm) to excite broadband single-photon f luorescence from the disodium f luorescein jet by use of a conventional PLIF geometry. The f luorescence was collected along a direction orthogonal to the excitation beam through a long-pass f ilter with a 580-nm cutoff wavelength. Two hundred instantaneous images were collected with an intensif ied CCD camera. To validate the SPALI method for acquiring accurate concentration prof iles we carried out a parallel experiment in which a plane of two-photon f luorescence was excited by a femtosecond Ti:sapphire laser oscillator. The laser produced 150-fs pulses at a repetition frequency of 100 MHz and operated at a wavelength of 800 nm with an average power output of 1 W. The experimental setup is shown in Fig. 1 and is similar to a conventional two-photon microscope, except that the imaging optics were selected to produce a scan covering several millimeters. Galvanometer beam-scanning mirrors allowed the laser beam to be raster scanned over a 64 3 64 grid in the f low, and confocal optics were used to collect f luorescence through a dichroic beam splitter and a bandpass filter from 500 to 550 nm. A photomultiplier tube was used as the detector with high-speed analog-to-digital acquisition. Data were collected at each point in the f low at a sample rate of 200 kHz. However, only a single data point was recorded at each step of the scanned mirrors, meaning that a 5-mm sample was © 2004 Optical Society of America
1874
OPTICS LETTERS / Vol. 29, No. 16 / August 15, 2004
did not manifest themselves. In our configuration the detectability limit for the disodium f luorescein solution was approximately 0.5 mg兾l. Figure 3 shows the mean and rms concentrations with conventional LIF imaging and the SPALI techniques. In Fig. 3 the large image in the background is the conventional LIF image, and the images in the white boxes are from the SPALI experiment. The SPALI images show excellent agreement with the concentrations in the traditionally collected data, with the decay of the jet and the mixing to the free stream
Fig. 1. Experimental conf iguration used for single-pointaccess laser imaging. M1, pair of orthogonal galvanometer scanning mirrors; L1, 35-mm focal-length scan lens; L2, 250-mm focal-length tube lens forming a beam-expanding telescope with L1; L3, 75-mm focal-length objective lens; L4, 75-mm focal-length f luorescence collection lens; BS, dichroic beam splitter with high ref lectivity at 800 nm and high transmission at 500 nm; FC, f low cell; DSF, disodium f luorescein solution; PMT, photomultiplier detector; SPF, short-pass filter (l , 550 nm). The black rays represent the excitation beam (800 nm), and the gray rays show the emitted f luorescence (500 nm). The scanned beam is shown at one extreme of its range.
collected during each step. The time resolution of each image was limited by the 1-kHz scan rate of the mirrors. Four hundred image sweeps were collected, with 400 individual images over the 64 3 64 region being scanned. These images were processed in the software to display the mean and the rms dye concentration for each location in the f low. It should be noted that, although the full image is not time resolved, it is possible to gather a time-resolved data stream at each point. Traditional PLIF can yield instantaneous images, but it is unable to provide the temporal evolution of the f low. There remains a trade-off in the techniques between spatial and temporal structures. The quantitative nature of the f luorescence was verified with the laser power and tracer concentration. Figure 2(a) shows the expected quadratic dependence of the f luorescence intensity with laser power, where the f itted line has a slope of 2, conf irming that the f luorescence was the desired unsaturated two-photon f luorescence from the disodium f luorescein tracer. For the duration of the experiment the laser power remained unchanged within detectable limits of approximately 0.1%. Figure 2(b) shows the detected LIF intensity as a function of the jet concentration, and an understanding of this dependence was particularly critical for the performance of SPALI.5 Figure 2(b) shows that for more than 2 orders of magnitude of tracer concentration the LIF signal remained linear and that absorption problems within the probe volume
Fig. 2. Fluorescence signal intensity versus (a) laser power and (b) tracer concentration.
Fig. 3. Mean (left) and rms (right) concentration images in a turbulent water jet. The arrows indicate the position of the 2-mm jet. The images collected with two-photon LIF are superimposed on the PLIF images in white boxes.
August 15, 2004 / Vol. 29, No. 16 / OPTICS LETTERS
in good agreement. The rms concentrations show the same mixing structure, but the SPALI technique yields higher rms values than conventional LIF. The conventional and SPALI images were collected with the same f low cell but with different dye solutions and in different laboratories, so it is not clear whether the differences in the rms concentrations are a genuine result or an artifact of the SPALI technique. One explanation is that the tight focusing and high resolution of the SPALI technique might allow it to resolve smaller structures and detect intermittency in the f low that is not resolved by use of conventional LIF. Cross-talk background, in which a signal from one region of the f low corrupts the background light level in another region, is not present in the SPALI technique because only one region of the f low is probed at any instant.7 This might explain the higher rms in the SPALI images. The results presented so far in this Letter were collected with only one of a number of possible data collection strategies. A preferred method is now to collect a time-resolved data stream at each point before scanning the mirrors to the next point. This provides a time history of the f low in addition to the mean and rms images. A promising alternative approach is simultaneously collecting time-resolved data streams at two separate points, allowing the evolution of the f low to be recorded. To do this, we modif ied the apparatus by removing the scan mirrors and replacing them with a Wollaston prism, which created two focused excitation spots in the f low with a 1.2-mm separation. Two pinholes in the detection optics allowed the LIF signal from each spot to be collected separately on independent photomultiplier tubes. Preliminary data recorded in this way show high concentration correlations for the f lows examined, but a detailed analysis is beyond the scope of this Letter. A natural development of the two-photon liquid-jet mixing reported here is to extend the technique to imaging gas-phase mixing. In a proof-of-concept experiment we imaged three-photon f luorescence in acetone vapor (a common
1875
tracer used in PLIF imaging) by use of high-energy femtosecond pulses, and this result demonstrates the feasibility of implementing multiphoton SPALI in a practical engineering environment. In conclusion, we have demonstrated that the SPALI technique shows great potential for imaging f lows through only a single access window with a resolution, linearity, and detectivity comparable with conventional PLIF imaging, provided that suitable tracers are available for the f low under investigation. The single access window required to collect images with SPALI should make the technique invaluable in providing data from geometries that have in the past proved too complex for orthogonal viewing. The authors thank Wilson Sibbett from St. Andrews University, Scotland, for his support and his assistance in developing the initial stages of the research. The work was supported by the U.K. Engineering and Physical Sciences Research Council under grant GR/ R28461/01. J. Kelman’s e-mail address is j.kelman@ cranfield.ac.uk. References 1. J. B. Kelman, G. Sherwood, F. O’Young, M. Berckmueller, M. C. Jermy, A. R. Masri, and D. A. Greenhalgh, Proc. SPIE 4076, 55 (2000). 2. R. K. Hanson, J. M. Seitzman, and P. H. Paul, Appl. Phys. B 50, 441 (1990). 3. M. Berckmueller, N. P. Tait, and D. A. Greenhalgh, in SAE Technical Paper 961929 (Society of Automotive Engineers, Warrendale, Pa., 1996), pp. 77 –99. 4. M. Aldén, S. Wallin, and W. Wendt, Appl. Phys. B 33, 205 (1984). 5. J. M. Seitzman, J. Haumann, and R. K. Hanson, Appl. Opt. 26, 2892 (1987). 6. W. Sibbett, D. T. Reid, and M. Ebrahimzedah, Philos. Trans. R. Soc. London Ser. A 356, 283 (1998). 7. D. A. Greenhalgh, D. J. Bryce, R. D. Lockett, and S. C. Harding, in Advanced Non-Intrusive Instrumentation for Propulsion Engines, AGARD Conf. Proc. 598, 25.1 (1998).
