Ultimate sensing resolution of water temperature by remote Raman spectroscopy Myoung-Kyu Oh,1,* Hoonsoo Kang,1 Nan Ei Yu,1 Bok Hyeon Kim,1 JoonHeon Kim,1 JoonSeok Lee,2 and Gi Woo Hyung2 1
Advanced Photonics Research Institute, Gwangju Institute of Science and Technology, Gwangju 500-712, South Korea 2
Dongyang HiTech Industry Co. Ltd., Gwangju 502-881, South Korea *Corresponding author: [email protected]
Received 21 November 2014; revised 24 February 2015; accepted 24 February 2015; posted 25 February 2015 (Doc. ID 228351); published 24 March 2015
The limit of sensing resolution of water temperature by remote Raman spectroscopy was investigated experimentally. A remote Raman spectrometer, which employed a telescope of 20 cm in pupil size and the second harmonic generation (SHG) of a Q-switched Nd:YAG laser, was used for the measurement. By analyzing the broad O–H stretching Raman band located near 3500 cm−1, a parameter which is in secondorder polynomial relation with water temperature from 13°C to 50°C could be obtained. The resolution of our remote Raman temperature sensor was better than 0.2°C with measurement time shorter than 10 s. The influence of the Raman signal’s signal-to-noise ratio on the resolution and salinity effect on the accuracy of temperature sensing were also investigated. © 2015 Optical Society of America OCIS codes: (280.0280) Remote sensing and sensors; (280.6780) Temperature; (300.0300) Spectroscopy; (300.6450) Spectroscopy, Raman. http://dx.doi.org/10.1364/AO.54.002639
Since the late 1950s, the capability of Raman spectroscopy to monitor the temperature of a remote object has attracted much attention due to its high accuracy and high spatial resolution. Remote sensing of atmospheric temperature and liquid water temperature are one of the most famous examples [1–3]. The populations in the rotational energy levels of constituent molecules of air dependent on temperature result in the change of the Raman spectrum profile in the former case. While the populations of water clusters dependent on liquid water temperature lead to the change of the Raman spectrum profile in the latter. As for liquid water, the broad spectrum of O–H stretching Raman band near 3500 cm−1 is used as a temperature marker. The mechanism for the broad spectrum of the Raman 1559-128X/15/102639-08$15.00/0 © 2015 Optical Society of America
band has been under dispute until now. But the most prevailing explanation is that there are five O–H stretching bands in the spectrum. Four of them are symmetric stretching bands of different water clusters by hydrogen bond and the other one is an antisymmetric stretching band of a single molecule [2,3]. As temperature increases, the numbers of large water clusters decrease and those of small clusters or single molecule increase. This interpretation is supported by our experiment, the details of which will be discussed later. There have been several analysis methods proposed to quantify the Raman spectrum and determine the temperature of a given liquid water sample. The two-color method is the widely used one, where the whole Raman band is divided by two parts and the ratio between the areas under the two subspectra is used as a temperature indicator . Another important method is the deconvolution scheme, where component subspectra of Gaussian profiles are extracted by multifunctional fit, from 1 April 2015 / Vol. 54, No. 10 / APPLIED OPTICS
which a parameter linearly proportional to temperature is calculated . This is promising since the underlying physics can be directly reflected in the behaviors of the subspectra. Anyway, the variation of the temperature marker is about 0.5% per degree Celsius in each method. Therefore, it is expected that the resolution of temperature sensing by Raman spectroscopy depends mainly on the signal-to-noise (S/N) ratio of the Raman spectrum. Leonard et al. insisted that temperature accuracy better than 1°C can be realized for remote water sample or subsurface sea water with measurement time shorter than 1 s from theoretical calculation . They also developed a remote Raman spectrometer, which employed a nanosecond pulsed laser of 337 nm wavelength, and remotely measured sea water temperature in a sea shore. They also measured subsurface temperature to a depth of 30 m in the ocean moving on a ship. The estimated temperature accuracy of their Raman temperature sensor is around 1°C. It is true that gaining Raman spectrum of high S/N ratio for a distant target is challenging and there are many interference sources, such as salinity, surface wave, sunlight, depolarization effect, differential attenuation dependent on wavelength by water, etc., which seems to have prohibited them from obtaining temperature accuracy better than 1°C. In 1992, Lie et al. reported that they succeeded in acquiring temperature accuracy of 0.5°C in a field measurement by Raman LIDAR . Becucci et al. investigated the limits of accuracy in water temperature sensing comparing cw and pulsed excitation, where the salinity effect was also studied . The sensing resolution given in the report was ∼1°C in the case of cw excitation. The sensing resolution in pulsed excitation was reported worse than that in the cw case, to which the appearance of stimulated Raman scattering was attributed by the authors. No significant result was reported after these works, which reminds us the question on the feasibility of the Raman spectroscopy as a remote temperature sensor for liquid water. For the water Raman temperature sensor to be a robust tool, it should possess high temperature accuracy as well as broad application area. First, as for the accuracy, the ultimate limit of accuracy in the Raman-type temperature sensor and its dependence on instrument and measurement conditions are to be studied more. There cannot be found any sufficient evidence for sensing resolution much smaller than 1°C in all the previous reports until now irrespective of the measurement parameters. But high sensing resolution is often requested by environment scientists. Furthermore, the concept of accuracy limit in Raman temperature sensor seems to have been confused sometimes . Without doubt, the limit of water temperature sensing by Raman spectroscopy is determined by the intrinsic uncertainty of the relation between the temperature marker and water temperature as indicated by the authors . But the parameters defining the relation are specific to 2640
APPLIED OPTICS / Vol. 54, No. 10 / 1 April 2015
each Raman spectrometer and need to be determined in the best conditions possible prior to field measurements. So, to be accurate, the limit of sensing accuracy becomes clear only when the line shape (linear or polynomial of higher order) of the relation between the temperature marker and water temperature is well known and the influence of experimental parameters, such as resolution of a spectrometer, linewidth and stability of a pump laser, S/N ratio of Raman spectrum, etc., on the uncertainties of the parameters determining the relation are precisely understood. The laboratory experiment is free of outer interference sources and so provides an ideal environment to focus on the accuracy problem itself. Without environmental interference effects, accuracy has the same meaning as resolution in temperature sensing, the logic of which we followed in this report. In addition to the outer interference-free environment, a well-defined water sample as well as high sensitive Raman spectrometer employing low-noise array detector and high-power pump laser is needed for the precise investigation of the limit-ofresolution problem. With respect to the previous efforts for promoting the sensing resolution, one of the most important is that stimulated Raman scattering (SRS) instead of spontaneous Raman scattering was proposed by Leonard et al. in 1991 . From the report, SRS spectrum of water was found more sensitive to the change of water temperature than that of spontaneous Raman scattering. In addition, the high S/N ratio of the SRS signal due to the strong signal intensity provides more chance to increase the sensitivity of a Raman sensor  or the sensing resolution of liquid water temperature. However, the SRS scheme may not be adequate for quantitative water temperature sensing in the fields owing to the strict condition of the SRS (tight focusing of pump laser is needed) and the too high sensitiveness of the SRS signal to measurement parameters . Second, the application of the Raman-type water temperature sensor is another important point. One of the most promising applications is remote monitoring of the sea water temperature. Most of previous reports are about this. The current issue on global warming needs a remote and rapid sensing technique for sea water temperature, which is the main merit of the water Raman temperature sensor. Another good example is remote monitoring of cooling water temperature in power plants, such as nuclear power stations . In addition to telescopetype, a microscope-type noncontact temperature sensor based on Raman spectroscopy can find applications in laboratory experiments on bioscience, chemistry, material science, etc. In any case, it is out of question that the Raman temperature sensor will have much broader application when it provides higher measurement accuracy and fidelity established. In this work, a remote Raman spectrometer, where a telescope of 20 cm in pupil size and
532 nm nanosecond pulsed laser were employed, was built and used for a laboratory experiment, where sensing resolution smaller than 0.2°C (1 − σ error) was obtained. This is the highest resolution of the all experimentally confirmed values reported until now to our knowledge. In the following parts of this paper, the experimental setup and methods will be introduced first. Then, several analysis methods and their results will be given, which is followed by experimental results on the relation between the S/N ratio of the Raman spectrum and sensing resolution. And instrumental factors influencing the S/N ratio of the Raman spectrum will be discussed, after which experimental data on salinity effect will be given. Finally, conclusions will follow. 2. Experimental
A telescope-type remote Raman spectrometer was developed for sensing the temperature of a distant water sample. As a laser source, the secondharmonic generation of a Q-switched Nd:YAG laser (Quantel, Inc., model name: Brilliant B) was used. The time duration of the pulses is 7 ns. The repetition rate is 10 Hz and the maximum average power is 3 W. For signal collection, a telescope of 20 cm in pupil size was employed. An aspheric lens of 1 in. diameter and 5 cm focal length was inserted at the exit of the telescope to couple the collected Raman signal to a fiber bundle. The fiber bundle, which is composed of seven identical multimode silica fibers of 300 μm in diameter, guides the coupled Raman signal to a spectrometer. Before the aspheric lens, a notch filter (center wavelength: 532 nm, blocking band width: 17 nm, center optical density >6) was placed to block the strong Rayleigh scattering of the laser. A Czerny– Turner spectrometer of 20 cm focal length (Spectro, Inc., model name: Mmac200) was employed. Due to the entrance slit width of 300 μm from the geometry of the fiber bundle, the overall spectral resolution of the spectrometer is about 0.5 nm (∼10 cm−1 ) in the given wavelength range. A cooled silicon CCD array sensor (Andor, Inc., model name: iVac) was used for the acquisition of dispersed Raman signals. The set temperature of the CCD camera during the measurements was −60°C. For stabilizing and controlling the temperature of a water sample, a water cell was developed. The main body is made of stainless steel. It has two quartz windows of 2 cm diameter and 0.6 cm thickness. The optical path length is 30 mm. A temperature sensor chip (Thorlabs, Inc., model name: AD590) was planted in the water cell. Heating and cooling was executed by a thermoelectric cooler, which was driven by a temperature controller. The temperature of the cell was stabilized with resolution less than 0.1°C at any temperature between 5°C and 50°C and the water temperature was monitored by another temperature sensor (AD590), which was dipped in the water sample and connected to another temperature controller (TED200C). The accuracy of the temperature monitoring system consisting of the sensor chip and the controller is 0.1°C. And
Fig. 1. Schematic of the experimental setup. L, lens; M, mirror; W, window; NF, notch filter.
the temperature homogeneity of the water sample along the pumped region was also checked, from which the temperature variation along the region was confirmed as 10 mm in diameter lead to the high S/N ratio even with the short laser pulses of high energy prohibiting the occurrence of SRS. The low noise of the cooled CCD array detector also contributed to the high S/N ratio obtained. The well-defined temperature of the water sample within 0.1°C was also an essential part of the experiment. From this study, it was proved that the sensing resolution shows a shotnoise limit, the definition of which was precisely defined in this study, in the range of resolution over 0.2°C. Instead of linear function, the line shape of the relation between the temperature marker and water temperature showed polynomial function of second order but very close to linear. However, the limit of the sensing resolution is not clear below the resolution of 0.2°C until now, since it was
observed that the temperature resolution of the shot-noise limit was not improved with the increase of the S/N ratio below that level in the experiment. The more accurate line shape of the temperature marker dependent on temperature and the existence of the other effects than the S/N ratio of Raman spectrum limiting the sensing resolution can be investigated successfully only when the uncertainty of the reference water temperature is reduced. And it was proved that the resolution of the spectrometer as well as the linewidth and the stability of the pump laser are not so critical to the sensing resolution. When they are smaller than ∼1 nm, those parameters will not influence the sensing resolution of a Raman spectrometer, which has much implication in practice. While considering the instrumental factors determining the S/N ratio of the water Raman spectrum, accuracy as good as 1°C is expected for water 1 km away or subsurface water within 4.5 attenuation length in nearby detection with 1 s acquisition time and 1 m depth resolution. The sensing resolution will be much improved with longer acquisition time and the advanced pump lasers and detection devices seeming available in the near future. Remote temperature sensors with high enough accuracy like this will be very useful in many applications, such as wide range monitoring of ocean climate change, environmental monitoring for fishery, sequential monitoring of three-dimensional heat exchange in lakes or oceans for environmental science, industry process monitoring like remote sensing of cooling water temperature in nuclear power stations, precision temperature monitoring of water solution in microscopic scale for laboratory research, and so on. We note that there is no feasible alternative to this Raman-type temperature sensor for remote monitoring of water temperature. However, outer interference effects should be calibrated to get high temperature accuracy. Salinity effect, water pressure effect, surface water wave effect, differential (wavelength-dependent) attenuation and depolarization effect by turbid water, fluorescence signals, and sunlight will be the main sources waiting to be coped with. More laboratory experiments are needed for more precise measurement of such effects and development of efficient calibration methods. Anyway, this work will be a sound basis for the above advanced research. This research was a part of the project titled “Development of remote Raman spectroscopic sensor for the measurement of sea water temperature,” funded by the Ministry of Oceans and Fisheries, Korea, and also supported by the Public Health & Welfare Research Program (2010-0020794) and the “Ultrashort Quantum Beam Facility Program” of Gwangju Institute of Science and Technology. References 1. J. Cooney, “Measurement of atmospheric temperature profiles by Raman backscatter,” Appl. Meteorol. 11, 108–112 (1972). 1 April 2015 / Vol. 54, No. 10 / APPLIED OPTICS
2. Q. Sun, “The Raman OH stretching bands of liquid water,” Vib. Spectrosc. 51, 213–217 (2009). 3. D. M. Carey and G. M. Korenowski, “Measurement of the Raman spectrum of liquid water,” J. Chem. Phys. 108, 2669–2675 (1998). 4. D. A. Leonard, B. Captuto, and F. E. Hoge, “Remote sensing of subsurface water temperature by Raman scattering,” Appl. Opt. 18, 1732–1745 (1979). 5. M. Becucci, S. Cavalieri, R. Eramo, and L. Fini, “Accuracy of remote sensing of water temperaure by Raman spectroscopy,” Appl. Opt. 38, 928–931 (1999). 6. Z. Liu, J. Jhang, W.-Z. Chen, X.-S. Huang, and J. Ma, “Remote sensing of subsurface water temperature using Raman Lidar,” Proc. SPIE 1633, 321–329 (1992). 7. D. A. Leonard, “Method of remotely measuring subsurface water temperatures by stimulated Raman scattering using
APPLIED OPTICS / Vol. 54, No. 10 / 1 April 2015
9. 10. 11.
stimulated Brillouin backscattering,” U.S. patent 5,026,999 (25 June 1991). B. H. Hokr, J. N. Bixler, G. D. Noojin, R. J. Thomas, B. A. Rockwell, V. V. Yakovlev, and M. O. Scully, “Single-shot stand-off chemical identification of powders using random Raman lasing,” Proc. Natl. Acad. Sci. USA 2014, 12320– 12324 (2014). C. H. Chang, L. A. Young, and D. A. Leonard, “Remote measurement of fluid temperature by Raman scattered radiation,” U.S. patent 3,986,755 (19 Oct. 1976). G. W. Faris and R. A. Copeland, “Wavelength dependence of the Raman cross section for liquid water,” Appl. Opt. 36, 2686–2688 (1997). R. C. Smith and K. S. Baker, “Optical properties of the clearest natural waters (200–800 nm),” Appl. Opt. 20, 177–184 (1981).