Inelastic Scattering
Ken Voss, Ocean Optics Summer class, 2017 Introduction
Remember your freshman physics class and all that time you spent learning about collisions? Two types of collisions: elastic and in-elastic. What was the difference? Introduction
Remember your freshman physics class and all that time you spent learning about collisions? Two types of collisions: elastic and in-elastic. What was the difference?
Elastic collisions conserved Kinetic energy and momentum, inelastic just momentum….was energy lost? Introduction So elastic scattering: photons come in and out of the process with the same energy (wavelength)
In-elastic scattering: photons come in with one wavelength and leave with another wavelength. (remember Compton scattering of photons?). Difference between this and fluorescence is basically a short intermediate time for the event (which leads to other effects). prepared samples (see Fig. 1). All measurements were when recording Raman and luminescence spectra. done at a temperature of 21°C. This background-correcting spectrum was sub- A schematic for the experimental setup is shown in tracted from the raw experimental data. In the case Fig. 1. Laser light of wavelength 532 nm with a of Raman spectra, a straight line has been fitted in polarization controlled through the λ∕2 waveplate the spectral region from 700 to 745 nm. This straight was reflected by a dichroic mirror and focused onto line is shown in Fig. 2 and has been also subtracted the sample by a microscope objective. The degree from the spectrum. This subtraction has affected the of linear polarization of the incoming laser light is final estimate for the scattering coefficient by less at least better than 99.9%. The Raman scattering than 0.5%. Relating the total scattered light to the from pure water (impurity concentration <1 ppb) light collected from the cone intercepted by the mi- and luminescence of R6G were collected through the croscope objective requires knowledge of the angular same 0.55 NA microscope objective. A low-pass filter distribution of the scattered light. Generally the scat- (Semrock LP03-532RS25, optical density larger than tering from a single molecule is complex and must be 7 at 532 nm) removed laser light and a polarizer was described by a tensor [28]. We consider a geometry used to examine the polarization dependence of the where the exciting laser light, polarized along the sample’s radiation spectrum. The light was then x-axis travels along the y-axis through a bulk sample. focused into a spectrometer (SpectraPro 2300i) The optical axis of the microscope objective lies along with resolution 2 nm. The wavelength-dependent the z-axis of this coordinate system. The radiation transmissions of the optical collection system were scattered from a source much smaller than the wave- measured using a spectral lamp (Optronic Laborato- length of light but still containing a large number of ries Model: OL 200M) by comparing the measured randomly oriented molecules, which creates spheri- spectrum to the actual spectrum. The spectral lamp cal symmetry, can be described as the radiation manufacturer providesRaman a Scattering dataset for its spectrum of three orthogonal dipoles, with corresponding and the absolute intensities of the lamp have errors of less than 1% over a wider wavelength range than 1 1 covered in our experiment.Strongest, We assume most a generous (A)
0.5% error in the relative intensities of the lamp over ) the spectral range ofevident our interest. in natural − 1 0.8 0.8
The objective collects a cone of the scattered light nm with an opening angle θ (the NA of the objective used − 1 0.6 0.6
seawater, in-elastic (m in this experiment was 0.55). For each sample, six 5 spectra were measured,scattering corresponding process to all combi- is 0.4 0.4
nations of horizontal and vertical input and output ratio Depolarization 0.2 0.2 lights. For example,RamanSHV labels the scattering, spectrum of the vertically polarized component of the scattered light 0 0 obtained with horizontallyalthough polarized by laser most light. In 620 630 640 650 660 670 680 690 700 addition, spectra measured without any polarizers Wavelength (nm) are labeled by N. Weaccounts, measured the polarization water andhas intensity of laser light just in front of the cuvette. We 0.05 found that for H polarizationa weak the Raman light was slightly (B) attenuated and depolarized. We corrected this by Differential scattering coefficient x 10 adding weighting factors to the measured spectra. 0 cross section. (first 580 600 620 640 660 680 700 720 740 These factors are specified in the next section. Wavelength (nm)
seen 1920’s) 1500 2000 2500 3000 3500 4000 4500 5000 3. Analysis Frequency shift (cm−1) The raw spectral data had a relative standard deviation of 0.8% for R6G and 0.7%Excitation for Raman 532 in nm, AppliedFig. 2. Panel Optics, (A) shows Bray differential et al., 52,2503-2510, Raman scattering 2013 coefficient of water as defined in Eq. (11). We show this quantity since it water. These errors have been attributed to inconsis- has no dependence on the geometry of the optical collection tency in the sample preparation. We have used system. This curve equals the spectrum measured when the scat- Student’s t-distribution [27] to estimate the 95% con- tered light is collected with high numerical aperture (NA ≈ 1) fidence interval for the mean of six measurements. optics. The S-shaped line in the middle shows the wavelength- This interval turned out to be 0.85% of relative dependent depolarization ratio. Panel (B) shows the low intensity deviation for R6G luminescence and was the largest region for better visibility of weak lines and covers a wider range of single contribution to the final estimate of the uncer- wavelengths. The units of the vertical scale are the same as in tainty (the uncertainty in the mean of the Raman panel (A). All features (except for the several sharp spikes caused signal was 0.4% because 11 spectra were measured by noise in the depolarization ratio factor) are attributed to Raman scattering in water. The dashed line shows the Gaussian fit to the for water). The measured spectra have been treated band centered at 674 nm. Stokes detuning from the laser fre- to remove background contributions. This was im- quency is shown at the bottom of the figure for convenience of portant for relatively weak Raman spectra. A spec- the reader. A thin dashed-dotted line shows the linear fit sub- trum obtained using an empty cuvette has been tracted from the spectra to correct for a small gradient in the recorded under conditions identical to ones used background.
2506 APPLIED OPTICS / Vol. 52, No. 11 / 10 April 2013 Introduction
As opposed to the absorption process discussed earlier, the initial photon does not have to match an energy level to be absorbed, at least for a very short time, limited by the uncertainty principle (ΔEΔt<ħ). But the probablility of this happening is greatly enhanced if there is a nearby transistion to the virtual energy level(but if too close can cause issues of confusion with fluorescence).
RAMAN SPECTROSCOPY 423
Intensity, arb. units where s is the salinity in g/l, T is the temperature in °C; 1.0 s0 = 19 g/l and T0 = 21°C are two constants. Fitting (2) to the experimental data gives the following values for the parameters: 0.8 –3 –1 A ==1.451± 0.004, B ()10.8± 0.3 × 10 °C , –3 (3) 0.6 C = ()5.6± 0.2 × 10 l/g. In Fig. 2 the experimental results with relative error bars and the fitted curves are reported. It is important to 0.4 note that the overall agreement is obtained with only three parameters.
0.2 3. RESULTS OBTAINED WITH PULSED EXCITATION 0 The idea to perform a study of the temperature 2800 3000 3200 3400 3600 3800 4000 –1 marker with pulsed experiment is twofold: to be near to Raman shift, cm the actual experimental condition in field which used a pulsed excitation and to investigate the discrepancy Fig. 1. Typical Raman spectrum obtained in cw excitation between cw and pulsed results that have been reported superimposed to the fit based on a two gaussians model. in literature. For this purpose we have also studied the Raman Scattering RAMAN SPECTROSCOPY 423 presence of stimulated Raman scattering, possibly gen- Marker R erated in pulsed experiment and which may alter the Intensity, arb. units where s1.8 is the salinity in g/l, T is the temperature in °C; detected spectrum, as compared to the weak field (cw) 1.0 s0 = 19 g/l and T0 = 21°C are two constants.Salinity Fitting 38 g/1 (2) measurement. The scheme of the setup is reported in to the experimental1.7 data gives the following values for26 Fig. 3. The laser beam, of duration of 6 ns, diameter of the parameters: 0.8 1.6 6 mm and linearly polarized was the third harmonic of –3 –1 a pulsed Nd : YAG laser. The beam was passed through A ==1.451± 0.004, B ()10.8± 0.3 × 10 °C , 13 1.5 –3 (3) a 1-m-length cell of deionized water (s = 0) whose tem- 0.6 C = ()5.6± 0.2 × 10 l/g. perature was controlled within 0.1°C in the range 17– In Fig.1.4 2 the experimental results with relative 0 error 26°C. The back scattered light, collected by a pierced bars and the fitted curves are reported. It is important to mirror, and in some configuration, passed through a 1.3 0.4 note that the overall agreement is obtained with only polarizer, was focused onto the entrance slit of a 0.25 m three parameters. 1.2 focal length spectrometer. The setup is suitable for the 0.2 following scattering geometries: X(Y, YZ)X (spectrum 3.1.1 RESULTS OBTAINED WITH PULSED P + S); X(Y, Y)X (spectrum S); X(Y, Z)X (spectrum P). 0 5 10 15 20 25 30 35 40 EXCITATION The whole interesting spectrum was detected by a gated Water temperature, °C 0 The idea to perform a study of the temperature intensified OMA (Optical Multichannel Analyzer) and 2800 3000 3200 3400 3600 3800 4000 –1 marker with pulsed experiment is twofold: to be near to averaged over 500 laser pulses. The OMA was pro- Raman shift, cm Figuresthe actualFig. from2.experimental The ratioBecucci R obtainedcondition et from al., in fieldthe Laser measured which Physics, used spectra a with 9, cw excitation, as defined in the text [formula (1)], is vided with two array of photodiodes. We have used 422-425,pulsedreported excitation 1999 versus and temperature. to investigate The solid the lines discrepancy are the fit with these two arrays to contemporary detect the light scat- Fig. 1. Typical Raman spectrum obtained in cw excitation between cw and pulsed results that have been reported superimposed to the fit based on a two gaussians model. expression (2) and parameters (3). Each line corresponds to tered in axis respect to the laser beam or out of axis. We For water two main peaks, around 3200 incm-1 literature. For this purpose we have also studied the a different value of the salinity s: from top to bottom s = 38, will come back later on this point, crucial for our anal- and 3400 cm-1 (O-H stretch in water), ratiopresence depends26, of13, stimulated and 0on g/l. temp Raman andscattering, salinity. possibly gen- Marker R erated in pulsed experiment and which may alter the ysis in function of the laser energy. The determination Proposed1.8 as early as the 70’s to be useddetected with lasers spectrum, to as remotely compared to sense the weak water field (cw) of the instrumentation response function has been done Salinity 38 g/1 measurement. The scheme of the setup is reported in with a black body lamp. temperature.1.7 (Leonard, Caputo and26 Hoge)Fig. Also3. The usedlaser beam, to calibrate of duration ofa 6lidar ns, diameter return. of ture (from 3 to 40°C) and we extracted as a marker the We have performed two series of measurements: Note1.6 how to use this shift: 1/Final Lambda=6 mm 1/wavelength and linearly polarized – shift was the all third in harmoniccm or of a pulsedfollowing Nd : YAGadimensional laser. The beamexpression: was passed through changing the pulse energy at fixed temperature of the cm-1 13 1.5 a 1-m-length cell of deionized water (s = 0) whose tem- water cell and changing the temperature at a fixed perature was controlledR = (withinH1/W 10.1)/(°HC2 in/W the2), range 17– (1) energy. To be near to the experimental conditions of the 1.4 0 26 C. The back scattered light, collected by a pierced in field measurements, we have determined the marker where° H and W are the height and width of the Gauss- mirror, and i in somei configuration, passed through a from the P + S spectrum. 1.3 ians. The label i = 1, 2 indicates the higher and lower polarizer, was focused onto the entrance slit of a 0.25 m In Fig. 4 the integrated intensities of the Raman 1.2 focalfrequency length spectrometer. band respectively. The setup The is suitable choice for of thethe marker following(1), that scattering is an effective geometries: marker, X(Y, YZ)Xis leaded (spectrum by a previous spectra versus laser energies at T = 19.1°C for aligned 1.1 P +study S); X(Y, [6]. Y)X We (spectrum have then S); fitted X(Y, theZ)X marker (spectrum behavior P). with and unaligned detection (made with the two photodiode 0 5 10 15 20 25 30 35 40 arrays) are reported. In the figure the solid lines are fits Water temperature, C Thetwo whole variables interesting expansion spectrum of was the detected unknown by a functiongated R: ° intensified OMA (Optical Multichannel Analyzer) and with quadratic expression in the pulse energy, and the averaged over 500 laser pulses. The OMA was pro- Fig. 2. The ratio R obtained from the measured spectra with R = A + B(T – T0) + C(s – s0), (2) data are normalized to have equal linear contribution. cw excitation, as defined in the text [formula (1)], is vided with two array of photodiodes. We have used reported versus temperature. The solid lines are the fit with these two arrays to contemporary detect the light scat- expression (2) and parameters (3). Each line corresponds to tered inLASER axis respect PHYSICS to the laser Vol. beam9 No. or out1 of 1999 axis. We a different value of the salinity s: from top to bottom s = 38, will come back later on this point, crucial for our anal- 26, 13, and 0 g/l. ysis in function of the laser energy. The determination of the instrumentation response function has been done with a black body lamp. ture (from 3 to 40°C) and we extracted as a marker the We have performed two series of measurements: following adimensional expression: changing the pulse energy at fixed temperature of the water cell and changing the temperature at a fixed R = (H1/W1)/(H2/W2), (1) energy. To be near to the experimental conditions of the in field measurements, we have determined the marker where H and W are the height and width of the Gauss- i i from the P + S spectrum. ians. The label i = 1, 2 indicates the higher and lower frequency band respectively. The choice of the marker In Fig. 4 the integrated intensities of the Raman (1), that is an effective marker, is leaded by a previous spectra versus laser energies at T = 19.1°C for aligned study [6]. We have then fitted the marker behavior with and unaligned detection (made with the two photodiode two variables expansion of the unknown function R: arrays) are reported. In the figure the solid lines are fits with quadratic expression in the pulse energy, and the R = A + B(T – T0) + C(s – s0), (2) data are normalized to have equal linear contribution.
LASER PHYSICS Vol. 9 No. 1 1999 Raman Scattering Importance in the natural light field:
By the early 80’s the ocean optics community were getting multispectral instruments, either full blown spectrometers (for example Ray Smith and John Tyler, Vislab) or multi channel radiometers, such as the MER-1032 made by Biospherical instruments.
Seeing strange results: Diffuse attenuation coefficients less than water absorption. Raman Scattering K, which you expect has to be larger than absorption was being measured to be less than absorption when approaching red wavelengths.
Sought instrument problems: leaks in filters for example… hard problem in general where light is weak (red). Raman Scattering We knew Raman was there, particularly for laser excitation, most thought it was unimportant in natural light. Series of people figured out that we were wrong:
Sugihara, Kishino and Okami, J. Oceanogr. Soc. Japan, 1984.
Stavn and Weidemann, Applied Optics, 1988.
Marshall and Smith, Applied Optics, 1990
Raman Scattering
Characteristics of Raman scattering:
-4 -1 1) relatively weak, br =1.84 x 10 m at 532 nm, vs rayleigh scattering, 1.8 x 10-3 m-1. 2) scattering phase function like water Rayleigh, but higher depolarization factor. 3) wavelength shift approximately 3400 cm-1. 4) strength varies with wavelength, around λ-5, depending on whether you are talking excitation or emission wavelength and photon vs energy. Raman Scattering
Energy shift is constant (3400 cm-1), which causes a varying shift in wavelength, increasing towards red: Note how to use this shift: 1/Final Lambda= 1/wavelength - shift Raman Scattering Look at natural light field Water vapor, oxygen abs. lines along with fraunhofer lines Raman Scattering
Raman becomes important when the amount of light at the emission wavelength is reduced in elastic processes relative to the excitation wavelength ͑Eu or Lu͒ that we computed, the fraction with ab- sorbing gases, fw,isgivenby
fw͞o fw ϭ , fw͞o ϩ T͑1 Ϫ fw͞o͒
and we can see that near 720 nm, fw Ϸ 35–40%. As there is only weak gas absorption ͑principally due to ozone͒ for wavelengths Շ600 nm, the excitation wavelength for Ϸ750 nm, there is no need to con- sider gas absorption in the Raman excitation ͑i.e., the influence of absorbing gases on the Raman compo- nent of Eu or Lu is negligible͒.Otherthantheeffects of absorbing gases, we would be in nearly complete agreement with Waters had we used the same aw and the same spectral variation for br.
Fig. 3. Raman fraction of Lu for a water body consisting of pure 4. Raman Contribution to the Upwelling Light Field in seawater for 0 ϭ 20°, 37°, and 60°. Case 1 Waters We also made computations similar to those de- scribed in Section 3 for case 1 waters as a function of the pigment concentration C.Morel28 has shown more recent measurements of aw and a correct spec- 15 tral variation for the Raman-scattering coefficient. that the Morel and Gentili model performs well in ͑ ͒ Waters used the a values suggested by Smith and the computation of the downwelling irradiance Ed w attenuation coefficient K .BecauseEq.͑4͒ shows Baker22 for the entire spectrum. These are gener- d ally larger than those we used and significantly that the Raman source function is proportional to E0͑e͒,whichdecaysinamannersimilartoEd,the larger in the blue. The spectral variation of br was 15 10 3 Morel and Gentili model should provide a realistic taken by Waters to be ϰe͑e Ϫs͒ ,ratherthan 4.8Ϯ0.3 13 simulation of the Raman source function. In addi- ϰ ͑e Ϫs͒ measured by Bartlett et al. There are other differences with Waters’s compu- tion to the Jr,thescatteringphasefunctionforthe ͑ ͒ tations that can be traced to the fact that we do not particles at both e and is required to simulate the includeRaman the Scattering effect of absorbing gases in the atmo- Raman component of the light field at .Fortu- sphere. The most obvious manifestation of this is nately, the results are not strongly dependent on the the strong peak in the Raman contribution seen at phase function used for the particles. Following ear- lier research,21,33 we use the phase function mea- 720 nm in Waters’s computations, but are absent 26 from1) Raman ours. This important peak is due because to the atmospheric it shifts lightsured from by Petzold wherein it the is San abundant Diego Harbor. to To see water-vapor absorption at 720 nm that reduces the the influence of the particle phase function on the elasticwhere contribution. the “elastic” The part water-vapor is small. transmit- light field, we carried out simulations of the Raman component using the Petzold phase function as well LOWTRAN32 tance T there ͑provided by ͒ is Ϸ0.6 for 0 ϭ 34 60°. Calling f the fraction of Raman in the total as a Henyey–Greenstein phase function with asym- 2) Need tow ͞beo careful when interpreting metrymeasurements parameter g.Theasymmetryparameterfor at these the Petzold phase function is Ϸ0.924. The results wavelengths are presented in Table 1 where the notation HG͑g͒ ͑E or L ͒ that we computed, the fraction with ab- stands for a Henyey–Greenstein phase function with u u sorbing gases, f ,isgivenby asymmetry0.3 parameter g.NotingthatthePetzoldg w falls midway between HG͑0.90͒ and HG͑0.95͒ and 0.3 fw͞o that HG͑0.85͒ would be unrealistic for marine partic- fw ϭ , ulates, these computations suggest that the error in fw͞o ϩ T͑1 Ϫ fw͞o͒ the Raman component induced by a lack of precise knowledge of the particle phase function is likely to and we can see that near 720 nm, fw Ϸ 35–40%. As 3 ͑ be Շ Ϯ3% for C ϭ 0.1 mg͞m and Շ Ϯ5% for C ϭ 1.0 there is only weak gas absorption principally due to 3 Շ mg͞m .AstheRamancontributionisarelatively ozone͒ for wavelengths 600 nm, the excitation small fraction of the total ͑ϳ10%, see below͒,this wavelength for Ϸ750 nm, there is no need to con- error is not significant. sider gas absorption in the Raman excitation ͑i.e., the It is important to note that the Raman contribution influence of absorbing gases on the Raman compo- is also only weakly dependent on the particle scatter- nent of Eu or Lu is negligible͒.Otherthantheeffects ing coefficient. We varied B in the equation for b͑͒ of absorbing gases, we would be in nearly complete from 0.1 to 0.5 ͑the range 0.12–0.45 was suggested by agreement with Waters had we used the same aw and 400,0.0 Gordon and Morel20 for case 1 waters͒ and computed the same spectral variation for br. L at 520 nm ͑excitation ϳ443 nm͒.Theresulting Fig.u 3. Raman fraction of Lu for a water body consisting of pure 4. Raman Contribution to the Upwelling Light Field in Fig. 4. Raman fraction of E for a water body consisting of pure Raman contributions to L are presented for three u seawater for 0 ϭ 20°, 37°, and 60°.u Case 1 Waters seawater for 0 ϭ 20°, 37°, and 60°. values of C in Fig. 5 and show that Lu varies by at We also made computations similar to those de- scribed in Section 3 for case 1 waters as a function of 20 May 1999 ͞ Vol. 38, No. 15 ͞ APPLIED OPTICS 3171 Gordon, AO, 3166-3174, 1999 28 more recent measurements of a and a correct spec- the pigment concentration C.Morel has shown … … w 15 Note most important for cleartral (pure) variation water for the Raman-scatteringwhy? coefficient. that the Morel and Gentili model performs well in ͑ ͒ Waters used the a values suggested by Smith and the computation of the downwelling irradiance Ed w attenuation coefficient K .BecauseEq.͑4͒ shows Baker22 for the entire spectrum. These are gener- d ally larger than those we used and significantly that the Raman source function is proportional to E0͑e͒,whichdecaysinamannersimilartoEd,the larger in the blue. The spectral variation of br was 15 10 3 Morel and Gentili model should provide a realistic taken by Waters to be ϰe͑e Ϫs͒ ,ratherthan 4.8Ϯ0.3 13 simulation of the Raman source function. In addi- ϰ ͑e Ϫs͒ measured by Bartlett et al. There are other differences with Waters’s compu- tion to the Jr,thescatteringphasefunctionforthe ͑ ͒ tations that can be traced to the fact that we do not particles at both e and is required to simulate the include the effect of absorbing gases in the atmo- Raman component of the light field at .Fortu- sphere. The most obvious manifestation of this is nately, the results are not strongly dependent on the the strong peak in the Raman contribution seen at phase function used for the particles. Following ear- lier research,21,33 we use the phase function mea- 720 nm in Waters’s computations, but are absent 26 from ours. This peak is due to the atmospheric sured by Petzold in the San Diego Harbor. To see water-vapor absorption at 720 nm that reduces the the influence of the particle phase function on the elastic contribution. The water-vapor transmit- light field, we carried out simulations of the Raman component using the Petzold phase function as well LOWTRAN32 tance T there ͑provided by ͒ is Ϸ0.6 for 0 ϭ 34 60°. Calling f the fraction of Raman in the total as a Henyey–Greenstein phase function with asym- w͞o metry parameter g.Theasymmetryparameterfor the Petzold phase function is Ϸ0.924. The results are presented in Table 1 where the notation HG͑g͒ stands for a Henyey–Greenstein phase function with asymmetry parameter g.NotingthatthePetzoldg falls midway between HG͑0.90͒ and HG͑0.95͒ and that HG͑0.85͒ would be unrealistic for marine partic- ulates, these computations suggest that the error in the Raman component induced by a lack of precise knowledge of the particle phase function is likely to be Շ Ϯ3% for C ϭ 0.1 mg͞m3 and Շ Ϯ5% for C ϭ 1.0 mg͞m3.AstheRamancontributionisarelatively small fraction of the total ͑ϳ10%, see below͒,this error is not significant. It is important to note that the Raman contribution is also only weakly dependent on the particle scatter- ing coefficient. We varied B in the equation for b͑͒ from 0.1 to 0.5 ͑the range 0.12–0.45 was suggested by Gordon and Morel20 for case 1 waters͒ and computed Lu at 520 nm ͑excitation ϳ443 nm͒.Theresulting
Fig. 4. Raman fraction of Eu for a water body consisting of pure Raman contributions to Lu are presented for three seawater for 0 ϭ 20°, 37°, and 60°. values of C in Fig. 5 and show that Lu varies by at
20 May 1999 ͞ Vol. 38, No. 15 ͞ APPLIED OPTICS 3171 Raman Scattering
Table 1. Raman Component of Eu and Lu for Various Particle Phase Functionsa
433 nm 550 nm
Phase Function Eu Lu Eu Lu
C ϭ 0.1 mg͞m3 HG͑0.85͒ 0.5725 0.1577 0.2249 0.06847 HG͑0.90͒ 0.5412 0.1488 0.2135 0.06568 HG͑0.95͒ 0.5095 0.1410 0.2026 0.06393 Petzold 0.5268 0.1451 0.2083 0.06476 C ϭ 1.0 mg͞m3 HG͑0.85͒ 0.2152 0.06077 0.1509 0.04246 HG͑0.90͒ 0.1996 0.05600 0.1408 0.03986 HG͑0.95͒ 0.1803 0.05114 0.1294 0.03735 Petzold 0.1903 0.05377 0.1349 0.03848 a Eu is in milliwatts per square centimeter per micrometer Fig. 6. Raman fraction of ͑L ͒ as a function of C for ϭ 37° and L is in milliwatts per square centimeter per micrometer per w N 0 u wavelengths of interest in ocean color remote sensing. steradian. Should say…this modeling done without current knowledge of the UV. function of C over the range 0.027 Շ C Շ 5mg m3. 3 Gordon,͞ AO, 3166-3174, 1999 most Ϯ10% ͑at C ϭ 5mg͞m ͒ as B varies from 0.3 by The CZCS bio-optical algorithms were based on these Ϯ0.2. The fractional variation in Lu is even less for measurements,20,36–38 as was the semianalytic radi- smaller C.ThusthecontributionoftheRamanscat- ance model of Gordon et al.14 Gordon and Clark39 tering to L ͑and E ͒ is only weakly dependent on the u u used the measured Lu to compute the water-leaving elastic-scattering properties of the particles. How- 2 radiance Lw ϭ tLu͞n ,wheret is the Fresnel trans- ever, the elastic component of Lu is a strong function mittance of the sea surface ͑ϳ0.98͒ and n is the re- of B. fractive index of water. From Lw,theydefinedthe As mentioned in Subsection 2.D, the Morel and normalized water-leaving radiance ͓L ͔ through Gentili15 model cannot produce the correct depen- w N dence of the elastic component’s upwelling light field r͑͒ on C.Althoughprovidingexcellentvaluesofthe Lw͑͒ ϭ ͓Lw͔͑͒N cos 0 exp Ϫ ϩ O ͑͒ cos 0 , ͭ ͫ 2 z ͬͲ ͮ elastic Eu at 443 nm for small C,asC increases the decrease in Eu at 443 nm is much slower than ob- where ͑͒ and ͑͒ are the Rayleigh and ozone 3 r Oz served, e.g., at C ϭ 5mg͞m Eu is a factor of approx- optical thicknesses of the atmosphere, respectively. imately 3 too high. Therefore, rather than trying to Examples of ͓Lw͔N for ϭ443, 520, and 550 nm are simulate the upwelling field, we use direct measure- provided in Gordon et al.14 The original interpreta- ments at the emission wavelength. 39 tion of ͓Lw͔N by Gordon and Clark, as the water- On a series of cruises in support of the CZCS, leaving radiance in the absence of the atmosphere Clark35 measured the upwelling spectral radiance and with the Sun at the zenith, has been shown by just beneath the sea surface in case 1 waters as a Morel and Gentili17 to be invalid because of bidirec- tional effects in the subsurface upwelling radiance distribution. However, in the following we do not need this ͑or any͒ interpretation; we simply form the normalized water-leaving radiance of the Raman light field in a manner identical to that of Gordon and Clark39 for the total light field. We estimated the Raman contribution to ͓Lw͔N for 3 0.05 Յ C Յ 5mg͞m and 0 ϭ 37° using the model described above with the Petzold particle phase func- 35 tion. We then used ͓Lw͔N derived from Clark’s measurements for the total light field—the Raman component plus the elastic component. Figure 6 provides the Raman fraction of ͓Lw͔N as a function of C for the range 443 Յ Յ 595 nm. This restricted range is used because for Յ 443 nm, e Շ 380 nm, and the absorption of phytoplankton is not well es- tablished for wavelengths less than 400 nm. In fact, even for emission at 443 nm, we took the particle absorption to be that at 400 nm, and this is probably Fig. 5. Raman contribution to Lu ͑in milliwatts per square cen- 40 timeter per micrometer per steradian͒ as a function of the model’s too large and would lead to an underestimation of scattering parameter B for C ϭ 0.05, 0.5, and 5.0 mg͞m3.The the Raman contribution. The upper limit is re- solar zenith angle is 37°. stricted to 595 nm because, for տ 600 nm, instru-
3172 APPLIED OPTICS ͞ Vol. 38, No. 15 ͞ 20 May 1999 Measuring Raman Scattering
How do you measure?
In lab, 90 degree scattering experiments…must be careful to exclude excitation. Also careful of polarization effects on instrumentation.
In the field: 1) indirectly….make measurements and model how much Raman should be there given the measured excitation field.
2) directly….Ring effect….3 groups, basically simultaneously, but independently worked along these lines in early 90’s…..NOSC (now SPAWAR), TAMU, UM. TAMU only theory, NOSC specifically for a specific application, SLC, and UM both theory and experiment.
Key: Broad emission spectrum, sharp lines in the light field….Ring effect measures the filling of existing spectral lines by a broad emission source. Measuring Raman Scattering
653 483 488 659
Ge et al, JGR, 13227-13236, 1995 Using Ring effect
Can use this to also look at fluorescence.
4 10 11:02am, 480s A 11:12am, 180s
raw data 11:34am, 60s fitted data 3 10 12:06pm, 15s (relative units) (relative
d 13:01pm, 6s E
13:30pm, 3s
2 10 688 691 688 689 690 691 3 λ (nm) ! 5 10 688 691 10 11:02am, 150s E (0), 15s, 15:07pm w=0.202nm B d B 11:12am, 60s w=0.215nm Eu (0.8m), 8min, 10:55am raw data 11:34am, 30s 4 10 2 fitted data w=0.210nm 10 12:06pm, 15s Eu (3.0m), 9min, 11:15am (relative units) (relative
u 13:01pm, 6s Irradiance (relative (relative units) Irradiance L
3 13:30pm, 3s 10 517 518 519 520 1 λ (nm) 10 688 689 690 691 520 ! 517 λ (nm) ! Top 689 nm, bottom 518 nm, in shark river Above Brain Coral in Dry Tortugas 168 Hickman et al.
"q l
Incident beam ".. i/,,"k "-.. ~"
UB f~ ~ FB
,~" "t" q ~ .~ .,,,-~, .L ./" .- Using Eq. (2) and the fact that ]k[="q Ik'], we have curacy sufficientl to yield a good measurement of the following expressions:Other Inelastic scattering: sound velocity. Brillouin scattering From Eq. (5), the fractional root mean square 168 Hickman Incident et al. beam ".. i/,,"k "-.. 0 ~" qe = ]k'-k] 2 = k '2 + k 2 -2k'k' = 4k '~ sin e error in v, in terms of the errors in the other parameters can be written as "q (3) l UB f~ ~ FB and ,~" "t" q ~ .~ .,,,-~, .L ./" .- Since the phononsnamely,,~" propagate"t" q ~ at.~ .,,,-~, the .L speed of sound,./" .- Now Fluorescence Remember distinction is lifetime, longer intermediate state, more chance to “forget” information about incoming photon…. Phytoplankton fluorescence basically isotropically emitted (Gordon et al., L&O, 1993). Probably also completely depolarizing for phytoplankton, has been used separate polarization from natural light. Found one article using techniques on extracted compounds from Red Tide organisms. Polarization of Fluorescence generally used as an indication of the lifetime of the state and diffusion of the fluorophore. " The impact of algal fluorescence on the underwater polarized light field ", A. Tonizzo, A. Ibrahim, J. Zhou, A. Gilerson, B. Gross, F. Moshary, S. Ahmed (2010), Proceedings of SPIE Ocean Sensing and Monitoring II (5-9 April, Orlando, FL) , CDOM fluorescence Other fluorescent properties (besides Chl and pigments described in next lecture) CDOM Fluorescence, EEMS techniques (Paula Coble, USF) : 4092 J. Para et al.: Fluorescence and absorption properties of chromophoric dissolved organic matter (CDOM) [QSU] [QSU] 1 520 10 520 20 (Rhône plume intrusion) 480 480 8 16 23 2008 June 440 C 440 C 6 12 Note how emission depends 400 M 400 M 4 8 on excitation, and on sample 360 360 TT 320 2 320 TT 4 BB 2 m BB 280 0 280 5 m 0 200 250 300 350 400 450 200250 250 300 300 350 400 400 450 450 These samples 520 (photobleached water) 520 20 15 2 23 September 2008 480 From Bay of Marseilles 480 16 440 440 10 12 400 400 8 Peaks: 360 360 5 T T T 320 T ? 4 320 ? M UVA marine humic wavelenghtEmission (nm) 2 m 5 m 0 280 0 280 C UVA Humic like 200 250 300 350 400 450 200 250 300 350 400 450 Emissionwavelength (nm) 520 20 520 6 25 November 2008 3 (well mixed water) 5 B tyrosine like 480 16 480 440 C 440 C 4 T Tryptophine like Emissionwavelength (nm) 12 400 M 400 M 3 A UVC humic like 8 360 360 2 T T 320 4 320 1 5 m 280 2 m 0 280 0 Para et al., Biogeoscience, 2010 200 250 300 350 400 450 200 250 300 350 400 450 Excitation wavelenght (nm) Excitation wavelength (nm) 1045 Fig. 4. 2-D EEM contour plots of CDOM (in QSU) collected in the Bay of Marseilles (SOFCOM station) at 2 (left panels) and 5 m depths 1046(right panels)Fig. on4. 232-D June EEM (upper contour panels), 23plots September of CDOM (middle (in panels) QSU) and coll 25ected November in the 2008 Bay (bottom of Marseilles panels). These spectra illustrated fluorescent peaks positions observed during this study. Corresponding peaks fluorescence intensities are reported on Table 2. On middle 1047panels, question(SOFCOM marks indicatestation) a possibleat 2 (left slight panels) signature and of the5 m peak depths T. (right panels) on 23 June (upper panels), 1048HIX and23 September BIX values determined(middle panels) for marine and 25 (SOFCOM) November 2008ter samples (bottom came panels). from the These variability spectra concerning the pres- and freshwater (Arles) samples are presented on Table 3. ence of complex high molecular weight components (i.e., H, 1049During illustrated the study period, fluorescent for SOFCOM peaks positions samples observed at both duCVring 41%),this study. while lowCorresponding molecular weight peaks components part re- depths, HIX values were low and variable with 0.84 0.38 mained= more steady (i.e., L, CV 13%). The Rhoneˆ River ± = 1050and 0.90fluorescence0.35 at 2 m intensities and 5 m depths, are reported respectively, on Table while 2. Oirradiationn middle experimentpanels, question shows a marks strong decreaseindicate in a HIX at T1 BIX values± were high and stable: 1.10 0.17 and 1.09 and T2 triggered by H value decreased (H 1433, 1326, ± ± = 10510.05, respectively.possible slight These resultssignature suggest of the a predominantly peak T. au- 810 44 and 426 15 QSU at T0, Dark control, T1 and tochthonous origin of DOM in surface marine waters. BIX T2,± respectively) while± corresponding BIX remained con- maximum values were observed at 2 m depth on 23 June stant (Table 3), due to an alteration in similar proportion of 10522008 and 25 November 2008 while the lowest one was humic-like components, compared to T0 and dark control. observed on 7 July 2008 at 2 m depth. This date corre- Such results, strong decrease of HIX ( 50%) coupled to a 1053sponded also at the maximum values observed at both depths constant BIX between T0 (initial time)⇠ and T2 (final time) for HIX. At Arles station, Rhoneˆ River CDOM likely con- underline the higher photosensitivity feature of high molec- 1054tains compounds of higher molecular weight compared to ular weight DOM (i.e. humic-like components) compared to marine CDOM. Indeed, high HIX values (4.85 1.55) and low molecular weight DOM (i.e. protein-like components) low BIX values (0.74 0.05) suggest a predominantly± al- (Table 2). lochthonous origin of± DOM. The high variability of HIX (CV 32%), which is the ratio of H/L, for these freshwa- = Biogeosciences,46 7, 4083–4103, 2010 www.biogeosciences.net/7/4083/2010/ BLOUOHBTAL.: ABSORFrIONSPEC"IRA oF WATERS FROM ORINOCORIVBR 2275 3O a x arelarger than our analytical uncertainty, weare hesitant to concludethat thesevariations are real and not the resultof a sys- /x/x /x 0o tematicerror in our determination ofa x' Twopieces of evidence A AA AAA A point to a commonorigin for the COM from thesetwo regions: A ß A 2O (1) the indistinguishablevalues of S and(2) the similarityof their fluorescenceexcitation, emission spectra [Coble et al., 1990] (also S. A. Greenand N. V. Blough,manuscript in prep,3r.ation, 1992). .o :.,, Thevalues of a x andS forthe Gulf of Pariaand Orinoco River COMare tightly bracketed bythe values ofa* x and Sreported by 10 Zepp and Schlotzhauer[1981] for blackwater rivers in the CDOM southern United States and for soil-derived humic and fulvic acids.In contrast, thea* x reported byCarder etal. [1989] for marine fulvic and humic acids are approximately150-fold and CDOM typically0 measured in the field by10-fold eitherlower, respectively, absorption,than thoseof thewhich Orinoco and is Paria also its 3O COM. These comparisonsillustrate that the COM in these largest natural effect on the light field: regionsis dominatedby terrestriallyderived organic matter and o o 1.5 8 A These samples2O fromo% Orinoco Basin Blough et al, JGR, 1993 1.0 0.5 10 Absorption spectra typically: 0.0 o S(450nm−λ ) -0.5 o 1'o 2'o 40 aλ = a450nme SALINITY(ppt) -1'•2 Fig.5. Dependenceofa300 on salinity for transects (a)out of the mouth of 1.5 the OrinocoRiver and(b) throughthe Gulf of Paria. (a) Transecton September25, 1988(open circles); transect on September 27, 1988(solid circles);transect on October 9, 1988(open triangles); transect on October 1.0 S in these10, 1988 cases(solid triangles); on(b) the combined orderdata (open circles) of 0.014from transects on September29, 1988,October 7, 1988,and October 10, 1988along with the valuesobtained for stationsG, Y, Z, and M in the Caribbean 0.5 Typically(solid 0.014-0.02.squares). 0.0 Specificabsorption coefficients and valuesof S were measured for COM isolatedfrom the OrinocoRiver (stationR5) andthe Gulf of Paria (stationP4) to further examinethe spectral -0.5 propertiesof thesematerials and to estimatethe levelsof COM in the fiver and estuaries. The results of these measuremenu are -1.0 summarizedin Table 1 andFigure 9. The valuesof S were indistinguishablefor the COM isolated WAVELENGTH(n m) from the OrinocoRiver and Gulf of Paria and, within the uncer- Fig. 6. Log4inearizedoptical absorption spectra of surfacewaters col- tainties, were identical to those obtained for the waters in these lectedacross the salinitygradient for transects,(a) outof themouth of the regions.Over the range tested, changes in pH andionic strength OrinocoRiver and (b) throughthe Gulf of Paria. Parametersfor the didnot substantially alterthe values ofS and a*x (Table 1). These spectra in Figure6a are (downfrom the top) salinity15.06%o, $ = resultsare consistentwith the observedindependence of S on 0.0142nm q ßsalinity 24.56%o,$ = 0.0141nm q , salinity27.01 ø1oo, S = salinitybelow 30 O/oo (Figure 7). 0.0141nm '•' salinity31.10% o, $ =0.0141nm'• ß salinity 31.97% o, $= 0.0155nm q; andsalinity 35.29% o, $ = O.0190 nm -• . Parametersfor AlthoughS wasvery similar for the Orinoco and Paria samples, the spectrain Figure61> are (down from the top) salinity 4.82 o/oo, $ = the specificabsorption coefficients obtained for the COM isolated 0.0137nm '• ßsalinity 10.46%o, $ = 0.0138nm '• ßsalinity 16.92%o, $ = from the Gulf of Paria were ~ 40 to 50% lower than those ob- 0.0138nm '•; salinity17.27% o, $ =0.0138nm '•; salinity29.17% o, $= tainedfor theOrinoco River (Table1). While thesedifferences in 0.0154nm '•; andsalinity 33.19 ø/o0,$ = 0.0162nm q . 10654096 J. Para et al.: Fluorescence and absorption properties of chromophoric dissolved organic matter (CDOM) Marine humic like 2m Peak C 2m , Exc= 350 nm Peak T 11 (a) 11 (c) 0.90,9 0.90,9 23 Sept. 08 0.80,8 0.80,8 0.70,7 0.70,7 0.60,6 0.60,6 T2 (final time) 0,5 23 Jun. 08 0,5 0.5 0.5 T0 and Dark control 0.40,4 0.40,4 23 Sept. 08 0.30,3 0.30,3 25 Nov. 08 23 Jun. 08 0.20,2 CDOM0.20,2 0.10,1 0,1 0.1 25 Nov. 08 00 00 330 350 370 390 410 430 450 470 490 510 530 410Or 420 measured 430 440 450 460 with 470 480 fluorescence: 490 500 em 420-460 Marine humic like 5m Peak C 5m , Exc= 350 nm (b) 11 11 (d) Or single excitation and emission: 0.90,9 0.90,9 23 Sept. 08 0.80,8 0.80,8 Wetlabs ECO/FL: 370nm/460 nm 0.70,7 0.70,7 0.60,6 0.60,6 T2 (final time) 23 Jun. 08 0.50,5 0.50,5 T0 and Dark control 0.40,4 0.40,4 Fluorescence Fluorescence 23 Sept. 08 0.30,3 25 Nov. 08 0.30,3 23 Jun. 08 0.20,2 0.20,2 0.10,1 0.10,1 25 Nov. 08 00 00 330330 350370 370 390410 410 430450 450 470 490 490 510530 530 410 420430 430 440450 450 460470 470 480490 490 500 em 420-460 1066 Emission wavelength(nm) Emission wavelength(nm) Fig. 6. Normalized emission spectra of peak M at Ex 300 nm (a and b) and peak C at Ex 350 nm (c and d) acquired at SOFCOM station 1067 Fig. 6. Normalized emission spectra= of peak M at Ex = 300 nm= (a and b) and peak C at Ex = on 23 June (Rhoneˆ plume intrusion sample, black solid line), 23 SeptemberExcitation (photobleached at sample, 350 nm blue solid line) and 25 November 2008 (well mixed sample, green solid line) at 2 (upper panel) and 5 m (bottom panel) depths. Normalized emission spectra of peak C determined 1068at T0, dark350 control nm (red(c and solid d) line) acquired and T2 (duplicate,at SOFCOM orange station solid line) on of23 the June irradiation (Rhône experiment plume performedintrusion on sample, Rhoneˆ River sample collected on 7 February 2009 at 2 m depth were also plotted on both panels (c) and (d). These emission spectra were normalized to the 1069maximum black emission solid intensity line), in the23 rangeSeptember 380–400 (photobleached nm for the peak M andsamp 430–450le, blue nm solid for the line) peak C.and These 25 spectraNovember were smoothed by a moving average order 3 which imposes a red shifted of 5 nm. 1070 2008 (well mixed sample, green solid line) at 2 (upper panel) and 5 m (bottom panel) depths. observed offshore (Blough and Del Vecchio, 2002) or in all dates (except on 23 June 2008 and 25 November 2008) at 1071an oligotrophic Normalized coastal emission area not influenced spectra of by peak river C inputs. determined2 and at 5T0, m depths dark control (Table 2) (red was insolid accordance line) and to thatT2 reported High SCDOM could reflect either CDOM photobleaching if in surface Ise Bay in the Pacific coastal area (Yamashita and 1072aCDOM ( (duplicate,) is low as observed orange duringsolid line) summer of the (10 irradiation July 2008 experimentTanoue, 2003). performed Interestingly, on Rhône during River Rhoneˆ sample River plume in- and 23 September 2008) or fresh biological CDOM produc- trusion and mixing events in Marseilles Bay, fluorescence in- 1073tion in surfacecollected waters on if7 aFebruaryCDOM( ) is2009 high at as 2 observedm depth on were alsotensity plotted of peak on T both was onepanels order c ofand magnitude d. These higher at 2 m 23 June 2008. By contrast, low SCDOM with high aCDOM( ) depth (Table 2). The origins of peaks T and M have been at- as observed on 25 November 2008 suggest the presence of tributed to planktonic activity (Determann et al., 1998; Myk- 1074aged CDOMemission in surface spectra that couldwere be normalized the consequence to the of maximum the lestad, emi 2000;ssion Nieto-Cidintensity etin al.,the 2006; range Romera-Castillo 380-400 et al., strong mixing of deep water that was reported at this period. 2010) while the origin of peak C is known to be terrestrial 1075Concerning nm for CDOM the peak fluorescence M and 430-450 properties, nm ourfor studythe peak andC. Th thusese coming spectra from were freshwater smoothed inputs by (Sierraa moving et al., 1997, showed the dominance of recent autochthonous compounds 2005; Komada et al., 2002). However, peak C which is rela- 1076(peak T, BIXaverage>1) order and extremely 3 which low imposes values a of red humic shifted sub- of 5tively nm. abundant in deep waters could also originate from the stances (peaks C and M, HIX 1) within marine surface humification of marine DOM and thus may reach surface wa- ⇡ CDOM pool.48 Fluorescence intensity of peak T observed on ters during upward mixing events (Coble et al., 1998; Parlanti Biogeosciences, 7, 4083–4103, 2010 www.biogeosciences.net/7/4083/2010/ CDOM In most models of upwelling radiance, CDOM fluorescence is ignored, why? (hint on next slide?) Would be useful to have the quantum fluorescence efficiencies, which are rarely measured (one exception Hawes et al, Ocean Optics XI, 1992, on concentrated samples.) Most measurements are related to quinine sulfate or some other fluorescent material which makes it difficult to include in optical models CDOM flurorescence Oils Other materials: Oils fluoresce when illuminated by UV…can get fingerprints for these with EEMS systems. (Stelmaszewski, Optica Applicata, 405-418, 2004) Fluorescence methodfor the determinationof oil identity 409 Fig. 1. Examples of the normalized total spectra Φ of hexane solutions of oils. Minerals Minerals (such as calcite) fluoresce when illuminated by UV light. 365nm Ex looks pink, 320 nm looks red, 254 nm looks blue. Typically not enough UV to matter (except to geologists, and the Fluorescent Mineral Society). http://en.wikipedia.org/wiki/File:Fluorescent_minerals_hg.jpg Conclusion Inelastic scattering: Raman important to take into consideration when modeling light field and comparing models with data. Brillouin scattering: could be an interesting technique for lidar remote sensing temperature. Being applied by Thomas Walther (Germany now after Fry’s lab). CDOM fluorescence important for measuring CDOM, not as big a deal in the natural light field (in my data and view…not necessarily everyone). Oil fluorescence important for detecting, fingerprinting oil, not as much in the natural light field Mineral fluorescence important for identifying minerals…not in natural light field.