wwee LIBRARY Michigan State “Unlveuity This is to certify that the dissertation entitled MATRIX EFFECTS IN THERMAL LENSING SPECTROMETRY presented by Charles Mark Phillips has been accepted towards fulfillment of the requirements for Ph.D . degree in Chemistry Major rofessor Date August 30, 1985 "flu-nth»: . ' '- IA . . . . 042771 MSU LIBRARIES m U RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. MATRIX EFFECTS IN THERMAL LENSING SPECTROMETRY By Charles Mark Phillipa A DISSERTATION Sub-itted to Michigan State University in partial fulfill-ant of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Che-iatry 1985 ABSTRACT MATRIX EFFECTS IN THERMAL LENSING SPECTROMETRY 3? Charles Mark Phillips A naJor goal of developments in analytical spectroscopy is to improve analyte detectability. However, serious linitations in ultratrace methods arise fro. the precision of the measure-ant itself as well as the precision with which the analyzed sample is defined. Questions of sample reproducibility, background contributions, impurities, and other interferences due to the matrix must be investigated in assessing the utility of a proposed technique. An ultratrace technique which has received luch recent attention is thermal lensing spectrometry which relies on the measurement of a radial refractive index gradient created in a (normally) liquid sample through heating caused by a partially-absorbed Gaussian laser bean. The strength of the gradient formed is directly proportional to the heat created and thus to the absorbance of the sample. Ultratrace amounts of analyte (on the order of 10'10 M) have been routinely measured by thermal lensing under laboratory conditions. The applicability of the technique to Charles Mark Phillips environmental samples has yet to be rigorously investigated. This thesis describes an investigation of solution salinity on the thermal lensing signals created by laser light absorption of a chronogenic product of phosphorous. Dissolved salts are a com-on environmental natrix, especially in the regime of marine chemistry, and their effects on thermal lensing signals have not been previously considered. The dissertation is divided as follows: 1) a brief historical introduction of the thermal lensing technique, 2) a description of the prevailing theories of therlal lens formation, 3) experimental details surrounding the construction of a thermal lensing spectrometer and of the colorinetric analysis of phosphorous, 4) results pertaining to the effects of saline matricies on the thermal lens signal, and 5) suggestions for further work. In addition to a definable saline effect, sensitivity limitations due to thermal flucuations and inpurities are addressed. The efficacy of techniques utilized to alleviate the. and thus extend the power of thermal lensing spectronetry are described. Dedicated to the menory of my father ii ACKNOWLEDGMENTS I would like to thank ‘Profs. George E. Leroi and Stanley R. Crouch for their guidance and expertise during the course of this work. Special thanks go to Prof. Alexander I. Popov for his helpful suggestions. I an also indebted to Dr. Thomas V. Atkinson and Dr. Ralph Thiim who aided in the writing of the colputer software used in this work. Thanks also go to Carol Zink, who typed the equations presented in this dissertation. I cannot forget the love and support of my family, especially my mother who saw me through the good and bad times. I also want to mention ny brother, Ron, who instilled the love of science in me. Another group of people which have neant much to me are all the friends (both scientific and non-scientific) I have met throughout ny life. They have taught me a great deal about living and have challenged ne to think about my responsibility to humanity. I humbly appreciate the caring and concern they have shown for me. I only hope that I have been fully able to give the same concern to then. I also thank my special compatriot, Maria, who proved to me that God’s love is never-ending. I am truly grateful for knowing her. iii Extra thanks to: ny stereo system (for keeping my mental stability), Wherehouse Records and Compact Disc Emporium (for supplying the music), MAC tours (for the best concerts), and Michigan (for being a wonderful state to live in)....and BOOM!!! My final thanks go to God; He has given me the faith and strength to carry on when it could have been just as easy to quit. iv TABLE OF CONTENTS LIST OF TABLES .......................................... viii LIST OF FIGURES. ............ . .............................. x CHAPTER 1 INTRODUCTION ............... . ....... . ........... 1 CHAPTER 2 THEORY ......................................... 6 Introduction ................. ..........6 CHAPTER 3 Characterization of Gaussian Laser Bea-SOOOOOOOOOOOOOOOOO .......... .6 Creation of the Thar-a1 Gradient ....... 9 Transformation of the Laser Heam......13 Lens Signal Expression ................ 18 Modifications to the Previous TheoryOOOOOO......OIOOOOOOOOO0.......026 EXPERIMENTAL ....... . ...... . ........ .. ........ .33 Introduction ............ .... ........ ..33 Laser.. ........... . .......... . ........ 33 Shutter. ...... ........................39 Lens.. . ...... ... ...... ..... . . 40 CHAPTER 4 CHAPTER 5 Cell .................................. 47 Mirrors ............................... 48 Detector Assenbly ..................... 49 Software ...................... . ....... 53 Preparation of the chromogenic compound ...... ................. ....... 59 RESULTS AND CONCLUSIONS ............... . ....... 65 IntroductiOn ...... .... ................ 65 Characterization of the Spectra-eter........ ........ . ....... ..65 Initial Investigations ................ 71 Detection Linits ...................... 78 Contaminant Effects ................... 81 Saline Matrix Effect ............ . ..... 83 Solvent Extraction .......... ..........88 Conclusions......... .................. 94 SUGGESTIONS FOR FURTHER INVESTIGATIONS ........ 95 Reduction of Thermal Fluctuations.....95 Phototherlal Deflection Spectrometry........... ........ . ...... 96 vi Dual Beam Techniques .......... . ....... 97 APPENDIX FORTH SOFTWARE... ............................. 98 REFERENCES ............................................... 108 vii LIST OF TABLES Table 2939 1 Ray transfer matrices for common optical elements. ......... . ........................... 15 2 Calculated enhancement factors for some common solvents ........... . ................... 26 3 Properties of the SDC SD 041-12-12-211 photodiode (all specifications measured with 0 bias voltage) ........... . .............. 50 4 Important FORTH words in the Thernal Lensing experiment ............................ 54 5 Typical values of 0, tc, and 1(0) obtained for one concentration of phosphorous analyte ........................... 74 6 Results of filtering analyte solutions........78 7 Results of the relative signal change in a 400 pptr phosphorous solution with a change in the background NaCl concentration...83 8 Test of interference of NaCl on the chromogenic reaction of phosphorous ........... 84 9 Calculated signal enhancement for various NaCl solutions (referenced to 0.0 M NaCl).....86 10 Calculated values of tc for selected NaCl concentrations.............. ............. 88 viii 11 Results of IBA solvent extraction ...... .......90 12 Results of IDA extraction from solutions of varying NaCl concentration. ...... .... ...... 93 ix Titers 1 6a 6b 9a LIST OF FIGURES Bax Diagram of TEMoo laser node ..................... 8 Ray diagram for a simple lens ......... ... ...... 16 Diagram of a thermal lens.... ............. .....19 Positional dependence of the thernal lens signalflOOOOOOOOOOOO00......0.0.0.0000000000000023 Parameters for diffraction theory of thernal sy.t°..00COOOOCOO...OOOOOOO0000.00.00.00000000028 Thermal lens apparatus with dye laser syste-OOOOOOOIOOOOOOOO00.00.00.000...0000.0000034 Thar-a1 lens apparatus with Er’ laser .y.te..0..........O...O..0...00.0.000000000000034 Diagram of astigmatically compensated cavity for CM dye laser (note the relative tilt of folding mirror to output lirror)........ ..... ..36 Diagram of‘a Ronchi ruling including a Iagnified view of the representative Pmax and Pltn measurements............ ..... ....43 Plot of 03 versus position relative to the focussing lens..........................46 9b 10 ll 12 13 14 15 16 17 18 19 20 21 Plot of 03 versus 23... ........................ 46 Schematic diagram of detector assenbly.........51 Photodiode response with laser power (note: power scale is lower than later observed powers due to use of opal glass diffuser in these experiments).............................52 Flowchart of thermal lensing experiment........56 Salient reactions in the determination of phosphorous using the common colorimetric process, with proposed stoichiometries of important species......... ...... .. ...... . ...... 61 Absorption spectra of reduced lZ-MPA and lZ-MSbPA solutions (200 ppb phosphorous).. ..... 64 Thermal lensing transients acquired with differing time delays (5.0x10'9M Crystal Violet in methanol)............................67 Noise level versus number of scans averaged in one experiment (n= number of scans averaged)...... ...... ............. ...... . ...... 68 Dynamic range check of the thermal lensing spectrometer with crystal violet solutions.....70 Working curve for lZ-MSbPA solutions with data scaling...................................73 Thermal lens time profiles upon sample .tirrin‘OOOOOOOOOOOOOO......OOOO0.0.0.00.00000077 Working curve for lZ-MSbPA solutions without data scalin‘OOOOOOOOOOOOO0.00.0...0.00.00.00.0080 Working curve for lZ-MSbPA solutions with erroneous data point due to contamination......82 xi 22 23 Observed enhancements (std. dev. bars) and calculated enhancements (diamonds) in the thermal lens signal as a function of saline concentration..... ....... ................ ...... 87 Comparison of thermal lens signal response for lZ-MSbPA in: aqueous solution, 0.5 M NaCl solution, IBA extract of 0.5 M NaCl solution, IBA extract of aqueous solution......91 xii I. INTRODUCTION Lasers are rapidly becoming standard photon sources in all forms of chemical endeavour. The properties of laser light - monochromaticity, directionality, phase and time coherence, extremely high spectral radiance and diffraction-limited focal volumes - make lasers ideal sources for trace chemical analysis. Normally, laser techniques rely on phenomena induced in species through absorption, rather than measuring the attenuation of the incident beam itself due to the inherent problems in measuring small changes on an extremely large signal. One such technique is laser-induced fluorescence. One-photon fluorescence is the most prevalent type; however, because of the high photon fluxes available from lasers, nonlinear absorption phenomena can also be detected using multiphoton-induced fluorescence. In laser-induced fluorescence, the sensitivity is limited by: l) the overall quantum yield of the species being probed, 2) the background due to scattered light, and 3) luminescence from the ”blank”. However, with careful optical and electronic design and sanple preparation, Dovichi et. al.1 reported a detection liait of 1.4 x 10-13 M in an aqueous solution of the fluorescent dye Rhodamine 60. This was equivalent to l detecting 35,000 molecules in the 6 picoliter probing volume. Most species, however, do not possess a large fluorescence quantun yield. Rather, the absorbed photon energy is dissipated into non-radiative channels through vibrational-translational relaxation, thereby heating the sample. These non-radiative photothermal effects have led to new forms of spectroscopy and spectronetry. One branch of spectroscopy linked to these photothermal effects is photoacoustic spectroscopy (PA8)3. PAS measures pressure waves induced by the increased kinetic energy of molecules following absorption of modulated light. These pressure waves are detected by a microphone or another transducer such as a piezoelectric device. This technique has been used to probe liquids3 and gases‘; however, the majority of work in PAS is directed towards measuring absorption spectra in solidsz. A gas (usually the ambient atmosphere) is utilised as the energy transducer in PAS of solids. The energy absorbed by the solid is transferred by convection into the gas above the sample, creating pressure waves in the gas, which are then detected. Resonant acoustic cavities can also be utilised, further ienhancing the sensitivity of this technique. Another effect of localised temperature increases can be seen on a hot summer day; when looking at the road fro- a distance the image of the pavement seems to be blurred. This is due to local density changes in the atmosphere and with it changes in the refractive index. These refractive index effects form another class of photothermal spectrometry which includes: photothernal deflection spectrometrys'll, interferometric based spectrometrylz'li, thermal diffraction spectrometry15'15, and photothermal refraction spectroaetry17'1'. By far the most widely investigated photothermal technique (and the subject of this dissertation) is thermal lensing (or blooming) spectrometryl’. The thermal lensing effect was first reported by Gordon, et al. in 196430. In an attempt to maximize Raman scattering from organic liquids, cells were placed within the cavity of a helium-neon laser. When the laser was turned on, power transients on the order of seconds were observed; these would decay and reappear upon movement of the cell. These effects were shown to be due to the for-etion of a refractive index gradient within the cell, producing a lens-like element which caused the beam to diverge. Thermal lensing was shown to be able to measure extremely small absorptions in ostensibly ”transparent” liquids’l‘a‘. Thermal lensing has been used to understand physical properties such as: absolute fluorescence quantum yields35, vibrational-translational relaxation33'3°, theraal diffusion in gases3°v31, and photolytic reactions33'33. In eaddition, theraal lensing has also been widely used as a tool to perform spectroscopy on systems with inherently low absorptions; for example, high vibrational overtones3“3° and multiphoton transitions37-33. Extensive surveys of spectroscopic applications of thermal lensing have been published by Hliger3° and Fang and Swofford‘0. Thermal lensing has only recently come in the fore of analytical chemistry. The first analytical study involving thermal lensing was performed by Dovichi and Harris in 1979‘1. It has rapidly burgeoned into a widely-used analytical detection scheme with applications in: trace analysis‘2'53, chromatographic detections“°2 and flow injection analysis53. In addition, thermal lensing is a major effect in another laser-based technique; laser intracavity-quenching spectronetry6“°°. Inherent to the afore-mentioned techniques is the ability to detect ultra-trace amounts of analyte. However, as has been lucidly presented by Harris and Williams", "The analytically significant property is not how small an absorbance can be measured but how snall a change in absorbance can be quantified...It is therefore necessary to examine the available methods with a concern for instrumental precision as well as the controllable and uncontrollable source of sample absorbance not related to the analyte concentration. These concerns heve been ignored previously because instrumental sensitivity expired before precision was the limiting factor in the analysis.” In transnission nethods, reduced precision can arise from absorption and scattering of light from several sources: 1) excess reagent, 2) solvent, 3) natrix (such as dissolved salts, organics, etc.), and 4) impurities. In calorimetric methods such as thernal lensing spectrometry (where absorbed power is converted into heat), the previously-mentioned interferents may have an effect on the thernal properties of the system, modifying the resultant signal even in the absence of any change in absorbance. However, no study has been perforned on the magnitude of the effect or on procedures to alleviate it. This thernal interference is the primary concern of this dissertation. II. THEORY letregestiee This section is designed to give the reader a background of the existing mathematical fornalisms that describe the formation of a thernal lens. A comparision of the various treatments is included. In addition, experimental techniques which have been suggested by theory are detailed. A detailed description of Gaussian laser beam formation (including the characterisation of laser nodes) can be found in. any text on quantum electronics, such as the excellent text written by Yariv". Nith the exception of self-terminating lasers (such as nitrogen, excimer, etc.), which have no definable beam structure, most laser cavities operate in the fundamental transverse Gaussian mode, designated TEMoo. The description of the radial electric field amplitude, E(r,z) for a TEMoo beam travelling along the z axis (shown in Figure l) is: E w 2 2 o o 232 _ _ lifl“ _ I“ 5”“) ' u(z) exp [’1 ( x ”(2) ) m “2] (1) where Be is the electric field amplitude at the center of the been, 0 is the "beam radius” or spot size corresponding to the radial distance where the amplitude is reduced by e"1 to that of lo, A is the wavelength of the beam radiation, n is an arbitrary longitudinal phase factor, and R is the radius of curvature of the constant phase surfaces. R is z 2 R-z [3 + (Eg'):] (2) where z: is the confocal distance or "Rayleigh range": given by: I“) (3) In (3), tb is the ninimum spot size at the focal plane of the been (where z=0). At the focal plane R becomes infinite. The beam radius at any distance z fron the focus may be calculated using: 2 «3(z)-u§ [3 + (if-) :] (4) c .evel usesu oollh he lauwsnn .n eased.- —— -~ ‘ “ '-— —_~ D ’- -- v, “ -eui3° (13 Therefore, once the minimum spot size, so, and the focal plane, z=0, are specified, the entire Gaussian beam is characterized. In a paraxial ray approximation, the beam divergence is snall enough such that: “l A A 0beam . tan ( moo ) . nwo (5) An equivalent description of the Gaussian beam is nade through the use of the complex bean parameter, q: 1: - "j; (6) I“) .0 I" al.-a This will be the useful parameter later’ on in the description of the propagation of the Gaussian been through the thermal lens. As the Gaussian bean passes through an absorbing medium, a time-dependent temperature increase in the nedium will develop; this increase will be the greatest along the axis. for an induced temperature rise AT, the refractive index distribution follows: n(r,t) - no + (an/aw) AT(r.t) (7) no is the bulk refractive index of the medium and dn/dT is the refractive index-temperature coefficient. For most media dn/dT has a negative value; that is, a tenperature increase induces a density decrease and hence a refractive index decrease. This gives rise to the formation of a divergent lens. However, in certain cases69 dn/dT is positive, resulting in a convergent lens. The expression for AT(r,t) can be determined by solving the radially-symmetric, nonsteady-state, heat convection equation70: a - Cp 5E-[AT(r,t)] . Q(r) + k VZLAT(r,t)] r0 for a- converging lens and fko for a diverging lens. Every optical element has associated with it a ray transfer matrix; some of the more conmon elements are listed in Table 1. When a ray passes through a systen of multiple optical elements, the overall matrix of the system can be calculated 14 Optical Element Matrix -- (1) Straight section 1 2 length 2 0 l (2) Thin lens: 1 0 focal length f -l —- l f (3) Dielectric interface: 1 0 refractive indices "e - "- o .2. n. (4) Spherical mirror: 1 0 Radius of Curvature R -_2_ 1 R Table 1. Ray transfer matrices for common optical elements. 15 meco_a _oodd .alen eualue s new leaned! hen .« ensign 16 by taking the product, in reverse order, of the matrices characterizing each optical elenent. In the calculation of the Gaussian beam transformation, the parameter of interest is the complex bean parameter, q, described in equation (6). It can be shown°° that changes in this parameter are easily described by: Aqo 4- B o where A, B, C, and D are the elements of the ray transfer matrix for one medium. The sane rule for multiple elenents applies as before; namely, that: q - tgfiL:-55 24 x Cxqo e Dx ( ) where Ax. Ex, Cs. and Us are the elements of the product ray natrix: x x n n : 2 2 1 1 (25) for n optical elements. The origin of the thermal lens optical Asystem is conviently set at the beam focus (z=0); at this point R=~ and by (6), qo= i zc. With this boundary condition, the spot size through any optical system is: 2 ’ Br (26) ) When the optical medium of the last element in the system is the sane as the first element in the system, A‘D‘ - B‘Cx - l (27) thus, (26) simplifies to B2 ez-eglA:"';—’) (28) 1<3 The relative change in e?- is what is measured in a thermal lensing experiment. The expressions relating this change to analytically useful terms are discussed in the next section . A simplified diagram describing a thernal lens is shown in Figure 3; the details are exaggerated for clarity. The extracavity beam waist (focus) is defined with a converging lens. This becomes the origin (z=0) of the optical system. As the thermal lens forms, the curvature of the bean Phase fronts change, causing a concomitant change of the bean spot size in the far field. Since the beam intensity 1!. inversely proportional to the bean area, it will also c=1'aange as the lens forms. The change in intensity is deternined by sampling the axial beam intensity (Ibc) in 18 .ese~ aslhema s we Isnhsua .a eulwuh O A _ . _ _ . . . . _ /l_\ ml H ..TADI/ I 19 .aseu delhelu s we Isnhsqn .a eaflluh 19 the far field using a limiting aperture such as a pinhole; a fiber optic cable has also been successfully used. The relative change in the observed intensity is thus a direct measurement of the relative change in the spot size: Ale - Ibc(0) ' Ibc(.) . (”2(a) - (”2(OL - 9-02.:- (29) Ibc Ibc(.) w2(0) For the optical setup in Figure 3 the ray product matrix will be: (30) where !(t) is the time-dependent focal length of the thermal lens, given by (18). Calculation of the product matrix and substitution of the pertinent terms into (28) gives the calculated spot size at the detector, we: 2 22 2 (z1 - z1zzlf(t) + 22)2 z c Since the relative change in the spot size is most easily detected in the far field, z: is allowed to increase without bound. In this case (31) simplifies to: Z 22 1 2 h' (5)2 ud-uozz (m) + (32) CNN 2 20 The value of interest is AIbc/Iec= Anna/waz; recalling that f(0)= 0: 2 2 2 “‘6 .AI..:"_‘..31_Z__2 (ad I r(-) r(-)2 (33) If the absorption of the medium is very small, the amount of heat created will also be very small which means that the lens created will be very weak; {(0)3 ) 212 + zcz. In this case the quadratic term will be negligible. However, as the absorption of the nedium increases, the quadratic term dominates. In the regime of absorptivities covered by thermal lensing, the quadratic term can be safely ignored: Ibo f(') (34) Upon substituting for f(~) with (19), for 02 with (4) and using (3) to simplify, (34) becomes: dn Alba -.2l P (dT) to 2z 2 . -——-———--- ---' (35) Ibc AK 22 . z An experimental characteristic of the thermal lens suggested by (35) is that the resulting signal is dependent upon the relative position of the thermal lens (cell) to the beam focus. Since 21 can take on positive or negative values, Alec/Inc may either be positive or negative. That 21 is, dependent on its placement, the lens may cause a relative divergence or convergence of the laser beam. This effect is easily explained; a divergent lens placed after the beam focus causes an increase in beam divergence, whereas the same lens placed before the beam focus causes a decreased convergence and thus a decreased divergence after the focus - in essence the bean focus is shifted further along the z axis. Hy differentiating (35) with respect to zi, the magnitude of the lens is found to be maximized when zi= 12:. The resulting function is an antisymmetric curve centered about the beam focus. .A representative example of this behavior is shown in Figure 4. A technique designed to take advantage of this asymmetry is differential thermal lensing spectrometry'l. Since most "transparent" solvents do have some absorption in the visible region, they can cause lens formation which becomes an unwanted background. However, if cells which both contain the solvent being used are placed an equal and opposite distance from the beam focus (according to Equation 35, +zc and -z:), the result will be a balance of convergent and divergent phenomena, ”cancelling” any common lensing due to the solvent. If the solvent in the cell located past the beam focus is replaced with a solution of analyte, any lensing then must be due exclusively to the analyte. Of course, this technique works only if the solvent has a small absorption coefficient (a) such that the lens created by the 22 'o-|./lu 5 + -O.15 1L J. «in w 1- AL. 10.0 70.0 Sample Position (cm) Figure 4. Positional dependence of the thermal lens signal. 23 solvent is weak; for most common solvents this requirement is met. If the cell is positioned for maximum sensitivity (zi= zc), (35) simplifies to: dn dn AI -.2“P(—) -.ZuP(.—) .23.. d7 . l 6‘1” I 36 Ibc lk in 2.303 AK A ( ) where the absorbance scale has been changed from base a to base 10. A direct comparision between thermal lensing and conventional absorption spectrometry can be made here. If ' the absorbance is very small: I-I AI I 0 __ 103(I-;)-2.303A- Io - I (37) whereas by grouping all the pertinent terms in the thermal lensing expression (36): dn AI -.2HP(—) Ibc - 2-303 [WA-L] A - 2.303 EA (38) be E (which encompasses all the terms in the parentheses) 'becomes the factor by which thernal lensing allows enhanced sensitivity over normal absorption spectrometry. The reason for this is quite straightforward. No matter the power of the source, in conventional techniques this factor is always ratioed out. In thernal lensing the signal is directly proportional to the source power; when lasers are used the enhancement can be quite large. Table 2 lists the 24 thermo-optical parameters and the calculated enhancement factors for some widely used solvents. Most non-polar solvents have a larger enhancement factor than polar solvents. This is due to the long range dipolar interactions between polar solvent molecules which can transfer absorbed energy quite efficiently. Since this transfer is so efficient (reflected in the relatively large thermal conductivity parameter, k), heat is conducted into the surrounding solvent rather than remaining localized around the beam area; this decreases the lens strength. Unfortunately, the most common solvent, water, is among the worst solvents for thermal lensing. One other solvent that should be mentioned here is supercritical 00231. Under the standard operating conditions of a supercritical CO: chromatograph the solvent exists close to a phase transition; in this case dn/dT * C. This situation results in enhancement factors which are extremely large (a value of 250 ml!"1 has been suggested as an upper reproducible bound for E/P). These equations represent an early description of the thermal lens phenomena. Although they are qualitatively correct land proved to be analytically useful, there were some corrections to be made which slightly changed the description of the lens. These modifications are discussed in the next section. 25 Table 2. Calculated enhancement factors for some common solvents. Solvent E_ls!_es:i§:ii deli!-i§:ii §£E_ls!:£i 0014 1.02 -5.8x10‘4 8.93 Benzene 1.44 -6.4x10“| 7.02 Acetone 1.60 -5.0x10" 4.97 Methanol 2.01 -3.9x10" 3.06 Water 6.11 -0.8x10" 0.21 002(1) >250 Eeéiiisetiees-te_ibe-£:ezieue-1!eesx In 1982, Sheldon, et a1.72 modified the theory of Gordon, et al.20 in two important aspects which led to differing quantitative results in both time- and position-dependence. The first modification was in the description of the temperature change induced by the laser beam. Initially, the same equations describing the heat evolution and convection [(8)-(l3)] were incorporated as before. Gordon truncated the series expansion after the quadratic term since it was determined that 87% of the beam energy was confined within one beam radius. In effect, spherical aberration of the lens was ignored; however, this turned 26 out to be a significant property. Therefore, the exact integral solution to (8): t _ 2 2 AT(r.t).a'—Zflj (c—1_-—) .exp(—M)dtt 211' C lepuz o l + 5‘;- l + (39) c was used by Sheldon as the term describing the temperature change of the system. The second modification was to relate the observed signal not to a spot size change, but to a change in the curvature of the beam’s constant phase front at the detector. Essentially, the complex phase amplitude at one point on an output plane (detector) is the superposition of all waves generated at the input plane (cell). Mathematically: 1 e 21 1 . cos exp [-1(Zsli)|zz-r|] Um“) - I ’0 L 010'.” ( ——-9'2 ) lzz‘rlfir dr d6 (40) is the integral describing this diffraction phenomenon, where all symbols are described in Figure 5. U:(r,t) is the complex phase and amplitude of the waves at the input plane. The second term is the inclination of the input plane edge to the beam center in the output plane. The third term is the attenuation and phase of the wave after traversing zz-r. ch(t) is then the resulting phase and amplitude of the superposed waves on beam center at the output plane (detector). 27 .esen Maine‘s no hassle sonuoshuhuv new shovelshsh .n shaman 2.2a ego... “sauna anac— ceaueaeo . a N .30 28 Since the radial dimension (r) is much smaller than sz, simplifications based on this approximation were made to Equation (40): 2! 2 . -iwr U (t)u-A ‘I I li(r,t)emp ( bc 0 o i izz ) r dr d6 (41) At this point, an expression for D:(r,t) was derived by considering the beam characteristics in the absence of a lens and then including the lens as a small pertubation in the phase of the unperturbed beam. This is reflected as a change in the optical path length of the medium: o(r.t) - 9.[n] (42) and from (7): um.) - 2. (33%} [AT(r.t) - mom] (43) Incorporation of O(r,t) into the phase variation of a fundamental Gaussian mode, D:(r,t), gave: 2 alum.) - 3 exp (2".2.) exp { -1 (I) (rZ/a + 24) } (44) U where B is a collection of constants and R and e have been defined previously. The first exponential term describes the radial amplitude of the beam at the input plane and the second term describes the phase change of the beam at the 29 input plane. A further approximation can be made about the phase change: 2n exp[-i(g1';')O]-l'1"x'° 4 %w<<1 (5) This is equivalent to saying that the lens formed is weak and thus the phase pertubation is small. For the range of weak absorptions probed by the thermal lens, this is a very good approximation. If one substitutes the exact solution of AT(r,t) (39) into (43), incorporates this into (44) and solves for (41) with the knowledge that Iec(t)= Dec(t)3, the fractional intensity change AI/I becomes: I(o)-I(-)- 1 _1 I(-) 1 _ a tan-1 (__2_r_2) (46) 3"; where (= z1/zc and 0 =.24Psl(dn/dT)/A k . One distinctive difference between this derivation and the previous one is that the maximum in (46) is located at (= J3; i.e. z1= J3zc. This may be contrasted with (35) which implies that the lensing effect is maximized at z1= zc. The itime dependence of the thermal lens also differs between the derivations. The time-dependent intensity 30 change derived by Sheldon is: -1 es I(t)-I(o)[1-etan (fi'fiffill (47) when the cell is located for maximum lensing (zi= JBzc). Notice that the denominator of the last expression is only dependent on t, whereas in Equation (18) the numerator is dependent on 2t, predicting a t: for the parabolic lens model twice that of the aberrant lens model. Since t: is a quantity derived from well known physical constants it was shown that the observed decay fit much better to an equation dependent on t (aberrant lens model) and not on 2t (parabolic lens model). In a recent paper, Carter and Harris73 have reinvestigated the discrepancies between the two theories. A comparision of the time-dependent signal expression was made, with a noticeable similarity. The parabolic lens model predicts: I(t) - 1(0) [1 - e ( W) i (43) whereas, if the tan‘1 term is series-expanded, the aberrant lens model predicts: 1 I(t) - 1(0) [1 - .577 e ( 771:7?) 1 (49) where 1(0), 0, and to are defined as before. 31 By comparing (48) and (49), the discrepancy of decay times is evident again. In addition, the parabolic lens model predicts a larger beam intensity change for a given 0 than does the aberrant lens model. Due to their similarity, either model can successfully fit an observed intensity change; however, after comparing the best fit variables to predicted values, the authors decided to modify the parabolic lens model by removing the factor of 2 from the time-dependent part of (48) and scaling 0 to come into accord with (49). This model also has the advantage of being based on a simpler and easily-visualized theory. The modified parabolic lens model was sucessful in correctly fitting position-dependent data, even in the presence of a strong lens (large 0) at and distances close to z=0. Apparently, higher order terms (in 0) incorrectly weight the position-dependent data. However, as was shown by Carter and Harris, when the cell is positoned for an optimum signal there is essentially no difference in the fit values of 0 up to 0~1.5. Therefore, in the studies which follow, the time-resolved data are fit to (47) using a non-linear least squares routine (to be described in the next chapter). 32 III. EXPERIMENTAL Introduction This chapter gives a detailed description of the thermal lensing apparatus, characterizing the optical, electronic, and software aspects of the experiment. In addition, the chromogenic reaction used for determination of phosphates is discussed. Block diagrams of the thermal lensing apparatus described in the text are shown in Figures 6a and 6b. The first studies employed used an ion laser-pumped dye laser system (Spectra Physics Model 164 Art ion laser/ Coherent Model 599-01 dye laser). The dye used was Rhodamine 6G (R6G) dissolved in an ethylene glycol/ethanol solvent mixture (80:20). Normally the dye laser produced between 50 and 150 mW in continuous wave (CW) output over tine spectral range between 610 and 630 nm. It was hoped tllat this laser system could implement the use of the laser 33 Laser H3 em I! Hi Pump Loser (Kr) O I Dye : Laser : cm I E I I I I I . x L : Plahele M: I5 M7 Phetedlode Shutter M2 ““ “0 \. Cell '3 [D I O : : “‘—— '3" Pinhole - “3 us Photodlede to micro Figure 6b. Thermal lens apparatus with Hrt laser system. 34 dye Styryl 9 (also called LDS 820) which has a maximum conversion efficiency of 162 at 820 nm and is directly pumpable with an Art ion laser. This wavelength regime corresponds to the absorption maximum of the product formed in the colorimetric determination of phosphorous ( mentioned later in this chapter). In this way, the maximum sensitivity of the thermal lens (as applied to this reaction) could be achieved. However, when performing feasibility experiments using R6G as the dye two related problems became evident. Most modern CW dye lasers (including the Coherent 599-01) utilize astigmatically compensated cavities'N (shown in Figure 7). This cavity has a three mirror configuration designed for long cavity length and small focal volume while minimizing Fresnel and reflective losses. In such cavities the folding mirror is fixed at an angle oblique to that of the output mirror, which introduces an astigmatic distortion. In addition, the dye jet is canted at Brewster’s angle so as to minimize reflective losses; this introduces another astigmatic distortion. What is done in the compensated design is to offset the distortion due to the dye stream with the distortion due to the folding mirror (if the cavity were an "in-line” design, the astigmatism introduced by the dye stream could not be compensated). In such cavities the TEMoo mode is not circular, but elliptical. All existing theories of thermal lensing assume a circular beam output from the laser. Therefore, the 35 .Ahohhnl unease ea weasel msuuuom he ends senasneh emu euosv semen one In new hearse weasemenlou unseemuslmuuss he Ishmsua .9 shaman Eoem not; aEam aEaa \Ai couoetom :3: use exo cecc 5 Lot.) 33:0 9:2: 36 cavity had to be adjusted slightly away from an optimum configuration to achieve a circularly-shaped beam. This obviously lowered the 0 factor of the cavity, making it much more prone to power losses. Dropouts of power (on the order of 1-2 ms) were frequent. Another undesired result was long-term power drift due to the pump mirror "relaxing” from its initial position. This effect (and how to deal with it) is discussed in the next chapter. In addition to the aforementioned problem, another characteristic of the output beam (even under optimum conditions) is spatial noise. This is due to imperfections in the dye jet being transformed on the output beam. Such imperfections arise from the following sources: I) dye jet jitter from noise created by the dye circulating pump, 2) dust or other particulates not efficiently filtered in the dye circulator, and 3) air bubbles introduced into the dye stream. These sources lead to beam shape (mode) changes and localized "flashes” on the output which are easily observed in the far field. Considering the difficulties with a visible beam (R6G), it was determined that investigations with an invisible beam (Styryl 9) would be prohibitively difficult. The replacement laser system used for later experiments was a Coherent Model 165 Er‘ ion laser which had been refurbished in-house". Single-line operation gave an output power up to 500 mW at 647.1 nm and up to 150 mW at 676.4 nm. The laser can be operated in either a 37 current-regulated or light-regulated mode. Laser power stabilization requires a warnup time of approximately 30 minutes. A slight 60 Hz ripple was noted on the dc output when the laser was current regulated; it varied between 1-22 (measured p-p) of the dc level. The magnitude of the noise was wavelength dependent, being slightly lower at longer wavelengths. The source of the noise is apparently local gain variations around the ionizing filament although the reason why the variations exist is not clear. It is known that Ert lasing processes are very sensitive to the pressure of the gas in the tube. This may be a related reason. When the laser is operated under light regulation, a photodiode monitors the fluctuations in power, controlling an error amplifier in the current regulator. This reduces the magnitude of the noise by about one-half and increases the frequency of the noise. In most cases, the laser was operated under light regulation except when low power levels are desired (20-30 mW); the minimum power level attainable under light regulation is approximately 50 mW. Any noise under either type of regulation is effectively dealt with through signal averaging and nonlinear fitting of the time-resolved intensity change. The output beam has no observable mode or intensity change in the far field; TEMoo is also circular. The power stability afforded by the Ert ion laser allowed for the 38 detection of an unwanted side effect - convective currents in the sample (described in the next chapter). §huttsr As suggested by the equations in Chapter 2, an initial time (t=0) must be experimentally defined. In pulsed laser experiments the initiation of the G-switch (when the laser fires) determines t=0. In quasi-CW experiments a rapidly rotating sector (normally, from 50-150 Hz) chops the beam, and lock-in amplification is implemented in signal detection; by adjusting the phase between the chopper and lock-in amplifier, t=0 can be easily determined from the position of maximum intensity. Fast chopping or pulsed experiments are performed when real-time data are required, such as in chromatographic analysis. Since real-time analysis was not required in the experiments described in Chapter 4, a computer-controlled shutter (Uniblitz Model 23XOA1T5; 3 mm diameter aperture) was used to establish the time origin. A toggling TTL pulse generated by setting one bit of an eight bit parallel I/O device (Intel 8255) is used to trigger the opening and closing of the shutter. This pulse activates a switching transistor on a controlling board (Uniblitz Model 100-23) which sends a 60 VDC open pulse and a 5 VDC hold level to the electromagnetically-controlled shutter mechanism. The manufacturer specifies a maximum of 39 1.2 ms between the time of activation and the time of full opening of the shutter; this delay is compensated in the controlling software (described later). The‘ same software also controls the length of time which the shutter remains open. Normally, this was set at 300 ms for reasons explained later in the software section. The shutter housing also contains an LED/phototransistor combination, placed on opposite sides of the shutter, which is used as an external synchronizer. At first, that system was used as a trigger for manual monitoring of the thermal lens with an oscilloscope. It was obviated when the shutter was placed under computer control. II." l0 IE I. The optic used to define an extracavity beam waist is an Oriel 350 mm focal length borosilicate crown glass planoconvex lens (f/13.8) set in a special Teflon holder which is mounted on an optical rail (described later). The focal length chosen was a compromise. A short focal length lens would cause z: < I, introducing a nonlinearity in signal response since different portions of the sample would experience vastly different spot sizes. A long focal length lens would force the need for an extremely long optical rail, since the beam waist would be located a long distance from the lens. A long focal length lens would also make no larger; as is seen in Equation 14, tc would also be longer, 40 thus increasing the time needed for an equivalent intensity change. In the first characterization study, the defining lens was placed close to the laser (dye), in hopes of making the system compact. However, anomalies in the positional behavior (apparent non-symmetry) of the signal response led to an investigation of the dye laser beam characteristics. It was noted that the dye laser itself had an extracavity beam waist and the lens had been inadvertently placed at the position of the waist. This did not permit the natural parabolic variation in beam spot size with distance. In effect, the beam was now allowed to converge and diverge symmetrically. This was easily rectified by increasing the distance between the laser and the lens with a folded mirror configuration (M3 and M4, Figure 6a). This same configuration was used in the studies performed with the Er’ ion laser (M1 and M2, Figure 6b). The defining lens forms an extracavity beam waist in which the beam spot size follows Equation 4. The confocal distance, zc, is a parameter of interest in this equation since it suggests the optimum placement of the cell. Therefore, by measuring spot size (0) versus position (z) a ‘walue of the confocal distance can be calculated. There exist several techniques which can be employed to measure the spot size of a Gaussian been, among them: scanning Pinhole", scanning knife edge", and scanning slit"""’°. However, these methods require precise and complex Domitioning techniques. A simple and accurate method which 41 can be used to measure Gaussian spot sizes is a scanning Ronchi ruling7°. A Ronchi ruling is made up of evenly spaced parallel bars etched on a glass substrate with a baked-in opaque black fill, forming an alternating pattern of opaque and transparent bars. The Ronchi ruling is placed in the Gaussian been nearly perpendicular to the beam axis (the ruling is tilted slightly so as to avoid interference from specular reflection off the ruling). A detector is placed immediately behind the ruling to monitor transmitted beam intensity. The Ronchi ruling is then moved in its own plans, such that alternate opaque bar and transparent space regions are intercepted by the beam. When the beam is centered on a transparent space, the detector sees‘ the maximum transmitted intensity (Pmax); conversely, when the beam is centered on an opaque bar, the minimum transmitted intensity (Pmin) is detected (see Figure 8). An aspect ratio, R: Pmts/Pmax can then be calculated. Under the assumption of a Gaussian beam and given a space (or bar) width, w, the l/e3 beam diameter, d(=2e) may be calculated over a reasonable range of E (0.103ESD.75) from: --2.2K+1 (50) which is accurate to 1127’. The simplicity of this method stems from the fact that no precise micrometer drive is required to measure the spot 42 nflfllnflfllm nflfllflflfllfl... . e a H m P nflflIflflflIm . nflfllnflfllm mflflllflfllt.... w 2 mflflllflfllt mflflllflfllr mflfllmflfllfl n u ing a magnified of a Hon the repr nts. iagram view of measurema size; that is, a precise knowledge of the relative position of the measuring device with respect to the beam is not required. The precision of the measurement is dependent only upon the precision with which the rulings of the Ronchi ruling are made. Another reason that the Ronchi ruling method is simpler is that only two measurements are required (Pmax and Pmin), whereas with other techniques many measurements must be made to completely characterize the beam spot. However, there are some disadvantages with the Ronchi ruling method. The first is that fine spatial variations of the beam cannot be resolved since the spacing of the measuring device is on the order of the beam spot size. Another is that Equation (50) only holds for a certain range of spot sizes for a given Ronchi ruling (l.2$d/w$2.7); therefore, if a beam has a widely varying spot size (tight focus), multiple Ronchi rulings must be used to characterize the beam accurately. A final disadvantage is that the beam must be in the fundamental Gaussian mode; any other beam intensity distribution will give errant results. Characterization of the Krt ion laser beam was performed with two Ronchi rulings manufactured by the Edmund Scientific Company: a 100 lines/inch ruling (#30513) equivalent to a space width (w) of 127 um, and a 200 lines/inch ruling (#30516) equivalent to a space width of 63.5 pm. The dectector apparatus used here is the same one which monitors thermal lens formation (described later). It, too, was tilted to avoid specular reflection, and an 44 interposed piece of opal glass diffused the laser light to prevent detector saturation. The results for a 50 mW, 647.1 nm beam are graphically depicted in Figure 9a. Once these data are plotted, the minimum in the parabola can be easily visualized; this corresponds to the focal point (z=0) of the system. All other points can then be scaled to this origin (2 scale) by addition or subtraction. If Equation 4 is rearranged, m2 2 uZ-~§e(-%)z <51) 2° it can be seen that a plot of 02 versus z2 should give a straight line with intercept 002 and slope eoz/zcz. A plot of .2 versus 22 is shown in Figure 9b. The least squares analysis of these data gives eo= 83.9 um and zc= 3.49 cm. The values calculated suggest a position of maximum enhancement at z= 6.0 cm, corresponding well with the observed maximum lying in the range of 2 between 5.5 and 6.5 cm. Another value which can be inferred from the least squares data fit is the characteristic time constant, tc. At the position of maximum sensitivity, 2: J3zc; thus 02: 44h2. Placing this value into Equation 14 with the tabulated constants for water, a value of 49.6 ms is calculated for tc, agreeing quite well with the observed range of to from 48 to 55 ms. 45 30000.0 25000.0 + N E 200000 1 + 15000.0 1' N + 3 + 10000.0 + + + + 5000.0 .30 .35 40 45 Position from lens (cm) Figure 9a. Plot of .2 versus position relative to the focussing lens. 25000.0 r- .1 A 20000.0 2.. N .4)— E I 1 15000.0 +- w i 3 10000.0 - llAlLllllLll —1—+—+—+-+—H—H—+—+-++—l—l—H—+-++i+o—I—++H—+—+—+-+—+—H 5 10 15 20 25 30 35 Z 2 ( c m 2 ) 5000.0 0 Figure 9b. Plot of e3 versus s'. 46 10 IO 10-0 IF The cells used in these studies were Hellma 110-08 fused silica cells with a light path of 10.0010.01 mm. They were chosen for their low absorptivity in the visible—UV region, thus minimizing any significant cell wall heating which could be transferred into the solution. The cells were held in a specially-made brass cell holder possessing a pivoted platform, such that the cell could be tilted with respect to the beam axis. This prevented any back-reflection of beams which could cause an optical interference; otherwise the cell could become a Fabry-Perot interferometer. Any such optical interference would mimic signal changes induced by the formation of the thermal lens. The cell holders were mounted on a Newport Research Corporation Model URL-36 low profile optical rail via an NRC Alb-2 translating carriage, allowing for free motion of the cell along the rail with the ability to lock the mount onto the rail using knurled screws once the optimum position was found. The rail facilitated the placement of a matched cell in a differential optical mode, mentioned in the previous chapter. However, due to background absorption in the blank, optical nulling of blank solutions was not possible. Transients with an initial intensity rise followed by the Characteristic intensity decay were observed in the 47 differential mode. Apparently, the stronger background reagent absorption caused the beam waist to shift significantly during the formation of the thermal lens, invalidating the simplifying assumptions in the differential lens theory. However, the investigations mentioned in the next chapter easily accommodated a single cell configuration; scaling was done by subtracting the observed value of 0 for the blank from the analyte 0 values. In fact, observation of the absolute O values proved very useful in explaining some of the data enumerated in the next chapter. 5122222 Mirrors placed after the cell fold the beam path to the detector, providing a large distance between the cell and the detector. This gives the detector a true far-field observation [see Equations (32), (40), and associated text], allowing for an easily detectable beam size change while keeping the apparatus within reasonable dimensions. The mirrors used in these studies were Melles Griot square flat reflectors (02 MFG 001; aluminum coated Pyrex, A/4 flatness) which have reflectivities greater than 902 in the visible region. The mirrors were mounted in specially-made aluminum holders, consisting of a mounting plate affixed to a backing plate through the use of steel springs. Three knurled adjusting screws pass through the backing plate and touch the mounting plate; these screws 48 control the horizontal and vertical steering of the reflected beam. The length of the folded path between the cell and the detector measured approximately 4.9 m, satisfying any far-field requirements. The mirror positions were optimized by placing a sample solution in the cell and adjusting the attitude of the mirrors until all reflected images were centered on all the mirrors. The mirror facing the detector was then adjusted until a maximum signal was seen on either an oscilloscope or the controlling-computer display, ensuring detection of the beam center in the far field. Qetesser_ésesshlr To assure that only the beam center intensity is observed, a limiting aperture was used, which in these studies was a thin stainless steel disc with a laser-drilled precision pinhole 10014 um in diameter (Melles Griot 04 PIP 015). Since the distance from the detector to the beam waist is approximately 5.0 m, the area of the spot at the detector can be easily calculated using Equation (4); it was found to be 4.54 cm‘. The area of the pinhole was then calculated to be 0.0023 of the area of the spot. It is thus safe to approximate the sampled intensity as the ”true” beam center intensity. The pinhole was affixed to a rubber grommet which acted both as the mount and light shield for the photodiode. With 49 the grommet as a template, the process of aligning the pinhole with the active area of the photodiode was very easy. The grommet was attached to a small aluminum box which contained the photodiode and associated electronics. The photodiode selected was a Silicon Detector Corporation model SD 041-12-12-211, chosen for its extremely low noise specification. In fact, the photodiode producedi no detectable dark current. Table 3 lists the salient characteristics of this photodiode. Table 3. Properties of the SDC SD 041-12—12-211 photodiode (all specifications measured with 0 bias voltage) NEP ( 1 kHz)....................... 7.3 x 10‘15 W Hz‘1/3 Active surface area................ 0.85 mm2 Response time ( 108- 902).......... 18 ns Responsitivity ( 300-700 nm) ...... . 0.12 - 0.44 A W-1 The photodiode was configured in a photovoltaic mode; that is, the photodiode was used purely as a photocurrent source, with no forward bias voltage. The current was transformed to a voltage using a simple operational amplifier-based (LF 351) current-to-voltage converter. The schematic diagram of the detector assembly is shown in Figure 10. The voltage output was found to be linear with laser power, as shown in Figure 11. A feedback resistor (Rt; 680 kn, 1/4 W) was empirically chosen to give an output voltage between 0 and 50 680 k0 i by ——> L \N i LF 351 ' V(n=-””‘LR1; Figure 10. Schematic diagram of detector assembly. 51 i-V Reading (V) as ‘P I. «ls as u d #1 0.” l 0.00 23.00 Power (mw) Figure 11. Photodiode response with laser power (note: power scale is lower than later observed powers due to use of opal glass diffuser in these experiments). 52 10 volts for laser powers between 0 and 150 mW. When the dye laser was employed, high frequency (>10 kHz) dye jet noise was prevalent. Therefore, a 20 pF capacitor was placed in parallel with the feedback resistor, resulting in an upper 3 dB point of 11.7 kHz, which reduced the high frequency noise without significantly distorting the intensity decay profile. This filter configuration was retained in the studies in which the Ert ion laser was used. The output voltage was sent via a ENC cable to an Intel 8088-based microprocessor designed by Enke and Newcome°°. At the microprocessor end the voltage was sent into one channel of an eight channel differential multiplexer (MUX; Datel MVD 807) whose output was amplified with a programmable gain amplifier (PGA; Burr Brown 3606 EG). The amplified analog output was then digitized using an analog-to-digital converter (A/D; Analog Devices AD 574). Unipolar operation of the A/D gives digitization of voltages between 0 and 10 volts; this is why the value of R: was carefully selected. Microsecond timing of the experiments was handled with a 1.0 MHz crystal and a counter/timer chip (Intel 8253). The controlling software is explained in the next section. Software The 8088 microprocessor is controlled via the FORTH computer language. The controlling software is listed in 53 Appendix A. The function of FORTH words pertinent to these experiments is listed in Table 4. Table 4. Important FORTH words in the Thermal Lensing experiment IQETE-!259 9292-122: Bernese (A12) Assembler Acquire one A/D value 1A12 Assembler Store one data value on floppy OSHUTTEH/ FORTH Open/close shutter CSHUTTER CURR FORTH Array containing last transient SUMMR FORTH Array containing 2 s)‘ .i DEVR FORTH Array containing 2(a) l )2 J TAKE FORTH Acquire one TL transient ADD FORTH Store 2 s in memory (32 bit) snsv roars Store 2 .2 in memory (32 bit) AVG FORTH Store 2 s and 2 s3 on floppy EXPS FORTH Runs a complete . experiment 54 The data acquisition routine is menu-driven, with the experimental variables clearly listed on the screen. A typical experiment proceeds as follows (a flow chart is diagrammed in Figure 12): the operator interactively selects the experimental parameters [number of data points per intensity decay transient, time between data points in a transient, number of transients to be averaged together in one experiment, relaxation time between transient acquisitions, and gain setting of the amplifier (PGA)]. The operator may at this time also choose to graphically display the transient and/or store the data on floppy disk. Once the experiment starts, the microprocessor triggers the opening of the shutter by setting one bit of the PIO, a software delay of 1.25 ms is then implemented which compensates for the opening time of the shutter. The analog 'signal is then sampled, digitized, and stored in the first cell of array "CURE”. The user-selected delay is implemented and the next point in the transient is sampled, digitized, and stored in a succesive cell of CURB. This procedure is repeated for the number of user-selected points in a transient. At the end of one scan, CDRH contains the A/D values of one time-resolved transient. Between the recording of each transient in an experiment (the ”dead time" - when the thermal gradient is allowed to relax) each value in CDHH is sequentially summed and stored in array "SEMI!” (each point in time is summed separately). Subsequently, the values in CDHH are squared, summed, and 55 / \ lee ‘. esperlreent? yes Change 9“ input psremeter/ I parameters I \ I JO! .MSCM \ la 1 ,MPI'S Aeeelre A/D velee Celeelete ts ilst so yes Stere 2 2! ts Figure 12. Flowchart of thermal lensing experiment. 56 stored in array "DIVE" (again, each point in time is handled separately). This process is repeated for the selected number of transients averaged in the experiment (a maximum of 127 transients is allowed to prevent integer overflow). If data storage was user-selected at the end of the experiment each 32 bit element of the SDMMH array is written to disk, followed by the corresponding 32 bit element of the DIV! array. A header containing the file name, experimental parameters, and comments for each experiment (which were entered through the menu before data acquisition) is written in the first 70 cells of a FORTH block, followed by the data. Once this is done, a FORTH disk directory is updated to include the addition of a new experiment. All experiment names and associated numbers may be retrieved by using the DIR command. If specific information (parameters) is requested, the number of the experiment is entered, followed by ED; this gives the experiment header which was written to disk. Another feature of the FORTH software is the ability to scan and plot stored data, allowing results to be observed graphically within a few seconds of an experiment’s completion. After the experiment is completed, the FORTH formatted floppy is placed in the floppy drive of an L81 11/23 minicomputer. The FORTRAN program "TL” searches the FORTH block for the parameters of a selected experiment. With this information, the program reads each 32 bit element of 57 SUMMH and DEVH and then calculates the average: i (gsJ )/ N (52) and the variance: 1/2 2 (s3)2 - N( Z s; )2 (53) N(N-1) for each time increment in the experiment; in this formalism s:* is the i-th time channel signal (A/D value) in the j-th scan, the i-index runs up to the number of time increments in one transient and j is the number of transients in one experiment (N). This information is written to an RSX-based file in either ASCII or binary format. The raw data are then read by the fitting routine ”DATFIT”. DATFIT is based on a non-linear least square (x2) minimization derived by Levenburg81 and modified by Marquardtaz. The core routines used here are contained in program ”MIIPACH”°3 which uses an algorithm devised by More°‘ to calculate the gradient search of a solution to a non-linear equation with minimal storage. The program requires a theoretical model [in these experiments, Equation (47), which fits I(o), 0, and tc] and a numerical calculation of the Jacobian matrix (derivative of the residual with respect to the parameter of interest). The MIHPACH routine was modified in DATFIT to accept the 58 calculated variance of each point in the transient and to calculate a normalized weighting factor: 1 of ’ E (54) °l r—m—o DATFIT allows the user to make initial guesses to the solution, simulate with this solution vector, and plot the simulated results versus the raw data. When the initial parameters are found to be reasonable, a fit may be performed. The fit window may be adjusted, a fit performed, and comparison of best-fit data to raw data for the entire data set attempted. This is extremely useful in cases where severe aberrations of the thermal lens at relatively long times are evident (approximations break down), but where at short times aberrations are negligible. Results of the fit, including uncertainties (the square root of the diagonal elements of the variance-covariance matrix), are printed out at the user’s discretion. Output files containing the raw data and the best-fit data compatible with the departmental plotting routine (MULPLT) are easily generated. Breecrstiee_2£-ihe-ebreseseeis_22222229 The particular preparations discussed here deal with the colorimetric determination of phosphorous. Phosphates are an environmentally relevant class of compounds due to their eutrophic properties. Many techniques have been devised to detect phosphorous in environmental samples, with 59 the majority based on the colorimetric determination of molybdophosphoric acid°5 or its analogs°°. Another procedure involves the spectrophotometric determination of a reduced form of the molybdophosphate species- the so called ”molybdenum blue” nethod". Originally, strong reducing agents such as tin(II) chloride and hydrazine sulfate were used; however many interferences were possible due to co-reduction of background species93. A gentler reducing agent, ascorbic acid, was later chosen which eliminated many of these interferences. The kinetics of molybdophosphate reduction were slow; to speed up the reduction, solutions were heated in a boiling water bath for 10 minutes". Later research found that certain trivalent members of Group V elements (such as bismuth) sped up the reduction of molybdophosphate in what was thought to be a catalytic process30. Subsequently it was shown that the bismuth was incorporated into the molybdophosphate complex". Another member which has been succesfully used to increase the reduction rate is antimony(III) ion ( as potassium antimonyl tartrate hemihydrate; E(SbO)Ce HeOs ’iflsO) . The accepted proceedure for determining phosphorous in marine samples, devised by Murphy and Riley°°, is based on the inclusion of antimony(III) ion followed by reduction with ascorbic acid. Going and Eisenreich" investigated the stoichiometry of antimony in this formation and reduction process and set forth optimum conditions for the preparation 60 of the reagent used in this proceedure. Repeated attempts by this author at preparation of the reagent enumerated in the afore-mentioned reference always resulted in the formation of a fine dusky-yellow crystalline precipitate. Upon closer investigation of the reaction conditions, it was found (with immeasurable chagrin) that the procedure set forth in Reference 91 resulted in an antimony(III) ion concentration 80 times higher than the maximum solubility limit of basic antimony oxides (mentioned two paragraphs earlier in the article)! Overall schemes for the reactions described above are given in Figure 13. “'0' I) P + MoOez' --* [PM012040]3' "12’ “PA" [PMosteol3' + red --+' [PMoszeol5‘ + ox ”heteropoly blue” H? II) P + M0042“ + Sb3* --* [PszM01oOeo]°' (7) "12‘ "SOFA" [PszMoxoOeo]°'+ red '-*’[P8b8u010040]11'+ ox ”mixed heteropoly blue" Figure 13. Salient reactions in the determination of phosphorous using the common colorimetric process, with proposed stoichiometries of important species A modification of the previous procedure was made in order to alleviate precipitation problems. A stock solution 61 of mixed reagent is prepared by dissolving 8.5 g of sodium molybdate dihydrate ( NazMoOc’thO) in 400 to 500 mL of distilled, deionized water (with stirring). To this solution is added 0.1371 g of potassium antimonyl tartrate. Then 83.2 mL of concentrated sulfuric acid is carefully and slowly added. At this point the solution may become cloudy, but it should clear after a period of about one hour. The solution is allowed to cool and diluted to 1000 mL. This stock reagent solution should ”age" for at least 24 hours, allowing the formation of molybdate polymeric species in the acidic solution’z. Fresh solutions of ascorbic acid are prepared prior to analysis by dissolving 17.5 g of the reagent in distilled, deionized water and diluting to 1000 mL. A stock solution of phosphorous is prepared by dissolving 0.4395 g of primary standard (dried and dessicated) potassium dihydrogen phosphate (EH2P04) in distilled, deionized water and diluting to 1000 mL; this is equivalent to a 100 ppm phosphorous solution. ' Successive dilutions of this stock solution are performed to give a 10 ppb phosphorous solution. Selected aliquots (0-10 mL) of the phosphorous stock solution were introduced into 100 mL volumetric flasks, followed by the addition of 5 mL of the antimonylmolybdate reagent and the addition of 5 mL ascorbic acid solution. The solutions were allowed to react (they turned a very light bluish-green in the process) and then diluted to 100 mL. Representative spectra of the reduced (”blue”) form for 62 both the molybdophosphate (l2-molybdophosphoric acid; 12-MPA) and the antimonylmolybdophosphate (lZ-MSBPA) species are presented in Figure 14. Another advantage of the addition of antimony(III) is seen here; since there is stronger absorption on the shoulder to the blue of has: (the region of available laser radiation), enhanced thermal lensing should result. A further modification to the previous procedure is to extract the molybdophosphate solution (or its antimonyl analog) with an oxygenated organic solvent such as propylene carbonate’3; the most common extractant is isobutyl alcohol°‘. In the studies discussed in the next chapter, 25 mL of isobutyl alcohol was introduced to 50 mL of an aqueous antimonylmolybdophosphate solution, shaken for two minutes, separated, and diluted to 50 mL. The reasons for extraction are discussed in the next chapter. .IO r ....... '2- MFA _ 12“ MSDPA .08 +- 11wa 500 600 700 800 900 k,nm Figure 14. Absorption spectra of reduced l2-MPA and lZ-MSbPA solutions (200 ppb phosphorous). 64 IV. RESULTS AND CONCLUSIONS letredsetiee This chapter describes many of the properties and anomalies of the thermal lensing experiments performed in the course of this thesis research. Among the topics covered are: characterization of the spectrometer, problems associated with the dye laser source, effects on reproducibility resulting from sample contamination and thermal convection inside the cell, and the investigation of a saline matrix effect in the thermal lensing determination of phosphorous - including results from solvent extraction of the saline solution. ghsrsstsrissties_ef-ths_§ess£resstsr When any new device is constructed, preliminary measurements must be made to optimize the experimental parameters with respect to such factors as signal-to-noise ratio, acquisition time, data processing constraints, etc. There are three parameters under computer control which 65 affect the afore-mentioned considerations: l)time delay between data acquisitions within a transient, 2) number of transients averaged in an experiment, and 3) time delay between transients ("dead time”). With the instrument designed for this work, acquisition times could be varied continuously between 250 pa and 32 me; a convenient upper limit is 1 ms. Figure 15 shows identical transients with time delays of 250, 500, and 1000 us respectively. A value of 1000 us was chosen since this provides a longer sampling time per equivalent number of points. Storage requirements of the fitting routine DATFIT allow a maximum of 300 points. Thus, with a 1 me time delay a transient 300 me long can be fit. This is adequate to fit the inflection region of the intensity accurately for such common solvents as water which has a tc between 50-60 ms (dependent on the beam size). The number of transients averaged in one experiment affects the signal-to-noise ratio, which improves as (N)1/3. This is demonstrated in Figure 16. An upper limit of 127 transient acquisitions per experiment was fixed to prevent the possibility of integer overflow in FORTH data storage. A value of 100 transient acquisitions per experiment proved optimal; the slight increase in S/N did not justify the further increase in experimentation time. The time between acquisitions (”dead time”) can be set from one second to hundreds of seconds (the latter being quite unfeasible with respect to experiment time). Tests 66 ( A r 5 U n it s ) l n t e n s it y K ?. .‘:. 2. x. .35 I- .0,‘ ' 1. ”is. I. akKiNnéqmfinflwfindnewwwducdquflflfi’ use» p" "'3" e :‘3 . ..“2 0e o e’. e 5A“. . g e .~.«we:~a-xa.aeasxvn<~p*e’nréw~ae 500» o“. ... .. «.... ... .. ...... .... n 0e 1000” a 4 L a a L 1 7 j l ' r If I I 1 V v ' j 0 128 T irn e (In 3 ) Figure 15. Thermal lensing transients acquired with differing time delays (5.0x10'9M Crystal Violet in methanol). 67 Units) e . s . ~.:’=O.‘.‘. . e O 3...~. . ... . e .. f 0' " e .. o‘eue ' 5.. . \ e. :00. ' e 0. 0 ’ ' g .. _....: g a”... '. e ': ‘ee s.‘ ...o’.& . ....) ... ~01 ... .e ;.\"e .Qetéie ‘;’ 0‘. ... .’.‘ .‘f.’ 0.. ..."... i :0 s~ “v‘ 0 ..s s I . . s . e . fl . . 0 e i . ~ . ' ' e m .‘ ‘eO‘ .... 0 ' 2“. :“f ; ...:dO: ...? e'Ifl‘ "" .~:'.e: 0 "-1 s ‘0 . .... . .0 .0. . '0‘. ' .fl ’ e"’..sw.". ‘2’ kg . f . :- .....H 0. : f .e w. .. ° 0 . 3" e e ...a t '“M’vf-f: 3.9: Wmetmmbfih‘ifflfirk’r'f "‘5 "ee ""95 s . ..'*W s ~a'e “ . . O ‘ -. , MWmnmwwWehknfi-MW‘ n-20 (Arb '3' y {A v ‘ #u- .m- -0 .w -.. “a " 'r ‘vw- v‘v‘va—W s t 17 l A" ..., “he - A _ A- ~ n-‘m . ... A AA; . ' ‘v ‘ u A AA‘ A ... vw w, 'vw- w—w W W ‘V"“' T" «(l- I as ntens l ... 3 0 3 U) Figure 16. Noise level versus number of scans averaged in one experiment (n= number of scans averaged). were performed on a typical phosphate solution to determine whether a change in dead time would alter the magnitude of 8 in case the thermal lens was not given adequate time to ”relax"; this could result in ”memory effects” from one transient to another. However, no statistically discernible difference in 0 was observed; the lens seemed to ”relax" within 1 second. Therefore, to reduce experimentation time, a dead time of 1 second was used throughout (except where noted). Another characteristic which is of major importance in any spectrometric observation is the linear dynamic range of the spectrometer. One example of such a test is shown in Figure 17. This test was performed with a 50 mW incident beam at 590 nm (R6G dye laser) on solutions containing accurately-known concentrations of the dye Crystal Violet in methanol. Multiple measurements at each concentration were performed, and le bars are plotted for each concentration. There is a distinct bend-off in the response due to a partial breakdown of the simplifying assumptions, which occurs at strong absorption levels. The size of the dynamic range and the ultimate detection limit will vary with specific analyte/solvent pairs due to the concomitant changes in s, k, and dn/dT. In the case diagrammed in Figure 17, a detection limit (2tse: 95x confidence level) of approximately 2x10“1°M was calculated. Checks were also performed on the spectrometer system in order to determine whether: 1) 0 was linear with 0.8 0.6 ‘b 0.4 0.2 0.0 Figure ._ I A... —>- 4+- AJAIJJIIIJILAI 11]] T I T T I l 1 I F I U I Y 1 17 r fl T 1 0.0 2.0 4.0 6.0 8.0 10.0 Concentration [M] x 109 17. Dynamic range check of the thermal lemsimg spectrometer with crystal violet solutions. 70 incident laser power and, 2) 0 was independent of the gain setting of the programmable gain amplifier (at constant laser power). Within the uncertainty of the measurements, both of these requirements were met. Attesting to the optical sensitivity of the spectrometer was the detection of oscillatory intensity fluctuations resulting from air convectively heated by the shutter fine as they blocked the laser beam. The fluctuations were eliminated by circulating cooling air directly on the shutter via a small fan. Following completion of these preliminary characterization tests, an investigation of the detection of phosphorous was initiated. Isiiisl-lezestissiiess The initial laser system used in these studies was an Art ion laser-pumped dye laser operating with Rhodamine 6G dye. Three separate phosphorous concentrations (0 pptr, 100 pptr, and 1000 pptr) underwent reactions for development of the reduced 12-MPA species. However, no statistically significant difference in 0 was found among any of the solutions. It was thought that: l) the reduction might not have been complete and 2) the ratio of analyte absorption to reagent (background) absorption was not favorable for ultratrace analysis. Therefore, the investigation turned to the antimonyl complex of reduced l2-MPA. 71 Initial results with lZ-MSbPA did prove hopeful in that a noticeable increase in 0 was observed as the phosphorous concentration was increased. However, with the dye laser system power loss during an experiment was common (for the reasons mentioned in the previous chapter). Since the linear relationship between 0 and incident laser power was well established, scaling of the 0 value with respect to the fit I(o) value was performed during data analysis. Scaling proved quite useful for situations in which there was a significant change in observed intensity (>152); quite reasonable results were obtained for calibration curves through scaling, as is shown in Figure 18. However, when relatively small changes in the observed intensity were noted (3-53 within a single concentration run), scaling seemed to be erroneous; it was frequently found that 0 values did not correlate well with the relative intensity (higher intensity values had lower 0 values). In these cases the correlation decreased upon normalization. Normalization was thus not a solution to the problem of power fluctuation. It was determined that the Er’ ion laser would be a better source because: 1) the output is in the red region of the visible spectrum (A = 647.1 nm or 676.4 am) .where major absorption of the reduced lZ-MSbPA species lies, and 2) the output can be servo-regulated such that the power incident on the sample remains constant with time. When the analyte samples were probed with the Er’ ion laser the large changes in intensity (>152) were eliminated, 72 0.130 -r- 0.120 0.110 A l l l L l l l I can I . r* , r , ri r,_j o to) an an an an Concentration P (pptr) Figure 18. Working curve for l2-MSbPA solutions with data scaling. 73 but the small changes in intensity (~52) still remained, along with the concomitant fluctuations in 0 and tc. Table 5 shows results of the fit for one run of multiple measurements on a 500 pptr phosphorous solution (each experiment is an average of 100 transients). Table 5. Typical values of I, tc, and I(o) obtained for one concentration of phosphorous analyte. 9 islesl Iiel-ié£9-seiisl 0.13211.0004 52.9to.7 2863.3to.8 0.12891.0004 51.6to.7 2910.8t0.8 0.12701.0004 49.9:o.6 2926.610.7 0.13231.ooos 49.8tO.5 2888.610.6 o.1311:.ooos 52.6io.6 2854.3to.6 o.134at.ooos 52.5to.7 2850.3i0.9 0.13611.0004 53.5:o.7 2884.2io.8 o.1377t.ooos 52.7t0.6 2924.6:o.7 0.13902.0004 55.1:o.7 2955.lto.8 0.1376i.0005 55.3to.9 2976.4tl.l The uncertainties listed are 1‘ values calculated by the fitting routine. The data suggest that the fluctuations in 0, tc and 1(0) are statistically significant. There also seems to be a periodicity to the fluctations. Checks were made of the Hr* ion laser power stability by recording multiple exposures of laser light and averaging (with statistics). Typically, the standard deviations of recorded 74 power were less than 12. Therefore, another mechanism must be responsibile for the observed fluctuations. This same effect was also observed by Buffett and Morris". They determined that it was due to convective stirring of the solution which resulted in oscillatory signal fluctuations, with periods on the order of minutes. Attempts were therefore made to break up the periodicity of the currents, or at least lessen their magnitude. First the cell was ”baffled” with glass rods dispersed in a random fashion. This was done in hopes of creating a turbulent break-up of the convective' currents. Unfortunately, no evidence of increased precision in the average of multiple determinations of a single sample was observed with this configuration. More closely-spaced baffling (in the form of glass wool) was tried. The glass wool was inserted into the cell and a space 3.to 4 mm in diameter was formed in order to allow the laser beam to pass through the cell. This also did not increase the precision of the average. Since the convective currents did not seem to be diminished by baffling their removal by stirring was investigated. The "dead time" of the experiment was increased from 1 second to 4 seconds. During this time, a magnetic stirring bar was activated either through the use of a magnetic stirring motor (which produced rotary motion of the bar) or by manually manipulating an external bar magnet (thus producing linear up-and-down motion of the stirring bar). The succeeding transient was acquired and 75 the stirring was repeated during the next "dead time". Particulate matter in the solution was noted through "flashes" as the particles scattered the laser light, causing large fluctuations in intensity. The sample (in this case tests were performed on distilled, deionized water) was filtered through 0.45 pm Nylon 66 filters to remove the majority of the particulate matter. The stirring experiment was repeated on the filtered sample. The results are shown in Figure 19; they are neither reproducible nor monotonic. The reason for the distortion of the transients is as follows: the thermal convection described by Equation 8 is radial and if this is disturbed the signal will be drastically affected. The turbulence from stirring had not subsided by the time the next transient acquisition was to be made. To give the stirred sample enough time for the turbulence to subside sufficiently would make the experiment prohibitively long. It has been suggested°° that the convective phenomenon is exacerbated by the presence of particulate matter, notwithstanding the previously-mentioned scattering effects. Therefore, the analyte solutions were filtered through 0.45 um Nylon filters prior to analysis. The results are shown in Table 6. From Table 6 it is apparent that, in addition to particulate matter, the analyte is retained on the filter, and does not pass through with the rest of the solution. 76 (a ‘1 2700 -_ c 1 :3 o \ < )4 n c 0 c O 100 200 300 — T i m e ( m s ) Figure 19. Thermal lems time profiles upon sample stirring. 77 Table 6. Results of filtering analyte solutions §ssels-dsssriztiee Eleverase of 7 scans) Unfiltered 100 pptr P 0.124t.002 Unfiltered 200 pptr P 0.1251.004 Unfiltered 500 pptr P 0.1361.004 Unfiltered 700 pptr P 0.1361.004 Unfiltered 1000 pptr P 0.147t.004 Filtered 100 pptr P 0.1163.006 Filtered 200 pptr P 0.114i.004 Filtered 500 pptr P 0.116t.005 Filtered 700 pptr P 0.1121.004 Filtered 1000 pptr P 0.lllt.007 Further substantiation of this conclusion was the observation of a light blue tinge given to the white filters by the reduced 12-MSbPA. It is not clear whether l2-MSbPA is filtered by or is adsorbing to the filter. Estesiiee-Lisits The question of the true detection limit of thermal lensing measurements of 12-MSbPA must be addressed. How valid are the statistical standard deviation values determined' for the studies performed? What are the 78 systematic contributions of scaling, background absorption, convection currents, etc.? Scaling should not have been performed since the fluctuations did not reflect true fluctuations in the incident laser power. Moreover, the problems with thermal convection are not reflected in the statistical error estimates. Realistic detection limits for the 12-MSbPA species in aqueous solution were therefore reinvestigated in light of these factors. The Er+ ion laser, operating in light-regulated mode, was employed as the pump source. Solutions of 12-MSbPA were prepared as before; they were neither filtered nor stirred. Typical results are diagrammed in Figure 20. A detection limit of approximately 750 pptr phosphorous (95x confidence limit) was determined for this system. This limit is unfortunately poorer than previously thought; the detection limit calculated for the data in Figure 18 was approximately 200 pptr. In comparision, Fujiwara, et al.53, using the same basic colorimetric procedure, calculated a detection limit of approximately 70 pptr phosphorous in aqueous solution. However, a recent study by Nakanishi, et al.53 questioned those results, since it was determined that background absorption due to the reagents and the solvent was equivalent to the absorption of several ppb of analyte. This implied that the earlier workers could detect one one-thousandth of the overall signal. Nakanishi, et al. performed thermal lensing on 12-MPA aqueous solutions using 79 0.14 0.13 0.10 L l l L l L L L A 1 J l l l l l L l A I I’ r '— l I I U I j I V I j T T j r T I I 1 00 200 300 400 500 000 700 000 900 1 000 Concentration P (pptr) Figure 20. Working curve for lZ-MSbPA solutions without data scaling. 80 a 10 n” seniconductor laser with output corresponding to the absorption naxinun (A: 820 us). The detection linit found in these studies was 2200 pptr phosphorous. If corrections are nade to this linit to account for the power and wavelength change, their results suggest detection linits at 647.1 nn of approxinately 350 pptr phosphorous. That extrapolation is consistent with the detection linit found in this study. Egaseaigsgs-§££essg Due to the high sensitivity of a technique such as thernal lensing, contaninants can have an innense effect on the nagnitude of the signal observed. A contaninant can itself exist at trace levels, preventing its direct observation. The contanination can occur at any point in the analytical deternination: preparation and nixing of reagents, chronogenic reaction, or during the transfer of solution into the cell. The last case probably accounts for the singular point in the results illustrated in Figure 21. One particular calibration run of aqueous lZ-MSbPA had a wildly variant value of 0 at 700 pptr phosphorous (point a). The sanple was replaced with fresh solution taken fron the sane volunetric flask reservoir, giving a value nuch nore consistent with the predicted response (point b). Sonewhere in the transfer the first solution in the cell becane 81 0.25 ‘1 I 4 1 L Concentration P (pptr) Figure 21. lorking curve for lZ-lePA solutions with arronaons data point due to contanination. 82 contaninated. This exenplifies the stringent cleanliness requirenents placed on the laboratory environnent as well as on the analytical procedures when ultratrace analysis is to be perforned. §9l129-!ss£15-§££sss The effect of the saline natrix of connon environnental sanples on the thernal lensing neasurenents was investigated. Several identical solutions each containing 400 pptr phosphorous were prepared as before, the color was developed, and varying anounts of 1 M NaCl were then added prior to dilution to create final solutions of 0.0, 0.01. 0.1 and 0.5 M NaCl, respectively. Initial experinents were perforned with the dye laser systen (A: 620 nn, power= 70 a“). After fitting, 0 values at each NaCl concentration were referenced to the 0.0 M NaCl value. The results are presented in Table 7. Table 7. Results of the relative signal change in a 400 pptr phosphorous solution with a change in the background Incl concentration. fieanls zele£ixs-si 0.0 M NaCl 100.013.5 0.01 M NaCl 100.815.3 0.1 M NaCl 111.914.9 0.5 M NaCl 120.015.o ' relative to 0.0 M NaCl 83 The results suggest a definite enhancenent in the thernal lens upon the addition of NaCl to control the ionic strength. The salinity of the world’s oceans is approxinately equivalent to that of a 0.5 M NaCl solution. Thus the analysis of Fujiwara, et al.52 is highly suspect since no apparent account was nade for the ionic strength change between standard solutions and the analyzed seawater sanples. As a check for possible Na+ or Cl‘ interference in the chronogenic reaction, solutions of 400 pptr phosphorous were again prepared, and NaCl was added either before or after color developnent; results are listed in Table 8. Table 8. Test of interference of Incl on the chronogenic reaction of phosphorous. §saele 9 0.1 M added before color 0.1051.004 developnent 0.1 M added after color 0.1091.003 developnent , 0.5 M added before color 0.114i.004 developnent 0.5 M added after color 0.1191.004 developnent Since the 0 value for a given NaCl concentration is statistically invariant,there is no discernible interference of Na’ or 01' on the chronogenic reaction within the tine frane of the neasurenent. 84 Another possible explanation for the thernal lens enhancenent is a change in the bulk refractive index which night alter the absorption coefficient of the reduced lZ-MSbPA species. A change in nolar absorptivity with the refractive index of the solvent was proposed nany years agol°°, and a correction factor to the nolar absorptivity , n /(n2 + 2)2, was given by the authors. Fron this factor and the accepted values for the refractive index of water (1.333) and a 0.5 M NaCl aqueous solution (l.338)1°1, a relative change in the nolar absorptivity of 0.333 was calculated. Since this is two orders of nagnitude snaller than the observed enhancenents, the refractive index change contributes insignificantly to the saline natrix effect. In the expression for 0 [0: 0.24 P0£(dn/dT)/xk], the only other paranaters which can vary are dn/dT and k. Values of dn/dT for pure solvents are abundantzz; however, for electrolyte solutions there is a paucity of reliable data. Recently, dn/dT values for KBr solutions were deternined by an interference refractoneterlOz. Due to the sinilarity of the electrolyte solutions, extrapolated values of dn/dT should reasonably approxinate NaCl solution data. Since the naJor constituent of seawater is NaCl, tabulated values of thernal conductivity values (k) for seawater”3 will also reasonably approxinate NaCl solutions. The extrapolated values for k and dn/dT (at 25° C) and the calculated enhancenent in 0 are listed in Table 9. 85 The calculated values in Table 9 are graphically conpared to the observed values of Table 7 in Figure 22. The experinental values are consistently higher than the predicted values. However, due to the approxinations nade, this is not surprising. The nagnitudes of the calculated and observed enhancenents are approxinately equal, and it is apparent that the variation in the therno-optical paranaters is the prinary reason for the change in 0 between aqueous and NaCl solutions. Table 9. Calculated signal enhancenent for various laCl solutions (referenced to 0.0 M Incl) [Hegllinl kical 8“c-“K“l' 921 TLK’11” 5-218921 0.0 14.60x10"I 1.135x10" 100.0 0.01 14.60x10" 1.137x10“ 100.2 0.1 14.53x10“ 1.160x10" 102.7 0.5 14.48x10" 1.260x10" 112.0 ‘ extrapolated fron data in reference 103 b extrapolated fron data in reference 100 It was thought that the saline natrix night also have a neasurable effect on the characteristic tine constant of the systen, tc. Fron equation (14), tc= ech/4k. Calculations were perforned for solutions of 0.0, 0.01, 0.1 and 0.5 M NaCl concentrations, using tabulated values of heat capacity (c; cal g‘1K’1) for seawater”1 as an approxinate value for NaCl solutions. The p values (g cn‘3) were obtained 86 130.0 '1!- 120.0 *1- 110.0 Signal Enhancement (Z) 100.0 90.0 «+— .L L.— -r—- 0.0 [NaCl] (M) figure 22. Observed enhancenents (std. dev. bars) and calculated enhancenents (dianonds) in the thernal lens signal as a function of saline concentration. 87 fron the CRC Handbook101 and the values for k (cal s‘lcn‘lk'l) were as before103. These paraneters, along with a bean radius of 170 un (the approxinate radius of the bean in these experinents), were used to calculate the values of tc (ns) given in Table 10. Within the known experinental uncertainties in the neasurenent of tc these differences are insignificant and therefore no further investigation of this paraneter was attenpted. Table 10. calculated values of t: for selected laOl on concentrations. 1219911111). 9' 2” l.“ is 0.0 1.0000 1.0000 14.60x10“ 49.5 0.01 1.0000 1.0007 14.60x10“ 49.5 0.5 0.9910 1.0043 14.53x10“ 49.5 0.5 0.9607 1.0207 14.48x10" 48.9 ' extrapolated fron data in reference 104 U fron reference 101 c extrapolated fron data in reference 103 Since to the Solvent Extraction saline natrix does the response of the systen, have a noticeable effect the next course of action was attenpt to separate the analyte fron the natrix. Solvent extraction was enployed to separate the analyte fron the saline solution natrix. Much literature exists on the subject of extracting either lZ-MPA or its antinonyl analog fron aqueous solution with oxygenated organic solvents". Isobutyl acetate or propylene carbonate can be used, _ however, the nost cannon solvent is 2-nethyl-l-propanol, also known as isobutyl alcohol (IBA)°5'°7. For this reason, IBA was chosen for these extraction studies. Phosphate solutions ranging fron 0 to 1000 pptr phosphorous were forned in both distilled, deionised water and 0.5 M NaCl solution (the latter being chosen because it closely resenbles the ionic strength of seawater and gives the nost noticeable enhancenent). Both series of solutions (50 nL) were extracted with IRA (25 nL) and shaken for two ninutes. The extract was then diluted to 50 nL. The aqueous solutions were analyzed first, followed by the extracted solutions to allow: 1) adjustnent to the lower laser power needed for the IDA solvent since its enhancenent factor is one order of nagnitude larger than that of water and 2) adjustnent of the nirror assenblies to allow the bean center to fall on the pinhole (there is a bulk refractive index change). The results are enunerated in Table 11. With the exception of a few anonalies (for reasons nentioned previously), each series of experinents gave a signal response that varied linearly with concentration. The results for each series in Table 11 are and plotted in Figure 23 (after blank subtraction) in order to show 89 Table 11. blank aqueous solution 100 pptr P aqueous soln. 200 pptr P aqueous soln. 500 pptr P aqueous soln. 700 pptr P aqueous soln. 1000 pptr P aqueous soln. extract of blank aq. soln. 100 soln. extract of pptr P, aq. 200 soln. extract of pptr P, sq. 500 soln. extract of pptr P, aq. 700 soln. extract of pptr P, aq. 1000 soln. extract of pptr P, aq. 0.136i.003 0.136t.006 0.138i.006 0.142t.005 0.1561.008 0.156i.010 0.5821.008 0.5911.012 0.5501.009 0.654t.019 0.6861.013 0.735t.019 90 blank 0.5 Results of IDA solvent extraction. M 0.174t.007 NaCl solution 100 pptr P in 0.1661.006 NaCl solution 200 pptr P in 0.177t.010 NaCl solution 500 pptr P in 0.2151.005 NaCl solution 700 pptr P in 0.2081.002 NaCl solution 1000 pptr P in 0.208i.009 NaCl solution extract of 0.3001.020 blank 0.5 M NaCl extract of 100 0.3l3t.003 pptr P,0.5 M NaCl extract of 200 0.319t.013 pptr P,0.5 M NaCl extract of 500 0.339t.008 pptr P,0.5 M NaCl extract of 700 0.309t.014 pptr P,0.5 M NaCl extract of 1000 0.364t.007 pptr P,0.5 M NaCl 0.00 0.10 0.05 0.00 0.10 0.05 0.00 0.05 extract. aq. SOln. extract. saline soln. saline soln. 0 1 L A r I I I l I I A aqueous soln. 0 100 200 500 400 500 600 700 800 900 1000 Concentration P (pptr) Figure 23. Conparison of thernal lens signal response for lz-MSbPA in: aqueous solution, 0.5 M NaCl solution, IIA extract of 0.5 M laCl solution, IDA extract of aqueous solution. 91 ‘0 relative enhancenents as reflected by a change in slope of the constrained least squares line (for clarity the error bars have been renoved and data which vary by nore than 2‘ fron the least squares line have been edited). The najor observations fron this portion of the study are as follows: 1) the saline effect is definitely real and inportant [an average enhancenent of 28158 was calculated between corresponding concentrations of aqueous and saline solutions; taking power and wavelength differences into account, this result agrees exactly with the value given in Table 7]; 2) an enhancenent in response over that of the parent aqueous solution is observed upon extraction into IBA; 3) nost inportantly, solvent extraction does not free this analysis fron the effects of a saline natrix effect. Large changes in 0 are observed for corresponding concentrations between extracts fron aqueous solutions and extracts fron saline solutions and the slopes of the working curves are also affected. The ionic strength of the solution apparently affects the efficiency of the extraction. Three separate 500 pptr phosphorous solutions were prepared in solutions of 0.1, 0.5, and 0.8 M NaCl respectively. Each solution was extracted with IRA .as before; the extracts were then diluted and analysed. The results of this experinent are given in Table 12. Another experinent involved extraction of a 0.5 M NaCl solution with IRA, dilution of the extract to 50 IL, 92 placenent of 10 nL into a previously-weighed drying bottle and volatilisatian of the solvent. Table 12. Results of IDA extraction frou solutions of varying Incl concentration. §22212 9 500 pptr P in 0.1 M NaCl 0.5051.004 500 pptr P in 0.5 M NaCl 0.344i.006 500 pptr P in 0.8 M NaCl 0.3121.005 Visual inspection of the bottle revealed a white coating, obviously NaCl; the bottle was reweighed and by subtraction the solubilized anount was deternined. A solubility of 0.009 M NaCl in IBA was calculated. This value is 270,000 tines larger than the naxinun possible concentration of the lZ-MSbPA species in IBA. Fran these experinents it is clear that the activity of the highly charged lZ-MSbPA species is altered by the change in ionic strength. The higher ionic strength of the aqueous phase stabilizes these species, as predicted by the Debye-Ruckel theory. In addition, the salt which is extracted partially excludes the lZ-MSbPA species, resulting in a lowered activity. Therefore, solvent extraction cannot be used in the analytical procedure for the deternination of phosphorous in environnental sanples by thernal lensing spectronetry. The experinenter is relegated to prepare a working curve in a natrix approxinating the environnental sanple or alternatively the nethod of standard 93 additions could be perforned on the environnental sanple. 92921921292 The initial purpose of this investigation was to deternine possible linitations to thernal lensing analysis introduced by a foreign natrix. However, in the course of this work nany other linitations also cane to light. Background absorption, thernal convection, inpurities, environnental natrices- all of these sources lead to a reduction in the precision of ultratrace analysis by thernal lensing spectronetry. An application such as HPLC detection would appear to be sore favorable for the thernal lensing technique since ultrapure solvents with low background absorbances can be utilized. This would allow one to exploit the high sensitivity of the thernal lensing technique. The experinental conditions for thernal lensing nust also be inpraved with respect to thernal fluctuations before practical detection linits can be reasonably lowered. Colorinetric nethods should be reinvestigated, not only to optinize the nolar absorptivity of the colored conplex, but also to nininize the anount of background absorption associated with the reagents used. If these problens can be overcone, thernal lensing nay becane a viable practical technique in the future. 94 V. SUGGESTIONS FOR FURTHER INVESTIGATIONS 222921129-21-I929921-119219211292 The nost pressing problen is the thernal fluctuations set up in the call; these invariably reduce the precision of the thernal lensing technique. The use of cells with very snall dinensions night prove helpful since the convection currents seen to occur over very snall distances (at least in aqueous solution). Another inpravenent night result fron continual circulation of fresh analyte into the cell during the ”off" tine of the laser; this will lengthen the experinent since allowance nust be nade for the relaxation of turbulence created by the circulation. This extra tine nust be considered when evaluating and conparing thernal lensing to other techniques. Recently, Jansen and Earris1°5 have developed a nethod for reducing the spatial noise in thernal lens nethods through the use of a parabolic transnission nask. Essentially the purpose of this sank in to nultiply the transnitted intensity by r3, resulting in the neasurenent of the second spatial nonent (spot size) at the detector. Since nost of the transnitted bean is now 95 detected (a lens placed after the transnission nask focuses the light passed by the nask onto a detector), any spatial noise is averaged. This is in contrast to the earlier work of Miyaishi, et al.5°, who neasured the evolving been with a linear array detector and fit the data to a theoretical bean profile. This proceedure essentially calculated the first spatial nonent (linear slice) of the bean. Although this technique also averages spatial noise, the fitting is done nathenatically, whereas averaging perforned with the nask is done optically (and in two dinensionsl). The ability to average spatial noise instantly reduces the effect of optical inhonogeneities, including those of thernal fluctuations. In addition, this night also significantly reduce spatial noise associated with the tunable CW dye laser, naking it a viable laser source. For these reasons, the transnission nask technique should be investigated. £92221929221-!211221129-§2221922229! The apparent failure of filtering the phosphorous solutions suggests another possible technique for the detection of phosphorous. Photothernal deflection spectronetry is a technique designed to neasure absorption of species located on surfacess. This is done through detection of a linear refractive index gradient created above a surface which has partially absorbed laser radiation. Since the reduced lZ-MSbPA trapped in the white 96 filter is colored, it night be feasible to perforn phothernal deflection experinents on the filter surface. In this situation, the analyte would be preconcentrated, and lower detection linits night be attainable. D921-9299-I222912922 Another avenue which can be explored is the use of two laser beans in the thernal lensing experinent. Nornally, one nore powerful bean is used to forn the thernal gradient in the sanple and a nuch weaker bean probes the gradient. The punp bean can be intensity-nodulated and the changes inparted to the probe bean can be detected through sensitive lock-in techniques, such as second harnonic detection°°. A problen associated with this technique is the change of the relative overlap of the two beans which occurs gradually. To alleviate this, a single laser/dual bean thernal lensing spectroneter has been devised105 in which the punp and probe beans have been given opposite polarization. By using a polarizer, discrinination betweeen these two beans is possible and thus the problen of decreased overlap is solved. These dual bean/lock-in techniques can effectively reject laser shot noise, although any optical noise enanating fron the sanple cannot be conpensated for with this technique. However, a conbination of the transnission nask and the dual bean technique night give quite good results. 97 APPENDIX APPENDIX. FORTH SOFTWARE1°7 Block Nunber: 120 0 ( THERMAL LENSING ROUTINES--VARIABLES) ’ CR ’EXP ! 1 2 ( LOAD GENERIC VARIABLES AND CONSTANTS) 40 LOAD 3 ( TL SPECIFIC CONSTANTS AND VARIABLES) 121 LOAD 4 ( GRAPHIC VARIABLES) 190 191 THRU 5 ( NUMBER DISPLAY) 41 42 THRU 6 ( TL STUFF) 122 LOAD 7 ( LOAD DATA ACQUISITION ROUTINES) 48 50 THRU 52 55 THRU 8 ( TL STUFF) 123 126 THRU 9 ( LOAD GRAPHIC SOFTWARE) 192 199 THRU 10 ( TL STUFF) 127 137 THRU ll l2 l3 14 15 Block Nunber: 121 O ( THERMAL LENSING ROUTINES--VARIABLES) l 2 2 CONSTANT MULTIPLIER 3 2 CONSTANT (EXP) 4 VARIABLE P/S 5 VARIABLE TBPS 6 VARIABLE CSCAN 7 VARIABLE S/E VARIABLE LFREO 8‘ VARIABLE DT VARIABLE CTBPS 9 VARIABLE CURR MULTIPLIER 512 # ALLOT 10 VARIABLE SUMMR MULTIPLIER 1024 * ALLOT ll VARIABLE DEVR MULTIPLIER 1024 * ALLOT 13 72 ’ lST-DATA-OFFSET ! 98 Block Nunber: com-amoubwlon—o ( THERMAL LENSING ROUTINES--PRINTING PRIMITIVES) ’.DATA-S 1000 122 IZE P/S O 8 1024 t/ O ; .DATA-SIZE 62 07 22X SNAP OVER DABS R) OVER - SPACES Block Nunber: ( THERMAL LENSING--DISPLAY ROUTINES) PARAM-DISPLAY CDQQQOIfiQONr-IO 23 18 66 66 64 CREATE R) 3 + 12 P/S l4 TBPS l6 DT 12 S/E 123 , C, C, DOES CO SNAP 22X PARAM-DISPLAV PARAM-DISPLAV PARAM-DISPLAV PARAM-DISPLAV 13 CSCAN PARAM-DISPLAY l6 LFREO PARAM-DISPLAY .GAIN .COMMENT 13 18 22X? GAI 13 07 22X? ..DATE TODAY 0 63 06 .SPECTRA .CSCAN 12 06 !2XY # 36 + 2: Y ’.DATA-SIZE 06 )R ( 46 HOLD 8 SIGN TYPE .” Ebytes” ; ) )R I 0 O I 2+ 00 Y 04 RU.R ; .P/S .TBPS .DT .S/E ’.CSCAN .LFREG N C0 I)G 04 RU.R ; ."COMMENT 34 TYPE SPACE ; !ZXY SPACE .DATB ; .”SPECTRA 9 TYPE GPhP NOT IF ’.CSCAN THEN ; ) Block Nunber: 124 ~ DQQQOIQWND-‘O H O 11 Hour-n— 0|:th ( THERMAL LENSING ROUTINEsr-PROMPT WORDS) ?PROMPT ( NEEDS X-Y POSITION AND ADDRESS OF 32 BIT) CREATE , C, C, DOES) )R I G I 2+ CO R) 3 + C0 SNAP ?CHANGED IF SNAP ! .DATA-SIZE ELSE DROP THEN ; 23 14 TBPS ?PROMPT ’?TBPS 18 16 DT ?PROMPT ?DT 66 12 S/E ?PROMPT ?S/E 64 16 LFREO ?PROMPT ?LFREO ?TBPS ’?TBPS TBPS DUP 0 250 MAX SNAP ! ; ?P/S 23 12 ?CHANGED IF 00 MAX 256 MULTIPLIER * MIN DUP P/S ! T P ! .DATA-SIZE THEN ; ?GAIN 13 18 ?CHANGED IF 00 MAX 1024 MIN G)I DUP (PGA) C! GAIN C! THEN ; Block Nunber: 126 CDQQQO-thI-‘O ( THERMAL LENSING ROUTINES--.PARAM AND CHANGES) CREATE CHANGE-LOOP ’ ?P/S , ’ .P/S , ’ ?TBPS , ’ .TBPS , ’ ?DT , ’ .DT , ’ ?GAIN , ’ .GAIN , ’ ?S/E , ’ .S/E , ’ ?LFHEO , ’ .LFREG , 47 LOAD .PARAM .SPECTRA ..DATE .COMMENT .DATA-SIZE 00 05 .PARAMS 3 ?.VALUE 01 MAN 06 MIN 1- 2* 2* CHANGE-LOOP + DUP 0 EXECUTE 2+ 0 EXECUTE ; ' ?.VALUE8 SWAP DO I ?.VALUE LOOP ; 100 Block Nunber: 127 0 ( THERMAL LENSING ROUTINES--?MAX-GAIN) 1 2 CSHUTTER AMASE CO 00 AND AOUT ; 3 4 . OSHUTTER AMASH C0 01 OR AOUT ; 5 EXIT 6 ?MAX-GAIN ( FINDS MAX GAIN FOR THERMAL LENS EXP) 7 ( 2 DUMMY VALUES) 99 01 15 0 DO I (PGA) C! 8 I GAIN ! .GAIN OSHUTTER (A12) DUP 4095 ( IF 9 I ZSNAP 2DROP ELSE DROP LEAVE THEN 10 CSHUTTER LOOP DUP (PGA) C! GAIN ! .GAIN DROP 3 ll 12 13 14 15 Block Nunber: 128 ( THERMAL LENSING ROUTINES--DATAOFF) .DON ( DISPLAY ON) CMND CO 02 OR CMND C! 15 20 !2XY .” [" .RV” ON” .” ,OFF]" ; .DOFF ( DISPLAY OFF) CMND C0 13 AND CMND C! 15 20 !ZXY .” [ON,” .RV" OFF” .” I” ; DATAOFF 53 20 !ZXY .” [Y," .RV” N" .” ]" CMND CO 01 AND IF D>D lST-BLOCE O BLOCK UPDATE )R 10 ?NEXT-BLOCE l- I 1+ C! CSCAN O R) 54 + ! 11 DIRECTORY-UPDATE FLUSH THEN GPhP NOT II 12 66 13 !ZXY 04 SPACES THEN CMND CO 06 AND CMND C! ; OQQQWQNNHO 101 Block Nunber: 129 ( THERMAL LENSING ROUTINES--DATAON AND SAVE-PARAM) mmqmmeuwwo U ... O O 3' GDQQGOIIFWNHO SAVE-PARAM ?NEXT-BLOCH DUP BUFFER )R 250 + IDENTIFY UPDATE I 1024 ERASE I C! ."SPECTRA I 2+ 09 (CMOVE ."COMMENT I 12 + 32 (CMOVE S/E O I 54 + ! TBPS O I 48 + ! P/S O I 50 + ! DT G I 52 + ! GAIN O I 46 + ! LFREO O I 56 + ! ODATE (DATE) I 60 + SNAP (CMOVE R) lST-DATA-OFFSET + (DATA) ! lST-DATA-OFFSET 2/ POINTS ! ?NEXT-BLOCE IST-BLOCE ! DATAON ?DATA-DISE ?ENOUGH-SPACE CMND CO 08 OR CMND C! 53 20 !ZXY .” [" .RV" Y” n ,N]” SAVE-PARAM ; Nunber: 130 ( THERMAL LENSING ROUTINES--HD) .SPACEShU.R SPACES 04 U.R CR ; 60 MIN OFFSET G 250 OFFSET ! SNAP l BLOCE 48 + SNAP 16 t + 10 + CO BLOCK )R CR ." NAME:” 13 SPACES I 2+ 09 O O O O O O O O 0 TYPE CR ." COMMENTS:” 9 SPACES I 12 + 32 TYPE CR GAIN" I 46 + C0 I)G 13 .SPACES&U.R TIME BTN POINTS” I 48 + O 02 .SPACEShU.R POINTS/ SCAN” I 50 + O 05 .SPACEShU.R DEAD TIME” I 52 + O 08 .SPACEShU.R LASER FREQ" 01 SPACES I 56 + O 10 U.R CR NUMBER OF SCANS:” I 54 + O 01 .SPACEShU.R DATE:” 13 SPACES I 60 + 8 TYPE CR TOTAL OF: " I 1+ CO R) CO - 1+ 07 SPACES 04 U.R BLOCKS" CR OFFSET ! ; 102 Block Nunber: 131 0 ( THERMAL LENSING ROUTINES--EXPS) l 2 TAEE CURR P/S O 21 OVER + SWAP OSHUTTER 1250 US 3 DO GET-DATA I 2 CTBPS 6 US 02 +LOOP ; 4 5 ADD P/S O 0 DO I 2* CURR + O 00 I 4 t SUMMR + 6 0+! LOOP ; 7 8 SDEV P/S G 0 DO I 2* CURR + O DUP Mt I 4 # DEVR + 9 D+! LOOP ; 10 ll : AVG P/S C 0 DO I 4 # SUMMR + 20 12 POINTS 0 512 = IF D>D THEN 2A12 1A12 13 I 4 t DEVR + 20 14 POINTS 0 512 = IF D>D THEN !Al2 2A12 LOOP ; 15 Block Nunber: 132 0 ( THERMAL LENSING-- LENSE-PLOT) l 2 LENSE-PLOT P/S O 0 DO I 100 Mi 3 CURR I 2* + O PLOT-POINT 4 LOOP 20 PYL +2 20 PYH +2 ; 5 6 TLBEGIN SUMMR 1024 MULTIPLIER t ERASE 7 DEVR 1024 MULTIPLIER * ERASE 8 CURR 512 MULTIPLIER * ERASE ; 9 10 : TLEND COUNTER CMND CO 02 AND IF LENSE-PLOT THEN ll COUNTER SWAP - UT 0 1000 t SNAP - 0 MAX MS ; 12 13 l4 15 103 Block Nunber: 133 @QQQGhWNI-‘O ( THERMAL LENSING ROUTINES-- EXPS) ACCUM TLBEGIN TBPS 0 I76 - CTBPS ! S/E O 0 DO I 1+ CSCAN ! .CSCAN TAKE CSHUTTER ADD SDEV TLEND ?LEAVE-LOOP LOOP ; SETUP TLBEGIN S/E O 0 DO I 1+ CSCAN ! .CSCAN OSHUTTER P/S G 0 DO TBPS 0 121 - US LOOP CSHUTTER TLEND ?LEAVE-LOOP LOOP ; Block Nunber: 134 ‘DQQmele-IO ( THERMAL LENSING ROUTINES-- EXPS) EXPS CMND CO 08 AND IF CMND CO 01 OR CMND C! THEN CMND CO DUP 02 AND IF 100 PYL ! 425 PYH ! ?DISPLAY-ON THEN ACCUM 01 AND IF AVG THEN GP&P IF HEY-PROMPT 01 [’] ?MENU-UP C! EXP ELSE DATAOFF THEN ; 104 Block Nunber: ( THERMAL LENSING ROUTINES--.HEAD) meQO-h :wNo-IO Block @QQQGéQNt-IO 135 .HEAD 00 05 75 16 BOX 01 08 12X? 75 ZSOLIDS 01 09 !2XY 23 ZSOLIDS .” THERMAL LENSING SPECTROSCOPY 22 ZSOLIDS 01 10 !ZXY 75 ZSOLIDS 01 19 22X? 75 ZDASHS XZG 01 06 !ZXY .” EXPER NAME:” 42 SPACES ." DATE: ” 03 07 £28? ." COMMENTS:" 37 SPACES ." DATA SIZE:” 04 12 szxr ." 1. POINTS PER SCAN” 27 SPACES ." 5. SCANS PER EXP” 04 14 szxv .” 2. TIME BTN POINTS” 04 16 22X? .” 3. DEAD TIME” 33 SPACES .” 6. LASER FREO:" 04 18 !ZXY .” 4. GAIN” 04 20 SZXY .” DISPLAY” 35 SPACES ." RECORD DATA" 04 20 lZXY .DOFF DATAOFF ; Nunber: 136 ( THERMAL LENSING ROUTINES--1LOOP) ?O : 44 LOAD lLOOP DUP 68 ( D) = IF DROP CMND CO 02 AND 0: IF .DON ELSE .DOFF THEN ELSE DUP 82 ( R) = IF DROP CMND CO 08 AND IF DATAOFF ELSE DATAON THEN ELSE 3LOOP THEN THEN ; 105 Block Nunber: 137 madmmthI-IO ( THERMAL LENSING ROUTINES--EXP) COLD-START 1000 TBPS ! 256 MULTIPLIER * DUP P/S ! T P ! 100 S/E ! 4096 LYH ! O LYL ! OO GAIN ! 01 DT ! 647 LFREO ! 04 CMND C! 00 (MUX) C! ."SPECTRA 10 BLANK ."COMMENT 34 BLANK .PARAM CLLINE .” ANY CHANGES? [Y,N]" KEY 89 = IF 01 O7 ?.VALUES THEN ; EXPERIMENT FLUSH .HEAD CMND CO 04 AND IF .PARAM ELSE COLD-START THEN 01 [’] ?MENU-UP C! 25OFF 25ON O5 24 !ZXY .RV” X=EX H=HEAD E=EXP D=DISPLAY R=RECORD T=TNK P=PAINT G= GRAPHICS” BEGIN CLLINE ." Con-and? ” KEY ILOOP 0 END ; EXPERIMENT ’EXP ! Block Nunber: 138 QQQQUOMNHO ( TL-SCAN) SCAN IST-DATA-OFFSET POINTS ! OFFSET O SNAP 250 OFFSET ! ENTRY>BLK )R I 50 + O l- LXH ! 0 LXL ! I 54 + O TEMP ! ( S/E) I 50 + O DTEMP ! ( P/S) 32000 00 ( MINMAX) I 1+ CO 1+ R) CO DO I BLOCK DUP POINTS 0 + SNAP 1024 + SNAP DO I 20 TEMP 0 M/ DUP IF MINMAX ELSE DROP THEN 08 +LOOP 0 POINTS ! ?LEAVE-LOOP LOOP ZDUP LYH ! LYL ! SNAP - TMY 0 MIN TMY ! LXH 0 LXL 0 - TMX 0 MIN TMX ! OFFSET ! ; 106 Block Nunber: 139 comqayoabwiov—o ( TL-PLOT) TL-PLOT-END DTEMP 0 O< IF LEAVE THEN ; ’PLOT FRACTION-FIX IST-DATA-OFFSET POINTS ! OFFSET G SNAP 250 OFFSET ! ENTRY>BLH )R I 54 + O TEMP ! ( S/E) I 50 + O DUP DTEMP 2! ( P/S) I 1+ CO 1+ R) CO DO ?PLOTTING-BLK I BLOCK DUP POINTS 0 + SNAP 1024 + SNAP DO DTEMP 20 - DUP 100 MS I 20 TEMP O M/ PLOT-POINT LXH O ) IF LEAVE O DTEMP ! THEN -I DTEMP +! TL-PLOT-END 08 +LOOP 0 POINTS ! ?LEAVE-LOOP TL-PLOT-END LOOP OFFSET ! CR AMODE ; PLOT VCLR AXES GMODE ’PLOT 100 PYL ! 425 PYH ! GT? IF KEY-PROMPT 24*80 00 [’l ?MENU-UP C! THEN ; 107 REFERENCES REFERENCES l. N.J. Dovichi, J.C. Martin, J.H. Jett and R.A. Heller, Science 219, 845 (1983). 2. A. Rosencwaig, Anal. Chan. 41, 592A (1975). 3. S. Oda and T. Sawada, Anal. Chan. 54, 471 (1982). 4. L.B. Ereuzer, Anal. Chan. 46, 235A (1974). 5. N.B. Jackson, N.M. Aner, A.C. Boccara and D. Fournier, Appl. Opt. 29, 1333 (1981). 6. W.C. Mundy, R.S. Hughes and C.E. Carniglia, Appl. Phys. Lett. 43, 985 (1983). 7. H. Coufal, Appl. Phys. Lett. 44, 59 (1984). 8. E.V. Raddy, Rev. Sci. Instrun. 55, 422 (1983). 9. A.C. Boccara, D. Fournier, N. Jackson and N.M. Aner, Opt. Lett. 5, 377 (1980). 10. T. Masujina, A.N. Shards, L.B. Lloyd, J.M. Harris and E.M. Eyring, Anal. Chen. §§, 2975 (1984). 11. E. Peck, F.H. Fotiou and M.D. Morris, Anal. Chan. 21. 1359 (1985). 12. J. Stone, Appl. Opt. 12. 1828 (1973). 108 13. D.A. Creners and R.A. Heller, Appl. Opt. 2 , 1654 (1982). 1.— l4. H.B. Lin, J.S. Gaffney and A.J. Canpillo, J. Chronatog. 295, 205 (1981). 15. M.J. Pelletier, H.R. Thorshein and J.M. Harris, Anal. Chen. 54, 239 (1982). 16. M.J. Pelletier and J.M. Harris, Anal. Chen. 55, 1537 (1983). 17. N.J. Dovichi, T.G. Nolan and R.A. Rainer, Anal. Chan. 22. 1700 (1984). 18. T.G. Nolan, R.A. Weiner and N.J. Dovichi, Anal. Chen. §§o 1704 (1984). 19. J.M. Harris and N.J. Dovichi, Anal. Chen. 52, 695A (1980). 20. J.P. Gordon, R.C.C. Leite, R.S. Moore, S.P.S. Porto and J.R. Nhinnery, J. Appl. Phys. 25, 3 (1965). 21. R.C.C. Leite, R.S. Moore and J.R. Nhinnery, Appl. Phys. Lett. 5, 141 (1964). 22. D. Solinini, J. Appl. Phys. 21, 3314 (1966). 23. D. Solinini, Appl. Opt. 5, 1931 (1966). 24. J.R. Nhinnery, Acc. Chen. Res. 1, 225 (1974). 25. J.R. Brennan and D. Magda, J. Phys. Chan. 52, 705 (1978). 26. J.R. Barker and T. Rothen, Chen. Phys. 55, 331 (1982). 27. P.L. Trevor, T. Rothen and J.R. Barker, Chen. Phys. 55, 341 (1982). 109 28. M. King-Xiao and X. Zhu-De, Chen. Phys. Lett. 55, 563 (1983). 29. T. Inasaka, E. Shinanoe and N. Ishibashi, J. Chen. Phys. 19. 3201 (1983). 30. R.T. Bailey, F.R. Cruickshank, D. Pugh, A. McLeod and N. Johnstone, Chen. Phys. 55, 351 (1982). 31. R.T. Bailey, F.R. Cruickshank, D. Pugh and N. Johnstone, J. Chen Soc. Faraday Trans. 2 11, 1387 (1981). 32. E. Fuke, A. Hasegawa, M. Ueda and M. Itoh, Chen. Phys. Lett. 22. 176 (1981). 33. E. Fuke, M. Ueda. and M. Itoh, J. An. Chen. Soc. 195, 1091 (1983). 34. H.L. Fang and H.L. Swofford, J. Chen. Phys. 22, 6382 (1980). 35. H.L. Fang and R.L. Swofford, J. Chen. Phys. 22, 2607 (1980). 36. R.L. Swofford, M.E. Long and A.C. Albrecht, J. Chen. Phys. §§o 179 (1976). 37. A.J. Twaroski and D.S. Kliger, Chen. Phys. 29, 253 (1977). 38. A.J. Twaroski and D.S. Kliger, Chen. Phys. 29, 259 (1977). 39. D.S. Kliger, Acc. Chen. Res. ;§, 129 (1980). 40. H.L Fang and H.L. Swofford, ”The Thernal Lens in Absorption Spectronetry” in Ultrasensitive Laser §pectrosc022u ed. by D.S. Eliger (Acadenic Press, New York), 1983, pp. 175-232. 110 41. N.J. Dovichi and J.M. Harris, Anal. Chen. §1. 728 (1979). 42. -H. Mori, T. Inasaka and N. Ishibashi, Anal. Chen. 22. 1075 (1983). 43. T. Higashi, T. Inasaka and N. Ishibashi, Anal. Chen. §§o 1907 (1983). 44. K. Mori., T. Inasaka and N. Ishibashi, Anal. Chan. 22. 2034 (1982). 45. J.A. Alfhein and C.H. Langford, Anal. Chen. §Zo 861 (1985). 46. G.R. Long and 8.3. Bialkowski, Anal. Chan. 55, 2806 (1984). 47. J.P. Haushalter and M.D. Morris, Appl. Spectrosc. 22. 445 (1980). 48. H. Miyaishi, T. Inasaka and N. Ishibashi, Anal. Chin. Acta 224, 381 (1981). 49. T. Inasaka, H. Miyaishi and N. Ishibashi, Anal. Chin. Acta 245, 407 (1980). 50. E. Miyaishi, T. Inasaka and N. Ishibashi, Anal. Chen. §Sa 2039 (1982). 51. J.V. Beitz and J.P. Hessler, ”Oxidation State Specific Detection of Transuranic Ions in Solution", presented at The Argonne National Laboratory Specialists’ Septenber 5-6, 1979, Argonne National Laboratories, Argonne, IL. 52. E. Fujiwana, H. Lei, H. Uchiki, F. Shinokoshi, K. Fuwa and T. Eobayashi, Anal. Chen. 54, 2026 (1982). 53. E. Nakanishi, T. Inasaka and N. Ishibashi, Anal. Chen. §Z. 1219 (1985). 111 54. R.A. Leach and J.M. Harris, J. Chronatog. 215, 15 (1981). 55. N.J. Dovichi and J.M. Harris, Anal. Chen. 52, 689 (1981). 56. C.E. Buffett and M.D. Morris, Anal. Chen. §fl, 1824 (1982). 57. S.L. Nickolaisen and S.E. Bialkowski, Anal. Chen. g1, 758 (1985). 58. M.J. Sepaniak, J.D. Vargo, C.N. Hettler and M.P. Maskarinec, Anal. Chan. 55, 1252 (1984). 59. C.E. Buffett and M.D. Morris, Anal. Chen. 55, 376 (1983). 60. T.-E.J. Pang and M.D. Morris, Anal. Chen. 55, 1467 (1984). 61. R.A. Leach and J.M. Harris, Anal. Chen. §§o 2801 (1984). 62. C.A. Carter and J.M. Harris, Anal. Chen. 55, 922 (1984). 63. Y. Yang and R.E. Hairrell, Anal. Chen. 55, 3002 (1984). 64. J.S. Shirk, T.D. Harris and J.R. Mitchell, Anal. Chan. 52, 1701 (1980). 65. T.D. Harris and J.R. Mitchell, Anal. Chen. 52, 1706 (1980). 66. V.I. Grishko, I.G. Yudelevich and V.P. Grishko, Anal. Chin. Acta 159, 159 (1984). 112 67. T.D. Harris and_A.M. Willians, Appl. Spectrosc. 55, 28 (1985). 68. A. Yariv. Q229192--§12212291221_22-222 (Wiley. New York), 1975. 69. F.R. Dabby and J.R. Whinnery, Appl. Phys. Lett. 19. 284 (1968). - 70. H.S. Carslaw and J.C. Jaeger, Qperational Methods in 2991122-!2122221122 (Dover. New York). 1963. pp- 106-110- 71. N.J. Dovichi and J.M. Harris, Anal. Chen. 52, 2338 (1980). 72. S.J. Sheldon, L.V. Knight and J.M. Thorne, Appl. Opt. 21. 1663 (1982). 73. C.A. Carter and J.M. Harris, Appl. Opt. 25, 476 (1984). 74. H.R. Eogelnik, E.P. Ippen, A. Dienes and C.V. Shank, IEEE J. Ouantun Electronics 95:5, 373 (1971). 75. C.M. Phillips, R.L. Haas, R. Thiin, M. Moini, S. Bankroff and G.E. Leroi, Appl. Spectrosc. 32. 360 (1985). 76. Y. Suzaki and A. Tachibana, Appl. Opt. 15. 2809 (1975). 77. J.E. Pearson, T.G. McGill, S. Hurtin and A. Yariv, J. Opt. Soc. An. 55. 1440 (1969). 78. Model 1080 Bean-Scan, Photon Technology, San Jose, CA. 79. L.B. Dickson, Opt. Engineering 55, 70 (1979). 80. B. Newcone, Ph.D. Thesis, Michigan State University, East Lansing, MI, 1983. 113 81. C. Levenburg, Ouart. Appl. Math. 2, 164 (1944). 82. D.N. Marquardt, SIAM J. Appl. Math. 11: 431 (1963). 83. J.J. More, NTIS Publication ANL 80-74, Argonne National Laboratories, Argonne, IL, 1980. 84. J.J. More, ”The Levenburg-Marquardt Algorithn: Inplenentation and Theory." in 1221292-92122-19__H2292221122 2391---992291221__29912212. G.A. Hat-on. ed. (Springer Verlag, New York), 1977, p. 105. 85. D.F. Boltz and M.G. Mellon, Anal. Chan. 25, 749 (1948). 86. D. Gullstron and M.G. Mellon, Anal. Chan. 25, 1809 (1953). 87. D.F. Bolts and M.G. Mellon, Anal. Chen. (1947). “-1 I‘D m s] 00 88. R.H. Canpbell and M.G. Mellon, Anal. Chan. 52, 54 (1960). 89. L.G. Hargis, Anal. Letters A, 799 (1968). 90. J. Murphy and J.P. Riley, Anal. Chin. Acta 21, 31 (1962). 91. J.E. Going and S.J. Eisenreich, Anal. Chin. Acta 15, 95 (1974). 92. S.R. Crouch and H.V. Malnstadt, Anal. Chen. 55, 1084 (1967). 93. R.J. Jakubiec and D.F. Boltz, Microchin. Acta [Nien], 1199 (1970). 94. D.F.U Bolts and T.A. Howell, in 9595;931_Ag§1ygig‘ 2122-221221221212-2222221922129-_21--!29222212 (wiley. New York), 1978, p. 348. 114 95. C.M. Proctor and D.W. Hood, J. Mar. Res. (1954). ll—- I00 122 96. E. Stephens, Linnol. Oceanog. 5, 361 (1963). 97. A. Henriksen, Analyst 29. 29 (1965). 98. C.E. Buffett and M.D. Morris, Appl. Spectrosc. 5 455 (1983). IQ 99. R. Anthore, P. Flanent, G. Gouesbet, M. Rhazi and M.E. Neill, Appl. Opt. 21. 2 (1982). 100. G. Eortun and M. Seiler, Ang. Chan. 52, 687 (1939). ‘ 101. 939-92922222-21-992912191_292_£9x2122. R-C. Wee-t.ed. (CRC Press, Cleveland, OH). 102. R. Appleby, D.W. Janes and C.A. Bowie, Spectrochin. Acta 454, 785 (1984). 103. D.T. Janieson and J.S. Tudhope, Desalination 5, 393 (1970). 104. D.T. Janieson, J.S. Tudhope, R. Morris and G. Cartwright, Desalination Z, 23 (1970). 105. H.L. Jansen and J.M. Harris, Anal. Chen. 5], 1698 (1985). 106. Y. Yang, Anal. Chen. 55, 2336 (1984). 107. For further details surrounding the FORTH words written for these experinents, see R. Thiin, Ph.D. Thesis, Michigan State University, East Lansing, MI, 1985. 115 I! III ll I‘ll... III. ”'7ITll‘Ilu1l1uillJlTl:fillllffl'lillflillllfil“