JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1994 VOL. 9 67 Rapid Survey Analysis of Polymeric Materials by Laser-induced Plasma Emission Spectrometry* David R. Anderson and Cameron W. McLeod Division of Chemistry School of Science Sheffield Hallam University Sheffield UK S 7 7 WB Trevor A. Smith Arun Technology Limited Unit 7 6 South water Industrial Estate Station Road South water Horsham Sussex UK RH 13 7UD Laser-induced plasma emission spectrometry has been applied to the analysis of poly(viny1 chloride) mate- rials. Methodologies used to optimize the monitoring of the time-dependent emission from the laser-induced plasma are discussed and the effects of key parameters such as laser energy sample position and repetitive firing at given sites were examined. Basic performance data are reported for Ca [limit of detection 0.016% m/m and 4.8% relative standard deviation (RSD)] and Sb (limit of detection 0.04% m/m and 4.8% RSD) and the potential use for rapid survey analysis is demonstrated.Keywords Laser-induced plasma emission spectrometry; gated diode array detection; laser ablation; polymer analysis Polymeric materials generally contain a wide variety of addi- tives which are used to give specific properties to the material.' These include pigments for colour (e.g. white Ti02) fillers (e.g. CaCO,) stabilizers (e.g. Zn Pb and Ba) flame-retarding agents (e.g. Sb P) and smoke suppressants (e.g. Zn) among others. These inorganic components vary in concentration from trace (pg g-') to minor (%) levels. Analytical techniques currently in use for such analyses include solution-based atomic spectrometric and X-ray fluorescence spec- t r ~ m e t r y .~ An alternative approach is direct spectrochemical analysis by laser ablation (LA) either by direct spectral measurement of the laser-induced plasma or by coupling to another analytical technique. For example LA with inductively coupled plasma (ICP) atomic emission spectrometry has been used for survey analysis of paints and polymer^,^ while ICP mass spectrometry has been applied to polymeric and refrac- tory materiak6 Direct spectrochemical measurement of the plasma laser- induced plasma emission spectrometry (LIPS) enables rapid in situ analy~is,~.' and is particularly suited to process measure- ment.' The emission signals from the laser-induced plasma are complex and vary greatly with time.Time-resolved measure- ment is usually essential for the separation of the analyte emission response from the intense plasma background." Basic characterization of the laser-induced plasma and systematic studies concerning the effect of key parameters such as the wavelength of the laser light and the type and pressure of the buffer gas have been reported.''-16 Niemax et a2.,I6 concluded that a wavelength of 1064 nm was more suitable than 266 nm for analysis of glass and steel matrices. Other workers's'5 found that a reduced pressure of Ar gas typically 133.32 x lo3 Pa offered improved analytical performance. For process monitoring situations where it may not be feasible to achieve partial pressures atmospheric pressure has been used.Cremers and Archuleta' reported the in situ measurement of molten steel for five elements (Ni Cr Si Mn and Cu) using a laser wavelength of 1064nm by LIPS and LA-ICP and concluded that at least semiquantitative analysis was feasible. Lorenzen et a1.17 described the in situ monitoring of S Si and Zn in rubber production using an excimer laser operating at 248 nm and discussed the choice of laser wavelength for this application. The present study examines the application of LIPS to the *Presented in part at the 1993 European Winter Conference on Plasma Spectrochemistry Granada Spain. January 10-15 1993. rapid survey analysis of poly(viny1 chloride) (PVC) samples and provides information about the emission characteristics of the laser-induced plasma in particular the emission-time pro- files of analyte and background emission.Basic performance data for the determination of Sb Ca and P in PVC samples are presented. Experimental Instrumentation A schematic diagram of the LIPS system is shown in Fig. 1 and details of the instrumentation and operating parameters are given in Table 1. The system consisted of a Q-switched Nd:YAG laser with output wavelength of 1064 nm an optical multi-channel analyser (OMA) that comprised a spectrometer an intensified photodiode array (PDA) detector a com- puterized control system and a master pulse generator. Operation of the laser and the OMA was synchronized elec- tronically using the master pulse generator. This controlled the timings of the laser flash lamp and Q-switched laser firing and the detector gating and scanning of the detector array.Both the laser flash lamp and the OMA detector were operated at 10Hz. The laser was fired during alternate flashes of the flash lamp i.e. at 5 Hz and the OMA recorded a blank spectrum during the flash lamp cycles when the laser did not fire. Each blank spectrum was automatically subtracted from Laser Mirror 1064 nm r Pulse generator P ,rk+?Qbro asrn a Fibre optic Fig. 1 Schematic diagram of laser and spectrometer system68 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1994 VOL. 9 Table 1 Instrumentation and parameters used in this study Laser Spectra Physics DCR I1 Wavelength 1064 nm Repetition rate 10 Hz Laser flash lamp energy Pulse-width 10 ns Reflecting mirror Focus lens Ablation gas Sample position (relative to laser focal point) Spectrometer 0.28 m Czerny-Turner Grating 2400 g mm-' Spectral window 25 nm Effective resolution 0.20 nm Detector Intensified photodiode array Light collection and transfer 40 J 45" (Newport 10 QM 20 HM-15) 500 mm focal length (Newport SBX 040 AR.18) Air at room temperature and pressure -0.5 mm (0.5 mm towards lens) OMA EG&G OMA I11 Fibre optic bundle of 19 x 200 pm UV-grade fused silica Fibre optic observation position From laser focal point Above sample surface 1 mm 12 mm Pulse generator EG&G 1310 the previous emission spectrum to yield a net signal from the plasma.The run time for 5 laser shots was 1 s. Laser light was delivered to the sample by a mirror and a lens was used to focus the light onto the sample surface.The laser-induced plasma was formed at the surface of the sample and radiation was transmitted to the OMA by a fibre optic cable. The light was dispersed by the grating of the spec- trometer and detected by an intensified photodiode array. The detector was time-gated to allow precise control of the inte- gration time. Various spectral regions could be monitored by setting the grating to the required centre wavelength. This was controlled by the grating-drive mechanism which rotated the grating until the required wavelength window covered the detector elements. Three spectral windows 237-262,267-292 and 307-332 nm of centre wavelengths 250 280 and 320 nm respectively were used to detect the elements selected (Al Ba Ca Cu Fe Mg Pb P Sb Sn Ti and Zn).The emission lines monitored are given in Table2. In addition two barium lines were utilized (230.423 and 233.527 nm) at centre wavelength 230 nm. Data Acquisition The OMA hardware and software allowed versatile collection storage and manipulation of data. For example individual scans of the photodiode array representing separate laser shots could either be stored in separate memories or accumu- lated into a single memory. The time-gating capability of the OMA was used in two ways to address the problem of complex time-dependent emission from the laser-induced plasma. Two data acquisition modes of OMA operation are described below incremental mode and fixed time mode (FT). An incremental program was used for preliminary studies in which each scan of the array was stored in a separate memory and the integration window (100 ns) was stepped sequentially through the lifetime of the plasma by the increment time (100 ns).This enabled the spectra to be time-resolved to produce an emission- wavelength-time (E-W-T) profile of emission from the plasma. Examination of the transient signal responses within these profiles enabled suitable values for delay and integration times to be selected for use in an FT program. The FT data acquisition mode exploited the separation in time of the analyte response from the background signal and allowed greater measurement sensitivity to be realized. Here Table 2 List of element emission lines for each spectral window monitored in this study Wavelength range/nm 237-262 240.549 Cu 241.949 Sn 247.857 C I 250.200 Zn I1 250.911 C I1 251.203 C I I 251.743 Ti I1 252.560 Ti I1 252.852 S b I 253.401 P I 253.565 P I 254.480 Cu I1 255.328 P I 255.493 P I 255.796 Zn I1 256.253 FeII 259.806 S b I 259.881 Cu I1 261.418 Pb 267-292 279.553 Mg I1 280.199 P b I 280.270 Mg I1 283.304 P b I 283.999 S n I 285.213 MgI 286.333 S n I 286.426 Pb 287.792 S b I 307-332 308.215 A11 308.802 Ti I1 309.271 A11 315.887 C a I I 317.502 S n I 317.933 C a I I 322.579 Fe I1 322.775 Fe I1 323.252 S b I 323.452 Ti I1 323.612 Ti I1 323.904 Ti I1 324.199 Ti I1 324.754 C u I 326.233 S n I 326.751 S b I 327.396 C u I 328.233 Z n I 328.321 Sn I1 330.259 Z n I the start of integration was delayed by a set time the delay time to enable rejection of the initial intense background signal and the integration period captured the analyte emission signal response; typical parameters were delay time 500 ns and integration time 1 ps.Two FT programs FT1 and FT2 were used to acquire the laser generated spectra. Method FT1 consisted of firing 30 laser shots at the same spot each shot being stored individually i.e. 1 scan in 30 memories (30 scans total). Method FT2 accumulated five laser shots into one memory at one sample site and this was repeated seven times in total with a fresh site each time i.e. 5 scans in 7 memories (35 scans total). The run times for FT1 and FT2 were 6 and 7 s respectively with a further 25 s required for sample translation for FT2. Materials Poly(viny1 chloride) samples (A-F) were supplied in sheet form by FMC Process Additives (UK).Sample X was industrial grade PVC (Darvic) obtained locally. Procedure Samples were analysed as received with no sample preparation. The sample was mounted on an XYZ manipulator and a fresh area of material postioned at the laser spot. The manipulator was moved in height (2) until the sample was at the laser focal point 500 mm from the focusing lens and then moved 0.5 mm towards the lens. With the laser operating at a suitable flash lamp energy e.g. 40 J the acquisition was started. The laser fired under control of the OMA software and the master pulse generator and the resulting scans were stored on the computer and printed. Separate routines were used off-line to perform individual pixel analyses to give emission-time profiles and statistical information.Results and Discussion Preliminary Studies Previous studies1* of LIPS with metallurgical samples found that emission signal responses and analytical performance were greatly influenced by the complex interdependent relationship of several parameters e.g. laser energy sample position relative to laser focal point position of the fibre optic and OMAJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1994 VOL. 9 Fig. 2 Talysurf stylus-profile of a crater in PVC (sample X) produced by ten laser shots laser flash lamp energy 40 J. The profile was recorded by 164 scans in a 1.63 mm square grid detector settings. For example ablation with a relatively high laser energy (flash lamp energy 70 J irradiance x3.1 x 10" W cmP2) produced a more intense longer-lived plasma that required a different fibre optic viewing position and OMA settings for optimal performance compared with ablation with a lower energy (flash lamp energy 40 J x8.0 x 10'' W crnT2).For this preliminary study with poly- meric materials initial selection of laser lamp energy was made by examining the amount of laser-damage .to the sample and the variation in emission intensity produced by different values of laser energy. For optimum LIPS performance it may be necessary to characterize the experimental parameters listed above for the ablation of polymeric materials and also consider the effects of polymer properties such as transparency and colour upon ablation. A relatively low laser energy (flash lamp energy 40 J) was chosen which inflicted minimal laser damage to the sample but produced suitable emission responses.The laser appeared to couple well with coloured opaque samples. A Talysurf stylus-profile of the crater produced in PVC (sample X) by ten laser shots is depicted in Fig. 2. It shows the round crater to be approximately 1 mm in diameter and 50 pm deep. A dark- t B h > v) c a3 c c 0 v) v) w c .- c .- .- .- E 1 307 319 3 Wavelengthlnm 69 12 Fig. 3 LIPS spectra for ablation of A Ti metal; and B PVC sample X (laser flash lamp energy 40 J). OMA parameters delay time 900 ns; integration time 100 ns ened region about 2mm in diameter was observed around each crater. The LA of an opaque polymer was compared with the ablation of metal using conditions previously established for metal samples. Fig. 3 depicts spectra from the ablations of sampleX and Ti metal taken from the respective E-W-T profiles.The spectra are very similar and most emission lines can be identified as Ti indicating the presence of Ti in the polymer probably as the pigment titanium dioxide. The spec- trum for the polymer is more intense than that for the metal probably due to a combination of factors such as greater coupling of the laser and plasma with the polymer and the lower temperatures needed for volatilization and decompo- sition of the polymer compared with the metal. In the case of transparent polymeric samples no emission signals were obtained when material was ablated with the operating parameters specified under Experimental. Faint tun- nelling through the material and ablation of the metal support underneath the sample were observed.This would suggest that a plasma was not induced on the surface of the polymer and that the laser light was transmitted through the plastic to the metal. The tunnelling was due to the self-focusing of the laser t ~. - - Wavelength - / 1 Fig. 4 Emission-wavelength (237-262 nm)-time (0-900 ns) profile for ablation of sample A using the Incremental program (increment time 100 ns; integration time 100 ns) A carbon I 247.9 nm; and B carbon I1 250.9 and 251.2 nm; C antimony 252.9; and D phosphorus 253.4 and 253.5 nm70 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1994 VOL. 9 140 I 1 0 400 800 1200 1600 2000 Delay timehs Fig.5 Emission-time profile for ablation of sample A with the Incremental program (increment time 100 ns; integration time 10011s) A carbon I 247.9; B antimony 252.9; C phosphorus 253.4; and D background 245.0 nm beam within the material which trapped the light and pre- vented it from spreading to produce a waveguide.'' This was not evident in the opaque samples because the laser light was absorbed leading to ablation and production of the laser- induced plasma.When the transparent sample was moved away from the laser focus position i.e. 3 mm away from the lens (+ 3 mm) emission signals and laser damage on the sample surface were observed. The plasma was induced in the air above the sample surface and the plasma was responsible for volatilizing sample material and creating the crater. Results indicated that it was feasible to gain elemental composition data from transparent polymer material.These conditions were not adopted for opaque samples because greater damage was inflicted upon these materials without any significant improve- ment in performance. An E-W-T profile recorded with the OMA incremental program from the ablation of sample A (Fig. 4) shows the intense background continuum at early times and the emerg- 400 I/ L .- 0 v) c 2500 E 2000 .- .- - w 1500 C D -.-.- -.._.._.. _..- ..-*.- 500 1 ..-.._.._.._.._.._.._.. -..-.. E _-- - - - - - - - - - _ _ _ _ _ _ _ _ _ - I I I 1 I L 0 1 2 3 4 5 6 Memory No. Fig.6 Emission-time profile for ablation of sample A with FT program (delay time 500 ns; integration time 1 ps) (a) FT1 30 shots at same spot; and (b) FT2 5 shots in each memory 7 memories fresh site each memory A carbon I 247.9; B antimony 252.9; C antimony 259.8; D phosphorus 253.4; and E background 245.0 nm ence of atom/ion emission at later times.Carbon ion emissions (250.9 251.2 nm) were evident initially and carbon atom emissions (247.9 nm) later. The intense background signal and carbon ion emissions reflect the high temperatures of the plasma at early times but as the plasma expanded and cooled the background signal decayed rapidly and atomic emission lines for Sb and P became prominent. An E-T profile for selected wavelengths (Fig. 5) shows the responses of C Sb and P to be indistinguishable from the background up to 400ns. After this emission signals for these three elements remain above background up to about 1.5 ps. These observations enable suitable detector settings to be obtained for the FT data acquisition mode; the background signal is reduced significantly by 500ns so selection of a time delay of 500ns would minimize background contributions and a signal inte- gration of 1 ps would be appropriate for monitoring analyte emission.Analytical Performance Full quantitative measurement of inorganic additives in poly- meric materials is considered to be difficult because of the absence of suitable certified reference materials (CRMs) neces- sary to prepare calibration graphs. Quantitative measurements in this work are based on using characterized samples as calibration standards. To obtain analytical performance data OMA FT programs were devised using the system operating parameters established above. Initial studies used two FT programs FT1 and FT2 to examine the effect of repetitive firing of the laser on signal response and to test for sample homogeneity prior to making performance measurements.Results from the ablation of sample A by both methods are shown in Fig. 6 (laser flash lamp energy 40 J delay time 500 ns integration time 1 ps). With FT1 the Sb emission responses were not similar throughout the experiment although the C and P emission signals were reasonably constant. Both Sb lines increased in intensity at first and then decreased until A t >- v) C al C C 0 v) v) 4- .- 4d .- .- .- E w Wavelength --c Fig. 7 Emission-wavelength (237-262 nm)-memory number profile for ablation of sample A with FT2 with delay time 500 ns; integration time 1 ps 5 shots in each memory 7 memories fresh site each memory A carbon I 247.9; B antimony 252.9; and C antimony 259.8 nm Table 3 Comparison of data from fixed time methods FT1 and FT2 for the ablation of sample A Parameter No.of shots per site RSD C 247.8 nm RSD Sb 252.8 nm RSD P 253.4 nm S/N (P 2.6%) Value of n for statistical calculation S/N (Sb 4.9%) Method FT1 30 10.6 29.7 6.9 13.4 4.5 30 Method FT2 5 4.8 4.8 7.0 97 33 7JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1994 VOL. 9 71 1000 about memory number 9 after which a relatively constant signal response was observed. Emission responses for all elements were fairly constant with FT2 [Fig. 6(b)]. The emission response for Sb was not linear with concen- tration for method FT1 but FT2 did produce a linear response.It would appear that the Sb in the vicinity of the laser spot was selectively volatilized during the first few laser shots possibly through formation of volatile antimony chloride leading to a depletion of Sb for the remaining shots. The effect was not due to higher concentration levels of Sb at the surface as similar E-T profiles were obtained from the ablation of material beneath the sample surface. Use of a small number of laser shots i.e. five on several sites with method FT2 enabled a linear calibration to be produced. Reasonably con- stant signals for C Ca and P were obtained by FT1 suggesting that these elements were not selectively volatilized and it was still possible to make representative measurements after firing 30 laser shots at the same site.The emission responses for C and Sb were more reproducible Mg Sn - n L - 3000 I 3000 1500 A v) C 3 0 ). CI - Y C 01 .- 3000 C 0 cn .- .- E W 1500 3000 1500 n C C 237.0 2000 1000 0 2500 1250 with method FT2 compared with FT1. An emission-wave- length-memory number profile (Fig. 7) for ablation with FT2 clearly indicates good repeatability for successive laser firings and results suggest that the sample is relatively homogeneous. Precision [relative standard deviation (RSD)] improved from 10.6 and 29.7 (FTl) to 4.8 and 4.8 (FT2) for C and Sb respectively but there was no significant change for phos- phorus RSD 6.9 (Table 3). Using the C signal (247.86 nm) as an internal standard RSD was further improved to 2.3 and 3.4 for Sb and P respectively. This approach is only applicable for element emission signals that are within the spectral region 237-262 nm that contains a C emission signal.The signa1:noise ratio for C Sb and P was considerably improved using FT2 due to the accumulation of five laser shots into each OMA memory compared with one shot with FT1. Using method FT2 linear calibration graphs were generated for Sb (to 4.9% m/m) and Ca (to 6.8% m/m) emissions (Sb 252.8 and 258.8 nm; Ca 315.9 and 317.9 nm) with three data points for each element. The correlation coefficients were 8oo 1 2ooo i- 2ooo I 800 400 Ca 800 400 n Al 262.0 267.0 292.0 307.0 Wavelengthh m 332.0 Fig. 8 Laser-induced plasma emission spectra from samples A B D and E (from top to bottom). Spectra were recorded with fixed time method FT2 delay time 500 ns integration time 1 ps 5 shots in each memory 7 memories fresh site each memory.Emission wavelengths are given in Table 272 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY FEBRUARY 1994 VOL. 9 0.9936 and 0.9999 for Sb 259.8 nm and Ca 317.9 nm respect- ively. Limits of detection based on 30 (measurement of blank PVC sample) were estimated at 0.04 and 0.016% m/m for Sb 252.8 nm and Ca 317.9 nm respectively. The data are similar to values obtained for elements in steel (Ni Cr Si and Mn) by Cremers and Archuleta’ using LIPS. Survey Analysis To demonstrate the rapid analysis capability a range of PVC samples was examined. (The composition values quoted were supplied by the manufacturer.) Spectra from four samples are shown in Fig. 8 for three spectral regions. Samples A and B are clearly seen to contain Sb as the emission lines for Sb are identified within all three spectral windows for both samples.The greater emission intensity for A compared with B indicates the higher concentration present in A (Sb A 4.9% m/m; B 2.8% m/m). Tin and P are present in sample A (P 2.6% m/m Sn 0.05% m/m). Calcium emission lines are evident in samples B and D the greater emission intensity of B indicating the higher concentration present (Ca B 6.8% m/m; D 0.4% m/m). Emission lines of magnesium were observed for all four samples. Barium was detected in samples B (Ba 0.1% m/m) the Ba emission lines (230.423 and 233.527 nm) are not shown in Fig. 8. Lead (3.0% m/m) is present in sample D and Al P Sn and Zn are evident in E. Samples C and X (not shown) contain Sn and Zn and Ti and Mg respectively. These results show that the technique can be used for the rapid analysis of samples of PVC for a range of elements without the need for time-consuming sample dissolution and possible loss of volatile elements.Conclusions Laser-induced plasma emission spectrometry has been applied to the rapid survey analysis of polymeric materials. This study monitored a range of twelve elements (Al Ba Ca Cu Fe Mg Pb P Sb Sn Ti and Zn) and provided analytical data for key elements (Sb and Ca). Quantitative measurement was limited by the availability of characterized samples. To address this problem it is proposed to prepare a series of samples covering a wide range of concentration levels for selected elements which will enable additional calibrations and further quantitative measurements to be made.The speed of analysis compactness of instrumentation and simplicity of operation suggest that the technique has the potential for compositional monitoring of polymeric materials in industrial processes such as manufacture and recycling. We would like to thank Richard Dellar (FMC Corporation Manchester UK M17 1WT) for providing samples. We are grateful to Richard Burdett (EG&G Sorbus House Wokingham UK) for valuable assistance in configuring the OMA system. We thank Mike Simpson for helpful discussions and Mac Jackson (School of Engineering) for the talysurf analysis. The authors thank the DTI and SERC for funding this project under the LINK TAPM scheme. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 References Plastic Additives Handbook eds. Gachter and Muller Hanser 1983.Wong K. L. Anal. Chem. 1981 53 2148. DiPasquale G. and Casetta B. At. Spectrosc. 1984 5 209. Warren P. L. Anal. Proc. 1990 27 186. Booth P. K. and McLeod C. W. Mikrochim. Acta 1989 3 283. Marshall J. Franks J. Abell I. and Tye C. J. Anal. At. Spectrom. 1991 6 145. Moenke-Blankenburg L. Laser Micro Analysis Wiley 1989. Laser-induced Plasmas and Applications eds. Cremers D. A. and Radziemski L. J. Marcel Dekker New York 1989. Cremers D. A. Archuleta F. L. and Dilworth H.C. Proceedings of the 5th Process Tech. Con$ on Meas. and Control Instrum. in the Iron and Steel Industry April 14-17 1985 pp. 157-162. Iron and Steel Society Warrendale PA USA. Cremers D. A. and Radziemski L. J. in Applications of Laser Spectroscopy eds. Cremers D. A. and Radziemski L. J. Marcel Dekker New York 1986 ch. 5 pp. 380-395. Niemax K. Leis F. Sdorra W. and KO J. B. Mikrochim. Acta 1989 2 185. Niemax K. KO J. B. and Sdorra W. Fresnius’ Z . Anal. Chem. 1989 335 648. Niemax K. and Sdorra W. Mikrochim. Acta 1992 107 319. Iida Y. Anal. Sci 1991 7 61. Kuzuya M. and Mikami O. J. Anal. At. Spectrom. 1992 7 493. Niemax K. Sdorra W. and Brust J. Mikrochirn. Acta 1992 108 1. Lorenzen C. J. Carlhoff C. Hahn U. and Jogwich M. J. Anal. At. Spectrom. 1992 7 1029. Anderson D. R. and McLeod C. W. unpublished work. Ready J. F. EfSects of High Power Laser Radiation Academic Press New York 1971. Paper 3/02845C Received May 19 1993 Accepted November 8 1993