Anal. Chem. 1981, 53, 665-676
possibility of the application of extremely sensitive methods of HzOz detection which employ fluorescence or chemiluminescence, and (8) may be applicable to a wide range of biologically important molecules including haptens and antigens. Although the work reported herein illustrates the PGLIA concept by use of a simple haptenic molecule, theophylline, it was also envisaged that the method would be applicable to much larger molecules including protein. Work has thus been in progress which demonstrates that the PGLIA can be used to measure IgG concentration and will be reported in a subsequent paper (22). LITERATURE CITED Rubenstein, K. E.; SchneMer, R. S.; Ullman, E. F. Blochim. Blophys. Res. Commun. 1072, 47, 846-657. Haimovich, J.; Hurwitz, E.; Novik, N.; %!a, M. Biochim. Blophys. Acta 1070, 207, 115-124. Leute, R. K.; Ullman, E. F.; Goldsteln, A.; Herzenberg, L. A. Nature (London) New Biol. 1872, 236, 93-94. Dandliker, W. B.; Schapiro, H. C.; Maduski, J. W.; Alonso, R.; Feigen, G. A.; Hamrick, J. R. Immunochemistry 1884, 1 , 165-191. Schroeder, H. R.; Vogelhut, P. 0.; Carrico, R. J.; Boguslaski, R. C.; Buckler, R. T. Anal. Chem. 1878, 48, 1933-1937. Carrico, R. J.; Christner. J. E.; Boguslaski, R. C.; Yeung, K. K. Anal. Blochem. 1078, 72, 271-282.
885
(7) Burd, J. F.; Wong, R. C.; Feeney, J. E.; Carrlco, R. J.; Boguslaski, R. C. Clin. Chem. ( Winston-Salem, N.C.) 1077, 23, 1402-1408. (8) Swoboda, B. E. P. Biochim. Biophys. Acta 1888, 175, 365-379. (9) Harvey, R. A.; Damle, S. FEBS Lett. 1872, 28, 341-343. (10) Koberstein, R. Eur. J. Blochem. 1878, 87, 223-229. (11) Whitby, L. G. Biochem. J. 1053, 54, 437-492. (12) Cook, C. E.; Twine, M. E.; Meyers, M.; Amerson, E.; Kepler, J. A.; Taylor, G. F. Res. Commun. Pathol. pharmacal. 1078, 13, 487-505. (13) Swoboda, B. E. P.; Massey, V. J. Biol. Chem. 1068, 240. 2209-2215. (14) Barham, D.; Trinder, P. Analyst (London) 1872, 97, 142-145. (15) Trayer, I. P.;Trayer, H. R.; Small, D. A. P.; Bottomley, R. C. Blochem. J. 1074, 139, 609-623. (16) Johnson, R. D.; LeJohn, L.: Carrico, R. J. Anal. Blochem. 1078, 88, 526-530. (17) Zappelll, P.; Pappa, R.; Rossodivita, A.; Re, L. J. Eur. Bhhem. 1078, 89, 491-499. (18) Mkenko, P. A.; Ogihrie, R. I. N. Engl. J. Med. 1073, 239, 600-603. (19) Jenne, J. W.; Wyze, E.; Rood, F. S.; MacDonald, F. M. CNn. PharmaCol. Ther. 1872, 13, 349-360. (20) Weldner, N.; McDonald, J. M.; Tlebar, V. L.; Smith, C. H.; Kessler, Q,; Ladenson, J. H.; Dietzler. D. N. Clin. Chim. Acta 1070, 97, 9-17. (21) Forester, R. L.; Wonall, J.; Robertson, W. R.; Watali, L. J.; Wliheim, D. Ther. Drug Monitoring 1070, 1 , 381-385. (22) Yeager, F. M.; Ngo, T. T.; Carrico, R. J.; Morris, D. L.; Boguslaski, R. C.; Hornby. W. E., In preparation.
RECEIVED for review July 1, 1980. Accepted December 22, 1980.
Chemical Derivatization in Electron Spectroscopy for Chemical Analysis of Surface Functional Groups Introduced on Low-Density Polyethylene Film Dennis S. Everhart and Charles N. Reilley' Kenan Laboratories of Chemistry, Universiiy of North Carollna, Chapel Hill, North Carolina 27514
Derlvatlzatlon reagents contalnlng an elemental tag facllltate ESCA analysts of surface functlonal groups Introduced on low-denslty polyethylene film exposed to radlo frequency Inductively coupled N2or Ar plasma. Angulardependent ESCA measurements of polyethylene treated wlth nltrogen plasma Indicate a surface that Is vertically Inhomogeneous In oxygen and nitrogen. Angular-dependent spectra help characterize the reactlons of some of these tagged reagents wlth the plasma-modified polymer. Several tags decompose during ESCA analysis wlth the DuPont 650 B spectrometer, but all derlvatlves are stable to ESCA analysls wlth the PHI 548 spectrometer. Na accelerates decomposltlon. Modest reductlons In anode power can retard sample decomposltlon and help remedy some dire effects whlch sample lnstablllty has on the semlquantltatlve (f15 % ) potentlal of ESCA.
The surface properties of organic polymers are of considerable importance in many industrial, biomedical, and technological applications (1-3). Wettability, adhesion, abrasion resistance, biocompatability, and reverse osmosis character are a few areas where the chemical properties of the surface have a prodigious role. Often modification of polymer surfaces is required while leaving the bulk properties unchanged. For example, surface carbonyl and carboxyl groups are incorporated in low-density polyethylene films upon mild chemical oxidation, and these functionalities are at least in part responsible for the increased wettability of the previously hydrophobic surfaces (4,5). Numerous reports on the identi0003-2700/8 1/0353-0665$01.25/0
fication and chemical modification of functional groups introduced on polyethylene film with chemical oxidation have been made (6-10). Surface modification by exposure to glow discharge (low temperature) plasma has received considerable attention (11-18). The distinguishing feature of surface modification effected by plasma is that the process is experimentallysimple and can be selectively controlled to produce drastic alterations in surface properties without affecting the overall quality of the material. A two-component, direct and radiative energy transfer, model has been proposed to characterize argon gas plasma modification of polymers (19). This model predicts that the outermost monolayer cross-links rapidly as a consequence of Hz elimination induced by direct energy transfer from argon ions and metastables. Despite the large amount of effort in this area, the details of plasma-surface interactions are complex and consequently plasma modification of polymers has largely been developed empirically. One technique especially suited for studying surface modification is electron spectroscopy for chemical analysis (ESCA) (20). Clark et al. (21-23) have extensively elaborated on the use of ESCA to study polymers. The utility of ESCA to elucidate changes in structural and chemical features of a modified surface is well substantiated. Unfortunately, ESCA analysis often lacks the resolution required to unambiguously identify a specific functional group. To help overcome this problem, curve fitting procedures are utilized which attempt to decompose the ESCA spectrum into a collection of individual peaks. The binding energies (BE) and areas of the composite spectra are then used to identify the presence of 0 1981 American Chemical Society
666
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
a particular group. Although useful, these decomposition procedures lead to results that are often speculative. The differential charging associated with the ESCA analysis of insulators results in indeterminate increases in peak positions and widths, making these procedures even more questionable. For enhancement of identification of specific surface functional groups, derivatization methods with group specific reagents containing an ESCA “tag” are becoming increasingly popular. Methods specifically developed for the analysis of modified polymers have recently appeared (24-28). We have initiated an investigation of this approach to identify groups introduced on low-density polyethylene during exposure to rf inductively coupled plasmas of Ar and N2 The stability of these tags during ESCA analysis at moderate vacuums (lo-’ torr) as well as the quantitative features of these methods are examined. Angular-dependent studies allow for more thorough characterization of these reactions and provide a better understanding of the reactions of surface groups in a polymer film.
EXPERIMENTAL SECTION ESCA Spectra. ESCA spectra (Mg Ka) were recorded with a DuPont 650B and a PHI 548 electron spectrometer. Both instruments were calibrated for a Au 4f 712 binding energy of 83.8 eV. Unless otherwise stated, binding energies are not corrected for sample charging. The number of scans for each frame illustrated is indicated in parentheses following the core designation (i.e., F(ls)(4)). The DuPont spectrometer was interfaced to a microprocessor to facilitate data manipulation. Maximum X-ray power was 330 W (30 mA; 11kV). Spectra acquired at reduced X-ray power (9 kV) still maintained a saturated emission current of 30 mA. Vacuum levels during acquisition with maximum power were torr while reductions in anode voltages normally (2-4) X resulted in slightly lower analyzer pressures. Polymer films were secured to cylindrical brass probes with double sided tape. Each 19 eV scan required about 60 s to complete. Spectra were recorded on the PHI 548 spectrometer at 300 W (10 kV; 30 mA; pass energy = 50 eV). Vacuum levels were below 5.0 X torr. Samples were secured to the ESCA probe with copper clamps to avoid use of tape. Each 20-eV scan required ca 30 s to complete. Angular-dependent measurements were recorded with a Physical Electronics (PHI) 548 spectrometer (90”angular aperture; pass energy 50 eV). Details of the angular resolved feature are found elsewhere (29). Surface biased intensities were measured at grazing angles of about lo”, measured parallel to the surface. “Bulk” intensities were collected from electron emission about 87” parallel to the surface. Because these angular-dependent measurements involve large solid angles, they are used only for a qualitative assessment of surface homogeneity. Computer curve fitting utilized a damped, nonlinear leastsquares routine previously described (30). The calculated spectra were constructed from Gaussian-Lorentzian peak shapes (50% Gaussian fraction) and an S-curve background. The S curve accounts for the high binding energy features which result from inelastically scattered photoelectrons Minimum deviation from the experimental spectrum was used as a “best-fit” criterion. Because of sample charging and possible differential charging artifacts, peak widths are uncertain and consequently, widths of the composite spectra were allowed to vary during the fitting procedure. Reagents. Low-density polyethylene film was a product of Northern Petrochemical Co. (density = 0.917 g/mL; M, = 2.59 X lo4; A?, = 4.95 X lo5) and contained trace amounts of antioxidants. All chemicals were obtained from Aldrich. Trifluoroethanol (TFE), trifluoroacetic anhydride (TFAA), and pentane were “gold label”; all other chemicals were reagent grade and used without further purification. 2-Propanol was dried with anhydrous K2C03. CaC12was used as a desiccant during Soxhlet extractions. Procedure. All films were Soxhlet extracted with 2-propanol (temperature ca. 75 “C) for 18 h as previously suggested (7) to remove antioxidants prior to plasma treatment. After extraction,
films were air-dried and stored dark in a vacuum desiccator until use. After this cleaning procedure, samples were handled with stainless steel tweezers. Plasmas were generated with an rf (13.56 MHz) Harrick plasma cleaner (ModelPDC-3XG) operated at lowest power setting (ca. 10 W). The diameter and length of the plasma chamber are 7.6 cm and 20 cm, respectively. A dry ice-2-propanol trap was used between sample and two-stage mechanical pump. Prior to discharge, samples (3 X 5 om2films) were evacuated to less than 10 mtorr and subsequently flushed at 300 mtorr for 10 min with the gas subsequently utilized for the plasma treatment. Five minutes at 250 mtorr and 1min at 150 mtorr were the exposure conditions selected for N2 and for Ar plasmas, respectively. These plasma-modified samples are referred to as PE-N and PE-Ar, respectively. After the plasma was extinguished, the system was then flushed for an additional 5 min before exposing the plasma-treated sample to the atmosphere. Samples were normally used several hours after plasma treatment. Reaction 1. Formation of Schiff Base with Pentafluorobenzaldehyde (PFB). To a solution containing 300 pL of PFB (0.1M) in 15 mL of pentane a 3 X 5 cm2sample of PE-Ar or PE-N was reacted for 2 h at 35-40 “C. The sample was then
washed with pentane and Soxhlet extracted with pentane (temperature ca. 33 “C) for 12 h. This sample is designated by the abbreviation PE-Ar-PFB or PE-N-PFB. Reaction 2. Coupling 2,2,2-Trifluoroethanol (TFE) with
Dicyclohexylcarbodiimide (DCC). To a solution containing 500 pL of TFE, 1mL pyridine, and 200 mg of DCC in 15 mL of CH2C12,a 3 X 5 cm2sample of PE-Ar or PE-N was reacted for 15 h at 25 “C. The sample was then washed with anhydrous ethyl ether and Soxhlet extracted for 12 h with ethyl ether. This sample is designated as PE-Ar-TFE or PE-N-TFE. Reaction 3. Reaction of Trifluoroacetic Anhydride (TFAA). To a solution containing 1mL of TFAA (0.4 M) and 1 mL of pyridine in 15 mL of benzene, a 3 X 5 cm2 sample of PE-Ar was reacted for 1.5 h at 25 “C. The sample was washed with benzene and then Soxhlet extracted for 12 h with ethyl ether. The sodium salt of this sample was prepared by immersing the film in a dilute 2-propanol-NaOH solution for 5 s. Longer times resulted in significant hydrolysis of the trifluoroacetyl group. The sample was then rinsed with 2-propanol. These samples are designated as PE-Ar-TFAA and PE-Ar-TFAA(Na+). Reaction 4. Formation of Pentafluorobenzyl Esters with Pentafluorobenzyl Bromide (PFBBr). PE-Ar(K+)or PE-N(K+)was prepared by immersing a sample of PE-Ar or PE-N in alcoholic KOH as described in reaction 8 (substitute KOH for NaOH). To a solution containing 0.5 mL of PFBBr (0.1 M) in 15 mL of benzene, a 3 X 5 cm2sample of PEAr (K+)or PEN(K+) was reacted for 15 h at 25 “C. The sample was washed in benzene and then Soxhlet extracted for 12 h with ethyl ether. This sample is designated PE-Ar-PFBBr or PE-N-PFBBr. Eliminating the alcoholic KOH treatment resulted in much slower reaction; without K+, the product of the reaction of PE-Ar with PFBBr had a F(1s) area only 5% of the F(1s) area of PE-Ar-PFBBr. Reaction 5. Reactions of (Pentafluoropheny1)hydrazine (PFPH) and (Tetrafluoropheny1)hydrazine (TFPH). To a solution containing 150 mg of either TFPH or PFPH (ca. 0.1 M) and 1drop of concentrated HCl in 15 mL of 95% ethanol, 3 X 5 cm2sample of PE-Ar or PE-N was reacted for 2 h at 25 “C. The sample was then washed with 100% ethanol and Soxhlet extracted for 12 h with ethyl ether. Samples are designated as PEAr-PFPH and PE-Ar-TFPH. Reaction 6. Reaction of (Monofluoropheny1)hydrazine Hydrochloride (MFPHSHCI). A 200-mg sample of MFPH-HC1 was dissolved in 15 mL of 95% ethanol by the dropwise addition of 1M KOH. KC1 precipitated from the yellow solution which was subsequently filtered and adjusted to pH 3 with 0.1 M HC1. In this solution, a P E A sample was reacted for 2 h at 25 “C. The sample was washed with 100% ethanol and Soxhlet extracted with ethyl ether for 12 h and designated as PE-Ar-MFPH. Reaction 7. Reaction of Mercuric Trifluoroacetate [Hg(TFA)2]-2,2,2-Trichloroethanol (TCE). To a solution containing 400 mg of Hg(TFA)2(0.06 M) and 500 p L of TCE in 15 mL of benzene, a 3 X 5 cm2sample of PE-Ar or P E N was reacted for 2 h at 25 “C. The sample was then washed with benzene and Soxhlet extracted for 12 h with pentane and designated as PE-
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
Table I. Results of Curve Fitting C( 1s) Signal of Polyethylene Exposed to Nitrogen or to Argon Plasma posi tionP width, possible sample peak eV eV area assignment PE-N
PE-Ar
1 2 3 4 5
285.0 286.0 286.9 288.1 289.4
1.86 1.80 2.06 1.74 1.89
8817 1010 935 605 605
1 2 3 4 5 6
285.0 285.6 286.6 287.0 288.1 289.3
1.72 1.70 1.91 2.02 2.18 1.98
10757 1950 1073 7 89 69 1 455
CH,; C=C C-NH, C-OH
667
% total
carbon 74 8 8 5 5
c=o
CO,H CH,; C=C C-NH, C-OH
68 12 7 5 4 3
?C
c=o
CO,H This peak may be assigned to conjugated carbons such as or$a Reference CH, of polyethylene to 285.0 eV. unsaturated carbonyls, 1,3-dienes,etc. as well as to carbons containing the amine group. Including this addition peak was necessary to keep FWHM in the acceptable range (1.7-2.1 eV) although its assignment is not certain. One explanation is that this signal arises from the enol tautomer of peak 5. Ar-Hg(TFA) or PE-N-Hg(TFA). Reaction 8. Reaction of Alcoholic NaOH. A dilute solution of NaOH was prepared by dissolving ca. 200 mg of NaOH in 300 mL of anhydrous 2-propanol (ROH). PE-Ar(Nat) or PE-N(Na+) was prepared by immersinga 3 X 5 cm2sample of PE-Ar or PE-N in 20 mL of this solution for 30 s. The sample was rinsed with anhydrous 2-propanol. The amount of Nat incorporated was strongly dependent on sample washing. This point will be addressed in the “Results and Discussions” section. Reaction 9. Coupling Tribromoacetic Acid (TBrA) with Dicyclohexylcarbodiimide (DCC). To a solution containing 100 mg of TBrA and 200 mg of DCC in 15 mL of CHZCl,, a 3 X 5 cm2sample of PE-Ar was reacted for 12 h at 25 “C. Immediately upon addition of TBrA to the DCC/CH2C12solution, a white precipitate formed. This precipitate dissolved after about 5 min. The reacted sample was washed with anhydrous ethyl ether and Soxhlet extracted for 12 h with ethyl ether. This sample is designated as PE-Ar-TBrA. Diborane Reduction. Argon plasma modified polyethylene (PE-Ar)was reduced with 0.1 M diborane/THF by reaction,under Nz, for 3 h at 25 O C . Subsequently, the sample was removed, washed with THF (2X), and then washed with CHBOHas previously suggested (7). The alkylborane was hydrolyzed with 4 N H2S04at 50 “C (2X). The film was then washed with H20(2x) and finally with acetone. The reduced sample (PEAr’)was stored in a vacuum desiccator until use. Throughout this study, reactions will be referred to by their abbreviated form. The reactions of PE-Ar are illustrated in Figure 2. Reaction times were chosen to maximize ESCA “tag”intensity. RESULTS AND DISCUSSION Plasma Modification of Polyethylene. The ESCA spectra of polyethylene (PE) exposed to Ar or Nz plasma have been previously reported (13). These spectra indicate that both plasmas oxidize the polymer surface as evident by the increased O(1s) intensity and the appearance of a high binding energy (BE) shoulder in the C(1s) region. In addition to oxidized carbon, both plasmas cause incorporation of nitrogen-containing functional groups into the P E film. The C(ls), N(ls), and O(1s) spectra of P E N are shown in Figure la. The N(ls) area measured for PE-Ar is considerably lower than the N(1s) area for PE-N (see Table 11). For both PE-Ar and PE-N, the charge-corrected N(1s) BE of 400.5 eV (referenced to CH, a t 285.0 eV) suggests organic nitrogen groups (Le., amine, imine, amide, nitrile) but specific chemical identification cannot be determined from the ESCA spectrum alone. It is at first surprising that such large amounts of oxygen are incorporated with Ar and Nz plasmas. However, it is reasonable to expect oxygen uptake through the reaction of surface radicals and surface intermediates with 0, and HzO during exposure to atmosphere (31). Because exhaustive attempts to remove residual oxygen and HzO from the plasma gasses were not made, oxygen constituents in the plasma
a > r C lS(1) r N lS(4)rO lS(1)
t it
1
H
5e.v.
B I N D I N G ENERGY (E. V . )
1
287.8
1
BINDING ENERGY (E. V . ) (a) C(ls), N(ls), and O(1s) ESCA spectra of polyethylene after a 5-mln exposure to nibogen plasma (E-N). (b) c(ls), N(ls), and F(1s) ESCA spectra of PE-N after treatment with pentafluorobenzaldehyde (PENPFB: reaction 1). AI spectra were taken w b a Wont MOB spectrometer at 330 W. Figure 1.
volume could also react with the polymer. The low-pressure plasmas used in this study were advantageous because they allowed for a modified surface containing oxygen and nitrogen functional groups. For these plasmas, modification is best described as resulting from an Ar/N2/02 or a N2/02plasma mixture rather than from a pure Ar or a pure Nz plasma. Substantial evidence indicates that ethylene, alcohol, carbonyl, and carboxylate groups Carl be formed on PE during plasma treatment (2,11,26). Identification of specific groups from the ESCA spectra is supported by the results of a peak fitting procedure as summarized in Table I. For a reasonable
068
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
range of arbitrarily chosen peak widths (1.7-2.1 eV), the C(1s) signal of PE-N can be reconstructed from a minimum of five peaks; PE-Ar requires a minimum of six peaks. Possible assignments are based on BE values from homopolymers which contain a known group, for example C(1s)-OH in polyvinyl alcohol. Clark and Thomas (32) have reported a systematic investigation of core level binding energies for a series of homopolymers. The relative peak areas for each composite signal determine the percent carbon associated with a particular group. The values in Table I are similar to those reported by Clark (21)for the Oz plasma modification of PE. For each plasma, approximately 18% of the total carbon area results from carbon-oxygen functions. Within experimental error (f15%), this oxidized C(ls) area accounts for all the measured oxygen area. It should be emphasized that ESCA peak fitting is not a deconvolution (33)but only suggests the presence of a particular functional group. Clearly, a more objective approach to functional group analysis is desirable. Chemical Derivatization of Surface Functional Groups. Chemical derivatization is a clever approach to analyzing specific functional groups with ESCA. Derivatization reagents which contain a unique elemental tag can readily be identified in subsequent ESCA analysis, In principle, these reagents should react selectively and analytically with only the intended functional group. The elemental tag should have a favorable photoelectric cross-section and should be stable to ESCA analysis. For example, PE-N containing surface amines would react with pentafluorobenzaldehyde to form the Schiff base derivative via reaction 1. Pentafluorobenzaldehydereacts with PE-N C G F ~ C H O PE-N=CHCGF~ H2O (1)
+
-
+
primary and secondary amines and hydrazines but is not expected to react with amides, imines, nitriles, or nitrogencontaining heterocyclics. A F(1s) signal during subsequent ESCA analysis of (1)would be strong evidence for surface amines and/or hydrazines on the polymer. The ESCA spectrum of PE-N treated with 0.1 M pentafluorobenzaldehyde (PFB) in pentane is illustrated in Figure lb. Several changes in the ESCA spectra are recognized (1) a large F(1s) area, (2) a decrease in the N(ls) area, and (3) additional area in the high BE region of C(1s). A control sample not treated with Nz plasma (PE-H) does not show any fluorine after repeated signal averaging. This strongly suggests that a large portion of the nitrogen functions detected in Figure l a are primary amines (or hydrazines), and complements the previous work of Hollahan et al. (34)who reported that ammonia and Nz/Hzplasmas can introduce amine groups on PE. The specific nature of these amines (primary, secondary, or tertiary) could not be determined at that time. For PE-N, the hydrogen necessary to complete amine formation is introduced into the plasma volume from H2 eliminated during Nz plasma etching (11)and can also be produced with abstraction from CH2by atomic nitrogen radicals incorporated in the polymer chain. The nitrogen groups of PE-N are readily protonated with 0.1 M HzS04(35).The resulting ESCA spectrum consists of a sharp N(1s) signal with a charge-correctedbinding of 401.5 eV (reference C(1s) of polyethylene CH2to 285.0 eV). Because protonated hydrazines would show a broad N(ls) signal characteristic of a protonated and a free nitrogen, this sharp N(1s) signal indicates that hydrazines are not present on PE-N. Furthermore, the sharp N(ls) peak which results upon protonation indicates that amines are the principal nitrogen-containing functional groups on PE-N. This point will be addressed later in the discussion. Besides allowing for the more positive identification of a specific group, derivatization offers a significant decrease in detection limits. The increased sensitivity for PEN=CHCPS
Table 11. Atomic Densitya Ratios for PE-H, PE-Ar, and PE-N before and after Derivatizationd (A) Atomic Density Ratios before Derivatization sample PE-H PE-Ar PE-N
c/o 55 5.0
4.5
C/N b
162 26
O/N b
30 7.8
(B) Atomic Density Ratios after Derivatization reaction product C/tagc PE-H 1 no reaction 2 no reaction 3 PE-H-TFAA 330 4 no reaction 5 PE-H-PFPH 810 7 no reaction 8 PE-H-(Na') 87 9 no reaction PE-Ar 1 PE-Ar-PFB 105 2 PE-Ar-TFE 66 3 PE-Ar-TFAA 36 4 PE-Ar-PFBBr 50 5 PE-Ar-PFPH 170 7 PE-Ar-Hg(TFA) 49 8 PE-Ar-(Na+) 4.7 9 PE-Ar-TBrA 381 PE-N 1 PE-N-PFB 24 180 2 PE-N-TFE 4 PE-N-PFBBr 32 5 PE-N-PFPH 80 7 PE-N-Hg(TFA) 112 8 PE-N-(Na+) 3.6 a Atomic density calculations are described later in the discussion. Less than detectable amounts of nitrogen after four scans. Corrected for stoichiometry of tag. All spectra acquired with the DuPont 650B spectrometer at full power. is achieved by a combination of two factors, a 5-fold stoichiometric enchancement and a 2.4-fold cross-section enhancement of the fluorine tag. The decrease in detection limits is illustrated with the reaction of PE-Ar with pentafluorobenzaldehyde to form PE-Ar-PFB. The dilute concentration of nitrogen groups in PE-Ar is easily characterized as primary amine functionals. For both PE-Ar-PFB and PE-N-PFB, the C(ls)/F(ls) area ratio is ca. 14 times smaller than the C(ls)/N(ls) area ratio. This is in good agreement with the 12-fold enhancement predicted from the stoichiometric and cross-section considerations of the PFB derivatives. Chemical derivatization permits more readily identification of specific functional groups. Figure 2 illustrates a series of reactions to identify groups expected on PEAr (2,11,%). The ESCA spectra of these derivatized surfaces are illustrated in Figure 3. Atomic density ratios of the elemental ESCA tag for each reaction are summarized in Table 11. Before the results of these derivatizations are discussed, the question of reaction selectivity will be addressed. The Selectivity of Derivatization Reactions. In principle, derivatization reagents should react quantitatively and specifically with only the intended functional group. Reactions of interfering functional groups must be accounted for if an accurate qualitative analysis is to be achieved. This requires that each reaction should be evaluated with surfaces known to contain specific groups. Unfortunately, standard polymer surfaces are often difficult to formulate. Several methods to generate standard polymer surfaces could be envisioned. A wide variety of commercial polymers and copolymen me available and these could help characterize reaction specificity. For example, polyvinyl alcohol might be
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
SPECTZA
FR 0D b C C 1s -NH2
-
1
CsFsCHC
-C@2H
2 -
C F3C Y2 Gi-
0
is
N
is
F 1s
> H
5 E.V.
C6dllNCNC6H11'
a)
-C02 H
-OH
r 0 2 C C F3 1 ) KOP (ROH)
-C02H
2)CgF5CH2Br
tc-0
5 -
C6 F5 N H N H2
'C-
7 -
Hg(CF3CO2 12 C 13C H2Gh
-C@*H
-
8
NaOH ( R O H )
,c-
669
'
PE-Ar 69p. 1
-CG2CH2CgF5
b) 3C:NNLICgF5
-C-HQ(C F3C 0 2 ) -YOCH2CC13
Figure 2. Reactions of PE-Ar with various tagged reagents. useful to evaluate reactions with the alcohol function. Although this methodology appears straightforward, commercial polymers have several disadvantages which complicate their use as standard surfaces. These include residual surface oxidation and/or contamination which can result during processing, surface segregation of functional groups, and blooming of bulk additives to the surface. These problems can be remedied by synthesizingand extensively purifying a polymer just before use in a derivatization scheme. This synthetic approach has been suggested by Hammond et al. (25,28), reportedly with successful results. This method suffers however, because it requires the synthesis of a large number of polymers. An alternative approach to generate polymer surfaces containing known functional groups is proposed. This approach employs reagents known to selectively reduce or oxidize certain surface groups. For example, mild reduction with 0.1 M diborane/THF converts carboxylic acids to alcohols without affecting carbonyls (7,36).Extensive reduction with diborane will however, reduce carbonlys to alcohols. In comparison to an unreduced surface, subsequent derivatization of a mildly reduced surface should reveal an increase in alcohol derivatives and no change in carbonyl derivative concentrations. No carboxylic acids should be detected on the reduced sample with a carboxylic acid specific reagent. Because carboxylic acids are reduced only slightly faster than olefins, decreases in the surface olefin concentrationcould be expected after mild (BH&/THF reduction. Marked deviation from these anticipated results will raise serious questions concerning the selectivity of the derivatization reactions employed. Although this is an indirect approach to generating a standard polymer surface, diborane reduction/subsequent derivatization can provide supporting evidence for reaction selectivity. The results of chemical derivatization/ESCA analysis of PE-H, PE-Ar, and PE-Ar' (mildly reduced PE-Ar) are summarized in Table 111. Only the reaction of (pentafluoropheny1)hydrazineincorporates fluorine in the control sample, indicating that residual carbonyls could be present on PE-H. In contrast to PE-Ar, derivatives of the mildly reduced sample show an increase in surface hydroxyl concentration, a reduction in surface olefins, and no significant change in the amounts of carbonyls detected. No carboxylates are found on PE-Ar'. The reactions of PE-Ar and PE-Ar' with pentafluorobenzyl bromide (PFBBr) introduced surface fluorine concentrations that differ by an order of magnitude. Because PFBBr reacts slowly with alcohols (37) and possibly with amines, some derivatization of PE-Ar' with this reagent is expected. In contrast, the carboxylate salts of PE-Ar rapidly react with PFBBr, explaining the high F(ls) area measured
d)
PE-Ar-TFAA
i J I_ i(-
I .
285.8
Hg 4F C l
1 zT'8 ,nA I
P E - A r - H g (TFA)
F 1s
201.0
640. 7
1 A?VL&& Na 1s j A 285.8
r-\
2P
103.0
1
533 1
1074.0
---
BINDING ENERGY (E. V.) Flgure 3. ESCA spectra of PE-Ar after treatment with various tagging reagents. Ail spectra were taken with a DuPont 650B spectrometer at 330 W. Counts full scale (cfs),in thousand count units, for each core level is contained in brackets followingthe core designation (Le., C(1s) [2K cfs]): (a) C(1s) [32K cfs], q l s ) [32K cfs], and N(1s) 2K cfs] spectra of E - A r ; (b) C(1s) [16K cfs], qls) [4K cfs], N(1s) 2K cfs], and F(1s) [4K cfs] spectra of PE-Ar treated wlth pentafluorobenzaldehyde (PE-Ar-PFB: reaction 1); (c)C(1s) [8K cfs], qls) [8K cfs], N(1s) [2K cfs], and F(1s) [4K cfs] spectra of PE-Ar treated with 2,2,2-trlfluoroethanoiand DCC (E-Ar-TFE: reaction 2); (d) q l s ) [65K cfs], O(1s) [32K cfs], and F(1s) [32K cfs] spectra of E - A r treated wlth trifiuoroacetic anhydride (PE-Ar-TFAA: reaction 3); (e)C(ls) and K(2p) [8K cfs], q l s ) [8K cfs], and F(ls) [16K cfs] spectra of PE-Ar treated with pentafluorobenzyi bromide (PE-Ar-PFBBr: reaction 4); (f) C(1s) [l6 k cfs], O(1s) [8K cfs], N(1s) [2K cfs], and F(1s) [2K cfs] spectra of PE-Ar treated with (pentafluoropheny1)hydrazine (PE-ArPFPH: reaction 5); (g) C(1s) [16K cfs], Hg(4f) [4K cfs]. CI(2p) [2K cfs], and F(1s) [2K cfs] spectra of PE-Ar treated wlth mercuric trC fluoroacetateand 2,2,2-trlchloroethanol(E-Ar-Hg(TFA): reaction 7); (h) C(1s) [16K cfs], O(ls) [8K cfs], and Na(1s) [16K cfs] spectra of PE-Ar treated with alcoholic NaOH (PE-Ar(Na+): reaction 8).
1
670
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
Table 111. Derivatization of Control Polyethylene (PE-H), Polyethylene Oxidized by Exposure to Argon Plasma (PE-Ar),and PE-Ar Mildly Reduced for 3 h with Diborane/THF (PE-Ar')
Table IV. Angular-Dependent Area Ratios for CH,(ls), O(IS),and N(IS)of PE-N
or deg
C(lS)/
functional group
derivative
-OH
I-O,CCBr,
-c=o
I>C=NNHC,F, I-CO,CH,CF,
I
-CO,H
-C-Hg(CF,CO,)
9
tag atomic ratio PE-H PE-Ar PE-Ar' 170
127
64
34 22
43
32a
313
10
108
Mercury tag.
for PE-Ar treated with this reagent. Pentailuorobenzylethers will also be formed on PE-Ar but will be only a minor contribution to the F(1s) intensity of PE-Ar-PFBBr. The results of Table I11 are in good agreement with the expected chemistry resulting from mild diborane reduction and support the indicated selectivity of the derivatization reactions. Although a more detailed investigation of group specificity is required, Table I11 indicates that derivatization can extend the capabilities of ESCA for the detection of specific surface groups. Similar procedures could be useful for evaluating the selectivity of other chemical derivatizations. NaBH4will reduce carbonyls in the presence of carboxylic acids (38). Complementary studies with mild BH,/THF reductions, extensive BH,/THF reductions, and NaBH, reductions should prove helpful in further characterizing reaction selectivity. Reactions of a particular functional group with several different reagents could also be helpful in this respect. Chemical Derivatization of Plasma Modified Polyethylene. The previous section emphasized the importance of reaction specificity and presented preliminary evidence supporting the selectivity of several derivatization reactions. These reactions were used to compare the chemical functionality introduced on PE with Ar and with N2 plasma exposure. Small amounts of tag intensity are detected in control samples with reactions 3,5, and 8, but no tag signal is detected with reactions 1, 2, 4, 7, and 9. Under the conditions employed, reaction 5 is believed to be specific for aldehyde and ketone groups (7). In contrast, trifluoroacetic anhydride reacts with a variety of groups having an active hydrogen. Pentafluorobenzyl bromide reacts rapidly with carboxylate salts and more slowly with alcohols. Dicyclohexylcarbodiimide is an effective coupling agent to form esters from alcohol and carboxylate functions. The results of these reactions with PE-H indicate that residual carbonyl and possibly alcohol functions are present in PE-H. This is not surprising because P E is susceptible to slow environmental oxidation (39). The failure to detect tag intensity after reactions 1, 2, 4, 7, and 9 indicates that PE-H does not contain amine, carboxylic acid, alcohol, and olefin functions. Residual oxidation is further supported by the small O(1s) intensity measured in PE-H. It is known that oxidation of PE occurs during ESCA analysis with the DuPont 650 B spectrometer (40),and consequently, the extent of oxidation before analysis is smaller than that indicated by the C(ls)/O(ls) ratio of PE-H. The identification of alcohol groups is complicated by enol-keto tautomerization of carbonyls which is believed to
35 87 %AQ
CH,(ls)/ O(1S)
CH,(ls)/ N(1s)
O(ls)/ N(1s)
1.0 1.4 1.6 37
10 16 21 52
10 12 10 0
% change in area ratio measured at e = 9" and e = 87" ; integration error is estimated at ? 10%.
play an important role in the surface properties of oxidized PE (41). Numerous arguments supporting a keto-enol hydrogen bond association have appeared (26, 42). The enol form of residual carbonyl groups on PE-H could react with TFAA. Reactions of alcoholic NaOH with PE-H, P E A , and PE-N to form sodium carboxylate salts have been unsuccessful. Sodium incorporation is not reproducible and is highly dependent on the conditions used to wash the sample. Mild washing with wet 2-propanol rapidly removes the Na+ tag. Because the derivatized polymers have to be rigorously washed to ensure removal of absorbed reagent, covalent attachments have an intrinsic advantage over ionic derivatives. In this respect, ester formation of surface carboxylic acids is reproducible and results in a product that is stable to extensive Soxhlet extraction; sodium salt formation is not. For this reason, the results of reaction 8 are very suspicious. The results summarized in Table I1 indicate that surface amines, carboxylic acids, alcohols, carbonyls, and olefins are present on PE-Ar and PE-N. Not surprisingly, the surface amine concentration of PE-N is almost 5 times larger than the amine concentration of PE-Ar. PE-Ar shows an olefin concentration twice that found on PE-N. This result is also not surprising because an important pathway of amine formation is the reaction of NH2radicals with olefiis (43). Inert gas plasmas are most effective in producing surface crosslinking and surface unsaturation (11). The surface concentration of carbonyls on PE-N is twice that measured for PEA, whereas the carboxylic acid concentration of PE-Ar exceeds that on P E N almost 3-fold. Comparing the resulta of reaction 4 is complicated by the possible reaction of PFBBr with amines. It is tempting to use the results of Table I1 to compare and speculate on differences in the mechanisms of Ar and N2 plasma modifications. This generalization was presently not attempted because of problems associated with the quantitative analysis of modified polymer surfaces. In our studies with chemical derivatization of surface functional groups, it was found that quantitative analysis of a modified polymer surface is complicated by segmental mobility of polymer chains (35). This mobility results in the transport of functional groups to and from the ESCA sampling depth. Mobility will obviously affect interpretation of ESCA results. A germane observation is that solvent assistance is not required to change the concentration of surface functionality. This dynamic nature of a polymer surface raises important questions concerning the significance of a polymer surface analysis. Currently, methods to reduce this functional group mobility have been unsuccessful. Angular-Dependent Studies. Angular-dependentESCA measurements are well established as a qualitative way of distinguishing surface from subsurface regions. The principles of angular-dependent ESCA have been described (44). To understand qualitatively the vertical nature (surface subsurface) of the PE-N surface, we determined angular-dependent ratios of C(1s) for CH2,O(ls), and N(1s) transitions and summarized them in Table IV. The C(ls)/O(ls) and +
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
288.0
Table V. Time Dependence for the Reaction of PE-N with Pentafluorobenzaldehyde reaction C( Is)/ F( Is)/ O(Is)/ time, min 15 30 100 240
F( 1s) 4.2 3.6 2.4 2.3
n
5 e.v.
F(ls) 2.9 2.1 1.8 1.6
2.3 4.5 9.0 8.0
671
>
I-
H
C(ls)/N(ls) ratios increase with increasing 8, indicating that oxygen- and nitrogen-containing groups are preferentially located at the surface. This vertical inhomogeneity is interpreted as a gradient of functional groups whose concentration decreases with increasing depth into the polymer. The invariance of O(ls)/N(ls) with 9 suggests that the relative distribution of these elements is similar. This conclusion is further illustrated by the C(1s) spectra of PE-N recorded at different collection angles (Figure 4). Emission from 8 = 8 7 O contains only a small relative amount of oxidized carbon; grazing angle collection, 8 = loo, shows a relatively high amount of oxidized carbon, indicating that these groups are located closer to the surface. At 8 = 3 5 O , intermediate levels of high BE C(1s) are detected. By monitoring increases in the F(1s) signal during the formation of PE-N-PFB, we can determine the time required for complete derivatization of the amines contained in the ESCA sampling depth (e70 A). Table V summarizes important core level ratios for the reaction PE-N + PFB PE-N-PFB proceeding for four different times. In all ratios, total C(1s) intensity was used and not corrected for contributions arising from C(1s)-F of the PFB tag. Initially, large amount of fluorine is incorporated; however, no significant increase in F(1s) area develops after 100 min. The sampling depth is fully derivatized after 4 h of reaction time. To substantiate these results, we conducted angular-dependent studies with PE-N-PFB after reaction for 15 and 240 min. Table VI summarizes the results. In comparison t ,onger reaction times, treating PE-N for 15 min with PFB esta lishes a surface significantly rich in fluorine. The O(ls)/F(ls) ratio is increased by greater than 50% when collecting “bulk” rather than surface emission. Because the vertical distribution of oxygen and nitrogen groups is similar, this suggests that a large number of amines contained in the ESCA sampling depth have not been derivatized after 15 min. 0 0 s ) area was used because of the poor S/N of the N(1s) area in the angular collection mode. After 4 h of reaction, the O(ls)/F(ls) ratio is much less dependent on 8. The slight dependence of this ratio after 4 h reaction results from the vertical gradient of nitrogen and oxygen groups. By use of the F(ls)/N(ls) area ratio measured after 15 min of reaction time, it is calculated that approximately 30% of the amines in the first 50-70 A have reacted after 15 min.
-
9
Ln
Z
w Iz
H
d, ..L, c
BINDING
ENERGY (E. V . )
w e 4. Angular-dependentC(ls) spectra of PE-N. Emlssbn detected at (a)8 7 O , (b) 35O, and (c) 8 O , measured parallel to the surface. PaSS energy was 50 eV; each frame was scanned three times. Similar conclusions were obtained for reactions of PFPH, TFAA, and Hg(TFA)2with PE-Ar; ca. 1.5-2.0 h reaction time were required to saturate the ESCA sampling depth. Angular-dependent area ratios for PE-Ar-TFAA and for PE-ArPFPH are summarized in Table VI. The angular-dependent ratios of pertinent core levels of PE-Ar-Hg(TFA) are given in Table VII. These results illustrate the potential of angular-dependent measurements for the study of surface reactions in a polymer film. They help enormously in characterizing the vertical surface features of a plasma-modified polymer and allow for the study of reaction kinetics of groups contained in the outer 70 A of a polymer surface. Quantitation a n d Decomposition of Tags i n t h e DuPont 650B Spectrometer. Decomposition of polyethylene, polyacrylonitrile, and polytetrafluorcethylene(Teflon) occurs during ESCA analysis in the DuPont 650B spectrometer (40). Decomposition is characterized by reductions in F(1s) and CF2(ls)of Teflon during X-ray exposure. It is theorized that decomposition is initiated by what resembles a low-pressure plasma established above the sample. This plasma is the result of the high secondary electron flux characteristic of the DuPont spectrometer. Many of the derivatized surfaces studied were not stable to ESCA analysis in the DuPont 650B spectrometer, A discussion of sample instability is reserved for the conclusion of this manuscript, although sample decomposition will surely affect any quantitative analysis using derivatization methods. The several strong photoelectric transitions of Hg can be used to illustrate the dire effects that rapid decomposition
Table VI. Angular-Dependent Area Ratios of C(ls), O(ls), and F(1s) for Various Reactions of PE-Nand of PE-Ar sample PE-N-PFB PE-N-PFB
reaction time, min
15 240
PE-Ar-TFAA
30
PE-Ar -PFP H
60
-9, deg
C( ls)/F(Is)
O(ls)/F(Is)
9 35 87 9 35 87 9 35 87 9 35 87
2.5 4.1 6.1 2.0 2.8 3.3 2.5 3.4 4.7 6.0 5.9 6.5
1.8 3.4 4.0 1.4 2.0 1.9 1.3 1.3 1.5 1.8 1.6 1.6
C( ls)/O( 1s)
1.4 1.2
1.5 1.5 1.4 1.8 2.0 2.6 3.1 3.3 3.6 4.2
672
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
Table VII. Angular-Dependent Area Ratios of C( Is), F(ls), and Hg(4f7/2,5/2) of PE-Ar-Hg(TFA) Reacted for 2 h 0,
deg 9 35 87
Hg( 4f7/2*5/*)/ Hg(4f’/2-5/2)/ C(1s) 0.96 1.5 0.81 1.3 0.90 1.3
F( Is)/ (71s) 1.5 1.5 1.4
can have on any quantitative potential of the ESCA experiment. Table VI11 compares the normalized Hg 4f, 4d, and 4 p areas measured for PE-Ar-Hg(TFA)with the DuPont 650B and the PHI 548 spectrometers. Normalized areas, I,,,are expressed
I
I, = -
nU@A where n is the number of scans, I is the uncorrected ESCA area, u is Schofield’s (45) calculated cross section, @ is the angular correction necessary for limited solid angle detection (46)(using Fadley’s (44) notation, a is 67O for the DuPont 650B spectrometer and 90° for the PHI 548 spectrometer (47, 48)), and A is the mean free path as calculated from the theoretical principles outlined by Penn (49).It is seen that spectra taken with the PHI 548 spectrometer are useful for semiquantitative analysis (& 15% ) while rapid decomposition of PE-Ar-Hg(TFA)in the DuPont 650B spectrometerprevents this advantage from being realized for the DuPont 650B spectrometer. Modest reductions in X-ray power (9 kV; 30 mA) can often be helpful in attenuating sample decomposition. In favorable cases, stability was sufficiently enhanced to allow for semiquantitative analysis with the DuPont 650B spectrometer. To illustrate this point, the reactions of various fluorinated phenylhydrazines were used to compare experimental F(ls)/N(ls) area ratios with those predicted from the stoichiometry of the carbonyl tag. The quantitative expression used is _ -- IAUB@BABTBCB DA DB IBUA@AAATACA where D is the atom density, T the spectrometer efficiency for electrons of a given kinetic energy, and C is the attentuation factor for adsorbed comtamination on the sample. Other quantities have already been defined. AA/AB is estimated from A a (KE)0.75which, for KE 1 300 eV, is a good approximation to the more exact expression derived from Penn (49). It is assumed that TBCB/TACA-1.0 (47). The reactions of (4-fluoropheny1)hydrazine (MFPH), (2,3,5,6-tetrafluorophenyl)hydrazine(TFPH), and (penta-
fluoropheny1)hydrazine (PFPH) with PE-Ar were used to compare the experimental and the theoretical atomic F / N ratios. The data summarized in Table IX were acquired with the DuPont 650B spectrometer at reduced anode voltage (9 kV; 30 mA), keeping the total exposure time below 5 min. The theoretical sensitivity factor (uBAB/uAAA) for an F(ls)/N(ls) area ratio is 0.575; the average experimental sensitivity factor for each of the three reactions is 0.51. This agreement is within the expected limits for quantitative ESCA analysis. When the same experimentwas repeated at full anode voltage (>300 W) and more prolonged exposure (ca. 10 min), only *50% agreement between calculated and experimental sensitivity factors could be obtained. These results illustrate the adverse effects of decomposition on quantitative ESCA analysis and how reductions in anode power can often remedy this problem. The reaction of PFB with PE-N saturates the ESCA sampling depth after 1.5 h; no further increase in fluorine intensity develops in PE-N-PFB after reacting for this time, In view of this, the F(ls)/N(ls) area ratio obtained for the PE-N-PFB sample can be used to estimate the amount of nitrogen reacted with the PFB tag. A F/N ratio of 5.0 is expected if all of the original N(1s) area of PE-N results from amines which quantitatively react with the PFB tag. By use of the average experimental sensitivity factor for F(ls)/N(ls) area ratios measured with the DuPont 650 B spectrometer, the atomic F / N ratio for PE-N-PFB is calculated to be 6.1, 18% higher than that expected for a surface of fully derivitized amines. The average value for the F(ls)/N(ls) area ratio of PEN-PFB (formed under similar conditions) measured with the PHI 548 spectrometer is 8.5. With the theoretical sensitivity factor (0.575), F/N atomic ratios of 6.9 and 4.9 are calculated for the DuPont 650B and PHI 548 spectrometers, respectively. The value for the atomic F / N ratio of PE-N-PFB determined with the PHI spectrometer indicates that the amines on the surface =e fully derivatized. It is thought that decomposition of PE-N-PFB during ESCA analysis in the DuPont 650 B spectrometerincreases the surface fluorine concentration (vide infra). This would explain the fluorine rich, F/N atomic ratio measured for PE-N-PFB with the DuPont 650B spectrometer. Decompositionof PE-Ar-TFAA and PE-Ar-Hg(TFA). As previously mentioned, many of the derivatives studied were not stable to ESCA analysis. Figure 5a illustrates the behavior of the F(1s) signal of PE-Ar-TFAA as a function of X-ray exposure time in the DuPont 650B spectrometer. After 8 min, the F(1s)area has decreased by 50% of its original value. An induction period precedes the decay and its magnitude is dependent on the spectrometer vacuum. Reduced anode voltages attentuate the decomposition. Figure 5b shows the effect of Na+ in the decomposition of PE-Ar-TFAA(Na)+.
Table VIII. Normalizeda Hg Cross Section Areas for PE-Ar Reacted with Mercuric Trifluoroacetate-Trichloroethanol instrument DuPont 650B PHI 548c
Hg( 4 f 7 / f * 5 / 2 ) 31 1.9
Hg(4d5I2) 12 1.5
Hg(,,,I2) 14 1.4
Hg(4P) 10 1.9
Total exposure time was 5 min; each frame scanned a Z/nu@h; see quantitation section for explanation of symbols. Total exposure time was 20 min; each once in the order Hg(4f), Hg(4d5l2),Hg(4d3j2),and Hg(4p3’,). Power = 330 W. frame scanned twice in the same order as footnote b. Power = 300 W. Table IX. Theoretical and Experimental Atomic Ratios of Several (Fluoropheny1)hydrazine Tags Reacted with PE-Ara tag
C( ls)/F( 1s)
F( ls)/N(Is)
DF/DN(exptl)
DF/DN(theoretical)
C,F,H,NHNH, C,F,HNHNH, C,F,NHNH,
16.5 12.6 10.8
1.1 3.8 4.6
0.63
0.50 2.00 2.50
2.2 2.7
a All spectra were acquired with the DuPont 650B spectrometer at 9 kV and 30 mA; X-ray exposure time was kept below 5 min.
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4,APRIL 1981
I
a>
F
1s (4)I
F
i
1691. I 1
t tH
C lS(5)F lS(5 Na l S ( 3 )
lS(4) .
673
1
I
.
U
57.".
5 e.v.
1691.0
t
H
v,. Z W
cr, Z w
t-
z
t-
Z H
L
.
k
&
A
d
d
h
U
SINDING ENERGY -
c
H
cn Z W
t-
Z
H
L 690.8
t
1690.6
t
b--
BINDING ENERGY (E. V . ) Figure 5. Decomposition of PE-Ar-TFAA during ESCA analysis: (a) F(1s) spectra of PE-Ar-TFAA after (1) 2 min, (2) 8 min, and (3)22 min of exposure with the DuPont 6508 spectrometer; (b) F(1s) spectra of PE-Ar-TFAA(Na+) after (1) 2 min, (2) 8 min, and (3)22 min of exposure with the DuPont 650B spectrometer. All spectra were acquired at 330 W.
Accompanying decay of the organic F(1s) (691.2 eV) signal is the development of an inorganic fluoride (686.8 eV) signal. The area of this latter signal increases during the first 8 min but levels off after longer exposure times (15-25 min). The binding energies of the CH2(ls)and fluoride F(1s) signals remain invariant; however, the binding energy of the organic F(1s) signal decreases during the decay by 0.6 eV. This process is accompanied by an increase in analyzer pressure, rising from 2 x IO-' to 1 X lo4 torr. These observations are analogous to those for the decomposition of Teflon (40). We believe that the fluoride eventually results in the evolution of HF. The formation of NaF is not substantiated by either the N a b ) or F(1s) binding energies. Presently, details of the interaction of Na+ during the decomposition are not known. Decomposition is much more prevalent in spectrometers which achieve only modest vacuum levels. The ESCA spectrum of Teflon acquired with the PHI 548 spectrometer (pressure ca. 5 X lo4 torr) shows no detectable decomposition after 30 min of exposure (40). A similar situation occurs in the analysis of PE-Ar-TFAA and PE-Ar-TFAA(Na+). Figure 6 contrasts the ESCA spectra of PE-Ar-TFAA(Na+)recorded with the PHI 548 and with the DuPont 650B spectrometers. No noticeable decomposition is evident in the PHI system; decomposition in the DuPont 650B spectrometer is obvious. Important differences in the corrected Na(1s) binding energies (reference to 285.0 eV) are noticed: 1073.6 eV and 1071.2 eV with the DuPont and PHI spectrometers, respectively. Compared to the value of 1070.9 eV for the Na(1s) of sodium acetate (50),the binding energy of 1071.2 eV measured for
BINDING ENERGY (E. V . 1 Flgure 6. Comparison of the ESCA spectra of PE-Ar-TFAA(Na+) acqulred with the PHI 548 and DuPont 650B spectrometers. (a) C(ls), F(ls), and Na(1s) spectra taken with PHI 548 spectrometer (pass energy was 25 eV, 300 W). Total exposure time was 15 min. (b) C(ls),F(ls),Na(ls), and O(1s) spectra taken with DuPont 6508 spectrometer, total exposure time was 9 min. Power = 330 W.
PE-Ar-TFAA(Na+) with the PHI 548 spectrometer is in agreement with the expected group (Na+ salt of carboxylate group). The BE value of 1073.6 eV measured in the DuPont 650B spectrometer, however, is unusually high and out of the range reported for most sodium-containing compounds (50, 51). To elucidate the chemical nature of this sodium species, we studied its X-ray excited KLL Auger transition. Auger transitions often show binding energy shifts that are larger than their ESCA counterparts and are particularly sensitive to the polarizability of the immediate atomic environment of the emitter. Wagner (51)has recently reviewed the distinct advantages that certain Auger lines can offer over their ESCA counterparts. Figure 7a shows the F(1s) and F(KLL) Auger spectra of PE-Ar-TFAA recorded with the DuPont 650B spectrometer. During decomposition,only organic fluorine is detected. The F(KLL) transition is broadened because the final L hole occurs in the bonding orbitals (52). Further evidence for the formation of fluoride during decomposition of PEAr-TFAA(Na+) is presented in Figure 7b. The F(KLL) Auger spectrum is a composite of a sharp and an appreciably broad component. In contrast to covalently bonded atoms, ionic species produce relatively sharp Auger transitions when the final hole develops in a valence orbital. The binding and kinetic energies for F(ls), Na(ls), F(KLL), and Na(KLL) transitions of PE-ArTFAA(Na+) acquired with the PHI 548 and DuPont 650B spectrometers are used to calculate and contrast the modified Auger parameters (51),a', listed in Table X. This parameter combines photoelectron and X-ray excited Auger lines to increase the utility of ESCA for identifying chemical states. Values of a' for NaF are included for comparison.
ANALYTICAL CHEMISTRY, VOL. 53, NO. 4, APRIL 1981
674
a)
IF lS(4) [ F KLL(2)
r”il
690.5
602.5
w Z H
Ig4F (1)F 1s (1 F 1s (14 Hg4F (1) 103,. 2 1
w
5 e.v.
I---
5 e.v
uu
BINDING ENERGY (E. b)
Ah 690‘5 102.7
V.)
BINDING ENERGY (E. V . )
b) Hg4F(2 F lS(2 F 1s (21 Hg4F ( 2 ) 106. 5 106.5
-F 1SQ)
692.7
600.0 I
I
1
A%k 692.7
5T.v
265.5
B I N D I N G ENERGY (E. V .
BINDING ENERGY (E.V .
)
Flgure 7. Comparison of the fluorine ESCA and fluorine Auger spectra of PE-Ar-TFAA and PE-Ar-TFAA(Na+), (a) F(1s) ESCA and F(KLL) Auger spectra of PE-Ar-TFAA taken with DuPont 6508 spectrometer. (b) F(1s) ESCA, F(Kii) Auger, and Na(KLL) Auger spectra of PE-ArTFAA(Na+) taken with DuPont 6508 spectrometer. All spectra were acquired at 330 W.
Table X. Modified Auger Parameter, a‘,for Sodium and for Fluorine of PE-Ar-TFAA(Na+)and of NaF and Comparison of Values from DuPont 650B and PHI 548 Spectrometers sample a‘ DuPont 650B PHI 548 PE-Ar-TFAA(Na+) 2062.6 2060.5 (1073.6)“ a ’ ~ 1342.1 LY’F
NaF “IF
(689.3)” 1340.4 (684.8)‘ 2060.7b 1339.4b
(1071.2)a 1341.9 (688.6)” 2059.7c 1339.4c
a Corrected binding energy in eV referencing CH, of PE Sample prepared by evaporating 1 drop of to 285.0 eV. an alcoholic NaF solution on a PE film mounted on sample probe. Differential charging prevented accurate BE
measurements.
Values taken from “PHI Handbook of
Photoelectric Spectroscopy” (48).
Values of d determined from spectra acquired with the PHI 548 spectrometer are consistent with the expected groups of PE-Ar-TFAA(Na+)and NaF. The analysis of NaF with the DuPont 650B spectrometer similarly gives reasonable values for a’. However, in reference to Wagner’s two-dimensional chemical state plots (51),the a’ (Na+) measured for PE-ArTFAA(Na+) with the DuPont 650B spectrometer is significantly different from most values reported. It more closely resembles that expected for NazO. This finding is interesting in view of an observation made by Wendt et al. (53)who studied the oxidative stability of various carbon fibers. These
)
Flgure 6. Hg(4f7/z*s’2) and F(1s) areas of E-Ar-Hg(TFA) as a functlon of X-ray exposure. (a) Hg(4f7/2*5’2)after 1 and 5 min of exposure (1 and 4, respectively), and F(1s) after 2 and 4 min of exposure (2 and 3, respectively) with DuPont 6508 spectrometer (330 W). (b) Hg(4f7/z05’2) after 3 and 10 min of exposure (1 and 4, respectively) and F(1s) after 5 and 12 min of exposure (2 and 3, respectively) with a PHI 548 spectrometer (pass energy was 25 eV); ail spectra were acquired at 300 W.
authors concluded that trace Na+ (as Na2S04)present in the polymers acted to decrease their stability and appeared to be catalytic in the degradation process. It was hypothesizedthat NazS04was initially converted to NazO which then acted as the catalytic agent. Numerous examples of similar catalytic roles of metal oxides are known (39). Shirley et al(54) recently reported that certain alkali salts are converted to the corresponding oxides during irradiation with 0.3-1.4 keV electrons, It was suggested that oxidation may proceed via reduction of M+ to MO, which subsequently reacts to form the metal oxide. Although the details are different from those reported here, this supports the suggestion that a low-pressure plasma induces decomposition and would explain the catalytic effect of sodium and the unusual sodium species observed. Because an oxide O(1s) signal is not evident, some other form of Na+ is suggested; however, its nature is unknown. A survey scan did not help to identify the anion associated with the Na+ present on PE-Ar-TFAA(Na+). The value of a’ (Na+) indicates association with a highly polarized anion. Sodium carbide is a possible candidate. An alternative explanation for the unusual value of a’(Na’) determined from the spectrum of PE-Ar-TFAA(Na+)with the DuPont 650B spectrometer is considered. The value of a’ (Na+) reported by Wagner (51) was measured for granular NaF crystals. The extra atomic environment of Nat ions should differ for NaF in a homogeneous NaF crystal and a single NaF molecule isolated in a PE matrix. Because Auger shifts are dependent on extraatomic relaxation, the a‘ (Na’) for NaF in significantly different environments should not be identical. This seems the more likely explanation for the
ANALYTICAL CHEMISTRY, VOL. 53,
)F
KLL(1) H
5 e.v
1 600.5
BINDING ENERGY (E.V . )
NO. 4,
APRIL 1981
675
occur. With 8 rnin of X-ray exposure, the Cl(2p) area has increased by 250% of its value after 3 min of exposure. The increases in F(ls) and Cl(2p) areas indicate that decomposition occurs below the ESCA sampling depth. (Control polyethylene and Au samples run immediately after PE-Ar-Hg(TFA)K+did not show any signs of fluorine or chlorine contamination). One interpretation of these events is that decomposition of derivatized alkenes below the sampling depth produces fluoride ions. Being insoluble in the PE matrix, Fmigrates to the surface. The observed increases in analyzer pressure suggest that evolution of HF may occur. This is in agreement with the decomposition of Teflon (40).
ACKNOWLEDGMENT The authors thank M. Umafia for her many helpful discussions and editorial suggestions. LITERATURE CITED
BINDING ENERGY ( E . V .
)
Figure 9, Effects of K+ on the decomposition of PE-Ar-Hg(TFA)-K+ during ESCA analysis in the DuPont 6508 spectrometer: (a) F(1s) spectra of PE-Ar-Hg(TFA)-K+ afler (1) 2 min, (2) 4 mln, and (3) 9 mln of X-ray exposure and (4) fluorine KLL Auger line of PE-Ar-Hg(TFAbK+ after 10 min of X-ray exposure; (b) CI(2p) spectra of PE-Ar-HS(TFA)-K+ afler (1) 3 min and (2) 8 min of X-ray exposure.
results contrasted in Table X. One additional point should be made. Decomposition in the DuPont 650B spectrometer is a common problem observed to some extent in all the modified polymers listed in Table 11. It does not seem to be associated with any particular group, is dependent on spectrometer vacuum, and is catalyzed by sodium. It can be significantly reduced by modest reductions in anode voltage and by limiting sample exposure to X-rays. Rates of sample decomposition vary considerably. PE-ArHg(TFA) and PE-N-Hg(TFA) are extremely prone to rapid decomposition in the DuPont 650 B spectrometer. Figure 8 contrasts the behavior of the Hg(4f) and F(1s) signals of PEAr-Hg(TFA) during analysis in the DuPont 650B and PHI 548 spectrometers. In the DuPont system, only 4 rnin is required to reduce the Hg(4D area to 50% of its original value; similar reductions in the F(ls) signal area are observed. The analogous experiment with the PHI 548 spectrometer shows no changes in either core-level area, indicating that no decomposition has occurred. K+ accelerates the decomposition of PE-Ar-Hg(TFA). Figure 9 illustrates the F(1s) and Cl(2p) spectra of PE-ArHg(TFA)K+ during analysis in the DuPont 650B spectrometer. A fluoride F(ls) signal develops after 2 min of exposure, and the total F(1s) signal area increases rapidly with X-ray exposure. After 4 min of exposure, the F(1s) area has increased to 500% of its value after 2 min of exposure. With 9 min of X-ray exposure, the F(ls) area levels off at 900% of its value in Figure 9a(l). The fluorine KLL Auger line after 10 min of exposure (Figure 9a(4)) shows evidence for a number of inorganic fluoride components. The Cl(2p) area behaves similarly, although no evidence for changes in chemical state
(1) Lee, L. H., Ed. “Advances In the Characterlzatlon of Polymer and Metal Surfaces”; Academic Press: New York, 1976. (2) Clark, D. T., Feast, W. J., Eds. ”Polymer Surfaces”; Wlley: New York, 1978. (3) Lee, L. H., Ed. “Adhesion Science and Technology”; Plenum Press: New York, 1975. (4) Blab, P.; Darlson, D. J.; Cosllog, G. W.; Wiles, D. M. J . ColloMInterface Scl. 1974, 47, 636-649. ( 5 ) Brlggs, D.; Brewis, D. M.; Konieczko, M. B. J. &fer. Sci. 1973, 7 7 , 1270-1277. (6) Kato, K. J. Appl. folym. Sci. 1974, 78, 3087-3094. (7) WhiiesMes, 0.M.; Rasmussen, J. R.; Stedronsky, E. R. J . Am. Chem. SOC.1977, 99, 4736-4745. (8) Benham, J. V.; Pullukat, T. J. J . Appl. Polym. Scl. 1978, 20, 3295-3311. (9) Johnson, M.; Wllllams, M. E. Eur. folym. J . 1976, 72, 843-846. (IO) Burfield, D. R.; Law, K. S. Polymer 1979, 20. 620-626. (11) Bell, A. T., Hollahan, J. R., Eds. “Techniques and Appllcatlons of Plasma Chemistry”; Wlley: New York, 1974. (12) Clark, D.T.; Dllks, A. J. J. folym. Scl., folym. Chem. Ed. 1978, 16, 957-976. (13) Yasuda, H.; Marsh, H.C.; Brandt, E. S.; Reilley, C. N. J . Polym. Scl. folym. Chem. Ed. 1977, 75, 991-1091. (14) Yasuda, H.; Hsu, T. S.; Brandt, E. S.; Rellley, C. N. J . fokm. Sci., folym. Chem. Ed. 1978, 76, 415-425. (15) Yasuda, H.; Marsh, H. C.; Brandt, E. S.; Rellley, C. N. J. Appl. folym. Scl. 1976, 20, 543-555. (16) Yasuda, H.; Morosoff, N.; Brandt, E. S.; Rellley, C. N. J . Appl. folym. Sci. 1979, 23, 1003-1011. (17) Morosoff, N.; Yasuda, H.; Brandt, E. S.; Rellley, C. N. J . Appl. folym. SCi. 1979, 23, 3449-3470. (18)Morosoff, N.; Yasuda, H.; Brandt, E. S.; Rellley, C. N. J . Appl. folym. sci. 1979, 23,3471-34a8. (19) Clark, D. T.; Dllks, A. J. J. Polym. Chem. Ed. 1977, 75, 2321-2345. (20) Slegbahn. K. et al. “ESCA, Atomic, Molecular and Solid-State Structure Studied by Means of Electron Spectroscopy”; Almqulst and Wlksells: Uppsala, 1967. (21) Clark, D. T., In ref 1. (22) Clark, D. T., Chapter 16 In ref 2. (23) Clark, D. T. A&. folym. Sci. 1977,24. (24) Batich, C. D.; Wendt, R. W. 179th Natlonal Meeting of the American Chemical Soclety, Houston, TX, 1980; American Chemlcal Society: Washington DC, 1980;POLY 101. (25) Hammond, J. S. 179th National Meetlng of the Arnerlcan Chemlcal Society, Houston, TX, 1980 POLY 118. (26) Brlggs, D.; Kendall, C. R. Polymer 1979, 20, 1053-1054. (27) Riggs, W.; Dwight, D. J . Electron Specfros. Relat. fhenom. 1974, 5 ,
447-453. (28) Hammond. J. S.; Holubka, J. W.; Durisln, A. M.; Dickie, R. A. 176th Natlonal Meeting of the Amerlcan Chemical Society, Honolulu, 1979; American Chemical Society Washington, DC, 1979;Division of Polymer Chemistry.
(29) Umah, M.; Murray, R. W., Unlverslty of North Carollna (Chapel HIII), private communlcatlon, 1979. (30) Smith, D. R.D. Thesis, University of North Carolina (Chapel Hill), 1978. (31) Evans, J. F.; Kuwana, T. Anal. Chem. 1979, 57, 358-365. (32)Clark, D. T.; Thomas, H. I?.J. folym. Scl., folym. Chem. Ed. 1978, 76,791-820. (33) Brundle, C. R., Baker, A. D., Eds., “Electron Spectroscopy-Theory, Technlques and Applications”; Academic Press: New York, 1978 p 23. (34) Hollahan, J. R.; Stafford, B. B.; Falb, R. D.; Payne, S. T. J. Appl. folym. Sci. 1969. 73,807-816. (35) Everhart, D. S.; Rellley, C. N., submltted for publlcatlon In Surf. Interface Anal.
(36) Brown, H. C.; Korytnyk, W. J. Am. Chem. SOC. 1960, 82, 3866. (37) Blau, K., Klng, G. S., Eds. Handbood of Derivatives for Chromatography”; Heyden: London, 1977. (38) Fleser, L.; Fieser, M. ”Reagents for Organic Synthesis”; Wlley: New York, 1967;Vol. 1. (39) McKellar, J. F.; Allen, N. S. ”Photochemistry of Man-Made Polymers”; Applied Sclence Publishers: London, 1979.
676
Anal. Chem. 1981, 53, 676-680
(40) Everhart, D. S.;Reiiiey, C. N., unpublished results. (41) Owens, D. K. J. Appl. Polym. Scl. 1075, 19, 265-271. (42) Briggs, D.; Rance, D. G.; Kendaii, C. R.; Biythe, A. R. Polymer 1080, 21, 895-900. (43) Demissy, M.; Lescalux, R. J. Am. Chem. SOC. 1080, 702, 2897-2902. (44) Fadiey, C. S. J. Electron Spectrosc. Relat. Phenom. 1074, 5 , 725-754. (45) Scofieid, J. H., Report No. UCRL-51326; Lawrence Livermore Laboratory: Berksley, CA, 1973. (46) Reiiman, R. F.; Msezane, A.; Manson, S. T. J. Electron Spectrosc. Reht. Phenom. 1076, 8 , 389-394. (47) Moses, P. R.; Wier, L. M.; Lennox, J. C.; Finkiea, H. 0.; Lenhard, J. R.; Murray, R. W. Anal. Chem. 1078, 50, 576-585. (48) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F. "Handbook of X-ray Photoelectron Spectroscopy"; Muilenberg, G. E., Ed.; Perkin-El-
mer Corp., Physical Electronics Division: Eden Prairie, MN, 1979. (49) Penn, D. R. J. Nectron Spectrosc. Relat. Phenom. 1976, 9 , 29-40. (50) Wagner, C. D. Faraday Discuss. Chem. SOC. 1975, 60, 291-300. (51) Wagner, C. D.;Gale, L. H.; Raymond, R. H. Anal. Chem. 1079, 57, 466-482. (52) Wagner, C. D. Anal. Chem. 1072, 4 4 , 967-973. (53) Wendt, R. C.; Gibbs, H. H.; Wilson, F. C. Polym. Eng. Scl. 1979, 19, 342-349. (54) Shirley, D. A.; Sasaki, T.; Williams, R. S.; Wong, J. S. J. Chem. Phys. 1070, 71, 4601-4610.
RECEIVED for review May 9,1980. Accepted January 13,1981. The financial assistance of Hercules, Inc., is greatly appreciated.
Determination of Light Rare Earth Elements in Apatite by X-ray Fluorescence Spectrometry after Anion Exchange Extraction Iwan Roelandts Department of Geology, Petrology and Geochemistry, University of Liege, B 4000 Sart Tilman, Liege 1, Belgium
The results of an investigation, the purpose of which was to study the feasibility of analysis of La, Ce, Pr, and Nd in apatite minerals, utilizing X-ray fluorescence spectrometry (XRF), are discussed. After sample dissolution in HNO,, rare-earth elements (REEs) are separated by a mixed solvent anion exchange using the batch extraction technique. The anion exchanger employed is the strongly basic resin Dowex 1x8, 100-200 mesh (nitrate form). REEs are fixed on the resin from a medium consisting of 95% CH,OH-5% 7 M HNO, (v/v). The dried resin is uniformly spread over a disk of self-adhesive foil backed with a cellulose support. The thin resin film formed is then used for XRF analysis. Synthetic standards prepared following the same procedure are employed for calibration. Anaiytlcal results of two apatite samples are presented. Good correlation Is obtained between XRF and neutron activation analysis data. Limits of detection, precision, and accuracy of the results are discussed.
Over the last few years, quantitative mathematical models of trace-element partitioning in igneous rocks have been proposed (1).Because of their mutually similar behavior, the rare-earth elements (REEs) are sensitive indicators of magmatic differentiation processes (2).Several analytical techniques have been developed for the determination of these elements that are of geochemical interest. Trace amounts of REEs in geological materials are usually obtained by neutron-activation analysis (NAA) (with or without radiochemical separations) (3-6)or by mass spectrometry [isotope dilution (MSID), spark source (SSMS)] (7-12). However, all these methods are delicate and depend on the ability of the geoanalyst. Furthermore, these specialized instruments are costly and are not available in the majority of geochemical laboratories. Other methods of analysis such as optical emission (OS) (13),atomic absorption spectrometry (AAS) (13,14),X-ray fluorescence (XRF) (15-17), electron microprobe (EM) (18),charged particle induced X-ray emission spectroscopy (PIXE) (19), and inductively coupled plasma spectroscopy (ICP) (20)have also been applied under favorable circumstances. The use of XRF analysis has rapidly increased in the geosciences and is now an accepted method for the accurate
determination of many major and minor constituents. This technique can also be extended to some trace elements in common silicate rocks. In order to enhance both sensitivity and accuracy, chemists have proposed different procedures for preliminary chemical enrichments and separations of the trace elements from the original sample for use with XRF analysis, e.g., precipitation (15),solvent extraction (21),and ion exchange [resins such as Amberlite and Chelex-100 or filters such as membranes or resin-loaded papers supplied by Whatman and Gelman-Hawksley (17,21-31)]. Selectivity and reproducibility of the separation method and uniformity of the target exposed to the X-ray beam are factors that should be considered. The method of ion exchange is rapid and provides good separations and the low atomic numbers of the elements constituting the resins make them suitable for X-ray measurements. Automated ion-exchange separations permit simultaneous sample analyses. This is important in analytical geochemistry where multiple analyses are required to obtain statistically valid results. With regard to the ion exchange methods, the batch equilibration process provides some advantages over the column technique: it is faster, a smaller quantity of resin is required, and a homogeneous distribution of the ion of interest on the resin is achieved, which simplifies sample handling prior to analysis. In the present work, the possibility was investigated of combining a preliminary collection of REE group by a mixed-solvent anion exchange technique (32)and subsequent use of XRF spectrometry for the direct determination of these elements on the resin. Apatite is known to contain amounts of REE on the order of tenths of a percent or higher. Two apatite samples of different REE concentrations, previously analyzed by NAA, are analyzed here.
EXPERIMENTAL SECTION Apparatus, Batch equilibrations were performed by means of a Turbula system Schatz WAB shaking machine. Measurements of the pH of solutions were made with a Beckman pH meter, Model H-2. Absorbance measurements were carried out in a Beckman spectrophotometer, Model B, using 1-cm cells. A stainless steel mold and a hydraulic press (Weber, StuttgartUhlbach, Germany) were used for pelletizing the cellulose supports. All X-ray measurementswere performed by using a CGR-alpha 2020 semiautomatic spectrometer (Compagnie GBnBrale de Ra-
0003-2700/81/0353-0676$01.25/00 1981 American Chemical Society