Quantitative determination of paint additives with fast atom

Oct 1, 1984 - T. L. Riley, T. J. Prater, J. L. Gerlock, J. E. DeVries, and Dennis. ... Charles L. Wilkins , Jeremiah D. Hogan , Tracy L. Donovan , Dav...
0 downloads 0 Views 412KB Size
Anal. Chem. 1984, 56,2145-2147 (15) Ross, M. M.; Kidwell, D. A,; Colton, R. J., submitted for publication. (16) Ryan, T. M.; Day, R. J.; Cooks, R. G. Anal. Chem. 1980, 52,

2054-2057. (17) Cook, K. D.; Chan. K. W. S. I n t . J . Mass Spectfom. Ion Proc. 1983, 5 4 , 135-149. (18) Dang, T. A.; Day, R . J.; Hercules, D. M. Anal. Chem. 1984, 56, 866-871. (19) Unger, S.E.; Ryan, T. M.; Cooks, R. G. Anal. Chim. Acta 1980, 778, 169-174. (20) Busch, K. L.; Unger, S. E.; Vincze, A.; Cooks, R. G.; Keough, T. J . Am. Chem. SOC. 1982, 704, 1507-1511. (21) Day, R. J.; Unger, S. E.; Cooks, R. G. J . Am. Chem. SOC.1979, 707, 499-501.

2145

(22) Pierce, J. L.; Busch, K. L.; Cooks, R. G. Inorg. Chem. 1982, 21. 2597-2602. (23) Wheeler, 0.H. J . Chem. Educ. 1968, 4 5 , 435-437.

RECEIVED for review April 2, 1984. Accepted May 25, 1984. The National Research Council is acknowledged for its support of M. M. Ross and D. A. Kidwell as Resident Research Associates. This work was supported by the Office of Naval Research and the Army Chemicaland Research Office.

Quantitative Determination of Paint Additives with Fast Atom Bombardment Mass Spectrometry T. L. Riley,* T. J. Prater, J. L. Gerlock, J. E. deVries, and Dennis Schuetzle Ford Motor Company, Research Staff, Dearborn, Michigan 48121

Quantitative fast atom bombardment (FAB) mass spectrometry methods were Investigated for the analysis of the photostabllizer bis(2,2,6,6-tetramethyi-4-piperldlnyl) sebecate (Tlnuvln 770) in cured coating systems. External and Internal standardization procedures using both isotopically labeled and nonlsotoplcally labeled internal standards were attempted. Excellent accuracy ( f 10 % ), precision (RSD < 5 % ), and detection llmlts (1 ng) were obtained for the isotopically labeled Internal standardizationprocedure. Shortcomings of the external standardization and nonisotoplcaliy labeled procedures are discussed.

In the brief period since its introduction in 1981 ( I ) , fast atom bombardment mass spectrometry has become a widely used technique for the analysis of high molecular weight and/or thermally labile compounds (2). The fast atom bombardment ionization process appears to be identical in mechanism with the more well established secondary ion mass spectrometry (SIMS) ionization process (3, 4). The SIMS technique, however, is generally used to examine the surface of a solid sample directly, whereas FAB is generally used to desorb ions from a sample suspended in a liquid matrix like glycerol. For the analysis of organic molecular compounds, the liquid matrix effectively provides a renewable sample surface which greatly reduces primary beam surface damage problems. This permits an extended period of analysis and allows the use of higher primary particle currents which improves sensitivity. It would also appear that the liquid matrix would greatly simplify quantitative analysis since internal standards can be easily incorporated into this medium. Relatively few studies have been made on the feasibility of quantitative FAB analysis. Gaskell et al. ( 5 ) ,Millington e t al. (6), Murphy et al. (7),DeStefano et al. (8),and Ho et al. (9) have described quantitative FAB analysis procedures using isotopicaly labeled internal standards of steroid sulfates, acylcarnitines, blood platelet activating factor, surfactants, and dipalmitoylphosphatidylcholine, respectively. Teeter (10)has described an external standardization procedure for the quantitative FAB analysis of alkylbenzene and aliphatic sulfonate salts.

Recently in our laboratory the need arose to develop a quantification procedure to monitor the photostabilizer additive bis(2,2,6,6-tetramethyl-4-piperidinyl)sebecate (Tinuvin 770) (11)in cured paint as a function of weathering time. This paper describes a systematic study in which quantitation of this compound by FAB mass spectrometry was investigated using external, nonisotopically labeled internal, and isotopically labeled internal standardization procedures.

EXPERIMENTAL SECTION All FAB analyses were performed on a Vacuum Generators ZAB-2F mass spectrometer interfaced to an INCOS 2300 data system. The mass spectrometer was equipped with a Vacuum Generators modified Ion Tech, Ltd., B50 power supply and a B11 saddle field ion source as a primary atom gun. Xenon was used as a bombarding gas with an energy of 8 kV and at a neutral atom “current”of approximately 20-30 WA(1mA power supply limiting current). Industrial grade Tinuvin 770 obtained from Ciby-Geigy,Ltd., was triple-recrystallized from ethanol prior to use. The tetradeuterated analogue of Tinuvin 770, bis(2,2,6,6-tetramethyl-4piperidinyl) 2,2,9,9-tetradeuteriosebecate,was synthesized by transesterification of the dimethyl ester of 2,2,9,9-tetradeuteriosebasic acid (Sigma Chemical Co.) with 4-hydroxy2,2,6,6-tetramethylpiperidine (Aldrich Chemical Co.). Tricyclohexyl citrate (2-hydroxy-1,2,3-tricarboxylicacid tricyclohexyl ester) was used as supplied (Chem. Service, Inc.). Two coating systems were tested a thermosetting acrylic melamine (12) and a hydroxy ester melamine (13). Glycerol used as a FAB matrix was diluted to 150 pg/kL in methanol to alleviate problems associated with handling small volumes of material. Tinuvin 770 was extracted from a known weight (approximately 40 mg) of cured or uncured coating sample by soaking in 5.0 mL of methylene chloride for approximately 15 h. In internal standard experimenk a known amount of tricyclohexyl citrate or deuterated Tinuvin 170 was also added to the mixture at the beginning of the extraction. The external standard FAB analysis was performed by placing 2.0 FL of extract or standard solution onto a 2 X 5 mm gold sample probe tip, allowing it to dry, then adding approximately 2 pL of glycerol/methanol solution (300 pg of glycerol). The sample and glycerol were mixed and most of the methanol was allowed to evaporate prior to insertion into the FAB source. The internal standard FAB analysis was accomplished in a similar manner except that only an approximate volume of sample and standard was initially added to the probe tip. Quantification was achieved by integrating the relevant ion

0003-2700/84/0356-2145$0 1.50/0 0 1984 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 56, NO. 12, OCTOBER 1984

2146

1

4811

4L,

1

i

Table I. Recovery and Precision of the Deuterated Internal Standard Procedure QIQ QIQ Recovery Rel.Std.Dev.

T l N U V i N - 770

60

%I

342

Amount (ne)

Number Samples

- -

Polymer-free Matrix

94

4.1

200

7

Uncured Polymer Extract

96

3.4

100-500

10

Cured Polymer Extract

80

2.4

100-200

8

20

212

1 8

0

326

l " ' ' ' ~ " ' 1 " ' ~ ' ' ~ ' ' l 1 ' " ' "

100

200

300

400

I] 500

rn / a

Flgure 1. FAB glycerol.

mass

spectrum of 100 ng of Tinuvin 770 in neat

currents associated with the sample and internal standard (when present) over the lifetime of the sample signal which was generally on the order of 3-4 min.

RESULTS AND DISCUSSION Tinuvin 770 proved to be well suited for FAB analysis in a glycerol matrix. The addition of reagents to the glycerol to enhance the surface activity or cationization efficiency of Tinuvin 770 was not found to be necessary. One nanogram of Tinuvin 770 provided a signal-to-noise ratio of approximately 2 to 5. The spectrum shown in Figure 1 was obtained from approximately 100 ng of Tinuvin 770. The most intense ions in the spectra derive from the (M + 1)+parent ion at mass 481 and the fragment ions a t mass 140 and 342. The origin of these ions was not thoroughly investigated. The mass 342 fragment is best explained by a McLafferty-type rearrangement of the ion produced by the loss of a piperidinyl ring from the (M + 1)' parent species. The lower molecular weight fragment could be accounted for by the piperidinyl ring itself through simple cleavage from the (M + 1)+parent species. This is somewhat unexpected since formation of an odd electron fragment ion from an even electron (M + 1)' parent species would require the energetically unfavorable splitting of an electron pair (14). As shown in Figure 1, the spectra were generally free of glycerol cluster ions. The surface activity of Tinuvin 770 appears to be adequate to suppress sputtering of the glycerol matrix. Quantification of Tinuvin 770 was first attempted by using an external standardization procedure. An excellent linear relationship initially was obtained (slope = 2.8, y intercept = 4.6, correlation coefficient = 0.9992) between integrated response and weight of Tinuvin 770 over the 2-200 ng range. This indicated that external standardization is at least feasible and that the concept of quantification by integrating the Tinuvin 770 ion current over the lifetime of the sample is sound. However, the reproducibility of the external standard curve proved to be very difficult. Response varied by over 50% in a neat glycerol matrix on a day-to-day basis. Analysis of coating extracts containing known amounts of Tinuvin 770 demonstrated that minor changes in the sample matrix also had significant effects on response. Response also was affected by minor changes in source tuning and even the manner in which the sample was distributed on the probe tip. These latter effects preclude the possibility of accurate quantification by a standard addition procedure. To overcome these problems in reproducibility, quantification was next attempted by using an internal standardization procedure. Tricyclohexyl citrate was chosen as an internal standard because its response (signal intensitylweight of analyte) was similar to Tinuvin 770 in a neat glycerol matrix and it was free of interferences. A classical internal standard

plot of the quantity (Tinuvin 770 response area/tricyclohexyl citrate response area) vs. the quantity (weight of Tinuvin 770/weight of tricyclohexyl citrate) provided an excellent linear relationship (slope = 0.74, y intercept = -0.01, and correlation coefficient = 0.9999) over the Tinuvin 770 weight range of 2-200 ng in the presence of 50 ng of internal standard. The day-to-day reproducibility of this relationship was excellent. This technique appeared to overcome the problems associated with instrumental drift and sample distribution on the probe which plagued the external standardization procedure. However, problems were encountered when actual coating extracts were analyzed. The response of tricyclohexyl citrate proved to be highly dependent on the sample matrix and this dependency did not parallel that of Tinuvin 770. Thus, the relative response between Tinuvin 770 and tricyclohexyl citrate was significantly different (2- to 10-fold change) for the neat glycerol matrix and the coating extract-glycerol matrix. These results suggested that the internal standard procedure wauld work if an internal standard could be chosen that would more closely simulate the chemical properties of Tinuvin 770 in the glycerol matrix. The deuterated Tinuvin 770 analogue was therefore synthesized and its use as an internal standard has thus far provided excellent quantitative results. The relationship between the quantity (Tinuvin 770 response area/deuterated Tinuvin 770 response area) vs. the quantity (Tinvin 770 weight/deuterated Tinuvin 770 weight) has consistently been linear and reproducible over the 5-1000 ng Tinuvin 770 weight range. This relationship has proven to be free of matrix effects for the two paint systems, both weathered and unweathered, which we have investigated. Table I illustrates the results of a study in which Tinuvin 770 was measured in a polymer-free matrix, an uncured coating extract and a cured coating extract using the deuterated internal standard procedure. The indicated number of samples containing a known amount of Tinuvin 770 were analyzed over a period of several days. The precision of these measurements fell within a 5% relative standard deviation. The percent recovery information in the table was derived by comparing the known concentration of Tinuvin 770 in each sample to the amount measured. The lower recovery obtained for the cured coating extract samples is accounted for by the conversion of Tinuvin 770 to other compounds in the curing process. The presence of cured polymer resulted in no observable interference to the FAB measurement process or to the methylene chloride extraction efficiency. The analysis of neat solutions containing known amounts of Tinuvin 770 and sample extracts containing known additions of Tinuvin 770 indicated that the accuracy of the technique is within

&lo%.

ACKNOWLEDGMENT The authors thank P. J. Schirmann of Ardsley Laboratory, Ciba-Geigy, Ltd., for his helpful discussions concerning the synthesis of deuterated Tinuvin 770.

Anal. Chem. 1984, 56, 2147-2153

Registry No. Tinuvin-770, 52829-07-9. LITERATURE CITED (1) Barber, M.; Bordeli, R. S.;Sedgwlck, R. D.; Tyler, A. N. J. Chem. Soc., Chem. Commun. W81, 7 , 325-327. (2) Barber, M.; Bordeli, R. S.;Elliott, G. J.; Sedgwlck, R. D.; Tyler, A. N. Anal. Chem. 1082, 5 4 , 645A-657A. (3) Bennlnghoven, A. Int. J. Mass Spectrom. Ion Phys. 1983, 4 6 , 459-462. (4) Magee, C. W. Int. J. Mass Spectrom. Ion phys. 1983, 4 9 , 21 1-221. (5) Gaskell, S.J.; Brownsey, B. G.; Brooks, P. W.; Green, B. N. Int. J. Mass Spectrom. Ion Phys. 1083, 4 6 , 435-438. (6) Milllngton, D. S.,presented at 31st Annual Conference on Mass Spectrometry and Allied Topics, Boston, MA, 518-13183. (7) Murphy, R. C.; Clay, K. L.; Stene, D. O., presented at 31st Annual Conference on Mass Spectrometry and Allied Topics, Boston, MA, 5/ 8-13/83.

2147

(8) DeStefano, A. J.; Keough, T., presented at 31st Annual Conference on Mass Spectrometry and Allied Topics, Boston, MA, 518-13183. (9) Ho,9. C.; Fensehu, C.; Hansen, G.; Larson, J.; Daniel, A. Clln. Chem. (Wlnston-Salem, N.C.) 1083, 2 9 , 1349-1353. (IO) Teeter, R. M., presented at 31st Annual Conference on Mass Spectrometry and Allied Topics, Boston, MA, 518-13183. (1 1) Shlyapintokh, V. Y.; Ivanov, V. B. I n "Developments in Polymer Stabilisation-5"; Scott, G., Ed.; Applied Science Publishers: London, 1982; Chapter 3. (12) Bauer, D. R.; Budde, G. F. Ind. Eng. Chem. Prod. Res. Dev. 1981. 20, 674-679. (13) Chattha, M. S.;Cassatta, J. C. J. Coat. Techno/. 1083, 55, 39-46. (14) McLafferty, F. W. "Interpretation of Mass Spectra", 2nd ed.; BenjaminlCummIngs: Reading, MA, 1973; p 42.

RECEIVEDfor review December 8,1983. Accepted May 7,1984.

Analysis of Chronocoulometric Data and Determination of Surface Concentrations James F. Rusling* and Margaret Y. Brooks University of Connecticut, Storrs, Connecticut 06268 Department of Chemistry (U-60),

A general method is described for analysis of data from double-potential-step chronocoulometry by simultaneous nonlinear regression onto exact equations for forward and reverse branches of O-f curves. Parameters determined include diffusion coefficients, surface concentrations, and double-layer charge. Applications to analysis of diff usloncontrolled reductions of Ti(1) and U(V1) and to determination of the surface concentration (I?,) of Cd(I1) adsorbed on mercury from thiocyanate soiutlons are described. With step wldths (7) 10.1 s and data equally spaced on the f axis, improvements in accuracy and precision in determining were realized over those obtained from a conventional iinearplot analysis. Data were most significant in the time range of 25% of T following the initial potential step, and, for determination of for 25% of 7 on either side of f = T . Clustering of data for nonlinear regresslons in these regions when T was 20.1 s. A provided accurate computation of diagnostic test based on deviation-pattern recognltlon was successful in detecting reactant adsorption.

ro

r,,

ro

Since its inception in the 1960s (I-3), double-potential-step chronocoulometryhas made possible significant contributions to the understanding of adsorption of electroactive molecules and ions ( 3 , 4 ) . A particular advantage of the method is that the charge for electrolysis of adsorbed species can be separated from the electrode double-layer charge, enabling the use of chronocoulometry for determining surface concentrations. The method has also found considerable use in elucidating mechanisms of electrode reactions and in determining rate constants of chemical reactions coupled to electron-transfer steps ( 3 , 5 ) . In a typical double-potential-step chronocoulometric experiment, the potential a t the working electrode is held at an initial value (EJwhere no electrolysis occurs, and, at t = 0, is rapidly stepped to a value (Ef)at which the desired electrode reaction takes place at a diffusion-limited rate. At t = T , with T ranging from milliseconds to several seconds, the potential is returned to Ei. The quantity of electricity, or 0003-2700/84/0356-2 147$01.50/0

charge (Q), which has passed through the electrochemical cell, is measured vs. t during the period 0 to 27 (Figure 1). The usual mode of analyzing Q vs. t data for electrochemical reactions uncomplicated by chemical steps assumes that plots of Q(t < T ) vs. t1izand Q, = Q(7) - Q(t > T ) vs. 6, where 6 = d 2 (t - 7)112 - t1Iz,are linear and that charging of the electrode double layer and electrolysis of any adsorbed reactant are instantaneous. The linear-plot model is based on semiinfinite linear diffusion to a planar electrode and, when adsorption is involved, approximates a nonlinear function in the theoretical expression for Q with a linear one. As pointed out by its originators (1,2), the assumptions of the linear-plot method are approximately followed within optimum ranges of experimental conditions (e.g., small T for spherical electrodes) and may lead to considerable errors under other conditions. We felt that regression analysis of the data onto the exact theoretical expressions for the Q-t curves would provide a more general approach to analyzing chronocoulometric data and, moreover, would avoid the approximations and limitations of the linear-plot method. For electrode processes uncomplicated by chemical reactions, Q is a linear function of all the parameters in the equations for the forward and reverse potential steps. In principle, then, the data could be analyzed by multiple linear regression. However, in the interest of developing a general method with an approach extending to systems with nonlinear relationships between Q and parameters, and because common factors appear in the forward and reverse equations, we have chosen to evaluate simultaneous nonlinear regression of the data onto equations for forward and reverse branches of the Q-t curves. In this paper, we describe applications of the new method to electrode reactions involving only diffusion and to those featuring adsorption of the electrochemical reactant.

+

THEORY Consider the electrode reaction

0+ne = R

Eo' (1) where only 0 is present initially in the solution at a concen0 1984 Amerlcan Chemical Society