Carbon dioxide permeable tubings for post-suppression in ion

Department of Analytical Chemistry, University of Umek, S-901 87 Umek, Sweden. Darryl D. Siemer. Exxon Nuclear Idaho Company, Idaho Falls, Idaho 83401...
0 downloads 0 Views 1MB Size
1085

Anal. Chem. 1984, 56, 1085-1089

Carbon Dioxide Permeable Tubing for Postsuppression in Ion Chromatography T h o m a s SundBn* and Anders Cedergren Department of Analytical Chemistry, University of UmeA, S-901 87 UmeA, Sweden D a r r y l D. Siemer

Exxon Nuclear Idaho Company, Idaho Falls, Idaho 83401

Gas permeable poly(tetrafiuoroethylene) (PTFE) tubings were used to lower the background conductivity from hydrogen carbonatekarbonate eluents in suppressed ion chromatography (IC). The carbonic acid concentration in the effluent from the suppressor column is lowered as a result of carbon dioxide permeating through the porous tubing. Up to 90% suppression compared to normal suppressed IC was achieved with a properly designed postsuppressor. The additional band broadening accompanied by the introduction of the extra volume is controlled by the use of different mass transport devices inside the permeable tubings. Advantages gained with this postsuppressor are enhanced sensitivity, e.g., almost 40% for sulfate, and the use of gradient elutions with great differences In eluent strength wnhout causlng much shM in the base line. Also, the reduced background level improves essentially the conditions for quantitative evaluation of ions coeiuting with the carbonate dip.

In a previous paper (I) we described a gradient system for ion chromatography with bicarbonate/carbonate eluents. This system was designed in such a way that essentially no base line shift was obtained for a change in eluent from 5 mM NaHC03 to 5 mM Na2C03. Unfortunately, this change in eluent strength is too small to be effective for more strongly retarded anions. In order to extend the range of the gradient the total background conductivity from the carbonic acid has to be lowered. One way to do this is to displace the carbonic acid-carbon dioxide equilibrium ( 1 ) to the right by removing the carbon dioxide in some way.

If all gaseous carbon dioxide can be taken away, the remaining effluent will consist of deionized water with very low conductivity. It is well known that several polymeric materials are permeable to gases. For example, silicone rubber has been used in cells where gases permeate from a donor to a recipient stream across a membrane. In recent years a microporous poly(tetrafluoroethy1ene) material has been available under the trade name GORE-TEX, in which gases can diffuse through gas-filled pores between the fibrils of the hydrophobic matrix. These membranes are available as tubings as well as sheets. The aim of the work presented in this paper was to make a systematic investigation of the possibilities of removing the carbon dioxide from the eluent stream by use of porous tubings. The analytical advantages gained by lowering the carbonic acid concentration, i.e., the background conductivity, will be discussed.

Table I. Permeable Tubing Useda dimensions max (mm) i d . / pore wall porosity, size, % wm tubing thickness GORE-TEX TA 001 GORE-TEX TB 001 GORE-TEX TA 002 GORE-TEX 1.7 X 0.30t Silicone rubber tubing

1.0/0,4 1.0/0.4 2.0/0.4 1.7/0.3

50 70

30-40

2.0 3.5 2.0 b

1.0/1.0

c

C

50

a Specifications according to manufacturer. known by the authors. Not specified.

Not

EXPERIMENTAL SECTION Instrumentation. The ion chromatograph described in a previous paper (1) was used with an additional postsuppressor device. The separator columns used were either one (4 X 250 mm) anion separator column (No. 030827 Dionex Corp., Sunnyvale, CA) or two (4 X 50 mm) anion precolumns (No. 030825 Dionex Corp., Sunnyvale, CA). The two precolumns were joined together by a special connector with no dead volume in order to get one (4 X 100 mm) separator column. The connector joins the column tube ends with only a thin Teflon washer between them. The ion exchange material forms hereby a uniform packing and works as a single column of double length. The suppressor columns used were either a laboratory made (5.7 X 150 mm) glass column or an Omni (3 X 100 mm) glass column (Omnifit Ltd., Cambridge, England) both packed with AG 50W-X16,200-400 mesh (Bio-Rad Labs., Richmond, CA). The postsuppressor device is shown in Figure 1 and is described below. All experiments were carried out in a thermostated room at 20 f 0.2 "C. Chemicals. All solutions were prepared in doubly deionized water using reagent grade chemicals. Postsuppressor Construction. Four types of microporous poly(tetrafluoroethy1ene) (GORE-TEX) tubing (W. L. Gore & Associates, Inc., Elkton, MD) were examined in this work; see Table I. The GORE-TEX tubing was placed inside a Plexiglas cylinder with inlet and outlet for compressed air which was used to sweep away the permeated carbon dioxide. The permeable tubing was connected to the ion chromatograph by flanged end couplers through threaded (1/4 in. X 28) holes in the Plexiglas tube stopper. This stopper was a laboratory constructed Plexiglas/Teflon valve, which was used to bypass the postsuppressor device when comparative studies were performed. In order to remove carbon dioxide from the compressed air, the inlet port was equipped with an Ascarite tube. A manometer and a needle valve were placed at the outlet port to control the pressure outside the permeable tubing. RESULTS AND DISCUSSION Postsuppressor Performance. Table I shows the different types of gas-permeable tubings that have been examined during this work. The desirable properties for this kind of postsuppressor are (i) high permeability for carbon dioxide,

0003-2700/84/0356-1085$01.50/00 1884 American Chemical Society

1086

* ANALYTICAL CHEMISTRY, VOL. 56. NO. *From

7.

JUNE 1984

Suppressor Column

.O-ring

AirrC02

out

.

cylinder, 30mm i d x 85 rnrn

9

*Compressed

Air

I"

Flgure 1. Schematic diagram of the postsuppressor.

Table 11. Background Suppression with Different Postsuppressors' supprestubing filling sion 1 mTB001

1 m TB 001 3 m TB 001

1 0 m TB 001 1 m TA 001 1 m TA 001 1 m TA 001

empty empty, coiled empty empty, coiled empty 0.7 mm n y l o n fishing line

notched 0.7 mm nylon fishine line twisted stainless steel wire 2 x 0.4 mm knotted 0.28 mm nylon fishing line GORE-TEX Teflon thread thread Y 10176 twisted stainless steel wire 2 x 0.4 mm ~~

1 m TA 001

1 m TA 001 2 m TA 001 2mTA001

Eluent flow rate

=

30 45

74 89

30 53

68

~0

70 77

89

90

2.5 mL m i l l '

(ii) small volume in order to minimize the extra hand broadening, (iii) good mass transport of the gaseous molecules from the bulk to the membrane, and (iv) capacity to withstand the actual pressure without leakage. However, it is impossible to fulfill all these demands with a single manifold since (i) high permeability also means low water entry pressure which will lead to leakage and (ii) the total amount of permeated gas is dependent on the volume of the device or rather the area of the membrane. Thus, a large membrane area means a long piece of tubing and a large pressure drop if a small diameter tubing is used in order to achieve good radial mass transport. A large pressure drop leads to leakage if it exceeds the water entry pressure for the particular tubing. Large area also implicates a large volume which causes undesirable hand broadening. This means that there are some compromises to he made. Fortunately, good mass transport is usually obtained in small volumes. Table I1 lists different configurations of the selected tubings with and without fillings used to disturb the laminar flow and enhance the radial mass transport to the membrane wall. The percentage suppression is calculated with the following equation: suppression = (1 - (Gps- Gw)/(Gn8- G,)) X 100% (2) where G, is the conductivity with additional postsuppressor and G,, is the conductivity with normal suppressor column and G, is the conductivity when deionized water, originally with resistance of a t least 5 Ma, was pumped through the ion chromatograph in normal suppressor mode (G, = 1.2 pS m-'). As seen in Table I1 both tubing types A and B have the same degree of suppression, when no filling is used, although their porosity (see Table I) and air permeability differ. This result indicates that is is the convective transport in the hulk

I

Flgure 2. Mass bansport devices: (Ai two twisted stainless steel wires. each 0.4 mm 0.d.; (B) knotted 0.28-mm nylon fishing line; (C) Teflon thread (GORE-TEX Y 10175) approximately 0.5 mm 0 . d .

solution rather than membrane diffusion that limits the overall mass transport. Similar results were obtained by Stevens et al. (2). The most simple way to increase convection is to coil the tubing in order to achieve a secondary flow according to Tijssen (3). These tubings are very flexible, so when coiled, the cross-section area becomes elliptic rather than circular which shortens the distance in the secondary flow direction and increases the positive effect of coiling. The flexibility of the tubing and problems to purchase beads of suitable size made attempts to make a single head string reactor (SBSR), described by Reijn et al. (4),of the 1 mm i.d. tuhing impossible. SBSR was made with the (1.7 X 0.3Ot)tubing, hut the pressure drop unfortunately exceeded the water entry pressure resulting in leakage. If the needle valve was adjusted in order to increase the pressure outside the permeable tubing and prevent leakage, problems arised due to air entry through the downstream parts of the postsuppressor. As alternatives to SBSR, different string-type fillings were examined in order to increase convection and decrease the volume. An ordinary 0.7 mm 0.d. nylon fishing line (ABU, Sweden) was inserted which increased the suppression and the effect was even larger when the fishing line was notched. This indicates that the laminar flow is not broken with the smooth fishing line but the decrease in volume and diffusion distance gave about 50% suppression. The most effective filling is a thin fishing line with half-hitches as close to each other as possible, see Figure 2B, which form a similar pattern as the beads in an SBSR. A major drawback with this knotted fishing line is the timeconsuming manufacturing (handmade). The GORETEX tubings in Table I which are not included in Table I1 were tested with different fillings with good suppression results hut their large volumes, in spite of fillings, gave unacceptable band broadening. The silicone rubber tubine (2 m) gave only 38% suppression and severe band broadening. Background conductivity was shown to he linearly dependent on flow rate with a slope of 0.62 pS min cm-' mL-' (standard error 0.016) within the working range for one of the most used devices, 2 m TAW1 equipped with a GORE-TEX thread (Y 10175). The preferences of this configuration are as follows: (i) the suppression is about 9070,(ii) it withstands the pressure at flow r a t e up to 3.5 mL min-', and (iii) the hand broadening is acceptably small. For flow rates up to 2.5 mL min-' a configuration with 2 m TA 001 and two twisted 0.4-mm stainless steel wires is preferred because of even lesser band broadening and slightly lower background conductivity. Band Broadening. The introduction of additional volume in a chromatographic system is inevitably accompanied by

ANALYTICAL CHEMISTRY, VOL. 56, NO. 7, JUNE 1984

1087

Table 111. Regression Data for Calibration Curves

I\

10s

std range, ppm FC1NO,. SO,’-

[ 3pscm1 iG

c 1-

regression slope 95% conf limit normal supp

postsupp

slope difference 99.9% conf limit

0.05-5 1010 i 30a 1140 i 35 127 i 0.1-10 314 i 15 417 i 14 103 f 1-100 27.6 i 0.17 39.8 i 0.80 1 2 i 1-100 37 t 1.4 51 ?. 1 . 2 14 f

81 35 1.4 3.3

a The fluoride peak for 0.05 ppm disappeared in the carbonate dip (see Figure 4): separator column, 4 X 250 mm; suppressor column, 5.7 x 150 mm; postsuppressor, 2 m TA 001 with 2 x 0.4 mm stainless steel wire; eluent, 3.0 mM NaHCOJ2.4 mM Na,CO,; flow, 2.5 mL min-’. Nine four ion standards and blank with at least two injections of each concentration were used.

s 0;

Flgure 3. Peaks for F-, GI-, and Sod2-: separator column, 4 X 250 mm; suppressor column, 3 X 100 mm; postsuppressor, (A) none, (B) 2 m TA 001 wRh stainless steel wire (see Figure 2A), (C) 2 m TA 001 (empty); eluent, 4.0 mM NaHC0,13.6 mM Na,CO,; flow, 2.5 mL min-’.

extra band broadening (see Figure 3). The contribution is severe for peaks with small k’values, fluoride and chloride, but is less pronounced at higher k’values, sulfate. With 2 m of 1mm i.d. tubing, Figure 3C (approximately 1.6 mL of extra volume), the peak width a t half height is increased to more than 3.5 times and the height is reduced to approximately one-third of normal values for fluoride. It should be pointed out that these changes are relative and thus dependent on the band broadening without postsuppressor which is influenced by the eluent strength and size of the suppressor column. The fillings utilized to increase mass transport mentioned above will also prevent additional band broadening as it reduces the volume of the postsuppressor device and interfere with the laminar flow in the tubing. When a GORE-TEX thread of approximately 0.5 mm 0.d. (see Figure 2C) is inserted in the 1mm i.d. porous tubing, the width at half height of the fluoride peak is reduced to three times that obtained without postsuppressor. A still better situation occurs when two 0.4 mm 0.d. stainless steel wires twisted round each other, see Figure 2A, was used as mass transport device. The peak width is now only 1.5 times larger than normal, see Figure 3B, but the peak height is increased due to the reduction of the eluent matrix effect and the carbonate dip. For chloride (k’ = 2.2 without postsuppressor) the peak widths increase with 10% with stainless steel wire, 50% with GORE-TEX thread, and 65% with no filling. For sulfate (k’ = 12 without postsuppressor) there are no significant differences in peak width for the different postsuppressor configurations. The peak heights for chloride and sulfate will be discussed in the next section. Sensitivity Enhancement. As thoroughly discussed by Pohl and Johnson (5)the reduction of the eluent matrix effect is the reason for greater sensitivity in suppressed IC compared

U

Figure 4. Peaks for F- (0.05 ppm) and GI- (0.1 ppm) obtained with (4 X 250 mm) separator and (5.7 X 150 mm) suppressor columns: (A) no postsuppressor, (B)2 m TA 001 with stainless steel wire. Eluent and flow are given in Flgure 3.

with nonsuppressed IC. With the postsuppressor device described herein the background conductivity of the eluent is lowered even further compared to suppressed IC. Hence, the reduced matrix effect should give rise to greater sensitivity (defined as the slope of the calibration graph) if no other factors, such as extra band broadening, were affecting the chromatographic band. If an empty GORE-TEX tubing is used as postsuppressor, the band broadening for early eluting ions is too severe and influences the peak height in the negative direction to a greater extent than does the reduced matrix effect, in a positive direction. But for ions with long retention time, e.g., sulfate, the extra band broadening is overtaken by the matrix effect and thus higher peaks are obtained with the postsuppressor; see Figure 3. With the Teflon thread (GORE-TEX Y 10175) as mass transport device, the extra band broadening is too large so the sensitivity for chloride is lower regarding peak height, but with respect to peak area the sensitivity is greater with postsuppressor than in normal suppressor mode. The sensitivity for sulfate is greater with respect to both peak height and peak area. With the best performing configuration, i.e., the stainless steel wire, the slopes regarding peak height for sulfate, for nitrate, and as for chloride and fluoride are significantly (99.9% confidence level) greater in postsuppressor mode compared to normal suppressor, as seen in Table 111. When the proper mass transport device is utilized, it is possible to enhance the sensitivity not only for peaks with long retention times but also for ions, such as fluoride, eluting with k’values as small as 0.2. This is achieved, in spite of approximately 50% band broadening, by the combined action of the reduced eluent matrix effect and the decreased carbonate dip.

1088

ANALYTICAL CHEMISTRY, VOL. 56, NO. 7, JUNE 19O1 3.5

a 50

r B

6

3.0

8

40

t

1

30

c.--l

20

60s 10

I

2

Flgure 5. Water (1) and carbonate (2) dips obtained with 5.7 X 150 mm suppressor column: (A) no postsuppressor, (B) 2 m TA 001 with stainless steel wire. Eluent and flow are glven in Figure 3.

" a

\

0

0

20 Time (min)

Pi I

0

I

I

20

I

I

40

Time (min)

0

3

6

Time ( h )

Flgure 6. Peak height for IO3- (5 ppm) during suppressor column no postsuppressor, (0) 2 m TA 001 with Teflon thread lifetime: (0) (see Figure 2C) as postsuppressor. Columns, eluent, and flow are given in Figure 4.

Water and Carbonate Dip Problems. The dips occurring in suppressed IC known as the water and carbonate dips (6) cause difficulties in quantitative determinations of ions coeluting with the carbonate dip. For example, peaks for fluoride or iodate can, in the worst case, totally disappear, which is exemplified with fluoride in Figure 4A. At this low level (50 ppb F-) there are also problems with the psotsuppressor if quantitative determination is to be made, but the improvement gained with this device is remarkable. The reason for this is the diminished carbonate dip (see Figure 5) due to the lowered background conductivity level. The explanation for the problem to make quantitative evaluations in the normal mode of operation is that the retention time for the carbonate dip is dependent on the suppressor column exhaustion (7) which means that the peak height for a coeluting ion varies with the time after suppressor regeneration. Figure 6 shows the response for iodate as a function of suppressor column depletion. A solution with equal concentrations of iodate and bromate was injected during the lifetime of a suppressor of moderate size, with and without postsuppressor. The retention time for the bromate peak was longer than that for the carbonate dip and was therefore not influenced by the degree of depletion of the suppressor column. In normal suppressor mode, the iodate peak height time dependence shows a characteristic shape with a minimum at about 4 h of use, when the retention times for the iodate peak and the carbonate dip coincide completely. Almost the same shape, but to a far lesser extent, is seen when the postsuppressor is used. If a mean value of iodate peak height is calculated, the relative standard deviation is less than 5% with a concentration of iodate of 5 ppm. This situation deteriorates when iodate concentration

Figure 7. Chromatograms of (1) F-, (2) HCOO-, (3) CI-, (4) Br-, (5) SO3'-, ( 6 ) SO,'-, (7) I-, (8) SZO,'-: separator column, 4 X 100 mm, suppressor column, 5.7 X 150 mm; flow, 2.0 mL min-'; (a) gradient elutions, eluent A (start) 0.25 mM NaHC0,/0.24 mM Na,CO,, eluent B (finish) 18 mM NaHC0,/3 mM Na'CO,; (upper curve) no postsuppressor, (lower curve) 2 m TA 001 with Teflon thread as postsuppressor; (b) isocratic elution without postsuppressor, eluent 3.0 mM NaHC0,/2.4 mM Na2C03.

decreases, but quantitative determinations are possible if standard solutions of approximately equal concentration as the sample are injected within an appropriate time before or after the sample injection. Gradient Elution. As mentioned in the introduction, the main reagon for this work was to develop a system for gradient elutions with great changes in eluent strength, without causing a large base line shift. Figure 7 shows two chromatograms with gradient elution where the carbonic acid concentration leaving the the suppressor column at the end of the chromatographic run is 40 times higher than the start concentration. The gradient consists of two linear parts with a change to a steeper slope at the time for sulfite peak apex. The benefits of the postsuppressor are easily noticed, namely, (i) the minor change in background conductivity level and (ii) the greater peak heights compared to normal suppressor mode. One drawback, the slightly deteriorated separation between fluoride and formate is also observed with this postsuppressor configuration. As a comparison, Figure 7b shows the same ions in an attempt to separate them with an isoratic run using the "standard eluent" (3 mM NaHC03/2.4 mM Na2C0,). The resolution for early eluting ions is far from acceptable, only six peaks for eight ions, with iodate and thiosulfate badly tailing. Registry No. COz, 124-38-9; HCOO-, 71-47-6; PTFE, 900284-0.

LITERATURE CITED (1) Sundln, T.; Lindgren, M.;Cedergren, A.; Siemer, D. D. Anal. Chern. 1983, 55, 2-4. (2) Stevens, T. S.;Jewett, G. L.; Bredeweg, R. A. Anal. Chern. 1982, 5 4 , 1206- 1208.

1089

Anal. Chem. 1984, 56, 1089-1096

(7) Hanaoka, Y.; Murayama, T.; Muramoto, S.; Matsuura, T.; Nanba, A. J . Cbrornatogr. 1982, 239, 537-548 (Figure 14).

(3) Tijssen, R. Sep. Sci. Techno/. 1978, 73, 681-722. (4) Reiin. J. M.: van der Linden, W. E.; Poppe, H. Anal. Chirn. Acta 1981, 726, 1-13. (5) Pohl, C. A.; Johnson, E. L. J . Chrornatogr. Sci. 1980, 78, 442-452. (6) Stevens, T. S.; Davis, J. C.; Small, H. Anal. Chem. 1981, 53, 1488-1492. \

I

for review November

77

lgg3* Accepted February

2, 1984.

o-Phthalaldehyde Derivatives of Amines for High-speed Liquid Chromatography/Electrochemistry Laura A. Allison, Ginny S. Mayer, and Ronald E. Shoup* Bioanalytical Systems, Inc., Technical Center, West Lafayette, Indiana 47906

The conventlonal problems of derlvatlve stablllty and fluorescence quantum ylelds In the case of amlnes reacted with o-phthalaldehyde and varlous thlols were clrcumvented by a liquld chromatography/electrochemlstry (LCEC) approach. The degradative hydrolysls of the Ssubstltuted IsoIndole was Investigated by varylng the sterlc bulk of the thlol used In the o-phthalaldehyde reagent. No appreciable lmprovement In stability accrued until the bulky terf-butyl group was Incorporated, whereupon half-llves In excess of several hours were appreclated. I n contrast to fluorescence, the thlol had llttle Influence on the electrochemistry of the derlvatlve, whlch was characterired as a very rapld, Irreversible oxldatlon of the Isolndole. Gradient Separations of both the 0mercaptoethanol and fert -butyl derlvatlves on short 3-pm reversed-phase columns permitted LCEC detectlon llmlts of less than 500 fmol In the gradlent mode. Separations of 22 amlno aclds could be accompllshed In under 10 mln, but reequlllbratlon lengthened the total analysis time to about 30 mln. Detectlon llmlts were lowered to 30-150 fmol In the lsocratlc mode.

The quantitation of amines in varied samples is a problem that has prompted the development of numerous analytical schemes. The most successful approaches have been those which couple derivatization of amines with a separation based on liquid chromatography, as illustrated by the amino acid method first proposed by Spackman, Stein, and Moore ( I ) . Their amino acid analyzer was basically a specialized liquid chromatograph, utilizing ion-exchange chromatography and derivatization. With the advent of high-performance liquid chromatography, newer systems have emerged which capitalize on the higher level of chromatographic capability. Generally, detection of chromophoric derivatives combined with continuous two-component gradients on hydrophobic octadecylsilyl or hydrophilic cation exchange stationary phases have been employed. For amino acids, derivatization schemes include those based on ninhydrin (Z), o-phthalaldehyde (3-6), fluorescamine (Z), dansyl chloride (7),dabsyl chloride (8),and 7-fluoro-4-nitro-2,1,3-benzoxadiazole (NBD-F) (9). In particular, fluorescence methods utilizing o-phthalaldehyde as a derivatizing reagent in both the pre- and postcolumn modes have become popular, although they continue to face certain limitations. The reaction of o-phthalaldehyde (OPA) with amino acids in the presence of a thiol reducing agent to produce fluorescent products was first re0003-2700/84/0356-1089$0 1.50/0

ported by Roth (10); the fluorescent products were subsequently determined (11)to be l-(alkylthio)-2-alkylisoindoles (I). The S-substituted isoindoles formed by the reaction of

amines with OPA and either p-mercaptoethanol or ethanethiol are amenable to reversed-phase liquid chromatography in conjunction with precolumn derivatization. However, a significant problem with precolumn OPA approaches has been the instability of the derivatives (5, 6, 12), causing careful timing or even instrumental automation to ensure reproducibility. Postcolumn derivatization schemes utilizing OPA must scrupulously avoid impurities in the reagents and mobile phase buffers which can contribute to high background fluorescence (13). Postcolumn schemes also result in some loss of both resolution and sensitivity due to mixing the mobile phase with diluent (14, 15). Several researchers have investigated the possibility of enhancing OPA derivative stability by varying the structure of the thiol compound (15, 16). Although a number of alternate thiols were utilized in the OPA reaction, some of which improved stability, it was found that the fluorescence properties of the derivatives were also markedly dependent on the structure. The trade-offs between stability and fluorescence prevented a true optimization of the OPA/thiol derivatization reaction. Recently, it was reported by Joseph and Davies (17) that OPA/P-mercaptoethanol derivatives of amino acids undergo anodic oxidation a t moderate potential, permitting the use of liquid chromatography/electrochemistry (LCEC) for their determination. The present study was suggested by our speculation that the electrochemical properties of the derivative will be far less susceptible than fluorescence to changes in the derivative's structure. Hence, alternative thiols were utilized to search for a more stable isoindole derivative which would be suitable to a simple, precolumn LCEC method without stability limitations. A second important advantage to the implementation of an electrochemically active derivatization is the compatibility of LCEC with high-speed chromatographic separations, such as those obtained with short, small particle size columns. Since detection occurs on a surface, rather than homogeneously in solution, it is convenient to reduce the cell volume with little degradation in detector performance. Concurrent with our study of alternate thiols, 0 1984 Amerlcan Chemical Society