388
Anal. Chem. 1083, 55, 388-390
to the Henderson equation, the potential difference of the solution I against solution I1 is given by the equation
i
i
where u,is the mobility and Cithe molar concentration of the ith ions. In this case, the test solution corresponds to solution I and the outer electrolyte solution of the reference electrode corresponds to solution 11. Then, the potential (Emr) corrected for Ej is given by = Eobsd - E j
opposed by C1- ISE in a constant CuC1, concentration M, for example), the minimum would appear upon changing ionic strengths with NaN03. Following the same procedure mentioned above, we should be able to obtain the values for mean activity coefficients without uncertainty caused by liquid-junction potentials. The same should be true for the Ca(II)-Cl- ISE combination. A t present, activity coefficients of ions in solutions made up of a mixture of electrolytes of known ionic strengths such as serum and seawater rather than of a simple single electrolyte solution are the ones which are needed to be determined accurately. The method proposed in this paper seems to fulfill this requirement.
(8)
The values of E , are listed in Table 11. The minimum point in Figure 1 shifted slightly to higher ionic strength for the Cu(I1) ISE but remained almost the same for the Ca(I1) ISE. The least-squares fitting calculation was applied and four parameters, S A ( = X ( l ) ) , B (=X(2)), S C (=X(3)), and Ed (=X(4)), in eq 6 were determined. The calculation was done by using the program SALS (statistical analysis with leastsquares fitting) ( 1 7 ) . Thus, A , B , and C in eq 2 were determined empirically, which are listed in Table I11 along with the theoretical values. The theoretical values for A and B were calculated as follows: A = 0.511 X 4, 0.511 is the constant of the Debye-Huckel equation for water at 25 O C and 4 the square of the charge of divalent ions; and B = (0.329 X lo8) x (6 x IO4), 0.329 x IOs is the constant of the DebyeHuckel equation and 6 X lo4 the ion-size parameter of Cu2+and Ca2+. The figures suggest that the leasbsquares fitting was successful to a considerable extent. From this, it may be concluded that the observed potential of Cu(I1) and Ca(I1) ISEs can be represented as the form of eq 6. This result indicates that the activity coefficient can be estimated experimentally by using the empirically determined parameters A , B, and C in eq 2. The extended DebyeHuckel equation of the form of eq 2 was originally derived for a single electrolyte, and therefore, it should be important to point out that the equation could also be applied empirically for the mixture of two electrolytes as the present case. The theoretical values for A and B in Table I11 are those for the single electrolyte which is contradictory to the present experimental condition where CuSO., and CaCb exist as the form of a mixture with concentrated NaN03 and KC1 solutions, respectively. This must be a major reason why the theoretical and experimental values for A and B differ from each other to a greater extent. The conclusion drawn above further indicates that the activity coefficient of relevant ions at any given ionic strength can be estimated “in situ” with the similar solution conditions to the following ISE experiment using the obtained parameters of A , B , and C. Also, it should be mentioned that if we measure the potential of a Cu(I1) ISE
ACKNOWLEDGMENT The authors acknowledge S. Fujiwara and R. Tamamushi for their generous support and valuable discussion.
LITERATURE CITED (1) Bates, R. G.; Alfenaar, M. “Ion-Selectlve Electrodes”; Durst, R. A., Ed.; US. Government Printing Office: Washington, DC, 1969; NBS Spec. Publ. No. 314, Chapter 6. (2) Bates, R. G.; Guggenheim, E. A. Pure Appl. Chem. 1980, 1 , 163. (3) Moody, G. J.; Thomas, J. D. R. “Selectlve Ion Sensitive Electrodes”; Merrow Publishlng Co.: England, 1970. (4) Roblnson, R. A.; Stokes, R. H. “Electrolyte Solutions”, 2nd ed.;Butterworths: London, 1969; Chapter 9. (5) Bates. R. G.; Staples, B.R.; Robinson, R. A. Anal. Chem. 1970, 42, 867-87 1. (6) Bates, R. G. Pure Appl. Chem. 1973, 36,407-420. (7) Bates, R. G. Denki Kagaku t978, 4 6 , 480-484. (8) Shatkay, A.; Lerman, A. Anal. Chem. 1969, 41, 514-517. (9) Shatkay, A. Anal. Chem. 1987, 71, 1056-1065. (IO) Shatkay, A. Siophys. J. 1968, 8, 912-919. (11) Shatkay, A. Electrochim. A C f 8 1970, 15, 1759-1767. (12) Bagg, J.; Rechnitz, G. A. Anal. Chem. 1973, 36, 407-420. (13) Neff, 0. W. Anal. Chem. 1970, 42, 1579-1582. (14) Harned, H. S.; Owen, B. B. “The Physical Chemlstry of Electrolytlc SOlUtiOnS”; Van Nostrand-Reinhold: Princeton, NJ, 1958. (15) Lakshminarayanaiah, N. ”Membrane Electrodes”; Academic Press: New York, 1976. (16) Sawatarl, K.; Imanlshi, Y.; Umerawa, Y.; Fujiwara, S.Sunseki Kagaku 1978, 27, 160-183. (17) Nakagawa, T.; Oyanagi, Y. SALS, Program Libraries, The Computer Centre, The University of Tokyo, 1979.
’
Present address: Department of Chemistry, College of General Education, The Unlverslty of Tokyo, Komaba, Meguroku, Tokyo 153, Japan.
Isamu Uemasul Yoshio Umezawa* Department of Chemistry Faculty of Science The University of Tokyo Hongo, Tokyo 113, Japan RECEIVED for review August 16, 1982. Accepted October 18, 1982.
Cross Polarization and Magic Angle Spinning Carbon- 13 Nuclear Magnetic Resonance Spectrometry for Quantitative Analysis of Coal and Coal-Derived Solids Sir: Coal and other macromolecular fossil fuels are not sufficiently soluble in solvents and hence the use of conventional NMR spectrometry has been restricted. Recent advances in solid-state NMR spectrometry (1) have made it
possible to reveal the chemical structures of solid organic substances including fossil fuels (2-10). However, few studies on the quantitative aspects of solid-state spectrum have been reported. Maciel et al. (8) and Yoshida et al. (11) evaluated
0003-2700/83/0355-0388$01.50/00 1983 Amerlcan Chemlcal Soclety
ANALYTICAL CHEMISTRY, VOL. 55, NO. 2, FEBRUARY 1983
t
\
389
tractor. The extract (C, 87.5%; H, 7.1%; 0, 2.6%) solidified at room temperature but was completely soluble in chloroform. Solid-state spectra were obtained on a JNM-FX6OQ FT NMR spectrometer (JEOL Ltd.) with combined high-power proton decoupling and CP/MAS techniques over a range of 5 kHz with 4096 data points at 15 MHz. Chemical shifts were first calibrated against a low-field peak at 38.7 ppm of external adamantane and then recalculated to give shifts from tetramethylsilane. About 500 mg of sample was pressed into a rotor, which was adjusted to the magic angle on a test sample of hexamethylbenzene. The spinning rate of the rotor was approximately 2.1 kHz. Pulses were accumulated 500 to 800 times.
4 .
RESULTS AND DISCUSSION
1
I
I
180
160
140
-
120
I
I
60
40
1
100
80
I
20
0
Chemlcal shifts from TMS (ppm)
Flgure 1. Comparison of CP/MAS solid-state spectrum with solution spectrum of hexane extract. (a)
0.5
k-l 8
f
12 ,
-
I
16I
Pulse! reDetitlon time, S
'[
0.50--4
(b) d a t a l r o m solution spectrum
6'
'
'
8
'
'
IO
C o n t a c t time, ms
Flgure 2. Change in fa value of hexane extract with pulse repetition time and contact tlme.
the reliability of carbon aromaticity (fa) values obtained in cross-polarization (CP) and magic angle spinning (MAS) experiments. The purpose of this work is to elucidate the effects of contact time and pulse repetition time on the quantitative aspects of CP/MAS spectra of coals and coal-derived solid products and to determine the optimal conditions for quantitative analysis. For a coal-derived solid product that was soluble in hexane, fa was obtained from both the CP/MAS solid-state spectrum andl solution spectrum. EXPERIMENTAL SECTION A sample was prepared by extracting hydrogenolysisproducts of Yubari coal (Hokkaido, ,Japan) with hexane in a Soxhlet ex-
Figure 1 compares the CP/MAS solid-state spectrum with the solution spectrum of the extract. It is clear that the combined use of C P and MAS techniques provides considerable improvement, giving resolution nearly as high as that of the solution spectrum. As shown in Figure la, the aromatic and aliphatic regions are clearly separated from each other. The signals in the aromatic region are insufficiently resolved and the distinction between protonated and substituted aromatic carbon atoms is difficult. On the contrary, the aliphatic carbon atoms can be divided into three groups, each being assignable to the terminal methyl groups in alkyl side chains (ca. 6-17 ppm), the methyl groups a to aromatic rings (ca. 17-27 ppm), and methine and methylene groups (ca. 27-51 ppm) on the basis of assignments in solution spectra (12). Thus, quantitative CP/MAS spectrometry appears to have the potential of becoming one of the most useful methods for the structural investigations of coal and other fossil fuels. Aromaticity is the most basic parameter for the structural analysis of natural organic substances, and hence we studied the effects of measurement conditions on fa values. Figure 2 shows the change in fa value with pulse repetition time and contact time. In CP experiments, the 13C spin system is polarized by the lH spin system under the Hartmann-Hahn condition and therefore the long I3C spin-lattice relaxation time (TJis replaced by the shorter 'H spin-lattice relaxation time (TIH). Nevertheless, Figure 2a indicates that pulse repetition times shorter than 6 s reduce the apparent fa value as the result of the progressive saturation of signals from aromatic carbon atoms. The polarization of I3C spin system is a function of proton TIPH and cross-polarization time ( T e H ) (13). The polarization at a given contact time should be different for aromatic and aliphatic carbon atoms because of the difference in their relaxation times. As shown in Figure 2b, fa values change with contact time. Compared with CP/MAS experiments with coals, the dependence of the fa value of the extract on contact time was small. In order to determine the optimal contact time for quantitative analysis, we compared the fa values obtained from CP/MAS solid-state spectra of the extract with that obtained from the conventional solution spectrum. The spectrum of the solution sample was measured by using a sufficient pulse repetition time and a gated-decoupling technique without nuclear Overhauser enhancement (NOE), to give a reliable fa value of 0.71. This value is close to the maximum fa value obtained in the CP/MAS experiments. Therefore, the contact time of 5 to 7 ms which gives the maximum fa value for the extract should be employed for the quantitative analysis. At the same time, this agreement suggests that the hexane extract used in the present study has a relatively low population of the fused ring aromatic species that give rise to the greatest difficulties in the quantitative determination of fa, even though it has a high fa value. Furthermore, we made one other check of the fa value from the NMR experiments. Namely, NMR with MAS and proton gated-decoupling without NOE, without
390
Anal. Chsm. 1983, 55,390-391
using the CP technique, provided quantitative spectral data a long measurement time was needed* The fa value was 0.70 when a pulse repetition time of 30 s was used. In conclusion, the maximum fa value obtained at various contact times in the CP/MAS experiments should be closest to the real value. Thus, the optimum values for contact time and pulse repetition time must be determined for each sample to obtain the most reliable measurement of fa.
LITERATURE CITED Pines, A.; Gibby, M. G.; Waugh, J. S. J. Cbem. Pbys. 1973, 5 9 , 569. VanderHart, D. L.; Retcofsky, H. L. Fuel 1976, 55, 202. Bartuska, V. J.; Maciei, G. E.; Schaefer, J.; Stejskai, E. 0. Fuel 1977, 56. 354. Retcofsky, H. L.; VanderHart, D. L. Fuel 1978, 5 7 , 421. Resing, H. A.; Garroway, A. N.; Haziett, R. N. Fuel 1978, 5 7 , 450. Maciei, G. E.; Bartuska, V. J.; Miknis, F. P. Fuel 1978, 57, 505. Ziim, K. W.; Pugmire, R. J.; Grant, D. M.; Wood, R. E.: Wiser, W. H. Fuel 1979, 58, 11. Maciei, G. E.; Bartuska, V. J.; Miknis, F. P. Fuel 1979, 58, 391. Ziim, K. W.; Pugmire, R. J.; Larter, S. R.; Aiian, J.; Grant, D. M. Fuel 1981, 60, 717.
(IO) Yoshida, T.; Nakata, Y.; Yoshida, R.; Ueda, S.;Kanda, N.; Maekawa,
Y. Fuel 1982, 61, 824. (11) Yoshida, T.; Narita, H.; Yokoyama, S.;Yoshida, R.; Maekawa, Y., submitted for publication in Fuel. (12) Johnson, L. R. F.; Jankowski, W. C. "Carbon-I3 NMR Spectra"; Wiiey: New York, 1972. (13) Schaefer, J.; Stejskai, E. 0.; Buchdahi, R. Macromolecules 1977, 70, 384.
Tadashi Yoshida* Yosuke Maekawa Government Industrial Development Laboratory, Hokkaido 2-17 Tsukisamu-Higashi, Toyohira sapporo 061-01, Japan
Teruaki Fujito JEOL Ltd. 1418 Nakagami, Akishima Tokyo 196, Japan RECEIVED for review June 7,1982. Accepted October 21,1982.
Reagent Stability in the Modified Pararosaniline Method for the Determination of Formaldehyde Sir: The selectivity and sensitivity of the pararosaniline method make it the preferred procedure for determining the formaldehyde content of dilute aqueous solutions or of the atmosphere ( I ) . However, when performed in the usual way the method suffers from the significant drawback that a poisonous tetrachloromercurate (TCM) solution has to be used to stabilize the sulfite reagent ( 2 , 3 ) .References have recently appeared in the literature concerning a modified pararosaniline method without mercury ions which is used to determine both sulfur dioxide and formaldehyde in air. When determining sulfur dioxide in the atmosphere, formaldehyde is used to stabilize the sample solution ( 4 , 5). When determining formaldehyde, however, no stabilizer is added to the sulfite solution ( I ) . This correspondence describes a study of the stability of sulfite reagents in pararosaniline methods. EXPERIMENTAL SECTION Reagents. The concentrations of the various components of the reagents were chosen in such a way that the final concentrations in the solutions used for measurement would always agree with those of Lahmann and Jander (2). Except for the concentration of the hydrochloric acid, these concentrations do not differ from those reported by Miksch et a1 ( I ) . The tetrachloromercurate-sulfite (TCMS) solution was made up by dissolving 100 mg of sodium sulfite in 100 mL of TCM solution. The TCM solution consisted of 5.8 g of sodium chloride and 13.6 g of mercuric chloride dissolved in 1L of water. The pararosaniline (P) solution consisted of 160 mg of pararosaniline (Merck) in 100 mL of 2.88 M hydrochloric acid solution; this reagent will keep for several months (3). The sulfite (S) solution consisted of 200 mg of sodium sulfite heptahydrate (NazS0,.7H20) dissolved in 100 mL of water. The combined TCMS-pararosaniline (TCMSP) reagent was prepared by mixing equal volumes of TCMS solution and P solution. The sulfite-pararosaniline (SP)solution, Le., the combined reagent without the mercury ions, consisted of 200 mg of sodium sulfite heptahydrate and 160 mg of pararosaniline in 200 mL of 1.44 M hydrochloric acid solution. The standard formaldehyde solutions were always freshly prepared from a 37 wt % formaldehyde solution (Baker) in a 0.005 M hydrochloric acid solution. The solution strength was determined by a potentio-
metric sulfite method (1, 6) or by an iodometric method (6). Procedure. The procedure of Lahmann and Jander (2)was followed. For the analysis 2 mL of TCMS or S with 2 mL of P or 4 mL of the combined reagent TCMSP, or 4 mL of SP, was added to 20 mL of formaldehyde solution. Absorbance was measured at a wavelength of 578 nm and an optical path of 1 cm (Varian 634). The sensitivity, expressed as absorbance units (AU) per microgram of CH20per milliliter of solution under investigation, was calculated from at least three measurement points for different formaldehyde concentrations (at 0.5,1.0,and 1.5 AU) and from one blank measurement (0.09 AU).
RESULTS AND DISCUSSION By application of the method described by Lahmann and Jander ( 2 ) with freshly prepared TCMS and P reagents (TCMS-P method) over a period of 20 months (34 measurements) a mean sensitivity of 0.472 AU mL bg-l with a relative standard deviation, V, of 2% and a mean correlation coefficient, r, of 0.999 94 for the calibration curve was found. A precipitate appears in the TCMS reagent after 24 h (3). The sensitivity of the formaldehyde determination with the filtered reagent deteriorates slowly (2% over 48 h). This agrees with the slow decrease in the sulfite concentration which Doklddalovii and Stankova (7) confirmed in this reagent, in spite of the presence of the [HgCl2SO3I2-complex. If the formaldehyde determination is carried out with an S and a P solution as reagents, erroneous results may be obtained. With a freshly prepared S solution, the same sensitivity was measured as with the TCMS-P method (mean value of three measurements: 0.468 AU mL kg-l, V = 3%, r = 0.999 90). The reagents were added to the analysis solution in immediate succession; no influence from the order of addition of the reagents was found. However, when the S solution was added first and the P solution only after 15 min, it appeared that the sensitivity had decreased by about 10%. Also, the relation between absorbance and formaldehyde concentration is not linear any more in that case. This phenomenon is due to the slow formation of hydroxymethanesulfonic acid in the weakly acid solution after addition of the S reagent (5).
0003-2700/83/0355-0390$01.50/00 1983 American Chemical Society
