retention indices. The problems associated with coinjection during solid-state pyrolysis have been overcome by a simple copyrolysis procedure using one compound to produce an entire spectrum of reference materials. As more sophisticated GC/MS computer techniques are developed, with greater accuracy in computer identification of hydrocarbon peaks, the concentrations of standards can be reduced so the original chromatogram and spiked chromatogram will be similar in appearance.
ACKNOWLEDGMENT The authors thank J. H. Futrell for his helpful suggestions. LITERATURE CITED (1) See references in W. H. McFadden, "Techniques in Combined Gas Chro-
(2) (3) (4)
(5) (6) (7) (8)
matography/Mass Spectrometry: Application in Organic Analysis", Wiley Interscience, New York. and others, 1973. H. Kau and K. Biemann. Anal. Chem., 46, 426 (1974). C. C. Sweeley, N. D. Young, G. F. Holland, and S. C. Gates, J. Chromatogr., 99, 507 (1974). Gas Chromatographic Data Compilation, Am. Soc. Test. Mater.. Spec. Tech. Pub/., No. DS25A, (1967). E. Kocats, "Advances in Chromatography", J. C. Giddings and R. A. Keller, Ed., Marcel Dekker, New York, 1965, p 229. E. L. Levy and D. G. Paul, J. Gas Chromatogr., 5 , 136 (1967). M. B. Evans, J. Chromatogr., 12, 2 (1963). S. L. Madorsky, "Thermal Degradation of Organic Polymers", Wiiey-lnterscience, New York, and others, 1964, Chapter IV.
RECEIVEDfor review January 13, 1975. Accepted August 20, 1975. This research, conducted at the Flammability Research Center of the University of Utah, has been supported by the National Science Foundation, Grant Number GI33650, under its RANN (Research Applied to National Needs) program.
Thermal Field-Flow Fractionation: Extension to Lower Molecular Weight Separations by Increasing the Liquid Temperature Range Using a Pressurized System J. Calvin Giddings, LaRell K. Smith, and Marcus N. Myers Depariment of Chemistry, University of Utah, Salt Lake City, Utah 84 I12
The rationale for using a large temperature Increment, AT, in thermal fleld-flow fractionation Is given. Thls serves to extend the tractable molecular weight range of complex solute mixtures. It ls shown that the lowest molecular weight subject to analysis changes with the Inverse square of AT, and that the resolutlon between low molecular weight components Increases directly with the square of AT. The varlous possible means for augmenting AT are discussed. The method chosen here is a system pressurized to elght atmospheres, which ralses the normal bolllng point and thus the potential AT by almost 100 O C for common solvents. Experlments are descrlbed for AT = 158 O C , approximately twlce as high as used previously. The confirmation of the theoretlcal conclusions is verlfled uslng polystyrene fractions of mol wt 600 and 2100. Samples of crude oil, asphalts, and asphaltenes are also characterlzed. It Is concluded that this methodology might be advantageous In characterlzing complex materials because of combined solvent sensitlvlty and molecular-weight sequenclng.
The general methodology of field-flow fractionation (FFF) (1-5) and the specific characteristics of the thermal field-flow fractionation (TFFF) subclass (6-10) have been discussed in earlier papers. In a recent study of retention in T F F F ( 9 ) , polystyrenes of various molecular weights were differentially retained in a variety of solvents. However, the low molecular weight species, mol wt 2000 and 5000, were only slightly retained, exhibiting R values, a t best, between 0.9 and 0.95. In this range, a significant loss of resolution is expected, as will be shown presently. The use of a variety of solvents had only a moderate effect on this result; therefore solvent variation shows limited promise in improving retention significantly. Here we report another approach to the improvement of retention, based on increasing the temperature increment,
A T , between hot and cold walls. This approach is suggested by theory. According to the simple theory of TFFF, basic retention parameter X is inversely proportional to AT and to the square root of mol wt. We can write (9, 10) X = @/AT(molwt)1/2
(1)
where the constant, @,depends on the solvent and the type of polymer. In the case of polystyrene in ethyl benzene solvent, @ is determined empirically to be 1420 "C (g/mo1)1/2. The above equation shows that an increase in AT will reduce X proportionally. Changes in X are reflected in R through the basic retention equation
R = 6X[coth (1/2X) - 2x1
(2)
which is quite accurate, although it ignores a few secondary effects related to changes in viscosity and thermal conductivity across the column (9). Under conditions of strong retention, where R and X approach zero, Equation 2 reduces to ( 4 ) R = 6X (3) Thus R and X are related proportionally in this important limit in which retention is substantial. I t will be shown in the following section that resolution becomes progressively weaker as R approaches unity. Depending on the nature of the problem, one can specify a maximum R value, R,,,, for which the retention is regarded as useful. This corresponds to a certain maximum X value, A,, specified by Equation 2. Thus if R,,, = 0.95, A,, = 0.56; if R,,, = 0.90, A,, = 0.37, and so on. A plot of R vs. A, corresponding to Equation 2, is shown in Figure 1. This can be used to translate the limiting retention behavior, R,,,, into a limit on theoretical retention parameter A, namely A,,. With A,, established by retention criteria, Equation 1, slightly rearranged, shows the minimum molecular weight, mol wtmin,that can be successfully retained in the system
ANALYTICAL CHEMISTRY, VOL. 47,
NO. 14,
DECEMBER 1975
2389
\
\
01
0
02
04
06
08
I0
R Figure 2. Deterioration of resolution with increasing R values in TFFF and in chromatography
X = Tlaw(dT1dx)
A
Figure 1. Plot of retention ratio R vs. basic retention parameter X for TFFF
mol wtmin = ~2/(AT)2X,,,2
(4)
This equation shows that the lower limit of the applicable mol wt range can be changed in inverse proportion to the square of the temperature drop, AT. Thus, for example, the process of doubling AT should cause a fourfold reduction in workable molecular weight. The implementation of this concept is described in this paper. A pressurized TFFF system has been developed to increase the liquidus range, making possible an approximate doubling of AT values over those available before.
RATIONALE: RESOLUTION LOSS AT LOW RETENTION It has long been recognized that low retention, reflected in an R value approaching unity, is detrimental to resolution in chromatographic systems ( 1 1 ) . The same conclusion can be reached for TFFF. The magnitude of the deterioration of resolution with increasing R is established below. The resolution of peaks can be defined in the same manner as for chromatographic systems
R s = Az/4u
(5)
where Az is the increment in distance between peak centers and CT is the average standard deviation of the peak. Quantity Az can be replaced by ARLIR, u by (HL)II2 and LIH by N , as in chromatography (11).Therefore we get Rs = (N/16)'I2ARIR
(9)
where T is temperature, w is column width, and dTldx is the temperature gradient. We get
When this is substituted back into Equation 8, we obtain
-=dR d l n R -da R dlnX( a
)
The negative sign has arisen because R rises as CY falls in value. We are interested only in the absolute values of increments, and if we define A!? = I dRI and Aa = 1 dal , we get the limiting incremental relationship
AR dlnRAcu -=-R
dlnX
CY
The resolution of Equation 6 therefore can be expressed approximately as
The three terms on the right represent contributions to resolution from, respectively, column efficiency as reflected by the number of plates, the fractional gain in R values relative to the fractional increase in X, and last, the selectivity between components as measured by the fractional increment, A d a , in the thermal diffusion factor. A comparable equation exists for chromatography (111. I t can be written
N
(6)
Rs =
(16)
112
AK
(1 - R ) -
K
where N is the number of theoretical plates and L the column length. Basically, separation in TFFF results from some finite increment, Aa, in the thermal diffusion factor, a (7, 9). We can best see the effect of a given Aa on resolution by relating the AR of Equation 6 to Aa. T o do this we use differential quantities, dR and da. The differential increment, dR, can be written as
where AK is the increment in distribution coefficient K for two substances. The center term of the last two equations interests us here. This term shows the rate of deterioration of Rs with increasing R at constant N and at constant a and Aa (TFFF) and K and AK (chromatography). In other words, the center term represents a relative resolution, expressible as
dR dX da dR =-(7) dX d a Therefore dRIR, corresponding to the ARIR of Equation 6, is
center term = Rs/Rs(R = 1)
dR d l n R d X d a (8) R dX d a The dependence of X on CY is given approximately by Equation 9 -e--
2390
* ANALYTICAL CHEMISTRY, VOL.
(15)
The magnitude of d In Rld In X in the TFFF case can be derived from Equation 2. Quantity 1 - R for chromatography is calculated directly. These two functions are plotted in Figure 2. This graph shows clearly the rapid deterioration of resolution with increasing R. The deterioration is worse for chromatography than TFFF but, in both cases, the resolution approaches zero as R approaches unity. Compensation by Increased Plate Numbers. The res-
47, NO. 14, DECEMBER 1975
olution lost due to increasing R values can, in theory, be regained by an increase in the number of theoretical plates, N, as, for instance, by increasing column length. If one regains exactly the same level of resolution, Rs, it can be shown by rearranging Equation 13 that the relative increase in required plates is of magnitude
50 '
'I1
40 L
IY
N(R = 1) For chromatography, the equivalent expression is obtained by the rearrangement of Equation 14. --=- N N(R = 1)
1 (1 - R)'
(17)
These quantities are plotted in Figure 3. This figure shows clearly the enormous magnitude of the additional plates needed to offset a decreasing retention. At R = 0.9, for example, N must be increased almost 30-fold above the limiting value necessary a t R = 0. Chromatography is even worse: one-hundred times more plates are needed to offset an R value of 0.9. Relative Gains in Resolution by Increasing AT. If additional plates are not provided to compensate for increases in R, then resolution must be sacrificed, as indicated in Figure 2. The slope of the curve a t R = 1 expresses the limiting rate of this deterioration; likewise, it indicates the rate of improvement in resolution for increasing retention or decreasing R . It will be shown elsewhere that Equation 2 acquires the form
as X
-
and R
-
R = 1 - 1/60X2
(18)
1. We find from this that (dRldX) = 1/
30X3 and therefore that the center term of Equation
comes
d In R --
-
d In X
or, in the limit as X
13 be-
1/30X2
1 - 1/60X2
dlnR -=dln X
1
(20)
30X2
Inasmuch as X = +/AT(mol wt)1/2,Equation 1, we have d In_R_--AT2 mol wt _ d In X
(22)
-
-)
Thus, resolution in this range increases with the square of the temperature increment, AT. However, limitation of this conclusion to conditions of poor retention (high A) must be kept in mind. The limiting expression, Equation 18, which underlies this conclusion is valid to within 1.5% a t R = 0.9 and 9.1% a t R = 0.8. Therefore, the validity of Equation 22 and any conclusions based on it deteriorates rapidly below R = 0.9. Resolution can also be expressed in terms of R instead of X in this low retention limit. The combination of Equations (23) Therefore Equation 13 becomes
N Rs =
1/2
(16) (R
-
ACY 2(1- R ) CY
1)
Chromatography
;:/
I
R Figure 3. Relative increase in number of theoretical plates needed to maintain a constant resolution as R increases
-
This shows explicitly how resolution degenerates as R 1. Interestingly, this is identical to Equation 14 for chromatography except for the factor of 2 preceding 1 - R. Thus, for identical plate numbers and relative increments in the physical properties (CY and K ) responsible for separation, TFFF has a resolution twofold better than chromatography in the low retention limit. It is recognized, of course, that low retention is not an inherent limitation of chromatography except for particular systems and for very small solute molecules. Despite the limitations of Equation 22, it clearly provides a rationale for increasing the temperature increment AT in TFFF. The practical means for augmenting AT will be discussed next.
MEANS FOR AUGMENTING AT
30+2
Resolution, of course, is proportional to this term, Equation 13. Use of the latter gives
(A
I0
The value of AT = T H- T , can be increased by increasing the hot-wall temperature, T H ,or decreasing the coldwall temperature, T,. Increases in T H are limited by the boiling point of the solvent. A deterioration of peak shape and retention, along with the onset of bubble formation, appears to occur a t 20-30 "C below the boiling point. Therefore, even with a high boiling solvent like ethyl benzene (bp = 136 "C), T H cannot much exceed 100 "C. In theory T , is limited by the freezing point of the solvent. This is -95 "C for ethyl benzene. A value near this limit could be achieved through a refrigerated coolant. However, the heat flux in typical TFFF devices is so large (up to 1000 watts in normal systems; 2500 watts in high AT systems) that it precludes convenient refrigeration to lower the cold-wall temperature. We have therefore retained the use of tap water as a coolant. With tap water at 10-13 "C, the cold wall temperature rises to approximately 20 "C as a result of the heat transfer. In view of these considerations, we have chosen to augment AT by raising T Hthrough the increased boiling point of a pressurized system. This method, of course, could be used in conjunction with a reduction of T , to obtain a double increase in AT. In the present work, however, we have used only the pressure-induced boiling point elevation.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 14, DECEMBER 1975
2391
TOLUENE + PS 600
A AT = 00'C
- Colculoted using Troutmb Ruk
,
IO0 I
2
3
4
5
6
,
,
,
,
,
7
8
9
IC
I
,
,
12
,
13
I4
1
,
1
,
,
16
,
I5
I7
18
19
,
B
2021
TOLUENE
P, atmospheres
Flgure 4. Elevation of boiling point, Tb, as a function of pressure, P, in atmospheres
A simplified treatment can be used to show the order of magnitude of the gain in T b . Integration of the Clapeyron equation under the assumption that the enthalpy of vaporization, AH,, is constant and that the vapor is ideal yields
where
TbO
is the normal boiling point occurring a t pressure
PO = 1 atm and R, is the gas constant. Upon solving for
Figure 5. The increase in retention of a 600 mol wt polystyrene frac-
tion through a doubling of AT
Tb/TbO, we have
where A s , is the entropy of vaporization, m,/Tbo, at TbO. This parameter is normally about 21 cal mol-l deg-l, according to Trouton's rule. Pressure P , of course, must be expressed in atmospheres. Equation 26 can also be rearranged to show the incremental gain in boiling point, A T b = T b - TbO, as a function of amlied Dressure P. I.
Figure 4 shows a plot of Tb/TbO from Equation 26 using both ASv = 20 and 22. For comparison, experimental curves are shown for several solvents employed in TFFF (12). These solvents fall rather symmetrically between the two theoretical curves; we, therefore, find it practical to use AS, = 21 f 1 cal mol-' deg-l to describe the augmentation of Tb for typical TFFF. Figure 4 shows that significant increases in T b can be gained by the application of only modest pressures. At 5 atm, one has a gain of better than 15%;for a solvent whose normal boiling point is 400 K, this represents a 60 K increase in T b . At 10 atm, the gain exceeds 25% or 100 K. In the case of ethyl benzene, the solvent used in this study, the normal boiling point is 136 "C. This becomes 208 and 246 OC a t 5 and 10 atm, respectively. In this study, we chose a minimum operating pressure of 8 atm, which yields a T b of approximately 233 OC. This provides a good margin for working at a hot-wall temperature, TH, of 185 "C. While in theory a somewhat higher value is possible under these conditions, a different spacer material would be required as Mylar begins to soften a t 200 "C. This temperature is adequate, however, to yield a AT of almost 160 O C , almost twice as high as any previous value used in TFFF.
EXPERIMENTAL The general configuration of the TFFF column was the same as that of the copper column described in a previous paper (9). The 2392
channel was 2 cm in breadth by 0.0245 cm wide by 45.5 cm long, with the ends tapered smoothly a t 45 degrees to the column axis in order to converge on the 1-mm inlet and outlet holes in the upper bar. Injections were made with a Carle Micro Volume Chromatographic Injection Valve. Sample loops were 5 pl and 7 p1 in volume. The quantity of solute injected was between 0.033 and 0.060 mg for polystyrene and 0.165 mg for crude oil samples. Solutes were dissolved in toluene or ethyl benzene prior to injection. The pressure in the column was controlled by a Nupro Fine Metering Valve a t the outlet of the column. Carrier flow and pressure were provided by a Chromatronix CMP IV Pump. Pressure in the column was maintained in the range 105-125 psi above atmospheric. Linear fluid velocity ( u ) was 0.025 cmhec. Detection was accomplished by means of a Waters Associates R401 Refractive Index Monitor. Ethyl benzene was used as the solvent. Heating was accomplished by the use of two 1500-watt cartridge heaters controlled by variable transformers. Because of the large heat input necessary to produce the 185 "C temperature of the upper bar, some variance in temperature across the breadth of the channel was observed. This was in all cases less than 4 "C with the maximum being in the center of the channel. There was also a variation of 5 "C along the length of the column. The retention ratio, R , was measured by comparing the elution time of the sample to that of toluene or other essentially nonretained solvents in which the sample was dissolved. In the case of samples not well resolved from the inert peak, a check was made by alternating injections of sample and of unretained solvent. The average retention time of the unretained peaks appearing directly before and after the sample peak was used for the void value. The two void peak values in this case always agreed within 2%. Samples of crude oil were obtained through the courtesy of the Phillips Petroleum Company refinery in Woods Cross, Utah, and also from the University of Utah, Department of Fuels Engineering. Asphalt samples were also obtained from the Phillips Petroleum Company. The asphaltenes were derived from the asphalt samples by standard extraction procedures. Polystyrene samples were obtained from Pressure Chemical Company.
RESULTS AND CONCLUSIONS Figure 5 supports the general theoretical conclusion that a large increase in A T significantly increases the low molec-
ANALYTICAL CHEMISTRY, VOL. 47, NO. 14, DECEMBER 1975
A
TOLUENE
TOLUENE
A
AT = 80°C
TOLUENE
B
B
TOLUENE
ps
2/00
A T = 158°C
Flgure 6. Illustration of the principle that a doubling of A T will allow a reduction of approximately four in mol wt without loss of retention and resolution
ular weight range of TFFF. This figure illustrates the case in which a series of runs was made using only polystyrene of molecular weight 600. At first (Figure 5a), runs were made at a A T of 80 "C. For comparison, the same sample material was run a t a A T of 158 OC (Figure 5b). The only evidence of the existence of 600 mol wt polystyrene a t A T = 80 "C is a slight broadening of the void peak when compared to a void peak with no polystyrene. At A T = 158 OC, by comparison, the 600 mol wt polystyrene peak has an R value of 0.90. While not completely resolved from the void peak, it is, nevertheless, easily detected. The fractograms of Figure 6 compare the elution behavior of 2100 mol wt polystyrene at A T = 80 O C AT and 600 mol wt at 158 O C . The average R value of the former was 0.91 and the latter 0.90. The two polystyrenes are almost equally resolved from the inert peak, although the retention and resolution are slightly superior for the 600 mol wt fraction. This result is in good agreement with the conclusion deduced from Equation 4 that a fourfold reduction in workable molecular weight can be obtained through a doubling of A T . In this instance we have reduced the mol wt by a factor of 3.5 while essentially doubling AT, and have retained full resolving power and retention. Figure 7 illustrates the general improvement with increasing A T of retention and resolution with a mixture of two low mol wt polystyrenes. In Figure 7a, corresponding to A T = 80 "C, no peak resolution is evident, although a distinctive shoulder was formed due to the presence of the 2100 mol wt component. Both peaks are moderately well resolved in 158 "C, Figure 7 b . The substantial decrease in tractable molecular weight occasioned by high LITTFFF prompted a brief investigation to see what characterization might be made of some naturally occurring substances. To this end, we have employed samples of crude oil, asphalt, asphaltenes, and pine tar. Characteristic elution patterns were observed for each of these. The results of injecting a number of crude oil samples AT = 158 "C are shown in Figure 8. A toluene (void) peak and a 2100 mol wt polystyrene peak are also shown for
Figure 7. Fractograms of toluene (inert), 600 mol wt polystyrene and 2100 mol wt polystyrene at (a) A T = 80 OC and ( b ) A T = 158 OC
Figure 8. Fractograms of crude oil samples at A T = 158 OC, show-
ing the generation of characteristic elution patterns. Void peak and 2000 mol wt polystyrene peak shown for comparison TOLLEVE
,I
Figure 9. Fractogram of asphalt and asphaltenes run at A T = 158
OC. Void peak shown for reference
comparison. All the crude oil samples show the same general trend of producing a negative peak at the void volume followed by a positive peak which gradually returns to base line. There are, however, definite differences in both the relative sizes of the positive and negative peaks as well as in the shape of the positive peaks, indicating differences in the composition of the samples, particularly in the relative amounts of some of the higher molecular weight constituents. From this, one can readily identify specific crudes or, more generally, identify those samples with the largest percentage of the low molecular weight fraction. The differ-
ANALYTICAL CHEMISTRY, VOL. 47, NO. 14, DECEMBER 1975 * 2393
ences observed correlate with a qualitative observation of the viscosity differences of the samples. The runs of the asphalt sample and of the asphaltenes extracted from it are shown in Figure 9. Again one gets a distinct distribution of the different materials along the elution-volume axis. Some comments are appropriate at this point on the expected properties of the high-AT TFFF in the characterization of complex materials. We note, first, that the elution sequence is from low to high molecular weight, unlike exclusion (gel) chromatography. The low molecular weight components tend to clump as a single peak while resolution improves continuously for the heavier components. Gel permeation chromatography (GPC) also shows better resolution for larger molecules, providing these are in the active partitioning range (13).However, resolution in GPC eventually declines with further increases in molecular weight, whereas with TFFF resolution is expected to reach a steady plateau (10). More important are the molecular characteristics which determine elution. A steady and fairly predictable increase in elution volume occurs with increase in molecular weight for any homologous series. However, unlike GPC, TFFF exhibits a solvent effect, in which a change of solvent will alter the magnitude of retention but not the pattern with respect to molecular weight. Presumably a change of solvents will shift different solute families with respect to one another, although this has not been confirmed. If so, variations in solvent could become a useful tool in identifying
different chemical families and the molecular weight distribution within families. This possibility is not immediately present in chromatographic methods. This results because the exclusion techniques are basically solvent independent and the other chromatographic techniques do not yield broad, readily-characterized molecular-weight spectrums. The TFFF method will, in all likelihood, permit the advantageous combination of these two characteristics.
LITERATURE CITED (1)J. C. Giddings, Sep. Sci., 1, 123 (1966). (2)J. C.Giddings, J. Chem. Phys., 49, 1 (1968). (3)E. Grushka, K. D. Caldwell. M. N. Myers, and J. C. Giddings, Sep. Purif. Methods, 2, 129 (1973). (4)J. C. Giddings, J. Chem. Educ., 50, 667 (1973). (5)J. C. Giddings, Sep. Sci., 8, 567 (1973). (6)G. H. Thompson, M. N. Myers, and J. C. Giddings, Anal. Chem., 41, 1219 (1969). (7)M. E. Hovingh, G. H. Thompson, and J. C. Giddings, Anal. Chem., 42, 195 (19701. -, (8)J. C. Giddings, M. E. Hovingh, and G. H. Thompson, J. Phys. Chem., 74, 4291 (1970). (9)M. N. Myers, K. D. Caldwell, and J. C. Giddings, Sep. Sci., 9, 47 (1974). (10)J. C. Giddings. Y. H. Yoon. and M. N. Myers, Anal. Chem., 47, 126 (1975). (11)J. C. Giddings “Dynamics of Chromatography. Part 1. Principles and Theory”, Marcel Dekker. New York. 1965. (12)“Handbook of Chemistry and Physics”, 45th ed., Chemical Rubber Co., Cleveland, Ohio, 1964,p D119. (13)J. C. Giddings, Anal. Chem., 40, 2143 (1968). \
RECEIVEDfor review July 3, 1975. Accepted August 26, 1975. This investigation was supported by National Science Foundation Grant Number MPS74-05260 A03.
Proton Magnetic Resonance Study of the Effect of Water, Acetonitrile, and Benzonitrile on Diprotonated 2,2’Dipicolylamine M. R. Olds and R. T. lwamoto Department of Chemistry, Universityof Kansas, Lawrence, Kan. 66045
The protonation of 2,2’-dipicoiyiamine in water, acetonitrile, and benronitriie has been examined by proton magnetic resonance. The protonation scheme of the amine in each of these solvents has been established, and by the use of substituent shielding constants for the methylene group, obtained from appropriate model compounds, the distribution of protons among the three basic sites for the diprotonated stage in each of the three solvents has been determlned.
A significant number of compounds of biochemical importance are multibasic substances, which undergo successive protonation or metal ion coordination. It is generally accepted that the protonation scheme of multibasic compounds or the proton distribution pattern of some of the multiprotonated species will vary with the nature of the medium. There is, however, no information as to how large a variation can be expected. T o obtain a better appreciation and understanding, in particular a quantitative picture, of the effect of solvent on the protonation of multibasic substances, we have examined the protonation of 2,2’-dipicolylamine, a tribasic amine, in water (DzO) [dielectric constant, 78; dipole moment, 1.86 D], acetonitrile 2394
[dielectric constant, 37.5; dipole moment 3.44 D], and benzonitrile [dielectric constant, 25.2; dipole moment, 4.02 D]. Future studies with metal ions are planned. The protons of a diprotonated symmetrical noncyclic tribasic compound can be located on neighboring or terminal basic sites. Differences in the ability of solvents to solvate the protonated sites should lead, through differences in relief of coulombic interaction, to varying extent of isomerism of position, Le., varying percentages of the two diprotonated forms. Grunwald, Lowenstein, and Meiboom ( 1 ) , and Lowenstein and Roberts (2) reported that the chemical shift of methyl and methylene hydrogens adjacent to basic groups is sensitive to changes in the electronic environment brought about by protonation and that there is a linear relationship between the shielding contribution of a basic site and the fraction of time it is protonated. Furthermore, Shoolery has demonstrated the additive nature of the shielding effect of substituents on the signal of the methylene protons of compounds of the type XCHzY ( 3 ) . Importantly, the chemical shift of the methylene hydrogens then should undergo a predictable change in going from XCH2Y to X’CHzY, X’CH2Y’, or XCH2Y’, where X’ and Y’ are designations for protonated basic sites.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 14, DECEMBER 1975
