J. Am. Chem. SOC.1992, 114,9527-9532
9521
Interpretation of Spectroscopic Changes upon Adduct Formation and Their Use To Determine E and C Parameters Russell S.Drago*.+and Glenn C.Vogelt Contribution from the Department of Chemistry, University of Florida, Gainesville, Florida 3261 1, and Department of Chemistry, Zthaca College, Zthaca, New York 14850. Received March 23. 1992
Abstract: Spectroscopic probes have been examined to determine whether the spectral shifts accompanying adduct formation are dominated by donoracceptor interactions. New spectral acceptors which employ IR, NMR, or visible spectral shifts are established. Since the CA*/EA*ratios for these spectral probes differ from those of acceptors whose spectral response is the change of their OH stretching frequency, these new spectral correlations greatly simplify the addition of new donors to the E and C database. Several spectral acceptors, reported to be useful indicators of donor strength, are shown to be insensitive to, or not dominated by, donor-acceptor interactions.
Introduction The recent extension of the electrostatic-covalent (E-C) approach' -AH = EAEB + CACB + W (1) to gas-phase, ion-molecule reactionsZand to bond energies3 has served to emphasize the importance of this quantitative application of the early Pauling" and Mullikens bonding models to an immense area of chemistry. In this article, the role that such considerations play in the understanding of a variety of spectroscopic shifts is evaluated. The finding of spectroscopic relations, dominated by bond energies, is very important not only to the expansion of the E-C approach to new areas of chemistry but also for the incorporation of new donors (bases), B, and acceptors (acids), A, into the E and C database. In order to add a new donor molecule, the enthalpy of reaction of this donor is measured toward four or more acceptors with known E A and CA parameters, and the four or more simultaneous equations are solved for E B and CB.If a constant energy contribution to the measured enthalpy is present or suspected, W is also solved for a third unknown and more measurements are needed. Otherwise, W is set equal to zero. The effort expended to carry out calorimetric measurements to add a new donor or acceptor to the correlation is extensive. Physicochemical properties, Ax, have been analyzed with the E-C approach.' For example, the spectral shifts, Ax, for a given acceptor bonding to donors with known EB and CBparameters are substituted into eq 2. The series of simultaneous equations AX = EA'Eg + CA*CB + W. (2) are solved for EA*, CA*, and WI. These values lead to the calculated A x by substituting them into eq 2 along with reported EB and CBenthalpy-based parameters. Thus, the fit of A x to eq 2 is a tit to enthalpy-based parameters. In treating spectral shifts, asterisks indicate the spectroscopic counterparts of the enthalpy parameters (EA, CA,and W)in eq 1 and are referred to as spectral acceptor parameters. Since EB and C, are the eq 1 donor parameters in units of (kilocalories mole-l)l/z, EA* and CA*also contain the pertinent conversion factors. The existence of spectral correlations with the E and C parameters is of value in understanding the spectroscopy. They also provide additional simultaneous equations (eq 2 ) that can be employed in the determination of the EB and CBparameters for a new donor that is not part of the E and C database. Spectroscopic data are easier to measure than enthalpy changes accompanying adduct formation, so that valid spectral acceptor correlations facilitate the expansion of the database to new donors. The value of the CA*/EA* ratio of a spectroscopic correlation is very important when adding a new donor to the E and C da-
'University of Florida. * Ithaca College.
tabase. If, for example, the CA*/EA* ratio is the same for two or more spectral acceptors, the two or more equations are not independent, and only one of them is in effect providing any information to solve for the two unknowns, EB and CB.When the cA*/EA* ratios of the acids used to solve for EB and cB differ only slightly, small errors in the measured shifts lead to a shallow minimum and large errors in the values of EB and CBfrom the solution of the simultaneous equations. Thus, acids with a wide range of values for their cA*/EA* and C A / E A ratios are needed to determine CBand EB. The change in the OH stretching frequency of phenol upon complex formation is a well-established correlation6 with EA* = 167, CA*= 109, W. = 205, and a CA*/EA* ratio of 0.65. Other alcohols give correlations with similar CA*/EA* ratios. Thus, if a series of measurements is to be carried out to add a new donor to the E and C database, there is no point in measuring more than two or at most three OH frequency shifts. Spectral acceptors with different values of CA*/EA* ratios and acceptors with different c A / E A ratios are needed to accurately define the EB and CBvalues for a new donor. In this article, we shall examine a series of spectral shifts with eq 2. In so doing, we gain an appreciation of the factors influencing the spectroscopy and evaluate the utility of various spectral measurements in determining EB and CB parameters for new donors. Such information is particularly essential for the interpretation of gas-phase ion-molecule reactions with the ECT a p proach? Because of the limited range of the c A / E A ratios found2 for gas-phase ions, it is difficult to use only gas-phase data to characterize the properties of new donors. Gas-phase measuren ~ ~ spectral ~ acceptors to add ments can be used in c o n j u n c t i ~ with new donors to the E and C database.
Results and Discussion Spectral changes have been used to assea reactivity parameters and to provide experimental justification of Mulliken's chargetransfer model.s In order to assess the electrostatic-covalent bonding contribution to a spectral change, data are selected in which the nonspecific solvation contribution to the measured (1) Drago, R. S. Coord. Chem. Rev. 1980, 33, 251. (2) Drago, R. S.; Ferris, D. C.; Wong, N. M. J . Am. Chem. Soc. 1990, 112,8953. (3) Drago, R. S.; Wong, N. M.; Ferris, D. C. J . Am. Chem. SOC.1991,
-I .13.1970. - , .. - . (4) Pauling, L. The Nature of the Chemical Bond, 3rd ed.; Cornell University Press: Ithaca, NY, 1960. (5) Mulliken, R. S. J . Am. Chem. Soc. 1950, 72,600,4493; 1952, 74,811. (6) (a) Joesten, M. D.; Drago, R. S. J . Am. Chem. SOC.1962,84,3817. (b) Epley, T. D.; Drago, R. S. J . Am. Chem. Soc. 1967,89,5770. (c) Drago, R. S.; O'Bryan, N.; Vogel, G. C. J . Am. Chem. SOC.1970, 92, 3924. (d) Doan, P. E.; Drago, R. S. J. Am. Chem. SOC.1982, 104,4524. (e) Nozari, M. S . ; Drago, R. S. J. Am. Chem. SOC.1970, 92, 7086-90.
0002-7863/92/1514-9527%03.00/00 1992 American Chemical Society
Drago and Vogel
9528 J. Am. Chem. SOC.,Vol. 114, No. 24, 1992 Table I. Exmrimental and Calculated Avnu Values (cm-l) for a Series of Adducts of Methanol and tert-BiCyl Alcohol (CCI, Solvent) CHqOH
126 (109) 150 (140) 83 (84) 286 (286) 429 (370) 304 (305) 160 (116) 205 (199) 274 (269) 115 (112) 75.5 (81) 179
(135) 118 115 152 127 128 87 75 77 284 23 1 231 (386) 307 (172) 143 145 205 181 (172) (265) 120 101 102 70 63 63 160 159 (189) 115 101 100 47 44 45 185 76 (165) 133 144 157 "Values from ref 7; values in parentheses in the experimental data column are from ref 8. *Calculated using EA* = 107.7, CA* = 70.0, and WC = -155.8. The % fit is 1.1, the CAI& range is 0.2-16.3, the value of A is 3 cm-I, and the experimental error is 3.5 cm-I. The values in parentheses under the A~oH(~a1cd) column are omitted from the fit. They are calculated from the E,*, CA, and WC parameters obtained by fitting the r a t of the data. CDatafrom ref 8. dCalculated using EA* = 88.95, CA* = 54.54, and WC = -121.0. The value of A is 3 cm-I, the experimental error is 3-5 cm-I, the 8 fit is 5
Table III. Change in C-I Stretching Frequencies v (cm-I) of Donor Adducts of ICN" and ICJ (1:l Adducts) donor (EtO),PO CH3C(0)CH3 HC(O)N(CH3)2 (CHhSO (C2H5)20 O(C2Hd20 C5H5N 3-CICsHLN
AvICN
A'ICN
(exptl)' 26.0 18.0 30.0 32.0 20.5 17.0 57.5 45.5
(calcd)b 23.1 23.4 26.5 30.0 (29.3)d 24.4 57.9 (49.5)'
AVlCll
AVIC21
AYICN
AYICN
donor (exptl)" (calcd)* 4-CH3CSH4N 61.5 63.5 88.4 (C2H5),N 88.0 (CH3)3N 96.0 (86) (36.9)d (CH2)IO 25.0 CH3CN 17.0 14.6 (CH2)dSO 34.0 32.7 (C2H5)2S 59.0 55.6 3-CHGHLN 61.5 (60.5)' AVlC I
AVlC I
donor (exptl)" (calcd)' donor (exptf)" (cal& (C2H5)20 6.5 8.3 HCON(CH3)2 12.5 (6.7)d 18.5 18.5 O(CH2)40 7 6.4 (C2HshS 7 6.0 4-CHICSH4N 22 21.7 (CHWO 31.5 31.5 (CH2140 9 (11.2)d (C2HS)3N 3-CHiCsHdN 21.0 (20.6)' "Reference9 and references therein. bCalculatedwith EA* = 5.2, CA* = 15.0. and W.= -4.6. Data in C6H6. The value of A is 2.8 cm-I. The Ifit is 3.8, and the CB/EBrange is 0.38-16.3. 'Calculated with E* = 2.10, C = 5.90, and W.= -5.09. Data in CS2solvent. The value of A is 0.6 cm-'. The % fit is 1.8, and the CB/EBrange is 0.69-16.3. dThe values in parentheses were omitted from the fit and were calculated from the resulting EA*, CA*,and W.spectral parameters. COmittedfrom fit because its E B and CBparameters are tentative.
and the shifts range over 260 cm-l (the calculated range is used). The percentage fit, %F,is given by %F = (A/response range)100 (3) Thus, (3/260) 100 is a fit of 1.1%. This percent fit is comparable to the enthalpy fit of f0.2 kcal/mol for enthalpies that range up to 10 kcal/mol or a fit of f0.3 kcal/mol for enthalpies that range from 15 to 30 kcal/mol. The above criteria indicate that the frequency shift is dominated by the donoracceptor bond energy. Three criteria should be used to judge the overall quality of the fit: (1) the range of CB/EB values in the measured data; (2) (A)/(average experimental error); and (3) the percentage fit. Table I1 gives the guidelines that have been successfully used in making judgements for spectral acceptors. Their usefulness will be shown with the ECW analyses that follow. Before proceeding, two points concerning the percentage fit should be made. If the percentage fit is very poor and a large range of CB/EB is used in the fit, the E*C*W* model has not failed, but the spectral shifts are not related to bond strengths. Experiments should be devised to investigate what factors other than bond strengths are controlling the shifts. The second point involves the situation in which the percentage fit is excellent but the CB/EB range is fair to poor. This situation is illustrated by the data for tert-butyl alcohol OH shifts in Table I. The percentage fit is