J. Phys. Chem. 1994,98, 12949-12957
12949
Understanding the Electronic Properties of Glycosylated Chromophores Using AM1 Semiempirical Calculations J. P. Rasimas and G. J. Blanchard* Department of Chemistry, Michigan State University, East Lansing, Michigan 48824-1322 Received: August 2, 1994@
We have investigated the electronic properties of carminic acid and glycosylated oxazines, phenoxazines, oxazones, and phenoxazones using AMI semiempirical calculations to predict and understand the spectroscopic behavior of these molecules. These calculations have been used to predict electronic state ordering, statedependent changes in charge distributions, and isomerization barriers for the molecules we report here. Comparison of fully protonated (neutral) carminic acid (H3CA) with the monodeprotonated form (HzCA-) shows that transition energies and state ordering change with the extent of protonation, but isomerization barriers for both forms are similar. We have also studied glycosylated oxazines and oxazones. The state ordering of the native and glycosylated oxazines and oxazones shows that glycoside modification does not affect the state ordering, and the rotational isomerization barriers for the glycosylated species show two minima separated by significant energy barriers. For the oxazines, there is a significant increase in SO S1 transition energy associated with glycosylation, indicating the importance of steric effects associated with the chromophore modification. These calculations predict the formation of a twisted internal charge transfer (TICT) S1 state in glycosylated oxazines arising from steric effects. These calculated data will be used as a tool for understanding the static and dynamic spectroscopic properties of these molecules as they are used to probe the nucleation and crystallization behavior of sugar solutions.
Introduction Research into understanding the interactions between dissimilar molecules has provided a great deal of information on structure and reactivity in both the solid and gas phases. A significant factor in achieving this progress has been the comparatively well-defined nature of intermolecular interactions in these two phases. In sharp contrast, achieving a fundamental understanding of intermolecular interactions in liquids and solutions has proven to be extremely difficult, primarily because of the short time and length scales over which molecular organization persists in this phase. The majority of experiments that provide information on the nature of intermolecular interactions in the liquid phase have used optical spectroscopy because of the short time scale(s) on which any organization is present in such systems. For these experiments, it is possible to examine neat liquids using stimulated transient Raman techniques,lq2 or more typically, a probe molecule is used to “read” information out of a s ~ l u t i o n . ~ -The ~ use of probe molecules for this purpose has found wide acceptance, and in many instances it is assumed that the probe molecule is a passive monitor of the surrounding medium, i.e., probe molecule intramolecular processes are well understood, and that there are no site-specific intermolecular interactions between the probe molecule and its surroundings. Both of these assumptions serve, of course, to simplify data interpretation and in many cases they have tumed out to be valid, but there exist a variety of experiments and classes of probe molecules for which one or both of these assumptions have proven to be limiting.8-9128 Understanding the optical and chemical properties of the probe molecule is therefore a necessary prerequisite for the successful interpretation of transient spectroscopic data aimed at understanding molecular scale processes in the liquid phase. The very properties that make probe molecules useful, such as a prominent optical response at frequencies accessible to short
* To whom correspondence should be addressed. @
Abstract published in Advance ACS Abstracts, November 15, 1994.
0022-365419412098-12949$04.50/0
pulse lasers or a large change in dipole moment on excitation, can also serve to complicate the optical response of the molecule and thus obviate straightforward data interpretation in terms of local organization. Indeed, there is a large body of information in the literature indicating the important role that intramolecular processes play in the transient optical response of certain probe molecules and that the transient absorption and/or emission characteristics of a given probe molecule may not be related to the local environment of the probe molecule in a straightforward way.3-5q8,28 Our previous work on several families of probe molecules demonstrates that there is an excellent correspondence between experimentally measured data and semiempirical calculation^.^^^^-^^ For example, the state-dependentreorientation dynamics measured for oxazines and thiazines in polar protic solvents were consistent with a site-specific solventsolute interaction, and semiempiricalcalculationsindicated that, on excitation, the oxazine chromophore accumulated significant electron density at its heterocyclic n i t r ~ g e n . ~ ~ In. addition, ~~J~ the transient optical response of the coumarins has been shown recently not to be accounted for simply by the presence of a single, uniform (shifting) electronic state.* Semiempirical calculations predicted the existence of several excited electronic states in close energetic proximity, and these predictions provided an explanation consistent with a large body of experimental data. Thus, there is compelling experimental evidence that validates the qualitative predictive power of semiempirical calculations for moderately large polar organic molecules. Most examinations of solvent-solute interactions have been performed in binary systems, i.e., dilute solutions of a probe molecule in a single organic solvent. Our present interest lies in understanding local structure in temary systems where, in addition to the solvent and the solute, there is a third constituent that exhibits limited solubility in the solvent. Specifically, we are interested in understanding molecular organization associated with the onset of crystallization in sugar solution^.'^^^^ In order 0 1994 American Chemical Society
12950 J. Phys. Chem., Vol. 98, No. 49, 1994
Rasimas and Blanchard
to examine these systems, we have chosen to use a family of probe molecules containing a glycoside moiety to facilitate the incorporation of the probe molecule into local structure(s) formed in the sugar solutions. The first step in this research program is to develop a detailed understanding of the probe molecules we will use and address several basic questions that cannot be answered directly by experiment. We report in this paper on a semiempirical computational study of three classes of molecules: carminic acid, glycosylated oxazines, and glycosylated oxazones (Figure 1). The questions we address regarding these molecules are the following: (i) Can we expect a simple linear optical response from these molecules, or might intramolecular relaxation effects be expected to play an important role in the experimental data? (ii) How does the pH of the solution affect the optical response and state ordering of carminic acid? (iii) For oxazines and oxazones, how does the addition of a glycoside moiety affect the optical response of the chromophore? (iv) Will rotation of the chromophore about its bond to the glycoside moiety affect its optical response? Our calculations indicate that all of these chromophores will exhibit a reasonably straightforwardlinear optical response, with the potential for intersystem crossing being the most significant concern. The deprotonation of carminic acid is calculated to cause a spectral blue shift and move the first triplet state into close energetic proximity to the first excited singlet state. For the oxazones, we calculate that the addition of the glycoside moiety will affect the optical response of the chromophore to only a small extent, while for the oxazines we observe, in addition to the previously reported accumulation of charge at the ring bound nitrogen on excitation, the formation of a stable twisted internal charge transfer conformation in the S1 state that arises from steric hindrance due to the presence of the glycoside moiety. This steric hindrance also gives rise to a substantial (-10%) blue shift of the SO S1 transition. Importantly, for all of the chromophores, rotation about the chromophoreglycoside bond does not affect the energy of the SO SI transition, and therefore we need not be concerned about any conformation-specific contributions to the linear response. These calculations serve to predict the qualitative features of the optical response(s) of these molecules and to answer fundamental questions about the effects of substitution on chromophores that, in their native forms, have received significant experimental attention. We note that, in addition to the uses we will have for these probe molecules, similar glycosylated chromophores have found application in fluorescent immunoassays and other biologically important fluorescencebased
-
-
Experimental Section Semiempirical calculations were performed using the Austin Model 1 (AM1) parametrization running on Hyperchem software (Release 3.0; Autodesk, Inc.) on an IBM compatible PC (Gateway 2000 486-66V). The AM1 p a r a m e t r i z a t i ~ n ~is~a- ~ ~ modification of the MNDO p a r a m e t r i z a t i ~ n that ~ ~ * is ~ ~more accurate for polar molecules and transition states. The calculation strategy for the molecules shown in Figure 1 was to perform an initial optimization of the structures using a molecular mechanics routine (MM+),” followed by geometry optimization at the semiempirical level using an SCF algorithm. Two structures for carminic acid were calculated, the fully protonated form (H3CA) and the singly deprotonated form (HzCA-). H2CA- was formed by removing the acidic proton from the carboxylic acid functionality. Semiempirical optimization was performed until the lowest energy conformation for each molecule was attained. Electronic energy calculations were
OH
0
Figure 1. Structures of molecules for which AM1 calculations were performed. Only one resonance form is shown for each structure. The glycosyl moiety is shown once for simplicity, its position in each molecule is noted: (a) glycosyl moiety; (b) carminic acid; (c) oxazine; (d) A = glycoside, B = H for 2-glycosyloxazine; A = H, B = glycoside for 18-glycosyloxazine; (e) phenoxazine; (f) A = glycoside, B = H for 2-glycosylphenoxazine; A = H, B = glycoside for 18-glycosylphenoxazine; (g) oxazone; (h) A = glycoside, B = H for 2-glycosyloxazone; A = H, B = glycoside for 18-glycosyloxazone;(I) phenoxazone; (m) A = glycoside, B = H for 2-glycosylphenoxazone;A = H, B = glycoside for 18-glycosylphenoxazone.
performed on the geometrically SCF optimized molecules. For these calculations, the ground state optimized geometry was used, and RHF closed-shell calculations were performed for single configuration interaction (CI) with 100 microstates. Previously, similar calculations using a large number of CI states were believed to provide a fair approximation of correlation effects in coumarins,28and we believe this condition will hold for these calculations as well. It is important to recall that these calculations assume the molecules are isolated and in an absolute vacuum.
Results and Discussion Carminic Acid. Calculations on carminic acid were performed to understand the electronic properties of both H3CA and H2CA- as well as to determine isomerization barriers for rotation of the glycosyl moiety about its bond to the chromophore. Understanding the pH dependence of the optical response of this fluorophore is important because pK,1 = 2.8 (deprotonation of the carboxylic acid) for carminic acid. Additionally, carminic acid possesses two other labile phenolic protons with pKd = 5.43 and pKd = 8.10. We discuss in this work only the protonated form, H3CA, and the monodeprotonated from, HzCA-, for reasons that we detail below. We report in Table 1 a summary of the electronic properties we have calculated for both the neutral H3CA and the negatively charged
Electronic Properties of Glycosylated Chromophores
J. Phys. Chem., Vol. 98, No. 49, 1994 12951
TABLE 1: Calculated Properties of Carminic Acid
carminic acid (neutral) carminic acid (- 1 charge)
-516.14 -551r.10
40000
40000
35000
s3
35000
19951 29858
22852 33 830
28355 30 140
29 176 34 118
32931 33 971
3.45 24.36 +0.03
1
0
0
1.92 15.94
1.56 8.87 a
0
s2
T, h
'7
6
0
30000
0
b
v
h
P w 25000
1
25000
L
I
T2
k-
20000
20000
4
o s0 :
Neutral Form (HsCA)
i
I
Figure 3. Calculated change in electron density, Agd, between the
0
Deprotonated Form (HzCA-)
Figure 2. Carminic acid electronic state ordering determined from AM 1 calculations using configuration interaction (CI).
H2CA- molecules. The dipole moments calculated for H3CA show a trend of decreasing polarity as the molecule is excited to higher electronic states. While it is typically held that the ground electronic states of polar organic molecules possess a smaller static dipole moment than the first excited electronic singlet states, it is entirely possible that the excited states of certain molecules will have a permanent dipole moment smaller than that of the ground state, owing to the change in charge distribution caused by excitation. The permanent dipole moments we calculate for H2CA- are, in our opinion, unreasonably large. We believe that these large calculated dipole moments result from limitations inherent to the AM1 parametrization and to the necessarily discrete way in which such calculations account for the presence of charges within the molecule. Despite these limitations, the qualitative trends predicted for other singly charged species have been proven correct by experimental data, and therefore we are confident that the data we report here for both H3CA and H2CA- are qualitatively useful. We show in Figure 2 energy level diagrams for H3CA and H2CA- determined from AM1 calculations using configuration interaction (CI). We note that the ordering of the states is quite different for the two species, with H3CA having two triplet states (AE(T1-So) = 19951 cm-l, AE(T2-So) = 22852 cm-'), between the ground and first excited singlet state (AE(S1-So) = 28 355 cm-I). The monodeprotonated species has a triplet state (AE(T1-So) = 29 858 cm-') in close energetic proximity to the first excited singlet state (AE(S1-So = 30 140 cm-l). State ordering for higher excited states also differs for the two forms, with the most dramatic difference being that H2CA- has triplet states slightly lower in energy, but always in close proximity to higher excited singlet states, suggesting the possibility of efficient coupling between the singlet and triplet
singlet excited state (SI) and the ground state (SO) for carminic acid (CA). Structure (a) is for carminic acid (H3CA), while (b) is for the monodeprotonated form (H2CA-). A positive sign indicates a decrease in electron density upon excitation from the SOto S1 electronic states. Values less than fO.O1 are not shown.
manifolds for this species. These calculations agree with observed spectroscopic properties of H3CA and HzCA-, as studied by Stapelfeldt et U Z . , ~ and ~ serve to provide a mechanistic explanation for their observations. Their experimental data showed that the fluorescence intensity of an aqueous solution of carminic acid at basic pH is less than the emission intensity of neutral or acidic solutions. As shown in Figure 2, there is a triplet state (TI) slightly lower in energy than the first excited singlet state (SI)for H2CA-, with no analogous state near the SI for H3CA. The lower fluorescence intensity observed for H2CA- suggests that depopulation of the S1 state occurs through efficient intersystem crossing to the TI state. Stapelfeldt et ~ 1 suggest that the main depopulation pathway for deprotonated forms involves an intersystem crossing to TI, followed by photooxidation and subsequent depopulation of triplet state H2CA-. Our calculations of electronic state ordering for H2CAshow that this singlet deactivation route is likely, although speculation on photooxidation subsequent to intersystem crossing is well outside the scope of this work. The larger energy separation between S1 and the triplet states TI and Tz for H3CA suggests that depopulation of the S1 state through intersystem crossing in the fully protonated species is less likely to be efficient. In understanding how a probe molecule will interact with its surroundings, it is important to consider how the electron distribution differs among the electronic states. We show in Figure 3 the calculated change in electron density (&d) between the SI and SOstates (heed = @&l) - gd(S0)) at each atom for H3CA (Figure 3a) and H2CA- (Figure 3b). For H3CA, the changes in electron density upon excitation occur within the anthraquinone chromophore while the glycoside group does not experience any signficant change in electron density. The primary implication of this result is that the molecular oribitals responsible for the optical response of H3CA are not associated with the glycoside moiety, and this finding is intuitively reasonable. The absorption maximum of H3CA is -550 nm in
.
~
~
12952 J. Phys. Chem., Vol. 98, No. 49, I994
Rasimas and Blanchard
::::1
a
Bo,
-300
5-5501 0 0 -200 Q
,
1
,
60
,
I
120
1
,
I
240
180
,
,
,
300
360
-250
b
-300 -350
n o
L
0
;cn
60
120
180
240
300
360
Twist Angle (degrees) Figure 4. Calculated energy dependence of the glycosyl group rotation for the SO (0),TI(0),and S I (0)states of carminic acid. Tracing (a) is for H3CA and (b) is for H;?CA-. The dihedral angle is the angle made by the glycosyl group with respect to the chromophore. Missing points indicate a failure of the calculation to converge at the given dihedral angle. Relative energies at a given dihedral angle were SI > TI > So.
polar organic solvents, indicating a highly conjugated chromophore. Because the glycoside group does not possess n orbitals that are conjugated with the n orbitals of the chromophore, we expect the glycoside group to produce steric, rather than electronic, perturbations to the optical response of this molecule. In an effort to understand any potential steric effects that the glycoside group may produce in the optical response of canninic acid, we have calculated the isomerization barriers for rotation of the glycoside moiety about its bond with the chromophore for the ground state, first triplet state, and first excited singlet state of both H3CA and HzCA-. These data are shown in Figure 4. The isomerization barriers are similar for the two forms of carminic acid, with the minimum energy of both species occumng at a glycoside-chromophore dihedral angle of 55". The calculated barrier height reaches a local maximum of -90 kcaYmol at 155" for H3CA and -68 kcaYmol for H2CA-. At 310", the barrier height is calculated to be -180 kcaYmol for H3CA and -127 kcaYmol for HzCA-. The heights of these barriers indicate that the glycoside moiety of carminic acid will be constrained to a relatively narrow range of angles and will not rotate freely. It is equally significant that the angular dependence of the isomerization barriers are virtually identical for all the electronic states we calculated. These calculations
indicate that the linear optical response of the chromophore will not depend on the dihedral angle it makes with the glycoside group, and therefore there is little or no information about this isomerization process available from electronic absorption or emission spectra. We note there are several points in the calculation of HzCA- isomerization barrier that failed to converge. This failure serves as another indication that the AM1 parametrization does not calculate charged species as effectively as neutral species. Additionally, the overall form of the isomerization calculated for H3CA is similar to the barrier calculated for HzCA-, indicating that the isomerization coordinate(s) of this motion are sterically, rather than electronically, mediated. Oxazines and Phenoxazines. We have calculated the electronic properties for oxazine, phenoxazine, 2-glycosyl- and 18-glycosyloxazine, and phenoxazine as well as the rotational isomerization barriers for the latter two compounds using the AM 1 parametrization. We included unsubstituted oxazine and phenoxazine in the calculation series so that effects of glycosylation on the electronic properties of the chromophore could be understood clearly. We have reported previously our semiempirical calculation results for several of these molecules using the MNDO parametrizati~n,~~ and the data we present here serve as a direct comparison of AM1 and MNDO parametrizations for complex polar organic molecules. In the previous work, heats of formation (Mf) of 187 and 203 kcal/ mol were calculated for oxazines and phenoxazines, respectively. These values are in qualitative agreement with the values presented in this work (208 and 233 kcal/mol, respectively), with the small differences being due to the different parametrizations used. We show in Table 2 a summary of the electronic properties we have calculated for the unsubstituted chromophores as well as the 2- and 18-glycosylated species. We calculated electronic properties for two isomers of glycosylated oxazines and phenoxazines because both of these compounds are being synthesized and will be used in future experimental studies of the crystallizationphenomena. These synthetic efforts may ultimately include several different species with slightly different substituents, and we have chosen to calculate the properties of the unsubstituted chromophores to maximize the utility of these results. Previous calculations show that small modifications of oxazines and phenoxazines have a minimal effect on their calculated properties. The dipole moments we have calculated for oxazine and phenoxazine show a trend of increasing polarity as the molecule is excited to higher electronic states, in agreement with previous experimental examinations of these molecule^.^^-^^ However, upon glycosylation, the dipole moments of the ground state are found to be, in general, larger than those calculated for excited states, a trend that we also observe in carminic acid. This prediction is important in the sense that it implies that the sensitivity of the glycosylated chromophore(s) to polar environments will be different than that of the native chromophore. We show in Figure 5 the results of AMl/CI calculations of energy levels for oxazine and glycosylated oxazine. Electronic
TABLE 2: Calculated Properties of Oxazines and Glvcosidvloxazines ~
~
~
_
_
_
_
_
_
_
energy relative to So (cm-I)
molecule oxazine 2-glycosyloxazine 18-glycosyloxazine phenoxazine
2-glycosylphenoxazine 18-glycosylphenoxazine
A& (kcaYmo1)
208.43 -6.91
-7.52 233.21 14.35 12.83
TI 12360 15 133 16 187 12780 16256 16715
T2
SI
20292
20292 23381 23520 19371 24235 24442
23349 23487 21 277 25223 25907
T3 25 269 26347
28849 24931 26087 28277
Sz
24212 25 153 25579 25 890 26307 26044
/@o)
(D) ,4Td (D) P(SI) (D) 2.52 3.78 3.64
10.89
19.50 3.25 10.77 17.51
10.30 19.16 4.03
9.46 15.34
13.92
14.75 6.00 4.10
11.95
Electronic Properties of Glycosylated Chromophores 30000
[
r,sS,
30000
I
[ I
J. Phys. Chem., Vol. 98, No. 49, 1994 12953 30000
S3,T,
30000 b
T
30000
,
S3'T.
I
T
s2
25000
5
20000
TZ'SI
20000
v
2.
e
c
20000
20000
Lu
15000
t
-t
lOO0OL
lOO0OL
-0 Oxazine
so
O
L so 2-glycosyl
o
t
c so
-0
15000
Figure 5. Oxazine and glycosyloxazine electronic state ordering determined from Ah41 calculations using configuration interaction (CI).
state ordering is similar for the three molecules. The most striking difference beween the oxazine and glycosylated oxazine is that the energy of the SO S1 transition in the native chromophore is lower than that of the glycosylates. This calculation, which is consistent with the calculated results for isomerization of the glycosylated species (vide infra), implies that the glycosyl moiety plays little or no electronic role in the SO S1 transition and that its dominant contribution to the optical response of the chromophore is sterically based. The blue shift is due to steric interactions of the glycosyl moiety with the neighboring amino group that serve to twist the amino group away from the orientation it prefers in the native chromophore. The close energetic proximity of the SI and T2 states in both the native and glycosylated oxazine suggests the possibility of efficient intersystem crossing. The energies of the T2 and SI states are calculated to be the same ( E = SO 20 292 cm-'), and in the glycosylated oxazines the calculated energy difference is negligible (AE(S1-TZ) = 32 cm-l). While this is obviously a negligible energy difference from an experimental standpoint (kT = 200 cm-' at 300 K), two sources of uncertainty in the calculations serve only to reinforce our assertion that these two states are functionally degenerate. First, these calculations rely on parametrizations derived from ground state experimental data, and thus the estimation of excited state parameters necessarily involves significant uncertainty. Second, where we are considering transition energies, it is important to realize that the excited state calculations are based on a molecule whose ground state geometry is optimized. There are many polar organic molecules that exhibit significant changes in geometry on excitation, as evidenced by their large static Stokes shifts,8J0*28and these calculations do not account for such changes. Substituted polar organic molecules are likely to exhibit different excited state geometries than that of the native chromophore, especially when the substitution gives rise to significant steric hindrance of a functionality that is part of the conjugated chromophore. These calculations, in effect, compare individual points on two different surfaces and cannot provide information on the locations of the minima of these two surfaces. We show in Figure 5 that the energy difference between the first excited state and the first triplet state for oxazine (AE(S1TI) = 7932 cm-') is different from 2-glycosyloxazine (AE(S1TI) = 8248 cm-') and 18-glycosyloxazine (AE(S1-TI) = 7333 cm-l). These energy differences are all large enough so that energy transfer from the S1 to the T1 state will be inefficient. Figure 6 displays energy level diagrams calculated for phenoxazine and glycoside modified phenoxazines. In all three species,
!
-0
lOO0OL
SO
2-glycosyl
o
c so 18-glycosyl
Figure 6. Phenoxazine and glycosylphenoxazine electronic state ordering determined from AM 1 calculations using configuration interaction (CI).
-
-
SZ'T2
20000
lOO0OL
so
Phenoxazine
18-glycosyl
r'
15000
r ooooL
1OOOOL
r
r IT t 30000
25000
T3
c
15000
r
a
-0.13 -0.24 -0.10 + o , 0 2 ~ N n + 0 . 0 4
+
b
I
0 C
+O 04
Figure 7. Calculated change in electron density, Aed, between the singlet excited state (SI)and the ground state (SO)for oxazine and glycosyloxazines. Structure (a) is for oxazine, structure (b) is for 2-glycosyloxazine, and structure (c) is for 18-glycosyloxazine. A positive sign indicates a decrease in electron density upon excitation from the SO to S1 electronic states. Values less than fO.O1 are not shown.
the state ordering is similar, with a triplet state between the ground and first excited state. Additionally, the S1 state for all species is well separated energetically from triplet states. The energy difference between the S1 and TI states for the glycosyloxazines are also similar (AE(S1-Tl) = 7979 cm-l for 2-glycosylphenoxazine, AE(S1-Tl) = 7727 cm-' for 18glycosylphenoxazine). We have calculated state-dependent electronic distributions for oxazine, phenoxazine, and the glycosylated species. Figure 7 shows the change in electron density upon excitation, where the changes are defined as the difference between the excited and ground electronic singlet states. A negative sign represents an increase in electron density upon excitation. Figure 7a represents the change in calculated electron density in oxazine.
Rasimas and Blanchard
12954 J. Phys. Chem., Vol. 98, No. 49, 1994 a -0 05 Cg
03/\*0
02
160
2oo \
+o 12
i
a
I b
+O 03
-o 03
no 06 I
a
r 200
160h 120
6 C
1047
I
.oo4
+O 13
1 -005
+O 03
Figure 8. Calculated change in electron density, Aed, between the singlet excited state (SI) and the ground state (SO)for phenoxazine and
glycosylphenoxazines. Structure (a) is for phenoxazine, structure (b) is for 2-glycosylphenoxazine, and structure (c) is for 18-glycosylphenoxazine. A positive sign indicates a decrease in electron density upon excitation from the SOto SI electronic states. Values less than fO.O1 are not shown. The heterocyclic nitrogen gains electron density, and its neighboring carbons lose electron density upon excitation. The other atoms do not show similarly large changes. We have previously determined that the nitrogen heteroatom in oxazine becomes negatively charged upon excitation, and this statedependent accumulation of electron density can be used as a probe of local solvent structure and We hope to use glycosylated oxazine as a probe of local polarity and hydrogen bond formation in the crystallization of sugars. Calculated excitation-dependent changes in electron density distribution for 2-glycosyloxazine and 18-glycosyloxazine are shown in Figure 7, b and c, respectively. The addition of the glycoside moiety to the oxazine chromophore affects the electron distribution similarly for both molecules. The heterocyclic nitrogen gains electron density upon excitation, as was observed in the "bare" oxazine. The amine nitrogen closest to the glycoside moiety loses significant electron density on excitation. We believe that the origin of this large change in electron density is due to the steric constraints placed on this amino group by the presence of the glycoside functionality. The excited state accumulation of charge at a (twisted) amino group is often referred to as a twisted internal charge transfer (TICT) SI The prediction of a TICT S1 state for the modified oxazines and phenoxazines is important because the existence of such a state could serve to alter the polarity dependence of the chromophore spectroscopic response substantially. The prediction of this TICT state is consistent with the data shown in Figures 5 and 6, where the SO SI transition energies of these molecules increase with modification. The electron distribution data for the phenoxazine and glycoside modified phenoxazines are presented in Figure 8. Figure 8a is unsubstituted phenoxazine, and we calculate that, like the oxazine, the nitrogen heteroatom accumulates electron density on excita-
-
Twist Angle (degrees) Figure 9. Calculated energy dependence of the glycosyl group rotation for the SO (0),TI (tso), and SI (0) states of glycosyl-substituted oxazines. Tracing (a) is for 2-glycosyloxazine, and (b) is for the 18glycosyl isomer. The dihedral angle is the angle made by the glycosyl group with respect to the oxazine chromophore. Missing points indicate a failure of the calculation to converge at the given dihedral angle. Relative energies at a given dihedral angle were S I > T1 > SO.
tion. The TICT state that was observed for the glycosyloxazines is also seen for the glycosylphenoxazines, consistent with the similarity of the unsubstituted chromophores. The glycoside moiety does not exhibit any change in electron distribution on excitation for either the modified oxazine or phenoxazine. Thus, steric rather than electronic factors will determine any change in optical response associated with the modification of these chromophores, just as we have calculated for carminic acid. As for carminic acid, we also calculated the isomerization barriers for rotation of the glycoside group about its bond with the oxazine chromophore. These calculations were performed for the modified oxazine. We expect the isomerization barriers observed for the oxazine will be similar to that for the modified phenoxazine. We report these calculations in Figure 9 for 2-glycosyloxazine (Figure 9a) and 18-glycosyloxazine (Figure 9b). These data are qualitatively similar, with minima occurring at a 50" glycoside-chromophore dihedral angle in both isomers. The calculated barrier height reaches a local maximum of -167 kcaYmol at 155" for 2-glycosyloxazine and -37 kcal/mol for 18-glycosyloxazine. At 325", the barrier height is calculated to be -121 kcaYmol for 2-glycosyloxazine and -136 kcal/mol for 18-glycosyloxazine. The heights of these barriers indicate that the glycoside moiety will be constrained to a relatively narrow range of angles and will not rotate freely. It is also significant that the state ordering is consistent throughout the range of dihedral angles calculated, indicating that the linear optical response of the chromophore will not depend on the dihedral angle it makes with the glycoside group. As previously discussed, it will not be possible to determine the position of the glycoside moiety through electronic absorption or emission experimental data. Oxazones and Phenoxazones. We have calculated the electronic properties for oxazone, 2-glycosyl- and 18-glycosyloxazone, phenoxazone, and 2-glycosyl- and 18-glycosylphenoxazone and the rotational isomerization barriers for the
J. Phys. Chem., Vol. 98, No. 49, 1994 12955
Electronic Properties of Glycosylated Chromophores TABLE 3: Calculated Properties of Oxazones and Glycosidyloxazones energy relative to So (cm-l) oxazone 2-glycosyloxazone 18-glycosyloxazone phenoxazone 2-glycosylphenoxazone 18-glycosylphenoxazone
t-
33.17 -196.49 -200.36 47.43 -182.95 -185.77
19441 19383 19804 20662 20802 21 129
T,
Oxazone
29030 28810 28710 31 143 30981 30818
31 661 31454 31 697 33 203 32637 32717
4.04 6.57 4.10 3.41 3.81 4.52
5.69 7.89 5.99 4.26 4.68 5.24
4.40 5.46 5.26 4.09 4.30 4.97
s2
p t
27 479 27410 27896 27 873 27778 28212
35000
25000
20000
27 177 27653 27836 29291 29 167 28732
20000
1
T,
I
30000
30000
30000
20000
zOOO:c o c s ,
1
20000
15000
Phenoxazon8
2-glycosyl
Figure 10. Oxazone and glycosyloxazone electronic state ordering determined from AM1 calculations using configuration interaction (CI). glycosylated species. As for the oxazines and phenoxazines, oxazone and phenoxazone were calculated as a reference point to determine the effect of the addition of the glycoside moiety to the chromophore backbone. We show in Table 3 the calculated properties for unsubstituted oxazone and phenoxazone and the glycosylated chromophores. The calculated dipole moments for these molecules follow a trend of increasing polarity on excitation, except for 2-glycosyloxazone, where the S1 state is calculated to have a smaller dipole moment than the ground state. The calculated dipole moment of the triplet state in these molecules has the largest value in the oxazone calculations, but examination of the charge distribution of the T1 state does not reveal a substantial charge accumulation for any of the heteroatom sites. Glycoside substitution at both the 2- and 18-positions causes an increase in the magnitude of the dipole moment compared to the unsubstituted chromophores. The AMl/CI calculated electronic state ordering of oxazone, 2-glycosyloxazone,and 18-glycosyloxazoneare shown in Figure 10. In contrast to the oxazines, the increase in the calculated SO S1 transition energy on substitution for the oxazone chromophore is comparatively small, despite the sterically mediated formation of a twisted terminal amino group in the glycosylated species. The increase in transition energy on substitution indicates that the rotated amino group does in fact play a small role in the transition, but because of the inherent asymmetry in the chromophore, the amino group plays a comparatively smaller role in determining the energy of the S1 state than for the oxazine species.30 Oxazone and substituted oxazones have similar state ordering, with the triplet state TI being found between the ground (SO)and excited singlet state (SI).For all species, the energy separation between the S2 and S1 states is large (AE(S2-Sl) = 4181 cm-' for oxazone, 4044 cm-' for 2-glycosyloxazone, 3801 cm-' for 18-glycosyloxazone), so that the dominant optical transition will be SO SI. The energy separation between the S1 and T1 states is large (AE(Sl-T1) = 8000 f 100 cm-l for all species), so that
-
-
k
T
1
2-glyC0sy?0
1"
20000
18-gly~0~yl
Figure 11. Phenoxazone and glycosylphenoxazone electronic state ordering determined from AM1 calculations using configuration interaction (CI). intersystem crossing from the S1 to T1 state not expected, but, as for the oxazines, the T2 is in close energetic proximity to SI. In oxazone, hE(s1-T~) = 302 cm-l, in 2-glycosyloxazone the AE(S1-TZ) = -243 cm-', and in 18-glycosyloxazone the AE(S1-TZ) = 60 cm-l. In close analogy to the oxazines, we take these values to indicate that for all species the S1 and T2 electronic states are essentially degenerate. The phenoxazone and glycosylated phenoxazones exhibit state ordering similar to that for the oxazones, but the energy separations between states is greater than for the oxazone series (Figure 11). The phenoxazone species have a T1 state between SO and SI, with AE(T1-So) = 20 662, 20 802, and 21 129 cm-' for phenoxazine, 2-glycosylphenoxazone, and 18-glycosylphenoxazone, respectively. The energy differences between SI and SOare also similar, with AE(S1-So) = 27 872, and 27 778, and 28 212 cm-l respectively. For the phenoxazones, the T2 state is higher in energy than the S1 state and is also separated by a larger energy than is seen for the oxazones. These calculations suggest that glycoside modification of phengxazone does not dramatically affect the electronic state ordering and that the optical response of all species will be largely similar, irrespective of the steric constraints imposed by the glycosyl moiety. This prediction is in significant contrast to that for the oxazines. We have calculated the state-dependent electron density distributions for oxazone, phenoxazone, and the glycosylated species of these chromophores. Figures 12 and 13 show the change in electron distribution for the oxazone series and phenoxazone series, respectively. It is important to note that these molecules do not undergo state-dependent charging on the atoms in the chromophore or in the glycoside moiety. The oxazone chromophore (Figure 12a) does not exhibit a significant spatial modulation of electron density on excitation, and the glycosyl modified chromophore (Figure 12b,c) behaves similarly. The phenoxazone calculations (Figure 13) predict behavior consonant with that of the oxazones. As with carminic
12956 J. Phys. Chem., Vol. 98, No. 49, 1994
Rasimas and Blanchard a 6o
N'
I
+007
a
rt
4
+020
b -0 07
pQp4yyo4
44; -A I& 0
0 +o 18
+002
b O
T0
o C
0
Figure 12. Calculated change in electron density, heed,between the singlet excited state (SI)and the ground state (SO)for oxazone and glycosyloxazones. Structure (a) is for oxazone, structure (b) is for 2-glycosyloxazone, and structure (c) is for 18-glycosyloxazone. A positive sign indicates a decrease in electron distribution upon excitation from the SO to SIelectronic states. Values less than fO.O1 are not shown. a
b -0.08
O
T0
-0.08
fi
0 C
60
120
180
240
300
360
Twist Angle (degrees) Figure 14. Calculated energy dependence of the glycosyl group and SI(0)states of glycosyl substituted rotation for the SO(0).TI (O), oxazones. Tracing (a) is for 2-glycosyloxazone, and (b) is for the 18glycosyl isomer. The dihedral angle is the angle by the glycosyl group with respect to the oxazone chromophore. Missing points indicate a failure of the calculation to converge at the given dihedral angle. Relative energies at a given dihedral angle were SI> TI> SO. The isomerization barrier of the glycoside moiety about its bond with the oxazone chromophore is qualitatively similar to that of the oxazines. Figure 14 shows the calculated isomerization barriers for 2-glycosyloxazone (Figure 14a) and 18glycosyloxazone (Figure 14b). For these molecules, the energetic minima occur at a -50" glycoside-chromophore dihedral angle. The barrier height reaches a local maximum of -184 kcal/mol at 155"for 2-glycosyloxazone and -40 kcdmol for 18-glycosyloxazone. At a dihedral angle of 325", the barrier height is calculated to be -125 kcdmol for 2-glycosyloxazone and -134 kcdmol for 18-glycosyloxazone. The magnitudes of these baniers indicate that the glycoside moiety is constrained to a narrow range and will not rotate freely, as was observed in the oxazine calculations. As seen in the oxazine calculations, the state ordering is constant throughout the range of dihedral angles calculated, with TI being between SOand SI in both oxazone isomers. Again, these data indicate that the linear optical response of the chromophore will not depend on the rotational position of the glycoside group.
Conclusions Figure 13. Calculated change in electron density, A@&,between the singlet excited state (SI)and the ground state (SO)for phenoxazone and glycosylphenoxazones. Structure (a) is for phenoxazone, structure (b) is for 2-glycosylphenoxazone, and structure (c) is for 18-glycosylphenoxazone. A positive sign indicates a decrease in electron density upon excitation from the SOto SI electronic states. Values less than f O . O 1 are not shown. acid and the oxazines, the glycoside moiety does not undergo any change in electron distribution upon excitation, and thus the role of this substituent in determining the optical response of these molecules is steric rather than electronic.
We have calculated the electronic properties of several unsubstituted and corresponding glycosylated chromophores. The calculated results for all molecules predict that the main spectroscopic features that will be observed arise from the SO S1 transition and that other transitions originating from SOto higher excited singlet states will not be overlapped with the SO S1 transition. The linear optical response of carminic acid is pH dependentz9 due to reordering of electronic states associated with the extent to which the molecule is protonated. Our calculations predict that glycosylated oxazines will exhibit state-dependent dynamical behavior arising from an excitationinduced accumulation of electron density at the heterocyclic nitrogen. The formation of a TICT S I state in the glycosylated
-
.-,
Electronic Properties of Glycosylated Chromophores oxazines is a direct result of steric interference between the terminal amino group and the glycoside moiety. Our calculations predict that the optical and dynamical response of glycosylated oxazones will not depend on state-dependent changes in electron density distributions in the same way as do the glycosylated oxazines. The presence of the glycoside moiety on these chromophores will affect their linear response through steric rather than electronic factors. An important consequence of this calculated prediction is that the optical responses of these chromophores will not depend on the dihedral angle between the chromophore and the glycoside substitutent.
Acknowledgment. We are grateful to the National Science Foundation for support of this research through Grant CTS-9407563 and to Professor G. L. Baker for use of additional computational resources. References and Notes (1) McMorrow, D.; Lotshaw, W. T. Chem. Phys. Lett. 1993,201,369. (2) McMorrow, D.; Lotshaw, W. T. J . Phys. Chem. 1991, 95, 10395. (3) Moms, D. L., Jr.; Gustafson, T. L. J. Phys. Chem. 1994,98,6725. (4) Butler, R. M.; Lynn, M. A.; Gustafson, T. L. J . Phys. Chem. 1993, 97, 2609. (5) Weaver, W. L.; Huston, L. A.; Iwata, K.; Gustafson, T. L. J . Phys. Chem. 1992, 96, 8956. (6) Hambir, S. A.; Jiang, Y.; Blanchard, G. J. J . Chem. Phys. 1993, 98, 6075. (7) Jiang, Y.;Blanchard, G. J. J. Phys. Chem. 1994, 98, 6436. (8) Jiang, Y.; McCarthy, P. K.; Blanchard, G. J. Chem. Phys. 1994, 183, 249. (9) Blanchard, G. J. J . Chem. Phys. 1991, 95, 6317. (10) Awad, M. M.; McCarthy, P. K.; Blanchard, G. J. J . Phys. Chem. 1994, 98, 1454.
J. Phys. Chem., Vol. 98, No. 49, 1994 12957 (11) Blanchard, G. J.; Cihal, C. A. J. Phys. Chem. 1988, 92, 5950. (12) Blanchard, G. J. J. Phys. Chem. 1988, 92, 6303. (13) Blanchard, G. J. J . Phys. Chem. 1989, 93, 4315. (14) Blanchard, G. J. J. Phys. Chem. 1991, 95, 5293. (15) Blanchard, G. J. Anal. Chem. 1989, 61, 2394. (16) Pan, B.; Chakraborty, R.; Berglund, K. A. J . Crysr. Growrh 1993, 130, 587. (17) Chakraborty, R.; Berglund, K. A. J . Crysr. Growth 1992, 125, 81. (18) Thompson, S. G.; Burd, J. F. Antimicro. Agents Chemother. 1980, 18, 264. (19) Bogulaski, R. C.; Burd, J. F.; Denning, C. E. J. Clin. Sci. 1980, 1, 215. (20) Li, T. M.; Benovic, J. L.; Burd, J. F. Anal. Biochem. 1981, 118, 102. (21) Burd, J. F. Methods Enzymol. 1981, 74C, 79. (22) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am. Chem. SOC. 1985, 107, 3902. (23) Dewar, M. J. S.; Dieter, K. M. J . Am. Chem. SOC. 1986,108,8075. (24) Stewart, J. J. P. J . Cornput.-Aided Mol. Des. 1990, 4, 1. (25) Dewar, M. J. S.; Thiel, W. J . Am. Chem. SOC. 1977, 99, 4899. (26) Dewar, M. J. S.; Thiel, W. J. Am. Chem. SOC. 1977, 99, 4907. (27) Allinger, N. L. J . Am. Chem. SOC. 1977, 99, 8127. (28) McCarthy, P. K.; Blanchard, G. J. J . Phys. Chem. 1993,97, 12205. (29) Stapelfeldt, H.; Jun, H.; Skibsted, L. H. Food Chem. 1993, 48, 1. (30) Blanchard, G. J. Chem. Phys. 1989, 138, 365. (31) Majumdar, D.; Sen, R.; Bhattacharyya, K.; Bhattacharyya, S. K. J . Phys. Chem. 1991, 95, 4324. (32) Wang, Y.; McAuliffe, M.; Novak, F.; Eisenthal, K. B. J . Phys. Chem. 1981, 85, 3736. (33) Hicks, J. M.; Vandersall, M. T.; Sitzmann, E. V.; Eisenthal, K. B. Chem. Phys. Lett. 1987, 135, 413. (34) Wang, Y.; Eisenthal, K. B. J . Chem. Phys. 1982, 77, 6076. (35) Sen, R.; Majumdar, D.; Bhattacharyya, S. P. Bhattacharyya, S . N. Chem. Phys. Lett. 1991, 181, 288. Nayuki, T.; Kobayashi, T. Chem. Phys. Lett. 1993, (36) Kajimoto, 0.; 209, 357.
