Conformational studies on 1,2-Di- and 1,2,3-trisubstituted heterocycles. A spectroscopic and theoretical study of 3-acylaminopicolinic acid derivatives and their N-oxides

Article abstract of DOI:10.1021/jo000993j

Theoretical studies involving minimization of model 3-propanoylaminopicolinic acids (10d-trans, 10d-cis), methyl ester (10a), and corresponding -N-oxide derivatives (10b, 10c-trans, 10c-cis) using AM1 gave conformations contrary to both sound chemical intuition and experimental data. RHF ab initio calculations using the 6-31G and 6-31G** basis sets on the other hand corroborated spectroscopic data. 3-Amidopicolinic acid derivatives (7a-9a, 7b-9b, 7c-9c, 9d) were prepared and studied by NMR and IR spectroscopy. The results show that a strong intramolecular hydrogen bond between amide-H and the 2-carboxyl substituent results in a planar molecular conformation. This is particularly profound in the 3-acylaminopicolinic acid N-oxides (c-series). When the 2-substituent is a methyl ester on the other hand, repulsion between N-oxide and ester functions induces twisting of the carbomethoxy group out of the plane of the aromatic ring. The type of method used in molecular modeling can have profound impact on the final theoretical result in the case of the above-mentioned class of compounds. Our results indicate, that it is advisable to employ ab initio methods for modeling these types of compounds, and further, that the choice of basis set used for such calculations should depend on the type of information required. Thus, for most purposes pertaining to molecular conformation the 6-31G basis set provides sufficiently sound data in relatively short CPU time. For data related to electronic properties such as involvement of the N-oxide function or spectroscopic information such as IR frequencies or ¹H or ¹⁵N NMR chemical shifts, the use of polarization functions as contained in the 6-31G** basis set seems to be a must.

Full text of DOI:10.1021/jo000993j

370
J . Org. Chem. 2002, 67, 370-382
Con for m a tion a l Stu d ies on 1,2-Di- a n d 1,2,3-Tr isu bstitu ted
Heter ocycles. A Sp ectr oscop ic a n d Th eor etica l Stu d y of
3-Acyla m in op icolin ic Acid Der iva tives a n d Th eir N-Oxid es
Ivoneide de C. L. Barros,† Claudia C. C. Bejan,‡ J oa˜o Bosco P. da Silva,*,‡
F. W. J oachim Demnitz,*,‡ Fernando Hallwass,‡ and Hans-Ulrich Gremlich§
Departamento de Quı´mica Fundamental, Universidade Federal de Pernambuco, Cidade Universita´ria,
50.670-901 Recife-PE, Brazil, Departamento de Qu´ımica, Universidade Federal do Amazonas,
Manaus-AM, Brazil, and Novartis Pharma AG, WSJ -503.1001, Lichtstrasse, CH-4002 Basel, Switzerland
jdemnitz@npd.ufpe.br
Received J une 30, 2000
Theoretical studies involving minimization of model 3-propanoylaminopicolinic acids (10d -tr a n s,
10d -cis), methyl ester (10a ), and corresponding -N-oxide derivatives (10b, 10c-tr a n s, 10c-cis) using
AM1 gave conformations contrary to both sound chemical intuition and experimental data. RHF
ab initio calculations using the 6-31G and 6-31G** basis sets on the other hand corroborated
spectroscopic data. 3-Amidopicolinic acid derivatives (7a -9a , 7b-9b, 7c-9c, 9d ) were prepared
and studied by NMR and IR spectroscopy. The results show that a strong intramolecular hydrogen
bond between amide-H and the 2-carboxyl substituent results in a planar molecular conformation.
This is particularly profound in the 3-acylaminopicolinic acid N-oxides (c-series). When the
2-substituent is a methyl ester on the other hand, repulsion between N-oxide and ester functions
induces twisting of the carbomethoxy group out of the plane of the aromatic ring. The type of method
used in molecular modeling can have profound impact on the final theoretical result in the case of
the above-mentioned class of compounds. Our results indicate, that it is advisable to employ ab
initio methods for modeling these types of compounds, and further, that the choice of basis set
used for such calculations should depend on the type of information required. Thus, for most
purposes pertaining to molecular conformation the 6-31G basis set provides sufficiently sound data
in relatively short CPU time. For data related to electronic properties such as involvement of the
1
N-oxide function or spectroscopic information such as IR frequencies or H or 15N NMR chemical
shifts, the use of polarization functions as contained in the 6-31G** basis set seems to be a must.
In tr od u ction
observed for H6 in acetanilides carrying hydrogen bond
acceptors in the 2-position have been attributed to
conformation 1.2c,3 In this conformation, a hydrogen bond
between amide-H and the C2 substituent forces the
amide carbonyl into a position able to deshield H6, and
the molecule assumes a planar conformation. In fact, the
magnitude of the downfield “acetylation shift” of H6 in
2-substituted acetanilides (1) after acetylation of the
respective aniline precursor has been deemed sufficiently
reliable so as to be used as a diagnostic tool to estimate
the extent of coplanarity in such molecules.2c,d
The conformation of ortho-disubstituted aromatics is
greatly influenced by the mutual interaction between the
substituents in question.1,2h In particular, hydrogen
bonding may induce the molecule to assume a conforma-
tion different from that which purely steric considerations
would otherwise dictate. Secondary aromatic amides with
hydrogen bond acceptors disposed in a 1,2-fashion have
been extensively studied, especially by NMR2 and IR.4
For example, the large downfield “acetylation shifts”
* To whom correspondence should be addressed. Fax: +55 81 3271
8442. E-mail: (J .B.P.d.S.) paraiso@npd.ufpe.br.
† Universidade Federal do Amazonas.
‡ Universidade Federal de Pernambuco.
§ Novartis Pharma AG.
(1) March, J . Advanced Organic Chemistry, 3rd ed.; J ohn Wiley &
Sons: New York, 1985; p 139.
(2) (a) Brown, R. F. C.; Rae, I. D.; Sternhell, S. Aust. J . Chem. 1965,
18, 1211. (b) Brown, R. F. C.; Radom, L.; Rae, I. D.; Sternhell, S. Can.
J . Chem. 1968, 46, 2577. (c) Rae, I. D. Can. J . Chem. 1968, 46, 2589.
(d) Appleton, J . M.; Andrews, B. D.; Rae, I. D.; Reichert, B. E. Aust. J .
Chem. 1970, 23, 1667. (e) Andrews, B. D.; Poynton, J . A.; Rae, I. D.
Aust. J . Chem. 1972, 25, 639. (f) Rae, I. D. Aust. J . Chem. 1974, 27,
2617. (g) Rae, I. D. Aust. J . Chem. 1974, 27, 2621. (h) Kolehmainen,
E.; Laihia, K.; Kauppinen, R.; Rasala, D.; Puchala, A. J . Phys. Org.
Chem. 1995, 8, 577.
(3) (a) Russell, R. A.; Thompson, H. W. Spectrochim. Acta 1956, 8,
138. (b) Brown, C. J .; Corbridge, D. E. C. Acta Crystallogr. 1954, 7,
711. (c) Aroney, M. J .; LeFe`vre, R. W.; Singh, A. N. J . Chem. Soc. B
1966, 1183.
We have recently become involved in an endeavor to
synthesize N-oxide derivatives of 3-aminopicolinic acid
(2) as new bidentate ligands for lanthanide ions. Such
ligands coordinating through the N-O and carboxyl
functions act as antennas5 in light conversion devices.5,6,7
The amino group serves as a point of attachment for the
introduction of appropriate substituents with a view to
modifying the physicochemical and/or electronic charac-
teristics of the respective lanthanide complexes. With
(4) Perje´ssy, A.; Rosata, D.; Loos, D.; Piorun, D. Monatsh. Chem.
1997, 128, 541.
10.1021/jo000993j CCC: $22.00 © 2002 American Chemical Society
Published on Web 12/29/2001

Studies on 1,2-Di- and 1,2,3-Trisubstituted Heterocycles
J . Org. Chem., Vol. 67, No. 2, 2002 371
regard to the former, long-chain hydrocarbon substitu-
ents would improve the solubility8 of such species, while
in the case of the latter property, one may envisage
Schiff-base or amide derivatives with the proviso, that
these be coplanar with the pyridine ring in order to be
able to manipulate excited state (π*) energy levels in the
ligands by variation of the substituent on nitrogen. The
fulfillment of this condition can be guaranteed by the
presence of an NH‚‚‚o-COOR hydrogen bond constraining
the amide in the plane of the aromatic ring (3), since
there is no reason to believe that 3-N-acylaminopicolinic
acids and esters will succumb to different conformational
dictates as the abovementioned ortho-substituted aceta-
nilides (1). Furthermore, the pyridine nitrogen introduces
an additional bias not present in the acetanilides, inas-
much as the CdO dipole lies preferably anti to the
pyridine nitrogen (as shown in 2 and 3), thereby well
disposed to hydrogen bond the neighboring amide pro-
ton.9
Thus, conformational predictions for 3-aminopicolinic
acid derivatives were expected to be straightforward. We
anticipated however, that the introduction of an N-oxide
moiety producing a 1,2,3-trisubstituted aromatic system
would result in additional interactions, which might
perturb molecular geometry. With a methyl ester deriva-
tive in the preferred s-cis conformation, mutual repulsion
of the alkyl-oxygen and pyridine-N-oxygen would oppose
the stabilizing influence of the intramolecular hydrogen
bond and ought to induce rotation about the C2-COOMe
bond in order to relieve this steric interaction, (4), with
concomitant departure of the 2-carbomethoxy-3-acy-
lamino system from planarity giving a “twisted” confor-
mation (4t or 4tt). In a 2-carboxy unit on the other hand
an s-trans configuration would permit an intramolecular
H-bond between the acid proton and N-oxide,10 a situa-
tion once again conducive to an overall planar shape for
the molecule (5). Bidentate complexation of such a
molecule through N+-O- and COO- to a lanthanide
metal cation (Ln3+) would similarly be expected to result
in an overall planar ligand geometry (5′).
We have examined a series of substrates including both
the parent 3-amino- and some 3-acylaminopicolinic esters
and acids as well as the respective N-oxides and eu-
ropium(III) complexes thereof by NMR and IR spectros-
copy in order to gain insight into their conformational
behavior. Furthermore, to underpin our experimental
analysis, we have compared our interpretations with
predictions gained from molecular modeling studies.
Sp ectr oscop ic Resu lts
To be able to synthesize lipophilic luminescent lan-
thanide complexes, we prepared a series of aryl- and long-
chain alkyl amide derivatives of picolinic acid-N-oxide to
serve as appropriate ligands. The ligands (N-oxide acids
(c-series)), intermediates (esters (a -series), N-oxides (b-
series)) and complexes under study here were synthesized
in standard fashion from methyl 3-aminopicolinate (6a ),7a
(Scheme 1). Surprisingly, of the various methods tried
for N-oxidation (CH3CO3H, CF3CO3H, MMPP, m-CPBA)
only the latter provided the N-oxides.7b The tris-Eu(III)
complexes (Eu (7c)3) and (Eu (9c)3) of the respective
N-oxide acids were prepared in refluxing EtOH in the
presence of 0.33 equiv of Eu(ClO4)3 and 1 equiv of base7b,c
(5) Lehn, J .-M. Angew. Chem., Int. Ed. Engl. 1990, 29, 1304.
Williams, A. F.; Floriani, C.; Merbach, A. E. In Perspectives in
Coordination Chemistry; Verlag Helvetica Chimica Acta: Basel, New
York, 1992. Alpha, B.; Lehn, J .-M.; Mathis, G. Angew. Chem., Int. Ed.
Enlg. 1987, 26, 266. Alpha, B.; Ballardini, R.; Balzani, V.; Lehn, J .-
M.; Perathoner, S.; Sabbatini, N. Photochem. Photobiol. 1990, 52, 299.
Prodi, L.; Maestri, M.; Ziessel, R.; Balzani, V. Inorg. Chem. 1991, 30,
3798. Alpha, B.; Balzani, V.; Lehn, J .-M.; Perathoner, S.; Sabbatini,
N. Angew. Chem., Int. Ed. Engl. 1987, 26, 1266.
Sch em e 1
(6) Bu¨nzli, J .-C. In Lanthanide Probes in Life, Chemical and Earth
Sciences; Bu¨nzli, J .-C., Choppin, G. R., Eds.; Elsevier: Amsterdam,
1989; Chapter 7. Horrocks, W. d. W., J r.; Albin, M. Prog. Inorg. Chem.
1984, 31, 1. Richardson, F. S. Chem. Rev. 1982, 82, 541. Horrocks, W.
d. W., J r.; Sudnick, D. R. Acc. Chem Res. 1982, 14, 384. Bu¨nzli, J .-C.;
Wessner, D. Coord. Chem. Rev. 1984, 60, 191. Balzani, V.; Sabbatini,
N.; Scandola, F. Chem. Rev. 1986, 86, 319. Lehn, J .-M. In Supramo-
lecular Photochemistry; Balzani, V., Ed.; Reidel: Dordrecht, The
Netherlands, 1987; p 29. Lehn, J .-M. Acc. Chem. Res. 1978, 11, 49.
Mukkala, V.-M.; Sund, Ch.; Kwiatkowski, M.; Pasanen, P.; Hogberg,
M.; Kankare, J .; Takalo, H. Helv. Chim. Acta 1992, 75, 1621. Mukkala,
V.-M.; Kankare, J . Helv. Chim. Acta 1992, 75, 1578. Mukkala, V.-M.;
Kwiatkowski, M.; Kankare, J .; Takalo, H. Helv. Chim. Acta 1993, 76,
893.
(7) (a) Mesquita, M. E.; de Sa´, G. F.; Demnitz, F. W. J . J . Alloys
Compd. 1998, 275-277, 844. (b) Bejan, C. C. C. Master’s Thesis,
Universidade Federal de Pernambuco, April, 2000. (c) Barros, I. d. C.
L. Master’s Thesis, Universidade Federal de Pernambuco, September,
2000.
(8) The parent complex tris(3-aminopyridine-2-carboxylato)-
diaquaeuropium(III)7a is difficult to purify and handle and, therefore,
has limited utility due to its extremely low solubility in all common
solvents.
1
The H NMR chemical shifts of 3-aminopicolinic acid
derivatives are depicted in Table 1. The very large down-
field shifts of 2.12-2.32 ppm observed for H4 upon acyl-
ation of 6a to 7a -9a (column 4) are clearly a result of
the neighboring amide CdO anisotropy and corroborate
earlier results2,3 obtained with ortho-substituted aceta-
nilides. The lesser effects upon protons H5 and H6 when
going from 6a to 7a -9a (columns 5 and 6) are almost
certainly through-bond effects.2b The large chemical shifts
of the amide NH protons (column 7, cf. o-carbomethoxy-
(9) Gandolfo, C.; Favini, G. Gazz. Chim. Ital. 1970, 100, 534.
Gandolfo, C.; Bueni, G.; Favini, G. Gazz. Chim. Ital. 1972, 102, 946.
(10) Dega-Szafran, Z.; Grundwald-Wyspianska, M.; Szafran, M. J .
Mol. Struct. 1992, 275, 159.

372 J . Org. Chem., Vol. 67, No. 2, 2002
Barros et al.
Ta ble 1. Exp er im en ta l 1H a n d Am id e 15N NMR Ch em ica l Sh ifts of 3-Am in op icolin ic Acid Der iva tives (Ser ies 6-9) in
CDCl3 a n d Cor r esp on d in g Ca lcu la ted Ch em ica l Sh ifts for Mod el 3-P r op a n oyla m in op icolin ic Acid Der iva tives (10 Ser ies)
Usin g a b In itio 6-31G a n d 6-31G** Levels of Th eor yf
a
b
d
Reference 7a. Spectrum in DMSO-d6. c Reference 2h. Reference 2f. e δ(CH3NO2) ) 0. f For 15N chemical shifts, the percent deviation
of the calculated from the experimental value for the corresponding alkylamide nitrogen (9 series) is given in parentheses (except for
10a g). Deviation from 7a .
g
acetanilide2f,h) imply that they are involved in hydrogen
bonding. To establish additional evidence for the in-
tramolecular nature of this hydrogen bond to the ester
carbonyl, the NMR spectra of the amides (7a -9a , 7c-
9c, and 9d ) were determined in high dilution. As ex-
pected, the positions of the amide-H resonances suffered
no alteration. The particularly large downfield reso-
nances of the carboxylic acid protons in 7c and 9c
(column 8) are an indication that, as anticipated, these
protons are hydrogen bonded by the N-oxide function.
concomitant change in the CdO anisotropy (column 4).
Taking pyridine as a model, it is established, that
N-oxidation results in an upfield shift of 0.29 and 0.23
ppm for the H4 and H2/6 protons respectively, while the
H3/5 protons in the meta positions remain relatively
unaffected.11 A comparison with methyl 3-dodecanoy-
laminopicolinate (9a ) and its N-oxide (9b) is instructive;
(Figure 1).
The H5 (-0.24 ppm vs +0.08 ppm) and H6 protons
(-0.49 vs. -0.23 ppm) of 9a suffer only moderately
greater upfield shifts upon N-oxidation than is the case
in pyridine. However, H4 is considerably more affected
after N-oxidation with an upfield shift of 0.79 ppm as
compared to only 0.29 ppm in pyridine. Assuming that
N-oxidation contributes approximately 0.30 ppm to the
upfield shift of the H4 proton we conclude, that the
additional effect of 0.49 ppm upon H4 comes from reduced
deshielding of this proton by the amide carbonyl in the
It can be argued, that oxidation of the pyridines (6a -
9a ) will result in mutual repulsion of the N-oxygen and
ester alkyl-oxygen atoms in the N-oxide products (6b-
9b). The ensuing rotation of the carbomethoxy group out
of the plane (4 f 4t) would weaken the hydrogen bond
to the amide-NH. The large upfield shifts (1.96-2.16
ppm) of the amide protons in N-oxides (7b-9b) as
compared to (7a -9a ) support this rationale. Further-
more, it is interesting to note, that in all cases the H4
aromatic protons also suffer an appreciable upfield shift
upon passing from the a - to the b -series implying a
(11) Hesse, M.; Meier, H.; Zeeh, B. Spektroskopische Methoden in
der organischen Chemie; Georg Thieme Verlag: Stuttgart, 5. Aufl.,
1995; p 191.

Studies on 1,2-Di- and 1,2,3-Trisubstituted Heterocycles
J . Org. Chem., Vol. 67, No. 2, 2002 373
similar to 4tt. The presence of the N-oxide, however,
provides a stabilizing hydrogen bond acceptor for the
acidic proton. Consequently, a trans conformation such
as depicted in 5 is possible and should “restore” the
planarity lost in the ester (b) series. An inspection of the
chemical shifts of both amide and aromatic H4 protons
(Table 1) in acids 7c-9c reveals that the deshielding
“lost” in going from a pyridine (a -series) to an N-oxide
(b-series) is, more than, restored by ester hydrolysis (c-
series). The amide protons now resonate strongly down-
field (> 12 ppm) and the chemical shifts of the H4 protons
are very close to what they were in the pyridines (7a -
9a ). Possibly the reason the chemical shifts for H4 in
N-oxide-acids (7c-9c) are consistently larger by about
0.14 ppm than in the corresponding pyridines (7a -9a )
lies in the fact that the effect due to deshielding by the
neighboring amide carbonyl is slightly potentiated in the
c-series by a reduction of the N-oxide +M effect in view
of this oxygen being compromised in a hydrogen bond.
The IR frequencies of the amide N-H, N-oxide as well
as amide-I and carbonyl CdO stretching modes for
benzoyl (7a -c) and dodecanoyl (9a -c) derivatives as well
F igu r e 1. Comparison of 1H NMR chemical shifts of pyridine
and methyl 3-dodecanoylaminopicolinate (9a ) as well as their
respective N-oxides. Relative upfield (-) and downfield (+)
shifts in N-oxides are given in parentheses.
N-oxide derivative (9b) as compared to 9a . Indeed, it has
been estimated, that anisotropic deshielding of ortho
protons by a coplanar amide group in simple acetanilides
does in fact contribute about +0.5 ppm to their chemical
shift.2b One may rationalize, that the effect of N-O T
MeOOC repulsion is “translated” around the periphery
of the molecule, resulting in twisting of both C2-CO2-
Me and C3-amide bonds. Such a rotation removes the
amide carbonyl deshielding cone from proximity to H4
and diminishes its through-space effect on this proton
(see 4t t and structure 10b in Figure 2). On the other
hand, H4 in the free amine (6a ), devoid of the anisotropic
CdO group, is, upon N-oxidation, deshielded by only 0.40
ppm a value only half that of the acylated series and close
to that for pyridine itself. Our modeling studies support
this hypothesis (vide infra).
as for the complexes (Eu (7c)
3) and (Eu (9c)3) are shown
in Table 2 together with the respective theoretical values
obtained from ab initio calculations using the 6-31G**
basis set. For comparison, pertinent stretching frequen-
cies for published spectra of methyl 2-acetamidoben-
zoate,4 N-phenylacetamide4 and methyl benzoate are
included. The results corroborate our conclusions based
upon NMR investigations. The positions of the N-H
bands clearly imply H-bonding and in all cases the
intramolecular nature of this amide-to-ester carbonyl
hydrogen bond (OdCN-H‚‚‚‚‚‚OdC(OR)) was substanti-
ated by solution-IR. The N-H stretching frequencies in
dichloromethane solution were unaltered by dilution, and
compare very well to the respective value in methyl
2-acetamidobenzoate4 as do the amide-I and ester car-
bonyl stretches in 7a and 9a .
Our proposal, that the N-oxide esters have a “twisted-
COOMe” conformation (4tt) implies a weaker H-bond in
the b-series. Insights into the relative strengths of the
hydrogen bond were gained from the respective peak
frequencies. Thus in both the benzoyl- (7) and dode-
canoyl- (9) amides a weakening of the H-bond in going
from the a - to the b-series was observed in dichlo-
romethane solution, (7a : νmax(N-H) 3311 f 7b: νmax(N-H)
3398 cm-1 and 9a : νmax(N-H) 3318 f 9b: νmax(N-H) 3401
cm-1) as would be expected as a result of rotation about
the pyridine(C2)-carbomethoxyl bond and concomitant
15N NMR spectra were obtained in deuteriochloroform
solution with an INEPT pulse sequence. Chemical shifts
are quoted relative to nitromethane. The influence of
hydrogen bonding upon the 15N chemical shift is best seen
by comparing the b- and c-series. As expected, NH‚‚‚‚‚
OdCOR hydrogen bonding shifts the 15N resonance to
lower field (column 9; 7b/9b vs 7c/9c). The N-oxide
oxygen has little effect on the amide-N chemical shift,
as shown by comparing 9c and 9d . Attempts to further
investigate the hydrogen bonding phenomena in this
class of compounds by 17O NMR spectroscopy proved
fruitless. No signals could be observed using a variety of
sequences (single pulse, RIDE) and acquisition condi-
tions. Thus, both in CDCl3 and CD3CN at room temper-
ature and 70 °C only the D2O peak was observed.
Finally, it was not possible to determine the structure
of the europium complexes (Eu (7c)3) and (Eu (9c)3) by
NMR spectroscopy. The former was virtually insoluble
(12) These complexes were prepared according to the procedure
given in ref 7a. All attempts to obtain single crystals suitable for X-ray
structure determination proved fruitless in the case of these com-
pounds. Eu (7c)3 could not be dissolved in any suitable solvent, and
those small quantities of material separating from large volumes of
very dilute methanol solutions were irregular and microcrystalline at
best. Eu (9c)3, on the other hand, provided an amorphous solid from
several different solvents. A more detailed discussion of these com-
pounds and their characteristics is being submitted elsewhere. The
limited solubility of similar types of europium complexes is well
documented. Typically they are obtained as powders. See ref 7a as well
as: de Sa´, G. F.; Barros Neto, B. d. B.; Ferreira, R. Inorg. Chim. Acta
1977, 23, 249. de Sa´, G. F.; Nunes, L. H. A.; Wang, Z.-M.; Choppin, G.
R. J . Alloys Compd. 1993, 196, 17. Boyd, S. A.; Kohrman, R. E.; West,
D. X. Inorg. Nucl. Chem. Lett. 1976, 12, 603.
(13) Oki, M.; Hirata, M. Nippon Kagaku Zasshi (J . Chem. Soc
J apan, Pure Chem Sec.) 1965, 86, 115.
(14) (a) Davies, M. In Hydrogen Bonding; Hadzi, D., Ed.; Pergamon
Press: London, 1959; p 393. (b) Oki, M.; Hirata, M. Bull. Chem. Soc.
J pn. 1964, 37, 209. (c) Oki, M.; Hirata, M.; Hirofuji, S. Spectrochim.
Acta 1966, 22, 1537.
1
in all common solvents, while the H NMR spectrum of
Eu (9c)3 was largely perturbed, displaying only a promi-
nent methylene envelope along with a series of very small
humps spread over the entire spectral region.12
The COO-H cis configuration in carboxylic acids is
favored by approximately 3 kcal mol-1 over the trans
form.13 However, it has been estimated that the enthalpy
of H-bond formation in o-alkoxybenzoic acids (leading to
a trans COO-H configuration) is about 3.3 kcal mol-1 14
.
We would expect the picolinic acid N-oxides (7c-9c),
were they in an s-cis configuration (vide infra), to be in
an overall conformation similar to that of the methyl
esters (7b-9b), i.e., a “twisted carboxyl”-conformation

374 J . Org. Chem., Vol. 67, No. 2, 2002
Barros et al.
F igu r e 2. Minimized conformations for selected 3-propanoylaminopicolinic acid derivatives using ab initio calculation in the
RHF/6-31G basis set. Gaussian numbering.
increase of the amide(H)-to-carboxyl(O) distance in the
N-oxides (7b/9b). Complete loss of intramolecular amide-
(H)-to-ester(O) hydrogen bonding would result in a
slightly greater shift to higher frequency as shown by
the N-H stretches of methyl 2-acetamidobenzoate and
attribute this to the presence of dimers4,15 in the KBr
matrix. Evidently, as the carbomethoxyl moiety rotates
from planarity in the b-series, the thus “liberated” amide
function is able to participate in intermolecular amide
association. The corresponding ester carbonyl stretching
vibrations in this case therefore suffer shifts to higher
wavenumbers (column 9) in accordance with this groups’
departure from involvement in a hydrogen bond to the
neighboring amide. Thus in all cases an increase in the
ester CdO stretching frequency in KBr upon going from
pyridine (a ) to N-oxide (b) implies the absence of an
intramolecular hydrogen bond in the latter species.
N-phenylacetamide4 (∆νmax(N-H) ) 121 cm-1 4
; Table 2,
column 4). The loss in H-bonding incurred between the
a - and b-series is regained in the planar N-oxide car-
boxylic acids (c-series), which display a particularly
strong hydrogen bond as shown by the low values for the
N-H stretch; (7c: νmax(N-H) 3114 cm-1, 9c: νmax(N-H) 3161
cm-1). This observation lends additional support to the
already mentioned very low field chemical shift of the
amide protons observed in these compounds, lower still
than in the similarly intramolecularly H-bonded precur-
sors 7a and 9a ; (Table 1; compare also the calculated
H11O22 distances for 10a and 10c-tr a n s in Table 4:
entry 33). A peculiar exception to this general trend is
observed in the KBr solid state IR spectra. In this case,
the surprisingly low amide N-H stretching frequency in
the N-oxides 7b and 9b (column 5) imply (contrary to
expectations) the presence of a strong hydrogen bond. We
In fact, the IR spectra of the N-oxide esters 7b/9b are
particularly intriguing. While a comparison of the char-
acteristic functional group frequencies for the a - and
c-series reveals very good agreement between KBr and
solution spectra, this is not so in the case of the
intermediate esters 7b/9b. Apart from the abovemen-
tioned discrepancy in the N-H stretches, the ester CdO
stretching frequencies differed substantially depending
(15) Perje´ssy, A.; Engberts, J . B. F. N. Monatsh. Chem. 1997, 126,
871.

Studies on 1,2-Di- and 1,2,3-Trisubstituted Heterocycles
J . Org. Chem., Vol. 67, No. 2, 2002 375
Ta ble 2. Selected P ea k F r equ en cies of IR Sp ectr a of 3-Am in op icolin ic Acid Der iva tives (7a -c, 9a -d ) in KBr Ma tr ix a n d
0.5% CH2Cl2 Solu tion a n d of Eu r op iu m (III) Com p lexes (Eu (7c)3, Eu (9c)3) in KBr h
a
b
Assignment of N-O stretching frequencies in CH2Cl2 not possible. Assignment tentative (see text). c Assignment tentative due to
low solubility. Reference 4. e Spectrum in CHCl3. f Spectrum in CCl4. g Aldrich Library of IR Spectra; Aldrich Chemical Co.: Milwaukee,
d
h
WI. Calculated ab initio 6-31G** values for model 3-propanoylaminopicolinic acid derivatives (10a -d ) and experimental peak frequencies
of pertinent reference compounds are given for comparison purposes. Values in cm-1
.
Ta ble 3. Com p a r ison of Selected Molecu la r Coor d in a tes for Op tim ized AM1, RHF /6-31G, a n d RHF /6-31G**
Con for m a tion s of 3-P r op a n oyla m in op icolin ic Acid N-Oxid es (10c-tr a n s a n d 10c-cis)a
10c-tr a n s
10c-cis
entry
coordinate
AM1
6-31G
6-31G**
AM1
6-31G
6-31G**
8
9
C21O23
C21O22
C12O13
N1O25
C2C21O23
C21O23H24
N1C2C21O22
C4C3N10H11
H7O13
1.3535
1.2382
1.2406
1.2275
122.54
113.15
148.18
-166.14
2.1254
2.0340
2.0211
1.3127 (+1.89)
1.2213 (+1.94)
1.2244 (+2.33)
1.3828 (+6.16)
118.75 (+0.90)
113.65 (+4.43)
179.99 (+0.00)
-180.00 (+0.04)
2.0893 (-1.23)
1.7881 (+0.02)
1.5918 (+3.89)
1.2883
1.1981
1.1965
1.3026
117.85
109.22
179.99
-179.96
2.1153
1.7877
1.5322
1.3568
1.2404
1.2407
1.2163
115.71
109.13
145.48
-166.04
2.1393
2.0711
1.3303 (+1.60)
1.2171 (+1.99)
1.2234 (+2.34)
1.3561 (+6.81)
113.97 (-0.05)
114.59 (+5.71)
128.15 (+0.55)
-158.38 (+0.54)
2.2198 (-1.62)
2.1371 (+1.12)
1.3094
1.1933
1.1954
1.2696
114.02
108.88
128.70
-157.84
2.2564
2.1135
22
23
31
47
54
58
70
71
72
73
H11O22
H24O25
O23O25
2.5941
2.8077 (+1.79)
2.7607
aBond lengths in angstroms, angles in degrees. In parentheses: deviation of RHF/6-31G values from corresponding RHF/6-31G** values:
in percent for bond lengths; in degrees for angles. Entry numbers correspond to the entries in the full Table S2 in the Supporting
Information.
on the medium in which the spectra were run. While the
KBr spectra for both compounds 7b/9b revealed single
strong absorptions in the expected regions (1749 and
1744 cm-1 respectively, column 9), the dichloromethane
spectra (column 8) show broad strong bands in the
carbonyl region (encompassing both amide-I and ester
CdO), with a weak shoulder at about 1750 cm-1, sug-
gesting the presence of two different conformations. As
before, one can rationalize these results in terms of the
twisting about the C2-ester bond in these compounds. As
already mentioned, in the solid-state intermolecular
amide association results in a hydrogen bond-free ester
twisted from coplanarity, with an appropriate high-
frequency CdO stretching vibration (≈1750 cm-1) as a

376 J . Org. Chem., Vol. 67, No. 2, 2002
Barros et al.
Ta ble 4. Com p a r ison of Selected Va lu es for Op tim ized RHF /6-31G a n d AM1 Con for m a tion s of 3-P r op a n oyla m in o
P icolin a tesa
10a
10b
10c-tr a n s
10c-cis
10d -tr a n s
10d -cis
1
9
total energy
C2C21
-718.869 352 2
1.4799
-793.605 134 9
1.4866
-754.617 711
-754.597 069 3
1.4841
-679.869 828 4
1.4862
-679.862 972
1.5072
1.4745
1.4879 (+0.54)
1.4797 (-0.46)
1.4880 (-1.27)
1.4761 (-0.54)
1.4907 (+0.30)
1.4850 (+0.71)
23 N1C2C21O23
24 C2C3N10H11
26 C4C3N10C12
32 H7O13
-0.01
-59.86
0.00
-59.09
-0.01
-0.01
-42.05 (+42.04)
0.00
15.57 (+15.57)
0.04
16.26 (+16.22)
2.1393
-40.96 (-18.90)
20.93
14.06 (-6.87)
18.86
14.62 (-4.24)
2.2264
-33.91 (+33.91)
0.00
13.36 (+13.36)
0.01
13.69 (+13.69)
2.0893
-36.54 (-22.55)
20.88
13.46 (-7.42)
18.57
14.19 (-4.38)
2.2198
-22.08 (+22.07)
0.00
11.27 (+11.27)
0.00
13.04 (+13.04)
2.1479
-38.49 (+38.48)
0.00
15.14 (+15.14)
0.01
15.96 (+15.95)
2.1382
2.1556 (+0.76)
1.8565
2.1441 (-3.70)
2.1442
2.1254 (+1.73)
1.7881
2.1393 (-3.63)
2.1371
2.1263 (-1.01)
1.9032
2.1511 (+0.60)
1.8672
33 H11O22
2.1193 (+14.16)
2.0874 (-2.64)
2.8129
2.0340 (+13.75)
2.0711 (-3.09)
2.8077
2.0376 (+7.06)
2.1047 (+12.72)
34 O23O25
2.6378 (-6.22)
2.5941 (-7.61)
35 H24O25
1.5918
2.0211 (+26.97)
a
Total energy in hartrees, Bond lengths in angstroms, dihedral angles in degrees. For each entry: first line: RHF/6-31G values.
Second line: AM1 values. In parentheses: deviation of AM1 value from RHF/6-31G value in percent for bond lengths, in degrees for
angles. Entry numbers correspond to the entries in the full Table S1 in the Supporting Information.
result of lower conjugation with the aromatic ring. In
dichloromethane an equilibrium exists between the
intermolecular “amide-association“ and intramolecular
“amide-to-ester H-bond” species with high frequency and
low-frequency ester CdO stretching bands, respectively.
This hypothesis was tested by measuring the IR spec-
trum of 9b in CH2Cl2 solution at different concentrations.
At concentrations of 0.5, 2 and 8%, the ratio of the peak
intensities at 1751 and 1709 cm-1 were 0.25, 0.31 and
0.68 respectively, thereby corroborating our conclusion
that at higher concentrations dimer formation is in-
creased.
In the c-series there is once again very good agreement
when comparing KBr spectra with those obtained in
dichloromethane solution. We reason, that in 7c/9c the
N-oxide effectively “freezes” the acid function into a single
orientation by means of an additional hydrogen bond. An
equilibrium between two different conformations, which
can only reasonably be rotamers about C2-COOH can
be ruled out in this case, the entire molecule being
effectively locked into place by two intramolecular hy-
drogen bonds on its’ periphery (5). Thus a single confor-
mation is observed irrespective of whether the IR-
spectrum is run in KBr or dichloromethane. Similar
agreement between solid state and solution spectra is
noted for 7a /9a , which is to be expected in view of the
absence of the N-oxide function and presence of an
intramolecular hydrogen bond.
Finally, the N-O stretch in the N-oxides (7b/9b) ought
to be shifted to lower frequency upon ester saponification,
as would be expected from the formation of a hydrogen
bond to N-O in derivatives 7c and 9c. Unfortunately the
preponderance of bands between 1200 and 1300 cm-1
precludes any unambiguous identification of the band in
question.
Complexation of the N-oxide acids (7c/9c) to Eu3+
occurs at the N+-O-/COO- site,7b,c providing Eu (7c)3 and
E u (9c)3 respectively. That coplanarity of the pyridine
and amide moieties was maintained in these complexes
follows from the N-H stretching vibrations, which
indicate strong hydrogen bonding comparable to that in
the precursors (7c/9c) (Table 2). The complexes were
highly luminescent, showing improved optical properties
over their unsubstituted, free amine predecessors.7,16
Molecular modeling using the Sparkle Program devel-
oped in this Department17 furnished planar ligand ge-
ometries in all cases upon minimization.16
Molecu la r Mod elin g
A. Con for m a tion a l Stu d ies. Having thus interpreted
our spectroscopic findings in terms of a convincing
conformational rationale we proceeded to underpin our
conclusions with an investigation of these molecules by
molecular modeling. To this end we subjected a model
series of 3-N-propanoylaminopicolinic acid derivatives
(10a -d ) as well as the propylamine (11) to a series of
semiempirical and ab initio calculations at different levels
of theory, performing an extensive conformational analy-
sis to study energetic, geometric and electronic density
aspects.
To guarantee the best compromise between the theo-
retical analysis of the molecules and our real computa-
tional availability, we decided to model the alkyl-amide
series. Beyond that, we used 3-N-propanoylaminopicoli-
nyl- as a prototype system for representing the 3-N-
dodecanoylaminopicolinate series. For the semiempirical
calculations the AM1 method implemented in the Mopac
6.0 package18 was used, whereas the ab initio calculations
at the Restricted Hartree-Fock (RHF) level using 6-31G
and 6-31G** basis sets19 were performed in the Gaussian
(16) Unpublished results; manuscript in preparation.
(17) Rocha, G. B. Master’s Thesis, Universidade Federal de Per-
nambuco, 1998. Andrade, A. V. M. d.; Longo, R. L.; Simas, A. M.; de
Sa´, G. F. J . Chem. Soc., Faraday Trans. 1996, 96, 1835. Andrade, A.
V. M. d.; Costa, N. B. d., J r.; Simas, A. M.; de Sa´, G. F. Chem. Phys.
Lett. 1994, 227, 349.
(18) Dewar, M. J . S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J . J . P.
J . Am. Chem. Soc. 1985, 107, 3902.
(19) Hehre, W. J .; Ditchfield, R.; Pople, J . A. J . Chem. Phys. 1972,
56, 2257.

Studies on 1,2-Di- and 1,2,3-Trisubstituted Heterocycles
J . Org. Chem., Vol. 67, No. 2, 2002 377
94 program.20 Both programs were run on Risc 6000
workstations in the Molecular Architecture Laboratory
of the Universidade Federal de Pernambuco.
no difference between the calculated 6-31G and 6-31G**
N1C2C21O22 (entry 54) and C4C3N10H11 (entry 58)
dihedral angles. Furthermore, the H7O13, H11O22 and
O23O25 nonbonding interatomic distances (entries 70,
71, and 73) predicted by the two basis sets also showed
acceptable agreement. The difference in the H11O22
hydrogen bond length for 10c-tr a n s was only 0.0004 Å
(0.02%) for these two calculations. Due to the discrepancy
in the predicted N-O bond lengths as well as C2C21O23
(entry 31) and C21O23H24 (entry 47) bond angles, both
of which are larger for 6-31G, there is a difference of
almost 4% in the predicted H24O25 hydrogen bond
lengths of 10c-tr a n s (entry 72). Significantly though, this
result does not perturb the overall planar molecular
conformation. In view of these results the 6-31G basis
set was considered adequate for the conformational
studies at hand, so that, to economize on computer time,
other ab initio calculations can be carried out with this
basis set only. It is quite clear on the other hand however,
that AM1 results were inadequate (see, for example,
entries 54, 58, 71, and 72). These are discussed at length
in the Supporting Information.
In Figure 2, we show the optimized RHF/6-31G struc-
tures for all the 3-N-propanoylaminopicolinate deriva-
tives modeled (10-series), while Table 4 depicts selected
geometric parameters together with the total energies
for the completely optimized structures at both ab initio
and AM1 levels of theory. A comparison of AM1 and
6-31G derived values quickly shows considerable differ-
ences between the two. (A more thorough discussion of
the differences between ab initio and AM1 results is given
as Supporting Information.)
Intriguingly, the AM1 models provided some highly
improbable structural results (see Table 4 and the
Supporting Information) contrary not only to sound
chemical intuition but, more importantly, to experimental
data. Indeed, Perni et al.21 recently disclosed that in a
theoretical study of pyridine-N-oxide at 10 different
theory levels, semiempirical calculations, including AM1,
gave unsatisfactory bond lengths when compared to
experimentally derived values, especially as regards the
N-O bond. The best method found in those studies was
the 6-31G** basis set. At this level of theory, the
calculated N-O bond length in pyridine-N-oxide was
1.16% lower than the experimentally derived one, while
6-31G provided a value 6.01% higher. There was thus a
discrepancy of 7.25% between the 6-31G** and 6-31G
derived N-O bond length values. In view of those results,
we ran the isomers (10c-tr a n s) and (10c-cis) (Figure 2)
through a series of different calculations. The semiem-
pirical AM1 method was clearly unsatisfactory, but we
found little overall difference between ab initio RHF
results obtained using STO-3G, 3-21G, 6-31G or 6-31G**
basis sets. The principal molecular geometrical param-
eters for AM1, 6-31G and 6-31G** pertinent to the
discussion at hand are depicted in Table 3 (a full table is
given in the Supporting Information). From an electronic
point of view, similar qualitative agreement between 10c-
tr a n s and 10c-cis was also found for Mulliken atomic
charges derived from 6-31G and 6-31G** basis sets.
In good agreement with Perni’s results,21 the N-O
bond lengths in the two conformers studied were between
6 and 7% higher in the 6-31G as compared to the 6-31G**
basis set (entry 23). This gives a slightly more open
C21O23H24 bond angle in the carboxyl group of 10c-
tr a n s as the longer N-O bond “pushes” the bound
hydrogen outward, (6-31G: 113.65° vs 6-31G**: 109.22°,
entry 47). Other differences in molecular coordinates
between the two ab initio basis sets were generally less
than 1% for bond lengths or less than 1° for angles in
both conformers (see the Supporting Information), with
the only exceptions being the polar C21O23 and C21O22
bonds in the carboxyl group and the amide carbonyl,
C12O13 (6-31G bond lengths 1.60-2.34% higher, entries
8, 9, and 22, respectively). These discrepancies, although
not as profound, are in accord with the trend observed
by Perni21 for the N-O function inasmuch as the 6-31G
basis set, devoid of polarization functions, gives rise to
an overestimation for the bond length in such (polar)
cases. Satisfyingly, the values of the molecular coordi-
nates most important to the main thrust of the discus-
sions herein were essentially immune to which of the two
basis sets was used to obtain them. As such, both the
acid and the amide functions in 10c-tr a n s were predicted
as being completely coplanar with the pyridine ring, with
As can be seen, the conformational predictions gleaned
from our spectroscopic results are corroborated by the
optimized RHF/6-31G structures for the compounds
shown in Figure 2. In particular, the starting pyridine
ester (10a ) as well as the N-oxide acid (10c-tr a n s) are
completely planar as borne out by the N1C2C21O23 and
C2C3N10H11 angles (entries 23 and 24). The H11O22
distances of 1.857 and 1.788 Å respectively (Entry 33),
clearly indicate a strong intramolecular hydrogen bond
in these substrates. Furthermore, the C4C3N10C12 zero
dihedral angles (Entry 26) show that the amide carbonyl
function is in the plane of the aromatic ring with the
NCdO group strongly deshielding H7 (Gaussian num-
bering) (H7O13 ) 2.139 (10a ) and 2.089 Å (10c-tr a n s),
entry 32), in agreement with the 1H NMR chemical shifts
in Table 1 for 9a and 9c. Our spectral data discussed
above indicate that the N-oxide esters (b-series) have
significantly weaker intramolecular hydrogen bonds than
the a - and c-series. As can be seen from the optimized
conformation of the 9b-model (the propanoylamino ana-
logue (10b) having the methyl ester in a preferred s-cis
conformation), the introduction of an N-oxide group
obliges the central carbomethoxyl to rotate from copla-
narity. We rationalize this observation as a result of
repulsion between the oxygen atoms O23 and O25. In
the optimized conformation the C2C21 bond (entry 9) is
slightly longer in 10b than in 10a (1.487 vs 1.480 Å)
reflecting a reduction in bond order as the carbomethoxyl
group is twisted by as much as -60 degrees (N1C2C21-
O23, entry 23) from the aromatic plane. O23 and O25
are thus 2.813 Å apart (entry 34). That this is so, and
that for example the methyl group in the ester has no
bearing upon this torsion was underlined by performing
the ab initio calculation on a hypothetical analogue with
(20) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
J ohnson, B. G.; Robb, M. A.; Cheeseman, J . R.; Keith, T.; Petersson,
G. A.; Montgomery, J . A.; Raghavachari, K.; Al-Laham, M. A.;
Zakrzewski, V. G.; Ortiz, J . V.; Foresman, J . B.; Cioslowski, J .;
Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala,
P. Y.; Chen, W.; Wong, M. W.; Andres, J . L.; Replogle, E. S.; Gomperts,
R.; Martin, R. L.; Fox, D. J .; Binkley, J . S.; Defrees, D. J .; Baker, J .;
Stewart, J . P.; Head-Gordon, M.; Gonzalez, C.; Pople, J . A. Gaussian
94, revision E.3; Gaussian, Inc.: Pittsburgh, PA, 1995.
(21) Perni, R. B.; Kowalczyk, P. J .; Treasurywala, A. M.; Tracy, M
Tetrahedron Lett. 1995, 36, 699.

378 J . Org. Chem., Vol. 67, No. 2, 2002
Barros et al.
F igu r e 3. Variation of total energy and H7- and H11 Mulliken
charges as a function of N1C2C21O23 dihedral angle for
N-oxide ester 10b.
F igu r e 4. Variation of H11-Mulliken charge, H11O22 and
H11O23 interatomic distances as a function of N1C2C21O23
dihedral angle for N-oxide ester 10b.
the acid constrained in the s-cis conformation (10c-cis).
In this case, the central carboxyl group twists from
coplanarity by the same amount (-59.09°, entry 23;
O23O25 ) 2.808 Å, entry 34) as in the s-cis ester (10b).
The twisting of the carbomethoxyl group in 10b neces-
sarily increases the H11O22 distance resulting in a
significantly weakened hydrogen bond (2.144 Å) as
compared to 10a (1.857 Å, entry 33). This is corroborated
by the infrared N-H stretching frequencies for the
corresponding dodecanoylamides 9a and 9b (νmax(N-H)
3318 and 3401 cm-1 respectively) as well as by the amide
proton NMR chemical shifts (δ(N-H) 10.98 and 8.83 ppm
respectively), the NH in the former substrate being
significantly less shielded.
(acyl-O) interatomic distances for each of the abovemen-
tioned minimized fixed angle conformations of compound
10b.
At the extremities of 0° and 180° H11‚‚‚‚O22 or H11‚
‚‚‚O23 hydrogen bonds respectively result in higher
charges at H11, while a Mulliken charge minimum is
observed at the center, corresponding to an orthogonal
orientation of the COOMe group with respect to the NH-
bond, with no possibility for hydrogen bonding to either
oxygen. As the N1C2C21O23 angle increases past -100°,
the H11O23 (amide(H)-to-(O)alkyl) distance falls below
H11O22 and an inversion of the C2C3N10H11 dihedral
angle takes place (data not shown) as H11 and O23
approach each other.
The twisting at hand of the ester group in 10b is
accompanied by a similar departure from planarity of the
amide function. Inspection of the minimized conformation
of 10b reveals that N-oxidation of 10a induces a disro-
tatory twisting about C2C21 and C3N10 so as to place
the H-bonding partners H11 and O22 on the same side
of the aromatic plane (cf. 4 f 4tt). Significantly, the
amide group is twisted at only about 20° (entries 24 and
26, Table 4) to the pyridine ring reflecting this groups’
reluctance to sacrifice delocalization energy. This amide
rotation however manifests itself in the considerable
upfield NMR shift for the para proton (H4 in 9b) in the
b-series (as compared to the a -series), for which the +M
effect of the N-oxygen is only partly responsible. In
addition, the rotation of the amide group necessarily
modifies the anisotropy exerted upon this protonsthe
distance between H7 and the amide carbonyl oxygen O13
is increased from 2.139 Å in 10a to 2.226 Å in 10b (entry
32) and the mutual orientation of these atoms is modified
resulting in a further upfield shift of the NMR reso-
nances.
In a simulated full rotation of the carbomethoxyl group
in 10b about C2C21 we have investigated the change in
total energy as well as the electronic Mulliken charges
on the para and amide protons, H7 and H11 respectively,
as functions of dihedral angle. To this end the molecule
was optimized for a series of fixed dihedral angle con-
formations at 30° intervals around the C2C21 bond. The
results are depicted in Figure 3.
Starting from a planar situation (i), the lowest energy
conformation (ii) for 10b is reached after a carbomethoxyl
rotation of -60 degrees (see structure 10b in Figure 2)
resulting in a slightly longer C2C21 bond as conjugation
is disrupted. Continuation to complete inversion (-180°),
placing N-oxide and ester carbonyl dipoles in parallel,
results in appreciable destabilization, (iii). The total
Mulliken electronic charge at the para proton (H7) varies
little during this rotation. Significantly though, the
charge is highest for planar conformations (i) and (iii) in
which hydrogen bonding to the amide, and therefore H7-
deshielding, are strongest. As anticipated, the variation
in Mulliken charge at the amide proton (H11) on the
other hand roughly parallels the change in total energy
as a function of the N1C2C21O23 dihedral angle. This
is to be expected because the rotation of the car-
bomethoxy group resulting in total energy lowering with
departure from planarity and relief of strain, also weak-
ens hydrogen bonding to H11, thereby decreasing the
Mulliken charge.
The calculated RHF/6-31G energy difference between
the N-oxide acid (10c-tr a n s) and its hypothetical s-cis
carboxyl isomer (10c-cis) is 12.95 kcal‚mol-1, with the
former more stable. While the s-cis conformer (10c-cis)
would have an almost identical conformation to 10b (cf.
Table 4, columns 4 and 6), the N-oxide acid actually exists
preferably in the 10c-tr a n s form attaining coplanarity
upon ester saponification of the twisted precursor (10b).
This state of affairs obviously results out of increased
conjugation coupled with the formation of two strong
Figure 4 depicts the variation in Mulliken charge of
H11 as a function of the N1C2C21O23 dihedral angle,
and at the same time the H11O23 (alkyl-O) and H11O22-

Studies on 1,2-Di- and 1,2,3-Trisubstituted Heterocycles
J . Org. Chem., Vol. 67, No. 2, 2002 379
intramolecular hydrogen bonds (H11O22: 1.788 Å,
H24O25: 1.592 Å, entries 33 and 35) effectively locking
the molecule into a planar conformation. This regaining
of coplanarity in going from the b- to the c-series was
previously inferred both from infrared (Table 2, shift to
lower N-H stretch frequencies) as well as NMR data
(Table 1, lowfield shifts for 9c) pointing to increased
amide-to-acid hydrogen bonding. The modeling results
at hand clearly substantiate this interpretation.
We also indirectly investigated the subtle conforma-
tional influences upon the H7 NMR chemical shifts by
virtue of calculation of H7 Mulliken electronic charges.
The H7 chemical shifts observed for compounds of the
c-series are a compromise between neighboring amide
carbonyl deshielding and N-oxide shielding, which in its
turn can be attenuated by hydrogen bonding to the
carboxyl-OH in the carboxylic acid (10c-tr a n s). The
electronic Mulliken charge on H7 as a function of rotation
of the amide and acid-OH groups in 10c-tr a n s and 11
was calculated, to obtain its’ variation as a function of
dihedral angle. The results are depicted in Figure 5.
the neighboring acid OH. As this group is rotated from
the s-trans to the s-cis conformation in 10c (10c-tr a n s
f 10c-cis, rotation B), hydrogen bonding is gradually
weakened and eventually lost, with a concomitant, albeit
small, decrease in the positive electronic Mulliken charge
on H7 as the N-oxide +M effect increases. That it is
largely the amide group (and not N-oxide hydrogen
bonding) which is responsible for the exceptional H7-
downfield shift is supported not only by comparing the
H7 chemical shifts in the free amine 6c with those for
amides 7c-9c, but also by performing a similar rota-
tional simulation (rotation C) of the acid OH group on
the propylamine derivative 11 devoid of amide CdO
anisotropy. In this case the electronic Mulliken charge
on H7 is not only much lower but a similar effect of only
a small(!) H7-charge lowering in going from a hydrogen-
bonded N-oxide (11) to a free one (11′) is seen. The
convergence of the amide group rotation curve for 10c-
tr a n s (rotation A) with that for OH rotation in amine
11 (rotation C) is interesting because it suggests a similar
environment for H7 in the two resulting conformers (10c-
tr a n s′ and 11′). In conclusion it can be said, that the
deshielding due to amide anisotropy outweighs the
N-oxide (+M) shielding effect in determining the NMR
chemical shift of the H7 proton in the para position, and
that reduction of the latter effect by H-bonding a neigh-
boring carboxyl-H results in minimal charge elevation
at the para hydrogen (H7).
As opposed to the N-oxides, the minimized conforma-
tions of the pyridine analogues 10d -tr a n s and 10d -cis
(methyl picolinates) are both planar, with the former
more stable by 4.30 kcal.mol-1. Devoid of the destabilizing
N-oxide T alkyloxy repulsion, 10d -tr a n s would assume
a planar shape and take full advantage of the amide-to-
carbomethoxyl hydrogen bond.
Cyclic dimers of the type (12) are common for second-
ary amides.15 In the case of N-oxide-ester (10b), possess-
ing twisted C2-carbomethoxyl and C3-amide bonds and
thus devoid of stabilizing intramolecular hydrogen bond
interactions, we entertained the possibility of amide
dimer formation both in CH2Cl2 solution and KBr matrix
during IR-studies (vide supra). With this in mind we
performed an ab initio RHF geometry optimization of the
dimer of 10b using the 6-31G basis set. To this end we
used a modified starting conformation in which the amide
configuration was set as trans. The optimized dimeric
arrangement is shown in Figure 6. The H11O41/O13H39
interatomic distances of 1.876 Å indicate moderately
strong hydrogen bonding, while the N1C2C21O22 and
C2C3N10C12 dihedral angles of -93.39 and 91.46°
respectively show that both carbomethoxyl and amide
groups are orthogonal to the pyridine ring. The total
energy of the dimer was calculated as -1587.967496 au,
a value 3.36 kcal‚mol-1 less than twice the calculated
energy of the monomeric cis-amide (10b), thereby indi-
cating that some gain is to be had from dimer formation
compensating the loss of intramolecular hydrogen bond-
ing to the ester carbonyl. However, we also performed
the RHF calculation with the 6-31G** basis set. The
geometric parameters were very similar (H11O41/
O13H39, N1C2C21O22 and C2C3N10C12 ) 1.976 Å,
-95.92 and 94.95°, respectively), although the energy of
the dimer was only 0.32 kcal‚mol-1 less than that of twice
the monomer energy. Despite the fact, that these values
satisfyingly support our idea of dimer formation, the
results suggest that thus determined H-bond energies
F igu r e 5. Variation of H7-Mulliken charge, as a function of
dihedral angles C4C3N10C12 and C2C21O23H24 in N-oxide
acid (10c-tr a n s) and as a function of C2C21O23H24 dihedral
angle in N-oxide propylamine (11).
The striking effect of removing the amide CdO deshield-
ing cone is apparent in a large drop in the H7 electronic
Mulliken charge in 10c-tr a n s on rotating the CdO group
away from this proton (rotation A). As anticipated above,
the N-oxide function para to H7 can hydrogen bond to

380 J . Org. Chem., Vol. 67, No. 2, 2002
Barros et al.
investigated. 6-31G** was much better. It is interesting
to note, that there were well-defined, almost fixed devia-
tions of calculated 15N chemical shifts from experimental
ones of 24% for 6-31G**- and 66% for 6-31G basis sets.
Comparing 7b/c and 9b/c it can be seen, that the amide
CdO substituent has only a small effect on the 15N
chemical shift (∆(δ-15N7b/9b) ) 1.8%; ∆(δ-15N7c/9c) ) 1.97%),
so that model-(10)-based δ-15N calculations provide good
predictability for arylamides as well, provided a +24%
correction is taken into account. Although the calculated
nitrogen chemical shifts are significantly less accurate
than the proton shifts, a coherent trend of these values
in accordance with hydrogen bonding can be discerned.
Thus, in the same way that involvement of amide-H in
hydrogen bonding in 9c leads to a 5.8 ppm downfield shift
of the 15N resonance with respect to precursor 9b, there
is a similar shift of 6.9 ppm in the corresponding
calculated resonances for the model compounds (10b) and
(10c).
F igu r e 6. Minimized RHF/6-31G structure of the dimer of
N-oxide ester 10b.
strongly depend on the basis set employed. Such differ-
ences are well recorded,22 and consequently undue im-
portance should not be imparted upon these values.
B. Spectr oscopic Stu dies. Selected calculated23 GIAO
1H and 15N NMR chemical shifts for the model 3-pro-
panoylaminopicolinates (10a -d ) using both 6-31G and
6-31G** basis sets are shown in Table 1. When compar-
ing with 9a -d , the data show two important trends.
First, 6-31G**-derived chemical shifts are uniformly more
reliable than 6-31G-derived ones. Thus, in the 6-31G**
basis set the calculated chemical shifts for the three
aromatic protons are, on average, about 4% higher than
the experimentally determined values, while 6-31G cal-
culations provide an average over-estimate of 12.5% for
these protons. Second, taking the aromatic protons
individually over all the derivatives (10a -d ), the esti-
mates for H5 are clearly the most accurate with an
average deviation from the experimental spectroscopic
values of +2.2% and +8.4% for 6-31G** and 6-31G,
respectively. H4- and H6-estimates are slightly less
accurate with 6-31G calculated chemical shifts deviating
about 2 to 3 times as much (in percent) as the 6-31G**
values. The larger deviations of calculated H4 and H6
shifts from experimental values are consistent with the
fact that these hydrogens are closer than H5 to electronic
perturbation originating in the polar pyridine nitrogen/
N-oxide and amide moieties. It is therefore to be expected,
that, as compared to 6-31G, employment of the 6-31G**
basis set will improve all three, but especially, as indeed
borne out, the chemical shift estimates for H4 and H6.
Although, strangely the 6-31G** basis set provides no
improvement over 6-31G with regard to accuracy in
predicting amide hydrogen chemical shifts, the deviations
from the experimentally derived values are of the same
order of magnitude in the case of the planar molecules
(10a ), (10c) and (10d ) (≈6-11%) as was seen with the
aromatic protons. The sole outlier was the “twisted”
derivative (10b), for which both basis sets resulted in a
20% overestimation of the true N-H chemical shift.
In the case of 15N NMR, 6-31G based ab initio calcula-
tions grossly over-estimate the absolute value of the
chemical shifts of the amide nitrogen in the compounds
The IR-spectra of model compounds (10a -d ) were
calculated with both 6-31G and 6-31G** basis sets. The
unambiguous attribution of calculated peak frequencies
in the former case was much more difficult, especially
for carbonyl stretches, which were often weak in intensity
as well as apparently coupled to more complex vibrational
modes. For this reason, Table 2 shows only the theoretical
peak values obtained from the 6-31G** basis set. The
calculated peak frequencies were on average uniformly
15-20% higher in model compounds (10a -d ) when
compared to the respective experimentally investigated
dodecanoyl analogues (9a -d ).
Despite the deviation of calculated values from experi-
mental ones, it was gratifying to observe that the shift
tendencies were always in the same direction (albeit
generally smaller) upon passing along the series a f b
f c f d . Thus, disruption of hydrogen bonding causes
an increase in (KBr)νN-H (7a f 7b: +2.6%, 9a f 9b:
+2.5%, 10a f 10b: +0.9%) and subsequent renewed
H-bonding results in a decrease (7b f 7c: -8.4%, 9b f
9c: -7.1%, 10b f 10c: -2.4%). Similar trends were
observed for the Amide-I bands. The ester CdO stretch
for 10b on the other hand is an interesting point. The
calculated νCdO stretch for this N-oxide ester (1975 cm-1
)
can only be one corresponding to the twisted, weakly
intramolecular hydrogen bonded nondimerized conformer
(as shown in 10b), and as such must be compared to the
peak at 1709 cm-1 for 9b. To further investigate this
point, we submitted the 10b-dimer to an ab initio
calculation, to obtain the calculated peak frequency for
ester νCdO in this species in the expectation that this
value would be higher than the calculated νCdO value for
10b. Unfortunately however, this calculation greatly
exceeded our computational possibilities.
Finally, as in the experimental series, the N-O
stretches were difficult to attribute, being coupled to
breathing of the pyridine ring. To aid in peak assignment,
we calculated the IR-spectrum of pyridine-N-oxide using
both basis sets. Indeed, as expected in view of Perni’s21
and our own results presented herein, 6-31G did not
provide an unambiguously attributable N-O stretch,
while the 6-31G** basis set yielded a value of 1368 cm-1
,
(22) Arau´jo, R. C. M. U.; da Silva, J . B. P.; Neto, B. B.; Ramos, M.
N. Chemo. Intel. Lab. Systems 2001, submitted.
(23) Wolinski, K.; Hilton, J . F.; Pulay, P J . Am. Chem. Soc. 1990,
112, 8251.
which compares well to the corresponding calculated
frequencies for 10b and 10c.

Studies on 1,2-Di- and 1,2,3-Trisubstituted Heterocycles
J . Org. Chem., Vol. 67, No. 2, 2002 381
J ) 7 Hz, 2H), 7.60 (dd, J ) 8.7, 4.5 Hz, 1H), 4.10 (s, 3H); IR
(KBr) νmax 3300, 1688 cm-1; MS m/z (rel intensity) 301 (M+,
1).
Con clu sion
3-Amidopicolinic acids and their N-oxides exist in a
planar conformation due to strong intramolecular NH‚‚
‚OdC and N-O‚‚‚H-O hydrogen bonds. In the case of
methyl 3-acylaminopicolinate-N-oxides, repulsion be-
tween N-O and alkyl oxygens results in twisting of the
carbomethoxyl and amide functions from the aromatic
plane. The conformation of such compounds can be
investigated theoretically, but care must be exercised in
choosing the molecular modeling approach for N-oxides,
since AM1 gives highly improbable results. The best
method found for pyridine N-oxide itself involves use of
ab initio RHF calculations with the 6-31G** basis set,21
although we have found that 6-31G works equally well
for determining molecular geometry. Discrepancies be-
tween 6-31G and 6-31G** derived results are found for
polar groups, especially the N-oxide bond. It seems
probable, that the former basis set is satisfactory for
modeling N-oxides, if a correction factor of -7% is used
for the estimation of the N-oxide bond length. Molecular
modeling studies of 3-propanoylaminopicolinic acid de-
rivatives show that the 1H NMR resonances of the proton
in the pyridine 4-position in such substrates are shifted
downfield as a result of deshielding by the neighboring
amide carbonyl group. For more reliable predictions
particularly with regard to ir and NMR spectroscopic
Meth yl 3-Dod eca n oyla m in op icolin a te (9a ). This mate-
rial was prepared in a fashion similar to that for 7a using CH2-
Cl2 at room temperature instead of refluxing CHCl3. Recrys-
tallization from MeCN provided a white, amorphous solid
(100%): mp 68-71 °C; 1H NMR (300 MHz, CDCl3) δ 10.98 (br
s, 1H), 9.15 (dd, J ) 8.7, 1.5 Hz, 1H), 8.41 (dd, J ) 4.5, 1.5 Hz,
1H), 7.49 (dd, J ) 8.7, 4.5 Hz, 1H), 4.04 (s, 3H), 2.47 (t, J )
7.5 Hz, 2H), 1.75 (m, 2H), 1.44-1.20 (m, 16H), 0.87 (t, J ) 6.6
Hz, 3H); 13C NMR (75 MHz, CDCl3) δ 172.5, 168.0, 143.3,
139.5, 131.4, 128.4, 128.2, 96.2, 53.0, 38.6, 31.9, 29.7, 29.5, 29.4,
29.2, 25.4, 22.7, 14.2; IR (KBr) see Table 2; MS m/z (rel
intensity) 334 (M+, 13).
Meth yl 3-Ben zoyla m in op icolin a te N-Oxid e (7b). To a
stirred solution of the amide 7a (2 g, 7.8 mmol) in MeCN (50
mL) was added ≈85% m-CPBA (3.37 g, ≈16.8 mmol, ≈2.2
equiv). After 2 h, filtration and evaporation gave a white solid.
Two cycles of slurrying in cold MeCN, filtration, and evapora-
tion removed most of the remaining m-CPBA and m-CBA.
Chromatography of the residue (EtOAc/MeOH) followed by
crystallization from cyclohexane/EtOAc furnished a white
1
powder (710 mg, 33%): mp 153-155 °C; H NMR (300 MHz,
CDCl3) δ 9.82 (s, 1 H), 8.55 (dd, J ) 8.7, 1.0 Hz, 1 H), 8.05
(dd, J ) 6.0, 1.0 Hz, 1H), 7.91 (m, 2H), 7.60 (m, 1H), 7.54 (m,
3H), 7.35 (dd, J ) 8.7, 6.0 Hz, 1H), 4.09 (s, 3H); IR (KBr) see
Table 2; MS m/z (rel intensity) 272 (M+, 6).
Meth yl 3-p-Nitr oben zoylam in opicolin ate N-Oxide (8b).
A procedure similar to that for 7b employing m-CPBA (≈4
equiv) for 2 days provided, after chromatography with EtOAc/
MeOH and crystallization from EtOAc, off-white needles
parameters the 6-31G** basis set is a must. H and 15N
chemical shifts for this method are on average overesti-
mated by about 4% and 24% respectively, while 6-31G
gives an almost three times larger average deviation.
1
1
(35%): mp 159-160 °C; H NMR (300 MHz, CDCl3) δ 10.24
(s, 1H), 8.55 (br d, J ) 8.4 Hz, 1H), 8.39 (m, 2H), 8.12 (m, 3H),
7.38 (dd, J ) 8.4, 6.5 Hz, 1H), 4.10 (s, 3H); IR (KBr) νmax 3096,
1757, 1693 cm-1; MS m/z (rel intensity) 317 (M+, 5).
Meth yl 3-Dod eca n oyla m in op icolin a te N-Oxid e (9b). A
solution of the pyridine ester (9a ) (1.08 g, 3.23 mmol) in CHCl3
(25 mL) was treated with ≈70% m-CPBA (980 mg, ≈5.68
mmol, ≈1.8 equiv) and heated to 50 °C for 1 h. Solvent
evaporation and flash chromatography (eluent: EtOAc/MeOH)
of the residue gave recovered starting material (360 mg, 33%)
followed by the desired N-oxide (9b) as a light yellow powder
(450 mg, 60% based on recovered starting material) which
could be recrystallized from i-Pr2O to provide pure material:
Exp er im en ta l Section
Melting points are corrected. Flash chromatography was
carried out using silica gel (Merck, Kieselgel 60, 230-400
mesh) and analytical TLC was done on precoated plastic sheets
(Macherey-Nagel, Polygram SIL G/UV254). Commercial AR
grade solvents and reagents (Fluka, Aldrich) were used directly
unless, where necessary, purified using standard published
procedures.24 The starting material, methyl 3-aminopicolinate
(6a ) as well as analogues (6b) and (6c) were prepared
according to ref 7a. NMR spectra were obtained on a Varian
Unity Plus 300 spectrometer in the Departamento de Quı´mica
Fundamental. Samples were dissolved in CDCl3. Chemical
shifts are quoted relative to TMS (δ ) 0) for 1H and ni-
tromethane (δ ) 0) for 15N.
1
mp 91-92 °C; H NMR (300 MHz, CDCl3) δ 8.82 (br s, 1H),
8.36 (br d, J ) 8.7 Hz, 1H), 7.92 (br d, J ) 6.3 Hz, 1H), 7.25
(dd, J ) 8.7, 6.3 Hz, 1H), 4.04 (s, 3H), 2.36 (t, J ) 7.5 Hz, 2H),
1.68 (m, 2H), 1.42-1.20 (m, 16H), 0.89 (t, J ) 6 Hz, 3H); 13C
NMR (75 MHz, CDCl3) δ 170.9, 163.4, 137.7, 134.8, 130.8,
126.3, 116.7, 96.0, 53.0, 37.7, 31.7, 29.4, 29.3, 29.2, 28.9, 24.9,
22.5, 14.0; IR (KBr) see Table 2; MS m/z (rel intensity) 351
(MH+, 31).
Meth yl 3-Ben zoyla m in op icolin a te (7a ). To a stirred,
refluxing solution of the amine 6a 7a (2 g, 13.2 mmol) in CHCl3
(20 mL) were added pyridine (2.1 mL, 2 equiv), DMAP (50 mg),
and a solution of benzoyl chloride (2.2 mL, 19.8 mmol, 1.5
equiv) in CHCl3 (10 mL). After 22 h at reflux, the solution was
cooled and washed sequentially with saturated aqueous Cu-
SO4, dilute NaOH, and brine, dried (Na2SO4), filtered, and
concentrated. Recrystallization of the residue from cyclohex-
ane/EtOAc provided a white powder (2.9 g, 86%): mp 134-
3-Ben zoyla m in op icolin ic Acid N-Oxid e (7c). The methyl
ester 7b (90 mg, 0.33 mmol) dissolved in MeOH (5 mL) was
treated with 0.2 M aqueous NaOH (1.8 mL, 1.1 equiv) and
stirred at room temperature for 4 days. The solution was
diluted with water and carefully acidified with dil HCl to pH
≈5 to precipitate the product. After being stirred for an
additional 2 h, the mixture was cooled and filtered, and the
white precipitate was washed with cold water. Drying and
recrystallization from MeOH furnished a white flocculent solid
(32 mg, 38%): mp 185-186 °C dec; 1H NMR (300 MHz, CDCl3)
δ 20.16 (s, 1H), 13.35 (s, 1H), 9.48 (br d, J ) 9 Hz, 1H), 8.16
(m, 1H), 8.09 (m, 2H), 7.68-7.53 (m, 4H); IR (KBr) see Table
2; MS m/z (rel intensity) 258 (M+, 1).
3-p-Nitr oben zoyla m in op icolin ic Acid N-Oxid e (8c).
This product was prepared from the ester in the same way as
7c with a reaction time of 6 days. Recrystallization from MeOH
furnished a fine, white powder (23%): mp 276-279 °C; 1H
NMR (300 MHz, CDCl3) δ 13.62 (s, 1H), 9.44 (br d, J ) 9 Hz,
1H), 8.39 (d, J ) 9 Hz, 2H), 8.24 (d, J ) 9 Hz, 2H), 8.22 (br d,
J ) 6.6 Hz, 1H), 7.68 (dd, J ) 9, 6.6 Hz, 1H); IR (KBr) νmax
3106, 1695, 1671 cm-1; MS m/z (rel intensity) 258 (MH+ - CO2,
25).
1
136 °C; H NMR (300 MHz, CDCl3) δ 11.98 (s, 1H), 9.35 (dd,
J ) 8.5, 1.2 Hz, 1H), 8.47 (dd, J ) 4.5, 1.2 Hz, 1H), 8.06 (dd,
J ) 8, 1.5 Hz, 2H), 7.57 (m, 4H), 4.10 (s, 3H); IR (KBr) see
Table 2; MS m/z (rel intensity) 256 (M+, 28).
Meth yl 3-p-Nitr oben zoyla m in op icolin a te (8a ). This
compound was prepared in a fashion similar to that of 7a ,
employing p-nitrobenzoyl chloride. Reflux for 4 h, workup, and
recrystallization from cyclohexane/EtOAc provided fine, off-
white crystals (46%): mp 201-204 °C; 1H NMR (300 MHz,
CDCl3) δ 12.20 (s, 1H), 9.30 (dd, J ) 8.7, 1.5 Hz, 1H), 8.52
(dd, J ) 4.5, 1.5 Hz, 1H), 8.40 (dd, J ) 7, 2 Hz, 2H), 8.21 (d,
(24) Perrin, D. D.; Armarego, W. L. F., Perrin, D. R. Purification of
Laboratory Chemicals, 2nd ed.; Pergamon Press: Oxford, 1980.

382 J . Org. Chem., Vol. 67, No. 2, 2002
Barros et al.
3-Dod eca n oyla m in op icolin ic Acid N-Oxid e (9c). This
product was prepared from the ester in the same way as 7c
with a reaction time of 8 h. Recrystallization from MeOH/water
Ack n ow led gm en t. Financial support was obtained
from CNPq (Conselho Nacional de Pesquisa e Desen-
volvimento, Brazilian Agency). We are indebted to the
analytical laboratories of Novartis S.A., Basel, Switzer-
land, for obtaining IR spectra. I.d.C.L.B., C.C.C.B., and
F.H. thank the Brazilian Agencies CAPES and CNPq
for scholarships. We are indebted to Professor Katia
Zaccur Leal/Universidade Federal Fluminense for help
with the NMR studies.
1
gave a white powder (66%): mp 103-104 °C; H NMR (300
MHz, CDCl3) δ 19.98 (br s, 1H), 12.34 (br s, 1H), 9.28 (dd, J )
9, 1 Hz, 1H), 8.10 (dd, J ) 6.3, 1 Hz, 1H), 7.55 (dd, J ) 9, 6.3
Hz, 1H), 2.50 (t, J ) 7.5 Hz, 2H), 1.74 (m, 2H), 1.42-1.20 (m,
16H), 0.88 (t, J ) 6.6 Hz, 3H); 13C NMR (75 MHz, CDCl3) δ
193.8, 173.6, 165.0, 142.9, 131.4, 127.7, 124.1, 123.3, 38.7, 31.9,
29.6, 29.4, 29.3, 29.2, 29.1, 25.1, 22.7, 14.1; IR (KBr) see Table
2; MS m/z (rel intensity) 336 (M+, 1).
3-Dod eca n oyla m in op icolin ic Acid (9d ). This product
was prepared from the ester (9a ) in the same way as 7c (from
7b) with a reaction time of 34 h. Recrystallization from MeOH/
water gave the free acid 9d as a white microcrystals (88%):
mp 134-135 °C; 1H NMR (300 MHz, CDCl3) δ 11.13 (br s, 1H),
9.27 (br d, J ) 8.7 Hz, 1H), 8.31 (m, 1H), 7.60 (dd, J ) 8.7, 4.5
Hz, 1H), 2.49 (t, J ) 7 Hz, 2H), 1.76 (m, 2H), 1.46-1.16 (m,
16H), 0.88 (t, J ) 6.6 Hz, 3H); 13C NMR (75 MHz, CDCl3) δ
173.4, 166.0, 139.8, 139.5, 130.7, 130.2, 129.3, 38.5, 31.9, 29.6,
29.4, 29.3, 29.1, 25.2, 22.7, 14.1; IR (KBr) see Table 2; MS m/z
(rel intensity) 320 (M+, 18).
Su p p or tin g In for m a tion Ava ila ble: Discussion of AM1,
RHF/6-31G, and RHF/6-31G** molecular modeling results for
10a , 10b, 10c-tr a n s, 10c-cis, 10d -tr a n s, and 10d -cis and full
molecular coordinate tables for 10c-tr a n s and 10c-cis. RHF/
6-31G results for 3-benzoylaminopicolinic acid N-oxide (7c-
cis). Full experimental characterization as well as 1H NMR
spectra for compounds 8b, 8c, and 9b. This material is
available free of charge via the Internet at http://pubs.acs.org.
J O000993J