Photochemical Reaction Mechanisms of 2-Nitrobenzyl Compounds: Methyl Ethers and Caged ATP

Article abstract of DOI:10.1021/ja039071z

The mechanism of methanol photorelease from 2-nitrobenzyl methyl ether (1) and 1-(2-nitrophenyl)ethyl methyl ether (2), and of ATP release from adenosine-5′-triphosphate-[P³-(1-(2-nitrophenyl)-ethyl)] ester ('caged ATP', 3) was studied in various solvents by laser flash photolysis with UV-vis and IR detection. In addition to the well-known primary aci-nitro transients (A, λ_max ≈ 400 nm), two further intermediates preceding the release of methanol, namely the corresponding 1,3-dihydrobenz[c]isoxazol-1-ol derivatives (B) and 2-nitrosobenzyl hemiacetals (C), were identified. The dependencies of the reaction rates of A-C on pH and buffer concentrations in aqueous solution were studied in detail. Substantial revision of previously proposed reaction mechanisms for substrate release from 2-nitrobenzyl protecting groups is required: (a) A novel reaction pathway of the aci-tautomers A prevailing in buffered aqueous solutions, e.g., phosphate buffer with pH 7, was found. (b) The cyclic intermediates B were identified for the first time as the products formed by the decay of the aci-tautomers A in solution. A recently proposed reaction pathway bypassing intermediates B (Corrie et al. J. Am. Chem. Soc., 2003, 125, 8546-8554) is shown not to be operative. (c) Hemiacetals C limit the release rate of both 1 (pH < 8) and 2 (pH < 10). This observation is in contrast to a recent claim for related 2-nitrobenzyl methyl ethers (Corrie et al.). Our findings are important for potential applications of the 2-nitrobenzyl protecting group in the determination of physiological response times to bioagents ('caged compounds').

Full text of DOI:10.1021/ja039071z

Published on Web 03/20/2004
Photochemical Reaction Mechanisms of 2-Nitrobenzyl
Compounds: Methyl Ethers and Caged ATP
†
‡
Yuri V. Il’ichev, Markus A. Schw o¨ rer, and Jakob Wirz*
Contribution from the Departement Chemie der UniVersit a¨ t Basel, Klingelbergstr. 80,
CH-4056 Basel, Switzerland
Received October 15, 2003; E-mail: J.Wirz@unibas.ch
Abstract: The mechanism of methanol photorelease from 2-nitrobenzyl methyl ether (1) and 1-(2-
3
nitrophenyl)ethyl methyl ether (2), and of ATP release from adenosine-5′-triphosphate-[P -(1-(2-nitrophenyl)-
ethyl)] ester (‘caged ATP’, 3) was studied in various solvents by laser flash photolysis with UV-vis and IR
detection. In addition to the well-known primary aci-nitro transients (A, λmax ≈ 400 nm), two further
intermediates preceding the release of methanol, namely the corresponding 1,3-dihydrobenz[c]isoxazol-
1-ol derivatives (B) and 2-nitrosobenzyl hemiacetals (C), were identified. The dependencies of the reaction
rates of A-C on pH and buffer concentrations in aqueous solution were studied in detail. Substantial revision
of previously proposed reaction mechanisms for substrate release from 2-nitrobenzyl protecting groups is
required: (a) A novel reaction pathway of the aci-tautomers A prevailing in buffered aqueous solutions,
e.g., phosphate buffer with pH 7, was found. (b) The cyclic intermediates B were identified for the first time
as the products formed by the decay of the aci-tautomers A in solution. A recently proposed reaction pathway
bypassing intermediates B (Corrie et al. J. Am. Chem. Soc., 2003, 125, 8546-8554) is shown not to be
operative. (c) Hemiacetals C limit the release rate of both 1 (pH < 8) and 2 (pH < 10). This observation
is in contrast to a recent claim for related 2-nitrobenzyl methyl ethers (Corrie et al.). Our findings are important
for potential applications of the 2-nitrobenzyl protecting group in the determination of physiological response
times to bioagents (‘caged compounds’).
Introduction
Here, we report a study of the model compounds 2-nitroben-
zyl methyl ether (1) and 1-(2-nitrophenyl)ethyl methyl ether (2)
by picosecond pump-probe spectroscopy, nanosecond laser
flash photolysis (LFP), and time-resolved infrared (TRIR)
spectroscopy. Two additional intermediates that are formed from
the primary aci-nitro photoproducts were identified and the rate
of methanol photorelease from nitrobenzyl-protected substrates
was found to be orders of magnitude slower than that for the
decay of the primary aci-nitro transients in aqueous solution at
The 2-nitrobenzyl functionality is the most widely used
1
photoremovable protecting group for application in synthesis,
2
photolithography (DNA microarrays), and biochemistry (“caged
3
compounds”). Physiological response times to bioagents such
as neurotransmitters may be determined, if the release of a
signaling molecule following pulsed excitation of the protected
4
precursor is faster than the response time under investigation.
The primary photoreaction of 2-nitrobenzyl compounds is
intramolecular H-atom transfer affording aci-nitro tautomers that
are readily detected by their strong absorption around 400 nm.
Their decay is often taken to indicate the release rate of the
protected agents. However, the subsequent reactions leading to
release of the protected leaving groups are not well understood.
5
pH ≈ 7. ‘Caged ATP’, the disodium salt of adenosine-5′-
3
triphosphate-[P -(1-(2-nitrophenyl)ethyl)] ester (3) was also
briefly investigated. A related study on derivatives of 1 and 2
6
7
was published after the completion of our work, and will be
referred to in the discussion. A well-known problem with the
use of 2-nitrobenzyl and 2-nitrophenethyl protecting groups in
biological preparations is that the release of the desired bioactive
compounds is accompanied by the formation of 2-nitrosoben-
zaldehyde or 2-nitrosoacetophenone, respectively, which may
inactivate the biological preparations or give rise to further,
undesired but harmful photoreactions. Secondary photoreactions
of the nitroso products were not investigated here. In a previous
publication, we have reinvestigated the largely reversible
†
Present address: Wichita State University, Department of Chemistry,
1
845 Fairmount St., Wichita, KS 67260-0051.
‡
Present address: MDL Information Systems GmbH, Theodor-Heuss-
Allee 108, D-60486 Frankfurt/Main, Germany.
(
1) Pillai, V. N. R. Synthesis 1980, 1-26. Greene, T. W.; Wuts, P. G. M.
ProtectiVe Groups in Organic Synthesis, 3rd ed.; Wiley-Interscience: New
York, 1999.
2) Chee, M.; Yang. R.; Hubbell, E.; Berno, A.; Huang, X. C.; Stern, D.;
Winkler, J.; Lockhart, D. J.; Morris, M. S.; Fodor, S. P. A. Science 1996,
(
2
74, 610-614. Pirrung, M. C. Chem. ReV. 1997, 97, 473-488. Pirrung,
M. C. Angew. Chem., Int. Ed. 2002, 41, 1276-1289.
(5) Kaplan, J. H.; Forbush, B.; Hoffman, J. F. Biochemistry 1978, 17, 1929-
1935.
(6) Corrie, J. E. T.; Barth, A.; Munasinghe, V. R. N.; Trentham, D. R.; Hutter,
M. C. J. Am. Chem. Soc. 2003, 125, 8546-8554.
(7) Schw o¨ rer, M. ‘Mechanismen der lichtinduzierten Freisetzung von Ab-
gangsgruppen aus 2-Nitrobenzylverbindungen’, Ph.D. thesis, University of
Basel, 2002, and ref 3.
(
(
3) Pelliccioli, A. P.; Wirz, J. Photochem. Photobiol. Sci. 2002, 1, 441-458.
4) Breitinger, H.-G. A.; Wieboldt, R.; Ramesh, D.; Carpenter, B. K.; Hess,
G. P. Biochemistry 2000, 39, 5500-5508. Givens, R. S.; Weber, J. F. W.;
Conrad, P. G.; Orosz, G.; Donahue, S. L.; Thayer, S. A. J. Am. Chem.
Soc. 2000, 123, 2687-2697. Pollock, J.; Crawford, J. H.; Wootton, J. F.;
Corrie, J. E. T.; Scott, R. H. Neurosci. Lett. 2003, 338, 143-146.
10.1021/ja039071z CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 4581-4595
9
4581

A R T I C L E S
Il’ichev et al.
Scheme 1. Phototautomerization of 2-Nitrotoluene
man-Harris apodization function, Mertz phase correction, zero-filling
factor of 4) gave a set of IR difference spectra (post-flash minus pre-
flash absorbance) at varying delay times, which was subjected to factor
analysis. A few significant factors (usually 2) were generally sufficient
to reproduce the time evolution of the spectra within experimental
accuracy. This procedure resulted in a substantial reduction of noise.
Mono- or biexponential decay functions, as required, were fitted to
the reduced set of spectra so obtained. The temperature in the sample
compartment was 30 °C during operation. Fast-scan spectra for
transients with a lifetime exceeding 100 ms were recorded on standing
solutions of 100-500 µm path length after exposing the sample to one
or several laser flashes.
phototautomerization of parent 2-nitrotoluene (Scheme 1) to
identify the elementary steps of the thermal back reaction in
8
aqueous solution as a benchmark. In a forthcoming paper, we
9
will discuss the photoreactions of 2-nitrobenzyl alcohols.
Materials. Doubly distilled water was used to prepare aqueous
solutions. All other solvents were of spectroscopic grade and were used
as received. The absorbance of the neat solvents at 248 nm was less
than 0.01 at 1-cm path length. Ethers 1 and 2 were synthesized by
methylation of the corresponding alcohols with dimethyl sulfate using
phase transfer catalysis.1
Molecular Probes.
3-15
Caged ATP (3) was purchased from
Experimental Section
Methods. Reaction quantum yields were determined by spectro-
photometric monitoring of the absorbance changes. A medium-pressure
mercury lamp equipped with a 365-nm band-pass filter was used as a
light source. Actinometry was done with a solution of azobenzene in
-Nitrosobenzaldehyde (4):16 A solution of anthranil (3 g, 25.2
mmol) in aqueous HCl (37 g, 23%) was cooled to -20 °C. Powdered
NaNO (1.75 g, 25.3 mmol) was added in small portions. The resulting
yellow-brown slurry was stirred for 20 min at -20 °C and filtered.
The solid was washed with cold aqueous HCl (6 mL, 23%) and water,
and redissolved in 10 mL of 2 M NaOH at 0 °C. Insoluble material
was removed by filtration and the filtrate was added to a cold solution
2
2
1
0
methanol.
The nanosecond kinetic and spectroscopic laser flash photolysis setup
was of standard design. An excimer laser operated mostly on KrF (248
nm), but also on XeCl (308 nm) or XeF (351 nm), was used as an
excitation source with a pulse duration of about 25 ns and pulse energies
of <200 mJ. Neither degassing nor changing the excitation wavelength
had a noticeable effect on the transient intermediates. Therefore, most
measurements were done with aerated solutions. All kinetic measure-
ments with aqueous solutions were done at ambient temperatures (23
of 4 mL of concentrated H
The resulting brown suspension was stirred at 0 °C for 3 h and filtered.
The brown solid was washed with 5 mL of aqueous Na CO (5%),
copiously with cold water, and dissolved in CHCl (20 mL). Addition
2 4
SO in 75 mL water with vigorous stirring.
2
3
3
of 10 mL petrol ether, bp 60-70 °C, precipitated brown flakes, which
were removed by filtration. The green filtrate was shaken with 2%
(
2 °C) and at ionic strength I ) 0.1 M, which was adjusted by addition
of NaClO . Consequently, all the equilibrium constants so determined
are molar concentration quotients, K , for I ) 0.1 M. Transients with
3
aqueous NaHCO , washed, evaporated to 5 mL and slowly cooled to
4
-
40 °C. 2-Nitrosobenzaldehyde (0.35 g, 2.52 mmol, 10.3% yield)
16
1
c
crystallized as bushy needles, mp 110 (113-113.5) °C. H NMR (300
lifetimes exceeding 1 ms were measured on a conventional kinetic flash
apparatus with a thermostated sample cell (25.0 ( 0.1 °C) using a
discharge flashlamp (e1000 J) for excitation.
The picosecond pump-probe setup has been described.11 Solutions
with an absorbance of 0.5 (1 mm path length) were circulated in a
flow system, pumped at 248 nm (4 mJ, 0.7 ps half-width, 10 Hz) and
probed by a delayed continuum pulse (310-700 nm) of the same
duration. Transient absorption spectra were recorded at ca. 50 different
time delays ranging from 2 ps to 1.8 ns relative to the excitation pulse
and subjected to factor analysis.
The step-scan instrument was built by the design of R o¨ dig and
Siebert12 using a Bruker IFS 66v/s Fourier transform infrared spectro-
photometer equipped with a globar IR source, a KBr beam splitter, a
nitrogen-cooled MCT detector (KV100-1-B-7/190) and an external DC-
coupled preamplifier from Kolmar Inc. (KA020-E6/MU/B). A Quantel
Brilliant W Nd:YAG laser (266 or 355 nm, pulse width 6 ns, pulse
energy e5 mJ, pulse frequency 10 Hz) was used for excitation and a
Germanium filter was positioned behind the cell to absorb the laser
beam. The spectral window was restricted to 1000-2000 cm-1 by using
a cut-on/cutoff filter combination from LOT Oriel. Solutions with an
MHz, CDCl
3
/TMS): δ ) 6.44 (dd, J
1
) 8, J
) 8, J
2
) 0.8 Hz, 1H), 7.68 (dt,
J
J
1
) 8, J
) 8, J
2
) 1.3 Hz, 1H), 7.91 (dt, J
1
2
) 0.8 Hz, 1H), 8.21 (dd,
13
1
2
) 1.3 Hz, 1H), 12.1 (s, 1H, CHO). C NMR (75 MHz,
3
CDCl ): δ ) 193.3, 162.0, 136.5, 134.0, 132.7, 127.7, 106.5 ppm.
MS (EI) m/e (relative intensity): 50 (44), 51 (98), 52 (34), 62 (6), 63
(
(
[
(
18), 64 (28), 74 (22), 75 (14), 76 (27), 77 (100), 78 (9), 79 (43), 91
50), 90 (8), 104 (3), 135 (61). UV/Vis (H O, 2%THF), λmax/nm (log
2
-1
-1
ꢀ/M cm ]) 238 (4.05), 287 (3.78 ( 0.02), 310 (3.73 ( 0.03), 750
1.34 ( 0.04).
Calculations. Density Functional Theory (DFT) calculations were
17
performed with the GAUSSIAN 98 package of programs. We used
the B3LYP hybrid functional, which combines Becke’s three-parameter
18
19
exchange functional with the Lee-Yang-Parr correlation functional
20
in a slightly modified form. Geometries were optimized using the
B3LYP functional with the 6-31G(d) and 6-31+G(d) basis sets for
(
13) Seebach, D.; Kalinowski, H.-O.; Bastani, B.; Crass, G.; Daum, H.; D o¨ rr,
H.; DuPreez, N. P.; Ehrig, V.; Langer, W.; N u¨ ssler, C.; Oei, H.-A.; Schmidt,
M. HelV. Chim. Acta 1977, 60, 301-325.
(
14) Merz, A. Angew. Chem. 1973, 85, 868-869.
(15) Gonzalez, M. A.; Meyers, A. I. Tetrahedron Lett. 1989, 30, 47-55.
(
(
16) Bamberger, E.; Fodor, A. Ber. Dtsch. Chem. Ges. 1910, 43, 3321-3335.
17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Rega, N.; Salvador,
P.; Dannenberg, J. J.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B.
B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara,
A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M.
W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.;. Replogle, E. S.; Pople,
J. A. Gaussian 98, Revision A.11, Gaussian, Inc., Pittsburgh, PA, 2002.
optical density of 0.4 were pumped through the sample cell (CaF
2
windows, 100-µm path length) to avoid re-irradiation of transient
intermediates and photoproducts. The Fourier transformed interfero-
-
1
grams (128 accumulations, spectral resolution 8 cm , 3-term Black-
(
8) Schw o¨ rer, M.; Wirz, J. HelV. Chim. Acta 2001, 84, 1441-1458.
9) Il’ichev, Y. V.; Kombarova, S. V.; Mac, M.; Schw o¨ rer, M.; Wirz, J. in
preparation.
(
(
10) Gauglitz, G.; Hubig, S. Z. Phys. Chem., N. F. 1984, 139, 237-246. Gauglitz,
G.; Hubig, S. J. Photochem. 1985, 30, 121-125. Persy, G.; Wirz, J. EPA
Newslett. 1987, 29, 45-46.
(11) Hasler, E.; H o¨ rmann, A.; Persy, G.; Platsch, H.; Wirz, J. J. Am. Chem.
Soc. 1993, 115, 5400-5409.
(18) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(19) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-797.
(12) R o¨ dig, C.; Siebert F. Appl. Spectrosc. 1999, 55, 893-901.
4582 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
A R T I C L E S
Figure 2. Transient absorption spectrum observed by picosecond pump-
probe spectroscopy of 1 in acetonitrile.
Scheme 2: Intermediates Observed in the Photoreaction of
Figure 1. Absorption spectra of 2-nitrobenzyl methyl ether (1, s),
2-Nitrobenzyl Methyl Ether (1)
2
-nitrosobenzaldehyde (4, ‚‚‚‚) and of the primary photoproduct of 1 (- -)
-
3
in 1 × 10 M aqueous HClO4 as determined by EFA.
neutral and anionic species, respectively. B3LYP/6-311+G(2d,p) single-
point energies include zero-point vibrational corrections that were scaled
2
1
with a factor of 0.9806. This level of theory gave satisfactory results
22
in a related study of 2-nitrotoluene. Calculated vibrational frequencies
were scaled with a factor of 0.9613 for comparison with IR spectra.21
Self-consistent reaction field (SCRF) calculations with the polarized
continuum model (PCM) of Tomasi and co-workers23 were utilized to
estimate solvent effects. The scaling factor for solvent accessible
surfaces and the initial number of tesserae on the surface of each sphere
were set to the default values of 1.2 and 60, respectively. Electronic
excitation energies were calculated using the random phase approxima-
Quantum yields for the conversion of 1 to 4 (φ ) 0.49 (
0
5
)
(
.05, aqueous acetate buffer 1:1, 3 measurements) and of 2 to
-
2-
(φ ) 0.48 ( 0.05, aqueous phosphate buffer H2PO4 :HPO4
1:1, 3 measurements) were determined spectrophotometrically
λexc ) 365 nm, λobs ) 310 nm) using azobenzene actinometry
and assuming clean reactions of 1 to 4 and of 2 to 5 at low
conversions. The extinction coefficients of 4 in water were
determined with an authentic sample (Experimental Section).
24
tion for a time-dependent DFT calculation with the B3LYP/6-311+G-
2d,p) basis set.
(
-1
-1
Those of 2-nitrosoacetophenone (5), λmax/nm (log [ꢀ/M cm ])
87 (3.78), 311 (3.82), 750 (1.3), were estimated from the
Results
Photoproducts and Quantum Yields. Irradiation of the
methyl ethers 1 and 2 at 254 nm (low-pressure mercury arc) in
acetonitrile or in aqueous solution at pH 3-6 generates
2
essentially clean conversion of 2 to 5. In the dark, the nitroso
compounds 4 and 5 were stable for hours in the pH-range of
-6, but decomposed at higher or lower acidities. The rate
3
coefficients for the acid- and base-catalyzed decays of 4 were
2
-nitrosobenzaldehyde (4) and 2-nitrosoacetophenone (5), re-
-
1
-1
5
-1 -1
found to be kH ≈ 20 M
s
and kOH ≈ 5 × 10 M s . The
spectively. The reaction produces strong absorption in the range
of 280-350 nm as well as very weak absorption at 750 nm.
Due to secondary photolysis of the photoproduct 4, 2-nitrobenzyl
methyl ether (1) could not be converted quantitatively. A series
of UV spectra obtained by monitoring the reaction progress
during irradiation of 1 was analyzed by the method of “Evolving
Factor Analysis (EFA)”.25 The method does not rely on a kinetic
reaction model. Concentration changes are obtained from the
weight of the linearly independent spectral components, which
are determined iteratively for increasing irradiation doses. The
result is shown in Figure 1. The spectrum of the first photo-
product, as estimated by EFA, is close to that of an authentic
sample of 2-nitrosobenzaldehyde (4). The photoproduct was
further identified by GC-MS analysis of the irradiated mixture
and comparison with an authentic sample of 4. On the other
hand, irradiation of 2 induced essentially clean and complete
conversion to 5.
decomposition products were not investigated. The equilibrium
with nitroso dimers lies far on the side of the monomers at the
-
4
26
low concentrations (<1 × 10 M) used in these experiments.
Picosecond Pump-Probe Spectroscopy. Broad transient
absorption (λmax ) 413 nm, Figure 2) appeared within 5 ps after
excitation of 1 with 248-nm subpicosecond pulses in acetonitrile
and persisted up to the maximum delay of 1.8 ns. Similar results
(
λmax ) 415 nm) were obtained with 2. The transient absorption
is attributed to the aci-nitro tautomers formed by intramolecular
hydrogen transfer in the singlet excited state. Only very weak
absorption was observed at wavelengths > 550 nm, and no
resolved growth of the aci-absorption due to a slower reaction
via the triplet state of 1 or 2 could be detected. Such observations
had been reported for related nitrobenzyl compounds.27
Nanosecond Laser Flash Photolysis (LFP) of 1. Five
kinetically distinguishable transient intermediates were detected
by nanosecond LFP of 1 in aqueous solution and in other
solvents. For the following discussion the transient intermediates
are labeled as A, B, and C (Scheme 2). Stereoisomers of A and
B exhibiting different reaction kinetics are distinguished as A1
(
20) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch M. J. J. Phys.
Chem. 1994, 98, 11 623-11 627.
(
(
(
21) Scott, A. P., Radom, L. J. Phys. Chem. 1996, 100, 16 502-16 513.
22) Il’ichev, Y. V.; Wirz, J. J. Phys. Chem. A 2000, 104, 7856-7870.
23) (a) Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55, 117-129.
(b) Miertus, S.; Tomasi, J. Chem. Phys. 1982, 65, 239-245. (c) Cossi, M.;
(26) Azoulay, M.; Wettermark, G. Tetrahedron 1978, 34, 2591-2596.
(27) Yip, R. W.; Sharma, D. K.; Giasson, R.; Gravel, D. J. Phys. Chem. 1984,
88, 5770-5772. Yip, R. W.; Sharma, D. K.; Giasson, R.; Gravel, D. J.
Phys. Chem. 1985, 89, 5328-5330. Yip, R. W.; Sharma, D. K.; Giasson,
R.; Gravel, D.; Wen, Y. X. J. Phys. Chem. 1991, 95, 6078-6081. Yip, R.
W.; Sharma, D. K.; Giasson, R.; Gravel, D.; Blanchet, D. Can. J. Chem.
1991, 69, 1193-1200.
Barone, V.; Cammi, R.; Tomasi, J. Chem. Phys. Lett. 1996, 255, 327-
35.
24) Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys.
998, 108, 4439-4449.
25) Gampp, H.; Maeder, M.; Meyer, C. J.; Zuberb u¨ hler, A. D. Talanta 1986,
3, 943-951.
3
(
(
1
3
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4583

A R T I C L E S
Il’ichev et al.
Figure 3. Kinetic traces obtained by LFP of 1 at 30 °C (for comparison
with the FTIR measurements). Trace 1: Acetonitrile. Trace 2: Acetonitrile
with 75 mM D2SO4. Trace 3: CH2Cl2.
Figure 4. pH-Rate profiles of reactions ab1 (b), ab2 (O), bc1 (+), bc2
(
×), and c4 (0) initiated by LFP of 1. The solid lines were obtained by
and A2 and as B1 and B2. The identification of these transients
30
nonlinear least-squares fitting of eq 1 to the data points (Table S1) for
ab1 and ab2 and of eq 3 to those for bc1, bc2, and c4. The resulting
parameters are given in Table 1.
-
as the aci-tautomers, A, or their anions, A , 1,3-dihydrobenz-
[c]isoxazol-1-ol intermediates, B, and hemiacetal, C, rests on
the TRIR data, absorption spectra, pH-rate profiles, and
observations of general acid and base catalysis reported below.
Reaction A1 f B1 is labeled as ab1, A2 f B2 as ab2, and so
on. The same symbols A-C will be used to designate the
corresponding intermediates formed from 2 and 3.
Rate constants determined with buffer solutions (pH 3.5-
0.5) were found to increase linearly with buffer concentration
1
indicating general acid and/or base catalysis. Data points for
the pH-rate profile were, therefore, obtained by linear extrapola-
tion of buffer dilution plots (vide infra) to zero buffer concentra-
tion. The pH-rate profiles for reactions ab1 and ab2 are shown
in Figure 4. The data points are given in Table S1 of the
Nonaqueous Solvents. Excitation of 1 in acetonitrile by a
nanosecond laser pulse (308 nm) produced transient A within
the duration of the laser pulse. The spectrum of A was identical
to that seen in the pump-probe experiment (Figure 2), and its
decay obeyed a biexponential rate law, with rate constants kab1
3
0
Supporting Information.
The rates of reactions ab1 and ab2 are constant at pH > 10,
and rise steadily with increasing acid concentration at pH <
10. The slope of log kobs for reaction ab2 clearly drops below
-1 in the pH range of 4-6. The solid lines were obtained by
nonlinear least-squares fitting of the parameters kab′′, kab′, and
kab/Ka,1 in eq 1 to the observed first-order rate constants of
reactions ab1 and ab2. Equation 1 will be derived in the
Discussion. The acidity constants Ka,2 of A1 and A2 were more
accurately defined by the buffer dilution plots (vide infra) and
3
2
-1
28
≈
6 × 10 and kab2 ≈ 9 × 10 s (trace 1 in Figure 3).
Reactions ab1 and ab2 were strongly accelerated in the presence
of a mineral acid. They merged to a monoexponential decay,
7
-1 29
kab ≈ 3 × 10 s , in the presence of 0.02 M H2SO4. The
decays were also, but less strongly, accelerated by acetic acid
5
4
-1 28
(
1.7 M), kab1 ≈ 2 × 10 and kab2 ≈ 5 × 10 s . For
comparison with the TRIR experiments discussed below, the
growth of absorbance at 310-330 nm (formation of C, vide
infra) was measured by LFP of 1 in CH3CN that was acidified
-
5
were thus fixed to the values Ka,2(A1) ) 3.2 × 10 M and
3
-1
-4
with 75 mM D2SO4, kbc ) 7.7 × 10 s (trace 2). LFP of 1 in
Ka,2(A2) ) 1.0 × 10 M for fitting of eq 1.
CH2Cl2 showed biexponential growth of absorption at 310 nm
-
1
-1
+
+ 2
(
formation of C), kbc1 ≈ 1.3 and kbc2 ≈ 0.3 s , and was
log (kobs/s ) ) log{(k ′′K + k ′[H ] + k [H ] /K )/
ab
a,2
ab
ab
a,1
-3 -1
followed by a slow decay of C to 4, kc4 ) 9 × 10
s
(trace
-
1
+
(
M s )} - log{([H ] + K )/M} (1)
a,2
3
).
pH-Rate Profiles in Aqueous Solution. The absorption
In pure water with no buffer added, a small time-resolved
increase in absorbance was observed following the sudden rise
in absorption at 420 nm. This is attributed to ionization of the
spectra of the aci-intermediates A observed immediately after
LFP of 1 in aqueous solutions were similar to the pump-probe
spectra obtained with 1 in acetonitrile (Figure 2), but the
absorption maximum was shifted to slightly longer wavelength,
λmax ) 420 nm. The decay of this transient absorption obeyed
-
aci-nitro transients A to the conjugate anions A , k-H ) (2.7
6
-1
(
0.3) × 10 s (ca 10% of the total amplitude), which appear
to show stronger absorption at 420 nm. This rate constant is
a biexponential rate law at pH > 5. The amplitudes of the two
components were approximately equal.28 The decay of the short-
nearly an order of magnitude lower than that of the aci-tautomer
8
of 2-nitrotoluene, indicating that the acidity constants of the
lived component was accompanied by a 15-nm blue shift of
the absorption maximum (measured in phosphate buffer, pH
.8). This indicates that transient A is a mixture of two
aci-tautomers A, Ka,2 ) k-H/ kH, are about an order magnitude
smaller than that of the latter, pKa,2 ) 3.6. Combination of k-H
with a typical rate constant for the reaction of protons with a
6
kinetically distinct species, A1 and A2.
1
0
-1 -1
weak oxygen base, kH ≈ 5 × 10
M
s , provides a first
(
28) The decay traces clearly deviated from a simple first-order rate law and
gave excellent fits with two exponentials, but the two close-lying rate
constants and their amplitudes were not defined with high accuracy.
29) This value is close to the time resolution of our instrument that was limited
by the 30-ns duration of the laser pulse.
estimate of the acidity constant of the aci-tautomers A, pKa,2 ≈
(
(30) Supporting Information: See paragraph at the end of this paper regarding
availability.
4584 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
A R T I C L E S
Figure 5. Kinetic traces obtained by LFP of 1 in aqueous solutions. Trace
-
4
1
: ionization of A in 5 × 10 M NaOH. Trace 2: decay of A1 and A2 in
phosphate buffer (0.0143 M KH2PO4, 0.0286 M Na2HPO4, pH 7.08). Trace
: reaction bc1 and bc2 in acetate buffer (0.0072M HAc, 0.00144 M NaAc,
pH 3.87). Trace 4: reaction c4 in 0.1 M HClO4.
Figure 6. UV-vis difference absorbance spectra monitoring reaction c4
in aqueous solution. The first spectrum was recorded about 10 s after LFP
3
-4
of 1 in 5 × 10 M HClO4.
f B1 f C and A2 f B2 f C is expected to obey a four-
exponential rate law.
4
.3.31 Ionization of the aci-tautomers A was accelerated by the
addition of NaOH or buffers. Measurements in the range of 5
-
4
-3
×
10 (Figure 5, trace 1) to 1 × 10 M NaOH gave a rate
-kabt
) - k (1 - e-kbct)
kbc(1 - e
ab
constant of kOH ) (8 ( 2) × 10 M s-1 for the deprotonation
9
-1
[C] ) [A]
(2)
t
t)0
kbc - k
ab
-
of A to A by hydroxyl ions. A very weak transient absorption
that was observed at 500 nm decayed with a rate constant of
However, eq 2 rapidly approaches a single-exponential rate law,
6
-1
ca. 5 × 10 s , independent of pH (up to 1 M NaOH) and
oxygen (up to 1 atm). We cannot offer a convincing assignment
for this transient.
-
kt
[C]t ≈ [A]t)0(1 - e ), as the two rate constants kab and kbc
become different in magnitude, with the observable rate constant
k becoming that of the slower (rate limiting) step. A reliable
determination of four rate constants was not possible from the
absorbance growth waveforms monitored at 330 nm. Growth
curves determined in the pH range of 5.5-6.5, where kab ≈ kbc
and a four-exponential fit would have been required, were
disregarded in the analysis. Traces obtained outside that range
were adequately fitted by a sum of two exponentials (Figure 5,
trace 3). The two rate constants for the formation of C that were
determined at 330 nm in solutions of pH > 6.5 are limited by
and equal to the rate constants of the reactions ab1 and ab2,
kab1 and kab2, whereas those at pH < 5.5 are slower and provide
the rate constants of reactions bc, kbc1, and kbc2.
The decay of transient A was clearly biexponential at pH >
(trace 2 in Figure 5). With increasing acidity the decay rate
5
of the slower component A2 increased more rapidly than that
of A1, such that the two decays merged to an essentially single
5
-1
exponential with a rate constant of about 6 × 10 s at pH 3
(cf., however, the buffer dilution plots discussed below). The
absorption due to transients A1 and A2 decayed to the baseline
in aqueous solutions with pH < 6, leaving no residual absorption
above 300 nm. Subsequently, a slow, biexponential growth in
absorbance was observed at 310-330 nm (trace 3 in Figure 5:
formation of intermediate C). These observations require
intervention of some intermediates, B1 and B2, which are
transparent above 300 nm. The pH-dependence of the two first-
order rate constants for the formation of C from B1 and B2,
determined at 330 nm, is shown in Figure 4 (reactions bc1 and
bc2). Formation of C is also accompanied by the growth of a
very weak absorbance at 740 nm, which is characteristic for
the formation of nitroso compounds.3 However, the measure-
ments at 330 nm gave more accurate results, because the
absorbance changes associated with the reactions B f C are
much larger at this wavelength. The decay of transients B is
catalyzed by base. At pH > 6 the decay rates of B1 and B2
exceed the rates of formation of these intermediates, so that
the decay rates of A1 and A2 coincide with the rates of
formation of C.
Finally, the lifetime of product C is around 40 s in the pH-
range of 3-6, and the absorbance changes arising from reaction
c4 were sufficiently slow to be recorded with a diode array UV
spectrometer. The absorbance at 320 nm due to C decreases,
while that at 230 nm increases strongly (Figure 6). A global fit
-
4
of the data obtained with 5 × 10 M aqueous HClO4 gave a
2
-
2
-1
rate constant of (2.57 ( 0.01) × 10
s
and provided UV
spectra of the initial (t ) 0) and final species. These were nearly
superimposable with the UV spectra of 2-nitrosobenzyl alcohol
(as a stable analogue for the hemiacetal C) and authentic 4,
respectively, in the same solvent. The higher rate constants kc4
outside the range of 3 < pH < 6 were determined by kinetic
flash photolysis analyzed at 320 nm.
The pH-rate profiles for reactions bc1, bc2, and c4 (Figure
In general, the time-dependent concentration of a product C
that is formed in a sequential reaction scheme of the type A f
B f C with rate constants kab and kbc is given by eq 2. Thus,
the formation of C by the two parallel reaction sequences A1
4
) indicate acid catalysis at pH < 3 and base catalysis at pH >
5
. Equation 3 describes a reaction that has a pH-independent
-1
+
+
-1
log (kobs/s ) ) log {(k [H ] + k K /[H ] + k )/s } (3)
H
OH
W
0
(
31) A similar estimate based on the ionization kinetics of 2-nitrotoluene’s aci-
nitro tautomer in pure water gave pK
≈ 3.4, in good agreement with pK
component, k0, but also exhibits catalysis by protons, kH, and
by hydroxide ions, kOH. The hydroxyl ion concentration was
a
a
8
)
3.6 determined independently.
(32) Walker, J. W.; Reid, G. P.; McCray, J. A.; Trentham, D. R. J. Am. Chem.
Soc. 1988, 110, 7170-7177.
+
-14
2
replaced by KW/[H ], where KW ) 1.59 × 10
M is the
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4585

A R T I C L E S
Il’ichev et al.
Table 1. Rate Coefficients Determined by Fitting of Eqs 1 and 3
to the pH-Rate Profiles of 1 to 3
those of A2. This led to a clear-cut separation of the two reaction
rates with increasing buffer concentration in acetic and formic
acid buffers, and proves that the near coincidence of the decay
rate constants kab1 and kab2 in wholly aqueous solutions at pH
1
ab′/s-1
(kab/Ka,1)/s-1
reaction
k
ab′′/s-
k
Ka,2/M
5
4
7
9
-4a,b
-5a,b
ab1(1) 2.1 ( 0.4
(8.7 ( 0.8) × 10
(5.0 ( 0.6) × 10 ∼1.0 × 10
9
ab2(1) 0.23 ( 0.08 (1.4 ( 0.3) × 10
(7.8 ( 1.5) × 10 ∼3.2 × 10
<
5 is fortuitous.
9
ab1(2)
(5.1 ( 0.2) × 10 ∼7 × 10
2
6
10
-4b
ab2(2) (9 ( 2) × 10 (1.1 ( 0.2) × 10
(1.8 ( 0.8) × 10
1 × 10
Solvent Kinetic Isotope Effects. The decay rates of A1 and
4
9
-5b
ab(3)
1.5 ( 0.5
(7 ( 2) × 10
∼5 × 10
3.2 × 10
A2 were determined in H2O and D2O solutions with equal
H
D
/M-1 s-1
/s-1
OH/M-1 s-1
k
nominal concentration of added acid or base: k /k ) 0.97 (
reaction
k
H
k
0
-
3
0
.03 in 1.1 × 10 M aqueous perchloric acid (single expo-
H D
4
2
10
bc1(1)
bc2(1)
c4(1)
bc(2)
c5(2)
bc(3)
(2.6 ( 0.4) × 10
(3.6 ( 0.3) × 10
23 ( 4
(1.8 ( 0.2) × 10
13 ( 2
∼1 × 10
4
10
6
nential), and k /k ) 1.08 ( 0.18 (ab1) and 0.84 ( 0.18 (ab2)
in 0.1 M sodium hydroxide. Relatively large error limits in the
latter values arise because of the uncertainty associated with
fitting decay traces with a sum of two close-lying exponentials.
(1.1 ( 0.2) × 10
(1.2 ( 0.2) × 10
-
3
(1.6 ( 0.3) × 10
3
2
10
∼1 × 10
(4.0 ( 2.7) × 10
∼1 × 10
2
7
-2
6
(2.3 ( 0.9) × 10
(2.0 ( 0.2) × 10
∼1 × 10
∼1.3 × 10
3
(7.7 ( 0.9) × 10
TRIR-Measurements of 1. The IR-spectrum of 1 in CD3-
a
Value taken from a fit of buffer slopes to eq 5. b Value was kept
CN is dominated by strong bands arising from the symmetric
constant in fitting.
-
1
-1
(
1343 cm ) and antisymmetric (1526 cm ) stretching vibra-
tions of the NO2 group. Medium intensity bands are seen in
-
1
-1
the range 2800-3000 cm (C-H stretch) and at 1109 cm
antisymmetric C-O-C stretch). The corresponding frequencies
(
-
1
predicted by the DFT calculations (1329 and 1353 cm for
the symmetric NO2 stretch, 1551 and 1564 cm for the
antisymmetric NO2 stretch, 1109 and 1111 cm for the
C-OMe stretch in the anti- and syn-conformations of 1,
respectively) are in satisfactory agreement with the experimental
data.
-
1
-
1
CD3CN and CD2Cl2 were used as solvents for the TRIR
experiments. Measurements with the same (nondeuterated)
solvents were done by optical LFP for comparison. Reaction A
f B in CD3CN was followed by the step-scan method with a
-
3
9
× 10 M solution of 1. The kinetic analysis of TRIR spectral
data is less precise than that of kinetic traces obtained by optical
LFP and the TRIR data for reactions ab and bc were adequately
fitted with a single exponential function. Hence, only a single
IR spectrum is available for A (representing a mixture of A1
and A2), and one for B.
Figure 7. Buffer dilution plot for the decay rate constants of transients
A1 (b) and A2 (O) in acetic acid buffer 1:1 (pH 4.57).
ionization constant of water at ionic strength I ) 0.1 M.33
Nonlinear least-squares fitting of eq 3 to the observed rate
constants for reactions bc1, bc2, and c4 provided the parameters
k0, kH, and kOH for each reaction, which are given in Table 1.
The coincidence of the reaction rates for ab1 and bc1 and for
ab2 and bc2 at pH > 6 (Figure 4) proves that intermediate A1
is the precursor of B1, and A2 of B2. Consistently, the amplitude
ratio of reactions ab1 and ab2 in the kinetic traces measured at
Reaction A f B. The spectral matrix covering a time range
from 2.5 to 500 µs after flash photolysis (Figure 8) was subjected
to factor analysis. Two components were sufficient to reconstruct
the spectra within experimental accuracy. Prominent features
are the decaying bands in the region of 1100-1400 and 1540-
-1
-1
1
650 cm . A band at 1080 cm grows in simultaneously, but
otherwise no strong absorption bands, particularly no NdO or
CdO bands, are present in the last spectra. The strong negative
bands in the difference spectra did not show a significant change
in position or intensity during the time span covered. The
positions of these negative bands correspond to those of the
absorption bands of 1 indicating depletion of the starting material
4
3
20 nm was about equal to that of reactions bc1 and bc2 at
20 nm.
Buffer Dilution Plots. Measurements in aqueous solutions
of pH 3.5-10.5 require the addition of buffers, which, however,
may influence the reaction rates by general acid and general
base catalysis. These measurements were made in series of
solutions of varying buffer concentration but constant buffer
ratio and, hence, constant pH. Catalysis by buffers was found
for all reactions and buffer dilution plots at constant buffer ratio
were linear (e.g., Figure 7). The intercepts represent the decay
rates at zero buffer concentration, which were used for the pH-
rate profiles (Figure 4). The buffer catalytic coefficients, kbu,
which correspond to the slopes of the buffer dilution plots, are
given in Table 2. As shown in Figure 7, the slopes kbu
determined for the decay of A1 were about 10-fold higher than
1
.
Intermediates A. The data matrix represented in Figure 8
was adequately fitted by a single-exponential rate law. The
3
-1
resulting rate constant kab ) (6.6 ( 0.2) × 10 s is in good
3
agreement with that for the major component, kab1 ≈ 6 × 10
-1 28
s , of the biexponential decay observed by LFP of 1 in
acetonitrile. The negative bands due to the depletion of 1 were
removed by adding an appropriate amount of the IR spectrum
of 1 to the first spectrum of Figure 8 to obtain the absorption
spectrum of transients A, Figure 9. It is in qualitative agreement
with the much more highly resolved spectrum that was reported
3
4
by Dunkin et al. for the same product formed by irradiation
(
33) Bates, R. G. Determination of pH, Theory and Practice; Wiley: New York,
1
973.
of 1 in Ar and N2 matrixes at 12 K.
4586 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
A R T I C L E S
Table 2. Buffer Slopes kBu Obtained with 1 by Linear Regression of Buffer Dilution Plotsa
bu/Mb
k
bu /(M s-1)
ab1
-1
k
bu /(M s-1)
ab2
-
k
bu /(M s-1)
bc1
-1
k
bu /(M s-1)
bc2
-1
k
bu /(M s-1)
c4
-1
buffer
x
HB
c
7
c
3
2
phthalic
formic
formic
formic
formic
acetic
acetic
acetic
acetic
acetic
phosphate
phosphate
phosphate
phosphate
phosphate
tris
1/2
4/5
2/3
1/3
1/5
5/6
2/3
1/2
1/3
1/9
6/7
3/4
2/3
1/2
1/3
1/2
1/2
0.01-0.05
0.05-0.50
0.03-0.30
0.015-0.15
0.0125-0.125
0.011-0.60
0.02-0.12
0.02-0.2
(2.70 ( 0.45) × 10
c
c
(7.1 ( 0.7) × 10
(4.5 ( 0.6) × 10
6
7
6
7
c
(4.8 ( 0.3) × 10
(6.2 ( 1.9) × 10
(3.1 ( 0.3) × 10
(1.3 ( 0.1) × 10
6
6
6
(5.0 ( 0.3) × 10
(1.9 ( 0.2) × 10
(1.4 ( 0.1) × 10
7
6
5
4
4
(1.18 ( 0.03) × 10
(1.16 ( 0.14) × 10
(2.9 ( 0.2) × 10
(7.2 ( 0.3) × 10
143 ( 9
6
5
(7.6 ( 1.1) × 10
(4.5 ( 1.2) × 10
261 ( 18
6
5
(3.5 ( 0.1) × 10
(1.46 ( 0.13) × 10
(1.10 ( 0.05) × 10
287 ( 17
0.09 ( 0.03
6
3
5
0.015-0.15
0.01-0.11
0.008-0.078
(1.1 ( 0.1) × 10
<6 × 10
(1.5 ( 0.1) × 10
385 ( 19
4
4
5
(5.8 ( 1.0) × 10
<6 × 10
(1.1 ( 0.1) × 10
431 ( 46
3
4
3
(8.6 ( 3.8) × 10
(1.9 ( 0.1) × 10
d
(5.9 ( 0.5) × 10
1.5 ( 0.4
<1
3
3
3
3
3
2
0.015-0.060
0.005-0.050
0.011-0.043
0.02-0.2
(1.5 ( 1.0) × 10
(3.6 ( 0.2) × 10
<3 × 10
(3.4 ( 0.2) × 10
(1.4 ( 0.2) × 10
(4.5 ( 0.8) × 10
3
3
3
<7 × 10
(1.4 ( 0.2) × 10
<5 × 10
3
2
3
<1 × 10
(3.2 ( 0.1) × 10
<2 × 10
<50
<50
6 ( 3
<5
55 ( 14
borate
0.01-0.05
a
Experimental data are given in Table S430. b Total buffer concentration, [HB] + [B ]. The decay rates of A1 and A2 were too close to be separated.
-
c
d
The buffer dilution plot was nonlinear, because reaction ab1 becomes rate-limiting at high buffer concentrations (Figure 4).
Figure 9, is hardly warranted, but the E,E-isomer appears to be
present in significant amount, because strong absorption in the
-1
range 1350-1400 cm was predicted only for this isomer. The
-
1
three bands in the 1550-1650 cm region suggest that other
isomers are also formed. The biexponential decay kinetics
observed for transient A (Figures 3 and 5, cf. Discussion) indeed
indicates formation of at least two geometrical isomers of A as
primary photoproducts in solution, whereas the rigid environ-
ment in an argon matrix appears to favor formation of a single
isomer. From the qualitative agreement of the IR-spectrum of
A in acetonitrile with the matrix spectrum reported by Dunkin
et al. and with the calculations, it is clear that the nitronic acids
A do not ionize in acetonitrile. The IR spectra calculated for
-
30
the two aci-anions A (Table S6) are substantially different.
Methyl vinylethers commonly show a strong asymmetric
C-O-C stretching band around 1200 cm . According to our
Figure 8. TRIR difference spectra of 1 in CD3CN measured by the step-
scan technique. Time after excitation increases from back to front.
-
1
DFT calculations the strongest feature in the IR spectra of the
-
1
E,E- and E,Z-isomers of aci-1 (1230-1245 cm ) corresponds
-
1
to the C-OMe stretch. Thus the band observed at 1248 cm
is attributed to this mode. The next strong band in the spectrum
can be assigned to the NOH bending mode, which is predicted
-
1
to be located in the 1270-1293 cm region. The band at 1211
-
1
cm was attributed to the NO stretching vibration which is
strongly coupled to ring deformations according to the DFT
results. The calculated frequencies for the N-OH stretch vary
-
1
from 928 to 952 cm (outside the range covered). The strong
-
1
band at 1144 cm seems to arise from the ring CH bending
vibrations and CH3 deformation modes. Our assignment of the
35
CdC and CdN bands differs from that of Dunkin et al. We
-
1
attribute the 1620-cm band to the CdN group, the band at
-1
-1
1
550 cm to the exocyclic CdC(O) group, and the 1580-cm
band to skeletal double bond stretching.
Intermediates B. Transients B do not absorb in the near UV.
In time-resolved optical measurements their presence was
inferred only from the delayed formation of product C. The IR
spectrum of B (Figure 10) was obtained by addition of the
Figure 9. IR spectrum of intermediates A (both upper and lower)
determined by adding an appropriate amount of the IR spectrum of 1 to
the initial TRIR difference spectra of Figure 8. Vertical bars indicate IR
frequencies and intensities predicted by DFT calculations for the four
isomers of aci-1 (Table S5). The first label (E or Z) denotes the
configuration of the exocyclic CdC bond, the second that of the CdN bond.
3
0
(
34) Dunkin, I. R.; Gebicki, J.; Kiszka, M.; Sanin-Leira, D. J. Chem. Soc., Perkin
Trans. 2 2001, 1414-1425. Dunkin, I. R.; Gebicki, J.; Kiszka, M.; Sanin-
Leira, D. Spectrochim. Acta, Part A 1997, 53A, 2553-2557.
Dunkin et al.34 attributed their matrix spectrum to the E,E-
isomer of aci-1. Spectra for all four possible isomers of aci-1
were calculated with DFT techniques (Table S5).30 An identi-
-1
(35) Three bands are observed in the range of 1500-1650 cm (1550, 1580,
1620 cm- ). The band at 1550 cm would be unusually low for either a
1
-1
CdC or a CdN stretching band. Additional conjugation tends to lower
the position of CdN bands less than that of CdC bands (ref 37). Our DFT
1
fication of structural isomers based on the solution spectrum,
calculations predict the CdN stretching vibrations near 1620 cm-
.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4587

A R T I C L E S
Il’ichev et al.
Figure 10. IR-Spectrum of intermediates B determined by adding an
appropriate amount of the IR spectrum of 1 to the final TRIR difference
spectra of Figure 8. Vertical bars indicate IR frequencies and intensities
predicted by DFT calculations for the two isomers cis-6 (red, dotted) and
trans-6 (blue).
Figure 12. Fast-scan IR difference spectra of 1 in CD2Cl2.
Figure 13. IR-spectrum of intermediate C in aqueous solution determined
by adding an appropriate amount of the IR spectrum of 1 to the initial
TRIR difference spectra of Figure 12. IR frequencies and intensities
predicted by DFT calculations for the hemiacetal 7 are shown as vertical
bars.
Figure 11. TRIR difference spectra of 1 in CD3CN with 75 mM D2SO4.
Time evolution is from front to back.
Cl2. Figure 12 shows the resolved growth of an intense CdO
-
1
band at 1700 cm . The NdO band is fully developed already
in the first spectrum, and its intensity hardly changes during
the reaction, but its position gradually shifts from 1495 to 1501
spectrum of depleted A to the last spectrum of Figure 8. It shows
-
1
only one prominent band at 1080 cm . Indeed, DFT-calcula-
tions for the two isomers of 3-methoxy-1,3-dihydrobenz[c]-
isoxazol-1-ol (6) predict only one strong band (O-CH3 stretch)
-1
cm as the CdO band develops. Global analysis of the spectral
matrix gave two significant components and their temporal
evolution was fitted well by a first-order rate law. The resulting
-
1
located around 1100 cm . The two other features in the
-
1
-2 -1
predicted IR spectra around 1180 and 1370 cm also coincide
with observed bands. Dunkin et al.34 report that irradiation of
aci-1 in an argon matrix yields a product with weak bands at
rate constant, kc4 ) 1.3 × 10 s , is close to that observed by
LFP in the same solvent, k ) 9 × 10-3
s
-1
(Figure 3, trace
c4
3).
-
1
1
637, 1091, and 756 cm , which they attributed to the
Removal of the negative peaks that are due to depletion of 1
from the first spectrum of Figure 12 provided the IR spectrum
of C (Figure 13), which is shown together with the calculated
frequencies for the conformer with an H-bond between the OH
group and the oxygen atom of the NO group. This conformer
formation of 6. The most important aspect in the spectrum of
B is the lack of any strong bands in the range of 1400-1700
-
1
cm that would indicate the presence of a CdO or an NdO
group. Thus, our assignment of B to the cyclic intermediate 6
rests mainly on the subsequent evolution of the TRIR spectra
described below.
is expected to be the primary product produced by ring-opening
22,36
of cis-6.
A similar experimental spectrum of C (not shown)
TRIR spectroscopy of 1 in CD3CN containing 75 mM D2-
SO4 showed partially resolved growth of an NdO band at 1500
cm (Figure 11), but still no absorption indicating the formation
of a carbonyl group. Hence, reaction B f C is attributed to
ring opening of the bicyclic compound 6 yielding the hemiacetal
methoxy-(2-nitrosophenyl)-methanol (7). The rate constant, kbc
was obtained from the last spectrum of Figure 11.
The IR spectrum of the final photoproduct 4 (Figure 14) was
obtained by correction for depleted 1 in the last spectrum of
Figure 12. Comparison with the spectrum of authentic 2-ni-
trosobenzaldehyde (4) unambiguously identifies the end product
as 4. The band positions and intensities calculated for 4 (Table
-1
4
-1
)
(1.1 ( 0.1) × 10 s , is in reasonable agreement with that
obtained by LFP in the same medium, kbc ) 7.7 × 10 s
Figure 3, trace 2).
Reaction C f 4. The final reaction c4 was followed by fast-
scan IR spectroscopy following 266-nm excitation of 1 in CD2-
30
S6) are in satisfactory agreement with experiment.
3
-1
Further support for the assignment of intermediate C to the
hemiacetal 7 stems from the IR spectral changes observed in
(
(36) Il’ichev, Y. V. J. Phys. Chem. A 2003, 107, 10 159-10 170.
4588 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
A R T I C L E S
Figure 14. IR-spectrum of photoproduct of 1 (final spectrum of Figure
1
2, blue dotted line) compared to a spectrum of authentic 2-nitrosobenzal-
dehyde (4) in CD2Cl2 (solid red line). IR frequencies and intensities predicted
by DFT calculations for 4 are shown as vertical bars.
Figure 16. pH-Rate profiles of reactions ab1 (b), ab2 (O), bc1 (+), bc2
(
×), and c5 (0) initiated by LFP of 2. The data points are given in Table
30
S2. The solid lines were obtained by nonlinear least-squares fitting of eq
for reactions ab1 and ab2 and of eq 3 for reactions bc1, bc2, and c5.
1
The resulting fit parameters are given in Table 1.
formation of C occurs in two waves, the slower one being
limited by the decay of the precursor A2 in basic solutions.
The rate of the faster component, bc1, increased linearly with
increasing base concentration up to about pH 12, where its
amplitude vanished and that of the slow component, bc2, was
doubled. The slopes of buffer dilution plots are given in Table
Figure 15. Left-hand graph: IR spectra of C (- - -, red) and of the final
-
1
products 4 and methanol (-O-, black) in the range of 2500-3800 cm
7
-1
obtained by TRIR. Right-hand graph: IR spectra of methanol (- - -, red)
and dilute methanol in CD2Cl2 (solid black line).
3. The intrinsic decay rate constant of A1, kab1 ) 5.1 × 10 s
is too fast to be affected by the acidic buffers. The decay of A2
exhibits general acid catalysis, the formation of C general base
catalysis (Table 3).
the OH-stretch region during the conversion of C to 4 in CD2-
Cl2. As shown in the left-hand part of Figure 15, the broad OH
Caged ATP (3). LFP (351 nm, pulse energy 100 mJ) of 3
-
1
band of C ranging from 3200 to 3600 cm changes to a
10-4 M) was done in aqueous solutions at various pH. Buffer
(
-
1
substantially narrower band at 3500 cm in the last spectrum.
The broadness of the OH band of C is attributed to intramo-
lecular H-bonding in the hemiacetal 7 and its narrowing in the
course of the reaction C f 4 to the release of methanol. Only
quite low concentrations of methanol are released in the IR
experiment, so that it will not form H-bonded methanol clusters.
The OH stretch bands of H-bonded alcohols are generally broad
concentrations were kept sufficiently high to damp any effect
of the disodium salt 3 on pH, but as low as possible to avoid
buffer catalysis. The rate constants for the first-order decay of
intermediate A (aci-3) were measured at 420 nm. The formation
of the 2-nitrosophenyl chromophore (C or 5) was monitored at
30
3
20 nm (Figure 17, Table S3). It was delayed with respect to
the aci-decay at pH < 6, indicating the intervention of a bicyclic
intermediate B that does not absorb above 300 nm. The acidity
3
7
and lie at lower frequencies than those of the free species. To
illustrate the band shift and broadening resulting from the
association of alcohols, the spectrum of a dilute solution of
methanol in CD2Cl2 is compared to a spectrum of neat methanol
in the right-hand part of Figure 15.
constant of A, Ka,1, was not adequately defined by the rate data
30
(
Table S3), and was therefore fixed at pKa,1 ) 4.5. Nonlinear
least-squares fitting of eqs (1) and (3) to the rate data for
reactions ab and bc, respectively, gave the fitting parameters
listed in Table 1.
LFP of 2. The pH-rate profiles for the transients A, B, and
C produced by LFP of 2 in aqueous solution are shown in Figure
1
6. The decays of the aci-transients A are biexponential with
Discussion
roughly equal amplitudes of the two components at 420 nm.
Only one of the aci-decays, A2, is pH-dependent. The faster,
pH-independent component A1 absorbs at longer wavelengths;
its amplitude dominates over that of A2 at λobs > 420 nm. The
decay rate of A2 approaches a constant value around pH 4, but
re-accelerates below pH 3, so that it escapes detection by
nanosecond LFP at pH < 2.6. The decay rates for A2 were
fitted with eq 1 and the resulting parameters are given in Table
We begin with a brief review of previous mechanistic studies
of the disodium salt of adenosine-5′-triphosphate-[P -(1-(2-
3
nitrophenyl)ethyl)] ester (3, ′caged ATP′), which was introduced
5
by Kaplan et al. The first detailed study of the reaction
32
mechanism of 3 was reported by Walker et al. The decay rates
of the aci-nitro intermediate were monitored by its absorption
at 400 nm, and appearance rates were determined for each of
the three products formed by LFP of 3, namely ATP (bioassay),
1
. The growth of product C at 320 nm was monoexponential at
+
2
-nitrosoacetophenone (absorption at 740 nm), and H (indicator
pH < 6, exhibiting both acid and base catalysis. The parameters
resulting from a fit with eq 3 are given in Table 1. At pH > 8,
dye). Two steps were kinetically resolved in aqueous solution
at neutral pH. A fast step indicated release of a proton due to
-
-1
ionization of aci-3 to 3 . In a slower reaction, k ) 86 s (pH
7.1, 21 °C), the free nucleotide and the side product 2-nitroso-
(
37) Bellamy, L. J. The Infrared Spectra of Complex Molecules, 3rd ed.;
Chapman and Hall: London, 1975.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4589

A R T I C L E S
Il’ichev et al.
Table 3. Buffer Slopes kBu Obtained with 2 by Linear Regression of Buffer Dilution Plotsa
bu/Mb
k
bu /(M s-1)
ab2
-1
k
bu /(M s-1)
bc2
-1
k
bu /(M s-1)
c4
-1
buffer
x
HB
c
8
7
4
4
4
acetic
acetic
acetic
phosphate
phosphate
phosphate
phosphate
tris
2/3
1/2
1/3
6/7
3/4
1/2
1/4
1/2
1/2
0.01-0.2
0.02-0.2
(3.0 ( 0.4) × 10
(2.2 ( 0.3) × 10
(3.6 ( 0.1) × 10
(5.2 ( 0.4) × 10
(6.9 ( 0.2) × 10
0.09 ( 0.03
0.3 ( 0.2
8
0.01-0.15
0.01-0.08
0.01-0.07
0.005-0.05
0.03-0.04
0.02-0.2
(1.19 ( 0.03) × 10
6
6
(9.1 ( 0.8) × 10
(1.34 ( 0.02) × 10
6
6
(7.7 ( 0.7) × 10
(2.3 ( 0.2) × 10
6
6
(4.4 ( 0.1) × 10
(3.1 ( 0.2) × 10
6
6
(2.4 ( 0.1) × 10
(2.5 ( 0.3) × 10
4
4
(7.0 ( 0.3) × 10
(8.4 ( 0.5) × 10
4
4
borate
0.005-0.05
(4 ( 1) × 10
(4.4 ( 0.8) × 10
a
Experimental data are given in Table S4.30
b
Total buffer concentration, [HB] + [B-].
authors, including ourselves,41 have implicitly or explicitly
assumed this mechanism to hold in general, and the aci-decay
rates were taken to indicate substrate release rates.
Several groups have questioned the validity of Scheme 3 in
recent years. Ab initio and density functional theory calculations
for the potential energy surface of 2-nitrotoluene and its isomers
predicted a much higher activation barrier for cyclization of
22
the aci-anion than for the neutral aci-tautomer. These calcula-
tions also predicted cyclization of the neutral aci-tautomer to
be highly exothermic, i.e., irreversible, whereas that of the anion
was calculated to be endothermic. Similar conclusions were
independently reached on the basis of semiempirical AM1
calculations for several 2-nitrobenzyl derivatives.42 This sug-
gested that the obserVed acid catalysis arises from a rapid
-
Figure 17. pH-Rate profiles for the dark reactions initiated by LFP of 3.
Scheme 3: Reaction Mechanism Proposed32 for Caged ATP (3)
preequilibrium of 3 with its conjugate acid aci-3, whereby the
concentration of the reactiVe species aci-3 increases with
increasing acidity. Experimental support for this hypothesis
43
comes from a recent paper by Walbert et al., who studied 2-(2-
nitrophenyl)-ethyl cages. From the dependence of the product
distribution on pH they concluded that cyclization of their
compounds proceeds only from the neutral aci-tautomer.
The present study deals with 2-nitrobenzyl (1) and 1-(2-
nitrophenyl)ethyl methyl ether (2), which are transformed to
2
-nitrosobenzaldehyde (4) and 2-nitrosoacetophenone (5), re-
spectively, releasing methanol with quantum yields of about
5
0%. A brief reinvestigation of caged ATP (3) follows.
TRIR Measurements with 1. Three transient intermediates,
A-C (Scheme 2), were detected by time-resolved infrared
spectroscopy (TRIR) of 1 in aprotic solvents. The first transient,
A, was formed within a µs after excitation of 1 in acetonitrile
acetophenone (5) were formed concomitantly with the decay
-
of 3 . An important conclusion of this work was that the release
(Figure 8) and is attributed to aci-1. The characteristic absorption
-
rate of ATP coincides with the decay of the aci-anion 3 under
bands due the aci-nitro group (1100-1400 and 1540-1650
the conditions of the experiment. This finding was since
-1
cm , Figure 9) disappeared within 1 ms (reaction ab) leaving
38
corroborated by time-resolved IR and by photoacoustic
spectroscopy.39 The reaction mechanism shown in Scheme 3
was proposed to explain the results.
-1
a single prominent IR band at 1080 cm , which is attributed
to the cyclic intermediate 1,3-dihydrobenz[c]isoxazol-1-ol (B).
The observed spectrum of B agrees well with theoretical
predictions, but its assignment rests mainly on the fact that the
bands characteristic for the nitroso and carbonyl group appear
only in the subsequent reactions. Ring opening of B yields the
-
Cyclization of 3 was assumed to be fast but reversible,
followed by rate-determining elimination of ATP. Acid catalysis
-
on the observed decay rate of 3 was attributed to protonation
of the triphosphate substituent, thereby accelerating its elimina-
tion from the cyclic intermediate. Walker et al.32 carefully
declared their hypothesis as a minimal scheme to explain their
results on caged ATP (3), and in a later review the same authors
have expressed doubt about its generality.40 Nevertheless, many
hemiacetal C (reaction bc, Figure 11). The IR spectrum of C
-1
(
Figure 13) shows the presence of a nitroso group (1500 cm ),
but carbonyl absorption is still lacking. Finally, formation of
(
40) Corrie, J. E. T.; Trentham, D. R. Bioorg. Photochem. Vol. 2, Morrison, H.,
Ed., 1993, 243-305.
(
38) Barth, A.; Hauser, K.; M a¨ ntele, W.; Corrie, J. E. T.; Trentham, D. R. J.
Am. Chem. Soc. 1995, 117, 10 311-10 316. Barth, A.; Corrie, J. E. T.;
Gradwell, M. J.; Maeda, Y.; M a¨ ntele, W.; Meier, T.; Trentham, D. R. J.
Am. Chem. Soc. 1997, 119, 4149-4159. Rammelsberg, R.; Boulas, S.;
Chorongiewski, H.; Gerwert, K. Vibr. Spectrosc. 1999, 19, 143-149.
39) Choi, J.; Terazima, M. Photochem. Photobiol. Sci. 2003, 2, 767-773.
(41) Peng, L.; Nachon, F.; Wirz, J.; Goeldner, M. Angew. Chem., Int. Ed. Engl.
1998, 37, 2691-2693.
(42) Schaper, K.; Dommaschke, D.; Globisch, S.; Madani-Mobarekeh, S. A. J.
Inf. Rec. 2000, 25, 339-354.
(43) Walbert, S.; Pfleiderer, W.; Steiner, U. E. HelV. Chim. Acta 2001, 84, 1601-
1611.
(
4590 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
-nitrosobenzaldehyde (4), which exhibits strong NdO (1501
cm ) and CdO (1700 cm ) bands, was observed on a time
scale of seconds in CD2Cl2 (reaction c4, Figure 12). The release
of methanol in the course of the last reaction is demonstrated
by a substantial narrowing of the OH-stretch absorption (Figure
A R T I C L E S
2
Scheme 4: Revised Mechanism for the Thermal Reactions of the
Primary Photochemical aci-Transients A Formed from 1 in
Aqueous Solution.a
-
1
-1
1
5). The pH- and buffer dependencies of reactions ab, bc, and
c4 in aqueous solution were studied in detail by optical LFP.
For comparison with the TRIR measurements, LFP of 1 was
also done in aprotic solvents and acetonitrile-water mixtures to
establish that the same sequence of reactions occurs in protic
and aprotic solvents.
A Revised Mechanism for the Light-Induced Reactions
2
-Nitrobenzyl Ethers in Aqueous Solution. The neutral aci-
8
tautomer of 2-nitrotoluene has a pKa of 3.6. We expected that
of aci-1 to be similar, because the delocalization of charge from
the methoxy substituent (+M) should be largely counteracted
by its inductive effect (-I). Indeed, the rate of ionization of
the aci-transients A, k-H ) (2.7 ( 0.3) × 10 s , indicated
pKa(A) ≈ 4. The IR-spectra of A show that these weak acids
do not ionize in acetonitrile, suggesting pKa g 3. In aqueous
solution at physiological pH the aci-transients will be predomi-
6
-1
-
a Only one isomer is shown for intermediates A and B.
nantly converted to their ionized form A , and the observed
acid catalysis may be attributed to an increase in the equilibrium
concentration of the more reactive neutral species A, as
discussed above. Assuming a pKa(A) of about 4, one would
then expect acid catalysis to saturate at pH < 4. However, the
decay rates of the two aci-nitro transients A1 and A2 formed
from 1 do not reach a constant value at pH < 4 but increase
further down to pH 2 (Figure 4), where they escape detection
by nanosecond LFP. Moreover, the addition of acid also
accelerated the decay rate of transients A in aprotic solvents,
To derive the rate law for the decay of intermediates A from
Scheme 4, we assume that the acid-base equilibria between
+
-
A , A, and A are fully established within the lifetime of the
+
aci-intermediates and that Ka,1 . [H ] for pH g 1, i.e., that A
is quantitatively protonated only at pH values well below the
4
5
range investigated. With these assumptions, analysis of
Scheme 4 leads to eq 4 for the pH- and buffer-dependent first-
order decay rate constant kobs of the aci-transients A. In the
-
absence of buffers, [B ] ) 0 M, eq 4 simplifies to eq 1, which
-
where ionization to A does not occur.
was used to fit the pH-rate profile of the aci-transients of 1
We conclude that the lifetime of the aci-intermediates is
shortened by an additional, acid-catalyzed reaction. In fact, the
slope of log kobs for reaction ab2(1) approaches a value of -2
between pH 4-6 in the pH-rate profile, Figure 4, indicating
that two protonation steps must be involved in the decay of
(
Figure 4).
k ′′K + k ′[H ] + (k + k [B ])[H ] /K
a,1
+
-
+ 2
ab
a,2
ab
ab
B
kobs
)
+
[H ] + K
a,2
-
(4)
A2 in that pH region. On the other hand, an onset of saturation
in acid catalysis is discernible from a short plateau at pH 4-5
in the pH-rate profile of transient A2 formed from 2 (Figure
Geometrical Isomerism of the Transient Intermediates A
and B Formed from 1. The decay of the primary aci-transient
A of 1 (λmax ) 420 nm) is clearly biphasic at pH > 5, in
1
6), but its decay rate starts to rise again below pH 4. An acid-
catalyzed reaction for transients A is introduced in the revised
mechanism shown for 1 in Scheme 4. Here, the aci-nitro
intermediate A is considered to be both an acid, pKa,2 ≈ 4, and
4
0,46,47
common with some other aci-nitro decays.
Below pH 5,
the two decay rates nearly merge, but they become clearly
separated with increasing buffer concentrations (Figure 7).
Several observations show that transient A is a mixture of two
species, A1 and A2. The two rate constants obtained from the
biexponential fits obeyed distinct pH-rate profiles and buffer
dependencies. Moreover, the decay of the short-lived component
was accompanied by a noticeable blue shift of the absorption
8
a weak base. In our previous study of parent 2-nitrotoluene
we have shown that protonation of aci-nitrotoluene leads to a
44
Nef-type reaction yielding 2-nitrosobenzyl alcohol. The pro-
+
tonated aci-cation A of 2-nitrotoluene was found to be a strong
acid, pKa,1 < -3. We may anticipate that the corresponding
+
cation A derived from 1 is substantially less acidic due to the
4
7b
maximum. Following a suggestion of Zhu et al., we attribute
stabilizing effect of the methoxy group. By analogy with
the biphasic decay of A to the presence of E- and Z-isomers
+
nitrotoluene, hydration of the protonated intermediate A is then
assumed to yield a nitroso hydrate B′. The acid-catalyzed
reaction via A accounts for the observation of acid catalysis
at pH values below pKa,2 in aqueous solution as well as for the
acceleration of the decay of neutral A in acetonitrile solution
by acids.
(
45) The preequilibrium assumption is not required for the protonation of A.
+
+
+
As long as Ka, 1 . [H ], A will always be present in low concentration,
+
and the steady-state approximation holds for A . In this case the term (kab
-
-
-
+ k [B
B
]) in eq 4 is replaced by (kab + k [B
B
])k /(k + k [B ] + k-H),
H
ab
B
H
where k and k-H are the rates of protonation of A and deprotonation of
+
A , respectively. The preequilibrium assumption clearly holds for inter-
mediates A and A- formed from 1, because reactions ab are rate-limiting
for the formation of C at pH > 6. The case of 2 will be discussed below.
46) Peng, L.; Goeldner, M. J. Org. Chem. 1996, 61, 185-191.
(
(
44) This product could not be isolated from the strongly acidic solutions, but
was tentatively identified on the basis of its transient UV spectrum and the
rate of dehydration of its nitroso hydrate precursor.
(47) (a) Schupp, H.; Wong, W. K.; Schnabel, W. J. Photochem. 1987, 36, 85-
97. (b) Zhu, Q. Q.; Schnabel, W.; Schupp, H. J. Photochem. 1987, 39,
317-332.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4591

A R T I C L E S
Il’ichev et al.
Figure 19. Buffer slopes kbu (Table 2) obtained for transients A1 (left)
and A2 (right) formed from 1 in acetic (b, red) and formic (O, blue) acid
buffers plotted as a function of buffer pH. The solid lines were obtained by
nonlinear least-squares fitting of eq 5.
do not support this, predicting longer wavelengths of absorption
for the E-isomers (Figure 18). Also, one would expect the
twisted Z-isomer to be the more reactive. However, SCRF
calculations in water predicted a smaller barrier for the cycliza-
tion of E,E-aci-1 (8.4 kcal/mol) than of Z,E-aci-1 (10.3 kcal/
36
mol). In view of these rather conflicting indications, we refrain
from attempting an assignment of transients A1 and A2 to the
structural isomers.
Catalysis by Buffers. Protonation Equilibria of Transients
A Formed from 1 in Aqueous Solution. Unambiguous
evidence for the occurrence of an acid-catalyzed path involving
+
Figure 18. Calculated structures of the four aci-isomers of 2-nitrobenzyl
protonation of A to A comes from the observation of general
-
1
methyl ether (1). The energies of the nitronic acids are given in kcal mol
relative to that of the anti-conformer of 1 (structure shown in Figure 1),
acid catalysis on the decay of A (Table 2). Buffer catalysis
affecting reactions ab1 and ab2 must be due to the acidic buffer
component, because the decay of A is not accelerated by
hydroxyl ions, the strongest possible base in aqueous solution.
Rapid equilibration of A with A, followed by rate-determining
cyclization of A to B, would give rise to specific acid catalysis
on the decay of A , because the general acid does not participate
in the rate-determining cyclization step A f B.
The additional reaction path via A introduced in Scheme 4
accounts for the effect of buffers on the decay rate of the aci-
transients. Protonation of A might be the rate-determining step
followed by rapid hydrolysis to B′ (Nu ) OH). This would
give rise to general acid catalysis, but is considered unlikely,
because even protonation of the weaker base aci-nitrotoluene
-
-
that of Z-aci-1 relative to E-aci-1 . The wavelengths (λ/nm) of the first
electronic transitions that were predicted by TD-DFT calculations are given
in brackets.
-
with regard to the dC-OCH3 group, which cyclize with
different rates.
The calculated geometries of the four isomeric aci-tautomers
of 1 and their energies relative to that of 1 are shown in Figure
-
+
1
8. Least-motion hydrogen transfer from the benzylic carbon
of 1 to the nitro group would form the E,Z- and Z,Z-isomers as
the primary photoproducts. The calculations indicate that the
E,E- and Z,E-isomers, in which the OH group of the nitronic
-
1
acid is situated trans, are several kcal mol lower in energy.
We may assume that the equilibrium between the two proto-
nation sites on the nitronic acid function is established on the
ns time scale in aqueous solution. Ionization of aci-1 occurs
8
+
is reversible. Alternatively, the cation A may be trapped
preferentially by the basic buffer component B to give B′ (Nu
-
6
-1
) B). The combination of preequilibrium protonation (specific
acid catalysis) with a rate-determining general-base-catalyzed
step results in apparent general acid catalysis.
Three buffer systems were studied at various buffer ratios,
namely formic acid/sodium formate, acetic acid/sodium acetate,
and potassium dihydrogen phosphate/disodium monohydrogen
phosphate. The slopes kbu, obtained from the buffer dilution plots
(e.g., Figure 7), correspond to the derivative of eq 4 with respect
with a rate constant of 2.7 × 10 s in water, and is accelerated
in the presence of buffers or base. Moreover, separation of the
ion pair formed upon protonation of water is not required for
proton transfer between the O atoms of the nitronic acid
function. A hydrogen-bonded water molecule may act as a
proton shuttle through a six-membered ring. Indeed, the
calculated barrier for intramolecular E to Z isomerization in the
-
1
aci-tautomer of nitrotoluene was 20.3 kcal mol in the gas
-
-
1
to the total buffer concentration, [bu] ) [HB] + [B ]. Making
phase, but was lowered to 10.1 kcal mol in the presence of
-
-
one water molecule.36 In the following discussion we will,
use of the relation [B ] ) xB[bu], where xB ) [B ]/([HB] +
-
[
B ]), we obtain eq 5.
therefore, assume that only geometrical isomerism at the
exocyclic CdC double bond, designated by the first label E or
Z, giVes rise to kinetically distinguishable aci-transients in
aqueous solution.
+
2
∂kobs
∂kobs
∂[B-] [H+]
x k [H ]
B B
k )
) xB
)
(5)
bu
(
)
(
)
+
∂[bu]
[
H+]
Ka,1(Ka,2 + [H ])
Steric congestion in the isomers with Z-configuration at the
methylene group is apparent from both the structures and the
relative energies given in Figure 18. The blue shift of the
absorption maximum associated with the decay of the short-
Plots of the dependence of kbu/xB on pH in acetic and formic
acid buffers are shown in Figure 19.
-
lived component A1 suggests that its exocyclic groups are
(
48) Quinkert, G. Wiersdorff, W.-W. Finke, M.; Opitz, K. von der Haar, F.-G.
twisted at the double bond.48 However, TD-DFT calculations
Chem. Ber. 1968, 101, 2302-2325.
4592 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
A R T I C L E S
which is not catalyzed by general base and only weakly by the
-
weak general acid H2PO4 . As we move to more acidic buffer
compositions, reaction ab2 becomes faster than bc2, and the
latter now limits the rate of formation of C. The acceleration
of reaction bc2 by the buffer at xHB ) 0.85 may be attributed
2-
to the remaining fraction of the general base HPO4 . Similarly,
the rate of reaction bc1 increases linearly in the acetate buffers
down to xHB ) 0.3, but then suddenly drops, because reaction
ab1 becomes rate-limiting.
pH-Rate Profiles of Transients A-C Formed from 1 in
Aqueous Solution (Figure 4). Inspection of Figure 4 reveals
the sequence of reactions from the primary aci-nitro transients
A to the final product 2-nitrosobenzaldehyde (4) and methanol
in aqueous solutions of various pH. The release of methanol
coincides with decay of the two isomeric aci-nitro transients
Figure 20. Buffer slopes kbu (Table 2) obtained for reactions ab2 (O) and
bc2 (b) in phosphate buffers (left-hand figure) and ab1 (O) and bc1 (b) in
acetic acid buffers (right-hand figure) plotted as a function of buffer
composition xHB. The curved lines for reactions ab were obtained by
nonlinear least-squares fitting of eq 5 to the buffer slopes for reactions ab.
+
To express eq 5 as a function of the variable xHB only, [H ] was replaced
bu
by Ka xHB/(1 - xHB). The red straight lines indicate the xHB-dependence
22,36
for reactions bc that would be expected in the absence of the rate-limitation
by reactions ab.
only at pH values > 9. Calculations
suggest that the reactions
-
-
of the aci-anions A1 and A2 proceed directly to the
hemiacetal C; the anions of the cyclic intermediates B were
found not to be stable intermediates. The hemiacetals C rapidly
eliminate methoxide in basic solutions.
+
+
Equation 5 simplifies to kbu ≈ xBkB[H ]/Ka,1 for [H ] . Ka,2,
+
2
+
and to kbu ≈ xBkB[H ] /(Ka,1Ka,2) for [H ] , Ka,2. Thus, in plots
of log(kbu/xB) versus pH, the slope should change from -2 to
Carbon protonation of A- by water to regenerate 1 is too
slow to compete with the formation of C. That reaction, which
was observed for the aci-tautomer of 2-nitrotoluene in basic
-
1 in buffer systems with pH ≈ pKa,2. That was found to be
the case and provided the most reliable estimates of pKa,2. The
values obtained for reaction ab2 were -2.23 ( 0.07 (phosphate
H
aqueous solutions, exhibits a large kinetic isotope effect, kab /
4
9
buffers, pH 6-7), -1.63 ( 0.15 (acetate buffers, pH 4-5),
and -1.14 ( 0.12 (formate buffers, pH 3-4). The pKa,2 of A2
must, therefore, be close to 4.5. The pKa,2 of A2 so obtained
agrees well with that estimated from the negative curvature
around pH 4.5 in the pH-rate profile of reaction ab2 (Figure
D
8
-
kab ) 7.2. The kinetic isotope effects for the decay of A1
-
and A2 , on the other hand, are unity within experimental error.
The decay of the hemiacetal C (compound 7) is the rate-
determining step for the release of methanol at pH < 8. At pH
values in the range of 3-6 its lifetime reaches nearly 1 min
and is 5-8 orders of magnitude longer than those of the primary
aci-intermediates A.
4
). Similarly, reaction ab1 gave slopes for log(kbu/xB) versus
pH of -1.96 ( 0.12 for acetate and -1.00 ( 0.08 for formate
the value in phosphate buffers was not defined accurately for
(
The cyclic intermediates B (compound 6) are observable only
at pH < 6 in water and in aprotic solvents. At pH > 6 they
decay more rapidly than they are formed. Assignment of reaction
bc to ring-opening of the cyclic intermediates 6 to the
reaction ab1), which indicates that the pKa,2 of A1 is about 4.0.
Acid and Base Catalysis of bc1 and bc2. Reactions bc1
and bc2 also exhibited buffer catalysis. In contrast to the rather
exceptional buffer dependence discussed above (eq 5), plots of
buffer catalytic coefficients, kbu, versus the molar fraction of
2-nitrosohemiacetal 7 is based on the TRIR measurements. Why
do we observe biexponential kinetics for reaction bc? The
-
the acidic buffer component, xHB ) [HB]/([HB] + [B ]), are
inversion barriers of amines increase strongly upon electrone-
commonly linear, eq 6. This allows separation of kbu into its
general acid, kHB (xHB ) 1), and general base, kB (xHB ) 0),
components.
5
0
gative substitution. Amines carrying two oxygen atoms reach
-
1 51
inversion barriers of up to 30 kcal mol . The nitrogen
inversion barriers for a number of isoxazolidine derivatives were
-
1 52a
found to be in the range of 14-16 kcal mol ,
1
that in
k ) k + (k - k ) x
(6)
bu
B
HB
B
HB
-1 52b
-hydroxy-2,2,4,4-tetramethylpyrrolidine is 12 kcal mol .
DFT calculations for the inversion barrier in 6 gave 16.8 kcal
Plots of the buffer coefficients obtained in acetic acid buffers
Table 2) were indeed linear in most cases, and indicated that
-
1
36
mol (Figure 21). Apparently, the flattening effect of the
conjugated aromatic ring in 6 is overcompensated by substitution
with a second oxygen atom on nitrogen. Thus, we may assume
that equilibration between the two stereoisomers of 6 is slow
compared to ring opening to the hemiacetal 7.
Ring opening of B might be concerted with elimination of
CH3OH yielding the aldehyde 4 directly. For aprotic solvents
this is ruled out by the results of the TRIR-measurements. The
intermediate C formed from B does not carry a carbonyl group.
In aqueous solutions the final reaction c4 was monitored by its
UV-spectral changes (Figure 6). The initial spectrum recorded
a few seconds after photolysis corresponds to that expected for
(
catalysis is due to the basic buffer component, kB(bc1) ) (2.3
5
-1 -1
2
-1
(
0.5) × 10 M
s
and kB(bc2) ) (5.5 ( 0.4) × 10 M
-
1
s . The intercepts at xHB ) 1 were not significantly different
from zero. The buffer slopes kbu for reaction bc2 in phosphate
buffers (Figure 20, left) and for reaction bc1 in acetate buffers
(right) changed in a strongly nonlinear fashion with xHB. These
anomalies arise from a change in the rate-limiting step of the
reactions. As can be seen from the pH-rate profile, Figure 4,
reactions bc limit the rate of formation of C at pH , 6, but
reactions ab become rate-limiting at pH . 6. Thus, in the basic
phosphate buffer compositions, xHB < 0.5, the rate-limiting step
for the slower component in the formation of C is reaction ab2,
(
50) Lehn, J. M. Top. Curr. Chem. 1970, 15, 311-377.
(
51) M u¨ ller, K.; Eschenmoser, A. HelV. Chim. Acta 1969, 52, 1823-1830.
(
49) The value somewhat below -2 may be due to a contribution of catalysis
by H PO in this buffer system. Loudon, G. M.; Ryono, D. E. J. Org. Chem.
975, 40, 3574-3577.
(52) (a) Hassan, A.; Wazeer, M. I. M.; Perzanowski, H. P.; Ali, S. A. J. Chem.
Soc., Perkin Trans. 2 1997, 411-418. (b) Perrin, C. L.; Thoburn, J. D.;
Elsheimer, S. J. Org. Chem. 1991, 56, 7034-7038.
3
4
1
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4593

A R T I C L E S
Il’ichev et al.
by R-substituents, from 20 kcal mol for the aci-tautomer of
-
1
-
1
nitrotoluene, to 14 kcal mol for that of 2-nitrophenethyl, and
-
1
to 8 kcal mol for that of 1. We have performed the same
calculations for the E,E- and E,Z-aci-isomers of 2, which gave
-
1
5
kcal mol for the more reactive E,E-isomer in water. This
low barrier, combined with an Arrhenius preexponential factor
1
3 -1
of A ) 10
s
provides an estimate of the rate of cyclization,
7
-1
kab′ ≈ 10 s , which is comparable to the rate of ionization,
7
-1
k-H ≈ 10 s . The fit of eq 1 to the observed rate constants
7
-1
gives kab′ ) 5.2 × 10 s for transient A1 formed from 2, but
5
4
-1
much smaller values, kab′ ) 8.7 × 10 and 1.4 × 10 s , for
transients for A1 and A2, respectively, formed from 1. (ii) The
formation of C(2) is biphasic at pH > 8. The rate of the slow
Figure 21. Geometries and energies (kcal mol-1) of intermediates B ) 6.
-
wave is limited by the decay of A2 , but that of the fast, major
the hemiacetal C (7), which has a lifetime of nearly one min
and limits the release rate of methanol from 1 in aqueous
solutions of pH < 8. There is no indication for a direct decay
path of B to 4.
Scheme 4 introduces an acid-catalyzed path affording nitroso
hydrates B′ via the protonated aci-nitro intermediates A+ in
strongly acidic and in buffered solutions. We cannot distinguish
between intermediates B and B′ (Nu ) OH or B) by optical
LFP. Thus, the rate constants determined for reaction bc at pH
wave continues to rise linearly with increasing base concentra-
tion until it vanishes at pH 12 (Figure 17). This proves that the
major part of intermediate B is formed directly from nascent
A1 by the fast, pH-independent reaction. (iii) Deprotonation of
A1 is accelerated by hydroxyl ions in aqueous base, kOH ∼ 1 ×
1
0
-1 -1
1
0
M
s , and this will diminish the amount of B formed
-
by the fast reaction. Deprotonation of A1 to A1 by hydroxyl
7
-1
ions becomes faster than cyclization, kab′(A1) ) 5.2 × 10 s ,
at about pH 12.5. Consistently, the fast wave in the formation
of C was still detectable at pH 12, but vanished, leaving only
the slow wave, at pH 13.
<
5 should probably be attributed to dehydration of B′ yielding
the hemiacetal C. The rate constants kH determined for these
reactions agree well with those reported for related nitroso
The pH-rate profile obtained by LFP of caged ATP (3) is
shown in Figure 17. First we note, in agreement with previous
5
3
hydrates.
The pH-rate profile of reaction c4 shown in Figure 4 exhibits
acid and base catalysis. In buffer solutions both general acid
and general base catalysis were observed. The corresponding
buffer slopes (Table 2) are comparable to those determined for
the decay of the methyl- and ethyl hemiacetals of benzalde-
hyde.5 Therefore, the release of methanol can be accelerated
slightly by the addition of buffers. The final product 2-ni-
trosobenzaldehyde (4) is sensitive to light as well as to acid
and base. In aqueous solution at room temperature, the coef-
ficients for acid and base-catalyzed decomposition are kH ≈ 20
3
2
work, that the decay of aci-3 obeys a first-order rate law
accurately. The bulky ATP substituent may favor formation of
a single aci-isomer in the primary photoreaction, or the
occurrence of a fast wave, as in 2, may have escaped observa-
tion. Otherwise, the analogy to the behavior of the aci-transients
A formed from 1 and 2 is evident. The decay rates increase
from pH 10 down to pH 2, where they exceed the time resolution
of nanosecond LFP. At pH > 6, the appearance rate of product
4,55
5
matches the decay rate of the aci-nitro tautomer A, as
3
2
established by Walker et al. HoweVer, for pH < 6 the decay
rate of A exceeds the rate of formation of 5, requiring an
additional intermediate B, which does not absorb aboVe 300
nm. Intermediate B does not delay the release of ATP at
physiological pH values, because triphosphate is a much better
leaving group than methoxide in 1 or 2. The decay of B could,
however, become rate-limiting in different media.
-
1
-1
5
-1 -1
M
s
and kOH ≈ 5 × 10 M s , respectively. Product 4
is no longer detectable at pH > 9, where its decay rate exceeds
that of its formation from 1.
pH-rate Profiles of Transients A-C Formed from 2 and
(Figures 16 and 17). The pH-rate profiles obtained for 2 show
3
many similarities with those of 1, but a qualitative difference
should be noted: The rate constant of the fast component A1
is independent of pH. Scheme 4 (adapted to 2 by adding a
methyl group to the structures) accounts for this change, if we
drop the preequilibrium assumption for the faster-decaying
transient A1. If cyclization of the primary photoproduct A1 is
Comparison with a Related Study of Derivatives of 1 and
2
. A study of several derivatives of 1 and 2, in particular the
glycolic acid ethers 8 and 9, by fast-scan FTIR and LFP was
6
published recently by Corrie et al. Their experimental data for
8
and 9 are largely in agreement with our results on 2 and 1,
-
faster than ionization to A1 , then we expect to see a
respectively, inasmuch as the conditions of measurement are
comparable. It should be noted that the IR-spectrum reported
biexponential decay of A, because the transient absorption by
A1 disappears to the transparent product B1, while A2 ionizes
6
by Corrie et al. (Figure 3) for the aci-nitro intermediate formed
-
to the longer-lived anion A2 , which still absorbs at 420 nm.
from 9 bears no relation to that of transient A determined here
Calculations and several observations strongly support the
(Figure 10). Their spectrum was obtained in aqueous solution
hypothesis that cyclization of intermediate A1 competes ef-
at pH 8.5. It is, therefore, the spectrum of the ionized aci-anion
-
36
fectiVely with ionization to A1 . (i) DFT calculations have
shown that the barrier of cyclization A f B is strongly reduced
-
A . Our spectrum was recorded in acetonitrile, where A does
not ionize. Surprisingly, Corrie et al. report first-order decay
rates for the aci-nitro transient formed from 9, while that from
1 exhibits dual exponential decay as shown here and else-
(
(
(
53) Gr u¨ nbein, W.; Fojtik, A.; Henglein, A. Monatsh. Chem. 1970, 101, 1243-
252.
54) Capon, B.; Nimmo, K.; Reid, G. L. J. Chem. Soc., Chem. Commun. 1976,
71-873.
55) McClelland, R. A.; Kangasabapathy, V. M.; Mathivanan, N. Can. J. Chem.
991, 69, 2084-2093.
1
40,56
where.
8
1
(56) Corrie, J. E. T. J. Chem. Soc., Perkin Trans. 1 1993, 2161-2166.
4594 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004

Photoreactions of 2-Nitrobenzyl Compounds
A R T I C L E S
-
1
-1
at 1497 cm , bands around 1100-1200 and 1420 cm ) as
that of the hemiacetal intermediate C formed from 1 (Figure
1
4). We surmise that the failure of Corrie et al. to detect the
hemiacetal formed from 9 may be attributed to their reaction
conditions: At pH 8.5 intermediate C decays more rapidly than
it is formed (Figure 4). We maintain that the findings of Corrie
et al. as well as our more extensive observations are fully
compatible with the single mechanistic Scheme 4 presented
above.
We concur with two important conclusions drawn by Corrie
et al. from their study of compounds 8 and 9. First, the
6
intermediate C formed from the nitrophenethyl ethers 2 and 8
is a hemiketal that is much longer-lived than the readily
observable aci-intermediates A, and determines the rate of
alcohol release under physiological conditions (pH ≈ 7). Second,
at pH > 7 the formation of the hemiketal C from 2 and 8 takes
place in two waves of substantially different rates, whereas only
a single wave is found with the nitrobenzyl ethers 1 and 9.
However, we disagree on essential points regarding the
reaction mechanism: (1) Corrie et al. claimed that the fast
formation of C from 8 proceeds directly from A, bypassing the
cyclic intermediate B (“two-path mechanism”). To explain some
difficulties with this hypothesis, they had to invoke reaction
from a vibrationally excited state of A. We have identified the
short-lived precursor A1 formed from 2 with a lifetime of about
Conclusion. Photorelease of methanol from 2-nitrobenzyl and
2
-nitrophenethyl methyl ethers (1 and 2) in wholly aqueous
solutions and in organic solvents proceeds via three intermedi-
ates, the primary aci-nitro transients A (λmax ≈ 400 nm), the
1,3-dihydrobenz[c]isoxazol-1-ol derivatives B observed here for
the first time, and 2-nitrosobenzyl hemiacetals C. A general-
acid-catalyzed reaction, which proceeds via nitroso hydrates B′,
prevails in buffered aqueous solutions. This process was not
previously taken into account. At pH values above 10, where
-
the lifetime of transients A becomes independent of pH, the
release of methanol occurs in a single step. The decay of
intermediate C limits the release rate of methanol up to pH 8
and 10 for 1 and 2, respectively. An intermediate of type B
limits the release rate of ATP from ‘caged ATP’ (3) at pH values
e 6. The rates of the reaction paths shown in Scheme 4 may
vary strongly with the nature of the reaction medium (solvent,
pH, and buffer concentration) and of the leaving group. The
lifetimes of the secondary intermediates B or B′ and C may
exceed that of the primary aci-nitro transients A by orders of
magnitude. Therefore, the easily observable decay rates of
transients A are not, in general, a good indicator for the release
rate of the leaving group.
2
0 ns, which is much too long for a vibrationally excited
molecule, and have shown that it does react via a cyclic
intermediate B1. (2) Buffer catalysis accelerating the decay of
-
the ionized aci-anions A , which was noted by Corrie et al.
and defined here as general acid catalysis, is not consistent with
any of the previously proposed mechanisms. An additional,
general-acid catalyzed reaction must be operating. (3) Corrie
et al. explicitly excluded reaction of 9 via a hemiacetal C, and
concluded that the cyclic intermediate B in this case decays
directly to the end product 2-nitrosobenzaldehyde (4). However,
intermediate B formed from 1 does react via a long-lived
hemiacetal C, which limits the release rate of methanol in
aqueous solution up to pH 8.57 The IR spectra of the hemiketal
Acknowledgment. This work is part of Project No. 20-
6
8087.02 of the Swiss National Science Foundation.
6
intermediates reported by Corrie et al. exhibit much the same
Supporting Information Available: Rate data for the pH-
prominent features (absence of a carbonyl band, nitroso band
rate profiles and the buffer dilution plots. Details of quantum
chemical calculations. This material is available free of charge
via the Internet at http://pubs.acs.org.
(
57) Observation of a long-lived hemiacetal intermediate formed by photolysis
of a 2-nitrobenzyl ether has been reported previously: Hirayama, Y.;
Iwamura, M.; Furuta, T. Bioorg. Med. Chem. Lett. 2003, 13, 905-908.
See also: Peyser, J. R.; Flechtner, T. W. J. Org. Chem. 1987, 52, 4645-
4
646.
JA039071Z
J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4595