Geminate recombination of hydroxyl radicals generated in 200 nm photodissociation of aqueous hydrogen peroxide

Article abstract of DOI:10.1016/j.cplett.2003.11.062

The picosecond dynamics of hydroxyl radicals generated in 200 nm photoinduced dissociation of aqueous hydrogen peroxide have been observed through their transient absorbance at 266 nm. It is shown that these kinetics are nearly exponential, with a decay time of ca. 30 ps. The prompt quantum yield for the decomposition of H₂O₂ is 0.56, and the fraction of hydroxyl radicals escaping from the solvent cage to the water bulk is 64-68%. These recombination kinetics suggest strong caging of the geminate hydroxyl radicals by water. Phenomenologically, these kinetics may be rationalized in terms of the diffusion of hydroxide radicals out of a shallow potential well (a solvent cage) with an Onsager radius of 0.24 nm.

Full text of DOI:10.1016/j.cplett.2003.11.062

Chemical Physics Letters 383 (2004) 481–485
www.elsevier.com/locate/cplett
Geminate recombination of hydroxyl radicals generated in 200 nm
photodissociation of aqueous hydrogen peroxide q,qq
*
Robert A. Crowell, Rui Lian, Myran C. Sauer Jr., Dmitri A. Oulianov, Ilya A. Shkrob
Chemistry Division, Argonne National Laboratory, 9700 S Cass Avenue, Argonne, IL 60439, USA
Received 9 October 2003; in final form 20 November 2003
Published online:
Abstract
The picosecond dynamics of hydroxyl radicals generated in 200 nm photoinduced dissociation of aqueous hydrogen peroxide
have been observed through their transient absorbance at 266 nm. It is shown that these kinetics are nearly exponential, with a decay
time of ca. 30 ps. The prompt quantum yield for the decomposition of H2O2 is 0.56, and the fraction of hydroxyl radicals escaping
from the solvent cage to the water bulk is 64–68%. These recombination kinetics suggest strong caging of the geminate hydroxyl
radicals by water. Phenomenologically, these kinetics may be rationalized in terms of the diffusion of hydroxide radicals out of a
shallow potential well (a solvent cage) with an Onsager radius of 0.24 nm.
Ó 2003 Elsevier B.V. All rights reserved.
HO þ H2O2 ! H2O þ HO2
ð2Þ
ð3Þ
1. Introduction
HO2 þ H2O2 ! O2 þ H2O þ HO
UV-light induced photodissociation of aqueous hy-
drogen peroxide
with rate constants of 2.7 ꢀ 107 and 7 ꢀ 109 Mꢁ1 sꢁ1
,
respectively [5]. Reaction (2) competes with the diffu-
sion-controlled recombination
hm
H2O2!2HO
ð1Þ
is widely used in water treatment, medical equipment
sterilization, bleaching, etc. [1,2]. These practical uses
are facilitated by the high oxidation potential of the
photogenerated hydroxyl radicals and large primary
quantum yield of H2O2 decomposition, 0.4–0.5 [3,4].
The overall quantum yield is further increased to 1–2,
due to Haber–Weiss chain reactions [1,5]
HO þ HO ! H2O2
ð4Þ
that proceeds with a rate constant k4 ¼ 4:7 ꢀ 109 Mꢁ1 sꢁ1
at 25 °C [5].
While the general photochemistry of H2O2 decompo-
sition is well understood, no mechanistic insight yet exists
as to the dynamics of geminate (HOꢂ ꢂ ꢂOH) pairs gener-
ated in photoreaction (1). In this Letter, the dynamics of
these hydroxyl radicals using 200 nm pump – 266 nm
probe ultrafast transient absorption spectroscopy are
reported. We also report an improved measurement for
the primary quantum yield of free hydroxyl radicals in
248 nm photodissociation of H2O2.
q Work performed under the auspices of the Office of Science,
Division of Chemical Science, US-DOE under Contract No. W-31-
109-ENG-38.
qq The submitted manuscript has been created by the University of
Chicago as Operator of Argonne National Laboratory (ÔArgonneÕ)
under Contract No. W-31-109-ENG-38 with the US Department of
Energy. The US Government retains for itself, and others acting on its
behalf, a paid-up, non-exclusive, irrevocable worldwide license in said
Letter to reproduce, prepare derivative works, distribute copies to the
public, and perform publicly and display publicly, by or on behalf of
the Government.
2. Experimental
2.1. Picosecond pump–probe kinetics
* Corresponding author. Fax: +630-2524993.
The ultrafast kinetic measurements reported below
were carried out using a 1 kHz Ti:sapphire setup. A
E-mail address: shkrob@anl.gov (I.A. Shkrob).
0009-2614/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2003.11.062

482
R.A. Crowell et al. / Chemical Physics Letters 383 (2004) 481–485
diode-pumped Nd:YVO laser was used to pump Kerr
lens mode-locked Ti:sapphire laser operating at
80 MHz (Spectra Physics Tsunami). The 45 fs FWHM
pulses from the oscillator where stretched to 80 ps in a
single grating stretcher, and single pulses were then
selected at 1 KHz with a Pockels cell. The 2 nJ pulses
were amplified to 4 mJ in a two-stage multipass
Ti:sapphire amplifier and passed through a grating
compressor that yielded Gaussian probe pulses of 60 fs
FWHM and 3 mJ centered at 800 nm. The amplified
beam was split into three parts. The first 800 nm beam
was used to generate 200 nm (fourth harmonic) pulses
used as a pump, as explained below: it was passed
through two tandem BBO crystals that generated the
second (Type I, 500 lm, 29°) and third (Type II,
100 lm, 58°) harmonic. The fourth harmonic is pro-
duced by upconversion of the third harmonic (266 nm)
and the second 800 nm beam in a third BBO crystal
(Type I, 100 lm, 68°). Up to 20 lJ of the 200 nm light
was produced this way (300 fs FWHM pulse). The
pump power, before and after the sample, was deter-
mined using a calibrated power meter (Ophir Optronics
model 2A-SH). The 266 nm probe was produced by
third harmonic generation with two 200 lm thick BBO
crystals using the third 800 nm beam. To remove the 1st
and 2nd harmonics, the resulting beam was passed
through a 3 nm FWHM notch filter and then reflected,
in succession, from four 266 nm dielectric mirrors
placed in each arm of the spectrometer. Test experi-
ments showed that no residual signal from the lower
harmonics reached the photodiode detectors.
jet is ca. 3). Even without 1 M H2O2 in the solution, the
266 nm transient absorption signal from the hydroxyl
radical and hydrated electron generated in the course of
the photoionization was <1% of the signal from the
hydrogen peroxide solution, under identical irradiation
conditions.
2.2. Nanosecond kinetics
Fifteen nanosecond FWHM, 1–20 mJ pulse from an
KrF excimer laser (Lamda Physik LPX 120i) was used
to photolyze N2-saturated aqueous solutions containing
20–40 mM of H2O2 and 1 M NaHCO3. The absorbance
of the 248 nm light by the bicarbonate was negligible
(<0.021 Mꢁ1 cmꢁ1). About 200 mL of the solution was
circulated through the cell using a peristaltic pump. The
1.36 mm optical path cell had 1 mm thick suprasil
windows. The laser beam was masked with a 3 mm ꢀ 6
mm rectangular brass aperture. This beam was normal
to the windows; the analyzing light (>500 nm) from a
superpulsed Xe lamp was crossed at 30° with this 248
nm excitation beam. The wavelength of the analyzing
light (600 nm) was selected using a 10 nm FWHM
narrow band interference filter.
A fast silicon photodiode (EG&G model FND100Q,
biased at )90 V) with a 1.2 GHz video amplifier
(Comlinear model CLC100) terminated into a digital
signal analyzer (Tektronix model DSA601) were used to
sample the transient absorbance kinetics (3 ns response
time). Two calibrated energy meters (Molectron model
J25-080) were used to measure the power of the incident
and transmitted 248 nm light.
The pump and probe beams were perpendicularly
polarized, focused to round spots of 100 and 20 lm
radius, respectively, and overlapped in the sample at
6.5°. The spot sizes were determined by scanning the
beam with a 10 lm pinhole. The signal and the reference
probe pulses (<1 nJ) were detected with fast silicon
photodiodes, amplified, and sampled using home-built
sample-and-hold electronics and a 16 bit A/D converter.
A mechanical chopper locked at half the repetition rate
of the laser was used to block the pump pulses on al-
ternative shots. No dependence of the transient absor-
bance on the pump and probe polarization was found.
Aqueous H2O2 solution was prepared by mixing
35 wt% hydrogen peroxide (Aldrich; stabilized with a
trace of tin) with deaerated nanopure water. The solution
was constantly purged with dry nitrogen. The experi-
ments were carried out using a 150 lm thick, 6 mm wide
high-speed (10 m/s) jet with a stainless steel nozzle. An all
316 stainless steel and Teflon flow system was used to
pump the solution through this jet at 0.5–0.75 dm3/min.
Importantly, two-photon photoionization of the sol-
vent (water) is negligible under the conditions of our
experiment because the two-photon absorbance of 200
nm light by water is orders of magnitude lower than
one-photon absorbance by 1 M H2O2 (OD200 across the
3. Results and discussion
3.1. 200 nm pump–probe kinetics
Fig. 1 shows transient absorption kinetics of hydroxyl
radicals observed at 266 nm in 200 nm photolysis of 1 M
H2O2. The short-lived ÔspikeÕ observed within the du-
ration of 200 nm light pulse is due to simultaneous ab-
sorption of both the pump and probe pulses (i.e., Ô1 + 1Õ
non-linear absorbance) and gives the instrument re-
sponse function; this ÔspikeÕ is a common artifact in UV
pump–probe studies [6].
As seen from Fig. 1, the kinetics observed after the
excitation pulse are nearly exponential. The hydroxyl
radicals rapidly escape from the solvent cage into the
water bulk (with a time constant of 29 ps); from the
exponential fit, the escape fraction of these radicals is ca.
68%. To estimate the quantum yield of photoreaction
(1), the following approach was used: the radial profiles
for the pump and probe beams were assumed to be
Gaussian, so that incident (time-integrated) photon
fluence for the pump beam was given by

R.A. Crowell et al. / Chemical Physics Letters 383 (2004) 481–485
483
10x10-3
From Eq. (9) we obtain a general formula
/ ¼ ð½ꢁDT=Tꢃ=eprobeÞ=ðIabs=p½ðrp2ump þ rp2robeÞꢃÞ:
ð10Þ
Thus, to obtain / using Eq. (10), the sample thickness L
and the absorption coefficient b of the sample at the
excitation wavelength need not be known; it is sufficient
to determine the optical density for the probe light and
the pump power before and after the sample. Note that
the prompt quantum yield for the decomposition of
H2O2 is 1/2 of the quantum yield of hydroxyl radicals.
For the kinetic trace shown in Fig. 1, Ipump ¼ 3 lJ,
rpump ¼ 100 lm and rprobe ¼ 20 lm. The molar extinc-
tion coefficients eprobe for hydroxyl radicals at 266 nm
are not known accurately: the estimates vary by as
much as 25%, from 370 to 465 Mꢁ1 cmꢁ1 [7–12]. Since
these coefficients were determined in pulse radiolysis
experiments, these coefficients depend on the estimates
for the radiolytic yields; the latter have been revised
repeatedly after the original experiments were carried
out [5–13]. Perhaps, the most reliable measurement is by
Nielsen et al. [8], eprobe ¼ 420 Mꢁ1 cmꢁ1. Using this
estimate, the average concentration of the hydroxyl
radicals across the jet was 0.8 mM and the quantum
yield of peroxide decomposition at 0 and 500 ps was
0.56 and 0.38, respectively. At 200 nm, hydrogen per-
oxide has an extinction coefficient of 200 Mꢁ1 cmꢁ1 [8]
and the pump light was absorbed in 22 lm (i.e., all of
the 200 nm light was absorbed by the jet), so that the
concentration of radicals near the jet surface was 5.6
mM. While this concentration is relatively high, given
the rate of reaction (4) the lifetime of free hydroxyl
radical was 40 ns, i.e., no cross recombination occurred
within the observation window (500 ps). Reaction (2) is
also too slow (k2 ¼ 2:9 ꢀ 107 Mꢁ1 sꢁ1 [5]) to affect the
geminate dynamics of hydroxyl radicals in 1 M hydro-
gen peroxide.
Our estimate of 0.38 for the primary quantum yield
for decomposition of H2O2 (after the escape of hy-
droxide radicals to the bulk) compares favorably with
the estimates given in the literature: 0.49 ꢅ 0.07 [3] and
0.47 ꢅ 0.03 [4] at 254 nm and 0.45 ꢅ 0.06 at 222 nm [2]
(also, see below). This estimate can be brought into a
better agreement with these previous measurements by
reconsideration of the extinction coefficient for the hy-
droxyl radical. The radiolytic measurements of Nielsen
et al. [8] and Czapski and Bielski [9] relied on the yield of
5.3 hydroxyl radicals per 100 eV of radiation absorbed
by N2 O-saturated water, whereas other work suggests a
15% higher yield (6.1 per 100 eV) [13]. This readjustment
decreases the molar absorptivity to 400 Mꢁ1 cmꢁ1
(which also compares better to 370 Mꢁ1 cmꢁ1 obtained
in the photolytic experiments of Boyle et al. [10]) and
increases our estimate of the quantum yield from 0.38 to
0.44 – in agreement with the previous work [2–4] and the
estimates given below (in such a case, the prompt
quantum yield of 0.56 should be increased to 0.65).
5
H2O2, 200 nm
0
1
10
100
time, ps
Fig. 1. Transient absorbance at 266 nm detected in 200 nm pulse ex-
citation (300 fs FWHM, Ipump ¼ 3 lJ; rpump ¼ 100 lm) of 1 M aque-
ous H2O2 in a 150 lm jet. The pump fluence was 9.5 ꢀ 10ꢁ3 J/cm2. This
signal is from hydroxyl radicals, 64% of which escape the recombi-
nation in the cage. These decay kinetics were simulated using ShushinÕs
semianalytical theory of diffusion in a shallow potential well (the solid
line is the least-squares fit of the kinetic trace for t > 3 ps).
2
JpumpðrÞ ¼ J0 expðꢁ½r=rpumpꢃ Þ
ð5Þ
and the beam power was Ipump ¼ prp2umpJ0. As the pump
beam penetrates the sample, it is attenuated as
expðꢁbxÞ; where b is the absorption coefficient of the
sample and x is the penetration depth. Neglecting the
absorbance of 200 nm light by photogenerated hydroxyl
radicals, the total number of 200 nm photonsabsorbed
by the sample of thickness L is given by Iabs ¼ Ipump
ð1 ꢁ expðꢁbLÞÞ. The sample-average yield PðrÞ of a
photoproduct across this sample is given by equation
PðrÞ ¼ /JpumpðrÞð1 ꢁ expðꢁbLÞÞ=L;
ð6Þ
where / is the quantum yield for the photoproduct.
Assuming that the absorption of the probe light is weak,
PðrÞLeprobe ꢄ 1 (which was the case in our experiment),
where eprobe is the extinction coefficient of the photo-
product (hydroxyl radical) at the probe wavelength of
266 nm, the loss DJprobeðrÞ in the intensity of the 266 nm
probe light is given by equation
DJprobeðrÞ ¼ ꢁJprobeðrÞPðrÞeprobeL;
ð7Þ
where JprobeðrÞ is the radial distribution function of the
probe light given by Eq. (5) with the beam radius of
r
probe. Combining all of these equations, and carrying
out the integration over r, the beam-average loss in the
probe transmission defined as
ꢀ
Z
Z
1
1
ꢁDT=T ¼
dr2prDJprobeðrÞ
dr2prJprobeðrÞ
0
0
ð8Þ
ð9Þ
is given by
ꢁ DT=T
¼ ꢁeprobe/ð1 ꢁ expðꢁbLÞÞIpump=½pðrp2ump þ rp2robeÞꢃ:

484
R.A. Crowell et al. / Chemical Physics Letters 383 (2004) 481–485
3.2. Quantum yield for 248 nm photolysis
absorbed by these radicals. Numerical simulations in-
dicate that this effect fully accounts both for the negative
curvature of the DOD600 plot and the observed linear
decrease in the transmission of the 248 nm light with the
laser power. Note that this effect is minor in our pico-
second experiment because the concentration of hydro-
gen peroxide (1 M vs. 20–40 mM) and its extinction
coefficient at 200 nm (200 vs. 26 Mꢁ1 cmꢁ1) are both
considerably higher. Thus, <0.5% of the 200 nm light
was absorbed by the hydroxyl radicals, which justifies
the use of Eq. (10).
All of the previous estimates for the quantum yield of
H2O2 decomposition were obtained by product analysis
(evolution of gaseous oxygen) at the end of chain reac-
tions (1)–(4). To improve on these indirect estimates for
the yield of reaction (1), we used 248 nm laser flash
photolysis to observe the primary yield of free hydroxyl
radicals through their fast reaction with bicarbonate
anions (k11 ¼ 1:5 ꢀ 107 [14] or 8.5 ꢀ 106 Mꢁ1 sꢁ1 [15])
HO þ HCOꢁ ! COꢁ þ H2O
ð11Þ
3
3
3.3. Kinetic analysis
At the pH 8.3 (1 M bicarbonate), all hydrogen peroxide
(pKa ¼ 11:7 [5]) is in the protonated form. The forma-
tion of the carbonate radical anion was observed at 600
nm (e600(COꢁ3 ) ¼ 1860 Mꢁ1 cmꢁ1 [14]); this formation
was complete in 200 ns (the apparent rate constant for
The most intriguing of our observations is that the
recombination kinetics of hydroxyl radicals are nearly
exponential, whereas a very different dependence was
expected for the geminate recombination controlled by
free diffusion of these radicals. This suggests a weak
interaction between the radicals (in other words, the
existence of a solvent cage around the radical partners).
Phenomenologically, this interaction can be described in
terms of a mean force potential UðrÞ having a profile of
a well [17,18]. As shown by Shushin [17], the decay ki-
netics of diffusional escape from such a potential well
exhibits two regimes: (i) an exponential decay on a short
time scale (with rate constant W ¼ Wr þ Wd; where Wr;d
are the recombination and dissociation rates of the
radicals in the potential well, respectively), and tꢁ1=2
behavior on the longer time scale, due to slow recom-
bination of radicals that escaped beyond the Onsager
radius a, at which UðaÞ ¼ ꢁkT. In this theory, the es-
cape probability pd of the radical partners is equal to
Wr=W and the effective radius Reff of the reaction in the
bulk is given by a product að1 ꢁ pdÞ [18]. Fig. 1 dem-
onstrates the fit of the experimental kinetics to ShushinÕs
theory expressions [17,18] for the survival probability of
a radical pair migrating out of the potential well. Using
a diffusion coefficient of 2.8 ꢀ 10ꢁ5 cm2/s for the hy-
droxyl radical [5], one obtains the parameters pd ¼ 0:64,
a ¼ 0:24 nm, and W ꢁ1 ¼ 20:6 ps, which in turn yield
Reff ¼ 0:09 nm. A direct estimate from the known rate
constant of reaction (4) in the bulk [5] gives the re-
combination radius of 0.11 nm.
reaction (11) was 2 ꢀ 107 Mꢁ1
s
ꢁ1). The second-order
decay of the carbonate radical anion (2k ¼ 1:25 ꢀ 107
M
ꢁ1 sꢁ1 [16]) was slow, with t1=2 of 5–10 ls, so the loss of
the 600 nm signal due to the cross recombination was
negligible. The reaction of COꢁ3 with H2O2 is also very
slow (k ¼ 8 ꢀ 105 Mꢁ1 sꢁ1 [16]). The highest yield of the
hydroxyl radicals estimated from our kinetic data was
140 lM; at this concentration, all of the reactions
competing with reaction (11) (such as rxn. (2) and (4))
are too slow to change the observed yield of COꢁ3 radical
anions at t ¼ 200 ns. Note that 2 lM hydroxide and
9.9 mM carbonate are present in the solution due to the
protic equilibria of the carbonate system. The hydroxide
anion rapidly reacts with the hydroxyl radical yielding
Oꢁ [5], but the concentration of the hydroxide is too low
to affect our measurement. While the carbonate anion
also reacts with the hydroxyl radical with rate constant
of 4.2 ꢀ 108 Mꢁ1 sꢁ1 [14], the product of this reaction is
COꢁ3 , i.e., no loss of the 600 nm absorbance from the
carbonate radical anion ensues.
The transient absorbance at 600 nm (at t ¼ 200 ns)
was plotted vs. the number of the absorbed photons.
The initial slope of this curve (for 248 nm laser fluence
<1016 photons/cm2) corresponds to a quantum yield of
0.44 ꢅ 0.01, in reasonable agreement with the previous
measurements. At higher 248 nm laser fluence ((2–
6) ꢀ 1016 photons/cm2), the DOD600 plot exhibited a
negative curvature (for example, the apparent quantum
yield decreases to 0.384 for the fluence of 4 ꢀ 1016 pho-
tons/cm2). This negative curvature is accounted for by
the absorption of the 248 nm light by hydroxyl radicals
that are generated within the duration of the laser pulse
(e248(HO) ¼ 500 Mꢁ1 cmꢁ1 [9]). The molar absorptivity
of H2O2 at 248 nm is very low compared to that of the
hydroxyl radicals formed in reaction (1) (we obtained
4. Conclusion
Photoexcitation of hydrogen peroxide at 200 nm
causes its rapid (<300 fs) dissociation with the forma-
tion of weakly interacting geminate hydroxyl radicals in
the solvent cage. At 25 °C, the decay of these radicals
from the cage takes ca. 20 ps, with 64% of the radicals
escaping to the water bulk and the rest recombining.
The prompt quantum yield for the photodissociation of
H2O2 is 0.56 (or 0.65, depending on the assumed esti-
e
248(H2O2) ¼ 26 Mꢁ1 cmꢁ1 using an Olis/Cary-14 spec-
trophotometer and 24.8 Mꢁ1 cmꢁ1 from the KrF laser
transmission data), i.e., under the conditions of our
experiment up to 20–40% of the 248 nm photons are

R.A. Crowell et al. / Chemical Physics Letters 383 (2004) 481–485
[4] D.H. Volman, J.C. Chen, J. Am. Chem. Soc. 81 (1959) 4141.
485
mate for the molar absorptivity of hydroxyl radical; see
above), and the Onsager radius of the attractive poten-
tial is 0.24 nm. The absolute primary quantum yield for
H2O2 decomposition (to free hydroxyl radicals) for 248
nm photoexcitation is 0.44.
[5] A.J. Elliot, Rate constants and G-values for the simulation of the
radiolysis of Ligth water over the range 0–300 °C, AECL Report
11073, COG-95-167, AECL Research, Chalk River Laboratories,
Chalk River, Ont., Canada, 1994.
[6] M. Lorenc, M. Ziolek, R. Naskrecki, J. Karolczak, J. Kubicki, A.
Maciejewski, Appl. Phys. B 74 (2002) 19.
[7] J.K. Thomas, J. Rabani, M.S. Matheson, E.J. Hart, S. Gordon,
J. Phys. Chem. 70 (1966) 2409.
Acknowledgements
[8] S.O. Nielsen, B.D. Michael, E.J. Hart, J. Phys. Chem. 80 (1976)
2482.
This work was performed under the auspices of the
Office of Science, Division of Chemical Science, US-
DOE under Contract No. W-31-109-ENG-38. We thank
Prof. S.E. Bradforth of USC and Dr. D.M. Bartels of
NRDL for providing the motivation for this study.
I.A.S. thanks Dr. S.V. Lymar of BNL for the intro-
duction to the perils of aqueous radiation chemistry.
[9] G. Czapski, B.H.J. Bielski, Radiat. Phys. Chem. 41 (1993)
503.
[10] J.W. Boyle, J.A. Ghormley, C.J. Hochanadel, J.F. Riley, J. Phys.
Chem. 73 (1969) 2886.
[11] A.J. Elliot, G.V. Buxton, J. Chem. Soc., Faraday Trans. 88 (1992)
2465.
[12] P. Pagsberg, H. Christensen, J. Rabani, G. Nilsson, J. Fenger,
S.O. Nielsen, J. Phys. Chem. 73 (1969) 1029.
[13] R.H. Schuler, L.K. Patterson, E. Janata, J. Phys. Chem. 84 (1980)
2088.
References
[14] J.L. Weeks, J. Rabani, J. Phys. Chem. 70 (1966) 2100.
[15] S.V. Lymar, H.A. Schwartz, G. Czapski, Radiat. Phys. Chem. 59
(2000) 387.
[1] S. Lunak, P. Sedlak, J. Photochem. Photobiol. A 68 (1992) 1, and
references therein.
[16] D. Behar, G. Czapski, I. Duchovny, J. Phys. Chem. 74 (1970)
2206.
[2] G.A. Loraine, W.H. Glaze, J. Adv. Oxid. Technol. 4 (1999) 424.
[3] J.L. Weeks, M.S. Matheson, J. Am. Chem. Soc. 78 (1956) 1273;
F.S. Dainton, J. Am. Chem. Soc. 78 (1956) 1278.
[17] A.I. Shushin, Chem. Phys. Lett. 118 (1985) 197.
[18] A.I. Shushin, J. Chem. Phys. 95 (1991) 3657.