Supporting Information

Photochemistry of Singlet Oxygen Sensor Green

Sooyeon Kim, Mamoru Fujitsuka, and Tetsuro Majima*

The Institute of Scientific and Industrial (SANKEN), Osaka University, Mihogaoka 8-1,

Ibaraki, Osaka 567-0047 (Japan)

Experimental Details

Materials

Singlet Oxygen Sensor Green® (Fl-An, Invitrogen/Molecular Probes®), 5(6)-carboxy-2’,7’-

dichlorofluorescein (Fl-Cl, 21882, Sigma), 5(6)-carboxyfluorescein (21877, Sigma), 9,10-

dimethylanthracene (DMA, D146706, Aldrich), and α,β,γ,δ-tetrakis(1-methylpyridinium-4-

yl)porphyrin (TMPyP4, A5014, TCI) were used without further purification. Anthracene (AC495,

TCI) and p-pheynylenediamine (P0170, TCI) were exploited as triplet state energy donor and one

electron reduction agents, respectively. Furthermore, we prepared endoperoxide form of Fl-An (Fl-

EP) either by continuous irradiation of Fl-An[1, 2] or chemical method that used Na2MoO4∙2H2O

(198-02471, WAKO) and H2O2 (086-07445, WAKO) as reported previously.[3] Using either a 50

mW cm-2 Hg lamp (REX-250, Asahi Spectra) with a bandpass filter (BP420-440, Olympus) or 100

mW cm-2 Xe lamp (LAX-C100, Asahi Spectra) with a bandpass filter (BA510-550, Olympus), 2

mL of Fl-An solution was irradiated in the wavelength range of 420–440 or 510–550 nm,

respectively, with a magnetic stirring.

Optical properties of fluorescein derivatives, including Fl-An and Fl-Cl, are known to be

sensitive to pH because its conjugation and energy level become differed by (de)protonation of

xanthene moiety (Xa) or forming a transparent lactone form. Therefore, we chose an aqueous

buffer with pH 8.0 in order to maintain Fl-An and Fl-Cl as a trianionic form. Throughout this study

except nanosecond laser flash photolysis, we used pH 8.0 Tris-K+ buffer, which was prepared with

200 mM KCl, 4 mM MgCl2, and 28 mM Tris-HCl. HCl and NaOH solution were added to adjust

pH.

Measurement of fluorescence lifetime (τfl) and quantum yield of fluorescence (Φfl)

τfls of Fl-An and Fl-Cl were measured using a time-resolved fluorescence microscope with a

confocal optics (MicroTime 200; PicoQuant, Berlin-Adlershof, Germany). In order to measure τfl,

70 µL of Fl-An and Fl-Cl solution in pH 8.0 Tris-K+ buffer was loaded into the micro-chamber

made by a 1-inch glass cover slip and Secure-Seal (S24733, Invitrogen). The samples were excited

through an oil objective (Olympus, UAPON 150XOTIRF; 1.45 NA, 150x) with a 485-nm pulsed

laser (PicoQuant, full width at half-maximum 120 ps) controlled by a PDL-800B driver

(PicoQuant). The excitation power of ~ 0.15 µW was used. The emission was collected with the

same objective and detected by a single photon avalanche photodiode (Micro Photon Devices,

PDM 50CT and 100CT) through a 75-µm pinhole for spatial filtering to reject out-of-focus signals.

The data collected using the PicoHarp 300 TCSPC module (PicoQuant) were stored in the time-

tagged time-resolved mode (TTTR), recording every detected photon with its individual timing.

To determine Φfl of Fl-An and Fl-EP, 5(6)-carboxyfluorescein (Φfl = 0.92) was used as a

reference. For the preparation of Fl-EP for Φfl determination, a chemical method to generate 1O2

was chosen because the direct irradiation of Fl-An causes the irreversible photobleaching of

fluorescein moiety (Fl), leading to the lower Φfl than the actual value. In order to decrease the error

and prevent the intermolecular interactions, we prepared the dye samples with absorbance of 0.005,

0.01, and 0.015 at 465 nm. Next, the integrated areas of fluorescence spectra of the samples were

calculated and linearly fitted against the responsive absorbance. Finally, Φfl was determined by

comparing the slope of the linear plots with that for 5(6)-carboxyfluorescein.

As summarized in Table S1, the Φfl of Fl-An and Fl-EP were found to be more than 20 times

different, which was large enough to distinguish between Fl-An and Fl-EP. In terms of the

“brightness”, Fl-EP was not as bright as carboxyfluorescein (Φfl = 0.92), but still moderately bright

(Φfl ≈ 0.4) suitable for the various experiments including single-molecule fluorescence

spectroscopy which requires higher Φfl than 0.1.[4]

Measurement of femtosecond laser flash photolysis (fs-LFP)

The sub-picosecond transient absorption spectra were measured by the pump and probe method

using a regenerative amplified titanium sapphire laser (Spectra Physics, Spitfire Pro F, 1 kHz)

pumped by a Nd:YLF laser (Spectra Physics, Empower 15) for each dyes dissolved by pH 8.0 Tris-

K+ buffer. The samples of Fl-An and Fl-Cl were prepared at the concentration where they have the

same absorbance of approximately 1.5 with 0.2-cm path length ([Fl-An] = 120 μM and [Fl-Cl] =

108 μM). An excitation pulse at 500 nm was generated by optical parametric amplifier (Spectra

Physics, OPA-800CF). A white continuum pulse, which was generated by focusing the residual of

the fundamental light to a sapphire plate after a computer controlled optical delay, was divided into

two parts and used as the probe and the reference lights, of which the latter was used to compensate

the laser fluctuation. Both probe and reference lights were directed to a rotating sample cell with

0.1 cm of path length and were detected with a charge-coupled device detector equipped with a

polychromator (Solar, MS3504). The pump pulse was chopped by a mechanical chopper

synchronized to one-half of the laser repetition rate, resulting in a pair of the spectra with and

without the pump, from which an absorption change induced by the pump pulse was estimated.

Measurement of nanosecond laser flash photolysis (ns-LFP)

The nanosecond transient absorption spectra were measured by the pump and probe method

using the third-harmonic oscillation (355 nm, 4 ns fwhm) and second harmonic oscillation (532 nm,

4 ns fwhm) from a Q-switched Nd:YAG laser (Continuum, Surelite II-10) as the excitation sources.

The light from a Xe flash lamp (Osram, XBO-450) was focused into the sample solution for the

transient absorption measurement. Time profiles of the transient absorption in the UV−Vis region

were measured with a monochromator (Nikon, G250) equipped with a photomultiplier

(Hamamatsu Photonics, R928) and digital oscilloscope (Tektronics, DPO 3054). Each decay

profile was collected after averaging 16 and 64 times for the 355- and 532-nm excitation,

respectively.

During the ns-LFP measurements, it should be noted that 45 volume% (v%) of acetonitrile

(ACN) mixed pH 8.0 Tris-K+ buffer was used because the absorption of Fl-An at 532 nm in pH 8.0

buffer was too small to obtain the acceptable transient absorption signal for the 532-nm excitation.

In the ACN mixed buffer, approximately 5 nm red-shift occurred in Fl-An, resulting in

approximately twice increase in absorbance at 532 nm. Meanwhile, the absorption spectral shape

of Fl-An in the ground state did not change, which indicated that the trianionic form of Fl was well

maintained.

In addition, 355-nm excitation caused significant photodecomposition of dyes during the

repetitive scanning. Due to the photo-instability of Fl-An under the 355-nm excitation, it was

impossible to obtain the whole transient absorption spectrum with good signal-to-noise ratio.

Instead, we measured only decay profiles at the peak position of the transient absorption (430 nm)

with 16 averaging times as shown in Figure 5a. On the other hand, in the case of transient

absorption spectrum of Fl-Cl (Figure 4b), most of the photons were absorbed by anthracene

molecules, triplet sensitizers. Thus, photodecomposition of Fl-Cl by direct excitation was largely

prevented.

Time-resolved phosphorescence measurement

The samples of Fl-An, Fl-EP, and TMPyP4 were prepared in pH 8.0 Tris-K+ buffer in a 1 × 1

cm2 quartz cell ([Fl-An] = 74.6 μM, [Fl-Cl] = 211.3 μM, and [TMPyP4] = 48.8 μM where they

have the same absorbance of 0.5 at 532 nm). The second-harmonic oscillation (532 nm, 4 ns fwhm,

6–6.5 mJ cm-1 pulse-1) from a Q-switched Nd:YAG laser (Continuum, Surelite II-10) was used for

the excitation light. The photoinduced luminescence from the sample cell was collected with quartz

lenses, passed through a monochromator, and then introduced into a near IR photomultiplier tube

module (Hamamatsu Photonics, H10330A-75). After amplified by 350 MHz amplifier unit

(Stanford Research, SR445A), the output of the photomultiplier was sent to a gated photon counter

(Stanford Research, SR400) under direct control from a PC via GPIB interface. To measure the

lifetime of 1O2, the signals were accumulated (10 repetitions) by changing the delay time from 0 to

70 µs with a gate width of 0.3 µs.

Calculation of ionization quantum yield (Φion)

In light of our previous report,[5] Φion of one-electron oxidized Fl (X) was determined with a

reference compound, benzophenone (BP). Briefly, Φion can be determined by

Φion = [X]/[3BP*] (Eq. 1)

where [X] and [3BP*] are the concentrations of X and BP in the triplet excited state (3BP*),

respectively. [3BP*] is equal to the photon concentration of the laser flash at 355 nm, determined

using chemical actinometry of the T1 → Tn absorption of BP as the standard according to the

method described. [X] was calculated using the previously reported molar absorption coefficient

(ε) of X, 5.0 104 M-1 cm-1 at 428 nm.[6] Since X is generated by the two-photon absorption, Φion

increases as the laser intensity is raised (Table S5).

Calculation of rate constant for bimolecular triplet-triplet energy transfer (kq) from Fl-An in

the triplet excited state (3Fl*-An) to O2

From the decay profile of transient absorption of 3Fl-An* excited at 532 nm (Figure 5a, below),

kq can be obtained by

kT = k0T + kq[O2] (Eq. 2)

where kT, k0T, and kq are reciprocals of the decay time of 3Fl*-An in the air- and Ar-saturated

conditions, and the bimolecular quenching rate constant between 3Fl*-An and O2, respectively.[7]

Considering that 45 v% ACN mixed pH 8.0 Tris-K+ buffer solution was used during the ns-LFP

measurement and the dissolved oxygen concentrations were different between ACN and water

([O2] = 1.9 and 0.27 mM at the room temperature, respectively),[8] kq = 1.2 109 M-1 s-1 was

obtained.

Supporting Tables and Figures

Table S1. Φfl and τfl of Fl-An, Fl-EP, fluorescein, and Fl-Cl.

Compound Φfl τfl / ns

Fl-An 0.009[a]

0.019 3.13[b]

Fl-EP 0.45[a]

0.37 n.d. [c]

Fluorescein 0.92[d] 3.6[e]

Fl-Cl 0.92 3.69

[a] Ref. 2. [b] τfl of Fl-An is considered to be originated from Fl impurities without anthracene

moiety (An) and/or preformed Fl-EP. [c] n.d. stands for no data available. [d] Ref. 9. [e] Ref. 10.

Table S2. Energy levels of Fl, An, and EP in the singlet and triplet excited states, and oxidation

and reduction potentials of Fl and An from the previous studies.

Moiety

(reference compound)

Singlet excitation

energy (kJ mol-1)

Triplet excitation

energy (kJ mol-1)

E (D+/D)

(V vs NHE)

E (A/A-)

(V vs NHE)

Fl (Fl-Cl) 230.9[a] 164.3[b] 1.66[c] -1.05[c]

An (DMA) 301[d] 168[d] 1.91[d] -1.90[e]

EP (DMA with

endoperoxide) 371.5[f] 299.1[f] n.d.[g] n.d.[g]

[a] Calculated from the 0-0 absorption of Fl-An. [b] Ref. 11. [c] Ref. 12. [d] Ref. 13. [e] Ref. 6. [f]

Ref. 14. [g] n.d. stands for no data available.

Figure S1. Chemical structures of Fl-An and Fl-EP analyzed by Gollmer et al.[2]

Fl-An Fl-EP

Figure S2. Chemical structures and abbreviations of the chromophores of Fl-An, reference

compounds, and radical species of Fl. The chromophore of Fl is Xa whose exact name is 6-

hydroxy xanthen-3-one. Since Fl is trianion in pH 8.0 Tris-K+ buffer, the net charges of one-

electron oxidized and reduced Fl are -2 and -4, respectively. The expected chemical structure of

one-electron oxidized and reduced Fl are as shown above and named as X and R, respectively.

Xa

Carboxyfluorescein Fl-Cl

One-electron oxidized Fl

(X)

One-electron reduced Fl

(R)

Figure S3. Fluorescence spectra of Fl-An excited at 350 nm (cyan), 375 nm (red), and 485 nm

(blue). Inset: absorption spectra of Fl-An in pH 8.0 Tris-K+ buffer (black) and in ACN (magenta),

and Fl-Cl in pH 8.0 Tris-K+ buffer (green). Cyan, red, and blue arrows indicate the excitation

wavelength of 350, 375, and 485 nm, respectively, used for the measurements of fluorescence

spectra.

Since the absorption peaks at 350–400 nm and 450–550 nm correspond to An and Fl,

respectively (Figure S3, inset), fluorescence of Fl-An would have the excitation-wavelength (λex)

dependence if energy transfer from An in the singlet excited state (1An*) to Fl does not occur. As

shown in Figure S3, however, neither An fluorescence nor enhancement of Fl fluorescence was

observed when Fl-An was excited at 350 and 375 nm (cyan and red, respectively). The weak Fl

fluorescence was resulted from the direct excitation of Fl because Fl also has weak absorption in

350–400 nm wavelength regions (inset, green line). The sharp emission peak observed in 400–500

nm range was water Raman scattering signal. Overall, the Fl-An fluorescence centered at 520 nm

was mainly induced by Fl, while the An fluorescence was so weak regardless of λex that its

intensity was weaker than water Raman scattering signal. This result indirectly suggests that 1An*

is mostly deactivated by the nonradiative processes since neither the intramolecular singlet-singlet

energy transfer from 1An* to the Fl nor An fluorescence were observed.

λex = 350 nm

λex = 375 nm

λex = 485 nm

Figure S4. a) Absorption and b) fluorescence (λex = 465 nm) spectral changes during the

irradiation of Fl-An with the continuous visible light in the range 510 to 550 nm (dark blue to

green). Black arrows show spectral change occurred with a conversion of Fl-An into Fl-EP.

Figure S5. Transient absorption spectrum of one-electron reduced Fl-Cl (analogues to R) observed

at 70 μs after a pulse during the ns-LFP of Fl-Cl at λex = 532 nm (20 mJ pulse-1) with 1-cm path

length, [Fl-Cl] = 100 μM, and [p-phenylenediamine] = 2 mM.

Figure S6. Transient absorption spectra of 1Fl*hot and relaxed 1Fl*, calculated from the global

fitting of 56 time profiles monitored during the fs-LFP of Fl-Cl (410–465 nm) with λex = 500 nm.

The fitting equation used for the global fitting of the spectra above is

y(x) = A1exp(-kcx) + A2{-exp(-kcx) + exp(-kd1x)} + A3exp(-kd2x) + A4 (Eq. 3)

where A1, A2, A3, and A4 are the amplitudes correlated to the maximum concentration of 1Fl*hot,

relaxed 1Fl*, Fl in the ground state, and a constant for the fitting, respectively. The detailed

descriptions and rate constants for each process are summarized in Table S3.

Table S3. Descriptions and rate constants for the relaxation pathways of Fl-Cl in the excited states

as shown in Figure S6.

Processes Description k / s-1

1Fl*hot-An → 1Fl*-An Vibrational cooling of 1Fl*hot kc = 2.3 1011

1Fl*-An →

Fl-An + heat and/or hν

Radiative and/or non-radiative

decay of 1Fl* kd1 = 2.5 109

Recovery of Fl in the ground state kd2 = 3.3 109

Figure S7. a) Diagram to illustrate relaxation processes of Fl-An in the excited states. kc, kCS1, kCS2,

kCR, and kd indicate the rate constants for vibrational cooling including solvation, intramolecular

charge separation between An and either 1Fl* in a Franck-Condon state (1Fl*hot) or relaxed 1Fl*,

charge recombination of R-An∙+, and other deactivation processes of 1Fl*, respectively. b) Time

profiles of transient absorption at 550, 560, and 570 nm where the stimulated emission of 1Fl*-An

and An∙+ formation can be observed (open circles) during the fs-LFP of Fl-An with λex = 500 nm.

Results of multi-exponential fitting using Eq. 3 are shown as black lines.

After the excitation at 500 nm, 1Fl*-An in a Franck-Condon state (1Fl*hot-An) is instantly

populated, followed by photoinduced electron transfer (PET, used as the same meaning of charge

separation in this report) and/or vibrational cooling processes to be relaxed via its charge separated

state and/or 1Fl*-An, respectively. In order to obtain rate constants accurately from the fitting of

transient absorption at 420 nm (absorption of 1Fl*) or around 550 nm (stimulated emission of 1Fl*),

a careful consideration is required because the signs for rise and decay in the fitting equation

become inverse if (kCS2 + kd) is larger than kc.[15] From the simple double exponential fitting of the

transient absorption at 720 nm, the formation of An∙+ (≈ 1012 s-1, Figure 3c) was found to be faster

than vibrational cooling of 1Fl* (2.3 1011 s-1, see Figure S6). At the same time, non-exponential

decay curves of 1Fl* absorption and stimulated emission necessitate at least one slow component

(< 1010 s-1 and/or a constant indicating longer than the time limit, 2 ns) for the accurate fitting.

Taken together, under the present experimental condition (pH 8.0 Tris-K+ buffer and λex = 500 nm),

charge separation takes place mainly in 1Fl*hot-An rather than as a sequential process after

vibrational cooling. Time profiles of transient absorption for the stimulated emission of 1Fl*-An

and An∙+ formation observed at 550, 560, and 570 nm (Figure S7b) were globally fitted by the

equation as shown below

y(x) = A1{-exp(-kCS1x) + exp(-kCRx)} + A2{-exp(-(kCS1+kc)x) + exp(-kdx)} + A3 (Eq. 4)

where A1, A2, and A3 are the coefficients correlated to the maximum concentration of An∙+,

stimulated emission from 1Fl*-An, and a constant for the fitting, respectively. Here, A1 and A2 are

b) a)

positive and negative, respectively. The detailed descriptions and rate constants for each process

are summarized in Table S4. As a result, kCS1 is determined to be 9.7 1011 s-1, and hence the

yields for the deactivation processes of 1Fl*hot-An, vibrational cooling and charge separation, are

0.2 and 0.8, respectively. Thus, R-An∙+ mainly originated from 1Fl*hot-An, and we denoted the kCS1

as the representative time constant for intramolecular PET (kPET) in the manuscript. It should be

noted that the charge separation of relaxed 1Fl*-An is still possible and may become the major

process to induce charge separated state provided that the different solvent and excitation

wavelength are used, resulting in the same consequence, effective fluorescence quenching.

Table S4. Descriptions and rate constants for the relaxation pathways of Fl-An in the excited states

as shown in Figure S7.

Processes Description k / s-1

1Fl*hot-An → R-An∙+ Charge separation to form R and An∙+

(kPET in the manuscript) kCS1 = 9.7 1011

R-An∙+ → Fl-An Charge recombination between R and An∙+ kCR = 2.0 1011

1Fl*hot-An → 1Fl*-An Vibrational cooling of 1Fl*hot (see Figure S7) kc = 2.3 1011

(Fixed)

1Fl*-An →

Fl-An + heat and/or hν Radiative and/or non-radiative decay of 1Fl* kd = 8.9 109

Figure S8. Time profiles of normalized fluorescence decay of Fl-Cl (black) and Fl-An (blue) (λex =

485 nm).

Figure S9. Transient absorption spectrum of DMA in the triplet excited state observed at 10 μs

after a laser pulse during the ns-LFP of DMA in Ar-saturated benzene (λex = 355 nm).

Figure S10. a) Time profiles of 1O2 phosphorescence decay of TMPyP4 (blue), Fl-Cl (red), and Fl-

An (blue) observed at 1280 nm (λex = 532 nm at 6.5 mJ cm-2) in pH 8.0 Tris-K+ solution. P. I.

stands for phosphorescence intensity. b) Determination of the singlet oxygen generation quantum

yield (ΦΔ) of Fl-Cl (red circles) as compared with TMPyP4 (black circles, ΦΔ = 0.74). By

comparing the slope of the two linear plots between TMPyP4 and Fl-Cl (black and red lines,

respectively), ΦΔ of Fl-Cl is calculated to be 0.06.

1O2 is generated by the bimolecular triplet-triplet energy transfer from a Sens in the triplet

excited state (3Sens*) to 3O2. Thus, the detection of 1O2 phosphorescence at 1280 nm indirectly

indicates the presence of 3Sens* in the photoreaction. Weaker phosphorescence of Fl-An than that

of Fl-Cl can be explained by (1) the presence of ultrafast deactivation pathway via the

intramolecular PET and (2) chemical quenching of 1O2 by An. Phosphorescence observed for Fl-Cl

in the triplet excited state (3Fl*-Cl) is in accordance with our suggested relaxation pathway as

shown in Figure 6.

Furthermore, ΦΔ of Fl-Cl was determined as compared to a reference compound, TMPyP4 (ΦΔ =

0.06 and 0.74, respectively) (Figure S10b). The same calculation method to that for Φfl was used

(see Experimental Details). Here, ΦΔ is the term containing the quantum yield of triplet formation

(ΦT) and fraction of triplet excited state quenched by O2 which gives singlet oxygen (fTΔ), ΦΔ =

ΦTfTΔ.[13] In addition, ΦΔ of Fl-An and Fl-EP could not be determined because of their negligible

phosphorescence signal. Taken together, ΦΔ of Fl in Fl-An and Fl-EP is smaller than that of Fl-Cl

(ΦΔ = 0.06), and it is tentatively assumed that smaller ΦΔ of Fl-An and Fl-EP is related to high

internal conversion quantum yield (ΦIC) than that of Fl-Cl. Detailed discussion on ΦIC of Fl-An can

be found in the manuscript.

TMPyP4, ΦΔ = 0.74

Fl-Cl, ΦΔ = 0.06

Figure S11. Laser intensity dependences of the absorbance at 430 nm observed at 100 ns and 5 μs

after a laser pulse during the ns-LFP of Fl-An with the excitation at λex = 355 (red) and 532 nm

(black), respectively, in the air-saturated 45 v% ACN pH 8.0 Tris-K+ buffer solution with 1-cm

path length.

Table S5. Φion of Fl-An at various laser pulse intensities (I).

I / mJ cm-2 Φion / 10-3

8.5 0.47

10.6 0.86

15.2 1.7

19.5 2.1

24.0 2.6

References

(1) X. Ragas, A. Jimenez-Banzo, D. Sanchez-Garcia, X. Batllori, S. Nonell, Chem. Commun.

2009, 2920-2922.

(2) A. Gollmer, J. Arnbjerg, F. Blaikie, B. Pedersen, T. Breitenbach, K. Daasbjerg, M. Glasius,

P. Ogilby, Photochem. Photobiol. 2011, 87, 671-679.

(3) a) K. Tanaka, T. Miura, N. Umezawa, Y. Urano, K. Kikuchi, T. Higuchi, T. Nagano, J. Am.

Chem. Soc. 2001, 123, 2530-2536; b) J. M. Aubry, B. Cazin, F. Duprat, J. Org. Chem. 1989, 54,

726-728.

(4) R. Roy, S. Hohng, T. Ha, Nat. Methods 2008, 5, 507-516.

(5) M. Hara, S. Samori, X. Cai, M. Fujitsuka, T. Majima, J. Phys. Chem. A 2005, 109, 9831-

9835.

(6) J. H. Song, M. J. Sailor, J. Am. Chem. Soc. 1997, 119, 7381-7385.

(7) M. Kuimova, S. Botchway, A. Parker, M. Balaz, H. Collins, H. Anderson, K. Suhling, P.

Ogilby, Nat. Chem. 2009, 1, 69-73.

(8) M. Montalti, S. L. Murov, Handbook of Photochemistry, 3rd ed., CRC/Taylor & Francis,

Boca Raton, 2006.

(9) D. Magde, R. Wong, P. G. Seybold, Photochem. Photobiol. 2002, 75, 327-334.

(10) M. Arık, N. Çelebi, Y. Onganer, J. Photochem. Photobiol. A 2005, 170, 105-111.

(11) T. Miura, Y. Urano, K. Tanaka, T. Nagano, K. Ohkubo, S. Fukuzumi, J. Am. Chem. Soc.

2003, 125, 8666-8671.

(12) X.-F. Zhang, I. Zhang, L. Liu, Photochem. Photobiol. 2010, 86, 492-498.

(13) F. Wilkinson, W. P. Helman, A. B. Ross, J. Phys. Chem. Ref. Data 1993, 22, 113-262.

(14) I. Corral, L. González, J. Comput. Chem. 2008, 29, 1982-1991.