# CHAPTER 3: RESULTS AND DISCUSSIONS

## 3.1 Characteristics of the System Used

### i) Benzyl Phenylacetate

In order to produce a reasonable concentration of benzyl radicals, it is necessary that the compound chosen as the source should absorb a large amount of the high energy from the flash. Further, since the rate determinations depend on monitoring the absorption of the benzyl radical at 318nm, it is essential that the source of the radicals should not have a strong absorption band near this region. Benzyl phenylacetate was chosen as the source of benzyl radicals because of its simple photolysis to give two benzyl radicals and one molecule of carbon dioxide, and its absorption spectrum (recorded on an SP 700 spectrophotometer using a 0.5cm path length and a concentration of 2.5 × 10-4M) is shown in Fig. 3.1. The spectrum shows a strong high-energy band at 215nm and no absorption at wavelengths greater than 280nm, thus confirming that benzyl phenylacetate is a suitable source.

### ii) Absorbing Species at 318nm after Photolysis of Ester

The spectrum of the transient responsible for the absorption at 318nm was determined as a check on the photolysis of the benzyl phenylacetate. The peak optical density after flashing was measured over a range of wavelengths and the optical density plotted as a function of wavelength (Fig. 3.2.). A comparison with the known spectrum of the benzyl radical (Section 1.4)confirmed that the transient absorption was due to the benzyl radical.

Since the benzyl radical has a narrow absorption band and is known to decay by a second-order process, it was necessary to know the value of the effective extinction coefficient for any particular slit-width on the monochromator when checking the value of the second-order rate constant. The procedure used is described in Section 3.2.

## 3.2 Determination of Effective Extinction Coefficients at 318nm

After each photoflash, it was found that the intensity of transmitted 318nm light from the monitoring source had decreased (see Section 3.3.iii), and so to maintain a constant initial intensity of 1.0V for each decay it was necessary to increase the slit-width of the monochromator. Such a procedure resulted in an increased band-pass and a constant decrease in the effective extinction coefficient, e, of the benzyl radical. Therefore the variation of e with slit-width was determined.

The infinitely-narrow 313nm emission line from a low-pressure mercury lamp was used as the basis of the determination. A range of wavelengths around 313nm was scanned at several slit-widths and the intensity of the transmitted light measured on the oscilloscope. A graph of intensity against wavelength is shown in Fig. 3.3. Each of these traces was approximated to a triangle which was superimposed on the known spectrum of the benzyl radical as measured my McCarthy and McLachlan[73], taking the value of the extinction coefficient at 317.7nm to be 1.92 × 104 M-1cm-1. The average extinction coefficient for each slit-width was then found by performing a series of summations of the products of area elements × extinction coefficients for each wavelength and dividing by the total area enclosed. A graph of mean extinction coefficient against slit-width is shown in Fig. 3.4.

## 3.3 Transient Decays

### i) General Observations

Benzyl phenylacetate was found to be insoluble in water but soluble in both methanol/water (5:1 v/v) mixtures and in cyclohexane. Solutions in both solvents were flash photolysed and a typical decay curve obtained is shown in Fig. 3.5.

The half-lives found for such decays were ca. 500μs and initial optical densities of the benzyl radical were in the range 0.01-0.05, giving initial radical concentrations ([Bz]t=0) of between ca. 0.5-2.5 × 10-7M-1. Assuming the literature values of k2, the second-order rate constant (i.e. about 2 × 109M-1s-1) such initial concentrations should lead to half-lives (t½ = 1/k2[Bz]t=0) of between 2 and 10ms. Such values are far greater than the observed half-lives and hence the assumption of a pure second-order decay must be invalid i.e. the kinetics show a composite first- and second-order decay rate.

### ii) Effect of Reagent Purity

Initial rate determinations were carried out using AR grade materials which were not further purified. After purification by the methods described in Section 2.2, analysis of the decays showed no significant change in the value of the half-life, although estimates of the second-order component (expressed as k2/ε) showed a decrease from (2.8 ± 0.3) × 106 cm+1s-1 with AR materials to (0.73 ± 0.01) × 106 cm+1s-1 with purified benzyl phenylacetate and methanol.

### iii) Nature of Long-Lived Species

As Fig. 3.5 shows, photolysis of benzyl phenylacetate resulted in a permanent absorption by some long-lived photolysis product. This product was suspected to be dibenzyl and was characterized by measuring its UV absorption spectrum on the flash photolysis apparatus. The base-line was found by plotting the intensity of the transmitted light against wavelength over a wide range of wavelengths (230-450nm) before the solution of benzyl phenylacetate had been photolysed. The solution was then flashed ten times and new values of the light intensities were measured. From these measurements the percentage absorptions and optical densities of the product at various wavelengths were found. The spectrum thus obtained was compared to the known dibenzyl spectrum (Table 3.1 and Fig. 3.6) and was in good agreement (except at shorter wavelengths, which was possibly because of the low intensities of light from the quartz-iodine lamp and the consequent inevitability of larger errors).

Since dibenzyl has a high extinction coefficient at 318nm (1.8 × 103 M-1cm-1) it was possible that the decay curves of the benzyl radical were being affected by the build-up of dibenzyl absorption to the maximum i.e. the observed absorption was due to both benzyl radicals and dibenzyl. The dibenzyl build-up would seriously affect the decay if it were built up over a relatively long period (about two half-lives). To check this, it was necessary to observe the dibenzyl build-up independently of the decay of the benzyl radical. This was possible by monitoring the photolysis at a wavelength where the extinction coefficient of dibenzyl was high and that of the benzyl radical low, and the region of maximum dibenzyl absorption (ca. 290nm) was convenient. Monitoring of the photolysis at 290nm showed that no appreciable build-up of dibenzyl occurred after the photoflash, suggesting that most of the dibenzyl was being formed by cage recombination of benzyl radicals (which cannot affect the observed decay curve) and so the only effect of the dibenzyl will be to displace the base-line of the decay (as in Fig. 3.5).

Table 3.1 Absorption Spectrum of Long-lived Product
Wavelength (nm) Io (V) Iabs (V) Log ε *
230 0.012 0.005 3.55
240 0.013 0.006 3.37
250 0.014 0.005 3.28
260 0.015 0.004 3.10
270 0.02 0.006 3.22
280 0.06 0.035 3.50
290 0.21 0.16 3.63
300 0.36 0.20 3.50
310 0.67 0.34 3.40
320 1.00 0.38 3.20
330 0.09 0.008 2.38
340 0.15 0.09 1.175
350 0.23 0.10 1.15
360 0.33 0.10 1.04

* Values of O.D. rescaled to give log ε at 290nm = 3.63

### iv) Analysis of Decays

As expected on the basis of half-life estimates in Section 3.3.i, attempts to fit the decays to pure first- or second-order rate laws were not successful over the full time span of the measurements. Results were analysed for first-order decay by plotting -log(optical density) against time (since -log O.D. = k1t/2.303 - log O.D.t=0) and for second-order decay by plotting (optical density)-1 against time (since 1/O.D. = 1/O.D.0 + k2t/εl) and results for the decay of a 1.5 × 10-3 M solution of benzyl phenylacetate in methanol/water are given in Table 3.2 and Figs. 3.7, 3.8.

In general, it was found possible to obtain a reasonable straight line from the first-order plots if values at small times (when the second-order contribution is very large) were ignored. The value of k1 obtained by this method could then be used to calculate a more accurate value for the second-order rate constant, according to the following analysis:

Rate of reaction with simultaneous first- and second-order decays is

$$-{dc \over dt} = k_1c + k_2c^2$$

Or, since O.D. = εcl

$$-{d \over dt} (O.D.) = k_1 (O.D.) + {k_2 \over \epsilon l} (O.D.)^2$$

Integrating from OD0 at t = 0 to OD at time t gives $$OD^{-1} = \Biggl({k_1 \epsilon l + k_2OD_o \over k_1 \epsilon l OD_0} \Biggl)^{e^{k_1t}} + {k_2 \over k_1 \epsilon l}$$

and hence a plot of OD-1 against $$e^{k_1t}$$ should have a negative intercept whose numerical value is equal to k2/k1 εl from which the previously determined value of k1 can be used to give a value for k2/ε. A plot of this kind for the decay detailed in Table 3.2 is shown in Fig. 3.9, and similar analyses of other decays in both methanol/water and cyclohexane give values for the rate constants as

$$k_1 = (0.78 \pm 0.06) \times 10^3 s^{-1}$$ $${k_2 \over \epsilon} = (1.44 \pm 0.1) \times 10^5 cm^{+1} s^{-1}$$

For typical slit-widths used in these determinations, the value of the effective extinction coefficient was about 0.95 to 1.1 × 104 M-1cm-1, which gives an upper limit to k2 of about 1.58 x 109 M-1s-1 which is on the low side of the literature values (ca. 2.0 × 109M-1s-1).

### v) Lower Limit of Benzyl Extinction Coefficient at 318nm

The amount of dibenzyl produced by each benzyl decay can be easily found by the equation

O.D. (dibenzyl) = εcl

since all quantities are known or measured except the concentration. Assuming the only benzyl radicals produced are those seen in the decay curve, the concentration of benzyl radicals produced is twice the concentration of dibenzyl formed. Hence, knowing the initial optical density of the benzyl radicals, a value for the extinction coefficient of the benzyl radical at 318nm can be found. However, since the main dibenzyl production is probably by cage recombination of benzyl radicals, the concentration of benzyl calculated will be too high, and so such an estimation gives a lower limit for the value of the extinction coefficient. Estimations by this method tended to give results varying between 3700 and 5700 M-1cm-1, although the majority of the calculations suggested a limiting value of (4.6 ± 0.3) x 103 M-1cm-1.

### vi) Flash Photolysis of Dibenzyl

Since dibenzyl has been shown to be formed as the permanent product after photolysis, it was necessary to determine whether subsequent flashing of the solution gave decays complicated by photolysis or electronic excitation of the dibenzyl. As a check on this, a 10-4 M solution of dibenzyl was flashed in both methanol/water and cyclohexane, but no transient absorption was observed in the region of 318nm, even at maximum vertical sensitivity on the oscilloscope.

Table 3.2 Analysis of Benzyl Decay in Methanol/Water
Time (ms) Voltage (mv) O.D. (×10-4) 1/OD -log(OD) ek1t
0.0 79.89362 27.621.44 2.25
0.1 71.37322 31.1 1.49 2.42
0.2 65.34294 34.0 1.53 2.61
0.3 59.45266 37.5 1.57 2.81
0.4 54.55244 41.0 1.61 3.02
0.5 49.29220 45.5 1.65 3.25
0.6 44.61198 50.5 1.70 3.50
0.7 40.69181 55.2 1.74 3.76
0.8 38.57171 58.5 1.76 4.05
0.9 35.58157 63.7 1.80 4.36
1.0 33.20146 68.5 1.835
1.1 31.52 139 71.9 1.85
1.2 29.40 129 77.5 1.89
1.3 27.31 120 83.3 1.92
1.4 25.45 111 89.7 1.95
1.5 23.95 105 94.8 1.97
1.6 22.41 98 102.0 2.01
1.7 21.10 93 107.5 2.03
1.8 19.38 85 117.6 2.07
1.9 17.88 79 126.6 2.10
2.0 16.56 72 138.9 2.14

## 3.4 Redox Studies

### i) Choice of Solute

Choice of metal species as oxidant for the benzyl radical was limited by two factors. Firstly, the species had to be those which were known to participate easily in redox reactions with free radicals, and initial possibilities were cupric salts (CuSO4, Cu(OAc)2, Cu(ClO4)2), ferric salts (FeCl3, Fe(CN)63-, FeCl4-, FeBr2+) and, as a possible alternative to metal compounds, certain organosulphur compounds such as mercaptans and ethyl thioglycolate. The second requirement was that the species used should have no strong absorption bands near the 318nm region used to monitor the reactions, and this proved to be a serious limitation for all the above compounds except those of Cu(II). Ease of solubility was a possible criterion for the choice of oxidant, but no solubility problems were encountered in the case of the Cu(II) compounds mentioned in the concentration range used (ca. 10-4M).

### ii) Analysis of Decays

Each of the metal salts cupric acetate, cupric perchlorate, and cupric sulphate were used in attempts to find pseudo-first-order rate constants for the (schematic) redox reaction

and some decay curves obtained are shown in Fig. 3.10 for the case of a 2.49 × 10-4M cupric sulphate solution in methanol/water. Typical results obtained for each of the cupric salts are given in Tables 3.3, 3.4, and 3.5, and Figs. 3.11, 3.12, and 3.13.

From a theoretical point of view, assuming that the redox reaction is proceeding via pseudo-first-order kinetics, the rate constants obtained should be directly proportional to the concentration of cupric ion. However, the results obtained indicated that, in spite of the variation in cupric concentration, the value of the first-order rate constant was effectively the same in all cases, and, also, the half-life increased to about 900μs. Such results seem to indicate that the reaction being followed was not the straightforward redox reaction expected.

### iii) Determination of the Nature of the Reaction with Cu(II)

Either of two possibilities for the process occurring in the presence of the cupric ion seems feasible. The first possibility is that some complex may be formed between Cu(II) and benzyl phenylacetate, and the observed photolysis may be that of this complex. To ascertain whether this was likely, absorption spectra were recorded on an SP 700 spectrophotometer for the ester alone and for the ester in the presence of cupric ion before and after several flashes (Figs. 3.14, 3.15). No change in the spectra were noted, making complex formation before photolysis appear unlikely.

The second possibility is that complex formation between the benzyl radical and the cupric ion occurs very rapidly after the photolysis of the ester and that the observed decay is a consequence of this complex breaking up. This possibility was tested by obtaining the spectrum of the absorbing species (as in Section 3.1.ii) and comparing it to the spectrum of the benzyl radical obtained by flashing the ester alone (Fig. 3.16). As the spectra show, it is impossible to draw a firm conclusion on this evidence alone about any difference in the nature of the absorbing species in each case.

Table 3.3 Benzyl Decay in Presence of Cu(OAc)2
Time (ms) Voltage (mV) O.D (× 10-4) -log(O.D.)
0.0 36.20 160 1.79
0.2 35.94 159 1.60
0.4 33.78 149 1.83
0.6 31.01 137 1.86
0.8 28.45 125 1.90
1.0 26.73 115 1.94
1.2 25.41 111 1.95
1.4 23.40 193 1.99
1.6 21.24 82 2.03
1.8 18.68 73 2.08
2.0 16.85 66 2.14
2.2 14.51 66 2.18
Table 3.4 Benzyl Decay in Presence of CuSO4
Time (ms) Voltage (mV) O.D (× 10-4) -log(O.D.)
0.0 37.92 168 1.77
0.2 35.43 156 1.80
0.4 33.78 150 1.82
0.6 32.25 142 1.85
0.8 30.05 132 1.88
1.0 27.57 121 1.92
1.2 26.18 115 1.94
1.4 23.47 104 1.98
1.6 22.08 97 2.01
1.8 20.37 89 2.05
2.0 18.54 81 2.09
Table 3.5 Benzyl Decay in Presence of Cu(ClO4)
Time (ms) Voltage (mV) O.D (× 10-4) -log(O.D.)
0.0 88.88 405 1.39
0.2 74.33 336 1.47
0.4 60.77 272 1.565
0.6 48.56 216 1.705
0.8 44.28 196 1.78
1.0 37.70 167 1.84
1.2 32.80 145 1.88
1.4 29.80 131 1.92
1.6 27.42 120 1.965
1.8 24.57 108 2.01
2.0 21.90 97 2.04
2.2 20.66 90 2.085
2.4 18.94 82 2.13
2.6 16.89 74 2.17
2.8 15.43 67 2.19
3.0 14.70 65 2.235
3.2 13.35 58 2.275
3.4 12.21 53 2.305
3.6 11.44 49 2.33

## 3.5 Discussion of Results and Suggestions for Future Work

The most obvious conclusion to be drawn from the results is that the overall situation is far more complicated than the expected simple second-order recombination of benzyl radicals. The nature of the first-order decay involving the benzyl radical was not clarified, but it is likely to be the result of a reaction with some impurity present (probably in the ester, since both methanol/water and cyclohexane give similar values for the first-order rate constant) which could not be removed by the purification methods attempted.

The analysis of the situation pertaining in the case of simultaneous first- and second-order decay described in Section 3.3.iv shows that, despite the kinetic complications of this first-order reaction, a value for k2/ε should be obtainable if k1 can be measured with sufficient accuracy. The graphical estimation of k1 used here gives reasonable results, but the values could be checked in either of two more accurate ways. Firstly, values for k1 and k2 could be found by computer-fitting the data into the equation containing k1 and k2 for each determination, and this method would be expected to give far greater precision in the values of both k1 and k2. Secondly, it seems reasonable that the first-order reaction should have a greater activation energy than the second-order reaction (since the latter is a diffusion-controlled process, which is expected to have a very low activation energy - ca. 3.5 Kcal mole-1) and so an increase in temperature should make the first-order reaction more dominant. Such a procedure would obviously be limited by the relatively low boiling-point of methanol (64°C), but a temperature increase of about 25° would be expected to increase the rate by a factor of approximately four, and would enable k1 to be determined more accurately and more convincingly. A series of such determinations at different temperatures would enable values of k2/ε to be extracted (as previously) as a function of temperature, and the value at 25°C could be obtained by extrapolation from a plot of log(k2/ε) against T-1 which is expected to be linear over the temperature range considered. If such more accurate values of k2/ε were obtained they would serve as a useful check on the literature values of the extinction coefficient at 318nm.

With regard to the possibility of studying cage recombination of the benzyl radicals as a function of temperature, it was decided, on the basis of the decay curves obtained with purified materials, that such an intended study would prove difficult. The amount of dibenzyl being produced was very small and consequently measurements of the dibenzyl optical densities would be susceptible to large errors, thus making any results obtained be necessarily of uncertain value.

In the case of the results in the presence of cupric ion, the nature of the decays obtained strongly suggests that something other than the decay of the benzyl radical is being followed. The effect of adding the salt should be to decrease the half-life of the decay, whereas an increase was observed in every case. Such an effect seems to be explicable in only two ways. The cupric salt could be functioning in some way so as to remove the impurity responsible for the first-order decay occurring in the absence of the cupric ion, and the decays observed are due to the subsequent reaction between Cu(II) and the free radical, but the independence of the first-order rate constant on cupric ion concentration appears to refute this. The second and most likely possibility has already been mentioned viz. that the observed decays are those of some cupric complex having a half-life of about 900μs. Formation of some species such as copper benzyl appears to be a reasonable suggestion, and the results can be rationalized on this basis if it is assumed that such a complex is formed very rapidly (within 100μs) after the initial photolysis of benzyl phenylacetate and that the complex has an absorption band around 318nm. The general impression obtained was that initial optical densities in the presence of the cupric ion were greater than those with the ester alone, which would suggest that the extinction coefficient of the complex is greater than that of the benzyl radical at 318nm i.e. it is greater than 104 M-1cm-1.

Such an explanation obviously requires further experimental investigation into the nature of the reaction. The kinetics of the reaction should be investigated over a wider range of concentrations (say 10-3 to 10-5 M) and the apparent independence of the first-order rate constant on cupric ion concentration confirmed or disproved. If this is found to be confirmed, it would remain to determine more satisfactorily the nature of the intermediate complex and to find its UV absorption spectrum more accurately in order to differentiate it sufficiently from that of the benzyl radical (this could probably be determined by using a flash spectrographic detection for each transient). The existence of copper alkyls has been proposed previously, and the suggested formation of copper benzyl is a consistent deduction from the data available at present, but, as stated, the problem warrants further attention before becoming a completely acceptable explanation of the photolytic reaction.