J . Chem. SOC., Faraday Trans. I , 1987, 83, 29-35 Two-dimensional Transient Electron Spin Resonance Spectroscopy Keith A. McLauchlan" and David G. Stevens Physical Chemistry Laboratory, South Parks Road, Oxford OX1 3QZ A two-dimensional field-time transient electron spin resonance spectrum is reported of the cyclohexyl- 1-01 radical. The spectrum shows linewidth alternation effects due to a ring-flip process. The radical is polarized and its spectrum displays symmetric E/A characteristics at early times, showing the radicals to originate in the reaction of a triplet state, but these invert to an A/E pattern in time. At longer times still the spectrum exhibits excess emission. This represents a fourth case of radicals whose spectra show these time-dependen t patterns. Using the continuous-wave (c.w.) and spin-echo methods developed in recent years, the high-resolution electron spin resonance (e.s.r.) spectra of transient free radicals can be obtained at times as short as 30 nslf after the radicals are created.When observed soon after radical formation most spectra exhibit electron spin polarization, which originates in chemically induced dynamic electron polarization (CIDEP) proce~ses.~ The two CIDEP mechanisms, the triplet mechanism (TM) and the radical pair mechanism (RPM), produce characteristic intensity distributions which allow the multiplicity of the radical precursor to be deduced. The radicals are identified via the positions of the lines in the spectra, these being unaffected by the CIDEP processes. As time elapses the initial intensity patterns change owing to a variety of physical and chemical processes and the observation of this time-dependence yields further information an the system, Most obviously, the initial polarized intensities diminish in absolute magnitude as a result of spin-lattice relaxation; in C.W.experiments the rate is enhanced by microwave-driven transitions. Often, as the radicals which were created together in geminate cages separate, the identity of the radical pairs formed by random diffusion (F pairs) differs from that of the original pair. These yield a different polarized intensity distribution which continues to be generated whilst reactive radicals persist in the solution and which discloses the identities of these radicals. Even where the type of radical does not change in time, there is growing evidence that for a given pair of radicals, the polarization pattern produced in a triplet geminate pair differs from that observed at later times in an F pair; this is contrary to accepted polarization theory.If the primary radical reacts within its lifetime to create a secondary species, this too may be identified. As a final example, it has been discovered that in C.W. experiments the development of the spectra of spin-polarized radicals in time is unusually sensitive to electron and proton exchange To observe such phenomena as these, the spectrum must be monitored for long periods of time following radical creation. Here C.W. methods have considerable advantages over the spin-echo one7 in that the intensity of a spin-echo falls rapidly with time (at the spin-spin relaxation rate) after radical creation, and also the measurable signal (the echo itself) is of very short duration. With C.W.methods a continuous signal may be sampled over a long period of time, and the signal height falls in time at an effective spin-lattice relaxation rate; in our experience b T, in solutions of normal viscosity. 2930 Two-dimensional Transient E.S.R. The C.W. methods, like the spin-echo one, are sampling techniques. Since it is not possible to sweep through a spectrum in a short time (cf. 30 ns) the spectrum is obtained by creating the radicals repetitively as the magnetic field is varied either continuously or in discrete steps. In the former case a sample, over a chosen period of time, is taken with a box-car averager and the continuous output from this as the field is swept constitutes the spectrum;8 it is important to realize that only that information pertinent to the chosen time period is recorded.In the latter, the entire decay curve is stored in a dedicated microcomputer and the period for which the sample is required is selected afterwards. In the time-integration spectroscopy (TIS) technique9 the signal is digitally summed over the chosen period and this sum is entered into an address which corresponds to a specific value of the magnetic field. As operated until recently, the contents of the store have then been destroyed before the field advances to its next value and the process is repeated. A curious feature of the TIS experiment has consequently been that all the information to define the time-behaviour of the system completely has been obtained, but most of this has been rejected subsequently.To obtain the TIS spectrum at a different time after radical creation has necessitated repeating the whole experiment, involving thousands more photolysis flashes. Moreover where the spectrum varies rapidly in time important behaviour may be missed by an unfortunate choice of sample periods. It is consequently better to store all the information obtained at each field value and to plot its intensity versus time and field on a three-dimensional surface. Following n.m.r. usage we define this as a two-dimensional plot since there are two independent experimental variables. Method The experiment has been described in a recent publication7 and is based upon our existing TIS apparatus to which a 10 Mb Winchester disc data store has been added.The radicals are created in U.V. pulses at 308 nm from a Lambda Physik EMG 103 MSC excimer laser; since our first paper an output control system has been added to ensure that the mean pulse energy (of ca. 50 mJ per pulse) does not vary during data acquisition, a period of ca. 2 h in the experiments reported here. The deoxygenated sample flows continuously through the quartz irradiation cell to avoid sample depletion. At each field position a chosen number (128) of transient decays are recorded and signal-averaged by the dedicated microprocessor described previously. The data are then output, together with information on the field position and a scaling factor, to allow them to be compared with subsequent data, to a PDP 1 1 /23 microcomputer where they are stored on the 10 Mb disc.The flash repetition rate is limited to 20 Hz by data- handling considerations. After data transfer, the data and add buffers of the micro- processor are cleared, the field is stepped and data collection recommences on the next laser pulse. The cycle is repeated until all the field addresses have been filled. The experi- ment is controlled by extensive software. In the spectra shown below 512 field points have been used and at each a decay curve has been obtained which consists of 2048 channels of 16-bit data, a total of ca. 2 Mbytes of information. This was transferred from the PDP 11 /23 to a Norsk Data ND-520 minicomputer for data manipulation. Whenever a laser beam strikes a tuned e.s.r.cavity it produces a spurious long-lived signal (even off-resonance) which originates in photoelectron emission and sample and cavity heating. At constant laser power the signal is reproducible at all field positions. It has consequently been subtracted from each decay curve used to construct the surface shown below. In addition pre-trigger subtraction has been employed to ensure that only pulse-correlated information is displayed.K . A . McLauchlan and D. G . Stevens 31 Fig. 1. A field-time-intensity surface of photolytically generated cyclohexyl- 1-01 radicals. The spectrum is spin polarized by the ST, RPM process to yield an E/A pattern at early times after radical creation, which changes to an A/E one later in time. Attention is also drawn to the marked alternating linewidth effect, which originates in the modulation of j? couplings by ring flipping.Results and Discussion Fig. 1 shows the two-dimensional field-time spectrum of the cyclohexyl-1-01 radical produced by photolysis of cyclohexanone in cyclohexanol. Its signal-to-noise ratio is sufficient that the normal spectrum at a specific time can be obtained simply by taking a point time sample (corresponding to a 10 ns sample period) across the field span (fig. 2). We stress that the surface and the derived figures display direct measurements of the magnetization in the y-direction of the rotating frame, and not its integral as in our TIS experiments. With weaker signals, time-integration can of course be applied to the contents of the data store, without the need for further experimentation, to yield TIS spectra.This can be invaluable too in studies of the variation of the absolute signal height with time, where equal integration periods are essential to compare signal heights at different times. In normal TIS experiments this integration window is often varied with time after radical creation to optimise the signal-to-noise ratio. As is obvious from the figures, the spectrum at earliest time is polarized and displays the emission at low field/absorption at high field (E/A) characteristics of a radical produced by reaction of the triplet state of the precursor ketone. This is consistent with a previous continuous-irradiation C.W.study of CIDEP in this radical. lo The reaction is consequently The spectrum when first observed is completely symmetrical in its intensities and exhibits the characteristics of the action of a pure ST, RPM polarization mechanism between identical radicals. The outermost lines (which would be weakest in the unpolarized 2 FAR I32 Two-dimensional Transient E.S.R. (a) absorption f I t II field 161 - 20G Fig. 2. Spectra obtained by taking a single point sample, at each of two chosen times after radical creation, from the surface shown in fig. 1. (a) At 2.5 ps, where the spectrum shows a symmetric E/A pattern and (b) at 75.0 ps, where the intensities have inverted to yield an A/E pattern. The spectra have been normalized to the same overall amplitude. spectrum) are strongly accentuated, whilst the central line (the strongest in the unpolarized situation) is of low intensity.Its finite width in this spectrum removes some of its intensity away from the exact centre of the spectrum and yields small signals in emission at low field and in absorption at high field. The appearance of the other lines is discussed below. Rather unusually, there is no contribution from the TM to the observed polarization. This is significant for previous studies of (CH3),cOH11 and (CH3),CHl2? l3 radicals displayed overall similar polarization characteristics to those discussed here, but with excess absorption at early times; it now appears that this originated in the TM process and that the early time behaviours of these systems are consistent with theory.At longer times after the flash [fig. 2(b) and 31 the nature of the electron spin polarization changes completely and, as with all other RPM-polarized radicals emanating from reactions of triplet states which have been monitored for a sufficiently long time, an absorption/emission (A/E) pattern is observed. This occurs with the radicals mentioned above besides (CH3)3c radicals.’* This is not as expected on simple theoretical grounds if this polarization arises in the encounters of freely diffusing radicals to form ‘F pairs’. As before in between the initial E/A and later A/E patterns the signals have low absolute intensities as the two opposite phases of polarization tend to cancel, whilst relaxation diminishes the large initial signal.In this region a completely absorptive spectrum is observed (but cannot be discerned clearly in fig. 1). The origin of the inversion of the phase of polarization with time remains incompletely understood. It is apparent that spin polarization continues to be generated at long timesK. A . McLauchlan and D . G. Stevens 33 - field absorption I c 1 emission 2OG Fig. 3. A time integration spectroscopy (TIS) spectrum of the cyclohexyl-1-01 radical obtained with sampling and summation between 160.0 ps and 172.5 ps after radical creation. It exhibits a basic A/E pattern, but shows excess emission. after the flash that creates the radicals since the A/E signal continues to grow in for some time, and the timescale suggests F pairs as its only reasonable source. Independent experiments seem to have confirmed this directly for (CH3)3c radicals,14 although other evidence has been submitted which apparently shows that radicals from F pairs observed at early times do yield the expected E/A pattern.15 The growing generality of the phenomenon makes it difficult to sustain the only model so far suggested to explain it, which was based upon a variation in the sign of the electron exchange interaction with inter-radical separation together with a difference in trajectories between identical radicals forming geminate and F pairs.16 It may be that electron or nuclear relaxation effects should be considered.However, a further advantage of the two-dimensional display is to allow a clear indication of the variation of the signal heights with time right across the spectrum. This allows an apparently facile comparison of the absolute magnitudes of initial and F pair polarizations which will yield further insight into this problem. To model such behaviour requires the calculation of the whole surface and, although it is fully defined through the Bloch equations, a considerable computational effort is required; it is already underway.At longer times, the polarization behaviour of the radicals has been investigated using our normal TIS technique (fig. 3). As with the radicals mentioned above, the spectrum then exhibits excess emission. The timescale of the observation suggests that the origin of the effect may be chemical. It is possible that the relative concentration of oxygenated radicals, for example peroxy radicals formed by oxygen-scavenging of primary radicals in incompletely deoxygenated solutions, becomes significant in time and that the nature of the majority radical pair alters. The oxygenated radical is not observed, possibly because of line broadening due to rapid relaxation, but an ordinary ST, RPM process with such a radical would yield excess emission in the cyclohexyl- 1-01 radical, as a result of the difference in g values between the two.Experimental attempts to confirm this have so far failed, and the phenomenon may well be an experimental artifact. For a radical with few hyperfine lines in its spectrum the display of the field- and time-dependence given by the surface in fig. 1 gives a very clear indication of its identity 2-234 Two-dimensional Transient E.S.R.absorption . . _ .. emission - 4 I I - 20 G field absorption a ._I - Fig. 4. A contour plot constructed from the surface shown in fig. 1. The peak positions are identified by a concentration of contours, with the magnetic field plotted horizontally, and the time vertically. In this black and white version of a colour-coded original information on the phases of the signals has been lost although reference to the previous figures allows them to be recognized. The contours have been drawn at equal intensity intervals with a grain chosen to illustrate the smallest peaks. In this way each line observed soon after the flash is seen to pass through near zero intensity to re-emerge later in time (actually in the opposite phase). and behaviour.Unfortunately as the complexity of the spectrum increases such a plot requires many more field points to define the peaks adequately and the time required to record the surface increases. At the same time the peaks become difficult to distinguish in this type of plot. It becomes convenient to use a different mode of display, the contour diagram. That corresponding to the surface shown in fig. 1 is given in fig. 4. In it the field is plotted horizontally, and the time vertically, and the peaks, which on the surface would project into (for emission) or out of (for absorption) the plane, are indicated by contours of equal height drawn through them. In the original the contours are colour-coded, with different colours used for emission and absorption, but this is not reproduced here.This type of display gives a very clean and clear picture but must be interpreted with caution, particularly with respect to the times at which the various lines apparently attain zero intensity. Even in spectra in which all the lines have identical effective relaxation times, this null point occurs at different instants owing to the different initial magnitudes of the various lines. These depend, of course, upon the CIDEP process occurring. The apparent null point also depends upon the contour density. It remains to discuss the striking linewidth alternation visible in these spectra. This originates in an interconversion between degenerate conformers of the radicals, with modulation of the couplings to protons B to the radical centre. The ring flip is not sufficiently fast to cause complete averaging of the positions of lines which vary in the process.Previous workers,17* * who studied the radical produced by continuous irradiation of a solution of di-t-butyl peroxide in cyclohexanol, were able to stop the interconversion by cooling to 183 K and the individual couplings to two pairs of identical /3 protons were measured. From the values obtained (35.5 and 10.3 G) it was deduced that the radical existed in a twisted-chair conformation at this temperature. AtK. A . McLauchIan and D . G. Stevens 35 higher temperatures the interconversion probably involves both twisted-chair and normal-chair conformations. In none of the previous observations of the radical has the linewidth alternation been observed as clearly at room temperature as in our experiments.To some extent this may reflect the different solutions used but it also owes something to the increase in intensities of the lines in spin-polarized radicals as compared with equilibrated ones. In this preliminary communication we have sought to introduce some of the important features of the polarized spectrum of the cyclohexyl- 1-01 radical, although our main purpose has been to demonstrate the two-dimensional e.s.r. method. We hope to have shown its great potential in studying a range of chemical and physical processes. A detailed study of the polarization behaviour in this and similar radicals, and of the interconversion kinetics, is in progress both experimentally and theoretically. D. G. S. is grateful to the S.E.R.C. for financial support. References 1 A. D. Trifunac and J. R. Norris, Chem. Phys. Lett., 1978,59, 140. 2 K. A. McLauchlan and D. G. Stevens, unpublished work. 3 A recent review is by C. D. Buckley and K. A. McLauchlan, Mol. Phys., 1985, 54, 1. 4 S. Basu, K. A. McLauchlan and A. J.,D. Ritchie, Chem. Phys. Lett., 1984, 105, 447. 5 K. A. McLauchlan and A. J. D. Ritchie, Mol. Phys., 1985, 56, 141. 6 K. A. McLauchlan and A. J. D. Ritchie, Mol. Phys., 1985, 56, 1357. 7 K. A. McLauchlan and D. G. Stevens, Mol. Phys., 1986, 57, 223. 8 e.g. A. D. Trifunac, M. C. Thurnauer and J. R. Norris, Chem. Phys. Lett., 1978,57, 471. 9 S. Basu, K. A. McLauchlan and G. R. Sealy, J. Phys. E, 1983, 16, 767. 10 P. B. Ayscough, T. H. English, G. Lambert and A. J. Elliott, Chem. Phys. Lett., 1975, 34, 557. 11 S. Basu, A. I. Grant and K. A. McLauchlan, Chem. Phys. Lett., 1983,94, 517. 12 A. I. Grant and K. A. McLauchlan, Chem. Phys. Lett., 1983, 108, 120. 13 K. A. McLauchlan and D. G. Stevens, J . Mugn. Reson., 1985,63,473. 14 I. Carmichael and H. Paul, Chem. Phys. Lett., 1979,67, 519. 15 M. C. Thurnauer, T-M. Chiu and A. D. Trifunac, Chem. Phys. Lett., 1985, 116, 543. 16 A. I. Grant, N. J. B. Green, P. J. Hore and K. A. McLauchlan, Chem. Phys. Lett., 1984, 110, 280. 17 C. Corvaja, G. Giacometti and M. Brustolon, 2. Phys. Chem. (Frankfurt am Muin), 1972,82, 272. 18 C. Corvaja, G. Giacometti and G. Sartori, J. Chem. SOC., Furuduy Trans. 2, 1974,70, 709. Paper 6/823; Received 28th April, 1986