5140
J. Phys. Chem. 1996, 100, 5140-5148
Femtosecond Infrared Spectroscopy of Low-Lying Excited States in Reaction Centers of Rhodobacter sphaeroides Klaas Wynne,† Gilad Haran,† Gavin D. Reid,† Chris C. Moser,§ P. Leslie Dutton,§ and Robin M. Hochstrasser*,† Department of Chemistry, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104-6323, and Johnson Foundation, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104 ReceiVed: September 28, 1995; In Final Form: December 12, 1995X
The mid- to near-IR difference spectra of optically pumped and unpumped reaction centers were recorded over a time range circa 0.1-20 ps allowing us to isolate kinetically the transient electronic spectra of the excited special pair P* and its charge-separated form P+H-. The spectrum of P* between 1 and 5.2 µm exhibits two main bands, one at 5300 cm-1 and another at 2710 cm-1, while P+H- exhibits two bands, one at 2600 cm-1 and another at 8000 cm-1. These new states do not appear as transitions of unexcited reaction centers at E(P*) + 5300 cm-1 or E(P*) + 2710 cm-1 (although the former is close to the QX transition region of the special pair), suggesting that they derive almost exclusively from charge transfer (PL+PM- or PL-PM+) states. This conclusion is strengthened by the anisotropies obtained by pumping P* and probing these transitions with polarized light. The transition of P+H- at 2600 cm-1 corresponds to the known hole transfer transition, and this assignment is confirmed by kinetic and anisotropy measurements: the transition dipole is along the direction of charge transfer between PL and PM. The anisotropy for the transition at 8000 cm-1 is consistent with a transition dipole along the y-axis of PM, in agreement with theoretical predictions for a trip-doublet transition.
1. Introduction The bacterial photosynthetic reaction center (RC) of Rhodobacter sphaeroides is a membrane bound protein that contains eight chromophores approximately arranged around a noncrystallographic C2 axis:1-3 two bacteriochlorophyll-a molecules (PL and PM) forming the so-called special pair, two accessory bacteriochlorophyll-a molecules (BL and BM), two bacteriopheophytin-a molecules (HL and HM), and two quinones (QM and QL). Excitation of the special pair in the lowest absorption band of the RC at ∼870 nm leads to a series of electron transfers with very high quantum efficiency down the L branch of the RC but not down the M branch.4-13 This indicates that the effective symmetry of the electron-transferring system is lowered from C2. To understand this fast and unidirectional electron transfer in the RC, it is of crucial importance to understand the nature of the electronic states that are involved in this process. The lowest excited singlet state of the RC is the lowest component of a pair of excitonic states, PY- and PY+, that are formed from the two singlet QY transitions on the two bacteriochlorophyll-a molecules that make up the special pair. On the basis of experimental evidence14,15 and electronic structure calculations,16-20 it can be concluded that the lowest energy component of the pair of excitonic levels is PY-. Considerable uncertainty remains, however, regarding the higher lying states of the RC. The various existing electronic structure calculations16-24 have two main features in common: (1) that the excited states of the RC are mixtures of excitations of the special pair, the accessory bacteriochlorophylls, and the bacteriopheophytins and (2) that the electronic states are mixed with the charge resonance states of the special pair. In a recent femtosecond time-resolved IR experiment,25 the onset of a band was observed at 1900 cm-1 that was assigned to the transition †
Department of Chemistry. Johnson Foundation. X Abstract published in AdVance ACS Abstracts, February 15, 1996. §
0022-3654/96/20100-5140$12.00/0
between the lower and upper exciton states. In the present paper we will consider the entire transient absorption spectrum of the excited special pair state (PY-) and the charge separated state (P+H-) from 10 000 to 1800 cm-1. By measuring the anisotropy in the experiments and by pumping both at 810 (B*) and 890 nm (PY-), we have been able to assign the experimental (transient) absorption bands. 2. Materials and Methods 2.1. Femtosecond Setup. The experiments were performed with a Ti:sapphire-oscillator and a Ti:sapphire based regenerative amplifier. They have been described in extensive detail elsewhere26,27 and will only be discussed briefly here. The source of our femtosecond pulses is an argon-ion laser pumped Ti:sapphire self-mode-locked laser as described before28 which produces ∼500 mW of 18-25 fs full-width half-maximum (FWHM) sech2 pulses centered at 810 nm. After stretching in a dispersive grating stretcher,29 the pulses are sent into a dispersion-compensated regenerative amplifier, resulting in 3050 fs pulses with an energy of 110 µJ after recompression, at a repetition rate of 4.3 kHz. Femtosecond IR pulses are generated using an optical parametric amplifier (OPA).27,30 Seed light is produced either by parametric generation in a 2 mm long β-barium borate (BBO) crystal cut at 20° or by single-filament white light continuum generation by a 1 µJ pulse in a 2 mm sapphire window. This seed light is recombined with approximately 10-50 µJ of 810 nm light in either a 5 mm BBO crystal cut at 20° (for type I phase matching) or in a 5 mm KTP crystal cut at φ ) 0°, ϑ ) 45° (for type II phase matching) or a 2 mm AgGaS2 crystal cut at 56° (for type I phase matching). With this system, light tunable between 1 and 8 µm and a pulse width less than ∼100 fs FWHM are produced. Probe pulses in the range 800 nm to ∼1.2 µm were made by white light generation in a sapphire plate. In most of the experiments, the RC samples were excited with 100 nJ (anisotropy data) or 200 nJ (magic angle data) pulses © 1996 American Chemical Society
IR Spectroscopy of Reaction Centers
J. Phys. Chem., Vol. 100, No. 12, 1996 5141
at 810 nm. For some of the experiments, pump pulses at ∼900 nm were made by pumping a traveling wave dye laser31-33 using either IR-125 or IR-144 as the lasing medium. In this way 100200 nJ linearly polarized pulses at 890 nm were obtained with a pulse width of approximately 1 ps. The anisotropy of the signals was obtained by rotating the polarization of the pump beam with a half-wave plate and measuring the signal with the pump parallel or perpendicular to the probe beam. The anisotropy of a signal is given by the equation
r(t) )
I| - I⊥ I| + 2I⊥
(2.1)
where I|| and I⊥ are the pump-probe signals with the probe polarization parallel and perpendicular to the pump polarization respectively. To confirm that this setup works as expected, pump-probe experiments were performed on solutions of the dyes IR-800, IR-26, and IR-5 with pump pulses at either 810 or 890 nm and probe pulses at 810 or 1200 nm. The anisotropy obtained with 810 nm pump pulses in IR-5 was 0.39 ( 0.02, nearly 0.4 as expected;34 the anisotropy obtained with 890 nm pulses in IR-5 was 0.36 ( 0.02, the difference from 0.4 being accounted for by the 800 nm half-wave plate in the pump beam. The sample cell consists of a thin (250-50 µm) layer of liquid pressed between two 2 mm thick CaF2 windows and is continuously translated and rotated to avoid sample heating by the laser beams. The pump beam is focused into the sample with an f ) 50 cm lens; the probe beam is focused with an f ) 5 cm off-axis parabolic reflector. The pump beam spot diameter in the sample is ∼250 µm, which is significantly larger than the 100-150 µm diameter of the probe beam. The probe beam is detected after the sample with a Ge photodiode (870-1.7 µm probes) or with a liquid nitrogen cooled MCT detector (1.6-6 µm probes, MCT ) mercury-cadmium-telluride). Spectral resolution is obtained by using either broad-band (∼70 nm) IR interference filters before or after the sample or a 22 cm focal length monochromator with a 300 or 150 L/mm grating blazed for 3 or 6 µm, after the sample. The photodiode signals are sent to a BOXCAR amplifier and digitized with fast A/D converters. The pump beam is chopped at half the laser repetition rate. Before an experiment the zero of delay and the pump-probe cross-correlation width were determined by pumpprobe in a slice of Si mounted behind a window identical to the window used to hold the sample. All data were fitted using a nonlinear-fitting program that fits the parallel and perpendicular signals simultaneously. The uncertainties in the fit parameters reported here are 68.3% joint confidence intervals35 under the assumption that the fit function is approximately linear in the fit parameters at the optimum fit values. 2.2. RC Preparation. RCs were purified from the bluegreen mutant of the photosynthetically grown bacterium Rb. sphaeroides, strain R-26, on the basis of the method of Clayton and Wang,36 using the detergent lauryldimethylamine (LDAO) to solubilize the photosynthetic membranes. An additional fast protein liquid chromatography ion-exchange purification step established a protein purity index (absorbance 280/800 nm) between 1.25 and 1.4. For experiments where the probe laser was tuned between 2.7 and 3.7 µm, bulk H2O was replaced by D2O by successive dilution in 10 mM Tris (pH ) 8), 0.1% LDAO D2O buffer, and the preparation was concentrated down to about 1 mM with Centricon-30 and Microcon-10 centrifugal filters (Amicon). No difference in dynamics or amplitude was observed between H2O and D2O in the region (i.e., 800 nm to 3 µm) where both were sufficiently transparent. Dithiothreitol (1 mM) was added to the solution to establish a reductive redox
buffer and to keep the primary quinone QA reduced. Under these conditions the kinetic state P+Q- is not formed from P+H-, and charge recombination is essentially complete on a microsecond timescale. Free bacteriochlorophyll-a was extracted with methanol and acetone from the lyophilized cells obtained from Rb. sphaeroides R-26 and purified chromatographically on (diethylamino)ethylsepharose CL-6B.37,38 The procedure was performed quickly in the dark with intermediate and final material held under an argon atmosphere to minimize the potential for irreversible lightinduced oxidation of the bacteriochlorophyll. Spectra confirmed the purity of the bacteriochlorophyll before storage in liquid nitrogen. Solutions of bacteriochlorophyll-a in acetone were prepared such that the optical density at 810 nm in a 50 µm path length cell was 1. Under these conditions no signs of dimer formation were present. 3. Results 3.1. Kinetics. As discussed in the methods section, the experiments were performed on two different ultrafast laser systems, using either white light generation or a BBO, KTP, or AgGaS2 based OPA to generate IR probe light in the range 870 nm to 6 µm. High time-resolution and high signal-to-noise transients were taken at ∼30 different probe wavelengths throughout this wavelength region, the main features of which will be discussed below. In most of the experiments presented here, the samples of the RC of Rb. sphaeroides R-26 were excited at 810 nm, and therefore it may be expected that energy transfer from excited B to the special pair will take place. It was found 39 that after excitation of the accessory bacteriochlorophylls the energy is transferred in 120 fs to the special pair. The reason why the energy redistribution time reported in this work is shorter than the 200 fs reported previously40,41 is not fully understood: the present work uses significantly lower excitation energies which increase the preservation of sample quality. A typical transient taken at a probe wavelength of 950 nm (data not shown) shows the rise of the stimulated emission from P* with 118 ( 6 fs and a decay with 3.4 ( 0.1 ps due to electron transfer. In this transient one can thus identify three kinetic states which we will refer to as B*, P*, and P+H- (whether electron transfer is from P* to either P+B- or P+H- is not considered here11). The issue of energy transfer from B* to P* will be discussed in much more detail in a forthcoming publication.39 However, as a result of this energy transfer or energy reorganization process, all the transients presented here show the 120 fs component. The transient obtained with 1.2 µm (8300 cm-1) probe pulses can again be fitted with a three-component model: a component, corresponding to B*, that rises instantaneously and decays with 120 fs,39 a component, corresponding to the excited special pair P*, that rises with 120 fs and decays with 3.4 ps, and finally a component, corresponding to P+H-, that rises with 3.4 ps and stays constant on the experimental time scale (