6924
J. Phys. Chem. 1993,97, 6924-6933
Photochemical Hole-Burned Spectra of Protonated and Deuterated Reaction Centers of Rhodobacter sphaeroides P.A. Lyle, S.V. Kolaczkowski, and G. J. Small’ Ames Laboratory-USDOE and Department of Chemistry, Iowa State University, Ames. Iowa 5001 I Received: January 28, 1993; In Final Form: April 2, 1993
Photochemical hole-burned spectra with improved signal-to-noise ratio (X20) are reported for the protonated and deuterated reaction center of the purple bacterium Rhodobacter sphaeroides. Spectra obtained as a function of burn frequency (WB) establish that the lifetime of P870*, the primary electron-donor state, is invariant to location of wg within the inhomogeneous distribution of P870 zero-phonon line transition frequencies. For both the protonated and deuterated RC, which exhibit P870 absorption widths at 4.2 K of only 440 and 420 cm-l, the zero-phonon holes yield a lifetime of 0.93 f 0.10 ps. This lifetime is independent of temperature between 1.6 and 8.0 K (range over which the zero-phonon hole could be studied). The invariance of the P870* lifetime to WB and other data indicates that the nonexponential decay of P870* (Voset al. Proc. Natl. Acad. Sci. U.S.A. 1991,88,8885) is due neither to a distribution of values from the electronic coupling matrix element associated with electron transfer, which one might expect from the normal glasslike structural heterogeneity of the RC, nor to gross heterogeneity. The higher quality of the hole spectra has allowed for more stringent testing of the theoretical model previously used to simulate the P870 hole profiles and absorption spectrum. Although the essential findings reported earlier (see, e.g., Reddy et al. Photosyn. Res. 1992, 31, 167) are not altered, it is concluded that the modeling of the distribution of low-frequency phonons (mean frequency -30 cm-I), which couples to P870*, in terms of a Debye distribution is inadequate. The anomalous low-frequency modes of glasses and polymers are suggested to be important also for proteins.
I. Introduction The structure that underlies the broad (- 100-500 cm-l at 4.2 K) absorption profiles of the photocatalytic excited state (Qy) of chlorophylls and pheophytins is relevant to the energy- and electron-transfer dynamics of photosynthetic units, e.g., temperature dependence, dispersive kinetics, coherence effects. Spectral hole burning has proven to be a versatile tool for the determination of such structure (for recent reviews see refs 1-3). The burn wavelength dependence of the hole profile yields the magnitudes of the site inhomogeneous broadening (I’I) and homogeneous broadening (I’H) contributions to the Qyabsorption profile. Studies of many protein complexes have yielded I’I 50-200 cm-1, depending on the complex and Qy band. Given the normal glasslike heterogeneity of the protein,4.s most of I’I is intrinsic, rather than solvent- or detergent-induced. That is, the existence of a large number of conformational substates of the protein6 leads to a broad distribution of transition frequencies. An important contributor to I’H is the linear electron-phonon coupling. Based on hole-burning studies on complexes of purple bacteria, cyanobacteria, green algae, and photosystem I and I1 of green plants, it is clear that coupling to a broad distribution (-40 cm-1) of protein phonons with a mea0 frequency of am = 20-30 cm-1 is u b i q u i t o u ~ . Furthermore, ~.~ the coupling strength is weak (S < 1) for antenna chlorophylls and moderately strong (S 2) for the special pair or primary donor state of reaction centers. Although not pertinent to this paper we note, for completeness, that, for antenna protein complexes characterized by structural subunits containing strongly exciton-coupled chlorophylls, the contribution to I’H from exciton level structure/ ultrafast interexciton level relaxation can be very signifi~ant.~-ll In this paper we present photochemical hole-burned (PHB) spectra of the bacteriochlorophyll (BChl) special pair or primary donor state Qy band, P870, of Rhodobacter sphaeroides. The results were obtained with a new experimental setup that has resulted in a 20-fold improvement in S/N ratio relative to
-
-
previously reported spectra from this laboratory.12-16 The main objectives of our studies were to further test certain assumptions and the values of the theoretical parameters used in previous work and simulations17of the hole spectra and to generate data pertinent to the question of dispersive kinetics for the primary charge-separation p r o c e ~ s . ~ * J ~ Our earlier studies, which are reviewed in ref 2, established that the absorption transition to P870* (asterisk denoting lowest excited Qydimerstateof thespecial pair, i.e., the primaryelectrondonor state) couples quite strongly to a 120-cm-l mode with an Svalue of 1.5. This mode was assigned17as an intermolecular mode of the special pair since the largest S value for the intramolecular modes of Chl monomers is -0.05.8.20 In the uppermost row of Table I the frequency and S value are denoted as wSpand S,,(sp special pair intermolecular “marker” mode). For the protein phonons, w, = 30 cm-l and S = 2.2. The values for P960*, the analogous state for Rhodopseudomonas viridis, are given in the second row of Table I. Some of the dynamical implications of these results are reviewed in ref 2. For example, it was argued that, along with the 30-cm-I phonons, the marker mode should be an important contributor to the reorganization energy associated with the initial phase of charge separation as well as the temperature dependence21 of this separation. On the other hand, it was suggested that the marker mode is not a promoting mode for charge separation since, if it were, it would lead to a coth( Aw,,/2kT) dependence for the kinetics, in contrast to the findings thatzI.22 the kinetics for Rb. sphaeroides and Rps. viridis speed up by a factor of 2 and 4 as the temperature is reduced from room temperature to 10 K. Another finding was that”16 the zero-phonon hole (ZPH) widths of P870 and P960 yielded P870* and P960* decay times equal, within experimental uncertainty (f0.2 ps), to those measured in the time domain in the low-temperature limit.21.22 This is important because the ZPH measures decay from the total zerepoint level of P*, whereas in the time domain measurements it is the marker and phonon modes that are predominantly excited. The agreement in the
-
-
0 1993 American Chemical Society
The Journal of Physical Chemistry, Vol. 97, No. 26, 1993 6925
Reaction Centers of Rhodobacter sphaeroides
WAVELENGTH(nm) 830 780
TABLE I: Structure of P870 end p9W w,n/sanc
P960 P870
13511.1 11511.5
wm/S
2512.1
3012.2
r
rr
40 30
170
910 1925’
120
r
890
730
860
4 From ref 16. bunits of wap, wm, I’, r1, and vm are in cm-l. Homogeneous width of the ZPL for the marker mode level wlpl is 50 cm-1 and j (50 cm-1) for the w,d 0’ L 2) levels. One-phonon profile on low- and high-energy sides is a Gaussian and Lorentzian, respectively;
C
cf. ref 16.
decay rates is consistent,with thermalization of the relevant lowfrequency modes preceding charge separation, an assumption in widely used electron-transfer theories.23 Until recently,’* the data from hole burning had not been used to address, theoretically, any facet of charge separation. Small et al. simplified the usual nonadiabatic expression for the electrontransfer rate to a form valid in the strongelectron-phonon coupling limit. As a result, they were able to study the effect of a distribution of values (stemming from normal glasslike structural heterogeneity) for the relevant P*-charge separated state energy gap on the T-dependence of primary charge separation. Their work also examined whether or not the distribution was sufficiently broad, relative to the “homogeneouswidth”of the phonon-nuclear factor associated with electron-phonon coupling, to yield P* decay kinetics sufficiently dispersive to account for the nonexponential kinetics recently reported for P870* at room temperature.2’29 It was concluded that it is not. It was suggested that a distribution of values for the pure electronic coupling matrix element Vmight be responsible for the nonexponentiality. This suggestion is investigated in this paper by means of experimental results that establish, for the first time, whether there is a dependence of the P870* lifetime on the burn laser frequency as it is tuned through the inhomogeneous distribution of P870 zero-phonon line absorption transition frequencies. The superior quality of the hole spectra allows us to also study, in greater detail, the underlying structure of P870.
II. Experimental Section
11000
12000 13000 WAVENUMBERS (cm-l)
-
WAVELENGTH(nm) 830 780
910
925
1.0
890
860
730
n
I
Samplesof deuterated wild type Rb. sphaeroides were prepared for the hole-burning experiments by diluting the reaction centers (RC) into a glycero1:water glass-formingsolvent along with 0.1% (w/v) LDAO (1auryldimethylamineN-oxide),20 mM Tris/HCl pH 8.0, and 250pM ascorbate. Protonated R-26 Rb. sphaeroides RC were also dissolved in glycerol:water, but with 0.05% Triton X-100 and 20 m M Tris/HCl pH 8.0. UQ2 (2,3-dimethoxy-5decylbenzoquinone),40 pM,was added to the protonated RC to refill any of the Qp sites left vacant in the isolation procedure. For both samples the room temperature reversible bleaching of the P870 band was used to indicate the extent of the Q. sites being filled. The photochemical bottleneck state is P+Q- (Q quinone), which relaxes back to PQ in 10 ms at 4.2 K. The optical density (OD) of the samples in this study was typically 0.25-0.4 at the maximum of P870. Burn irradiation (line width = 0.03 cm-’) was provided by a Coherent 899 Thapphire CW ring laser pumped by 15 W of the visible multiline output of a Coherent Innova Ar+ laser. Typical output at 870 nm was 200 mW, which was reduced to 510 mW/ cm2 with a variable neutral density filter to keep bleaching of the P870 band maximum at 4.2 K below 20%. Bleaching of up to 60% was possible with higher intensities. Absorption and photochemical hole-burned spectra over the range 24 OOW3750 cm-l were taken at 2-cm-l resolution with a Bruker IFS 120 high-resolution Fourier transform (FT) spectrometer. (Higher-resolution scans were not necessary.) Hole spectra are defined as the difference in absorbance with the burn laser on and off. The improved signal-to-noise ratio (SIN) is primarily due to the FT spectrometer, scan averaging for -8 min (200 scans), and with the use of improved sample handling
-
11000
12000 13000 WAVENUMBERS (cm-I)
F i p e l . (A, top) 4.2KabsorptionspectrumoftheQvregionofdeuterated RC of Rb. sphaeroides (wild type, strain 2.4.1). Inset shows simulation of P870 (fwhm = 420 cm-l) with parameter values listed in Table 11. (B, bottom) 4.2 K absorption spectrum of the Qv region of protonated RC of Rb. sphaeroids (R-26 strain). Inset shows simulation of P870 (fwhm = 437 cm-1) with parameter values listed in Table 11.
techniques. Only glasses of very high quality were used in order to avoid interferenceofscattered laser light (which passes through the interferometer before impinging on the detector) with the zero-phonon hole (ZPH) of the hole spectra. As emphasized in our earlier work,I2-l6 utilization of highquality RC samples and judiciously chosen host media/detergents, which minimize the site inhomogeneous broadeningof the P band, yield the most structured hole spectra. Typical 4.2 K absorption spectra for the protonated and deuterated RC samples used in the present study are shown in Figure 1. The fwhm of P870 is 440 and 420 cm-1, respectively, about 40 cm-1 sharper than in the highest-quality spectra previously reported.
III. Theory of the Hole Profde The theory of Hayes and Small28.29 (the basic Hamiltonian employed by Won and Friesner30.31 is identical to ours), which we employ, has been reviewed several In ref 2 a nonmathematical discussion of the underlying physics and spectroscopy is given along with model calculations, which illustrate how the single site absorption profile and site heterogeneity are related to the hole profile. We give here only a brief
6926 The Journal of Physical Chemistry, Vol. 97, No. 26, 1993
Lyle et al.
account of the theory valid in the low-temperature limit. A hole profile expression valid for all temperatures has been derived by Lyle.32 The absorption profile of a single site with a ZPL frequency of u can be written as
925
WAVELENGTH(nm) 890
860
L(Q - u ) =
-0.044
where j runs over the discrete pseudo-localized or localized phonons33 and, if necessary, the intramolecular Chl modes. S, and w j are the Huang-Rhys factor and frequency for the jth mode. The sum over r is isolated because it is meant to represent the contribution from the essentially continuous distribution of low-frequency “phonons” of the protein network. The function 1 is the line shape function. For rj = 0 and r = 0, L describes the ZPL associated with the total zero-point level of the excited electronic state. The ZPL is a Lorentzian with a homogeneous width we denote as y. As discussed elsewhere,12-16 we may set y (cm-l) = (2ATC)-l for P870* in the low-Tlimit, where T is the lifetime of P870* and c is the speed of light. The sequential nonzero values of r = 1, 2, 3, ... correspond to the one, two, ... phonon profiles obtained by convolving the one-phonon profile with itself r times.28329 The width of the one-phonon profile centered at wm (relative to the ZPL) is defined as I’. Guided by experimental data from the one-phonon profile in organic crystals, we have previously used Gaussian and Lorentzian line shapes for the low- and high-energy sides of the one-phonon profile rather than an unphysical Gaussian pr0file.3~ Within the harmonic approximation, the phonon sidebands that build on the ZPLs associated with total zero-point and the discrete modes are identical. In our previous simulations of P870 and its hole spectra we considered only the phonons and one discrete mode, the special pair marker mode with Huang-Rhys factor S,, and frequency os,(cf. Table I), because the hole spectra did not indicate the existence of more than one discrete mode. To obtain the absorption spectrum, the single site (RC) profile is convolved with a Gaussian zero-phonon site excitation distribution function centered at vm with a fwhm of I’I:
A , = Jdv No(v - v,,,)L(Q- u) where NO(U- v m ) / N is the probability of finding a site with a zero-phonon transition frequency equal to u. If a narrow band laser (the laser line width used in the present studies is less than 1/1000 of the homogeneous width of the ZPL) is tuned to WB with an intensity Z for a time T , the number of sites that remain a t frequency u is given by N,(u - vm) = No(uV m ) exp{-[uZt$tL(wB - u ) ] } , where u is the absorption cross section and t$ is the holeburning quantum yield. The absorption spectrum after burning is then A , = Jdu N,(u - u,) exp{-[uZqhL(wB- u)]L(Q- v))
(3)
The expression is valid for photochemical hole burning (PHB) and population bottleneck hole burning, but not nonphotochemical hole burning where the contribution to A , from the antihole must be taken into account. The hole-burned spectrum is defined as A , - Ao. With reference to the caption of Table I, we comment on the value of 50 cm-l previously used for the homogeneous width of the ZPL associated with the one-quantum transition of the marker mode, wspl. This value, which corresponds to a total dephasing time of 200 fs, was mandated by two observations: (1) burning a ZPH into the origin (usPo)band did not produce discernible
2 d 5w
-0.06-
a
-0.08-
10800
11000 11200 11400 11600 WAVENUMBERS (cm-l) Figure 2. PHB spectra of P870 (protonated, R-26) for wg = 10 921, 10 992, and 1 1 039 cm-I (top to bottom), T = 4.2 K. 16 = 10 mW/cm2 for all spectra. The %-absorbance change for the upmost spectrum measured at the peak of P870 is 6.1%. Simulations (3of these spectra were obtained with the parameter values of Table 11. The dashed arrow indicates the position where the satellite ZPH of olpl would occur if the marker mode were not damped; cf. text. The inset is a blowup of the region around the ZPH for the middle spectrum.
higher energy satellite ZPHs associated with the one, two, ... quantum transitions of the marker mode, and (2) burning directly into the region of the one-quantum transition did not produce a ZPH coincident with OB or a satellite ZPH in the region of the wspO band. Of course, if the marker mode feature of the hole spectra is due to two or more closely spaced and unresolved modes, then a dephasing time quite as short as 200 fs would not need to be invoked; vide infra. With regard to the simulations presented later, we emphasize that the wB-dependent hole spectra themselves provide the ZPH width and good first approximation for om,S, wsp, and Ssp.A reasonable estimate of I’I is available since the width of P870 is =r1+ &91ui.29 Thus, the simulations of the we-dependent hole spectra and the absorption spectrum itself should be viewed, to a considerable extent, as a refinement of parameter values obtained directly from the hole spectra. Finally, we remark that it should not be expected that the main conclusions of our earlier work would be altered by higher-quality hole spectra. These are that (1) the linear electron-phonon coupling is strong (with Stocal 3.5 and a total optical reorganization energy of -240 cm-l), (2) the low-frequency phonons have a mean frequency of -30 cm-l with S 2, and (3) there is an “effective” pseudo-localized mode, which we have called the special pair marker (frequency of 120 cm-l and &, 1.5) that contributes a Franck-Condon progression to P870. The only unresolved questions are related to the nature and modeling of the wm phonons and the dynamical nature of the marker mode (or modes) in the near vicinity of -120 cm-l. Simulations of the low-temperature absorption spectrum of P870 which ignore these facts and the absence of satellite ZPH hole structure from discrete modes3s have no physical basis.
-
-
-
-
IV. Results A. Experimental and SimulatedSpectra. Figure 2 shows three of the ten PHB spectra for the protonated RC obtained with WB valuesin therange 10 921-1 1 049 cm-1. This rangeencompasses the distribution of ZPL frequencies for the total zero-point level (Table 11). In each spectrum the weak ZPH is coincident with OB. Its weakness reflects the fact that the ZPL is quite highly Franck-Condon forbidden. For o g near the center of the above
Reaction Centers of Rhodobacter sphaeroides
The Journal of Physical Chemistry, Vol. 97, No. 26, 1993 6927
TABLE II: Structure of P870'
ZPL distribution (middle spectrum), the integrated intensity of the ZPH relative to that of the entire hole is -exp(-2Stopl).29 The ZPH could not be observed for WB > 1 1 050 cm-1. For such frequencies the probability of exciting ZPL transitions of us: becomes negligibly small. The absence of high-energy satellite ZPHs in each spectrum should be noted. This, together with the absence of a discernible ZPH coincident with OB for OB > 1 1 050 cm-l, proves that the total dephasing time of the one-quantum and higher levels of the marker mode (wsp) is considerably shorter than 1 ps. Hole spectra were obtained with burn intensities between 10 and 400 mW/cm2 (results not shown). It was determined that saturation broadening of the ZPH is absent for optical density changes less than 20%as measured a t the P870 maximum. Hole spectra with percent optical density changes 1 1 050 cm-l. Of particular interest is the increased structure in the region of the marker mode, in particular the doublet hole structure indicated by the two arrows. This structure tracks the ZPH a t OB, and furthermore, its intensity increases and decreases in parallel to the ZPH across the range 10800
-
-
6928
Lyle et al.
The Journal of Physical Chemistry, Vol. 97, No. 26, 1993
925
WAVELENGTH(nm) 890
cos =lo921cm'
860
~ ~ , = 1 0 9 cm" 2 1
0.0
w
W
$
-.02
d
-.04
d
5
-.06
n
-10
-.08 10800
11000 11200 11400 11600 WAVENUMBERS (cm-') Figures. PHBspectraofP870(deuterated,WT)forwe= 10 921,lO992, and 1 1 030 cm-1 (top to bottom); T = 4.2 K. The curves labeled with S are theoretical simulations;cf. Figure 4 caption. of W B values. This doublet structure is not observed for the protonated RC. The left- and rightmost arrows are displaced from the ZPH ( W B ) by 145 and 175 cm-l. The broad feature at 120 cm-1 should also be noted. This is where the satellite ZPH from wSpl would be expected if the marker mode frequency for the deuterated RC is about the same as that for the protonated RC. With reference to Figure 3, one is struck by the similarity of its second and third (from the top) simulated spectra with the just discussed structure of the spectra in Figure 4. However, in Figure 3, the "dimple" at OB 120 cm-1 is the satellite ZPH of wsPl. Thus, thedrawing of a parallel between thespectraof Figures 3 and 4 would require that the hole feature at the leftmost arrow in Figure 4 be the corresponding satellite ZPH (within the twomode (wm,os*)model). This, in turn, requires a marker mode frequency of 145 cm-l . With this in mind we performed extensive calculations for the PHB spectra of the protonated and deuterated RC for a marker mode frequency of 145 cm-1. For both it proved impossible to obtain an acceptable, overall accounting of the hole profiles for all W B values. We mention, for example, that an increase of the marker mode frequency from 120 to 145 cm-l requires a decrease in the value of S,,from 1.5 to 1.2 in order to account for the dependence of the center of gravity of the hole profile on OB as well as the total width of the entire hole profile. Such a decrease, however, led to poor fits in the region between the us: and wspl regions of the hole spectra. Figure 5 shows our best fits with the two-mode model to the PHB spectra of P870 for the deuterated RC obtained with the parameter values given in Table 11. The marker mode frequency used is 1 1 5 cm-1. As can be seen from Table 11, the parameter values obtained for the protonated and deuterated RC are very similar. For the deuterated RC, the fits in the region of the ZPH at W B are also unsatisfactory; see inset of Figure 5. In addition, the simulations fail to account for the aforementioned doublet structure of the spectra in Figure 4. B. Zero-Phonon Hole Widths. For the protonated and deuterated RCs, P870 ZPH widths were measured at 4.2 K for 10 and 9 W B values that spanned the range 10 921-1 1 049 and 10 921-1 1 068 cm-', respectively. From Table I1 it can be seen that these ranges encompass the distribution of ZPL excitation frequencies for the total zero-point level. For all WB values the quality of the spectra was comparable to those of the spectra in Figures 2 and 4. Examples of the ZPH profiles used for analysis are given in Figures 6 and 7. The ZPH profiles were systematically fit with a curve fitting program from Spectra C a l ~ . ~The 6 fitting procedure used included a fit to the ZPH including f 2 0 cm-1 of
-
0
10
20
WAVENUMBERS (cm-I)
-
Figure 6. Four representativeZPH profiles for the deuterated RC; T = 4.2K. Burn frequencies are given in the fisure. These profiles plus five others yielded at ZPH width of 11.5 i 1.0 cm-1 (95% confidence limit). The fits to the profiles (cf. text) are not shown because they would be
indistinguishablefrom the experimental profiles.
I
%=lo921cm-'
-
0,=10992 cm-'
wB=11039cm"
+
-
-10
0
10
20
WAVENUMBERS (cm-') Figure 7. Four representativeZPH profiles for the protonated RC; T = 4.2 K. These profiles plus six others yielded a ZPH width of 11.5 i 1.0 cm-l (95% confidence limit); cf. Figure 6 caption.
data with a Lorentzian line shape. Because the ZPH rides on the broad hole, a linear base line was fit simultaneouslywith the ZPH. The procedure was tested by simulating hole spectra with ZPH widths ranging from 5 to 20 cm-l and values of the other parameters as given in Table 11. The hole spectra were then fit with the curve fitting program, and the resulting widths were within h0.5 cm-1 of the input ZPH widths. For both the protonated and deuterated RC the ZPH width did not show any dependence on UB;the widths for both are 11.5 f 1.0 cm-I, corresponding to a P870* lifetime of 0.93 f 0.10 ps at 4.2 K. C. TemperatureDependencies of the ZPHIntensity and Width. Figure 8 shows the temperature dependence of the zero-phonon hole profile ( W B = 10 992 cm-1) for P870 of the deuterated RC. As can be seen, the ZPH peak intensity undergoes a marked decrease from 1.6 to 16 K. Johnson et a1.I6 reported a similar dependence on temperature between 4.2 and 14 K for the protonated RC (R26). As in their experiments, we could not observe the ZPH at temperatures higher than 16 K. Using the approximate expression29 for the Tdependence of ZPH intensity stemming from the linear electron-phonon coupling, I 0: exp{-2S(2ii l)]) when ii = [exp(hw/kT- l)]-I is the occupation number for an effectivemode frequency w, Johnson et al. obtained the value w = 23 4 cm-1 (for S = 2.2). On the basis of our results, we now believe the above expression to be an oversim-
-
+
Reaction Centers of Rhodobacter sphaeroides
The Journal of Physical Chemistry, Vol. 97, No. 26, 1993 6929
I
I
P870*. They suggested dispersive kinetics from a distribution of V values as another possibility. The results presented in section 1V.B prove that the electrontransfer time of P870* (from total zero point) is invariant to excitation frequency within the inhomogeneous distribution of zerephonon line frequencies. However, this negative result cannot be interpreted as meaning that nonexponentialdecay kinetics for P* could not arise from a distribution of values for the coupling matrix element V. The reason is that the inhomogeneous ZPL frequency distribution depends on the distribution functions for the ground- and excited-state energies @,E*) of the special pair, \\PI/ -8.0 K I -.025 fp(E) and fp*(E*), and the extent of correlation between them. -4.2K n (For a detailed discussion see ref 38.) Noting that at any given value of E of P there is a high degree of accidental degeneracy, .1.8K an absence of correlation means that I’l of P870 is the width of L fp*(E*), and, furthermore, at any excitation frequency one 10920 10940 10960 10980 11000 essentially samples the entire ensemble of RCs, i.e., all possible WAVENUMBERS (cm-l) Vvalues. At the other extreme thereis perfect positiveor negative correlation. In this case the width of fp*(E*) C l’l of P870 and, Figure8. Temperaturedependenceof the ZPH profile for the deuterated RC (WT). From top to bottom TB= TR = 16, 12, 8.0, 4.2, and 1.8 K also, different excitation frequencies would sample different where TBand TRare the burn and read temperatures. Note how intensity subsets of the RC ensemble and, therefore, also different parts from the ZPH is drained by the PSBH as TBis increased. of the Vvaluedistribution. The actual situation may lie between these two extremes. plification of the physics of the low-frequency modes of proteins However, the above negative result, when viewed in concert and postpone the discussion of our analysis of Figure 8 until with two other findings, strongly suggests that a distribution of section V. Vvalues is unlikely to be the source of the nonexponentialdecay With increasing temperature, Figure 8 shows that the intensity of P870*. The first finding, from wedependent PHB spectra of of the phonon sideband holes increase, especially on the highthe RCs from Rps. uiridis and Rb. sphaeroides, is that the site energy side of the ZPH in accord with the expression derived by excitation energies of P*(P-) and P+ (the upper excitonic dimer Lyle.32 However, for 1.8, 4.2, and 8.0 K the interference is component of the special pair) are positively correlated.15 The sufficiently low to permit a measurement of the ZPH width. We upper dimer component is the low-energy shoulder of the most find that the width is invariant in this range, that is, within f l . O intense band at 830 nm in absorption; cf. Figure 1. The second cm-1 of 11.5 cm-1; cf. Table 11. On the basis of the time domain finding is to be found in the following paper on the nonphotomeasurements of the temperature dependence of the P870* chemical hole burning of P960 of Rps. viridis, wherein a quite lifetime (10 K to room temperature)21and theoretical a n a l y s e ~ , l ~ * ~ remarkable, ~ persistent photoinduced structural transformation one would expect a weak (at best) dependence for T < 10 K. of the special pair is discussed. The transformation produces an Nonetheless, the present results are the first to establish that this antihole of P- that is shifted 150 cm-1 to the red of the P- hole is the case. and an antihole of P+ that is shifted 150 cm-1 to the blue of theP+ hole. This negativecorrelationis consistent with the simple V. Discussion excitonic dimer model. Thus, the above positive correlation indicates that the inhomogeneous broadening of P870 (P960) is A. Dependence of the ZPH Width on Excitation Frequency due primarily to statistical fluctuations in cage structure around (wg). Very recently, ultrafast time domain experiments have thespecial pair rather than a distribution of special pair structures. shown that the decay of P870* at room temperature is Noting also that the dimer splitting is large, 1300 cm-1 while n~nexponential.~‘~~ For example, Vos et al.24fit the decay for l’I for P870 is small, 130-150 cm-l, it is clear that variations in the R26 mutant with a biexponential, 2.9 ps (0.65) and 12 ps the structure of the special pair from RC to RC in the ensemble (0.35). Stimulated by these results, Small et a1.18 reported the are very small indeed. One can reasonably assert, therefore, that following expression for the electron-transfer rate of a single RC: the structure of the special pair and the two neighboring BChl monomers is also extremely well-defined and, as a consequence, k = 2aV2[2a3(u2 om2)]-1/2 x that the distribution of V values is very narrow. exp[-(n - som)/23(a2 + wm2)1 (4) Next we consider the possibility that the observed nonexponentiality of P870* is the result of heterogeneity that is above which is valid for strong !near electron-phonon coupling. Sand and beyond the type we have referred to as “normal glasslike”. o,areasdefinedbefore,S = Scoth(hom/2kT), Vis theelectronic That is, there are two distinct subsets of the RC ensemble, one coupling matrix element between P* and relevant chargeyielding a P870* decay time of -3 ps (0.65) and the other a separated state of the RC, !lis the adiabatic energy gap between decay time of 12 ps (0.35). We think this very unlikely since these two states, and 2u is the width of the one-phonon profile we have never observed a P870 ZPH profile that is a sharp feature of the nuclear factor. For a Gaussian distribution of Q values superimposedon a broader ZPH of the type shown in Figures 6 centered at !lo, the average value of the rate is and 7. Of course, one could suggest that the electron-phonon coupling for the (0.35) subset is significantly greater than that (k)= 2aV2[27r(P S(u2 wm2))]-l’2 x of the (0.65) subset, but there is no physical basis for this. exp[-(n, - ~w,)~/2(?’ + S(u’ + om2))] (5) B. The Underlying Structure of P870. A comparison of the results in Tables I and I1 shows that the main conclusions from where 2 f is the width of the !ldistribution. Equations 4 and 5 our earlier work on the underlying structure of P870 (and by are easily generalized to include the special pair marker mode. implication F960 of Rps. viridis, see following paper) are left Using hole-burning data and low-temperature optical data for unchanged by the present experimental results. However, the charge-transfer organic crystals, Small et a1.18 concluded that present hole spectra show that the simulations are inadequate in dispersive kinetics from the !ldistribution is very unlikely to be the region of the ZPH and, for the deuterated RC, also in the responsible for the aforementioned nonexponential decay of
\vi v -
1
- -
-
+
-
+
+
6930 The Journal of Physical Chemistry, Vol. 97, No. 26, 1993 region of the marker mode. We consider now possible reasons for the discrepancies. 1 Modeling of the Low-Frequency ‘Phonons”. The inserts of Figures 2 and 5 show that the simulation underestimates the contribution of the low-frequency modes (520 cm-I) to higher and lower energy of the ZPH. (Inspection of the experimental profiles just to the red of the ZPH in Figures 2 and 4 shows a distinct ‘bulging” at -30 cm-I which is the pseudophonon sideband hole.) As discussed earlier, we have used an asymmetric one-phonon profile centered at wm with the lower- and higherenergy sides described by a Gaussian and broader, Lorentzian, respectively. From Table I1 the respective half-widths for the protonated RC are 15 and 28 cm-I. By significantly broadening the Gaussian, it is possible to improve the agreement between experiment and simulation (results not shown). However, from the form of the single site absorption profile (eq l), it can be seen that such a broadening would lead to unphysical absorption to the red of the zero-phonon line at 0 K which, in turn, contaminates the region of the pseudophonon sideband in the hole spectrum. This is why in our simulations we have used Gaussian half-widths which are Swm/2. The failure of a Gaussian to describe the low-energy side of the one-phonon profile indicates that a Debye phonon densityof states, which is proportionalto 02,is inadequate for the RC; i.e., the density of states of low-frequency modes in the protein is anomalous and not Debye-like. This suggestion finds strong support from recent neutron inelastic scattering,39 Raman,M and h o l e - b ~ r n i n g 4studies ~ ~ ~ ~ of amorphous and crystalline polymers. For example, Kanaya et al.39 have reported that while crystalline polyethylene exhibits a Debye phonon density of states (580 cm-l), amorphous polyethylene exhibits an additional density from special modes that peaks at wp -30 cm-I, tails to higher energy, and, after a small decrease on the low-energy side of upappears to level off. The above groups associate the special modes with the microscopic disorder of the polymer. This being the case they should not be referred to as phonons (acoustic). A variety of experimental investigationsof inorganicglasseshave also indicatedtheexistence of disorder-induced low-frequency modes (see ref 43 and references,therein) which may be anharmonic at sufficiently low energy, several kelvin.“ We believe that the temperature dependence of the ZPH intensity shown in Figure 8 provides further evidence for the activity of such low-frequency anomalous modes. As discussed in section IV.C, Johnson et al.16fit the their T-dependent (4.2-14 K) intensity data to the standard theoretical expression (with S = 2.2) to arrive at a mean phonon frequency of om= 23 f 4 cm-I. Using the same expression (with S = 2.0, Table 11) to fit our data for 4.2 K an the higher temperatures, we obtained a value of w,,, = 17 f 2 cm-1. The problem is that with S = 2.0 and w, = 17 cm-I the theory predicts that the ZPH intensities at 1.6 and 4.2 K should be equal. This is at odds with the results of Figure 8. The significant decrease in intensity from 1.6 to 4.2 K indicates considerable activity of modes of very low frequency which, in turn, means that the theoretical expression used above is inappropriate. 2. The Special Pair Marker Mode wSp. We have previously assigned the wSp mode as an intermolecular vibration whose amplitude is highly localized on the special pair. The argument against the assignment as an intramolecular mode was based on the fact that the largest Svalue for an intramolecularmode (-740 cm-I) of Chl a and BChl a is 0.05,8*20 which is about a factor of 30 times lower than S,, (Table 11). Furthermore, in ref 20 no intramolecular mode with a frequency lower than 260 cm-1 was observed, which meant that their S values are 50.001. Very recently, nonphotochemical hole-burned spectra of the BChl a B870 and B895 bands of the antenna of Rb. sphaeroides have been reported.9 Both of these bands are associated with states that are characterized, in part, by a strong dimer interaction. I
Lyle et al. However, a mode comparable to the marker mode of the special pair is absent in the spectra and, furthermore, the electron-phonon coupling of the low-frequency phonons is weak (S< 1). Thus, it continues to be plausible thatl-3 the strong activity of the lowfrequency modes and marker mode is due to a significant amount of charge-transfer character for P870*.45-47 Before considering the marker mode further, it is appropriate to discuss first the doublet hole structure observed in the region of the marker mode for P870 of the deuterated RC; see arrowlabeled features of Figure 4. The second spectrum from the bottom of Figure 4 appears as the middle spectrum in Figure 5 along with the theoretical fit (S).It was obtained with WB = um (TableII) andcan bedirectlycompared with themiddlespectrum of Figure 2 for the protonated RC. The dashed arrow in both spectra is displaced by aspfrom the ZPH at OB. By smoothing out (visually) the doublet structure in the middle hole spectrum of Figure 5 and comparingthe result with the marker mode region of the middle spectrum of Figure 2, it is apparent that the two are different. For example, in the latter the maximum is at the position of the dashed arrow (UB+ asp) whereas, in the former, the maximum is near OB + wSp 30 cm-’. The possibility that the “bulge” at the position of the dashed arrow in Figure 5 and the aforementioned doublet components are the wSp1 120, w , ~ ~ 30, and wsPl 2 X 30 cm-l satellite holes, respectively, seems unlikely. First, it is difficult to understand why the one- and two-phonon sideband holes are sharper than the ZPH (bulge) at 12Ocm-1 and, second, why they are not resolved in the analogous region to the blue of the ZPH at OB. In addition, two-mode (um, asp) simulations based on the above assignment proved to be unsatisfactory. As mentioned in the preceding section, this was also the case when the above three satellite holes at 120, 150 and 180 cm-l were assigned as wSpl - 30, wSp1 150, and wlpl + 30 cm-I. We tentatively conclude that the two-mode (um,wlp) model for P870 of the deuterated RC is inadequate. Inclusion of two marker modes, one at 120 cm-1 and the other at 150 cm-I, in the simulations would probably lead to improved fits to the hole spectra when the damping constant for the 150-cm-1 mode is reduced from 50 cm-I to about 25cm-1 and Ssp 1.5 is properly apportioned between the two modes. We have postponed such simulations until we develop a more accurate description of the one-phonon profile (vide supra). The absence of the above triplet structure in the marker mode region for the protonated RC is interesting. The fact that the two-mode model yields simulated spectra in good agreement with the hole spectra (except in the near vicinity of the ZPH at WB) does not exclude the possibility that there are two closely spaced marker modes of the protonated RC which split further apart up& deuteration due to isotopedynamic mixing4 or anharmonic effects. One of the two modes could, of course, be weak for the protonated RC but strong for the deuterated RC due to isotopeinduced mode mixing?* As mentioned, the simulations show that the additional structure in the marker mode region of the hole spectra for the deuterated RC would not be observed if the homogeneous broadening for the marker mode(s) was as large as 50 cm-1 (cf. Figure 3). Its observationsuggeststhat deuteration may increase the vibrational dephasing times, in particular that of the mode at 150cm-l. This possibility finds support from the picosecond data of Dlott and c o - ~ o r k e ron s ~optical ~ phonon relaxation times in protonated and deuterated organic crystals. They were able toexplain thedependenceof therelaxation timeon optical phonon frequency using a simple phonon densityof states argument based, in part, on the fact that the effective Debye frequency in organic solids such as naphthalene is reduced by 20%upon deuteration. We turn next to a consideration of the relevance of the data from recent resonance Raman studies of P870 (protonated RC)353031to the absorption and hole spectra. These works have revealed considerablemode activity beiow 200 cm-I. The lowest-
+
+
-
+
-
- -
-
--
-
-
The Journal of Physical Chemistry, Vol. 97, No. 26, 1993 6931
Reaction Centers of Rhodobacter sphaeroides
-
frequency mode observed by Shreve et al. is 36 cm-1 with a width of 15 cm-1 comparable to that of the intramolecular modes, while Palaniappan et al. were not able to access the region below 50 cm-1. S factors, damping constants, and the inhomogeneous band broadening could not be obtained from the Raman spectra. Important also to the discussionis that hole burning has proven that a broad distribution of low-frequency modes with a “mean” frequency wm -30 cm-1 couple strongly to the P870* P870 transition with an S value of -2.0. The spectra presented here confirm the correctness of our earlier physics, e.g., direct observation of the 30-cm-1 pseudo-phononsideband hole (Figures 2 and 4) and the temperature dependence of the ZPH intensity (Figure 8) which cannot be understood by invoking discrete modes in the Raman spectrum with frequencies 1 3 6 cm-l. The reader is also referred to the following paper in which the real-phonon sideband hole for P960 of Rps. viridis at 30 cm-1 is resolved for the first time. Palaniappan et al. report resonance Raman spectra for P870 at 25 K that show 18 bands between 52 and 404 cm-1 with widths of I 1 0 cm-l. They assign the bands as intramolecular modes of BChl. The intensity distribution over this range is more or less uniform although the relative intensitiesare accurate to no better than about a factor of about 3. Because the Raman intensities cannot be directly used to model the P870 low-temperature aborption profile (or the hole spectra), Palaniappan et al. used a multiparameter model with variableweightingcoefficients(with most of the above modes plus several additional higher-frequency modes, up to 736 cm-1) to simulate the P870 bands (Le., a stickmodel with rl = 140 cm-1). The resulting S factors bear no relationships to the observed Raman intensities. Of interest, in view of our earlier and present results, is that their fits yield S factors of 0.63,0.43, and 0.43 (for a total of 1.5, cf. our Sspvalue in Table 11) for modes at 105, 130, and 140 cm-I. Their lowestfrequency modes at 52 and 80 cm-1 have S factors of 0.30 and 0.23. T h e S factor for the 15 modes with frequency >140 cm-I are a factor of 10 or more smaller than 0.63. The highestfrequency mode at 736 cm-1 has an Sfactor of 0.03. Palaniappan et al.’s 20-parameter fit arrives, more or less, at the results of Table 11, albeit with three modes in the near vicinityof our marker mode (Ssp= 120 cm-1) and the 52- and 80-cm-1 modes as substituentsfor thecritically important 30-cm-l phonons. Their model for P870 would lead to calculated hole spectra that bear no relationship to the experimental ~ p e c t r a . ~ ~Furthermore, -~l no data are provided that support the high S values of the 52-, 80-, 105, 130-, and 140-cm-1 modes. The intensities of the lowfrequency (5-200-cm-1) Raman modes of P870 are stated to be about a factor of 10 higher than those obtained for the two BChL monomers of the RC. This statement doesseemconsistentwith the Raman intensities reported by Shreve et al. and our measurements of intramolecular Franck-Condon factors. Shreve et al. report relative intensities for their modes at 36, 71, 94, 127, 202, and 337, 685, and 730 cm-1 of 0.4, 0.2, 0.6, 1.0, 0.6, 0.3, 0.2, and 0.7. (The origin of the discrepancies between the frequencies and number of the low-frequency modes reported by the two groups is not clear.”) Since a Franck-Condon factor of 0.05 for a pronounced intramolecular mode near 730 cm-1 has been measured for the Q, transition of antenna Bchl a and Chl a, and the FranckCondon factors for modes with frequency 5260 cm-1are 50.001, Franck-Condon factors of ~ 0 . 0 5for the 36-, 71-, 94-, and 127-cm-1modes of P870 are plausible. However, such a ualue is still an order of magnitude lower than those obtained from the P870 band simulation of Palaniappan et al. Indeed, if the 105-, 130-,and 140-cm-lmodescarriedsuchlargeSfactorvalues, one would expect to be able to observe their overtones and combinations since the above 730-cm-l mode with its S factor of -0.05 is readily observed. They are not observed. One further point is that the 36-cm-1 Raman mode reported by Shreve et al.
-
-
-
is too sharp to be assignableto the low-frequencyphonons observed in the hole spectra. It may be due to the 30-cm-l acetyl torsion modes of BChl calculated by Palaniappan et al. In conclusion, we are not inclined to change our assignment of the special pair marker mode as an intermolecular pseudolocalized phonon of the pair especially since charge-transfer states of stacked 1:1 *-molecular donor-acceptor complexes are well known to couple very strongly to intermolecular modes of the complex.52
VI. Conclusions The two main problems which have been considered are the nonexponential decay kinetics of the primary donor state P870* and the mechanisms that lead to the width and shape of the P870 absorption profile. With regard to the first, we have investigated further whether or not the nonexponentialdecay could be due to dispersivekinetics arising from the glasslike structural heterogeneity of proteins. In an earlier paper,I8it was concluded that the distribution of values for the relevant electronic energy gap(s) associated with electron transfer is too narrow to lead to kinetics sufficiently dispersive at room temperature to account for P870*’s nonexponentialdecay. In the present paper zero-phonon hole width data are presented which prove that the lifetime of P870* (from total zero point) is independent of the position of the P870* P870 zero-phonon line (ZPL) frequency within the inhomogeneous distribution of ZPL frequencies. For the protonated and deuterated RC samples studied, which exhibited nearly identical P870 absorptionwidths, the P870* lifetime is 0.93 f 0.10 ps for T I 10 K. On the basis of the just-mentionedindependenceand other results, we conclude that the nonexponential decay is unlikely to be due to a disorderinduced distribution of values for the relevant pure electronic coupling matrix element(s) responsible for the initial phase of charge separation. In addition, we conclude that it is unlikely that the nonexponential decay is due to gross heterogeneity. This conclusion is based on the assumption that the charge recombination kinetics of P+Q- for the “slower”decaying subset of the P870* ensemble are not significantly faster than those for the “faster” decaying subset. (We can conceive of no reason why they should be.) Our data and current thinking provide no insight on additional, plausible mechanisms that would be consistent with the nonexponential decay. We note, however, that it has been suggested that the slow decaying component may be due to a parking state, e.g. P+BM-(BM bacteriochlorophyll monomer on the M branch), which is populated following excitation of P870 and which undergoes charge recombination to P* in 12 ps.53354 We note also that time domain studies of certain mutants of Rhodobacter capsulatas have revealed that the amplitude of the faster and slower decaying components are weakly dependent on temperat~re.’~This is interesting since P870* undergoes a red shift of -300 cm-l as the temperature is reduced from 300 to 10 K.Z9 Concerningthesecond problem,the photochemical hole-burned spectra of P870 obtained as a function of burn frequency (see also Figure 4 of following paper on P960 of Rps. viridis) provide definitive proof that a broad (-40 cm-I) distribution of protein modes (phonons) with a mean frequency of 30 cm-l is critically important to the understanding of P870’s and (and P960’s) absorption and hole profiles. That such phonons couple to the P870* P870 transition is hardly surprising since such coupling is routinely observed for the Q, states of chlorophylls in antenna protein complexes. However, whereas the electron-phonon coupling is weak for the latter, it is strong (S 2) for P870*. This strong coupling, along with the pronounced coupling to the special pair marker mode with S,, = 1.5, leads to an optical reorganization energy of -240 cm-1, which is an order of magnitude or more greater than the values for antenna protein complexes.*J In our earlier works14J6 the strong coupling (Setal 3.5) was attributed to the large permanent dipole moment
-
-
-
-
-
6932 The Journal of Physical Chemistry. Vol. 97, No. 26, 1993
Lyle et al.
change (-9 D) associatedwith the P870* t P 8 7 0 tran~ition.4~9~ Thelarge dipolemomentofP870* suggested that P870* possesses significant charge-transfer character which could be the result of mixing with a nearby charge-transfer state@), e.g., P+BL,M-. The problem of how mixing between a neutral ?TU* state with a CT state leads to a concomitant strengthening of the electronphonon coupling is an old 0ne.5635~Lathrop and FriesneP8 have recently reconsidered it for the primary donor state. However, the permanent dipole moment of P870* could be due also to the charge resonance states of the special pair4'?59 since C2 is only a pseudosymmetry axis of the RC. We consider next the theoretical fits to P870's absorption and hole profiles. Except for the region in the very near vicinity of the ZPH at the burn frequency, good overall agreement was achieved for the protonatedRC by invokingonly the low-frequency phonons and a single special pair marker mode at 120 cm-l with a total dephasing time of 200 fs. The agreement from this model for P960 of Rps. viridis is even more striking since the experimental hole spectra for P960 are more highly structured (see following paper). However, the fact that the hole structure for P870 of the deuterated RC of Rb. sphaeroides is more complicated in the region of the marker mode leads us to tentatively conclude that two modes at -120 and -150 cm-l (with a total S,, 1.5) are necessary to account for its P870 hole profiles. The possibility that there are two closely spaced, unresolved marker modes for the protonatedRC was raised. In futureconsiderations, this isotope dynamic mode mixing48should be taken into account. Concerning the dynamical nature of the marker mode@), we continue2$8.20 to favor the interpretation that it is mainly a localized intermolecular vibration of the special pair on the basis of its unusually large Huang-Rhys factor. Very strong Franck-Condon activity of such modes in optical transitions to pure charge-transfer states of 1:l u molecular complexes is well documented.60 Concerning the relationship of the P870 resonance Raman data to the optical absorption and hole spectra, it was stressed that the Raman spectra do not access the critically important distribution of low-frequency phonons (w, 30 cm-*) and that the purely empirical determination of the S factors associated with the potentially relevant lower-frequencymodes leads tovalues that are inconsistent with the resonance Raman spectra. In future resonance Raman studies it will be important to directly determine the S factors and excitation profiles of the modes which have been observed and to ascertain the extent to which the lowfrequency phonons are silent in Raman. Consideration should also be given to the extent Herzberg-Teller vibronic coupling is operative in Raman as well as other effects that can make a connection between the resonance Raman data and absorption/ hole spectra difficult, e.g., the Dushinsky effect and nonconstant damping constants. Finally, we emphasize that our theoretical calculations and those of Won and Friesner" together with the experimental hole profiles (see also following paper) point convincingly to the special pair marker mode(s) having a total dephasing time significantly shorter than 1 ps. In our previous papers we have favored the interpretation that this dephasing is vibrational in nature since the lifetimes of P870* measured by hole burning (from total zero point) and by ultrafast spectroscopy agree within experimental error. (Given that Stal 3.5 for P870, excitation of P870 by an ultrashort pulse predominantly prepares P870* phononically and marker mode hot.) However, the proposal discussed abo~e,~3.54 that P870* may decay to a "parking state" such as P+BM-,itself inactive in electron transfer to bacteriopheophytin, raises the possibility that the fast dephasing of the marker mode may have an electronic contribution. That is, the marker mode, while being Frank-Condon active in absorption, could be vibrationally active in coupling to a dark state such as P+BM(but apparently not to P+BL-, since if this were the case one
would expect a coth(hw,,/kO dependence for the kinetics that lead to reduction of the bacteriopheophytinon the active branch (L) of the RC, which is not observed). This possibility is highly speculative. Clearly, much more experimental and theoretical work is needed to arrive at a firm understanding of the initial phase of charge separation in bacterial RC.
-
-
-
-
Acknowledgment. Research at the Ames Laboratory was supported by the Division of Chemical Sciences, Office of Basic Energy Science, U.S. Department of Energy. Ames Laboratory is operated for the Department of Energy by Iowa State University under Contract W-7405-Eng-82. Paul A. Lyle was supported by a W.E. Catron Fellowship. We thank David M. Tiede of the Argonne National Laboratory for providing us with thedeuterated RC samples and several useful discussions. References and Notes (1) Johnson, S.G.; Lee, I.-J.; Small, G. J. In Chlorophylls; Scheer, H., Ed.; CRC Press, Inc.: Bcca Raton, FL, 1991; pp 739-768. (2) Reddy, N. R. S.;Lyle, P. A.; Small, G. J. Photosyn. Res. 1992,31,
167. (3) Jankowiak, R.; Small, 0. J. In Photosynthetic Reaction Centers; Deisenhofer. J.. Noms. J.. Eds.: Academic Press: New York. in Dress. (4) Singh,'G. P.; Schhk, H. J.; Lohneysen, H.; Parak, F.f Hinklinger, S . 2.Phys. 1984,B55, 23. ( 5 ) Yang, 1.4.; Anderson, A. C. Phys. Reu. B 1986,34, 2942. (6) Frauenfelder, H.; Sligar, S.G.; Wolynes, P. G. Science 1991, 254, 15411
_ I _ - .
(7) Reddy, N. R. S.;Small, G. J. J. Chem. Phys. 1991.94, 7545. (8) Reddy, N. R. S.;Small, G. J.; Seibert, M.; Picorel, R. Chem. Phys. Lett. 1991, 181, 391. (9) Reddy, N. R. S.;Picorel, R.; Small, G. J. J . Phys. Chem. 1992.96, 6458. (10)Reddy, N.R. S.;Cogdell, R. J.; Zhao, L.; Small, G. J. Photochem. Photobiol., in press. (11) Johnson, S.G.; Small, G. J. J . Phys. Chem. 1991, 95,471. (12) Tang, D.;Jankowiak, R.; Gillie, J. K.; Small, G. J.; Tiede, D. M. J. Phys. Chem. 1988,92,4012. (13) Tang, D.;Jankowiak, R.; Small, G. J.; Tiede, D. M. Chem. Phys. 1989,131,99. (14) Tang, D.;Johnson, S.G.; Hayes, J. M.; Jankowiak, R.; Tiede, D. M.; Small, G. J. J. Phys. Chem. 1989,93,5953. (15) Tang, D.; Johnson, S.G.; Jankowiak, R.; Hayes, J. M.;Small, G. J.; Tiede, D. M. In Twenty-Second Jerusalem Symposium: Perspectives in Photosynthesis; Jortner, J., Pullman, B., Eds.; Kluwer Academic: Boston, MA, 1990; pp 99-120. (16) Johnson, S.G.; Tang, D.; Jankowiak, R.; Hayes, J. M.; Small, G. J.; Tiede, D. M. J. Phys. Chem. 1990,94,5849. (17) Reference 2 and references therein. (18) Small, G. J.; Hayes, J. M.; Silbey, R. J. J . Phys. Chem. 1992,96, 7499. (19) Kolaczkowski, S.V.; Lyle, P. A.; Small, G. J. In Structure, Function and Dynamics of the Bacterial Reaction Center, Breton, J., Ed.; Plenum Press: New York, 1992,pp 173-181. (20) Gillie, J. K.; Small, G. J.; Golbeck, J. H. J. Phys. Chem. 1989,93, 1620. (21) Fleming, G. R.; Martin, J.-L.; Breton, J. Nature 1988, 333, 190. (22) Breton, J.; Martin, J.-L.; Fleming,G. R.; Lambry, J . 4 . Biochemistry 1988,27, 8276. (23) Jortner, J. Biophys. Acta 1980,594, 193. (24) Vos, M. H.; Lambry, J. C.; Robles, S.;Youvan, D. C.; Breton, J.; Martin, J.-C. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 8885. (25) Zinth, W. Ultrafast YIIe Springer-Verlag: Berlin, 1992;abstract. (26) Holtzwarth, A. Ultrafast VIII: Springer-Verlag: Berlin, 1992; abstract. (27) Du, M.; Rosenthal, S. J.; Xie, X.;DiMagno, T. J.; Schmidt, M.; Schiffer, M.; Hanson, D. K.; Norris, J. R.; Fleming, G. R. Proc. Natl. Acad. Sei. U S A . 1992,89, 8517. (28) Hayes, J. M.; Small, G. J. J. Phys. Chem. 1986,90,4928. (29) Hayes, J. M.; Gillie, J. K.; Tang, D.; Small, G. J. Biochim. Biophys. Acta 1988932,287-30s. (30) Won, Y.; Friesner, R. A. Proc. Narl. Acad. Sci. U.S.A.1987,84, 551 1-5515. (31) Won, Y.; Fritsner, R. A. J. Phys. Chem. 1988,92,2214;1989,93, 1007. (32) Lyle, P. A. Ph.D. Dissertation, Iowa State University, 1992. (33) Small, G. J. J. Chem.Phys. 1970,52,656and many referencestherein. (34) Middendorf, T. R.; Mazzola, L. T.; Gaul, D. F.; Schenck, C. C.; Boxer, S . G. J. Phys. Chem. 1991, 95,10142. (35) Palaniappan, V.; Aldema, M. A.; Frank, H. A.; Bocian, D. F. Biochemistry 1992, 31, 11050. (36) Galactic Industria Corp., Salem, NH. (37) Bixon, M.; Jortner, J. Chem. Phys. Lett. 1989, 259, 17.
Reaction Centers of Rhodobacter sphaeroides (38) Sevian, H. M.; Skinner, J. L. Theor. Chim. Acta 1992,82, 29. (39) Kanaya, T.; Kaji, K.; Ikeda, S.;Inoue, K. Chem. Phys. Lett. 1988, 150, 334. (40) Saiken, S.;Kishida, T.; Kanematsu, Y.; Aota, H.; Harada, A.; Kamachi, M. Chem. Phys. Lett. 1990,166,358. (41) Furasawa, A.; Horie, K.; Mita, I. Chem. Phys. Lett. 1989,161,227. (42) Furasawa, H.; Horie, K.; Suzuki, T.; Machida, S.; Mita, I. Appl. Phys. Lett. 1990, 57, 141. (43) Buchenau, U.;Galperin, M.; Gurevich, V. L.; Parshin, D. A.; Ramos, M. A,; Schober, H. R. Phys. Rev. E . 1992,46,2798. (44) Buchenau, U.; Zhou, H. M.; Nilcker, N.; Gilroy, K. S.;Phillips, W. A. Phys. Rev. Lett. 1988,60, 1318. (45) L6sche.M.; Feher,G.; Okamura, M. Y. Proc. Natl. Acad.Sci. U.S.A. 1987,84,7537. (46) Lockhart, D. J.; Boxer, S . G. Proc. Natl. Acad. Sci. U.S.A. 1988,85, 107. (47) Parson, W. W.; Warshel, A. J. Am. Chem. SOC.1987,109, 6152. (48) Dlott,D.D.Annu.Rev.Phys.Chem. 1986,37,157-187andreferences therein. (49) Orr, G.; Small, G. J. Chem. Phys. Lett. 1973, 19, 574. (5O).Donohce, R.; Dyer, R. B.; Swanson, B. I.; Violette, C. A,; Frank, H. A.; Bocian, D. F. J. Am. Chem. SOC.1990,112,6716.
The Journal of Physical Chemistry, Vol. 97, No. 26, 1993 6933 (51) Shreve, A. P.; Cherepy, N. J.; Franzen, S.; Boxer, S.G.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1991,88,11207. (52) Haarer, D.; Philpott, M. In Spectroscopy and Excitation Dynamics of Condemed Molecular Systems; Agranovich, V. M., Hochstrasser, R. M., Eds.;North-Holland: Amsterdam, The Netherlands, 1983;pp 29-80. (53) Ogrodnik, A.; Langenbacher, T.; Eberl, U.; Volk, M.; Michel-Beyerle, M. E. In The Photosynthetic Reaction Center II. Structure, Spectroscopy and Dynamics; Breton, J., Vermeglio, A., Eds.; Plenum Press: New York, 1992;pp 261-270. (54) DiMagno, T.J.; Rosenthal, S. J.; Xie, X.; Du, M.; Chan, C.-K.; Hanson, D.; Schiffer, M.; Norris, J. R.; Fleming, G. R. In The Photosynthetic Reaction Center II. Structure, Spectroscopy and Dynamics; Breton, J., Vermeglio, A., Eds.; Plenum Press: New York, 1992; pp 209-207. ( 5 5 ) Chan, C.-K.; Chen, L. X.-Q.; DiMagno, T. J.; Hanson, D.; Schiffer, M.; Norris, J. R.; Fleming, G. R. Chem. Phys. Lett. 1991, 176, 366. (56) Beckman, R. L.; Small, G. J. Chem. Phys. 1977,30, 19. (57) MBhwald, H.; Sackmann, E. Z . Naturfursch. 1974, A26, 1216. (58) Lathrop, E.J. P.; Friesner, R. A. In The Photosynthetic Reaction Center II. Structure, Spectroscopy aird Dynamics; Breton, J., Vermeglio, A,, Us.; Plenum Press: New York, 1992;pp 183-192. (59) Thompson, M. A.; Zerner, M.C. J. Am. Chem.Soc. 1991,113,8210. (60) Hayes, J. M.; Small, G. J. Chem. Phys. 1977,21, 135.
