Electron Storage and Transfer in Organic Redox Systems with Multiple Electrophores

Electron Storage and Transfer in Organic Redox Systems w i t h Multiple Electrophores M. BAUMGARTEN,~ W. H U B E R and ~ K. M ~ L L E N ~ Max-Planck-lnstitut fur Polymerforschung, Ackermann- Weg 10,6500 Mainz, Germany Hoffmann La Roche Ltd, Grenzacher Strasse 124,4002 Basel, Switzerland

Introduction 1 Design and synthesis 5 Highly charged states via extended redox sequences 10 Mechanisms of successive electron transfers 14 Structural factors relevant to intramolecular electron transfer The nature of the subunit 22 The length of the bridging group 25 The conformation of the bridging group 29 Ion pairing 32 The mode of linking 36 6 Conclusion and outlook 39 References 40

1 2 3 4 5

1

17

Introduction

Unsaturated carbo- and hetero-cycles are known to constitute active electrophores that can be subjected to chemical or electrochemical redox processes forming persistent cations or anions. Depending on the number of electrons transferred, the charged products can be either diamagnetic or paramagnetic. Crucial questions in describing the redox-activity of cyclic 7c-systems concern: (i) the number of accessible redox-states (Meerholz and Heinze, 1989, 1990); (ii) the mode of charge distribution (Heilbronner and Bock, 1978; Salem, 1966; Eliasson et al., 1986, 1990); (iii) the stabilization of charge by ion-pairing and/or conjugative effects (Hogen-Esch, 1977; Miillen et al., 1990); (iv) the possibility of charge-induced configurational and conformational changes (Eliasson et al., 1986; Huber and Miillen, 1986); (v) the chemical reactivity (Hogen-Esch, 1977; Szwarc, 1968; Miillen, 1986, 1 ADVANCES IN PHYSICAL ORGANIC CHEMISTRY VOLUME 28 ISBN 0-12-033528-X

Copyright 0 1993 Academic Press Limited A / / rights of reproduction in any form reserved

2

M. BAUMGARTEN ET AL.

1987). A particularly attractive group of substrates are the annulenes, since electron transfer allows an interconversion of (4n + 2)n- and (4n)x-systems, and thus a switch between x-bond delocalization and n-bond localization (Mullen, 1984). A relevant step from the viewpoint of electron-transfer activity is the linking of two or more identical electrophores to bis-, oligo- or polyelectrophoric systems. In an attempt to extrapolate the description of the single electrophore to that of the higher homologues, the above questions maintain their significance, but additional attractive aspects arise. First of all, it is appropriate to subdivide multi-electrophoric species into two classes depending on whether the linkage between the building blocks is an unsaturated or saturated chemical unit. In the former case, with extended x-conjugation, one expects successiveelectron-transfer steps to create strongly interacting redox states in which the excess charge is distributed over the whole n-system. In the latter case, with electronically “decoupled” electrophores, one expects the electron-transfer steps to occur independently, with only electrostatic interactions of the redox states; the excess charge is localized on the electrophores (Smith et al., 1976; Flanagan et al., 1978). Within the domain of materials science, the classification of multielectrophoric organic systems according to the mode of linking of the subunits has led to the definitions of conducting polymers and redox polymers (Wegener, 1981; Baughman et al., 1982; Heinze, 1990; Nowak et al., 1980; Murray, 1984). The present review is restricted to “dimeric” or (low molecular weight) “oligomeric” redox systems with identical, electronically decoupled electrophores A. Consider the biselectrophore A-1- A, where 1 represents a saturated spacer (Fig. 1A). Injection of an electron into one subunit A under formation of a radical anion raises the immediate question of whether the electron will tend to localize in the original subunit or whether it will undergo a degenerate electron transfer to the neighbouring unit. In the terminology of electron-transfer kinetics such electron hopping between identical redox groups is termed self-exchange (Cannon, 1980). Depending upon the rate of this process, one will observe an “effective delocalization” over two or more units within the timescale of the experiment, or a localization of charge on one unit. Upon further contact with a redox reagent or at higher redox potentials, additional electrons can be transferred. After a two-electron transfer, each redox unit can accept one charge with formation of a singlet or triplet dianion, or (less favourably from an electrostatic point of view) both charges can enter one redox unit. Here again, an intramolecular electron-exchange process is possible. Further relevant questions concern the nature of the highest accessible

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

A-I-A

a.

3

a. or

m a. or

m

or

Fig. 1 Charging of a biselectrophore (A) and a triselectrophore (B).

redox state and the way in which the molecule copes with the increasing electrostatic repulsion. Clearly, the charging mechanism of the substrate and the particular electron-transfer kinetics are expected to be more complicated if the redox groups are different; this can be the case in the homologous tris-electrophore (Fig. lB), since the outer and inner electrophoric units now have a different substitution pattern, or in the bis-electrophore A-1-A when one unit A is slightly perturbed, e.g. by alkyl substitution. A major concern of this review is the tailoring of the redox behaviour of organic compounds, i.e. the optimization of such systems for electron storage and electron hopping. While the emphasis is on reduction and thus on anion formation, it has been shown on many occasions that oxidative cation formation leads to analogous conclusions (Meerholz and Heinze, 1990; Lewis and Singer, 1965). The structure of this text is thus obvious. (a) One will first have to consider the design and synthesis of suitable systems in which the structural conditions relevant for the energy profiles of inter- and intra-molecular electron-transfer processes can be systematically varied. The variation of the structure comprises the nature of the redox-unit A and of the spacer 1. It will be shown that

4

M. BAUMGARTEN ET AL.

this can be brought about readily by the synthetic technique of reductive alkylation (Mullen, 1984, 1986, 1987). (b) The second question is whether such species can serve as efficient electron acceptors in successive charging reactions. This aspect, if expressed in terms of structure, will focus on electrophores A that can be charged to a high charge density and on spacer groups 1 that tend to minimize the consequent Coulombic repulsion. It is obvious that this approach is closely related to the search for organic electronstorage materials that can be used as battery electrodes ( McDiarmid, 1979; Bitthin et al., 1987; Shacklette et al., 1987). (c) A third question concerns the sequence of the successive charging processes. It is clear that, depending on the particular charging mechanism and upon the number of electrons transferred, one can arrive at para- or dia-magnetic products that differ in the prevailing charge (spin) density distribution. (d) The fourt!? and main topic is how the intramolecular electron transfer between the redox groups of the systems depends upon structural phenomena. The energy profiles of intramolecular electron-transfer processes are important for many areas of chemistry. Some representative examples are (i) the photochemically induced electron transfer and the lifetime of charge-separated states in porphyrin-quinone diads, which serve as model compounds in photosynthetic studies (Gust et al., 1986, 1988; Wasielewski et al., 1985); (ii) the possibility of a long-range electron transfer and the relation between rate and reaction enthalpy according to the Marcus theory (Marcus, 1956, 1963, 1965) in benzenoid species attached to a steroidal spacer (Closs and Miller, 1988; Closs et al., 1989; Liang et al., 1990); and (iii) the influence of chain length and chain conformation on electron transfer between metal complexes (Reimers and Hush, 1990; Gray and Malmstrom, 1989), e.g. at the ends of an oligopeptide chain (Isied et al., 1988). In contrast with these approaches, the present account is restricted to degenerate, intramolecular electron transfer occurring in the charged ground state of the bis-electrophore A-1-A. It should be emphasized that our approach is a purely empirical one: knowledge of the structural dependence of the charge-storage capacity and of intramolecular electron-hopping processes might enable us specifically to design dimers, oligomers and polymers in which the electron-transfer rate, and thus the resulting charge distribution, can be controlled precisely.

5

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

2

Design and synthesis

It is appropriate to begin with an overview of the compounds that have been specifically selected and synthesized for the present purposes. They are made up of redox-active subunits such as the benzenoid hydrocarbons naphthalene [ 11 and anthracene [2], the [4n]annulene cyclo-octatetraene [3] and the bridged (aromatic = 4n + 2) [ 14lannulene [4]. Each subunit is known to form a radical anion and dianion in chemical or electrochemical reductions (Meerholz and Heinze, 1989; Rabinovitz, 1988). Cyclo-octatetraene is an outstanding electrophore because, in the course of an electron transfer, its tub-shaped structure undergoes a flattening of the ring, while the other hydrocarbons possess rigid n-systems that do not experience extensive structural change upon undergoing a redox process (Katz, 1960; Anet et al., 1964; Heinze et al., 1974). R

2 m

3 4

0 0 -

5

When linking these redox-active building blocks to form dimers and oligomers, care has to be taken to vary the steric and electronic interaction of the subunits systematically. In [51 an orthogonal arrangement of the two of the subunits systematically. In [51 an orthogonal arrangement of the two cyclo-octatetraene units is enforced (via the spiroconjugation). Compounds [6]-[ 103 contain a flexible connection of the anthracene species, while in the arrangement. Another type of face-to-face-arrangement is found in the multi-layered annulene systems [ 141 and [ 151. Here, unlike the para-cyclophanes, in which the phenyl units are connected by “external” alkanediyl groups, the stacking of annulene layers is achieved by “internal” linkages, with the bridge located inside the n-clouds. The advantage of the latter structure is that one can

M. BAUMGARTEN ET AL.

6

&yJ C6l

a:n=2 b:n=3 c:n=4 d:n=6 e:n=ll

3

&

‘ 0

c71

C8l 0

a:n=3 b:n=4

%I&c91

0

0

0

0

0

L

0

0

Jn

a: m = 3, n = 1 b:m=3,n=2 c: m = 11, n = 2

readily vary the inter-plane distance and that the multi-layered system can be created in a true polymer forming reaction (see below). The common feature of compounds [5]-[15] is that the electrophoric units are linked by saturated spacers, thus establishing only weak electronic (through-bond or through-space) interaction of the .n-systems. In contrast, the binaphthyl [ 161, the biperylenyl [ 171 and the bianthryl [ 181 as well as the structurally related homologues [19], [20] and [21] allow for a direct .n,.n-interactionof the subunits; it will be shown, however, that for both steric and electronic reasons the inter-ring conjugation can be weak and thus lead to electronically independent redox groups in a similar fashion as in [51-[151.

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

[14]a: R = CH3

b: R = 2-dodecyl

[ 15]a: R = CHs

b: R = 2-dodecyl

%p%%

&6

%%

'0 0

I/

c 171

21 a : R = R ' = hexyl 21 b: R = H,R'= lsopentyl

T p

In

c 191

21c: R = hexyl

7

a

M. BAUMGARTEN H A L .

[22]

a:n=l b:n=2

c:n=3

Compounds [6] and [7] have been described in the literature (Mullen, 1987; Fiedler et al., 1986; Huber et al., 1983). The ortho-anthracenophane [13] has been prepared by Diels-Alder cycloaddition as part of a project devoted to ladder-type polymers (Wagner et al., 1988; Wegener and Miillen, 1991; Pollmann et al., 1990). Compound [ 171 as well as the oligomers and polymers [ 19)-[22] have been prepared recently using various methods of aryl-aryl coupling (Bohnen et al., 1990; Fahnenstich et al., 1989; Koch and Mullen, 1991; Baumgarten et al., 1992a; Schenk, 1989). The most appropriate method, however, for the versatile linking of conjugated hydrocarbons by saturated spacer groups is reductive alkylation (Mullen, 1984, 1986, 1987; Bender et al., 1988, 1989; Bender and Mullen, 1988; Krummel and Mullen, 1988).This approach is based on the formation of carbanionic hydrocarbons by reduction of conjugated n-systems or by deprotonation of dihydroprecursors and their subsequent reaction with electrophilic reagents such as haloalkanes in SN2-type reactions. For the synthesis of the bis-cyclo-octatetraene compound [51 (Krummel et al., 1987; Auchter-Krummel and Mullen, 1991), cyclo-octatetraene dianion was quenched with tetrabromoneopentane to give the bis-adduct [231, which exists in an equilibrium between valence isomers [23a) and [23bl. Hexacycle [23a] was actually isolated in about 60% yield (Fig. 2) (Krummel et al., 1987). Accordingly, in the subsequent dehydrogenation, the formation of [23a] must be avoided by working at low temperatures; in this case it was possible to deprotonate the originally formed isomer [23b], obtaining a

Fig. 2 Quenching of cyclo-octatetraene with tetrabromoneopentane.

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

9

tetra-anion that was then subjected to oxidation with cadmium chloride to obtain the target molecule [51. This concept of reductive alkylation can be extended to polymer synthesis (Bender et al., 1988, 1989). A typical starting compound is the anthracene dianion, which is treated with a bifunctional electrophile such as a 1,n-dihaloalkane to obtain the intermediate [241. The latter constitutes both a nucleophile and an electrophile and one can now attach a second carbanion such as [25] to the electrophilic end. This process, after protonation and subsequent dehydrogenation, gives rise to the dianthrylalkane compound [6]. A related procedure provides dianthryl compounds such as [7] in which the oligomethylene spacer is replaced by an oxoethylene chain (Fiedler et al., 1986). Under appropriate experimental conditions the bifunctional intermediate [241 can undergo polymerization. Not surprisingly, the polymerization also proceeds upon mixing of the bis-nucleophile [261 and the bis-electrophile [27]. The yield of the polymer is higher than 95%, the average molecular weight is about 10000, and the product possesses high structural homogeneity. Finally, the poly( dihydroanthrylene) systems can be dehydrogenated to obtain the corresponding unsaturated polyanthrylene compounds [lo] (Bender et al., 1988, 1989).

A crucial aspect of the present carbanion alkylation is its high regioselectivity. Thus, under the prevailing experimental conditions, the kinetically controlled alkylation only proceeds at the positions of the highest local n-charge of the carbanions. In the anthracene dianion or in [25] these positions are at C-9 and C-10, respectively. An important consequence of this regioselectivity is given for the example of the pyrene isomer [28]. The corresponding dianion can be regioselectively alkylated at the two inner positions since these are the positions of the highest local n-charge. When applying a haloalkane as electrophile, the bridged [ 14lannulene [4] is obtained, and by carefully controlling the stoichiometry of nucleophile and electrophile one can, for example, obtain the doubly layered system [ 141

10

M. BAUMGARTEN H A L .

and the triply layered system [ 151 with varying spacer groups. This reaction can even be extended to polymer synthesis yielding [29b1, where, again, the crucial intermediate [29a] is obtained after alkylation of the dianion [28]’with a bifunctional species such as [271 (Irmen et al., 1984; Alexander et al., 1989). According to crystal-structure analysis, the products do indeed adopt a face-to-face arrangement of the annulene layers (Irmen et al., 1984; Maresch et al., 1989). In general, the technique of carbanion alkylation is the method of choice for providing a broad series of compounds for the study of electron-transfer reactions. In this way, indeed, it is possible systematically to vary (i) the type of the n-conjugation and the rigidity of the electrophoric subunits; (ii) the length and conformational behaviour of the spacer; and (iii) the relative orientation and the degree of overlap of the n-systems. These structural factors are relevant for the redox-activity of bis- and oligo-electrophoric systems and for the reorganization energy associated with intramolecular electron hopping.

3

Highly charged states via extended redox sequences

The first question is whether the redox systems can be subjected to successive electron-transfer reactions in extended redox sequences. What one needs to know thereby are the number of charges that can be transferred and what is the Coulombic repulsion arising between the charged subunits. The experimental methods that have to be applied are obvious. Cyclic

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

11

voltammetry (Heinze, 1984) provides information on the number of charges being transferred and on the energies of the available redox states. The ionic species formed along an extended redox sequence can be characterized spectroscopically: diamagnetic products by nmr spectroscopy, and the paramagnetic products by esr and ENDOR spectroscopy (Eliasson et al., 1990; Huber and Mullen, 1986; Mullen, 1987; Klabunde et al., 1987; Gerson, 1967, 1970; Kurreck et al., 1988). Quenching reactions can also be applied since they reflect the number of charges in the reduced system, and they can prove that the framework of molecules remains intact upon charging. From the pronounced tendency of cyclo-octatetraene toward dianion formation upon chemical or electrochemical reduction, it is not surprising that the rigid bis-electrophore [51 transforms into a tetra-anion upon chemical or electrochemical reduction. The geometry of the tetra-anion does not lead to significant Coulombic repulsion. The effective compensation of electrostatic energies in oligocyclo-octatetraenyls is even more astonishing if the subunits are brought into conjugative interaction. This is the case in the linear n-chains [30] and [31] that have recently become available and

r

c:n=2

c 321

12

M. BAUMGARTEN E T A L .

in which each cyclo-octatetraene unit accepts two electrons upon reduction

( Auchter-Krummel and Mullen, 1991; Staley et al., 1985).

The situation is less obvious for other electrophores such as anthracene, which upon electrochemical reduction forms a dianion salt, although at a much more negative potential than cyclo-octatetraene. The redox activity of a corresponding “dimer” will, of course, depend upon the chemical mode of linking. The example of l,n-di(9-anthryl) alkanes [6] is revealing (Huber et al., 1983; Becker et al., 1991). Here, the Coulombic interaction between two monocharged anthracene units depends sensitively on the length of the alkanediyl spacer. This energy is reflected by the difference between the first and second reduction potentials in the cyclic voltammogram (Heinze, 1986). Thus an interaction energy can be detected for spacers shorter than butanediyl. Further charging steps up to tri- and tetra-anions are distinctly shifted to more negative potentials than for the anthracene dianion itself and, additionally, a higher repulsion between the doubly charged redox groups is reported (Mortensen et al., 1991). Accordingly, dianthryl compounds with a close proximity, such as [6] and even [ 181,are capable of accepting four electrons although an appreciable electrostatic repulsion is built up (Becker et al., 1991; Huber and Miillen, 1980).When considering the question of how a bis-electrophore accommodates the extra charge it is important to note that, for example, the tetra-anion of di(9-anthry1)ethane [61 adopts an anti-conformation with respect to the central C-C bond, thus minimizing the electrostatic repulsion (Huber et al., 1983). In this context it is noteworthy to refer to the unsaturated analogue 1,2-di(9-anthryl)ethene [ 321 (Weitzel and Mullen, 1990; Weitzel et al., 1990). Like [6] (Becker et al., 1991), compound [32] forms a stable dianion and tetra-anion upon reduction. In the cyclic voltammogram of [321, the first two electrons are transferred at nearly the same potential, pointing to an effective minimization of the Coulombic repulsion between the charged anthryl units (Bohnen et al., 1992). This situation, which again corresponds to that in [6], could imply a torsion about the central olefinic bond (Bock et al., 1989). In contrast to the conformationally mobile dianthrylethanes, the rigid cyclophane [ 111, with a face-to-face arrangement of the n-layers, is electrostatically less favourable for reduction and only gives rise to a dianion upon alkali-metal reduction (Huber et al., 1983). It is straightforward in the charging of a layered electrophore to increase the interplane distance of the n-layers and thus to “relax” the resulting Coulombic strain; this approach is discussed for the doubly layered annulenes [ 141. In this particular case, a stable tetra-anion is available which can be characterized by a highly resolved ‘H-nmr spectrum (Irmen et al., 1984).

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

13

There are two remarkable findings: (i) While the neutral compound is made up of two diatropic [14]annulenes, the tetra-anion is composed of two strongly paratropic dianionic subunits; therefore the reduction is accompanied by drastic chemical shift changes, i.e. the ring protons of the tetra-anion appear at very high field (A& ca. - 8 ppm) and the bridge protons at very low field (AdH ca. 18 ppm). (ii) Unlike the neutral compound, the tetra-anion exists in two isomers that from an analysis of the 'H-nmr chemical shifts can be identified as syn- and anti-conformers of [ 141. Surprisingly enough, these two conformational isomers of the tetra-anion do not interconvert even at about + 80°C. Another question concerns the behaviour of redox systems containing more than two separate electrophores. It is characteristic that the tetra-anthrylene [ lOc] with a long undecanediyl spacer group can be reversibly charged with two electrons per electrophore without significant electrostatic interactions between the anthracenes (Becker et al., 1991; Mortensen et al., 1991; Bohnen et al., 1992). A related redox activity is found for the trisanthrylene [ lOa] with short propanediyl spacers (Bohnen et al., 1992). It should be noted, however, that in the cyclic voltammogram there is only one wave for a three-electron transfer; in other words, each anthracene can be charged with one electron without the creation of strong Coulombic repulsion, and only further reduction will then give rise to detectable electrostatic effects. Independent evidence for the redox capacity of these systems can be obtained from quenching reactions; the structure of the hydro-derivative obtained upon protonation of the charged hydrocarbon, i.e. the number of electrophiles being incorporated, reflects the number of charges in the anion. Figure 3 indicates that one can thus chemically prove the formation of both a dianion and a tetra-anion. This example is significant because the reduction of 1,2-di-(9-anthryl)ethanehas been claimed in the literature to cause a rapid cleavage of the ethane o-bond (Gerson et al., 1976; Hammerich and Saveant, 1979). In contrast, under appropriate experimental conditions, even the tetra-anion is a stable species (Huber et al., 1983). The homologous series of oligo( 1.4-naphthylene)~[ 191(Anton et al., 1992) and oligo(9,lO-anthrylene)~ [21] (Baumgarten et al., 1992a) are examples of directly connected electroactive units. In their cyclic voltammograms the first reduction and oxidation potentials are independent of the chain length due to the steric hindrance of conjugation. The second charging step is facilitated with increasing number of electroactive units because of the smaller Coulombic repulsion between the charges residing on the outermost units.

+

14

M. BAUMGARTEN E T A L .

Fig. 3 Quenching of the di- and tetra-anions of 1,2-dianthrylethane [6a] with protons.

At the end of a charging process, nearly every n-unit is singly charged in the case of the naphthylene derivatives and doubly charged in the oligoanthrylenes. The conclusion from the above examples is that under appropriate experimental conditions these systems can be subjected to successive electron-transfer reactions forming highly charged derivatives with intact molecular frameworks. 4

Mechanism of successive electron transfers

As has been pointed out already, the description of successive reduction processes meets with a fundamental problem. If one electron has been injected into a bis-electrophoric system (see Fig. l), the second charge can enter the same unit, forming a dianionic moiety, or it can enter the second neutral unit. The former case is only conceivable if the two extra charges on one redox group give rise to conjugative or ion-pairing effects compensating for the electrostatic repulsion. In the latter, electrostatically more favourable, case the dianion could exist either as a singlet or a triplet species. The situation is even more complicated when a tris-electrophoric system is charged. The first question is which subunit will be charged initially. Then, if a dianion with the two electrons in separate electrophores has been formed, the charges can reside either on neighbouring electrophores or on those allowing the greatest possible distance between charges (see Fig. 1 ). What experimental evidence is available to clarify the resulting mode of spin-density and charge-density distribution? If one characterizes the reduction of (i) a single annulene, (ii) a doubly layered and (iii) a triply layered analogue by cyclic voltammetry, one derives a first criterion (Alexander et al., 1989; Bohnen et al., 1992; Fry et al., 1985). The potential

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

15

wave for the diannulenyl [14] shows two very close redox steps, each describing a one-electron transfer. The simulation of these waves provides the potential difference for the first and second reduction; from this, after correcting for the entropy term, one obtains an interaction energy of about 1.7 kcal mol- '. In a similar fashion, the triply layered analogue [ 151 reveals one wave indicating a three-electron transfer. The interaction energy corresponding to a two-fold charging of [ 151 is below 0.2 kcal mol-'. Accordingly, one must describe the dianion of [l5] as a species in which the first two extra electrons reside in the outer layers. Only the injection of the third electron, then, creates a significant interaction energy. It is clear that the charging sequence can be affected not only by the (inner/outer) position of the electrophore within the oligomeric chain, but also by substituents slightly affecting the redox potentials. The oligo( 1.4naphthy1ene)s [ 191carry t-butyl groups only in the terminal units, in contrast to the corresponding oligo( 1,5-naphthylene)s [203, which are substituted with (cation-stabilizing) alkyl groups in each unit. This different substitution pattern explains why the oxidation of [20] occurs with greater ease ( A E z 110 mV) than that of [19] (Bock et al., 1989; Anton et al., 1992). The tris-9,lO-anthrylenes [21a] and [2lb] (Baumgarten et al., 1992a) possess a different number of solubilizing alkyl groups in the central anthracene unit. As a result, the first oxidation of [21a] occurs at a potential 120 mV lower than for [21b] owing to the second stabilizing group. When , alkyl substituents are placed turning to the tetrameric anthrylene [2 1 ~ 1the on the terminal units, which are then preferentially oxidized to the dication, and consequently no Coulombic repulsion is seen. The reduction, on the other hand, is favoured in the central unsubstituted units and the second charging step takes place against a higher interaction energy. As we have shown previously for two different series of oligo( p-pheny1ene)s [22] and [33] (Bohnen et al., 1991; Heinze and Meerholz, 1990), the substitution of central units of a chain can influence the charging behaviour drastically. In [331 with methyl substituents, the conjugation is interrupted as soon as two substituents interfere, leading to hindered rotation around the connecting bond. Thus, independent units of biphenyl ([ 33a], [33cl and [33e]) or terphenyl ([33b] and [33d]) - depending on the hindrance between the subunits -are charged, even in the case of long phenylene chains. While cyclic voltammetric experiments provide thermodynamic and kinetic information on the charging processes (Heinze, 1986), only indirect information on the structure of the redox products is available. Fortunately, independent evidence can be obtained from spectroscopic experiments. Figure 4 depicts the esr spectra recorded for both a dianion and a trianion of [21a] in solid solution. From the zero-field splitting D, which is proportional to the inverse distance of the unpaired electrons, one can roughly

16

M. BAUMGARTEN ET AL.

estimate the location of the charge; there is no doubt that in the dianion the two electrons have entered the outer anthracene moieties. Analogous findings can be obtained for the related tetramer [21c] (Baumgarten et al., 1992a). The detection of zero-field splitting for dianions of [ 181 or [21] is very important; it reveals not only the existence of a triplet state, but it also provides information on the mode of spin density distribution. Even more

Fig. 4 Esr spectra of high-spin trianthrylene [2la] at 120K in MTHF/K: A, biradical triplet state [2la]*-' (labelled 0); B, triradical quartet state [21aI3-' (labelled *); C , biradical triplet state [21aI4-' (labelled m).

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

17

subtle structural information is available from the esr spectra, however. Thus, for the conformationally mobile dianthrylalkanes [6]-[9], the zero-field splittings observed for solid solutions of the dianions together with model considerations indicate that the oligomethylene spacers adopt an all-anti conformation, so that the redox-active subunits have the largest possible separation. A related finding appears for the dianions of the doubly layered system [ 141; their esr spectra taken in the glass provide evidence for a triplet structure, and from the D-value the inter-plane distance in the charged species can be readily determined (Irmen et al., 1984; Alexander et al., 1989). The important conclusion to be drawn from the cyclic voltammetric and esr spectroscopic data is that the charging of redox systems with electronically separate units follows a certain sequence. Thus, according to the esr spectra, the first electron will enter the inner unit of the triply layered system [ 151 while, according to the cyclic voltammogram and zero-field splitting derived from the esr spectra, the corresponding dianion has the two electrons in the outer layers. Accordingly, transition from the monoanion to the dianion will require a redistribution of charge, i.e. an electron-hopping process (Alexander et al., 1989).

5

Structural factors relevant t o intramolecular electron transfer

The occurrence of intramolecular electron-hopping processes will now be discussed in detail, considering on an empirical basis systematic variation of relevant structural factors in the substrates. First, however, a few theoretical and experimental aspects will be summarized. In a general description of intramolecular electron-transfer (ET) processes one has to differentiate between charge separation in donor/acceptor (D/A) systems via the formation of photoexcited states and a “charge-transfer” or “charge-shift” reaction that is thermally activated (Cannon, 1980; Fox and Chanon, 1988; Meyer, 1978). For systems such as A-1-A, where each subunit A can act as donor or acceptor, the latter case may also be described in some cases as resonance as in ( l ) , because the electronic configuration can be written without a

difference in energy. If we think of a charge transfer (Ulstrup, 1979) between particle A+‘/-‘ and A, however, it is possible to define a rate constant k for

18

M. BAUMGARTEN E T A L .

this process, which, according to the Arrhenius equation, possesses an exponential temperature dependence (2), where A is a proportionality factor,

E , the activation energy,k , the Boltzmann constant and T the temperature. Following Eyring’s transition-state theory (Eyring, 1935a,b), one may replace A by Z K , where Z is either the collision frequency in bimolecular reactions or the vibrational frequency in intramolecular reactions in bridged systems, and K is the transmission coefficient, i.e. the probability that activated complexes (transition states) yield the product. By using the free energy of activation AG*, Marcus (1956, 1963, 1965) described the rate constant as a function of AG* and the thermal energy k,T as in (3). At large values of r, k = Zrc(r)exp( -AG*/k,T).

(3)

is assumed to change exponentially with r. The kind of process involved (adiabatic or non-adiabatic) depends on the J may be coupling J between the wavefunctions Y(+”-’lo)and Y(ol+.’-’). expressed as a function of overlap S of both states (Heilbronner and Bock, 1978; Hoffmann, 1971) by (4), in which the parameters are defined as in (4a) K

and He, is the Hamiltonian for the system in the Born-Oppenheimer approximation. The adiabatic and non-adiabatic processes may be visualized in terms of two intersecting potential curves, one of them representing the electronic configuration ( +./-.lo) and the other one the configuration (01 +./ -.). This is demonstrated in Fig. 5A-C for the cases of very strong, strong and weak coupling, where the first two examples are generally called adiabatic and the last non-adiabatic. A very strong coupling (large value of J ) results in the formation of a single minimum surface with a symmetric chargedelocalized ground-state ion (either anion or cation, Fig. 5A). For somewhat smaller values of J the process is still adiabatic, and the resulting surface has a double-minimum potential. Interconversion between the degenerate charge-localized states takes place via the avoided crossing region on the

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

4

A

B

19

C

nuclear coordinates

Fig. 5 Potential energy hypersurfaces as a function of the reaction coordinate for adiabatic (A, single-minimum potential; B, double-minimum potential) and non-adiabatic (C) electron-transfer reactions.

lower surface, while the electronically excited state surface has a single minimum. The magnitude of the avoided crossing region is given by twice the electron coupling integral J (Fig. 5B).In the non-adiabatic region with two separate curves an energy of 1would be necessary for electron transfer between the states, if no reorganization of the solvation shell takes place. Thermal activation of the solvent to a configuration in which the free energy of the ion is unchanged, is, however, possible at lower energies. For isoenergetic parabolic potential curves the energy of this point is $2 (Fig. 5C). Comparing J with the thermal energy, one can differentiate between the two processes. For adiabatic surfaces the coupling J >> kb’T; while in the non-adiabatic case J << kb i? A quantitative description of the two types of process can be made using the Landau-Zehner parameter g (Ziilicke, 1985; Onuchic et al., 1986) defined in ( 5 ) . Here, o,is the frequency of the vibrational g=-

2rcIJ(’ ho,1

mode determining the ET. If g >> 1, the reaction is adiabatic, and for g << 1 it is non-adiabatic. If we consider ET in the ground state, we shall deal with an adiabatic case where the occurrence of charge transfer depends on the activation energy between the double-minimum potential (Meyer, 1978). When defining the activation energy AG +, one typically uses the approximations of harmonic

M . BAUMGARTEN ET AL.

20

oscillators and parabolic potential curves, yielding a quadratic dependence (6) on 1 and AGO (Marcus, 1956, 1963, 1965; Closs and Miller, 1988; AG+

= (AGO

+~ ) ~ / 4 1

(6)

Closs et al., 1989; Liang et al., 1990; Hush, 1961, 1975), where 1 is the total reorganization energy, and AGO the difference in free energy of the two minima. The reorganization energy 1 is the sum of two terms, As + ili, the inner sphere reorganization ,Ii for changes of D/A (structure, geometry, and vibrational states) and the solvent reorganization 1,.In a simple model A, can be expressed by (7) in terms of the optical (cop) and static ( E , ) dielectric

(

1, = e2 +al + +a2 -

--)(& 1

-

t)

(7)

constants of the medium, the radii of donors and acceptors ai, the electronic charge e and the centre-to-centre distance r (Closs and Miller, 1988; Closs et al., 1989; Liang et al., 1990; Wasielewski, 1989). il may be determined experimentally, for example from charge-transfer bands ( Wasielewski, 1989; Siders and Marcus, 1981) or from the self-exchange rates for identical subunits where AGO is zero and k,, depends only on 1 through (8) (Grampp et al., 1990a).

k,,

= ZKexp[

-

+

'1

4AkbT

In the quantum mechanical formulation of electron transfer (Atkins, 1984; Closs et al., 1986) as a radiationless transition, the rate of ET is described as the product of the electronic coupling term J 2 and the Frank-Condon factor FC, which is weighted with the Boltzmann population of the vibrational energy levels. But Marcus and Sutin (1985) have pointed out that, in the high-temperature limit, this treatment yields the semiclassical expression (9). k

= 4n ~

h

J2(

-)'"

1 4nlk,T

exp[ -

+

41kb T

(9)

Thus the distance dependence of ET can be divided into the distance dependence of J and that of 1.For J an exponential decrease with increasing distance is assumed (Wasielewski, 1989). A principal distinction is made between outer-sphere and inner-sphere mechanisms in ET reactions (Kochi, 1988). In the outer-sphere reactions the

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

21

donor and acceptor are viewed as separate units even in the transition state of the transfer, and the kinetics show an exponential decrease with increasing distance (Kuznetsov and Ulstrup, 198 1). In the inner-sphere mechanism a compact transition state is formed where the coordination spheres of D and A penetrate each other (overlap), and the structure of the reactants is changed. The outer-sphere reactions may be described by the Marcus theory. This approach is widely used for explaining inter- and intra-molecular chargetransfer processes including systems such as A-1-A (Kuznetsov and Ulstrop, 1981) as long as no direct overlap occurs between the orbitals of each A (Fukuzumi et al., 1980; Hay et al., 1975). A number of different techniques have been applied to test the distance and orientation dependence of ET reactions (Closs and Miller, 1988; Closs et al., 1989; Liang et al., 1990; Reimers and Hush, 1990; Fox and Chanon, 1988; Wasielewski, 1989; Paddon Row and Jordan, 1988; Joachim et al., 1990; McConnell, 1961). Our method of analysing the mode of charge distribution in charged species is esr spectroscopy, which defines the timescale of the detectable dynamic species (Gerson, 1967; Kurreck et al., 1988; Wertz and Bolton, 1972). If an electron transfer is slow relative to the esr timescale ( < l o - ’ s) the spectrum corresponds to that of monomeric model compounds with a single electrophore. If the hopping process is rapid on the esr timescale, one will detect an effective delocalization. The charged electrophores are prepared by chemical or electrochemical electron transfer (reduction or oxidation) in solution and exist as ion pairs (Hogen-Esch, 1977; Huber and Miillen, 1986; Gerson, 1967). This situation differs significantly from that in which, for example, radical anions are prepared by pulse radiolysis (Closs and Miller, 1988; Closs et al., 1989; Liang et al., 1990). In the latter only “naked” anions exist, which possess no interaction with counterions. As has been described above, intramolecular hopping of electrons between electrophoric units requires a reorganization of the ion pair and of the solvent shell. Not surprisingly, therefore, the rate constants of such processes are similar to those observed in pulse-radiolysis experiments. It should be added that under the prevailing experimental conditions (e.g. high dilution) intermolecular electron-exchange processes are slow and do not affect the appearance of the esr spectra (Kurreck et al., 1988; Becker et al., 1991). In several cases studied in the literature the rates of exchange processes are within the esr timescale and give rise to dynamic line broadening (Ward and Weissman, 1957; Freed and Fraenkel, 1964). With the assumption of a short time for the jump itself, an adequate description of the exchange process is given by the modified Bloch equation (Gutowsky and Holm, 1956). The lineshape for an equilibrium is then described by the complex transverse magnetization M . The alternating linewidth effects have been particularly

M. BAUMGARTEN E T A L .

22

useful within kinetic analyses (Freed and Fraenkel, 1964; Sullivan and Bolton, 1970). Once the rate constants have been evaluated by spectral analysis for a series of temperatures, the determination of the activation parameters is achieved. While similar examples will also be reported herein, the emphasis is on the discrimination between localization of spin density in a single redox unit and effective delocalization over the entire molecule, i.e. on a yes/no decision between two borderline cases. Thus, in the present approach, the major focus is on the question of how we can influence the “external” parameters like solvent and counterion and the “intrinsic” structural parameters within the systems A-1-A to force the electron-hopping process into the timescale of the experiment, or at least to establish clearly the borderline cases. That we are still looking at an electron-hopping process in the case of “effective charge delocalization” over the entire molecule and not at a pure resonance phenomenon may be reassured by VIS/NIR spectroscopy of the neutral and charged species; the absorption of a single chromophore should be detected unless a very fast process > lo’, Hz is taking place. THE NATURE OF THE SUBUNIT

The first structural factor that is relevant for electron hopping is the electrophoric subunit. The reduction of the spirodicyclo-octatetraenylsystem [ 5 ] again serves as a suitable introductory case. Upon prolonged metal contact in solution [ S ] can be charged to a tetra-anion, where the number of H- and 3C-nmr signals reflects the D,, symmetry (Krummel et al., 1987; Auchter-Krummel and Mullen, 1991;Krummel, 1988).The intermediate ions such as the monoanion and the dianion are more interesting since they can represent “mixed”-valence states (Reimers and Hush, 1990; McConnell, 1961). It follows from the esr spectrum of the monoanion, which closely corresponds to that of a “monomeric” cyclo-octatetraene radical anion, that the unpaired electron is localized on one ring and that, even at elevated temperatures, there is no rapid hopping between the two rings. A completely analogous finding is obtained for the corresponding dianion. The 13C-nmr spectrum of the latter can be looked at as a superposition of a “neutral” and a “dianionic” spectrum. It is thus clear that the two excess charges are both localized on one subunit; even at +90°C the spectra fail to show any changes and to reveal any hopping process. Another important point can be deduced from the splitting pattern of the methylene protons in the ‘H-nmr spectrum of the dianion salt [5I2-/2Li+. It appears from the equivalence of, for example, H-9,11 and the nonequivalence of H-9’,11’ that the charged ring is planar, while the uncharged ring is tub-shaped (Fig. 6) (Krummel et al., 1987). The latter

’

’

23

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

H - 9 a .b

H-9'a.b

T

O°C

I

11 8

I

6

.

.

~

'

~

t

'

4

'

~

'

~

~

'

T = -6OOC

.

2

.

"

'

.

'

~

~

0

'

.

.

Fig. 6 Nmr spectra ofthe dianion of [ 51 with the molecular positions and assignment of protons.

"

'

'

24

M. BAUMGARTEN E T A L .

undergoes a rapid ring inversion whose slow-exchange domain is reached at -60°C. The localization of two electrons in one ring and the failure to observe an intramolecular electron transfer in [ 51 and [512- is again related to the specific redox-behaviour of the parent system, cyclo-octatetraene (Anet et al., 1964; Krummel et al., 1987; Staley et al., 1985).It is characteristic for the latter that (i) upon injection of an electron, the ring experiences a flattening and that (ii) the electronic stabilization of the dianion overcompensates the Coulombic energy associated with the second electron transfer upon transition from the radical monoanion to the dianion (Katz, 1960; Anet et al., 1964; Eliasson and Staley, 1985). Such a flattening is also likely to occur for the second (uncharged) ring of [5]’- in the course of an intramolecular electron transfer. This flattening will require some steric strain, an energy term that contributes to the inner-sphere reorganization energy Izi of the intramolecular electron transfer, and this will increase the overall activation barrier. Accordingly, if we want to bring about a rapid intramolecular electron transfer between identical electrophores, we shall have to introduce rigid electrophores that do not experience a significant structural change upon charging. The “extra stabilization” of charged cyclo-octatetraene and the reluctance of cyclo-octatetraenes to be involved in intra- or inter-molecular electrontransfer processes is also documented by the behaviour of the cyclooctatetraenylenevinylenes [301 and [311 ( Auchter-Krummel and Mullen, 1991; Staley et al., 1985). In the dianion of e.g. [30a] and [31a] the charge is localized on one eight-membered ring, although a complete 7c-charge delocalization within the conjugated systems is not inhibited by a saturated spacer. The reason is again the steric strain associated with the flattening of the cyclo-octatetraene moiety. Characteristically enough, a charge-shift occurs in [30a12 - at temperatures above - 20°C whereas [3 la] does not exhibit dynamic behaviour even above room temperature. A localization of the spin density on one redox-group of a bis-electrophore has also been observed for the radical anion of spiro [5.51undeca-1,4,6,9tetraene-3,8-dione [341 (Gerson et al., 1972). The authors emphasize, however, that in this case the localization is mainly due to the interaction of the oxygen atoms with the counterions. In contrast to this finding, the unpaired electron is delocalized over both z-systems in the radical anion of the spirobifluorene [35] (Gerson et al., 1974). This seems to support the assumption that an effective delocalization can be brought about by introducing two rigid electrophores for which injection of an extra electron induces no conformational change; an orthogonal arrangement of the two electrophores, however, does not necessarily inhibit rapid electron transfer. In contrast to the dianion of the spirodicyclo-octatetraenyl [51, the dianion of the spirobifluorene [35] exists as a triplet biradical. This spin state of the -’

’-

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

25

dianion can be rationalized in terms of a double degeneracy of the lowest antibonding orbitals of [351. THE LENGTH OF THE BRIDGING GROUP

1,n-Di-(9-anthry1)alkanes [61 are suitable for investigating the influence of the distance between the single redox groups A, since the length of the conformationally mobile alkane chain 1 can readily be varied. As pointed out above, injection of an electron into the bis-electrophore forming a radical anion raises two main questions: (i) is the unpaired electron localized in one subunit or does a rapid electron-hopping process occur so that within the timescale of the esr experiment one observes an effective delocalization over both units; and (ii) can one clearly distinguish between the extremes experimentally? The obvious criteria are the magnitude -of the esr/ENDOR hyperfine coupling constants and the multiplicity of the hyperfine coupling constants, which reflects the number of equivalent hydrogen atoms coupling with the unpaired electron (Kurreck et al., 1988; Wertz and Bolton, 1972). It is clear from Fig. 7 that the potassium salt of the dianthrylethane radical anion in a mixture of dimethoxyethane (DME) and hexamethylphosphoramide (HMPA) as solvent has the unpaired electron effectively delocalized over both units (Fiedler et al., 1986). This outcome, however, depends on the

M. BAUMGARTEN E T A L .

26

A

B

Fig. 7 Esr and ENDOR spectra of 1,2-dianthrylethane [6a] for different ion-pair conditions: A, DMEiHMPA, K'; B, MTHF, K + at 190 K in solution.

length of the alkanediyl spacer group. Thus, an analogous finding, i.e. effective delocalization over both anthracene units, can be made for the propanediyl-, and butanediyl bridged system, while the corresponding dianthrylhexane radical anion shows localization of the unpaired electron in one anthracene unit under all experimental conditions, both in different solvent systems and at different temperatures. In the latter case, the esr- and ENDOR spectra closely correspond to those of the 9-alkylanthracene radical anion (Fiedler et al., 1986; Huber et al., 1983). Thus an electron transfer is hampered by the activation barrier between the double-minimum potential; we can no longer differentiate between an adiabatic process with large activation and a non-adiabatic process (McConnell, 1961; Kuznetsov and Ulstrup, 1981). This occurs even for the flexible alkanediyl bridging chains, where a rapid rotation and reorientation of the spacer and electrophores is possible in solution. Furthermore, the delocalization is strongly distance-dependent, so that a through-space mechanism seems to be responsible for the exchange, while through-bond transfer is not favourable in longer alkanediyl chains. The average distance between the electrophores is too large for an effective through-space coupling; therefore, no Coulombic interaction (CV) and zero-field splitting (esr) is observed in the doubly charged species, which should be a biradical (Becker et al., 1991). The distance dependence of the electron-transfer rate can also be demonstrated for the non-benzenoid systems [ 141 and [ 151, which

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

27

incorporate [ 14lannulenes (Alexander et al., 1989; Irmen et al., 1984). Like the dianthrylalkanes, they possess flexible spacer groups; the connection of the Ir-layers, however, enforces a direct n,n-interaction (Irmen et al., 1984). The radical anion of the bridged [14]annulene [14] can again serve as reference system. Its esr hyperfine coupling constants are in accord with a simple MO model (Mullen, 1978). When one proceeds to the radical anion of the analogous doubly layered system with a propanediyl chain as a spacer [ 14a], one observes only half the magnitude of the coupling constants (Irmen et al., 1984). This is convincing evidence for an effective delocalization of the spin density over both layers, and thus for rapid electron transfer. When the inter-plane distance is increased, however, the esr and ENDOR spectra closely correspond to those of the monomer. Consequently, by setting the appropriate length of the spacer, one can control the rate of the degenerate electron transfer and switch between a spin-localized and spin-delocalized situation. Accordingly, from spectroscopic evidence we can clearly distinguish whether radical anions derived from bis-electrophores exhibit rapid or slow electron hopping on the timescale of the experiment. Furthermore, there is a certain maximum of the spacer length beyond which intramolecular electron transfer is no longer detectable. For long, flexible spacer groups and electronically “decoupled” electrophores, the contribution of spacer reorganization to the activation barrier of the exchanger process is too large. A theoretical treatment of electron hopping in bis-electrophores with flexible spacers will have to consider different conformations of the spacer. Inclusion of rigid biselectrophores (with a fixed distance) is therefore relevant. The question of “how far an electron can be transferred” in radical anions of bis-electrophores has been the subject of several articles (Harriman and Maki, 1963; Shimada et al., 1972; Shimada and Szwarc, 1974, 1975; Gerson et al. 1984); the systems studied possess both flexible and rigid spacer groups between the electrophores. The findings accumulated for a series of 1,n-dinaphthylalkanes (Shimada et al., 1972; Shimada and Szwarc, 1975) are in qualitative agreement with the above results on dianthrylalkanes. An analogous scaling for rigid spacers does not exist, but it was demonstrated for compound [36a] and higher homologues [36b] and [36c] with more than one spiro-bonded cyclobutane unit (Shimada and Szwarc, 1974; Gerson et al., 1984) that electron hopping can occur for distances up to 0.8 nm. Long-range electron transfer has also been reported in a recent esr/ENDOR study of stable radical anions of naphthalenes attached to rigid norbornane spacers of variable length (Gerson et al., 1990). No counterion or solvent effects were found for the dinaphthobicycloheptadienc [37al with a small edge-to-edge (e-e) separation [ r = 0.24 nm(e-e); r = 0.63 nm (centre-tocentre)], and complete delocalization could generally be detected. The esr/ENDOR spectra of dinaphthotetracyclododecadiene [ 37b] and di-

M. BAUMGARTEN ET AL.

28

naphthohexacyclohexadecadiene [37c] revealed that the electron transfer between the naphthalene moieties in the radical anion is slowed down, but still within the esr timescale. This occurs even though the centre-to-centre distances between naphthylene groups are r = 0.8 nm in [37b] and r = 1.03 nm in [37c]. The long-range interaction of n-systems across rigid saturated bridges has also been studied by PES in the gas phase, where the ionization potentials provide a direct measure of 25 (Paddon-Row and Jordan, 1988). These studies demonstrated that through-bond coupling is considerable and easily exceeds through-space coupling in rigidly linked olefins. Linear extrapolation of J-values for the cation radicals of [38a; n = 13 and [38b] as a function of the number of single bonds separating the alkene moieties allowed estimation of the rate constants for thermal electron transfer in the ground state cation radicals of [ 38a; n = 2, 3 and 41 by use of equation (9). In a study of long range ET between aromatic donor (biphenyl) and acceptor molecules separated by steroid spacers [391, pulse radiolysis and electron beam techniques have been used for the injection of electrons (Closs and Miller, 1988; Closs et al., 1989; Liang et al., 1990). Here, the reaction rates (observed by changes in the absorption spectra) pass through a

[37b]

a

a

c401

29

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

maximum for different acceptors, and the results confirm the inverted region predicted by Marcus (A < (AG'I) (Miller et al., 1984). For a cyclohexane and decalin bridge, the dependence of ET rates on the orientation of a biphenyl donor and naphthyl acceptor was studied. Although a considerable difference in the centre-to-centre distance results for two decalin isomers (2,7-a,a and 2,7-e,e) [40a] and [40b], the reaction rates did not change in a dramatic way. Through-bond coupling is therefore assumed to dominate over a through-space mechanism. THE CONFORMATION OF THE BRIDGING GROUP

Turning again to flexible spacers, it appears that in oligo( anthrylene) species [lo] a long, flexible (CH,),-chain improves the solubility of the compound, which is particularly important in the related polymers. The question therefore arises whether one can bring about rapid electron transfer even for long spacer groups and thus whether a fine-tuning of the electron-transfer behaviour is possible for a given spacer length. The solution of the problem comes from a conformational analysis of oligomethylene chains, which are known to adopt preferentially an all-anti conformation (Dale, 1974). A different situation is encountered in oxaethylene chains, which prefer a gauche conformation with respect to the carbon-oxygen bond, thus giving rise to a helical conformation of the chain. The crystal structure of the biselectrophore [7a] provides a convincing example (Fiedler et al., 1986). The corresponding hydrocarbon species with the same spacer length, the dianthrylundecane [6e], forms a radical anion that always exists as a spin-localized species. The esr spectrum obtained for the radical anion of the structurally related ether [7a] (see Fig. 8) is temperature-dependent, which

Ill

A. \

T=233K

Fig. 8 Esr spectra of the alkoxy-bridged dianthryl [7] at two temperatures: A, 190 K; B, 233 K.

30

M. BAUMGARTEN E T A L .

indicates that intramolecular electron transfer occurs on the timescale of the experiment. The measurement has been performed on the potassium salt in dimethoxyethane, and similar results have been obtained for the lithium salt in 2-methyltetrahydrofuran. By means of lineshape analysis, we can determine the free energy of activation of the intramolecular electron transfer as AG* = 3.5 kcal mol-’. A similar rate-increasing effect is detected with the homologous ether [7b] (Fiedler et al., 1986). When rationalizing the significant difference of the hydrocarbon- and ether-bridged radical anions, the main aspect will certainly be the conformation of the oxyethylene chain, which brings the electrophores into closer contact. An additional aspect follows from the ability of the oxygen centres along the chain to chelate the counterion and thus to fix the cation between the electrophores. It is not possible from the available experimental evidence to discriminate between the two effects. The role of ion pairing and the relative position of the counterion and carbanion will be dealt with below. There is, however, another obvious effect of conformation upon the electron-transfer rate. In the p-anthracenophane [ 111, the mode of linking enforces a face-to-face arrangement of the 7c-layers (Huber et al., 1983; Gerson et al., 1976). Correspondingly, for its radical anion effective delocalization has been detected under all experimental conditions in solution. It is not clear from the available evidence whether a hopping process occurs, thus giving rise to a double-minimum profile, or whether there is a true spin-delocalized structure representing a single-minimum energy profile. This question will be answered below. When comparing the esr hyperfine coupling constants of [ 111-. with those of the flexible dianthrylethane radical anion [61 -’, a close correspondence can be seen between flexible and rigid species (Becker et al., 1991). Even the coupling constants of the methylene protons are very similar, although they are known to be sensitively dependent upon conformation. What these data seem to suggest is that the dianthrylethane radical anion adopts a conformation in solution with an (at least partially) eclipsed position of the electrophores. Such a structure would differ from that of the triplet dianion [6a]’-., where the zero-field splitting suggests an anti-conformation of the anthryl units with respect to the central C-C bond (Becker et al., 1991). In agreement with this are the results for other “phane” structures, which may also be compared with their corresponding flexible alkanediyl bridged forms as in paracyclophane [41] (McConnell, 1961; Gerson and Martin, 1969), syn- and anti- [2,2] naphthalenophane [42] (Gerson et al., 1976; Hamacher et al., 1986; Konishi and Reddoch, 1977) and syn- and anti-[2,2]( 1,4)-anthracenophane [ 121 (Nemoto et al., 1980). For bridged naphthalenes with a flexible alkanediyl chain [43] and a more rigid cyclohexane bridge [441 it was found that the intramolecular electron transfer

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

31

is around five times faster within the open chain structure possessing the same number of bonds as for the rigid cyclohexane (Shimada and Szwarc, 1974). Aromatic spacers have also been used; thus it could be shown for a phenylene-bridged diporphyrinyl system that electron hopping between both subunits can occur (Huber et al., 1990). Here the phenylene bridge is twisted out of conjugation with the two porphyrin rings, and the effect is observed for 1,4-(para) [45a] and the 1,3-(meta) [45b] substitution. While the rate of an electron-transfer process and thus the mode of the resulting spin-density distribution can be controlled by the length and the conformation of the spacer groups, the ion pairing creates an additional factor that can serve as some “external” control even for a given choice of subunits and spacers.

Plr

Ar

[45a] Ar=CCHs,M=Zn

Ar

[45b] Ar=C6Hs,M=Zn

32

M. BAUMGARTEN €T AL.

I O N PAIRING

It has been pointed out already that formation of a radical anion by a redox process in solution produces an ion pair and that any hopping of the electrons will thus be bound to the migration of the cation, which then becomes rate-limiting (Gerson et al., 1972, 1974, 1990). The ,ion-pair structure of the radicals is mainly affected by the size of the counterions and the ion-solvating capability of the solvent (Hogen-Esch, 1977; Szwarc, 1968). The example in Fig. 7 is significant for the reorganization of the ion-pair structure in the course of an electron transfer. It has been shown already that the radical anion of dianthrylethane as its potassium salt in a DME/HMPA solution possesses an effective delocalization of the spin (Huber et al., 1983; Becker et al., 1991). If this solvent system is replaced by 2-methyltetrahydrofuran, leaving all other conditions unchanged, one can readily switch from a spin-delocalized to a spin-localized situation. It is known from various pieces of evidence (Huber et al., 1983; Szwarc, 1974) that 2-methyltetrahydrofuran has a poor cation solvating ability and strongly favours contact ion pairs, i.e. a tight interaction of counterion and carbanion, while, on the other hand, HMPA and dimethoxyethane with their strong cation coordinating ability generate solvent-separated ion pairs. If an intramolecular electron transfer requires a reorganization not only of the redox units but also of the ion-pair structure, it is clear that the existence of a tight ion pair increases the overall activation barrier of the electron-hopping process. Not surprisingly, therefore, increasing cation coordinating ability of the solvent allows one to accelerate the intramolecular electron transfer. This feature provides the opportunity of controlling the electron-transfer rate just by the experimental conditions. It thus constitutes an “external” factor in contrast to the “internal” factors such as the nature of the redox group and of the spacer. The same “control mechanism” can be put to work for the doubly layered electrophores [ 141. From esr measurements it appears that a change only of the ion pairing brings about a different hopping rate and creates a different spin-density distribution within the timescale of the experiment. The important role of ion pairing for the rates of intramolecular electron transfers has been addressed in nearly all articles concerned with this subject. The most direct pieces of evidence for the high influence of anion-cation interaction result from the investigation of radical anions of phane species such as [ l l ] , [12] and [41]-[43] (Huber and Miillen, 1980; Gerson et al., 1976; Gerson and Martin, 1969; Hamacher et al., 1986; Konishi and Reddoch, 1977; Nemoto et al., 1980). Owing to the strong interaction of the two n-systems as a result of the face-to-face arrangement, rapid electron

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

33

transfer is expected for all these derivatives. However, even for the radical anions, the intramolecular self-exchange is somewhat slowed down in solvents of low cation-solvating power. For defining the timescale of the actual electron-transfer processes, it is crucial to apply solvents of high solvating ability, since otherwise the exchange rate is tied in with the significantly lower rate of concomitant migration of the cation. Fixing the location of the counterion midway between two identical electrophores has been achieved in the radical anion of the dibenzo [ 18lcrown-6 derivative [46] (Mazur et al., 1980). When its radical anion exists as the ion pair with Na or K +,the intramolecular electron transfer becomes detectable on the esr timescale. The activation energy determined for the electron transfer (1.4 kcal mol-') clearly demonstrates that in this case a significant contribution to the activation barrier from ion pairing can be negated. In the radical anions of the norbornane-linked naphthalenes [ 371 mentioned earlier (Gerson et al., 1990) no counterion effects were detected for [37a], which has a small spatial separation, but the esr/ENDOR spectra of [37b] and [37c] -. indicate that the electron-spin transfer between the naphthalene moieties is determined by the rate of synchronous counterion migration (Gerson et al., 1990). For tight ion pairs the electron is localized, while for loose ion-pair conditions, e.g. by using solvents of high cation-solvating power, the transfer becomes fast on the hyperfine timescale (k > lo7 Hz). Esr spectroscopy has also been used to study pure solvent dynamics in electron self-exchange reactions (Grampp et al., 1990a; Grampp and Jaenicke, 1984a,b). When the systems are not linked by a spacer (i.e. TCNQ-'/TCNQ (TCNQ = tetracyanoquinodimethane), the homogeneous bimolecular rate constants khom are given by ( 1O), with kA the association constant and kET +

-'

the actual electron-transfer rate, comparable to our intramolecular exchange rate. The solvent dependence of the rate constant is explained in terms of Marcus theory by using an outer-sphere reorganization energy. This leads to an experimental reaction distance between the two molecules in the transition state for this adiabatic process (Grampp et al., 1990a). As an extension of the intermolecular self-exchange described above, the solvent-induced intramolecular electron exchange kinetics in radical anions of 1,3-dinitrobenzene [47] and benzene 1,3-dicarbaldehyde [48] have been studied by several authors (Freed and Fraenkel, 1964; Grampp et al., 1989, 1990b; Shohoji et al., 1987).The advantage of [47] and [48] is their structural simplicity and their high stability, which allows measurements even in protic

34

M.BAUMGARTEN ET AL NC

ray

CN

0

solvents; in particular, the fixed distance and orientation of the redox centres open the possibility of a straightforward application of theoretical models. From alternate line-broadening effects in the esr spectra, the solventdependent intramolecular exchange rates can be derived, which in turn allows the calculation of values for the outer-sphere reorganization energy. If, on the other hand, the electron transfer in solution is determined by some rearrangement within the ion-pair structure, it is crucial to investigate the feasibility of electron transfer for an immobilized ion-pair structure in the solid state. For three substrates, the doubly layercd annulene [ 14a], the dianthrylethane [6a] and the anthracenophane [ 111, the radical anions were prepared under experimental conditions bringing about a spin-delocalized structure in solution. The solutions were then frozen into a glass by cooling, and the solid-state ENDOR spectra measured using the Davis pulse method (Davis, 1974; Grupp and Mehring, 1990; Rautter et al., 1992). It appears from the interpretation of the experimental spectra that, independently of the temperature in the glass, there is only a spin-localized structure. This occurs in the case of a doubly layered system [14a] and for the dianthryl [6a] (Fig. 9) (Rautter et al., 1992). What is even more surprising is that it also holds for the anthracenophane [ 111 with a short interplane distance of about 3 A. In solution the slow-exchange domain of the hopping process could never be achieved. Accordingly, if in a rigid system with strong n,n-overlap of the electrophoric group the position of the cation is fixed, the localization of the electron on one side of a bis-electrophoric molecule is enforced. Computer simulation of the experimental spectra in the frozen solution points to a spin-localized structure independent of the temperature. This interpretation is unambiguous because, in the solid state, the dipolar contributions of the hyperfine interaction A( dip) for all three principal

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

35

Fig. 9 Pulse ENDOR spectra of 1,3-dianthrylpropane[6b] at different temperatures.

directions must sum to zero, leaving the isotropic components a(iso), which are directly proportional to the spin densities at the carbon centres [see ( 1 la,b)]. Furthermore, by warming the samples until the solutions liquify A

= a(iso)

+ A(dip)

( 1l a )

(ca. 200 K), a drastic decrease in hyperfine splittings and thus spin densities

to one-half of the original values can be observed within the same experiment. For the examples of [11]-'/K+ and [14]-'/K+, it could also be proved that the transition between localized and delocalized states takes place at the phase transition between solid and liquid state (Rautter et al., 1992). In order to obtain kinetic parameters for the electron transfer of [ 1l ] - ' / K + , the dephasing time t, of the electron-spin echo near the phase-transition temperature T, was measured. These"experiments gave a correlation time t , of 100 ns for the electron transfer at T, = 170 K. From the assumption of an exponential decrease oft, in solution, a value of 100 ps was estimated for t , at room temperature (Rautter, 1989; Rautter et al., 1992).

36

M. BAUMGARTEN ET AL.

It has been shown so far that “internal” and “external” factors can be combined in the control of the electron-transfer rate. Although in most cases a simple theoretical treatment, e.g. by the Marcus approach, is prevented by the coincidence of these factors, it is clear that the observed features for the isoenergetic self-exchange differ by the electronic coupling and the free energy of activation. Then it is also difficult to separate the inner- and outer-sphere reorganization energies. THE MODE OF LINKING

It will be shown that even more subtle structural changes can bring about drastic consequences in the electron-transfer kinetics. At first glance, the “dianthryl” compounds [61, [81 and [9] are closely related since they possess the same electrophoric subunit and the same ethanediyl bridge (Becker et al., 1991).Yet it appears that the different mode of linking profoundly affects the redox behaviour. The first evidence comes from cyclic voltammetric data. The question is whether the presence of the second redox group could facilitate the reduction oT the first one. This aspect could in principle be evaluated by comparison with the first redox potentials of suitably substituted monomeric analogues. More important in the present context is the interaction energy between the two electrophores, each carrying one extra electron. The experimental measure of this interaction is clearly the difference AE 1 , 2 between the first and second redox potentials (Bohnen et al., 1992). Considering first the anthracenophanes [ l l ] and [12], it is quite reasonable that the strong interaction of the n-systems as a result of the face-to-face arrangement should give rise to a significant potential difference upon dianion formation. More interestingly, a significant potential difference AE1,2can be detected for the di(9-anthry1)ethane [6] but not for the isomeric species [8] and [9] in which the bridge is attached to the centres C-1 and C-2 respectively. Even in the ortho-cyclophane [ 131 with two linkages there is no detectable interaction energy (Becker et al., 1991). It is thus possible to classify the closely related dianthryl systems according to the degree of interaction of the subunits; the question that immediately follows is whether such a clustering also develops from esr spectroscopic study of the corresponding radical anions. In the radical anions of the di(2-anthryl) system [9] and of the ortho-cyclophane [ 131 a spin-localized situation prevails independently of temperature and ion pairing. On the other hand, for the corresponding di(9-anthryl) species and even for the di( 1-anthryl) species a rapid hopping process can be detected if suitable ion pairing is established (Becker et al., 1991). Thus, according to both cyclic voltammetric and esr spectroscopic

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

37

evidence, the closely related dianthrylethane systems can be classified into three groups: (i) compounds [ 111 and [ 123 with strong interaction energy whose radical anions under all experimental conditions undergo rapid electron hopping in solution;

(ii) compounds [6a] and [8] whose radical anions show rapid or slow hopping, that is spin localization or effective delocalization, depending on the experimental conditions; and (iii) compounds [9a] and [ 131 with weak interaction, whose radical anions fail to exhibit electron transfer on the timescale of the experiment. When rationalizing the different behaviour of closely related biselectrophores, one must be aware that the intramolecular electron-transfer process requires some mixing of product and educt wave functions. The mode of linking seems to allow a fine tuning of the coupling integral J. The degree of interaction between the starting and the final situation will, of course, depend upon the spin density at the bridgehead position because these positions are closest in space. This may also be expressed by the distances between the average centre of spin density of the single electrophore. The lowest unoccupied molecular orbital of anthracene (see Fig. lo), i.e. the orbital that actually accepts the excess electron, has a high AO-coefficient at C-9 and a relatively small AO-coefficient at C-2 (Heilbronner and Bock, 1978). What is found experimentally is that rapid hopping is detectable in all those cases in which the linking is at C-9 or at C-1, i.e. at centres with a large or moderate AO-coefficient. This may be expressed in terms of through-space and through-bond mechanisms, which are both favoured in the case above, while a through-bond mechanism becomes much less probable as soon as the spin density at the bridgehead position is small as for [9] and [13]. Even though [9] is much more flexible than [13], which has two bridges, the situation encountered is comparable. It seems that a second bridge at a position of low spin density does not improve the through-bond exchange and thereby diminish the activation barrier. Thus it appears that

Fig. 10 Atomic orbital coefficients for the LUMO of anthracene.

M. BAUMGARTEN ET AL.

38

even with the same electrophore and with the same linking group very subtle structural changes, i.e. the position of attachment of the spacer to the electrophore, can influence the degree of interaction and the rate of electron hopping. Biaryl compounds, although formally possessing direct n-conjugation between the aryl moieties, can give rise to electronically independent subunits as outlined in Section 3. From the viewpoint of redox chemistry, there are two basic criteria by which to classify this situation: for "decoupled" electrophores (i) injection of one electron will give rise to spin localization and (ii) injection of a second electron may produce a triplet dianion. Clearly such a situation does not prevail in biphenyl and its analogue the 1,l'-binaphthyl derivative [ 161. In the radical anion of the latter, one detects a set of coupling constants [ a H= 0.39 mT ( n = 2), 0.225 mT ( n = 2), etc.], which indicates a simultaneous distribution of the spin density over both entities. From the asymmetry of spin density at positions 4 and 5, a twisting angle of 50" has been determined (Koch, 1991; Baumgarten et al., 1992b). Another extreme is 9,9'-bianthryl [ 181; as a result of the near orthogonality of the anthryl rings, the dianion exists as a triplet, and, in agreement with this finding, the esr spectrum of the biradical dianion shows a large zero-field splitting D = 15.5 mT and a half-field signal for the Ams = 2 transition (Baumgarten et al., 1992a). When considering the prevailing non-bonded interaction in the binaphthyl [ 161 and the corresponding biperylenyl [ 171,

MA h

T=240K

Fig. 11 Temperature-dependent esr spectra of the monoanion of biperylene [ 171 in MTHF.

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

39

a similar inter-ring torsional angle would be expected. It is characteristic, however, that the esr spectrum of [ 171 reveals also the presence of localized forms for strong counterion conditions in solution, while lowering the temperature to the frozen state (ca. 160K) yields nearly exclusively the localized structure (Fig. 11). If the inter-ring torsional angle is comparable in both cases, the differences in the exchange rates may be traced back to the A 0 coefficients and spin densities at the bridgehead position ( p = 0.2 vs. 0.13) and the large differences in the centre-to-centre distance (0.49 vs. 0.87 nm).

6

Conclusion and outlook

The rate of degenerate intramolecular electron-transfer processes in biselectrophoric redox systems, and the observed spin-density distribution over one or two units depend upon the overall reorganization energy and thus upon

(i) the nature of the subunit; (ii) the spacer between the subunits; (iii) the ion pairing; and (iv) the way in which the subunits are linked. In essence, these empirical findings allow control of the rate of electrontransfer processes by creating the appropriate structural conditions. It is, of course, straightforward to extend such a correlation of structure and electron-transfer kinetics to higher homologues. The esr spectra of the radical anion obtained for the triply layered annulene system [lS] points to a situation with one electron being effectively delocalized over three layers (Alexander et al., 1989). The same holds true for the higher homologues. A closely related finding can be made for the anthrylene structure [lo], i.e. a redox system with three anthracene electrophores. It is obvious from the esr data for the radical anion that, depending on the ion pairing, the unpaired electron resides on the inner anthracene unit or is effectively delocalized over the whole chain (Fiedler et al., 1986). Accordingly, when one proceeds to the related oligomers and polymers, by a proper choice of the subunit, the bridging group and the ion pairing, one can control the energy profiles of intramolecular charge-transport processes in an analogous fashion. Systems such as [6], [lo], [I81 or [21] have also been the subject of photophysical studies (Rettig, 1988; Mataga et al., 1989; Yao et al., 1989)

40

M. BAUMGARTEN € T A L .

in which energy transfer between the separate anthracene chromophores is considered as a function of the molecular geometry. Although a direct comparison is not straightforward, for example, in view of the role of the ion pairing, it is a tempting approach to compare the processes of energy and charge transport (Heine et al., 1990). Another aspect is that intramolecular electron transfer is a fundamental process in intramolecular electronics in which a single molecule is used to process electrical signals (Joachim et al., 1990; Joachim and Launay, 1990). In this context, the electron transfer through a spacer group has been related to its “conductance” if placed between two conducting wires. Thus, accepting the definition of conduction of a single molecule or spacer group, the control of an electron transfer between redox centres closely corresponds to the control of the current passing through an electrical circuit. Upon electrochemical oxidation at a platinum electrode, many polycyclic compounds such as naphthalene, fluoranthene and pyrene form deeply coloured, crystalline radical cation salts (Maresch et al., 1989; Enkelmaan et al., 1985; Endres et al., 1985; Krohnke et al., 1980). According to crystal structure analysis, these salts adopt a stack-type structure with a face-to-face arrangement of the n-layers and an alternating array of neutral and monocationic units. The complexes have a high electrical conductivity along the stacking axis, and one way of explaining this conductivity is based on the assumption of a charge-hopping process between the n-layers. Such a charge-transport mechanism is thus closely related to the electron transfer in multi-layered systems such as [lo], [l5] and [21].

References Alexander, J., Ehrenfreund, M., Fiedler, J., Huber, W., Rader, H.-J. and Miillen, K. (1989). Angew. Chem. 101, 1530; Int. Ed. Engl. 28, 1531 Anet, A. L., Bourn, A. J. R. and Lin, Y. S. (1964). J . Am. Chem. SOC. 86, 3576 Anton, U., Bohnen, A., Koch, K.-H., Naarmann, H., Rader, H.-J. and Miillen, K. (1992). Adu. Mat. 4,91-93 Atkins, P. W. ( 1984).Molecular Quantum Mechanics. Oxford University Press, Oxford Auchter-Krummel, P. and Miillen, K. (1991). Angew. Chem. 103, 996; Znt. Ed. Engl. 30, 1003 Baughman, R. H., Bredas, J. L., Chance, R. R., Elsenbaumer, R. L. and Shacklette, L. W. (1982). Chem. Rev. 82, 209 Baumgarten, M., Miiller, U., Bohnen, A. and Miillen, K. (1992a). Angew. Chem. 104, 482-485 Baumgarten, M., Koch, K.-H. and Miillen, K. (1992b). Accepted for publication Becker, B., Bohnen, A., Ehrenfreund, M., Wohlfarth, W., Sakata, Y., Huber, W. and Miillen, K. (1991). J. Am. Chem. SOC. 113, 1121 Bender, D. and Miillen, K. (1988). Chem. Ber. 121, 1187

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

41

Bender, D., Herbst, H., Schade, P. and Miillen, K. (1988). Chem. Ber. 121, 1177 Bender, D., Przybylski, M. and Miillen, K. (1989). Makromol. Chem. 190, 2071 Bitthin, R., Ely, G., Wolffler, F., Munstedt, H., Naarmann, H. and Naegele, D. (1987). Makromol. Chem. Macromol. Symp., 8, 51 Bock, H., Ruppert, K. and FenskelD: (1989). Angew. Chem. 101, 1717; Int. Ed. Engl. 28. 1685 Bohnen, A., Koch, K.-H., Liittke, W. and Miillen, K. (1990). Angew. Chem. 102, 548; Int. Ed. Engl. 29, 525 Bohnen, A., Heitz, W., Miillen, K., Rader, H.-J. and Schenk, R. (1991). Makromol. Chem. 192, 1679 Bohnen, A., Rader, H.-J. and Miillen, K. (1992) Synth. Met., 47, 37-63. Cannon, R. D. ( 1980). Electron-Transfer Reactions. Butterworth, London Closs, G. L. and Miller, J. R. (1988). Science 240, 440 Closs, G. L., Calcaterra, L. T., Green, N. J., Penfield, K. W. and Miller, J. R. (1986). J . Phys. Chem. 90, 3673 Closs, G. L., Johnson, M. D., Miller, J. R. and Piotrowiak, P. (1989). J . Am. Chem. SOC. 111, 3751 Dale, J. (1974). Tetrahedron 30,1683 Davies, E. R. (1974). Phys. Lett. 47A, 1 Eliasson, B. and Staley, S. W. (1985). Prepr. Am. Chem. SOC.Diu. Pet. Chem. 30, 620; Chem. Abstr. 104, 50494a Eliasson, B., Edlund, U. and Miillen, K. (1986). J . Chem. SOC.Perkin Trans. 11, 937 Eliasson, B., Johnels, D., Sethson, I., Edlund, U. and Miillen, K. (1990). J . Chem. SOC. Perkin Trans. 11, 897 Endres, H., Keller, H. J., Miiller, B. and Schweitzer, D. (1985).Acta Cryst. C41,607 Enkelmann, V., Gockelmann, K., Wieners, G. and Monkenbusch, M. (1985). Mol. Cryst. Liq. Cryst. 120, 195 Eyring, H. (1935a). J . Chem. Phys. 3, 107 Eyring, H. (1935b). Chem. Rev. 17, 65 Fahnenstich, U., Koch, K.-H. and Miillen, K. (1989). Makromol. Chem., Rapid Commun. 10, 563 Fiedler, J., Huber, W. and Miillen, K. (1986). Angew. Chem. 98, 444; Int. Ed. Engl. 25,443 Flanagan, J. B., Margel, S., Bard, A. J. and Anson, F. C. ( 1978). J . Phys. Chem. 82,1101 Fox, M. A. and Chanon, M. (1988). Photoinduced Electron-Transfer. Elsevier, Amsterdam Freed, J. H. and Fraenkel, G. K. (1964). J . Phys. Chem. 41, 699 Fry, A. J., Powers, T. A., Miillen, K. and Irmen, W. ( 1985). Tetrahedron Lett. 26,4879 Fukuzumi, S., Wong, C. L. and Kochi, J. K. (1980). J . Am. Chem. SOC. 102,4790 Gerson, F. ( 1967). HochauJlosende ESR-Spektroskopie. Verlag-Chemie, Weinheim; ( 1970). High Resolution ESR Spectroscopy. Verlag-Chemie, Weinheim Gerson, F. and Martin, W. A. Jr (1969). J . Am. Chem. SOC. 91, 1883 Gerson, F., Gleiter, R., Moshuk, G. and Dreiding, A. S. (1972). J . Am. Chem. SOC. 94, 2919 Gerson, F., Kowert, B. and Peake, P. M. (1974). J . Am. Chem. SOC. 96, 118 Gerson, F., Martin, W. B. Jr and Wydler, C. (1976). J . Am. Chem. SOC. 98, 1318 Gerson, F., Huber, W., Martin, W. B. Jr, Caluwe, P., Pepper, T. and Szwarc, M. (1984). Helu. Chim. Acta 67, 416 Gerson, F., Wellauer, T., Oliver, A. M. and Paddon-Row, M. N. (1990). Helu. Chim. Acta. 73, 1586

42

M. BAUMGARTEN ET AL.

Grampp, G. and Jaenicke, W. (1984a). Ber. Bunsenges. Phys. Chem. 88, 325 Grampp, G. and Jaenicke, W. (1984b). Ber. Bunsenges. Phys. Chem. 88, 335 Grampp, G., Sohoji, M. C. B. L. and Herold, B. J. (1989). Ber. Bunsenges. Phys. Chem. 93, 580 Grampp, G., Harrer, W. and Hetz, G. (1990a).Ber. Bunsenges. Phys. Chem. 94,1343 Grampp, G., Sohoji, M. C. B. L., Herold, B. J. and Steenken, S. (1990b).Ber. Bunsenges. Phys. Chem. 94, 1507 Gray, H. B. and Malmstrom, B. G. (1989). Biochemistry 28, 7449 Grupp, A. and Mehring, M. (1990). In Modern Pulsed and Continuous Wave Electron Spin Resonance, p. 195. Wiley, New York Gust, D., Moore, T. A,, Moore, A. L., Barrett, D. and Harding, L. 0. (1986). J . Am. Chem. SOC.108, 8028 Gust, D., Moore, T. A., Moore, A. L., Barrett, D., Harding, L. O., Makings, L. R., Liddell, P. A., De Schryver, F. C., Van der Auweraer, M., Bensasson, R. V. and Rougee, M. (1988). J . Am. Chem. SOC.110, 321 Gutowsky, H. S. and Holm, C. H. (1956). J . Chem. Phys. 25, 1228 Hamacher, V., Plato, M. and Mobius, K. (1986). Chem. Phys. Lett. 125, 69 Hammerich, 0. and Saveant, J.-H. (1979). J . Chem. SOC. Chem. Commun., 938 Harriman, J. E. and Maki, A. W. (1963). J . Chem. Phys. 39, 778 Hay, P. F., Thibeault, J. C. and Hoffmann, R. (1975). J . Am. Chem. SOC.97, 4884 Heilbronner, E. and Bock, H. (1978). Das HMO Model1 und seine Andwendungen. Verlag-Chemie, Weinheim/New York Heine, B., Sigmund, E., Maier, S., Port, H., Wolf, H. C., Effenberger, F. and Schlosser, H. (1990). J . Mol. Electron. 6, 51 Heinze, J. (1984). Angew. Chem. 96, 823; Int. Ed. Eng. 23, 831 Heinze, J. (1990). Top. Curr. Chem. 152, 1 Heinze, J. and Meerholz, K. (1990). Angew. Chem. 102, 695; Int. Ed. Engl. 29, 692 Heinze, J., Serafimov, 0.and Zimmermann, H. E. (1974). Ber. Bunsenges. Phys. Chem. 78, 652 Hoffmann, R. (1971). Acc. Chem. Res. 4, 1 Hogen-Esch, T. E. (1977). Adu. Phys. Prog. Chem. 15, 153 Huber, M., Kurreck, H., Maltzan, B. von, Plato, M. and Mobius, K. (1990). J . Chem. SOC.Faraday Trans. 86, 1087 Huber, W. and Miillen, K. (1980). J . Chem. SOC. Chem. Commun., 698 Huber, W. and Miillen, K. (1986). Acc. Chem. Res. 19, 300 Huber, W., Unterberg, H. and Miillen, K. (1983). Angew. Chem. 95, 239; Int. Ed. Engl. 22, 242 Hush, N. S. (1961). Trans. Faraday SOC.,57, 557 Hush, N. S. (1975). Chem. Phys. 10, 361 Irmen, W., Huber, W., Lex, J. and Miillen, K. (1984). Angew. Chem. 96, 800; Int. Ed. Engl. 23, 818 Isied, S. S., Vassilian, A., Wishart, J. F., Creutz, C., Schwarz, H. A. and Sutin, N. (1988). J . Am. Chem. SOC. 110, 635 Joachim, C. and Launay, J. P. (1990). J . Mol. Electron. 6, 37 Joachim, C., Launay, J. P. and Woitellier, S. (1990). Chem. Phys. 147, 131 Katz, T. J. (1960). J . Am. Chem. SOC.82, 3784 Klabunde, K.-U., Huber, W., Lex, J. and Miillen, K. (1987). Chem. Ber. 120, 1491 Koch, K.-H. (1991). Dissertation, Mainz Koch, K.-H. and Miillen, K. (1991). Chem. Ber. 124, 2091. Kochi, J. K. (1988). Angew. Chem. 100, 1331; Int. Ed. Engl. 27, 1227

ORGANIC REDOX SYSTEMS WITH MULTIPLE ELECTROPHORES

43

Konishi, S. and Reddoch, A. H. (1977). J . Magn. Res. 28, 41 Krohnke, C., Enkelmann, V. and Wegner, G. (1980). Angew. Chem. 92,941; Int. Ed. Engl. 19, 912 Krummel, G., Huber, W. and Miillen, K. (1987). Angew. Chem. 99, 1305; Int. Ed. Engl. 26, 1290 Krummel, G. (1988). Ph.D. dissertation, Mainz Krummel, G. and Miillen, K. (1988). Tetrahedron Lett. 29, 2809 Kurreck, H., Kirste, B. and Lubitz, W. (1988). Electron nuclear double resonance spectroscopy of radicals in solution. In Methods in Stereochemical Analysis (ed. A. P. Marchand). VCH Publishers, Weinheim/New York Kuznetsov, A. M. and Ulstrup, J. (1981). J . Chem. Phys. 75, 2047 Lewis, I. C. and Singer, L. S. (1965). J . Chem. Phys. 43, 2712 Liang, N., Miller, J. R. and Closs, G. L. (1990). J . Am. Chem. SOC. 112, 5353 Marcus, R. A. (1956). J . Chem. Phys. 24, 979 Marcus, R. A. (1963). J . Phys. Chem. 67, 853, 2889 Marcus, R. A. (1965). J . Chem. Phys. 43, 679 Marcus, R. A. and Sutin, N. (1985). Biochim. Biophys. Acta 811, 265 Maresch, G. G., Mehring, M., Schiitz, J. U. von, Werner, H.-P., Gokelmann, K., Enkelmann, V., Miillen, K. and Klabunde, K.-U. ( 1989).J . Chem. Phys. 91,4543 Mataga, N., Yao, H., Okada, T. and Rettig, W. (1989). J . Phys. Chem. 93, 3383 Mazur, S., Dixit, V. M. and Gerson, F. (1980). J . Am. Chem. SOC. 102, 5343 McConnell, H. M. (1961). J . Chem. Phys. 35, 508 McDiarmid, A. D. (1979). J . Chem. SOC.Chem. Commun., 594 Meerholz, K. and Heinze, J. (1989). J . Am. Chem. SOC. 111, 2325 Meerholz, K. and Heinze, J. (1990). J . Am. Chem. SOC. 112, 5142 Meyer, T. J. (1978). Acc. Chem. Res. 11, 94 Miller, J. R., Calcaterra, L. T. and Closs, G. L. (1984). J . Am. Chem. SOC. 106, 3047 Mortensen, J., Heinze, J., Herbst, H. and Miillen, K. (1992). J . Electround. Chem., accepted for publication Miillen, K. (1978). Helu. Chim. Acta 61, 2307 Miillen, K. (1984). Chem. Rev. 84, 603 Miillen, K. (1986). Pure Appl. Chem. 58, 177 Miillen, K. (1987). Angew. Chem. 99, 192; Int. Ed. Engl. 26, 204 Miillen, K., Heinz, W., Klarner, F.-G., Roth, W. R., Kindermann, I., Adamczak, O., Wette, M. and Lex, J. (1990). Chem. Ber. 123, 2349 Murray, R. W. (1984).In Electroanalytical Chemistry, Vol. 13 (ed. A. J. Bard), p. 191. Dekker, New York Nemoto, F., Ishizu, K., Toyoda, T., Sakata, Y. and Misumi, S. (1980).J . Am. Chem. SOC. 102, 654 Nowak, R. J., Schultz, F. A,, Umana, M., Lam, R. and Murray, R. W. (1980). Anal. Chem. 52, 315 Onuchic, J. N., Beratan, D. A. and Hopfield, J. J. (1986).J . Phys. Chem. 90, 3707 Paddon Row, M. N. and Jordan, K. D. (1988). In Modern Models of Bonding and Delocalization (ed. J. F. Liebman and A. Greenberg), Chap. 3. VCH Publishers, New York Pollmann, M., Wohlfarth, W., Lex, J. and Miillen, K. ( 1990).Tetrahedron Lett. 31,2701 Rabinovitz, M. (1988). Top. Curr. Chem. 146, 99 Rautter, J. (1989). Diploma thesis, Stuttgart Rautter, J., Grupp, A., Mehring, M., Alexander, J., Miillen, K. and Huber, W. (1992). Mol. Phys. accepted for publication

44

M. BAUMGARTEN ET AL.

Reimers, J. R. and Hush, N. S. (1990). Inorg. Chem. 29, 3886 Rettig, W. (1988). Mol. Struct. Energetics 6, 229 Salem, L. (1966). The Molecular Orbital Theory of Conjugated Systems. Benjamin, New York Schenk, R. (1989). Ph.D. thesis, University of Mainz Shacklette, L. W., Maxfield, M., Gould, S., Wolf, J. F., Jow, T. R. and Baughman, R. H. (1987). Synth. Met. 18, 611 Shimada, K., Moshuk, G., Connor, H. D., Caluwe, P. and Szwarc, M. (1972). Chem. Phys. Lett. 14, 396 Shimada, K. and Szwarc, M. (1974). Chem. Phys. Lett. 28, 540 Shohoji, M. C. B. L., Herold, B. J. and Novais, H. M. (1987). J . Chem. SOC. Perkin Trans. 11, 1465 Siders, P. and Marcus, R. A. (1981). J . Am. Chem. SOC.103, 748 Smith, T. W., Kuder, J. E. and Wychick, D. (1976). J. Polym. Sci. 14, 243 Staley, S. W., Dustman, C. K., Facchine, K. L. and Linkowsky, G. E. (1985). J . Am. Chem. SOC. 107,4003 Sullivan, P. D. and Bolton, J. R. (1970). Adu. Magn. Reson. 4, 39 Szwarc, M. ( 1968). Carbanions, Living Polymers and Electron-Transfer Processes. Wiley, New York Szwarc, M. ( 1974) Ions and Ion-Pairs in Organic Reactions. Wiley-Interscience, New York Ulstrup, J. ( 1979). Charge- Transfer Processes in Condensed Media. Springer-Verlag, Berlin Wagner, M., Wohlfarth, W. and Miillen, K. (1988). Chimia 42, 377 Ward, R. L. and Weissman, S. I. (1957). J . Am. Chem. SOC.79, 2086 Wasielewski, M. R., Niemczyk, M. P., Svec, W. A. and Pewitt, E. B. (1985). J. Am. Chem. SOC. 107, 5562 Wasielewski, M. R. (1989). In Photoinduced Electron-Transfer (ed. M. A. Fox and M. Chanon), Chap. 1.4. Elsevier, Amsterdam Wegner, G. (1981). Angew. Chem. 93, 352; Int. Ed. Engl. 20, 361 Wegener, S. and Mullen, K. (1991). Chem. Ber. 124, 2101 Weitzel, H.-P. and Miillen, K. (1990). Makromol. Chem. 191, 2837 Weitzel, H.-P., Bohnen, A. and Miillen, K. (1990). Makromol. Chem. 191, 2837 Wertz, J. E. and Bolton, J. R. (1972) Electron Spin Resonance. McGraw-Hill, New York; reprinted by Chapman and Hall, London, 1986 Yao, H., Okada, T. and Mataga, N. (1989). J. Phys. Chem. 93, 7377 Ziilicke, L. (1985). Quantenchemie, Bd.2 p 49ff. VEB Deutscher Verlag der Wissenschaften, Berlin