33
Ind. Eng. Chem. Rud. Res. Dev. 1084, 23, 33-40
CATALYST SECTION
Pulsed-Laser-I nltiated Photocatalysis in the Liquid Phase Ke-Jlan F u , ~ Robert L. Whetten,$ and Edward R. Grant. Department of Chemkby, Baker L a ~ a t o t yCornell , Unlvershy, Ithaca, New York 14853
We illustrate the use of pulsed-laser irradiation to promote homogeneous organometallic catalysis under mild conditions in the liquid phase. We concentrate on the Iron carbonyl catalyzed isomerization of olefins. Kinetics data on the Fe(CO),/lpentene system establish the sequence of photolytic events leading from Fe(CO), starting material through Fe(CO),(pentene) catalyst precursor to the true photocatalyst, Fe(CO)&pentene). Measurement of the average lifetime of the photocatalyst with respect to reversion to its Inactive, precursor form, combined with a knowled e of the photocatalytic quantum yield establishes the turnover rate under our conditions, which is a fast lo4.'-s We also find that Fe(CO)5 photocatalytic quantum yields are temperature dependent with an apparent advation energy of 2.8 kcal m T ' . We present a mechanistic analysis of the isomerization cycle, which compares and contrasts the reactMty of group 8 coordlnathrely unsaturated organometallic intermediates with those of group 6.
I. Introduction Photochemical methods have assumed a role of steadily increasing importance in the understanding of the reactions of organometallic compounds (Geoffrey and Wrighton, 1979). This is particularly the case for catalytic systems where, as abundantly demonstrated by transformations ranging from isomerization (Whetten et al., 1982a,b; Wrighton et al., 1975; Shroeder and Wrighton, 1976), to hydrosilation (Schroeder and Wrighton, 1977; Sanner et al., 1979; Austin et al., 1978), to the water gas shift reaction (King et al., 1981; Weiller and Grant, 1984), light can readily replace thermal energy in preparing an organometallic compound for entry into a catalytic cycle (Moggi et al., 1981). An especially attractive advantage of photocatalysis is the opportunity it affords to induce many hundreds of transformations for each quantum of light absorbed, thus making the consideration of industrial photolytic materials processing a practical one. Over the past two years we have conducted a program of research in the application of lasers to organometallic photocatalysis. Our effort has introduced two important developments. (1)We have shown (Whetten et al., 1982b) that nanosecond, pulsed laser photolysis separates initiation from propagation of the photocatalytic cycle isolating the sequence of events in a catalyzed transformation for time-resolved scrutiny. (2) We have additionally shown that, following pulsed laser initiation, cyclic catalytic processes analogous to those found in solution proceed homogeneously in the gas phase (Whetten et al., 1982a). One objective of our work is to exploit the natural time resolution of pulsed-laser initiation in an effort to directly identify and characterize primary products of photolytic initiation events and intermediates in subsequent catalytic Permanent address: Institute of Physics, Chinese Academy of Sciences, Beijing, The People's Republic of China. National Science Foundation Predoctoral Fellow. 0196-4321/84/ 1223-0033$01.50/0
Table I. Periodic Chart of the Transition Metal Block Elements Showing d Orbital Occupancies ~
Ti Zr Hf
V Nb Ta
Cr Mo W
Mn Tc Re
Fe Ru Os
Co Rh Ir
Ni Pd Pt
Cu(I1) Ag(I1) Au(I1)
processes. Our work so far on the Fe(C0)5-photocatalyzed isomerization and hydrogenation of olefins (specifically normal pentenes) has provided an indication of what to expect. The present report details the results of experiments that have studied quantum yields for photocatalytic isomerization of pentenes in the presence of Fe (CO),as well as selected group 6 metal carbonyls. Saturation methods combined with a relaxation kinetic analysis of the laser repetition rate dependence of catalytic activity establish turnover rates in the iron system as high as lo4 s-l, comparable with some of the fastest enzyme systems known. Measured rates are even faster at elevated temperature; the activation energy for the rate-determining step in the iron carbonyl/olefm isomerization cycle is found to be 2.8 kcal/mol. Comparison of this behavior with the less than unit quantum efficiencies always found for group 6 systems helps to establish a plausible sequence of steps for the laser-initiated organometallic mediated transformations which are photocatalytic. The following three sections provide background information illuminating the principles of photolytically induced pre-catalytic coordinative unsaturation in organometallic complexes. Section I11 contains an account of work to date in this program. It begins with a brief description of experimental method, followed by a repeating sequence of results and discussion that support the general observations mentioned above. 11. Background A. Coordinative Unsaturation. We are concerned exclusively with the properties of complexes of the tran0 1984 American Chemical Society
34
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984
Table 11. Stable Coordination Numbers for First Row Transition Metal Carbonyls Predicted by the 18-ElectronRule
Scheme I
- -
M-L
r*- a n t i bonding
M -L
.Ir-bonding
___.
_I____
complex Cr( CO) 6 Mn,(CO),o Fe(CO), Co( CO), Ni(CO),
metal d ligand electrons electrons (per contributed metal (per metal center) center)a total db ti d8 d9
cl'
12 11 10 9 8
18 18 18 18 18
a In bimetallic complexes the companion metal donates one electron through a metal-metal single bond.
sition metal elements, and primarily with those on the right of the d-block, from d6 to dO ' (Table I). The number of electrons in the metal-based d orbitals proves to be very important in correlating the structure and reactivity of transition metal complexes. We define organotransition metal complexes as transition metal complexes with organic ligands, of which the simplest examples are the high symmetry metal carbonyls. The basis of the stability and structure of organometallic complexes is now fairly well understood in terms of molecular orbital/ligand field theories. Perhaps the two most remarkable chemical properties of these complexes are their great utility as precursors to catalysts for the transformation of organic substrates and their extremely high photochemical reactivity (Geoffrey and Wrighton, 1979). A number of simple and widely useful rules concerning structure and reactivity of organometallic complexes have been deduced from experimental and theoretical studies. Of these rules, the 18-electron rule is certainly the most general and well-known (Cotton and Wilkinson, 1980) (cf. Table 11). Simply put, it states that kinetically stable organotransition metal complexes have 18 valence shell electrons. More generally, we can expect that 18-electron complexes will be more stable than the corresponding 16-electron or 20-electron complexes obtained by adding or subtracting one two-electron donating ligand. Conversely, the rule predicts that complexes with other than 18 valence electrons will be very reactive. As with any rule, there are exceptions, but among the d6-d10complexes with which we will be concerned, the rule has been remarkably successful and is essential to a discussion of catalysis. B. Catalytic Reactivity. Consider the metal carbonyls given as examples in Table 11. If one of these complexes loses a ligand (a process typically requiring 40-70 kcal/mol) through heating or absorption of light, it will have an open coordination site. This state is called coordinative unsaturation, and, as indicated before, the reduced electron count can be associated with high reactivity. This is the supposed basis of catalysis by organometallic complexes. Removal of one ligand opens a reactive site for the binding of substrates, for example, alkenes, viz.
Cr(CO),
+ alkene
-
Cr(CO),(alkene)
(2) The concept of a coordinatively unsaturated site for binding and transformation of substrates is entirely analogous to that of the active sites on catalytic metal surfaces or to the concept of the active site in enzyme catalysis, where a certain conformation of the protein molecule facilitates biochemical transformation (Bender, 1971).
t. - I t + --fc
u-bonding
U Oh
Metal -Ligand
Field
For effective catalysis the active site is hypothesized to have three essential properties: (1) it must be able to rapidly and specifically bind the substrate (unsaturation concept); (2) it must be able to perturb the electronic structure of the substrate in such a way as to make an ordinarily difficult reaction pathway energetically accessible; and (3) it must be able to rapidly release the transformed substrate. The energetic requirements are subtle and it has been observed that slight changes in the ligand field can have a pronounced effect on catalytic activity (Swartz and Clark, 1980). Thus here we face at least two difficult questions; e.g., what are the natures of energetic effects that control the dynamics of a catalytic process? How can these enertgetic effects be predicted and controlled? C. The Initiation of Catalysis/Photocatalysis. In order to study these catalytic transformations we require a highly resolved means of generating the active catalyst from its precursor. Simply heating a solution of precursor and substrate is the method generally used, but in recent years the method of starting with mildly stable coordinatively unsaturated complexes and mixing in substrate has gained favor. Neither of these methods is particularly suitable for dynamical studies (Hammes 1978 Eigen 1968). Fortunately, there is a viable third method. Consider the ligand field electronic structure of a transition metal complex, as diagrammed in Scheme I. The ligand field splitting of the metal-based d orbitals produces metal-ligand *-bonding orbitals as highest occupied orbitals and metal-ligand a*-antibonding orbitals as lowest unoccupied orbitals. We can thus conclude that optical excitation of an electron (?r a*) will greatly weaken metal-ligand bonding, leading perhaps to direct dissociation. Alternatively, the absorbed photon energy may be rapidly converted to vibrational excitation which upon randomization would ultimately lead to metal-ligand bond dissociation from the ground or near-ground electronic state. Experimentally it has been found that excitation to low-lying ligand field states leads to rapid dissociation with quantum yields near unity (Wrighton et al., 1973). This, of course, creates coordinative unsaturation and may generate the actual catalyst. Starting in the late 1960's, a series of systems has been found for which catalysis under mild conditions is either aided by light (photostoichiometric case) or for which a thermal catalyst is generated under mild conditions (photocatalytic case) (Moggi et al., 1981). This photolytic method of initiating catalysis has several clear advantages unique to photochemistry: (1)catalytic processes can be carried out at low temperatures, eliminating undesirable high-temperature side processes; (2) catalyst precursors can be excited specifically, saving en-
-
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 0
LOO.
R "\
Scheme I1
4
R\
35
Fe(C0I5 h v
\
fc;,t>
s t a b l e precursor
transformed substrate
substrate
substrate comDlexes 0
20
40
60
80
10
p Einsteins Absorbed
Figure 1. Time dependence of photocatalytic isomerization and change in optical density at 348 nm plotted as a function of peinateins absorbed under conditions of 2 X M Fe(C0)5in pure l-pentene, 300 K, 4.0-Hz laser-pulse repetition rate.
ergy; and (3) pulsed laser technology makes it possible to generate coordinatively unsaturated intermediates in a very short time interval, so that powerful relaxation specroscopic methods can be used to time- and frequencyresolve reaction processes. As indicated above, the role of light in aiding catalytic transformations under mild conditions may be either photostoichiometric or photocatalytic. In the first case light is needed for each complete cycle and the quantum yield-defined as substrate molecules transformed, divided by photons absorbed-cannot exceed unity. In the second case a photon creates the catalyst,which can undergo many cycles before reverting to inactive form. Figures below clarify this point. Both classes have examples, but the latter is definitely less common. 111. Experiments on Laser-Initiated Photocatalysis at Cornel1 A. Experimental Procedure. In our experimental work we have used pulsed radiation from a Lambda Physik 101 Excimer Laser, typically operating at the XeF 351 nm, XeCl 308 nm, or N2 337 nm line. Pulse lengths are aps, with repetition rates of 5-25 Hz and proximately average power of 5-100 mW. In a typical photocatalysis experiment, air-tight 3-mL Pyrex cuvettes are loaded with catalyst precursor, solvent, and substrate in a drybox under N2or argon gas. Radiation is focussed slightly by an iris-lens (40 cm focal length) combination through the cuvette and into a power meter. In all quantitative experiments pulse energies are less than 2 mJ/cm2, so that multiphoton events are negligible. Transformation to products is determined by gas chromatographic analysis and confirmed by infrared spectroscopy. B. Photocatalytic Isomerization by Fe(CO)@When a solution of Fe(C0)5and alkene is irradiated, it is found that the isomeric alkenes are rapidly formed. For example l-pentene
Fe(CO)&v,@%I
OC
cis-2-pentene + truns-2-pentane (3)
Using pulsed laser radiation, we find the same qualitative features as those seen by Wrighton (Wrighton et al., 1975; Schroeder and Wrighton, 1976). For example, under typical irradiation conditions an initial solution of neat l-pentene is equilibrated to the thermodynamic mixture of pentenes within 1 h. Figure 1 shows the time dependence of the isomerization reaction and illustrates several important features of the system.
a
L
\ 08-
\
c 0.6t-
'0
2
0'
c
2
c
u
-
'O
04-
0
C
0
0
\ 0
Q,
"A
0.2-
z
\ 0
@
? c
-0
:
-
0
Dvo
,
,
'. . @ ;
I
A
Gov-
v
oo 1
1
u E i n s t e i n s Absorbed Figure 2. Fractional conversion of neat l-pentene as a function of 1.5 x light absorbed for varying concentrationsof Fe(CO),: (0) M; ( 0 )7.4 x 10-3M;( 0 ) 5.5 x 10-3 M;(v)3.7 x 10-3 M; (A)1.9 x 10-3 M.
Perhaps the most remarkable of these features is the high quantum efficiency of the reaction. We have observed limited (minimum) quantum yields exceeding 700 for the pentene isomerization and the true quantum yield (defined as the number of substrates transformed divided by the number of photons absorbed by the precursor) undoubtedly exceeds 1000,when all inefficiencies, to be discussed further, are taken into account. Catalytic action terminates almost immediately after irradiation ends (Mitchener and Wrighton, 1981). We also observe turnover numbers per Fe complex (defined as substrates transformed divided by iron complexes originally present) of greater than lo4; i.e., a M solution of Fe(C0)5equilibrates neat (9.14 M) l-pentene, and added l-pentene is also quickly equilibrated. The same results are obtained when starting with either of the other isomers: solutions of cis- or trans-Zpentene are equilibrated to the known thermodynamic mixture, although these latter reactions are slower (they begin nearer the equilibrium mixture). From these facts it is concluded that the action of the laser pulse is to generate a catalyst of high activity at the temperature of the solution and that the catalyst reverts to a stable precursor, which, upon irradiation, becomes active again. This may be represented qualitatively by Scheme 11. Figure 2 shows the result of varying the concentration of Fe(CO),; activity is plotted in each case as a function of absorbed power. It is seen that catalytic activity depends only on the number of photons absorbed and not on the concentration of Fe(C0)5,except for a possible weak dependence during early stages of the reaction, which we have attributed to an induction period (to be discussed
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984
36
Scheme V
Scheme 111 Fe(C0I5
-%
Fe(CO),
Fe(CO), + Fe(CO)S
-
n
