Proc. Indian Acad. Sci. (Chem. Sci.), Vol. 107, No. 6, December 1995, pp. 673-689.
© Printed in India.
Heterosupramolecular chemistry and modulation of function in molecular devices
XAVIER MARGUERETTAZ, G A R E T H R E D M O N D , S NAGARAJA RAO and D O N A L D FITZMAURICE*
Department of Chemistry, University College Dublin, Dublin 4, Ireland
Abstract. An organised heterosupramolecular assembly is prepared by attach- ment of linked molecular components to previously organised condensed phase nano-components. The above approach yields an intrinsic substrate through which the resultant molecular device may be modulated. Further, the modulation state can be inferred from the bulk properties of the intrinsic substrate. Specifically, we report results for a heterosupermolecule consisting of a TiO 2 nanoerystallite attached to a linked viologen-quinone unit. The associated function is photo-induced vectorial electron flow. For the corresponding organised assembly we demonstrate the following: bandgap excitation of a semiconductor nanocrystallite results in light- induced vectorial electron flow; electrochemical modulation of the bulk properties of the intrinsic nanocrystalline semiconductor substrate modulates function; and the state of the intrinsic substrate provides information about the modulation state of the constituent units of the assembly.
Keywords, Supramolecular chemistry; molecular device; modulation; titanium dioxide; viologen; quinone.
1. Introduction
Supramolecular chemistry is distinguished from large molecule chemistry as follows (Lehn 1990; Balzani and Scandola 1991). First, the intrinsic properties of the molecular components are only slightly perturbed within the supermolecule; and second, the properties of the supermolecule are not a simple superposition of the properties of the molecular components. That is, there exists a supramolecular function. Another important aspect of supramolecular chemistry is organisation of supermolecules and consequent addressability of function.
Addressable supramolecular function offers the prospect of constructing molecular devices (Hopfield et a11988; Lehn 1988; Ashwell 1992). The constituent supramolecular entities may be photo-, electro-, iono-, magneto-, thermo-, mechano- or chemoactive, depending on whether they process photons, electrons or ions, respond to magnetic fields or to heat, undergo changes in mechanical properties or perform a chemical reaction.
However, despite many beautiful examples of organised supramolecular assemblies, progress toward realisation of practical molecular devices has, in fact, been slow.
The reasons for the limited progress to date are linked to the requirements that must be met by practical devices. First, that they function at a supramolecular level within an organised assembly; second, that the function of each constituent supermolecule in a given assembly can be modulated independently; third, that the state of modulation of
* For correspondence
673
674 Xavier Mar#uerettaz et al
each supermolecule in a given assembly can be determined individually, and finally, that the resulting device meet speed, reliability and cost specifications. In practice, problems encountered include the following (Hopfield et al 1988; Lehn 1990; Balzani and Scandola 1991; Ashwell 1992). First, constituent supramolecular units do not act independently; second, attempts to modulate supramolecular function typically in- volve incorporation of sub-units within a supermolecule that respond to external stimuli, an approach which to date has produced relatively small effects; third, substrates supporting an organised assembly that are capable of effecting stimulus of the above sub-units and providing information concerning their modulation state have proved difficult to identify; finally, techniques that permit individual addressing of the constituent supermolecules of the molecular device have not yet been realised although recent progress is encouraging (Bard et al 1993).
Regarding substrates supporting an organised assembly that are capable of effecting modulation of supramolecular function and providing information concerning the modulation state, recent work directed toward development of a heterosupramolecular chemistry appears to offer new opportunities (Marguerettaz et al 1994; Marguerettaz and Fitzmaurice 1994). Specifically, upon replacing a molecular component in a super- molecule by a condensed phase nanostructure, referred to as a heterocomponent, a heterosupermolecule is formed (Marguerettaz et al 1994a). By analogy, heterosupra- molecular chemistry is distinguished from the chemistry of molecules adsorbed at the heterocomponent surface as follows. First, the intrinsic properties of the heterocompo- nent and molecular components are only slightly perturbed within the heterosuper- molecule; and second, the properties of the heterosupermolecule are not a simple superposition of the properties of the heterocomponent and molecular components.
That is, there exists a heterosupramolecular function. As for supermolecules, addressa- bility is a consequence of organisation. Uniquely however, because one of the components of the assembled heterosupermolecules is a condensed phase nanostructure, organisation yields an intrinsic substrate capable of effecting stimulus of the above subunits and providing information concerning their modulation state. The realisation of this approach and implications for design of molecular devices are the subjects of this paper.
2. Experimental
2.1 Preparation of transparent nanocrystallite TiO 2 substrates
All semiconductor substrates consisted of a 4/an thick transparent layer of fused TiO 2 particles (12 rim. diameter) supported on fluorine doped SnO2 glass (0.5/an) supplied by Glastron. Preparation of these substrates has been described in detail elsewhere (O'Regan et al 1990; O'Regan and Graetzel 1991). Briefly, TiO2 was prepared by hydrolysis of titanium isopropoxide. The resulting dispersion was concentrated to about 160g/L and Carbowax 20000 (40% weight equivalent of TiO2) added yielding a white viscous liquid used to form a 4/~m thick layer on conducting glass. After drying in air for one hour, each substrate was fired at 450°C for 12 h and stored in a vacuum dessicator.
2.2 Synthesis of linked molecular components
The linked molecular components in scheme 1 were prepared as described in detail elsewhere and characterised by NMR and elemental analysis (Marguerettaz et al 1994a, 1995; Marguerettaz and Fitzmaurice 1994b).
Heterosupramolecular chemistry and modulation 675
I!
0
I!
O
SV
O
O O
SVQ o
0
sQ
S c h e m e 1.
2.3 Attachment of linked molecular components to transparent nanocrystalline TiO 2 substrates
To attach the linked molecular components in scheme I, to the surface of a transparent nanocrystaUine TiO2 substrate, an aqueous solution (typically 2 x 10- 3 mol dm - 3) of the required compound was prepared. The nanocrystalline substrate was placed in the above solution and adsorption monitored spectroscopically via the charge transfer absorption assigned to chelation of salicylic acid at TiO2 (Moser et al 1991). The derivatised substrate was washed carefully with distilled and deionised water prior to being stored in a darkened dessicator. For all experiments for which results are reported, previously unused samples were employed.
2.4 Potential dependent optical absorption spectroscopy
A derivatised TiO 2 substrate formed the working electrode (2 cm 2 surface area) of a closed three-electrode single compartment cell, the counter electrode being platinum and the reference electrode a saturated calomel electrode (SCE). Aqueous electrolyte solutions contained LiCIO 4 (0.2moldm -3) at pH 3-0 (added HC104).
Ethanolic electrolyte solutions contained LiCIO 4 (0-2 mol d m - a) acidified by addition of 10% by volume aqueous of HCIO4 (0.01 tool d m - 3) solution. Potential control was provided by a Thompson Electrochem Ministat potentiostat and a Hewlett-Packard 3310B Function Generator. The above cell was incorporated into the sample Compart- ment of a Hewlett-Packard 8452A diode array spectrometer. Difference absorbance spectra, recorded with respect to a background spectrum measured at ff00V, are plotted.
676 Xavier Marguerettaz et al 2.5 Cyclic voltammetry
All cyclic voltammograms were recorded under the following conditions: The working electrode was a transparent nanocrystalline TiO 2 substrate whose working area was defined by a hardened epoxy resin mask. The counter and reference electrodes were isolated platinum and SCE respectively. The aqueous electrolyte solution contained LiC10 4 (2 x 10-1moldm -3) at pH 3-0 (added HC104) and was deoxygenated by bubbling with argon for 30 min. All scans were recorded at a rate of 0.050V s-1 in 0.005 V steps between 0-200 V and - 0-800 V. The wait time between scans was 30 s. In each case a cyclic voltammogram (cv) was recorded for the substrate prior to derivatisa- tion. This cv was subtracted from that subsequently measured for the substrate following attachment of the linked molecular components.
2.6 Bandgap irradiation of transparent nanocrystalline TiO 2 substrates
To study the effect of bandgap irradiation on potential dependent absorption spectro- scopy measured for a given heterosupramolecular assembly, a derivatised substrate was incorporated as a working electrode in the electrochemical cell described above.
The ethanolic electrolyte solution employed contained LiC104 (0.2 mol din-3) acidi- fied by addition of 10% by volume of aqueous HC104 (0"01 tool dm -3) solution and was deoxygenated by bubbling with argon for 30 rain. The cell was placed in the sample compartment ofa Hewlett-Packard 8452A diode array spectrometer and the semicon- ductor substrate irradiated in situ from the conducting glass side using the pulsed output at 355 nm (10 Hz, stated pulse energy) of a Continuum-Surlite Nd:Yag laser.
Stated pulse energies are not corrected for reflection losses. The applied potential was as stated, otherwise open-circuit conditions applied.
To study the effect of bandgap irradiation on the cv measured for a given heterosup- ramolecular assembly, the derivatised substrate in question was removed from the electrochemical cell, washed carefully with distilled deionised water and placed in an ethanolic electrolyte solution containing LiCIO 4 (0-2 tool dm-3) acidified by addition of 10% by volume aqueous HC104 (0.01 moldm-3) solution and deoxygenated by bubbling with argon for 30 min. The derivatised substrate was irradiated from the conducting glass side using the pulsed output at 355 nm (10 Hz, stated pulse energy) of a Continuum-Sudite Nd:Yag laser. Stated pulse energies are not corrected for reflection losses. Following irradiation the derivatised substrate was again washed with distilled deionised water and replaced in the electrochemical cell used for cv measurements.
2.7 Real-time transient absorbance optical spectroscopy
Real-lime transient optical absorbance measurements were made using a spectrometer based on a previously described design (Fitzmaurice et a11993). Briefly, the continuous output of a Coherent At-ion laser (Innova 70-5) was used to pump a Coherent dye laser (Model 599-01 with Rhodamine 6G). The dye laser output was split (40/60%) into two beams. One beam (40%) was allowed to fall directly on one photo-diode of a dual diode detector. The second beam (60%) was passed through the sample before falling on the second photo-diode of the detector. The detectors used were United Detector Technol- ogy PIN-10D silicon photo-diodes protected against scattered 355 nm or 532 nm light by a Melles Griot OG-550 optical cut-on filter. The associated circuitry was configured to generate a d c voltage proportional to the difference in intensity of the two beams
Heterosupramolecular chemistry and modulation 677 incident on the dual-diode detector. This signal was amplified and digitised using a Le Croy 9410 oscilloscope. The sample was excited at right angles to the probe beam using the pulsed output (single-shot or 10Hz) at 355nm or 532nm of a Nd:YAG laser (Continuum-Surlite).
3. Results
~.1 Characterisation of heterosupramolecular assemblies
TiO 2 nanocrystallites when deposited on conducting glass and fired, yield a transpar- ent nanocrystalline semiconductor substrate in ohmic contact with the conducting support. Consequently, following incorporation as the working electrode of an elec- trochemical cell, potentiostatic control of the Fermi level within the substrate is possible (Rothenberger et a11992). At applied potentials more negative than that of the conduction band edge at the semiconductor-liquid electrolyte solution interface (SLI), referred to as the conduction band edge potential (V,a), electrons accumulate in the semiconductor substrate. By examining the potential dependence of an optical absorp- tion assigned to free conduction band electrons it is possible to determine Vc~. Further, Vcb exhibits the expected pH dependence, shifting to more negative potentials by (>060 V per pH unit and is given by
~ = - 0.40 - [0-060xpH](V, SCE) (1)
It has been shown that molecules such as salicylic acid are strongly chemisorbed at TiO 2 by chelating to surface Ti 4+ atoms (Frei et a11990; Moser et a11991). Therefore, this and related molecules are used to attach the previously linked molecular compo- nents to surface Ti 4+ sites of the constituent nanocrystallites of the semiconductor substrate. The concentration of such sites is estimated to b¢ about 5 x 101 acre-2 from the extinction of reduced viologen (Marguerettaz et al 1994a), see below. This value
E c - -0.58 V,
1302 Electrolyte Solution (_pH 3.0) Condu~on
Banu I le-
Ef - -0.80 V ~ 1
B
o E°~,-0.56V O ~ 2H+
E°'= - 0.22 V
E v - +2.62 V
Valence Band Scheme 2.
678 Xavier Marouerettaz et al
agrees well with the reported density ofTi* + states (Siripala and Tomkievicz 1982). The linked molecular components, for which results are reported are shown in scheme I.
The resulting organised heterosupramolecular assemblies are denoted H(TiO2)-S, H(TiO2)-SV, H(TiO2)-SVQ and H(TiO2)-SQ for substrates to which S, SV, SVQ and SQ respectively have been attached. Attachment is monitored spectroscopically via a visible charge transfer absorption assigned to chelation of salicylic acid at surface T P + atoms (Frei et al 1990; Moser et al 1991).
3.2 Potential-induced vectorial electron flow in H ( T i O 2 ) - S V Q
Recently, potential-induced vectorial electron flow in H(TiO2)-SVQ at applied potentials more negative than Vc~ has been reported, see scheme 2. These findings (Marguerettaz and Fitzmaurice 1994) are summarised in scheme 2. The first cathodic scan of a cv recorded at pH 3-0 for H(TiO2)-SVQ shows a wave at - 0.6 V due to l e - reduction of viologen and 2 e - / 2 H + viologen-mediated reduction of anthraquinone (Marguerettaz and Fitzmaurice 1994), see figure 1 a. The first anodic scan shows a wave at - 0.5 V, the result of l e - oxidation of viologen. Subsequent cathodic scans continue to show a wave at - 0.6 V due to reduction of viologen although no current due to viologen-mediated reduction of anthraquinone is observed. Subsequent anodic scans show a wave at - 0.5 V due to l e - oxidation of viologen and are similar to the initial anodic scan.
Analysis of the cvs in figure la indicates the integrated cathodic current due to direct reduction of viologen is less than half (about 40%) that due to viologen-mediated reduction of anthraquinone, i.e. close to the expected value of 33%. We conclude, therefore, that reduction of anthraquinone is greater than 90% viologen-mediated.
Following air bubbling for 20 min at the rest potential of + 0-20 V anthraquinone is regenerated and cvs similar to those initially observed are recorded. The voltammo- grams shown in figure la are typical of many such experiments and are consistent with previously reported cvs for polymeric viologens charge compensated by an- thraquinone (Hable et al 1989, 1993).
Also shown are cvs recorded for H(TiO2)-SQ under similar conditions, see figure lb.
We note, the formal potential for anthraquinone in aqueous solution at pH 3"0 is estimated to be - 0 - 2 V (Ksenzhek et al 1977). On the first cathodic scan the peak reduction current is observed at about - 0 . 6 V , consistent with a value of V~b of
- 0.58 V in aqueous solution at pH 3"0. On the first anodic scan partial reoxidation is observed and consequently the second cathodic scan shows a smaller peak current.
Subsequent scans do not differ appreciably. Of note is the fact that the integrated current is significantly smaller than that assigned to viologen-mediated reduction of anthraquinone. This suggests direct reduction of anthraquinone at the semiconductor electrode is, in the case of the model system under study, not appreciable. Following air bubbling for 20 min at a rest potential of + 0.20 V anthraquinone is regenerated and cvs similar to those initially observed are recorded. These observations represent further support for assignment of the cv in figure la.
Visible absorption spectra were measured between 0.00 V and - 0.80 V for H(TiO 2)- SVQ, see figure 2a. At applied potentials more negative than V~b a spectrum is measured which may be assigned to the summed spectra of reduced viologen (Kok et al 1965; Trudinger 1970; Wantanabe and Honda 1982) and reduced anthraquinone (Hayon et al 1972). However, as can be seen from figure 2b, the contribution by reduced anthraquinone to a spectrum measured at - 0 - 5 0 V is greater for spectra measured
Heterosupramolecular chemistry and modulation 679 a
b
,oo L
-'°° 1 / y Scan,
iii11 : , .
i-800 -600 -400 -200 0 200
Potential (mV, SCE)
100
<l:
::1 o -100' t.=
~'} -200'
t,,,.,.
rO -soo.
-400
Scan 2 ]
/
Scan 1
00 -600 -400 -200 0 200
Potential (mV, SCE)
Figure 1. (a) Cyclic voltammogram of a transparent nano,crystalline TiO2 substra- te derivatised by SVQ (adsorbed over a period of 30 min from a 1 x I 0 - 3 mol din- 3 aqueous solution) in a deaerated aqueous electrolyte solution containing LiCIO4 (0.2 mol dm-3) at pH 3-0 (added HCIO4). The background voltammogram meas- ured for the bare TiO 2 substrate has been subtracted. (b) Cyclic voltammogram of a transparent nanocrystalline TiO2 substrate derivatised by SQ (adsorbed over 30min from a 2 x 10-3moldm -3 aqueous solution) in a deaerated aqueous electrolyte solution containing LiCIO 4 (0"2 mol dm-a) at pH 3-0 (added HC104).
The background voltammogram measured for the bare TiO 2 substrate has been subtracted.
during the reverse potential sweep. This suggests that anthraquinone is reduced only at potentials sufficiently negative to reduce viologen, i.e. a viologen-mediated process. We note, that at 0.00 V the spectrum assigned to reduced anthraquinone persists for up to 5min and suggests long-lived charge-trapping. Oxidation of the reduced an- thraquinone is due to reaction with residual dissolved molecular oxygen and leads to the formation of peroxide (Callabrese et al 1983). The final spectrum is indistinguish- able from that measured initially at 0"00 V.
680 Xavier Marguerettaz et al a
b
(D O t- t,.- O .D
0.5
G) 0 p-
0.3' . 0
t , -
o u~
. o
,¢~ 0.1
-0.1 35q
//•
~ r o 5OR V (SCE) " -0.8O V (SCE) 0.40R V / - 0 . 6 0-0.50 0.40
Wavelength (nm)
750
0.2'
0.1
0.0 350
/(-0.50R - Viologen) V (SCE)
450 550
Wavelength (nm)
Figure 2. (a) Absorbance spectra at the indicated applied potentials of a transpa r- ent nanocrystalline TiO 2 substrate derivatised by SVQ (adsorbed over a period of 60rain from a 1 x 10-3moldm -a aqueous solution) in a deaerated aqueous electrolyte solution containing LiC104 (0.2 mol dm-3) at pH 3-0 (added HCIO~).
(b) Difference between absorbance spectra in (a) measured at - 0-40 V and - 0"50 V on changing the applied potential first in the cathodic and then in the anodic directions.
3.3 Light-induced vectorial electron flow in
H(TiO2)-SV Q
Based on the findings outlined above, it was expected that b a n d g a p irradiation of H ( T i O 2 ) - S V Q would result in vectorial electron flow, reduction of viologen and viologen-mediated electron transfer to anthraquinone. Further, it was expected that the resulting radical anion would add hydrogen and long-lived charge-trapping occur,
Heterosupramolecular chemistry and modulation 681
E e = -0.58 V
TiO 2 Electrolyte Solution (_pH 3.0) Conduction Band
l e -
Photon
/
59
~ 20-II v " ~ v
O E ° ' = - 0.56 V O ~, 2 H +
E°'= - 0.22 V
=
+2.62 V ~ 59 Ethanol
E v 1~)
Valence Band
Scheme 3.
see scheme 3. Outlined below are recently reported results consistent with these expectations (Marguerettaz et al 1994b).
H(TiO2)-SV Q was placed in a deaerated acidified ethanolic solution (pH 3.0) and irradiated from the conducting glass side at 355 nm for 60 s with the pulsed output of a frequency-tripled Nd:Yag laser. Following irradiation, the measured visible absorp- tion spectrum was principally that of reduced viologen (Kok et al 1965; Trudinger 1970; Wantanabe and Honda 1982) and reduced anthraquinone (Hayon et a11972) see figure 3a. An ethanolic solution was used to ensure efficient trapping ofphotogenerated holes and electron transfer to the viologen (Finklea 1988). The same experiment performed in aqueous electrolyte solution produced no detectable concentration of reduced viologen or reduced anthraquinone. After irradiation the decay of reduced viologen and anthraquinone was followed by recording spectra at 60 s intervals, see figure 3a. The absorbance assigned to reduced viologen decays completely in about 5 min. As before, we assume decay is by reaction with residual oxygen and electron transfer to vacant surface states of the semiconductor substrate. The spectrum assigned to reduced anthraquinone is then apparent. This decays in about 12 min, it is assumed, also by reaction with residual oxygen to form peroxide and by back-electron transfer to vacant surface states of the semiconductor substrate (Callabrese et al 1983). Spectra measured for H(TiO2)-SV under similar conditions are entirely consistent ~vith the above findings, see figure 3b. Briefly., for H(TiO2)-SV the spectrum observed following irradiation was that of reduced viologen and decayed over about 5 minutes.
Also examined were the kinetics of formation of reduced viologen by H (TiO2)-SVQ under pulsed irradiation at 355 nm by monitoring absorbance at 600 nm, see figure 4a.
It is known from real-time absorbance experiments that reoxidation of the viologen in H(TiO2)-SVQ is incomplete in the 0-I s interval between pulses. Therefore, successive pulses lead to accumulation of the reduced form of the viologen with maximum absorbance being observed after about 20 s. At long times the rate of formation of
682 Xavier Marguerettaz et al a
0 r - O3 . 0
0 O3
¢n
<
0.8'
0.6"
0.4"
0.2
-0.1 351
A
A
'/X
4~0 5~0 6~0 7~o
b Wavelength (nm)
O O r - J ~ O ¢/) ..Q <l:
0.8
o,1 A
0.4"
0.2"
"0.1
350
4~o 5~o 8~o 7~o
Wavelength (nm)
Figure 3. (a) Absorbance spectra measured at 60s intervals following pulsed irradiation for 60s with the frequency-tripled output (355 nm) of an Nd:Yag laser (10 Hz, 4.3 mJ pulse- ~) of a transparent nanocrystalline TiO2 substrate derivatised by SVQ (adsorbed over a period of 60min from a 1 x 10-3moldm -3 deaerated ethanolic electrolyte solution containing LiCIO4 (0.2 mol dm-3) and acidified by addition of aqueous HC10 4 (0-01moldm -3) (10% by vol.). Also shown is the baseline absorbance recorded prior to irradiation. (b) Absorbance spectra measured at 40 s intervals following pulsed irradiation for 60 s with the frequency-tripled output (355nm) of an Nd:Yag laser (10Hz, 4"3mJpulse -~) of a transparent nanocrystalline TiO; substrate derivatised by SV (adsorbed over 60min from a 2 x 10-3moldm -~ aqueous solution) in a deaerated ethanolic electrolyte sol- ution containing LiCIO4 (0.2 mol dm-3) acidified by addition of aqueous HCIO4 (0-01 moldm -~) (10% by vol.). Also shown is the baseline absorbance recorded prior to irradiation.
Heterosupramolecular chemistry and modulation 683 a
b
tl) o t-- ..o l b .
o t/) .Q <
0.2"
0.1' ~ "~°°~° ... ""
// k 2 -0.0044
// k 3 -4.0200
J k 4 -0.0001
y r -0.9944 r -0.9944
O0 i - _
-' 0 0 10 20 30 40 50 60
Time (s)
0.15.
~) o 0.10 c- JD
t.--
(/) o ..Q 0.05
0.00
//
k, .O.lOS,// k 2 -0.0044
._j ... r.-0.,935
0 0 10 20 30 40 50 60
Time (s)
Figure 4. (a) Plotted is the evolution of absorbance at 600 nm (solid line) for a transparent nanocrystalline TiO 2 substrate derivatised by SVQ during pulsed irradiation by the frequency-tripled output (355 nm) of an Nd:Yag laser (10 Hz, 4"3 mJ pulse- l ). Conditions are as in figure 3a. Also plotted is the predicted absorbance (dashed line) based on (2) to (5). (b) Plotted is the evolution of absorbance at 600 nm (solid line) for a transparent nanocrystalline T i O 2 substrate derivatised by SV during pulsed irradiation by the frequency-tripled output (355 nm) of an Nd :Yag laser (I0 Hz, 4.3 mJ pulse- 1 ). Conditions are as in figure 3b.
Also plotted is the predicted absorbance (dashed line) based on (6) and (7).
r e d u c e d viologen is slower t h a n the rate o f its d e c a y a n d a c c o u n t s for the o b s e r v e d a b s o r b a n c e decrease.
At i n t e r m e d i a t e times, a b s o r b a n c e g r o w t h for H ( T i O 2 ) - S V Q is sigmoidal, a l t h o u g h such b e h a v i o u r is n o t o b s e r v e d for H ( T i O 2 ) - S V , see figure 4b. T h i s b e h a v i o u r is
684 X a v i e r M a r g u e r e t t a z e t al
consistent with the existence of a process, other than reaction with dissolved oxygen and back-electron transfer to vacant surface states on the semiconductor substrate, leading to viologen reduction. If it is assumed that this process is electron transfer to the adjacent anthraquinone, the kinetic behaviour of H(TiO2)-SVQ will be described by the following system of differential equations:
d [MV 2 + ] / d t = - k 1 [MV 2 + ] + k 3 [MV +'] [AQ] + k 4 [MV +'] [AQH], (2) d [MV +']/dt= k~ ['MV 2 + ] - k 2 [MV +'] - k a [MV +'] [AQ]
- k4 [MV +'l [AQH], (3)
d [ A Q H ] / d t = k 3 [MV +'] [AQ] - k,, [MV +'] [AQH], (4)
d [AQH 2 ] / d t = k,, [MV +'] [AQH]. (5)
Values for kt, k2, k 3 and k 4 have been determined numerically by fitting the absorb- ance data at 600 nm for H (TiO2)-SVQ. These rate constants have been used to predict the measured absorbance at 600 nm and the agreement is excellent, see figure 4a.
A system of differential equations appropriate to H(TiOz)-SV and consistent with the system of differential equations used for H(TiO2)-SV Q, is given below.
d [MV 2 +]/dt = - k l [ MV2 + ], (6)
d [MV +']/dt = k x [MV 2 + ] - k 2 [MV +']. (7) These equations may be solved analytically and effective rate constants, k x and k 2 for formation and decay of reduced viologen respectively, determined by a fit to the absorbance data measured at 600 nm in figure 4b. These rate constants have been used to predict the measured absorbance at 600 nm and the agreement is excellent, see figure 4b.
These kinetic studies and accompanying analyses are consistent with light-induced vectorial electron flow to anthraquinone that is viologen-mediated. We note, an implication of these findings is that electron transfer from viologen to anthraquinone is slow and consistent with the findings of previous workers (Brun e t al 1990, 1992).
Further evidence supporting the assertion that light-induced vectorial electron flow to anthraquinone is viologen-mediated is provided by cvs measured for H(TiO2)-SVQ.
Specifically, cvs were measured following 10 minutes irradiation with the frequency tripled output (355nm) of a Nd:Yag laser in a deaerated and acidified ethanolic solution (pH 3"0}, see figure 5a. No current is observed during the first cathodic scan that may be assigned to viologen-mediated reduction of anthraquinone. It is concluded that light-induced vectorial electron flow has previously reduced anthraquinone. This is consistent with the fact that following irradiation under similar conditions the measured absorbance spectrum is that of reduced viologen and reduced an- thraquinone, see figure 3a. Support for this interpretation comes from the fact that in
s i t u regeneration of the anthraquinone, by air bubbling for 20 min, and argon bubbling
for 30 min, results in the cvs shown in figure 5b where during the next cathodic scan current that may be assigned to viologen-mediated reduction of quinone is observed.
It is concluded, based on the results presented above, that bandgap excitation does lead to formation of reduced viologen and reduced anthraquinone. Further, it is concluded that, for the potential induced process, reduction of anthraquinone is viologen mediated.
Heterosupramolecular chemistry and modulation 685 a
,°° t
-100' t'- Q)
~_ -200"
° t
-300 .,400/- 8 0 0
can 2 Scan 1
"600 -400 -200 0 200 Potential (mV, SCE) b
A
<
=L
t-- 100'
-100"
-200"
-300"
S c a n ,
-400
-800 -sbo -4bo -2bo 6 200
Potential (mV, SCE)
Figure 5. (a) Cyclic voltammogram of a nanocrystalline TiO 2 substrate de- rivatised by SVQ (adsorbed over 30min from a 1 x 10-3moldm -3 aqueous solution) in a deaerated aqueous electrolyte solution containing LiCIO4 (0.2 mol d m - 3) at pH 3-0 (added HCIO4) following 10 min prior irradiation by the frequency-tripled output (355 nm) of an Nd:Yag laser (10 Hz, 2.0mJpulse-t) in a deaerated ethanolic electrolyte solution containing LiCIO4 (0"2 mol dm-3) and acidified by addition of 10% by volume of aqueous HC10 4 (0.01 mol dm-3). The background voltammogram measured for the bare TiO2 substrate has been sub- tracted. (b) Cyclic voltammogram of substrate in (a) after bubbling aqueous electrolyte solution with air for 20 min and with argon for 30 min. The background voltammogram measured for the bare TiO 2 substrate has been subtracted.
3.4 Modulation of light-induced vectorial electron flow in H(TiO2)-SV Q
As discussed below, modulation of light-induced vectorial electron flow in H(TiO2)- SVQ m a y be effected by potentiostatic control of the bulk properties of the
686 Xavier Marouerettaz et al
intrinsic semiconductor substrate (Marguerettaz et al 1995), that is, modulation of the bulk properties of the substrate formed by the fused condensed phase nano-compo- nents of H(TiOz)-SV Q. Further, the modulation state of the heterosupermolecules
a,
(D 0 ¢- ..Q 0 . O U~
<
0.3'
0.2'
3 i
2 0,1'
0.0 - I
- 1 0 0 10 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0
b
¢- 0
0 ,.Q
<
0.05
0.04'
0.03'
0.02"
0.01
Time (s)
\3
0.00
-0.0001 0.0001 0.0003 0.0005
Time (s)
Figure 6. (a) Absorbance by a nanocrystalline TiO 2 substrate at 600nm de- rivatised by SVQ (adsorbed during 30rain from 1 x 1 0 - 3 m o l d m -3 aqueous solution) during irradiation by the frequency tripled output (355 nm) of an Nd:Yag laser (10 Hz, 5.5 mJ pulse - 1 ) in a deaerated ethanolic electrolyte solution containing LiCIO 4 (0-2moldm -a) and acidified by addition of aqueous HC10 4 (0"01 m o l d m -a) (10% by vol.). The applied potentials were (I) 0.00V, (2) open circuit and (3) open circuit following application of a cathodic step from 0.00 V to - 0 - 6 5 V (60 s), returning to 0.00 V (15 s) prior to irradiation. (b) Transient absorb- ance at 600nm for sample in (a). The applied potentials were (1) 0.00V, (2)open circuit and (3) open circuit following application of a cathodic step from 0.00 V to
- 0 - 6 5 V (60 s), returning to 0.00 V (15 s) prior to irradiation.
Heterosupramolecular chemistry and modulation 687 constituting the assembly may be inferred from the bulk properties of the above intrinsic substrate.
H(TiO2)-SVQ, forming the working electrode of an electrochemical cell and con- taining acidified ethanolic electrolyte solution (pH 3"0), was irradiated from the conducting glass side at 355 nm for 90s at the indicated potential with the pulsed output of a frequency-tripled Nd: Yag laser. The absorbance measured at 600 nm and assigned principally to reduced viologen is plotted in each case, see figure 6a. At the open circuit potential, behaviour is consistent with that observed under similar conditions, with sigmoidal growth of this transient indicating viologen-mediated reduction of anthraquinone, see figure 4a. At an applied potential of 0.00 V there is a small initial increase in absorbance by reduced viologen but the steady-state value is significantly less than that measured at the open circuit potential. This continues to be the case for the first number of pulses, however, continued irradiation results in accumulation of conduction band electrons and a shift in the Fermi level to more negative potentials. Eventually, interfacial electron transfer is as efficient as at the open circuit potential. However, since the rate of decay of reduced viologen is significantly faster, th e reduced viologen generated by a single pulse is almost entirely reoxidised in the interval between successive pulses. Following application of a cathodic potential step between 0.00V and -0"65 V (60s) and returning to 0.00 V (15 s) formation of reduced viologen at open circuit.potential is significantly faster and no longer sig- moidal. The faster formation of reduced viologen is due to trap filling and the absence of the sigmoidal feature is consistent with protonation of anthraquinone.
Similar experiments were performed at ns time resolution under single shot condi- tions. The transient shown in figure 6b, measured at the open circuit potential, is assigned to trapped electrons and to reduced viologen. The transient measured under similar conditions at 0.00 V is assigned principally to trapped electrons, residual
Ti%
Conduction Band
E c = -0.58 V
Ef - 0.00 V
Photon /
Electrolyte
Solution ~H 3.0)
E v - +2.62 V
0
I I°'" o v
7~ ~ Ethanol
Valence Band
E °: - 0.22 V
Scheme 4.
688 Xavier Marguerettaz et al
E c - -0.58 V
TiO 2 Electrolyte Solution (~oH 3.0)
Conduction Ban6J le-
'°- '3 J
a I
o E ° " - 0.56 V OH
E°'= - 0.22 V
Photon
m n
Valence Band Scheme 5.
absorbance assigned to reduced viologen is also observed. This is confirmed by transient spectra recorded under open circuit condition, at 0.00V and at 0.40V.
However, after about 100 pulses, charge accumulation leads to formation of reduced viologen. In short, at the open circuit potential light-induced vectorial electron flow is observed as in scheme 3. At an applied potential of 0-00 V light-induced vectorial electron flow is suppressed as in scheme 4. Following application of a negative potential step anthraquinone is reduced and light-induced vectorial electron flow is only to the viologen as in scheme 5. Therefore depending on the potential applied to the intrinsic substrate, constituted from the heterocomponents of H (TiOe)-SVQ, the function of the constituent heterosupermolecules may be modulated between three states. Further, the modulation state of the constituent heterosupermolecules of H(TiO2)-SVQ is known if the potential applied to the intrinsic substrate is known.
4. D i s c u s s i o n
4.1 Some oeneral comments
This paper describes preparation of a heterosupramolecular assembly, H(TiO2)-SVQ, for which we demonstrate the following. Bandgap excitation of a semiconductor nanocrystallite results in light-induced vectorial electron flow; electrochemical modu- lation of the bulk properties of the nanocrystalline semiconductor substrate modulates heterosupramolecular function; and the electrochemical state of the semiconductor substrate provides information about the modulation state of the constituent hetero- supermolecules.
More generally, the constituent heterosupermolecules of the proposed molecular device are assembled in a manner that determines their default function. As the constituent condensed phase nano-components are previously organised, the resulting heterosupramolecular assembly possesses an intrinsic substrate through which the
Heterosupramolecular chemistry and modulation 689 supramolecular function can be effectively modulated. Further, knowledge of the bulk properties of this intrinsic substrate provides information concerning the modulation state of the constituent heterosupermolecules.
Most generally, heterosupramolecular chemistry may be seen to be at the interface between condensed phase and supramolecular chemistry. Specifically, heterosupra- molecular chemistry offers the prospect of combining the function of supermolecules with the architecture of condensed phase assemblies. It is in this context that heterosup- ramolecular chemistry is of particular relevance with respect to artificial photosyn- thetic systems. It is likely that such systems will have to emulate the functional and architectural complexity of natural systems in order to achieve practical efficiencies.
Acknowledgements
This work was supported by a grant from the Commission of the European Union under the Joule II programme (Contract JOU2-CT93-0356).
References
Ashwell G 1992 Molecular electronics (New York: Wiley)
Balzani V and Scandola F 1991 Supramolecular photochemistry (New York: Ellis Horwood) ch. 2 Bard A et al 1993 J. Phys. Chem. 97 7147
Brun A M, Hubig S M, Rodgers M A and Wade W H 1990 J. Phys. Chem. 94 3869 Brun A M, Hubig S M, Rodgers M A and Wade W H 1992 J. Phys. Chem. 96 710 Callabrese G C, Buchanan R M and Wrighton M S 1983 J. Am. Chem. Soc. 105 5594 Finklea H 1988 Semiconductor electrodes (New York: Elsevier) ch. 2
Fitzmaurice D, Eschle M, Frei H and Moser J 1993 J. Phys. Chem. 97 3806 Frei H, Fitzmaurice D and Graetzel M 1990 Langmuir 6 198
Hable C T, Crooks R M, Wrighton M S 1989 J. Phys. Chem. 93 1190
Hable C T, Crooks R M, Valentine J R, Giasson R and Wrighton M S 1993 J. Phys. Chem. 97 6O6O
Hayon E, Ibata T, Lichtin N N and Simic M 1972 J. Phys. Chem. 76 Hopfield J, Nelson-Onuchic J and Beretan D 1988 Science 241 817 Kok B, Rurainski H J and Owens O 1965 Biochem. Biophys. Acta 109 347
Ksenzhek O S, Petrova S A, Oleinik S V, Kolodyazhnyi M V and Moskovskii V Z 1977 Elektrokhimiya 13 182
Lehn J M 1988 Angew. Chem., Int. Ed. Eng. 27 89 Lehn J M 1990 Angew Chem., Int. Ed. Eng. 29 1304
Marguerettaz X and Fitzmaurice D 1994 J. Am. Chem. Soc. 116 5017
Marguerettaz X, O'Neill R and Fitzmaurice D 1994a J. Am. Chem. Soc. 116 2629
Marguerettaz X, Rag S N, Redmond G and Fitzmaurice D 1994b Proceedings of the 13th International Symposium on Optical Maierials Technology for Energy EJJiciency and Solar Energy Conversion (Washington, DC: SPIE-Int. Soc. Opt. Eng.) vol. 2255, p. 793
Marguerettaz X, Rag S N, Redmond G and Fitzmaurice D 1995 Chemistry (in press) Moser J, Punchihewa S, Infelta P and Graetzel M 1991 Langmuir 7 3012
O'Regan B, Moser J, Anderson M and Graetzel M 1990 J. Phys. Chem. 94 8720 O'Regan B and Graetzel M 1991 Nature (London) 353 737
Rothenberger G, Fitzmaurice D and Graetzel M 1992 J. Phys. Chem. 96 5983 Siripala W and Tomkievicz M 1982 J. Electrochem. Soc. 129 1240
Trudinger P A 1970 Anal. Biochem. 36 222
Wantanabe T and Honda K 1982 J. Phys. Chem. 86 2617
