Electrochemiluminescence in low ionic strength solutions of 1,2-dimethoxyethane

JOU~.

OF

Journal of Electroanalytical Chemistry 396 (1995) 85-95

ELSEVIER

Electrochemiluminescence in low ionic strength solutions of 1,2-dimethoxyethane 1 Karolyn M. Maness, R. Mark Wightman * Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3290, USA Received 19 December 1994

Abstract

Steady-state electrochemiluminescence (ECL) arising from ion annihilation reactions was examined at a double-band microelectrode in 1,2-dimethoxyethane (DME) as a function of electrolyte concentration for the following reactions: the 9,10-diphenylanthracene (DPA) radical cation with the DPA, benzil and 2-naphthyl phenyl ketone (NPK) radical anions, and the rubrene radical cation with the rubrene radical anion. The ECL efficiency of all systems was found to increase as the added electrolyte concentration was decreased from 100 to 0.1 mM. These increases were found to correlate with changes in A E~/2 attributed to ion-pairing of the radical ions with their electrolyte counter-ions. For the energy-sufficient ECL reaction of DPA in DME, the increase in the Gibbs energy available for reaction as the electrolyte concentration was lowered produced an ECL efficiency approaching the theoretical limit of 25%. Keywords: Concentration effects; Electrochemiluminescence; 1,2-Dimethoxyethane

1. I n t r o d u c t i o n

Electrochemiluminescence (ECL) is the production of light as a result of highly energetic competing electron transfer reactions between electrogenerated radical anions and cations. This technique has been studied widely [1,2] and has found applications as a means of detection for liquid chromatography, polymerase chain reactions and immunoassays [3-10]. The following simplified ECL reaction scheme can be written for the generalized parent species A: A

,

A

A +"

k~ A +"

+

A-"

A-" ,

kkd

(1)

k+k_ d

The efficiency ~bECL of ECL production is a function of the relative rates of the above electron transfer reactions and can be expressed as

sk's 4'ECL = k,s + 3k,t + k'g

VA'+A

where ~bf is the efficiency of photon emission. The factor of 3 which appears before k't in Eq. (2) is due to the statistical nature of spin recombination, i.e. each radical ion recombination produces 25% singlets and 75% triplets. Marcus theory predicts that the rate of electron transfer will be a function of both the Gibbs energy AGet available and the reorganizational energy A required for a reaction [11]. Both AGet and A are in turn strongly dependent on the chosen solution conditions. Thus, through the appropri-

¢

A+A ~

k'

SA*+A

kg

SA*

where k's, k'l and k'g are the bimolecular electron transfer rate constants for production of the emitting excited singlet state SA*, the nonemitting excited triplet state "rA and the ground state A respectively. These bimolecular rate constants are related to the diffusion-controlled rate constants k d and k_ d for formation and dissociation of the encounter complex, and the unimolecular electron transfer rate constants k,, k t and kg as follows:

A + hv

Dedicated to Professors K. Honda, H. Matsuda and R. Tamamushi on the occasion of their 70th birthdays. * Corresponding author. 0022-0728/95/$09.50 © 1995 Elsevier Science S.A. All rights reserved SSDI 0 0 2 2 - 0 7 2 8 ( 9 5 ) 0 3 9 2 6 - 0

(2)

86

K.M. Maness. R.M. Wightman/ Journal of ElectroanalyticalChemistry396 (1995) 85-95

ate selection of ECL reaction conditions, it is possible to influence the relative electron transfer rates for the competing reactions and hence the ECL efficiency. Indeed, in past studies it has been shown that the efficiency ~bECL of ECL production can be improved by lowering the solvent dielectric constant [12-14] or decreasing the supporting electrolyte concentration [ 15-17]. In this study the steady-state ECL of a number of systems is examined in the low dielectric constant solvent dimethoxyethane as a function of supporting electrolyte concentration. By examining the ECL under conditions of low dielectric and ionic strength it should be possible to achieve very high ECL reaction efficiencies, a desirable aim for analytical applications of ECL. In addition, through the systematic variation of electrolyte concentration, it is also possible to gain insight into the ECL reaction mechanism for a given system. The production of steady-state ECL is accomplished at a double-band microelectrode in a constant-potential mode with one band serving as the anode and the other as the cathode. This electrode geometry has previously been shown to be very efficient at ECL production [18] and is relatively insensitive to problems with ohmic distortion [19], an important consideration for the production and study of ECL under these highly resistive conditions.

gap width, 4.5-5.3 I~m; length, 0.3-0.33 cm. The electrodes were polished with 1 I~m diamond paste and rinsed with water and acetone before use. Electrodes were affixed to glass standard-tapered joints (male), which allowed reproducible placement in the electrochemical cell. The cell was made from a glass joint (female) in which a platinum wire auxiliary electrode and a silver wire quasireference electrode were sealed. A quartz window was sealed on the end of the cell approximately 1 mm from the end of the electrode. 2.3. Electrochemical measurements

2. Experimental

The potential difference AEI/2 between the half-wave potentials of the reduction and oxidation reactions was determined using a 5 ixm radius Pt disk electrode (50 mV s -~). The half-wave potentials were obtained from the maxima of the differentiated cyclic voltammograms. For solutions containing less than 10 mM supporting electrolyte, AE~/2 determinations were made while maintaining a constant 10:1 ratio of electrolyte to analyte. Cyclic voltammograms at the microdisk electrodes were obtained using an EI-400 potentiostat (Ensman Instrumentation, Bloomington, IN), while a bipotentiostat of conventional design was used for the experiments with the microband arrays. Current and potential measurements were recorded on an x - y recorder or digital oscilloscope (Nicolet 310).

2.1. Chemicals

2.4. ECL production

1,2-Dimethoxyethane (DME) was saturated with N 2 and passed through a column of activated alumina to remove residual water prior to solution preparation. 9,10Diphenylanthracene (DPA) and benzil were recrystallized twice from absolute ethanol. Rubrene was recrystallized from a mixture of toluene (TOL) and UV-grade acetonitrile (ACN), both from Burdick and Jackson, Muskegan, MI. 2-Naphthyl phenyl ketone (NPK) was used as received. Tetrabutylammonium hexafluorophosphate (TBAPF6) was recrystallized twice from 95% ethanol. Solid chemicals were dried under reduced pressure at 60°C and stored in a desiccator before use. Except where noted, all chemicals were purchased from Aldrich (Milwaukee, WI) and were reagent grade or better. All sample preparation and experiments were performed in a nitrogen-containing glove-box.

Steady-state ECL at double-band microelectrodes was achieved with one band serving as the anode and the other as the cathode [18]. The applied potential for each band was chosen to be 250 mV past the Et/2 of the redox couple of interest to ensure generation of radical ions at a rate determined by diffusion-controlled mass transport.

2.2. Electrodes and electrochemical cell A platinum double-band electrode was constructed as described previously [18,20]. The dimensions of the electrodes used for all experiments were as follows (as measured from the current magnitudes during steady-state collector generator experiments [20] as well as by examination with an optical microscope): band width, 5-8 la,m;

2.5. Photometric equipment Electrochemiluminescence was measured using a Hamamatsu R928 photomultiplier tube (Bridgewater, NJ) operated at - 6 0 0 V. The electrochemical cell could be reproducibly positioned in front of the photomultiplier tube (PMT) which was mounted on the side of a light-tight steel box. The light-tight box and PMT were contained inside a glove box, and all electrical connections were made through standard BNC and MHV connections in the wall of the dry box. The photomultiplier current was measured using a Keithley model 427 current amplifier (Cleveland, OH) and recorded on a strip chart recorder or digital oscilloscope. The ECL coulometric efficiency was taken as the steady-state photomultiplier current 1 divided by the steady-state anode current. In the case of transient signals, the ratio was determined at the time of maximum intensity. To convert to an estimated ECL efficiency thECL the data

K.M. Maness, R.M. Wightman/ Journal of Electroanalytical Chemistry 396 (1995) 85-95

were compared with that measured for Ru(bpy)~ + in acetonitrile (emission maximum at 610 nm). The experimental efficiency of this system is approx. 0.05 at 25°C in acetonitrile [21], a value that is often taken as a standard [22]. Intensity measurements were corrected for differences in PMT sensitivity at the wavelength of maximum emission with the use of the manufacturer's specifications. Corrections were also made for the difference in refractive index between the standard ACN solvent and DME [23]. Because of the uncertainties in these corrections, the estimated error in the reported ~bEcL values is approximately 10%. 2.6. Spectral measurements

E u 0.15

TE o.lo

25

"7-

6"1

-5 E

~. 0.05 20

fN

E

0,004

0

2.7. Conductivity measurements The conductivity of solutions of TBAPF6 in DME was determined from ac impedance measurements using a Solartron-Schlumberger model 1250 frequency response analyzer and a model 1186 electrochemical interface (Farnborough, Hampshire, UK). A sine wave (100 mV peak to peak, centered at 0 V) was applied to a standard conductivity cell with the output taking the form of a real and imaginary impedance. The frequency of the sine wave was decreased until the imaginary impedance approached 0 fl. The real impedance at this frequency was taken as the solution resistance. The cell was standardized using an aqueous 0.1 M solution of KCI (molar conductivity A m of 129 S cm 2 mol- 1 at 298°C) [24] and found to have a cell constant C of 0.387 cm -l where C =/~K. Here R / I I is the measured solution resistance and x / l - l - l cm-1 is the known conductivity of the KC1 solution.

3. Results

3.1. Conductivity measurements In order to study ECL as a function of ionic strength, it was necessary to determine the concentration of free dissociated supporting electrolyte. Conductivity measurements of TBAPF6 in DME were undertaken in order to determine its degree of dissociation in this low dielectric constant solvent. The molar conductivity was obtained for a series of solutions of TBAPF6 in DME as a function of concentration. The value of the molar conductivity at infinite dilution, A 0 was estimated from the y intercept of a plot of 1 / A m vs. c A m (Fig. 1) according to the Oswald dilution law for weak electrolytes [25]

1/Ao= 1/Ao + cam/[ Ko( A0)

0.007

C Am// S crn "1 15

<

E 10

]

k

0.01

ECL spectra were obtained by positioning the electrochemical cell in front of a monochromator (2 nm bandpass, model H-20, JY Optical Systems, Metuchen, NJ) which was connected to the above PMT.

87

0,02

cl/2/

1

0.03

tool 1/21-1/2

Fig. 1. Plot of molar conductivity vs. (concentration) 1/2 for TBAPF6 in DME showing the strong association of this electrolyte in DME. The range of sample concentrations varied from 1 to 0.04 raM. The inset shows molar conductivity data treated according to the Oswald dilution law for weak electrolytes for determination of Ao:A o =274-3 S cm 2 mol- t.

where A m represents the molar conductivity obtained at electrolyte concentration c and K D is the dissociation constant for the electrolyte ion-pairing equilibrium TBAPF6 ~ TBA++ PF6-. The A 0 obtained in this manner (27 + 3 S cm 2 mol- l) was used to determine the degree of ionization a = A m / A o at each concentration. This quantity is related to KD:

Ko= [(3,+3,-)/3,TBApF0][ca2/(1where 3,+13,- and 3,rBAPF~ are the activities of [TBA+], [PF6] and [TBAPF6] respectively. The uncharged TBAPF6 will have approximately unit activity. The activities of the ions can be assumed to be equal and be replaced by 3, 2. Rearrangement then gives l o g [ c a V ( 1 - a ) ] = l o g K o - 2 log 3,

(3)

Through Debye-Hiickel theory, the last term on the righthand side of Eq. (3) is proportional to /z 1/2 for solutions of low ionic strength /x. In addition, for solutions of monovalent ions like those considered here, p. = ca. Thus an estimate of log K D can be obtained from the y intercept of a plot of l o g [ c a 2 / ( l - a)] vs. ( c a ) 1/2. For TBAPF6 in DME such a plot yields log K D = - 4 . 1 0 + 0.03 and K D = (8.0 + 0.6) × 10 -5. Such a high degree of association in DME is in direct contrast with the full dissociation of this electrolyte in ACN [24]. 3.2. ECL of DPA as a function of electrolyte concentration The ECL efficiency of DPA in DME was examined as a function of both the free electrolyte concentration (as determined from the K D of TBAPF6 in DME) and the

K.M. Maness, R.M. Wightman/ Journalof ElectroanalyticalChemistry396 (1995)85-95

88

(A)

®

(B)

(C)

;;

0.30

30

c 0.25

25

,'2

20

400

500 Wovelength/nm

0.20

qbEcL

At

15

-I qbECL

0.15 lO 0.10

0.05

5

~ -4

-3

-2

-1

~ 0.0

0.3

t

1

0.6

0.9

[DPA]-I/M

log ( r T B A H ] / M )

0

-1

Fig. 2. (A) Increase in ECL efficiency of DPA in DME with decreasing supporting electrolyte concentration: • y intercept from (C); • 15 mM DPA. (B) ECL spectra of 15 mM DPA: - 1 mM TBAPF6; . . . . . 100 mM TBAPF6. (C) Variation of ~bEcL as a function of DPA concentration in DME: • 100 mM TBAPF6; [] 10 mM TBAPF6: • 1 mM TBAPF6; © 0.1 mM TBAPF6.

D P A concentration. The ECL efficiency o f D P A has previously been shown to increase with increasing D P A concentration in a variety o f solvents [17]. This occurs when the electrogenerated ions undergo reactions other than the desired annihilation reactions [26]. A similar increase in the ECL efficiency o f D P A was also observed in D M E with increasing D P A concentration at all the electrolyte concentrations examined. This increase can be readily seen through a plot o f I / [ ~ E C L ] VS. I / [ D P A ] (Fig. 2). As described previously [18], such a plot should be linear with the y intercept yielding the ~/)ECL at infinite concentration where the effect o f any competing side-reactions which would lower the (~ECL is negligible. These intercept values for t~ECL are found to increase with decreasing free electrolyte concentration and are found to be approximately the same as the values o f (~ECL obtained in solutions o f 15 m M D P A (Fig. 2). Thus 15 m M is a sufficiently high concentration that the effect o f any possible interfering side-reactions which might deplete the electrogenerated radical ion reactant population is insignificant [17]. The ECL efficiency of 15 m M D P A was found to increase 2.44-fold from 0.096 to 0.234 as the added electrolyte concentration was decreased from 100 to 0.1 m M (Table 1) (Fig. 2), with the t~ECL at 0.1 m M being the highest reported for this system. The increase in (~ECL correlated with an observed increase in the absolute magnitude o f AEl/2 for D P A as the electrolyte concentration was lowered, as determined from cyclic voltammograms at a Pt disk electrode o f radius 5 ~ m (Fig. 3). These increases contrast with results previously obtained in 50: 50 A C N + TOL where only a 1.39-fold increase in t])ECL was observed accompanied by no change in A El~2 over the same electrolyte concentration range [17].

3.3. ECL spectra ECL spectra o f 15 m M D P A in D M E were measured in solutions containing 100 and 1 m M T B A P F 6 (Fig. 2). No shifts in the D P A ECL spectrum were observed as a function o f electrolyte concentration.

1.5 I

~,~

~:::::"

-3.25 AEI/2/V

1.0

-3.50 0.5 -4.0 -3.5 -3.0 -2.5 log ( [TBA+] / M )

I/I,ir%

o0

J

-0.5

-1.0

0.50

,

,

,

0.25

0.00

-0.25

/ /

,

,

-3,25

-3.50

-3.75

E / V vs DPA+/DPA

Fig. 3. Steady-state cyclic voltammograms of DPA in DME obtained at a Pt electrode of radius 5 p,m at 50 mV s- m: _ _ 1.12 mM DPA, 102.3 mM TBAPF6; . . . . . 0.07 mM DPA, 0.86 mM TBAPF6. The inset shows A Ell 2 of DPA in DME vs. the log of the free [TBA+ ] concentration. Linear regression yields a slope of 120-t-9 mV decade- ~, indicating ion pairing of both DPA radical ions with their respective electrolyte counter-ions. The [TBA+ ] concentration was determined with a K D of (7.85:0.6)× 10-5 as obtained from conductivity measurements. Estimated error in AE~/2 +0.01 V.

K.M. Maness, R.M. Wightman / Journal of Electroanalytical Chemistry 396 (1995) 85-95

89

Fable 1 ECL efficiencies and electrochemical data as a function of added electrolyte concentration ECL system a

Iog([TBAPF6]/M)

A El~ 2 b / V

AGs c / e v

tbECL d

DPA+'+ DPA-"

- 1 -2 -3 -4 - 1 -2 -3 -4 - I -2 -3 -4 - 1 -2 -3 -4

-3.33 -3.39 -3.46 -3.53 -3.08 -3.15 -3.23 -3.31 -2.55 -2.62 -2.69 -2.78 -2.53 -2.59 -2.65 -2.72

-0.22 -0.25 -0.30 -0.35 0.03 -0.01 -0.07 -0.13 0.56 0.52 0.47 0.40 -0.08 -0.12 -0.16 -0.21

0.096 0.140 0.209 0.234 0.0064 0.0089 0.018 0.054 0.0009 0.0014 0.0020 0.0038 0.029 0.033 0.039 0.044

)PA+'+ NPK- a

DPA+'+ b e n z i l - e

Rubrene+ + r u b r e n e -

+_ 0.001 + 0.001 + 0.001 _+ 0.001 + 0.0002 + 0.0001 +0.003 + 0.003 + 0.0002 _+ 0.0003 + 0.0003 _+ 0.0004 _+ 0.002 + 0.002 + 0.004 + 0.003

Concentrations of solutions: 15 mM DPA, 15 mM DPA and NPK, 5 mM DPA and benzil, 3.5 mM rubrene. Obtained at a Pt disk electrode of radius 5 p,m while maintaining 10:1 electrolyte to analyte concentration ratio. Estimated error, _+0.01 V. : AGs = AEI/2 - wr + E s, where E s is the energy level of the emitting excited singlet state. For DPA, Es = 3.0 eV; for rubrene E s = 2.3 eV. The work term wr was calculated according to Debye-Hikkel theory [23] with the following parameters: radii of reactants for DPA, D P A / N P K and DPA/benzil r = 5.15 X l 0 - 8 cm, and for rubrene r = 7.0 X 10 -8 cm [29]; Radii of counter-ions, 2.3 X 10 -8 cm; distance separating the reactants, for DPA and mixed ~ystems d = 10.3 x 10 -3 cm, and for rubrene d = 9.0 X 10 -3 cm [29]; dielectric constant, 7.2; temperature, 298 K. J Error in tbECL for different systems obtained as follows: DPA, standard deviation average of three successive potential steps; DPA + NPK, standard deviation of three successive potential steps; DPA + benzil, standard deviation of three separate solutions; rubrene, standard deviation of three separate ~olutions. Error in converting to ~bEcL, approximately 10%. : Excited singlet state of DPA is the emitting species [27].

3.4. ECL of mixed systems as a function of electrolyte concentration In addition to ECL produced from a single precursor molecule, two mixed systems were examined in DME as a function of electrolyte concentration. In these mixed systems DPA served as the radical cation and either benzil or NPK served as the radical anion. For these mixed systems

the emitting species is the excited singlet state of DPA [27]. The ECL efficiencies for both of these systems were found to increase as the electrolyte concentration was lowered, although the increases were much sharper than those observed for DPA. For DPA and benzil an increase in ECL efficiency of 4.15-fold was observed as the electrolyte concentration was lowered from 0.1 M to 0.1 mM, while an increase of 8.34 fold was observed for DPA and

-3.10 AEI/2/V

-2.8

0.004

AE1/2/V

-3.26

-2.B 0.003

@ECL

-4

-4

-3

log (r TBA+.I/M

0.002

-3

log ( [ T B A + ] / M )

-

0.03

qbEcL

0.02 0.001 B

-4

-3

-2

-1

Iog(ETBAH]/M )

-4

0

-3

-2

0.01 0 -1

Iog([TBAH]/M)

Fig. 4. (A) Increase in ECL efficiency of 5 mM DPA and 5 mM benzil in DME as a function of the supporting electrolyte concentration. The inset shows A El~ 2 of the benzil radical anion and the DPA radical cation in DME vs. the log of the free [TBA +] concentration. (B) Increase in ECL efficiency of 15 mM DPA and 15 mM NPK in DME as a function of the supporting electrolyte concentration. The inset shows A Et/2 of the NPK radical anion and the DPA radical cation in DME vs. the log of the free [TBA ÷] concentration. Estimated error in AE1/2, +_0.01 V.

K.M. Maness, R.M. Wightman/ Journal of Electroanalytical Chemistry 396 (1995) 85-95

90

ing the supporting electrolyte concentration and that these increases are accompanied by increases in the absolute magnitude of AEI/2. In turn, the increases in AE]/2 can be understood in terms of ion-pairing equilibria between the radical ions and the supporting electrolyte.

-2.55 W AE1/2/V -2.70

0.05

-4.0 -3.5 -3.0 -2.5 log([TBA+]/ M)

qbECL 0.04

1

0.03

1

0.02

_14

_15

_l2

_11

Iog([ZBAH]/M) Fig. 5. Increase in ECL efficiency of 3.5 mM rubrene in DME as a function of supporting electrolyte concentration (TBAPFt). The inset shows A E~/2 of the rubrene radical anion and cation in DME vs. the log of the free [TBA + ] concentration. Linear regression yields a slope of 114 + 4 mV decade- i indicating ion pairing of both rubrene radical ions with their respective electrolyte counterions. The [TBA + ] concentration was determined with a K o of (7.8+0.6)×10 -5 as obtained from conductivity measurements. Estimated error in A El~ 2, + 0.01 V.

NPK over this same electrolyte range (Table 1 and Fig. 4). Both increases were accompanied by corresponding increases in the absolute magnitude of A El~ 2.

3.5. ECL of rubrene as a function of electrolyte concentration The ECL efficiency of 3.5 mM rubrene in DME was also examined as a function of electrolyte concentration. As before, both ~bEcL and the absolute magnitude of A El~ 2 were found to increase as the electrolyte concentration was decreased from 100 to 1 mM (Table 1 and Fig. 5). The concentration of rubrene was limited by its solubility in DME. Thus higher concentrations, similar to those used for DPA in DME, could not be used. Therefore it might be expected that the reported values of ~bEcL for rubrene are rather low owing to the possible adverse effects of impurities on the measured thECL- However, an examination of the ~bEcL of rubrene as a function of rubrene concentration showed no increase as the concentration was increased from 1 to 3.8 mM (data not shown). Thus, unlike that of DPA, the thECL of rubrene is independent of the parent species concentration. The lack of dependence of the thECL of rubrene on the rubrene concentration indicates a greater stability of the rubrene radical ions compared with those of DPA.

4. Discussion

The data presented clearly illustrate that increases in ECL efficiency in DME can be readily achieved by lower-

4.1. Effect of electrolyte concentration on AEI /2 The strong degree of electrolyte association in DME suggests the presence of additional ion-pairing equilibria between the supporting electrolyte ions and the electrochemically produced radical ions as shown below for the generalized parent species A and the generalized supporting electrolyte ions M + and X-: A + e- = A-

(reduction)

A - e- = A ÷

(oxidation)

A++ X - = AX KAx = [ A X ] / [ A + ] [ X ] A-+M+=MA KMA = [ M A ] / [ A - ][M + ]

(4) (5) (6) (7) (8) (9)

In the presence of these ion-pairing equilibria, the A El~ 2 measured for the parent species will be a function of the free or dissociated electrolyte concentration. This was indeed the case for both DPA and rubrene as well as for the mixed DPA + benzil and DPA + NPK systems as shown in Table 1. These large changes in AE1/2 as a function of electrolyte concentration contrast with results obtained for DPA in 50 : 50 ACN + TOL where only very small changes in AEI/2 were observed, indicating the absence of ion pairing in this solvent [17]. In DME, for the ion-pairing equilibria shown above, the relationship between AE1/2 and free electrolyte concentration can be expressed as follows, assuming equal diffusion coefficients for all species [30]:

A El~ 2 = A E °' + R T / n F In( KAX KMA ) + 2 R T / n F In CM+

(10)

where AE °' is the difference between the formal potentials for reduction and oxidation of the parent species in the absence of electrolyte and CM+ is the concentration of free dissociated electrolyte, i.e. cM+ = [TBA +] = [PF6]. Thus a plot of A El~ 2 vs. log(c M+) should yield a slope of 118 mV per decade change in CM+. For DPA such an analysis yields a slope of 120 _+9 mV decade -1 (Fig. 3), while for rubrene, DPA + NPK and DPA + benzil such an analysis yields slopes of 114 ÷ 4 mV decade - l , 140 + 10 mV decade -l and 139+ 8 mV decade -1 respectively. These slopes indicate that the electrochemically generated anions and cations of all the compounds examined form ion pairs with their respective electrolyte counter-ions in DME, and it is this ion pairing which leads to the increases in the absolute magnitude of A El~ 2 with decreasing electrolyte concentration.

K.M. Maness, R.M. Wightman/ Journal of Electroanalytical Chemistry396 (1995) 85-95 At high electrolyte concentrations substantial ion pairing of the electrogeneratod ions with their electrolyte counter-ions occurs, leading to a decrease in the absolute magnitude of gEl~ 2 owing to the stabilization of the electrogenerated ions. However, as the electrolyte concentration is decreased this ion pairing becomes negligible, and the A El~ 2 values obtained at these low ionic strengths reflect the AE1/2 values that would be obtained in the absence of electrolyte and thus reflect the highest obtainable AEl/2s. That this is indeed the case can be ascertained from the following analysis. A slight rearrangement of Eq. (10) yields

exp

gEl~ 2 = e x p [ ( _ f / 2 ) A E O , l c M +

(11)

where f = nF/RT. Here KAx and KMA are assumed to be approximately equal to 1 / K D for TBAPF6, and hence are equal to 1.3 × 104. The values of A E °' can then be obtained from an analysis of the slope for a plot of e x p [ - ( f / 2 ) A E i / 2 ] vs. cM+. This yields a difference A E °' in the formal potentials for the oxidation and reduction of the parent species in the absence of electrolyte in DME of - 3 . 4 9 2 + 0.004 V and - 2 . 7 0 4 + 0.002 V for DPA and rubrene respectively. These values compare well with those obtained at 0.1 mM TBAPF6 ( - 3 . 5 3 V and - 2 . 7 2 V for DPA and rubrene respectively). Thus at this low concentration of electrolyte the electrogenerated ions are not appreciably paired with their respective electrolyte counter-ions and the measured A El~ 2 i s the largest obtainable in DME. Additional support for the absence of ion pairing at 0.1 mM TBAPF6 lies in the good agreement between the experimentally obtained values of A El~ 2 and those calculated from thermodynamic relationships (Table 2). The values of E ° / e V for standard one-electron oxidation and reduction can be related to solvation energies, ionization potentials and electron affinities as follows [33]: E ~ = 1 + AG°g + ,solv - AG°A.~ol~+ C

(oxidation) (12)

E ~ = - E A + AG°g,~olv - AG°g-,~olv + C

(reduction) (13)

91

where I is the gas phase ionization potential, EA is the gas phase electron affinity, C is a constant including the potential of the reference electrode on the absolute scale together with liquid junction potentials, and AG°A,solv, AG°g+ so|v and A G ° g -,solv are the solvation energies for the neutral, oxidized and reduced species respectively. The difference between these two relationships yields A E°: A E ° = - I - A + 2AG°g,solv - AG°A ,sony- AG°g * ,.~olv (14) The solvation energy of the neutral species will be negligible compared with that of the charged species and to a first approximation can be ignored. The solvation energies of the charged species are assumed to be equal and can be obtained approximately from the Born equation AG°~ ,solv = - l / 2 q 2 ( 1 / r ) [ 1 / e

- 1]

where q / e s u is the charge of the species, r / c m is the radius of the molecule and e is the dielectric constant of the solvent. For DPA and rubrene in DME the values of AE ° obtained from Eq. (14) (Table 2), - 3 . 5 7 V and - 2 . 7 6 V respectively, are very close to the values of AE~/2 obtained for DPA and rubrene at 0.1 mM TBAPF6. This good agreement further supports the conclusion that the values of gEl~ 2 measured at 0.1 mM TBAPF6 are the highest obtainable in DME.

4.2. Effects of increasing gEl~ 2 on the qbEcL of DPA The increases in t~ECL of DPA with decreasing electrolyte concentration and increasing AE~/2 can be understood in terms of the DPA ECL reaction mechanism, how this mechanism effects t~ECLand the relationship between the various rates of reaction and AEi/2. The ECL reaction mechanism of DPA is similar to that outlined in Section 1. The emitting species is the excited singlet state which has an emission maximum at 425 nm and q~f---1. As in Eq. (2), the ECL efficiency can be expressed as the ratio of the electron transfer rate of formation of the emitting species to the sum of all possible electron transfer rates. The unimolecular rates of electron

Table 2 Ionization potentials, electron affinitiesand AE° for rubrene and DPA in DME Compound lleV AleV h G ° h ±,solv ' / e V

A E° b/V

A El~2 c/V

DPA Rubrene

-- 3.57 -2.76

3.53 2.72

7.04 d 6.41 f

-- 1.06 e - 1.88 e

-- 1.24 -0.93

(15)

a Calculated using the Born equation with radii of 5.15 X 10-8 and 7.0 × 10-s cm for DPA and rubrene respectivelyand e = 7.2. b Obtained from Eq. (14). c Estimatederror, ±0.01 V d Ref. [31]. e Determinedby molecularorbital calculations using MOPAC with the AM1 Hamiltonianset. Values obtained were referencedto those of anthracene for which experimentallydetermined values were available[32]. f Ref. [32].

K.M. Maness, R.M. Wightman/ Journal of Electroanalytical Chemistry396 (1995)85-95

92

transfer which appear in Eq. (2) can be expressed through classical electron transfer theory in the following form:

ke, = Unffel e x p [ - - ( AG-- A)2/(4hkbT)]

(16)

where AG is the Gibbs energy available for reaction, c, is the electron transfer frequency, tel is the electronic transmission coefficient, k b is Boltzmann's constant, T is temperature and h is the sum of the inner-sphere (vibrational) reorganizational energy and the outer sphere (solvational) reorganizational energy h o [12]. This expression predicts that, for reactions occurring in the Marcus normal Gibbs energy region ( - A G < h), the rate of electron transfer will increase as the Gibbs energy available for reaction increases, i.e. as AG becomes more negative. However, increasingly negative values of AG will lead to an eventual maximum rate ( - A G = A) followed by a decreasing rate of electron transfer ( - A G > h). The values o f AG in units of electronvolts for the various competing ECL electron transfer reactions for DPA can be determined from the measured A El~ 2 through the following relationship:

AG = AEI/2

-

w r

÷ EO.o

(17)

where w~ is the work required to bring together the ion reactants as calculated from Debye-Hiickel theory [28] 1 and E0.0 is the energy needed to populate the excited singlet state (Es), the excited triplet state ( E t) or the ground state (Eg) as determined from the zero to zero energy level transition. For DPA E s, E t and Eg are 3.0 eV, 1.8 eV and 0 eV respectively [34,35]. For an estimated A of 0.546 eV 2 these energetics place the rate of formation of the excited singlet state in the normal region, while the triplet and ground states are in the inverted region. Thus k s will increase with increasing exothermicity, whereas k t a n d kg will decrease. However, if the rate to form the ground state is sufficiently inverted as to be

l

= - 0 2 [exp(Br~/-~-) ] wr

,a

[

1+

where fl = (8~rne2/lOOO~kbT)1/2, ~ is the ionic strength, • is the static dielectric constant, N is Avogadro's constant, e/esa is the electron charge, r/cm is the radius of the reactant A-'(or A + ") plus that of the dominant counter-ion in the ionic atmosphere, assuming rA-.= rA+. and a/cm is the distance between the reactants. 2 For large aromatic compounds such as DPA the inner reorganizational energy is much smaller than Ao and A~- Ao. This outer reorganizational energy can be estimated by the following equation: A0=e2 ~rA + 2r---~+--I ---rAa ] k %P where r^/cm and r a/cm are the radii of ions A and B respectively, raa/cm is the interionic distance, %p and a are the optical and static dielectric constants and e/esu is the electronic charge. For DME, aop = 1.89 and e = 7.2. The above equation gives h = 0.546 eV for DPA assuming ra = r B = 5.15 × l0 -s cm and tAB = 10.3 X l0 -8 cm.

negligible (kg = 0 ) and the rate to form the triplet state proceeds at its maximum rate (k t = kd), only k~ will be sensitive to changes in exothermicity. Support for the validity of these assumptions is found by comparison with other aromatic electron transfer reactions where electron transfer rates achieve their maximum value over a wide range of reaction exothermicities ( - A G up to 1.5 eV in ACN [36]), only decreasing at very high exothermicities. Thus for k t = k d, kg = 0 and thf = 1, Eq. (2) can be simplified to 4'ECL ----k J ( a k s + 3 k - d )

(18)

This equation predicts that thECL will approach a maximum of 25% as k s increases a n d / o r k_ d decreases. The values of thECL shown in Fig. 2 and Table 1 do indeed approach 25% as the added supporting electrolyte concentration is reduced from 100 mM to 0.1 mM, and this increase in (~ECLis in part due to increases in k s with increasing exothermicity. However, k_ d will also be a function of ionic strength, decreasing with decreasing ionic strength. Indeed, it has previously been shown that for DPA in 50: 50 A C N + TOL, a solvent system in which no changes in AEI/2 (and thus in k s) occur as the ionic strength is decreased, increases in t~ECL with decreasing ionic strength are directly correlated with decreases in k_ d as calculated by the eigen equation [37]. 3 Thus the increases in (~ECL for DPA in DME with decreasing ionic strength are due to both increases in k~ and decreases in k _ d. However, since both k s and k _ d a r e changing (approaching each other in magnitude), information about them cannot be directly obtained from t~ECL alone.

4.3. Effect of increasing AEI/2 on qbEcL of mixed systems Unlike D P A alone, the mixed systems examined here do not approach the maximum ~bEcL of 25%. Instead, as can be seen in Fig. 4, the ECL efficiencies are much lower and increase in a more exponential manner as the added electrolyte concentration is decreased. This differing response for the mixed systems can be understood in terms of the different energetics for these ion annihilation reactions. In the case o f the ion annihilation reaction between the D P A radical cation and the benzil radical ion, the energetics calculated using Eq. (17) indicate that the formation of the DPA emitting state is an endothermic process (energy deficient) with a very small k s. However, the rates of formation of the excited triplet and ground states are still

3 The eigen equation is expressed as 3(DA- + hA+ ) wrkbr k-a a2 I -exp( - wr/kbT ) where DA- and D^+ are the diffusion coefficients of the reacting species and a is the sum of the radii of the reactants. As wr becomes increasingly negative with decreasing ionic strength, k_ d decreases.

K.M. Maness, R.M. Wightman/ Journal of Electroanalytical Chemistry396 (1995) 85-95 predicted to be diffusion controlled and negligible respectively (see above). Since the reaction for the direct formation of the emitting excited singlet state is endothermic, additional reaction steps to those listed previously are usually invoked in order to account for the amount of emission [1]. These additional steps involve triplet-triplet annihilation (TTA) and are as follows: AT* + AT*

A s* + A

rate constant k'l

(19)

AT* + AT* = AT* + A

rate constant k~

(20)

=

AT* + AT* = A + A

rate constant k~

(21)

The efficiency of producing the emitting excited singlet state from these T I ' A reactions can be expressed as

6TTA = k'l/(

(22)

+ 3k' +

and the overall ECL efficiency is given by

+ 3k; +

+ (23)

where q~v is the efficiency of excited state triplet production and 8 is the fraction of excited state triplets that fail to find another excited state triplet reaction partner. Since the energetics of the ion annihilation reactions predict that k't = k d and k's < < k 't, ~v will approach unity for this system. However, ~6rr a is expected to be independent of ionic strength (only neutral species are involved) and very t small [29]. The above considerations (k't - - k d, k 8 = 0 , 4~f = 1, 4~r = 1 and k s << k_d), along with the assumption that 8 is also constant allow simplification of the above expression for ~bECL: ~bECe ----kJ3k_ d + (1 - 8 ) 4~TTA

(24)

ThUS for this system, as for DPA, increases in 4~ECL with decreasing ionic strength could be due to both increases in k s and decreases in k_ d. However, since the increases in k~ are much greater than the decreases in k_~d, k_ d can be considered constant 4 and changes in ~ECL are directly proportional to changes in k s. Thus the shape of the increases in 4~ECL for D P A and benzil with decreasing ionic strength (and increasing A Q ) (Fig. 4) contrast with the shape of the increase for DPA where k s is approaching diffusion control and a limit of 25% (Fig. 2). The qualitative arguments presented above also hold for the NPK +

4 If, to a first approximation, the radii, diffusion coefficients and solvent reorganizational energy A0 for both reactants are assumed to be equal to those of DPA in DIMEas a function of ionic strength, k_ d will decrease by 87% as the added TBAPF6 concentration is lowered from 100 mM to 0.1 mM, i.e. from 3.95×108 s -t to 0.51 × 10s s -l respectively. However, ks will increase by factors of 421 and 16 for benzil+ DPA and NPK + DPA, respectively over this same electrolyte concentration range. Although both the diffusion coefficients for benzil and NPK and the A0 for the electron transfer reactions will be slightly larger, in accordance with the slightly smaller size of these two molecules compared with DPA, the general outcome will be the same.

93

DPA mixed system as shown by the shape of the increase in ~bECL with decreasing ionic strength (Fig. 4).

414. Comparison of qbecL for the DPA-containing systems The magnitude of I~ECL for each system examined above correlates with the magnitude of AG s determined for that system. Thus, as can be seen in Table 1, the ECL efficiency is greatest for DPA and least for benzil and DPA. This correlation of ~bEcL with AG s for the different systems can be attributed to the dependence of thECL on k s and AG s as outlined above and is expected. These systems, amongst others, were also examined by Beideman and Hercules [27] in 5 0 : 5 0 ACN + benzene with 100 mM tetrabutylammonium perchlorate as the supporting electrolyte. In these experiments the ECL was produced by application of a 2 Hz square wave to a single electrode to produce radical cations and radical anions, alternately. Thus each ion was produced for a total of 0.25 s on each cycle. Beideman and Hercules reported the same qSECL for both DPA + benzil and D P A + NPK in 50 : 50 A C N + benzene (1.2% for both), although the measured available Gibbs energies for the two reactions were significantly different. Such an observation would be understandable if the majority of the emission arose from TTA which would be little affected by changes in the available Gibbs energy for the electron transfer reaction ( k t should be diffusion controlled for both systems). 5 The ECL efficiencies reported by Beideman and Hercules for these two systems in 50 : 50 A C N + benzene are higher than those obtained here in DME. This result is unexpected since the dielectric constant of DME (7.2) is much lower than that of the 50 : 50 A C N + benzene solvent mixture where the radical ions are preferentially solvated by the more polar solvent. 6 A solvent of lower dielectric constant should yield higher ECL efficiencies because of the lower solvent reorganizational energies necessary for electron transfer in such solvents [13,29]. The lower thECL reported here may be due to a difference in the values of 8 (Eq. (23)) for the two experimental systems since the value of 8 is a function of the manner in which the radical ion reactants are generated. A larger 8 will lead to a negligible contribution to the ~bEcL by TTA

5 Although Beideman and Hercules [27] presented evidence for the absence of TTA for the NPK+benz(a)anthracene system in 50:50 ACN + benzene through the lack of ECL quenching in the presence of the triplet quencher trans-stilbene, there was no direct evidence for the lack of TTA for NPK + DPA in this solvent mixture. 6 Kapturkiewicz [29] has estimated that the dielectric constant • of a 50:50 ACN + benzene mixture is 15.5, based on the relationship between 1/~ and the standard redox potentials for the oxidation and reduction of rubrene. However, in ACN + TOL mixtures the standard redox potentials of DPA were found to be constant as the percentage of TOL was increased, indicating the preferential solvation of the DPA radical ions by ACN [18].

94

K.M. Maness, K M. Wightman/ Journal of Electroanalytical Chemistry 396 (1995) 85-95

and thus a smaller t~ECL. That 8 is large during production of steady-state ECL at double-band microelectrodes is understandable in light of the steep concentration gradient for diffusion of products out of the thin reaction zone formed between the adjacent bands. Thus the excited state triplets formed upon ion annihilation will diffuse away and be quenched before encountering another excited state triplet and undergoing "l*rA.

4.5. Effects of increasing AEI/2 on the qbecL of rubrene The gradual increase in t~ECL observed for rubrene as the electrolyte concentration decreases and the A El~ 2 increases is reminiscent of that observed for DPA alone. However, although the rate of the increase is similar, the magnitude of the (J~ECL for rubrene is much less than that observed for DPA. As before, the observed increase in t~ECL can be qualitatively understood in light of rubrene's ECL mechanism and the energetics of the various electron transfer reactions. In the absence of TI"A the ECL mechanism of rubrene is similar to that outlined in Section 1 with one additional electron transfer pathway. This extra pathway arises as a result of the unique energetic relationship between the rubrene excited singlet and triplet states. The energy needed to populate the excited singlet (2.3 eV) [38] is exactly twice that needed to populate the excited triplet state (1.15 eV) [39]. Thus, when rubrene undergoes ion annihilation it is possible for two excited triplet states to form simultaneously, a process which is isoenergetic with the production of the excited singlet state: A + ' + A - ' = AT* + AT*

rate c o n s t a n t k'tt

(25)

The two triplets formed may then react to produce an excited singlet state. However, for rubrene in DME (25°C) the excited triplet states will diffuse apart before reacting [29]. Thus this extra electron transfer pathway is nonemitting and leads to a decrease in the magnitude of the ~bECL of rubrene compared with that of DPA. Incorporation of this extra electron transfer pathway step into the expression for thECL in Eq. (18) (for which all assumptions are still valid), accounting for the spin statistics of the electron transfer and assuming k~ = ktt, yields ~ECL = kJ(13ks + 3k_d)

(26)

Thus, as k~ increases with increasing AEI/2 and AQ, the maximum value of qbECL is predicted to be 0.077 for rubrene in DME at 25°C. The similarity in the rise and approach of ~bEcL towards a maximum for both rubrene and DPA implies that the rate to form the excited singlet state of rubrene, like that for DPA, is approaching diffusion control. However, unlike those for DPA, the experimentally observed values of

t~ECL for rubrene do not approach the theoretical maximum of 0.077. Instead, they approach a maximum of 0.044, approximately 40% lower than the theoretical value. This discrepancy in the maximum t~ECL cannot be completely attributed to the 10% error imparted in the conversion of the measured coulometric efficiencies into ~ECL values. Indeed, the t~ECL values for rubrene reported here are lower than those recently reported for the same system in DME (~ECL ~-- 0.041, 0.1 M TBAPF6, 25°C [29]). Such large discrepancies in the reported values for rubrene #bECL are common and may reflect the presence of an additional electron transfer pathway unaccounted for in the proposed rubrene ECL mechanism. 5. Conclusion The use of double-band microelectrodes for the production of ECL in this study has allowed ECL efficiencies to be examined as a function of ionic strength. All systems considered here show increases in ECL efficiency as the added electrolyte concentration is lowered, and these increases are accompanied by increases in AEI/2 and thus in k~ in accordance with electron transfer theory. The strong tendency for ion pairing of all the radical ions considered here with their corresponding supporting electrolyte ions in DME allows the available energy of reaction to be manipulated easily by changing the electrolyte concentration. This ease of manipulation has allowed the reaction Gibbs energy of the DPA ion annihilation to be increased, leading to ECL efficiencies approaching the maximum theoretical limit of 25%. This ability to increase the reaction Gibbs energy has also allowed insight to be gained into the ECL reaction mechanisms of NPK + DPA and benzil + DPA at the double-band microelectrode with the indication of an apparent lack of emission arising from TTA. Such manipulation of the reaction Gibbs energy should also prove to be a useful tool for further investigations of other ECL reaction mechanisms in low dielectric media.

Acknowledgments We gratefully acknowledge Dr. Richard Buck for use of the Solartron-Schlumberger model 1250 frequency response analyzer and a model 1186 electrochemical interface and Dr. Tal Nahir for help with the conductivity measurements. We also thank Dr. Gary Glish for help with the MOPAC calculations of ionization potentials and electron affinities. This research was supported by the National Science Foundation (CHE). References

7 Nine encounter-pair spin states are formally possible for two interacting triplets.

[1] L.R. Faulkner and A.J. Bard in, A.J. Bard (Ed.), J. Electroanalytical Chemistry, Vol. 10, Dekker, New York, 1977, p. 1.

K.M. Maness, R.M. Wightman/ Journal of Electroanalytical Chemistry 396 (1995) 85-95 [2] S.M. Park and D.A. Tryk, Rev. Chem. lnterrned., 41 (1981) 43. [3] 1. Rubinstein and A.J. Bard, J. Am. Chem. Soc., 103 (1981) 515. [4] M-M. Chang, T. Saji and A.J. Bard, J. Am. Chem. Soc., 99 (1977) 5399. [5] W.B. Beeker and A.J. Bard, Anal. Chem., 56 (1984) 2413. [6] J.K. Leland and M.J. Powell, J. Electrochem. Soc., 137 (1990) 3127. [7] E. Humphreys and D.J. Malcolme-Lawes, J. Chromatogr., 329 (1985) 281. [8] E. Hill, E. Humphreys and D.J. Malcolme-Lawes, J. Chromatogr. 370 (1986) 427. [9] T. Nieman in J.W. Birks (ed.), Chemiluminescence and Photochemical Reaction Detection in Chromatography, VCH Publishers, New York, 1989, p. 99. [10] J.B. Noffsinger and N.D. Danielson, Anal. Chem., 59 (1987) 885. [11] R.A. Marcus and N. Sutin, Biochim. Biophys. Acta, 811 (1985)265. [12] A. Pighin and B.E. Conway, J. Electrochem. Sot., 122 (1975) 619. [13] H. Tackikawa and A.J. Bard, J. Chem. Phys. Lett., 26 (1974) 245. [ 14] A. Kapturkiewiez, Z. Phys. Chem., 170 (1991) 87. [15] J. Kim and L.R. Faulkner, J. Am. Chem. Soc., 110(1988) 112. [16] C.P. Keszthelyi, N.E. Tokei-Takvoryan, H. Tachikawa, and A.J. Bard, J. Chem. Phys. Lett. 23 (1973) 219. [17] K.M. Manes, J.E. Bartelt and R.M. Wightman, J. Phys. Chem., 98 (1994) 3993. [18] J.E. Bartelt, S.M. Drew, and R.M. Wightman, J. Electrochem. Soc., 139 (1992) 70. [19] R.M. Wightman and D.O. Wipf in, A.J. Bard (Ed.), Electroanalytical Chemistry, Vol. 15, Dekker, New York, 1988, Ch. 3, p. 267. [20] B. Fosset, C.A. Amatore, J.E. Barlelt, A.C. Michael and R.M. Wightman, Anal. Chem., 63 (1991) 306. [21] N.E. Tokel-Takvoryan, R.E. Hemingway, and A.J. Bard, J. Am. Chem. Soc., 95 (1973) 6582.

95

[22] W.L. Wallace and A.J. Bard, J. Phys. Chem., 83 (1979) 1350. [23] J.N. Demas and G.A. Crosby, J. Phys. Chem., 75 (1971) 991. [24] J. Eliassaf, R.M. Fuoss and J.E. Lind Jr., J. Phys. Chem., 67, (1963) 1941. [25] P.W. Atkins, Physical Chemistry (3rd edn), W.H. Freeman, New York, 1986, p. 665. [26] R. Bezman, and L.R. Faulkner, J. Am. Chem. Soc., 94 (1972) 6317; 95 (1973), 3083. [27] F.E. Beideman and D.M. Hercules, J. Phys. Chem., 83 (1979) 2203. [28] N. Sutin, Acc. Chem. Res., 15 (1982) 275. [29] A. Kapturkiewicz, J. Electroanal. Chem., 372 (1994) 101. [30] M.E. Peover and J.D. Davies, J. Electroanal. Chem., 6 (1963) 46. [31] E. Anxolahehere, P. Hapiot and J.M. Saveant, J. Eiectroanal. Chem., 282 (1990) 275. [32] S.G. Lias, J.E. Bartness, J.F. Liebman, J.L. Holmes, R.D. Levin and W.G. Mallard, J. Phys. Chem. Ref. Data, 17 (Suppl. 1) (1988). [33] M.E. Peover in, A.J. Bard (Ed.), Electrochemistry of Aromatic Hydrocarbons and Related Substances, Electroanalytical Chemistry A Series of Advances, Vol. 2, Dekker, New York, 1967, p. 40. [34] L.R. Faulkner, H. Tachikawa and A.J. Bard, J. Am. Chem. Soc., 94 (3) (1972) 691. [35] J.S. Brinen, and J.G. Karen, Chem. Phys. Lett., 2 (1968) 671. [36] 1.R. Gould, R.H. Young, R.E. Moody and S. Farid, J. Phys. Chem., 95 (1991) 2068. [37] C. Chiorboli, M.T. lndelli, M.A.R. Scandola, and F.J. Scandola, J. Phys. Chem., 92 (1988) 156. [38] S.J. Stickler and R.A. Berg, J. Chem. Phys., 37 (1962) 814. [39] W.G. Herkstoeter, and P.B. Merkel, J. Photochem., 16 (1981) 331.