摘要:

Analyst, December 1996, Vol. 121 ( I 759-1 762) 1759 Recent Evolution of Luminescent Photoinduced Electron Transfer Sensors* A Review A. Prasanna de Silva, Thorfinnur Gunnlaugsson and Terence E. Rice School qf Chemistry, Queen’s Uniiiersity, Belfast, UK BT9 5AG The photoactive supermolecule ‘lumophore-spacer-receptor’ is shown to be capable of considerable tuning and growth to satisfy the requirements of a versatile sensing system. Keywords: Luminescence; jluorescence; sensors; switches; photoinduced electron transfer; supr-amolecu/ar photophysics Modularity brings multiple advantages to any functional system. While this truth has been appreciated by engineers for many decades, chemical designers are only beginning to exploit the possibilities that arise. We and others find that the field of chemical sensing can be considerably enriched by designing molecules which combine lumophores and receptors in a modular Since this paper in intended to review the results from our laboratory, we apologize to those in other laboratories who have produced much good work and we hope to redress the balance on a separate occasion.The photonic signalling of a host-guest recognition event can be easily arranged by coupling a receptor to a lumophore. However, the supramolecular nature of the system can be best preserved by interposing a spacer between the original units.8.9 Then, the components or modules, each with its characteristic attributes, are clearly recognizable. If the components were completely isolated, the system would be unable to transduce the guest recognition event into a photonic signal.Hence we need a lumophore-receptor interaction which can transcend the spacer. Photoinduced electron transfer (PET) fits the bill admirably since it is an enduringly long-range process. 10,1 PET brings with it another appealing property, straightforward thermodynamic planning with excited state energies and redox potentials. 12 Kinetic analyses are also possible.133l4 Our design relies on the fact that the redox potentials of a receptor module would be significantly perturbed on binding a guest, especially if the guest was ionic, whereas the excited state energy and the redox potentials of the lumophore module would be much less affected. Hence the PET thermodynamics can be arranged to switch from favourable to unfavourable (or vice versa) on guest binding.Emission from the lumophore is a constant competitor with PET as a means of excited state deactivation. Hence guest binding can cause switching of luminescence between ‘off‘ and ‘on’ states. This simple sensor design has proved to be reliable for two decades.Is-I7 In ideal cases (of which there are severa11x-2s)), the UV/VIS absorption spectra remain virtually untouched by the guest, as do the emission spectra except for their quantum yield. The experimental guest binding isotherm can be quantitatively predicted from conventional mass action and from the binding constant of the receptor module. Failure to meet these high standards is usually due to lumophore-receptor interactions ’ Presented at the 6th European Conference on Electroanalysis, Durham.UK, March 2-29. 1996. (other than PET) crossing the spacer.26-2y Nevertheless, even these systems are perfectly useful for practical sensing pur- poses. Lumophore modules with excitation wavelengths from ultraviolet to green and emission wavelengths from violet to orange have been incorporated into PET sensors. Receptor modules targeting protons (pH), calcium, magnesium, sodium, phosphate and glucose in concentration ranges of physiological relevance have also seen service in PET sensing systems. The schematic design and two examples are shown in Fig. 1. The thermodynamic planning for sensor 2 proceeds as follows.*~ NJV-Dimethyl-2-anisidine has an oxidation potential of 0.82 V (veiws SCE). This can serve as a model for the metal ion-free receptor module of 2 as far as one-electron oxidation is concerned.The oxidation potential of the lumophore module of 2 can be estimated from model rhodamine dyes as 1.44 V. Therefore, the photoexcited lumophore will receive an electron from the receptor with a thermodynamic driving force of about -0.62 eV (a spontaneous process). On the other hand, the Ca2+- bound receptor module of 2 can be estimated to have an oxidation potential of about 1.76 V, which corresponds to a thermodynamic driving force of about +0.32 eV (a non- spontaneous process) for PET in Ca2+-bound 2. Electron transfer is naturally subject to control by electric fields. The sensitivity of receptor redox potential to ionic guests, so necessary for the success of PET sensing, is an aspect of this Fig.1 Design and realization of fluorescent PET sensors. F = Fluor- ophore; S = spacer; R = receptor. Note that the two adjacent n-electron systems lie perpendicular to each other about the o-bond shown in red for structure 2.1760 Analyst, December 1996, Vol. I21 general truth. However, this also means that any photoinduced electric fields, such as those found in internal charge transfer (ICT) excited states of push-pull fluorophores, can seriously interfere with the simple thermodynamic planning of PET sensing, which remains innocent of such transient effects. Such interference can be either constructive or destructive, depending on whether the local electric field accelerates or retards the transiting electron. Hence one regioisomer, e.g., 3, can be a well behaved PET pH sensor, whereas the other, e.g., 4 is not (Fig.2). While this can be taken as a note of caution to sensor designers keen on mining the rich seam of push-pull fluor- 0phores,~7 this also allows us to mimic some aspects of the unidirectionality of PET seen in the photosynthetic reaction centre30 with much simpler supermolecules.28 Although the sharp ‘on-off’ switching of luminescence induced by the guest without any other spectral effects is ideal for chemical sensing under well defined conditions in the laboratory, more technical uses in hospitals or inside living cells require some method of internal referencing. Current sensors for use in these fields achieve such internal referencing either by ratioing intensity signals at two wavelengths, for systems that show guest-induced spectral wavelength shift^,^ * or by using emission lifetime rather than steady-state intensity as the sensory channel.32-j5 In wavelength ratioing, one population of sensor molecules (say, guest-bound) can be imagined to be Fig. 2 Regiocontrol in fluorescent PET sensors.F(1CT) = Push-pull fluorophore with internal charge transfer excited state. Note that the two cases employ different connectivities to the fluorophore. serving as internal reference for the other population of guest- free sensors. We have recently shown how an internal referencing module can be ‘added on’ to the basic PET sensing scheme.?6 This system is distinguished by the fact that internal referencing is achieved at the level of single molecule rather than a large population.Hence, these systems are true single molecule devices which are ready for the challenges of molecular information handling in the future.37 Fig. 3 outlines such a case (5) for pH sensing where one fluorophore (anthracene) is chosen to be PET active with the amine receptor whereas the second fluorophore (3-aminonaphthalimide) is chosen to be weakly PET active at best. When the anthracene module is preferentially excited, emission is observed from anthracene and 3-aminonaphthalimide in clearly distinguish- able spectral regions. The former emission is sharply switched off at basic pH while maintaining the anthracene band shape. In complete contrast, the 3-aminonaphthalimide band shows a much smaller pH effect. This band serves as an internal reference for the anthracene emission sensory signal and wavelength ratioing is easily achieved.The fact that the sensory and the reference signals can be accumulated over very wide wavelength bands is appealing. As a bonus, these triad systems allow an insight into electronic energy transfer (EET) across intervening electron pairs which can be ionically switched in or out of the EET path. As mentioned above, emission lifetimes can also be used for sensing when internal referencing is desired. Of course. such methods become more convenient if the lifetimes are signifi- cantly longer than nanoseconds. Sensors with long lifetimes can bring out another and more important advantage with regard to medical or biological contexts. The presence of matrix autofluorescence and light scattering can contribute sub- stantially to the background noise associated with the fluores- cence signal from the sensor. If the sensory emission is long- lived, the sensor signal can be time-resolved out from the background noise (Fig.4). While we have previously achieved this goal with ‘message in a bottle’ sensing systems utilizing organic phosphorescence,js we have also examined lanthanide lumophores as PET sensors components, especially because of their superior insensitivity to molecular oxygen. We modified a known luminescent label39 by building-in tertiary amine receptors for protons (6).40 The luminescence emission features due to the complexed terbium ion are switched on by a factor of 16 as protonation of the amine receptors takes place.Most important, the lifetime of the metal luminescence in acidic solution is 0.16 ins. In general, systems such as 6 will raise interesting issues regarding the initial reduction site during PET and the competition between EET and PET. Most molecular sensors target one chemical species or one environmental property. However, there are situations of Fig. 3 Internally referenced sensing. Fs = Sensory fluorophore; FIR = internal reference fluorophore.Analyst, December 1996, Vol. I21 1761 interest which involve two species coincident in space-time, e.g., an enzyme and its cofactor are both required before a substrate can be processed. While two sensors separately targeted to each chemical species can deliver the necessary information, it is more elegant if a single sensor can perform the same task.Also, the use of a single sensor eliminates any registration difficulties possible in the two-sensor experiment, i.e., the accurate overlapping of the data from each sensor for a given point in space-time. Such difficulties can be particularly serious when molecular-scale microenvironments are being addressed. If a PET-active receptor selective to a second guest is added on to the basic PET sensor as shown in Fig. 5 , we can develop the idea of coincidence sensing. Radiation physicists have used the related method of coincidence counting for many years. In the chemical arena, we have examined the related but distinct idea of two separate spectral parameters.41.42 Case 7 requires the simultaneous presence of protons and sodium ions at sufficient concentration in its locality before a fluorescence signal is generated.Although such systems are important as molecular AND logic gates43 for future information pro- cess0rs,~7we must not lose sight of their more immediate applications as coincidence sensors; 7 would give a direct visual indication of regions where both protons and sodium ions congregate. To round off this short review, we draw attention to the possibilities raised by truly molecular sensor systems. With their subnanometre spatial resolution, molecular sensors can Fig. 4 Time-resolved sensing. L = Lanlhanide ion lumophore; RJA = receptor for lumophore which also serves as a photon antenna; RA = receptor for analyte guest. (7) for coincident H+and Na+ Fig. 5 Coincidence sensing. Note that the two receptors R and R’ have mutually exclusive selectivity characteristics towards incoming guest species.(8) for H+ near membranes Fig. 6 Targeted sensing. AD = Anchoring and targeting unit; T’ = targeting unit for fine positioning of receptor.1762 Analyst, December- 1996, Vol. 121 penetrate to the very heart of biological action. When we fit transport modules to the basic PET scheme (Fig. 6), we can drive the sensor to a microlocation. On parking there we can receive space-selected information. For instance, 8 can be anchored in a membrane while its receptor module samples the membrane-bounded protons in the manner of a submarine periscope.44 Different regions near the membrane can be examined by structurally tuning the hydrophobicities of the targeting modules.These systems can aid studies in bio- energetics where protons gradients near membranes are all- important. We thank EPSRC, DENI, NATO (No. 921408 joint with Professor Jean Philippe Soumillion at the Universitk Catholique de Louvain), Queen’s University, Dr. Nimal Gunaratne, Tan Gibson and Frank Guthrie for support and help. References 1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 Bryan, A. J., de Silva, A. P., de Silva, S. A., Rupasinghe, R. A. D. D., and Sandanayake, K. R. A. S., Bio.sensors, 1989, 4, 169. Bissell, R . A. de Silva, A. P., Gunaratne, H. Q. N., Lynch, P. L. M., Maguire, G. E. M., and Sandanayake, K. R. A. S., Cheni. Soc. Rev., 1992, 21, 187. Bissell, R. A., de Silva, A. P., Gunaratne, H. Q, N., Lynch, P.L. M., Maguire, G. E. M., McCoy, C. P., and Sandanayake, K. R. A. S., Top. Cii1.i.. Cheni., 1993, 168, 223. Czamik. A. W., Adv. Siiprutriol. Clwm., 1993, 3, I3 I . Czamik, A. W., Arr. Chem. Kes., 1994, 27, 302. Fabbrizzi, L., and Poggi, A,, Chem. Soc. Rev., 1995, 24, 197. James, T. D., Linnane, P., and Shinkai, S., J . Chem. Soc., Chem. Balzani, V .. and Scandola, F., S~~~~ramoleciilur Photothemistry, Ellis Horwood, Chichester, 1990. Lehn, J.-M., Suprurnolecular Cheniistiy, VCH, Weinheim, 1995. Closs, G. L,. and Miller, J. R., Science, 1988, 240, 440. Paddon-Row, M. N., Act. Chern. Res., 1994, 27, 18. Weller, A., Pure Appl. Cheni., 1968, 16, 1 15. Marcus. R. A, Aticqe~4t. Chenz., Int. Ed. Engl., 1993, 32, 1 1 1. Rehm, D., and Weller, A., Isr. .I. Chern., 1970, 8, 259.Wang, Y. C., and Morawetz, H., .I. Anz. Chenz. Soc.., 1976, 98, 361 I . Selinger, B. K., A i m . .I. Clicrn.. 1977, 30, 2087. Shizuka, H., Nakamura, M., and Morita, T., J . Phvs. Chem., 1979,83, 2010. de Silva, A. P., and de Silva, S. A,, J . Clwm. Soc,., Chem. Comrnun., 1986, 1709. de Silva, A. P., de Silva, S. A., Dissanayake, A. S.. and Sandanayake, K. R. A. S., J . Chcni. Sot.., Chcw7. Coniniun., 1989, 1054. COn7fl?l4lZ., 1996, 28 I . 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 de Silva, A. P., and Gunaratne, H. Q. N., .I. Chem. S o c , . , Cheni. Comrnun., 1990, 186. de Silva, A. P., and Sandanayake, K. R. A. S., J . Chenz. Soc,., Chem. Conznzun.. 1989, 1183. de Silva, A. P., Gunaratne, H. Q. N., and Maguire, G. E. M., J .Cheni. Sot. .. Chem. Cornmiin., 1994, 12 13. Minta, A., Kao, J. P. Y., and Tsien, R. Y., .I. Biol. Clirni., 1989, 264, 8171. Kuhn, M., Bioprohes, 1992. 16, 5. de Silva, A. P., Gunaratne, H. Q. N., Kane, A. T. M., and Maguire, G. E. M., Chem. Lett., 1995, 125. de Silva, A. P., and Rupasinghe, R. A. D. D., J . Clieni. Soc., Clwtn. Cornnuin., 1985, 1669. de Silva, A. P., Gunaratne. H. Q. N., Lynch, P. L. M., Patty, A. J., and Spence, G. L., .I. Chern. Soc., Perkin Trai~s. 2, 1993, 161 1. de Silva, A. P., Gunaratne, H. Q. N., Habib-Jiwan, J.-L., McCoy, C. P., Rice, T. E., and Soumillion, J.-P., Angew. Chern., Int. Ed. EngI., 1995,34, 1728. de Silva, A. P., Gunaratne, H. Q. N., and Lynch, P. L. M., .I. Chem. Soi,.. Prrkin Trans. 2 , 1995, 685. Tlze Photosynihetic Reuclion Center. ed. Diesenhofer, J., and Norris, J. R., Academic Press, San Diego, 1993, vnls. I and 11. Tsien, R. Y., Chem. Eng. N e ~ i s , 1994, July 18, 34. Pardo, A., Poyato, J. M. L., Martin, E., Camacho, J. J., and Reyman, D., J . Lumin., 1990, 46, 381. Szmacinski, H., and Lakowicz, J . R., A n d . Cheni., 1993, 65, 1668. Draxler, S., and Lippitsch, M. E.. Sens. Actuator-s, B , 1995, 29, 199. Van den Bergh, V., Boens, N., deSchryver, F.C., Gallay, J., and Vincent, M., Photochem. Photoliol., 1995, 61, 442. de Silva, A. P., Gunaratne, H. Q. N., Gunnlaugsson, T., and Lynch. P. L. M., NCM, J . Chem., 1996, 20, 87 I . de Silva, A. P., and McCoy, C. P., Clzem. Ind. (London), 1994. 992. Bissell, R. A., and de Silva, A. P., J . Chem. So(.., Chcni. Cornmun., 1991, 1148. Toner, J. L.: US Put., 4 837 169, 1989. de Silva, A. P., Gunaratne, H. Q. N.. and Rice, T. E. AIZ~CMJ. Clicni., Int. Ed. Engl., 1996, 35, 21 16. de Costa, M. D. P., de Silva, A. P., and Pathirana, S. T., Can. J . Chenz., 1987, 65, 1416. Sandanayake, K. R. A. S., James, T. D., and Shinkai, S., Clwm. Lett., 1995,503. de Silva, A. P., Gunaratne, H. Q. N., and McCoy, C. P., A’ufurc (London), 1993, 364. 42. Bissell, R. A., Bryan, A. J., de Silva, A. P., and McCoy, C. P., J . Chem. Soc., Chern. Commun., 1994, 405. Puper 6/03726G Received Muy 29, 1996 Accepted July 12, I996

ISSN:0003-2654    

DOI:10.1039/AN9962101759

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 ( I 763-1 768) 1763 Sensing of Transition Metals Through Fluorescence Quenching or Enhancement* A Review Luigi Fabbrizzi, Maurizio Licchelli, Piersandro Pallavicini, Donatella Sacchi and Angelo Taglietti Dipartimento di Chimica Generale, Universita di Pavia, via Taramelli 12, I-271 00 Pavia, Italy A series of fluorescent sensors for transition metal ions were synthesized by linking a light-emitting subunit, anthracene, to a polyaza chelating subunit, either a dioxotetraamine or a tetraamine. Sensing of the divalent cations Cu", NiI1 and Zn" was investigated through spectrofluorimetric titrations in acetonitrile-N water (4 : 1) solutions. The selective recognition of Cu" and NiI1 among other transition and non-transition metal ions is signalled through full quenching of fluorescence; discrimination between the two ions can be achieved by performing titrations at controlled pH.The system containing the tetraamine fragment, whose interaction with CulI and Nil* induces fluorescence quenching, is also sensitive to Zn", but in this case the recognition is signalled through a fluorescence enhancement. Keywords: anthracene derivatives; copper(u) complexes; nickel(!!) complexes; zinc([!) complexes; fluorescent sensors; spectrofl uorimetry Introduction The modification of the luminescent emission of a given lumophore represents one of the most convenient ways to signal the occurrence of a recognition process in solution.' Thus, a luminescent sensor for a given substrate can be designed by following a multicomponent approach, i.e., by linking a lumophore to a receptor displaying selective affinity towards the envisaged substrate.2 Linking should not create any steric disturbance to the receptor's binding tendencies and, critically, the substrate-receptor interaction should trigger an intra- molecular process that modifies the emission of the proximate lumophore.Such a situation is easily achieved when the substrate is a transition metal. In fact, cations with partly filled d levels can be involved in electron transfer (eT) and energy transfer (ET) processes to/from the adjacent photo-excited lumophore, which quench its fluorescence either partially or completely. Therefore, the occurrence of the recognition process in solution can be visually perceived through the disappearance or decrease of the luminescence emission and monitored using an intensity versus equivalents plot.In this paper, the design of luminescent sensors for divalent 3d metal ions is considered in some detail. As transition metals display their highest affinity towards nitrogen donor atoms (in particular amines), the receptor component will be a polyaza ligand. In each case, an anthracene fragment will be used as a lumophore. Anthracene displays an intense emission band with a well defined and characteristic vibrational structure. A relatively high chemical stability allows a facile functionaliza- * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29, 1996. tion on its framework, which makes the synthesis of anthracene- based sensors especially convenient.The anthracene frag- ment3-9 and other polyaromatic fluorophoreslo have been utilized as reporter groups in several supramolecular systems used for sensing or signalling purposes. Fluorescence sensors bearing naphthyl luminescent fragments sensitive to copper(r1) and nickel(I1) have been reported recently.' Experimental Materials Diethyl malonate, 9-bromo-anthracene, 9-chloromethyl-anthra- cene, anthracene-9-carbaldehyde (Fluka, I3 uchs, Switzerland), 4-nitrobenzaldehyde and 4-(dimethy1amino)benzaldehyde (Al- drich, Milwaukee, WI, USA) were used without further purification. A solution of butyllitium in hexane (1.6 mol I-I) was used as purchased from Aldrich. A solution of ethylene oxide was prepared by dissolving 2 ml (40 mmol) of ethylene oxide (Aldrich) in anhydrous diethyl ether (20 ml).N,N'-Bis (2-aminoethy1)propane- 1,3-diamine (2.3-2-tet) was prepared as described for the analogous tetramine 3-2.3-tet,12 distilled at reduced pressure (125 "C; 5x10--2 Torr) and stored over NaOH in a refrigerator. Syntheses N,N'-Bis(2-aminoethyl)-2-anthracen-9-ylmethylmalonamide (lay3 and N-(2-aminoethyl)-N'{ 2-[ (anthracen-9-ylmethy1)- amino]ethyl}propane-l,3-diamine (2), (Scheme 1) were pre- pared as rep0rted.1~ Compound l b was prepared through the multi-step synthetic procedure shown in Scheme 2. 2-Anthracen-9-ylethanol (4a) 9-Bromoanthracene (7.8 mmol) was dissolved in anhydrous diethyl ether (70 ml) in a three-necked round-bottomed flask under a dinitrogen atmosphere. The solution was cooled to 0 "C and 6 ml of a solution of butyllithium in hexane (1.6 mol 1-1) were added over the period of a few minutes under magnetic stirring.After 30 min, a solution of ethylene oxide (40 mmol) in diethyl ether (20 ml) was slowly added and magnetic stirring was continued for 2 h. The resulting solution was treated with ice-water (50 ml) and diethyl ether (30 ml), then the aqueous phase was extracted with CH2C12 (3 X 30 ml). The combined organic phases were washed with water and dried over Na2S04. Solvents were distilled off using a rotary evaporator and the brown-yellow solid obtained was purified by liquid chromatog- raphy (SiO2, CHZC1,); yield, 70.1 %. Elemental analysis: found, C 86.19, H 6.51; calculated for Cl6HlrlO, C 86.45, H 6.35%. 1H NMR spectrum (CDC13): 6 8.43 (s, l), 8.37 (d, 2), 8.06 (d, 2), 7.59-7.49 (m, 4), 4.14 (t, 2), 3.99 (t, 2), 1.58 (broad s, OH).1764 Analyst, December 1996, Vol.121 Methunesulfonic acid 2-unthracen-9-ylethyl ester, (4b) Methanesulfonyl chloride (MsC1, 0.47 ml, 6 mmol) was added to a stirred solution of 4a (5.4 mmol) and triethylamine (1.1 ml) in anhydrous CH2C12 (30 ml) over a period of 10 min. The mixture was allowed to remain at room temperature for 2 h under magnetic stirring, then 30 ml of water were added. The aqueous layer was extracted with CH2C12 (3 X 25 ml) and the extract was consecutively washed with 10% HCl solution, aqueous NaHCO? solution and water. The combined organic layers were dried over MgS04 and the solvent evaporated in \~m40, giving a solid which was purified by liquid chromato- graphy [SiOz, hexane-CHCl3 (6 + 4)]; yield, 85.8%.1H NMR spectrum (CDC13): 6 8.46 (s, l ) , 8.31 (d, 2), 8.06 (d, 2), 7.64-7.59 (m, 2), 7.55-7.51 (m, 2), 4.63 (t, 2), 4.17 (t, 2), 2.84 (s, 1). 9- (2 -lodoe th y1)anth i-acen e (5 ) A solution of 4b (3.1 mmol) and NaI (15 mmol) in acetone (30 ml) was refluxed overnight. After evaporating the solvent, the n = i : l a ; n = 2 : Ib 2 R=NO,: 3a ; R=NMe2 : 3b Scheme 1 NH NH k NH2 l b 5 DMSO, NaH J CH,(COOEt), X=OH: 4a X=OMS: 4b kOEt OEt 6 Scheme 2 residue was treated with saturated aqueous solution of Na2S203. The colourless mixture was extracted with diethyl ether and the extract dried over Na2S04. The yellowish solid obtained after removing the solvent (yield 9 1.2%) was used directly in the next step.2-(2-Anthracen-9-ylethyl)malonic acid diethyl ester (6) A suspension of NaH (3 mmol) in anhydrous DMSO (20 ml) was heated at 55 "C and stirred for 1.5 h. After cooling to 0 OC, diethyl malonate (3.3 mmol) was added and the resulting mixture stirred at room temperature for 1 h. A solution of 5 (2.8 mmol) in DMSO ( 5 ml) was added dropwise, then the mixture was kept at 60 "C overnight, cooled to room temperature, poured into water (about 50 ml) and extracted with diethyl ether (4 X 25 ml). After drying the ethereal extract over Na2S04, the solvent was removed with a rotary evaporator, giving a white- yellow solid (yield 57%), which was used without further purification. N,N'-Bis(2-uminoethyl)-2-(2-anthracen-9-ylethyl)- malonamide, ( l b ) (Schemes I and 2) Ethylenediamine (30 ml, freshly distilled over CaO) and 6 (1.5 mmol) were stirred at room temperature under a nitrogen athmosphere for 7 d.Excess of ethylenediamine was evaporated under reduced pressure; on treating the yellow residue with diethyl ether, a pale yellow precipitate formed, which was isolated by in vucuo filtration; yield 74%. Mass spectrometry (70 eV, electron impact ionization, direct introduction, 200 OC): mlz 392 (M+, 68%), 363 ([M -CHNH2]+, 32%), 205 Elemental analysis: found, C 70.12, H 7.31, N 14.61; calculated for C23H2gN402, C 70.38, H 7.19, N 14.27%. ([CI4H&H2CHZ]+, 91%), 188 ([M -C~~H~CH~CHZ]+, 100%). N- {2-[(Anthracen-9-ylmethyl)umino]ethyl- } N'-[2- (4-nitrobenzyl)ethyl]propane-l,3-diamine (3a) (Scheme I ) Compound 2 (I .4 mmol) and 4-ni trobenzaldehyde ( 1.4 mmol) were allowed to react in ethanolic solution (40 ml) for 48 h at room temperature, then NaBH4 (15 mmol) was added portion- wise and the resulting solution was kept at 50 "C for 4 h.Ethanol was distilled under reduced pressure and the residue treated with water (30 ml) and extracted with CH2C12 (4 X 25 ml). The solution was dried over MgS04 and the solvent removed using a rotary evaporator, to give a semi-solid residue which was characterized as its tetrahydrochloride. The hydrochloride (3a.4HCl) was obtained by bubbling gaseous HCl into a CH2C12 solution of 3a and filtering off the pale yellow solid formed. Elemental analysis: found, C 55.80, H 6.13, N 10.51; calculated for C29H39C14N402, C 55.16, H 6.23, N 11.09%. N- {2-[(Anthracen-9-ylmethyl)umino~ethyl} N'-[2 -(4- dimeth ylamino henzylumino)ethyl]propune- I ,3 -diamine (3 6 ) (Scheme 1) The synthesis was performed under the same experimental conditions as for 3a.Mass spectrometry (electrospray ioniza- tion): mlz 484 ([M + HI+, 100%). Apparatus All fluorescence measurements were carried out on a Perkin- Elmer LS-50 luminescence spectrometer equipped with a 1 .O cm quartz cells. Emission spectra at 77 K were measured in dry ethanol by using quartz sample tubes and the same lumines- cence spectrometer equipped with a low-temperature lumines- cence accessory (Perkin-Elmer, Norwalk, CT, USA). UV/VIS spectra were measured on a Hewlett-Packard (Avondale, PA, USA) Model 8452 diode-array spectropho- tometer equipped with 1.0 cm quartz cells.Analyst, December 1996, Vol.121 1765 All pH measurements were made with an Orion 420A (Cambridge, MA, USA) digital pH meter using a combined glass-calomel electrode. Prior to performing measurements in aqueous acetonitrile, the pH scale was calibrated by the Gran method. 15 NMR spectra were obtained on a Bruker (Billeria, MA, USA) AMX-400 spectrometer. Mass spectra were determined on a Finnigan (Sunnyvale, CA, USA) TSQ700 or a Finnigan-MAT (San Jose, CA, USA) Model 8222 instrument. Spectrofluorimetric and Spectrophotometric Titrations Titrations were performed on acetonitrile-water (4 : 1) solutions (50 ml, 5 x 10-4 moll-1) with the pH adjusted to acidic values (G2) by adding small amounts of a standard aqueous solution of HC104 (0.1 or 1 moll-'). Then additions of a standard solution of 0.1 or I moll-1 were made until a basic pH value (3 1 1) was reached.Adsorption and emission spectra (excitation wave- length 372 nm; maximum emission intensities at 415 nm) were taken after each addition of base. Titrations at buffered pH values were carried out on acetonitrile-water (4 : 1) solutions (50 ml, 5 x mol 1-l). Buffered solutions were obtained with 2,6-lutidine (pH 7. l), sodium chloroacetate (pH 2.8) and morpholine (pH 8.0-8.2). M11(C104)2 standard (M = Cu, Ni, Zn) solutions were used and spectra were recorded after every addition. Results and Discussion Sensors Containing Dioxotetramine Receptors The two-component systems l a and l b contain a quadridentate diaminediamide donor set. At a given pH, a metal centre can promote the deprotonation of the two amide groups and induce coplanar chelation (see Fig. 1): the four-coordinating system made by two amine nitrogen atoms and two deprotonated amide-type groups exerts especially strong in-plane metal- ligand interactions.16.17 However, it should be noted that the deprotonation of the amide groups is a particularly endoergonic process which can take place only if outweighed by a large ligand field stabilization upon ion complexation, e.g., for metals late in the 3d series.On the other hand, metals earlier in the series are not able to induce amide deprotonation and are not complexed by amide-containing ligands. Fig. 2 illustrates the fluorescence intensity ( I F ) versus pH profile obtained in the course of a spectrofluorimetric titration of the two-component system l b , H2L (in which the anthracene fluorophore is linked to the dioxotetramine subunit by a -CH2CH2- spacer), in the absence of any metal ion (open circles).The experiment was carried out on an acetonitrile- 2H+ R = H , Fig. 1 (M = Cu, Ni) by a dioxotetraamine chelating agent. Mechanism of complexation mechanism of a divalent 3d metal ion water (4 : 1) solution. Under strongly acidic conditions, the system exists in the form H4L2+, in which both terminal amine groups are protonated and the anthracene fragment fully displays its emission band. On addition of standard base, progressive deprotonation of the ammonium groups takes place, but the H3L+ and HZL species keep their full fluorescence. The amine group displays reducing tendencies in photochemistry and can quench an excited anthracene subunit, An*, through an N-to-An* eT process.This does not happen in the present system, probably because of the large distance between the two primary $mine nitrogen atoms and the anthracene fragment (7.5 and 8.5 A to the C-9 atom of anthracene, as evaluated through molecular modelling). When the titration is carried out on a solution that contains also 1 equiv. of NiI1, the intensity of the fluorescent emission decreases according to a sigmoidal profile centred at pH 8.5 until complete quenching occurs (see Fig. 2, open triangles). Quenching has to be related to the metal complexation by the dioxotetramine subunit of l b . In fact, the sigmoidal I F versus pH decrease is accompanied by a sigmoidal increase in an absorption band centred at 450 nm, as shown by parallel spectrophotometric titration experiments. The d-d band corre- sponds to the yellow low-spin, square-planar complex [NiT1(L2-)].If Ni" is replaced with Cu" in the spectrofluorimetric titration experiment, a sigmoidal decrease in 1, until complete quenching is again observed, but the process is centred at pH 6.8 (see Fig. 2, diamonds). In this case also, a concomitant sigmoidal increase in a ligand field absorption band is observed. The band is centred at 520 nm and corresponds to the pink-violet square- planar complex [Cu"(L2-)]. In contrast, titration experiments involving divalent metal ions earlier in the transition series (Mn", Fell, Co") do not induce any modification of the fluorescent emission.This behaviour is rationalised by the inability of the metal to promote amide deprotonation and binding. Zinc(II), which cannot profit from ligand field effects, because of its dlo electronic configuration, does not induce amide deprotonation and therefore does not cause any IF modification in the titration experiment. Thus, the two-component system l b is able to recognize Nil1 and Cu" among all the other transition metals, identifying not the size of the metal, but its position in the Periodic Table. Analytical discrimination between Nil1 and Cu" cations can be achieved by utilizing the separation of the two I F versus pH profiles of the spectrofluorimetric titrations (about 1.7 pH units, see Fig. 2). In this connection, a solution of l b was adjusted to pH 8 with the morpholine buffer.At this pH value, which is 0 0 60- -0 2 4 0 - v 20 - 0 - 4 6 8 10 12 PH Fig. 2 pH dependence of the fluorescence intensity of MeCN-H20 (4 : 1) solutions containing: (i) l b (0); (ii) l b and equimolecular amounts of Nil* (0); and (iii) l b and equimolecular amounts of Cu" (0).1766 Analyst, December 1996, Vol. 12 I 60 s so 40 20 0 W intermediate between the two profiles, the less stable Ni" complex does not form, but the Cu" complex does. In fact, titration of l b with a standard Ni" solution did not affect fluorescence at all (see Fig. 3, triangles), whereas subsequent titration with Cu" (diamonds) induced a linear decrease in I F . Fluorescence quenching was observed after the addition of 1 equiv. of Cull. The previously investigated system la,I4 which differs from l b for the shorter spacer linking the dioxotetramine subunit and the anthracene fragment (-CH2- rather than -CHZCH2-), displays an analogous behaviour to lb: IF versus pH profiles are similar to those reported in Fig.2, but they are shifted to more acidic pH values, being centred at pH 5.9 in the case of Cu" and pH 7.4 in the case of Nili. The titration profiles ( I F versus n) obtained at buffered pH (7.1) superimpose on those reported in Fig. 3. The substantially similar behaviour is not surprising since the receptor component is the same. In addition, the modification of the linker does not alter the way the interaction is communicated to the fluorophore. Thus, la and l b are efficient fluorescent sensors for the Cur' ion, displaying unique sensing features which are related to the selective binding tendencies of the dioxotetramine subunit and to the powerful signalling proper- ties of the anthracene subunit.The nature of the signal transduction mechanism (whether eT or ET) ought to be made clear. In this connection, it should be noted that coordination by deprotonated amide (or peptide) groups favours access to the otherwise uncommon Nil1' and CulU states,'*.lg thus making a metal-to-fluorophore eT process feasible. In particular, the AGOeT value associated with the An* + MI' -+ An- + MI1' process is distinctly negative for both nickel (-0.35 eV) and copper (-0.5 eV). (AGOeT is calculated from the equation A G o c ~ = -EO-OA~* + FE'C~III/C~II -FEOAn/An-, where EO-O is the spectroscopic energy, obtained from the emission spectrum, and Eo is the electrode potential for the pertinent redox change, which can be obtained from the E1/2 value determined through cyclic voltammetric experiments on the separated components.) Moreover, the eT nature of the room temperature quenching mechanism is demonstrated by the fact that, when a solution of either the Nil1 or Curl complex is frozen at 77 K, the fluorescence of the anthracene fragment is fully restored.Immobilization in a glass prevents the rearrangement of the solvent molecules, thus raising the energy of the ion pair which forms following the eT process { An--[M"'(L2-)]+, in ' 801 0 0 0 1 I I I I I I 0.0 0.5 1.0 1.5 2.0 n Fig. 3 Discrimination of Cu" and Ni" by the fluorescent sensor l b in MeCN-H20 solution (4: l), adjusted to pH 8 with the morpholine buffer.The fluorescence intensity IF is not altered during the titration with Ni" (V), but decreases linearly on addition of Cu" (o), reaching almost 0% of the original value after addition of 1 equiv. n = Number of equivalents added. the present instance } . In these circumstances, the strongly thermodynamically disfavoured eT process does not take place and fluorescence is not quenched. In conclusion, the diamine-diamide chelating subunit of the two-component systems l a and l b displays interesting features for sensing purposes: (i) the endoergonic deprotonation of the two amide groups results in selective complexation among 3d metals; and (ii) strong in-plane interactions exerted by the dianionic tetraaza donor set promote one-electron oxidation of the metal centre and induce the transfer of an electron to the photoexcited fluorophore, thus providing an efficient signal transduction mechanism.Sensors Containing Tetramine Receptors Among saturated tetramines, 1,4,8,11 -tetraazaundecane (2.3,2-tet), gives the most stable complexes with transition metals.20.2' This may be due to the favourable alternating sequence of the chelate rings (5,6, S), the same sequence as for the dioxotetramine subunit of sensors l a and lb. When compared with other tetramines, 2.3.2.-tet is able to put its donor atoms in the positions required by the transition metal centre, i.e., at the corners of a square, according to a strain-free arrangement. Hence, the 2.3.2-tet subunit was chosen as a receptor and the signalling subunit, an anthracene fragment, was appended to one of its terminal amine groups through a -CHZ- linker, to give the two-component system 2.In the spectro- fluorimetric titration of a solution of 2 containing excess acid, a progressive decrease in the fluorescence intensity was observed after the addition of 1 equiv. of standard base (after that all the excess acid had been neutralized). The decrease in I F began at pH 3.5 (see Fig. 4, open triangles) and complete fluorescence quenching was observed at pH 10 (after the addition of 3 equiv. of base). Quenching can be ascribed to an eT process from the two secondary nitrogen atoms closest to the fluorophore (the closest one lies at 2.4 and the next at 5.9 A, as estimated from molecular modelling).In presence of 1 equiv. of either NiIr or Cu", fluorescence quenching was observed at distinctly lower pH values (see Fig. 4, open diamonds and circles, respectively) and took place according to well defined sigmoidal profiles. Again, the Cu" profile is centred at a pH value about 2 units lower than that of Nil', which reflects the greater stability of the CuI1 tetramine complex. This is a well known phenomenon in coordination chemistry and is expressed by the Irving-Williams sequence.22 In this case also, the separation of the two I F versus pH profiles (see Fig. 4) allowed metal discrimination, as I"Il 80 L I I I I 1 I 2 4 6 8 10 12 PH Fig. 4 pH dependence of the fluorescence intensity IF of MeCN-H20 (4 : 1) solutions containing: (i) 2 ( V); (ii) 2 and equimolecular amounts of Ni" (0); and (iii) 2 and equimolecular amounts of Cu" (0).Analyst, December 1996, Vol.121 1767 indicated by the consecutive titration of a solution of 2, buffered at pH 2.8, with Nil1 (no effect on ZF) and then with Cu" (linear decrease and full quenching after the addition of 1 equiv.). Metal-induced fluorescence quenching should be ascribed in the present case to an ET rather than an eT mechanism. In fact, on freezing at liquid nitrogen temperature, a solution of the [MII(2)]2+ complex (M = Ni, Cu), no regeneration of fluorescence was observed. It should be noted that transition metals possess empty or half-filled levels of low energy (d orbitals) suitable for a double electron exchange energy transfer mechanism (Dexter type). This electronic ET process, which does not induce any separation of charge and is not affected by solvent immobilization, also operates in a glass.It should be noted that both an M"-to-An* and An*-to-M" eT process are thermodynamically allowed for nickel and copper. Predom- inance of the ET over the eT mechanism can be tentatively related to the metal-to-fluorophore distance. It has been shown23 for a homogeneous series of two-component systems that at short distances the ET process prevails, whereas at longer distances it is the eT mechanism that redominates. Notably, the which fluorescence is quenched through an eT mechanism, the MIi-C-9 distances are considerably greater, 5.4 and 6.8 A, respectively . Whichever process occurs, the displayed signal transduction mechanism is very efficient and makes systems 1 and 2 powerful sensors for transition metals.MI1-C-9 distance in system 2 is 3.5 R . In systems l a and lb, in Sensing of the Zn" Ion Zn" (electronic configuration d10) is not a genuine transition metal ion: it does not profit from ligand field effects and, in the presence of a quadridentate ligand, it prefers to adopt a tetrahedral rather than a square-planar coordinative arrange- ment. Moreover, Zn" is photophysically inactive, as (i) it does not display any one-electron redox activity and cannot be involved in any eT mechanism; and (ii) having a completely filled d level, it cannot participate in any electronic ET process. However, Zn" can exert an indirect effect on the emitting activity of a proximate fluorophore, which may lead to fluorescence regeneration or enhancement.Such an effect, which was first documented with the complexation of two Zn" ions by a difunctional anthrylamine,3,4 is clearly illustrated by the titration of system 2 in the presence of 1 equiv. of Zn", whose IF versus pH profile is illustrated in Fig. 5. I 100 80 2? 60 P 3 h W 40 20 0 00 cc93 0 0 1 I I I 1 I I 2 4 6 8 10 12 PH Fig. 5 pH dependence of the fluorescence intensity IF of 2 (V j, 3a (0) and 3b (Oj in the presence of 1 equiv. of Zn" in MeCN-H20 (4: 1) solutions. It can be seen (open triangles) that fluorescence begins to decrease at pH 3.5. At pH 4.7, IF reaches a minimum (55% of the original fluorescence), then it increases again to reach a plateau.The incipient fluorescence quenching (pH range 3.5-4.7) can be ascribed to an eT process from the proximate amine groups which deprotonate after the addition of the first equivalent of standard base. At pH 4.7, the [ZnII(2)]2+ metal complex begins to form: the lone pairs on the amine nitrogen atoms become involved in the coordinative interactions and are no longer available for the eT process. At pH 6 (beginning of the plateau), 100% of the [ZnI1(2)]2+ complex has formed and fluorescence is fully restored. Both N?' and Cu" exert a negative effect on the emission properties of an adjacent fluorophore and signal their presence through the fluorescence quenching. Zinc(r1) has a positive effect: when interacting with an anthrylamine chelating agent, it stops the N-to-An* eT process and induces fluorescence.Fig. 6 shows the the variation of IF observed when a solution of 2, adjusted to pH 8.1, is titrated with a standard solution of Zn". ZF increases linearly to reach its limiting value (full restoration of fluorescence) after the addition of 1 equiv. of metal. Some further interesting effects take place when a substitu- ent, displaying either electron donor or acceptor tendencies, is appended at the other peripheral amine nitrogen atom of the tetramine subunit of system 2. Such is the case with the three- component system 3a, in which the substituent is the classical donor fragment nitrobenzene (NB). Fig. 5 shows the I , versus pH profile corresponding to the titration with standard base of a solution containing equimolar amounts of 3a and Zdl, plus excess acid (open diamonds).Fluorescence intensity begins to decrease at pH 3.7, owing to the occurrence of the N-to-An* eT process, as discussed previously. However, at pH 4.7, fluores- cence does not stop decreasing, as observed with system 2, but continues to decrease to reach less than 40% of its original intensity. We ascribe this behaviour to the fact that, on coordination of the tetramine subunit to the metal, An and NB fragments are brought to a distance apart small enough to allow an intramolecular eT process from An* to NB to take place. The real occurrence of such a process, which is favoured from a thermodynamic point of view (AGOeT = -0.5 eV), is demonstrated by the fact that on freezing the solution, the anthracene fluorescence is fully restored.A similar behaviour is observed (see Fig. 5, open circles) when the tetramine bears an electron donor substituent, NJ- dimethylaniline (DMA), 3b. An eT process (DMA-to-An*) takes place in this case also, the behaviour being accounted for on a thermodynamic basis (AGOeT = -0.4 eV).Z4 Note that in 500 400 h 300 2 : 200 100 A a A A A A A A 0 1 2 nz"" Fig. 6 Titration profiles obtained by adding Zn" to MeCN-H20 (4: 1) solutions of 2 (A>, 3a (O), or 3b (0) buffered at pH 8.2 with morpholine.1768 Analyst, December 1996, Vol. 121 the present case quenching of the excited fluorophore is more efficient than for 3a, as the fluorescence intensity is reduced to less than 20% of the original intensity. Systems 3a and 3b could hardly be proposed as fluorescent sensors for Zn".In particular, when a solution of 3a, adjusted at pH 8.2, is titrated with Zn" (circles in Fig. 6), only a moderate increase in fluorescence is observed in the range 0-1 equiv., then a plateau is reached. The fluorescence increase is even less pronounced when the titration is carried out on a solution of 3b owing to the more efficient quenching exerted by the DMA fragment (diamonds in Fig. 6). However, the two profiles described above are interesting from a mechanistic point of view: between 0 and 1 equiv. of ZnII, two different eT mechanisms (N-to-An* and An*-to-NB or DMA-to-An*) contribute, to a variable extent, to the quenching of the excited fluorophore. After 1 equiv., only the latter mechanism oper- ates. The mechanism of the intramolecular electron transfer process to/from the acceptor/donor substituent is shown in Fig.7. At low pH, the three-component system is fully protonated and is assumed to adopt a stretched arrangement. The distance between the nitrogen atom of the donor/acceptor substituent and the S-9 atom of the anthracene subunit can be estimated as 20.6 A. This distance is drastically reduced on metal coordination. Molecular modelling for a tetrahedral arrangement (one of the coordiqation mode preferred by Zn") gives an N-C-9 distance of 5.1 A. On the other hand, if a five- coordinate arrangement is assumed, with a water molecule occupying one of the equatorial positions of a trigonal bipyramid, the distance is 7.4 A. These values can be further Fig. 7 Suggested mechanism of the intramolecular eT process in the [Zn11(3b)]2+ system.The square coordinative arrangement of the tetraamine fragment around the Zn" centre has been drawn for the sake of simplicity. The Zn" stereochemistry is five-coordinate (with a water molecule completing the coordination polyhedron) or, more likely, tetrahedral. reduced on rotation of the -CH2- linker and, in any case, allow a through-space eT process to take place. Thus, in presence of bifunctional receptors of type 3, the photophysically inactive ZnII ion is able to exert, even if indirectly, a negative effect on the emitting properties of a proximate fluorophore. This work was supported by the Italian National Council of Research (CNR, Progetto Strategic0 'Tecnologie Chimiche Innovative'). References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Fabbrizzi, L., and Poggi, A., Chem. Soc.Rev., 1995, 197. Fluorescent Chemosensors for Ion and Molecule Recognition, ed. Czamik, A. W., ACS Symposium Series, no. 538, American Chemical Society, Washington, DC, 1993. Huston, M. E., Akkaya, E. U., and Czamik, A. W., J . Am. Chem. Soc. 1989,111, 8735. Akkaya, E. U., Huston, M. E., and Czamik, A. W., J. Am. Chem. Soc., 1990,112, 3590. Konopelski, J.-P., Kotziba-Hibert, F., Lehn, J-M., Desvergne, J.-P., Fages, F., Castellan, A., and Bouas-Laurent, H., J . Chenz. Soc., Chem. Commun., 1985,433. Fages, F., Desvergne, J.-P., Bouas-Laurent, H., Marseau, P., Lehn, J.- M., Kotziba-Hibert, F., Albrecht-Gary, A.-M., and Al-Joubbeh, M., J . Am. Chem. Soc., 1989,111, 8672. Fages, F., Desvergne, J.-P., Bouas-Laurent, H., Lehn, J.-M., Kono- pelski, J.-P., Marseau, P., and Barrans, Y. J., J . Chem. Soc., Chem. Commun., 1990,655. Fages, F., Desvergne, J.-P., Kampke, K., Bouas-Laurent, H., Lehn, J.- M., Meyer, M., and Albrecht-Gary, A.-M., J . Am. Chem. Soc., 1993, 115, 3658. Bissell, R. A., de Silva, A. P., Gunaratne, H. Q. N., Lynch, P. L. M., Maguire, G. E. M., McCoy, C. P., and Sandanayake, K. R. A. S., Top. Curr. Chem., 1993, 168, 223. Aoki, I., Harada, T., Kawahara, Y., and Shinkai, S., J. Chem. Soc., Chem Commun., 1992, 1341. Parker, D., and Williams, J. A. G., J . Chem. Soc., Perkin Trans., 1995, 1305. Barefield, E. K., Wagner, F., Herlinger, A. W., and Dahl, A. R., Inorg. Synth., 1975, 16, 220. Fabbrizzi, L., Licchelli, M., Pallavicini, P., Perotti, A., and Sacchi, D., Angew. Chem., 1994,106,2051; Angew. Chem., Int. Ed. Engl., 1994, 33, 1975. Fabbrizzi, L., Licchelli, M., Pallavicini, P., Perotti, A., Taglietti, A., and Sacchi, D., Chem. Eur. J . , 1996, 2, 167. Gran, G., Analyst, 1952, 77, 661. Fabbrizzi, L., Forlini, F., Perotti, A., and Seghi, B., Inorg. Chem., 1984, 23, 807. Kodama, M., and Kimura, E., J . Chem. Soc., Dalton Trans., 1979, 325. Fabbrizzi, L., Perotti, A., and Poggi, A., Inorg. Chem. 1983, 22, 1411. Kodama, M., and Kimura, E., J . Chem. Sol-., Dalton Trans., 1981, 694. Weatherburn, D. C., Billo, E. J., Jones, J. P., and Margerum, D. W., Inorg. Chem., 1970, 9, 1557. Paoletti, P., Fabbrizzi, L., and Barbucci, R., Inorg. Chem., 1973, 12, 1861. Irving, H., and Williams, R. J. P., J . Chem. Soc., 1953, 3192. Closs, G. L., Johnson, M. D., Miller, J. R., and Piotrowiak, P., J . Am. Chem. Soc., 1989, 111, 3751. Fabbrizzi, L., Licchelli, M., Pallavicini, P., and Taglietti, A., Inorg. Chem., 1996,35, 1733. Paper 6104261 I Received June 18,1996 Accepted July 23, 1996

ISSN:0003-2654    

DOI:10.1039/AN9962101763

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 (1 769-1 773) 1769 Direct Monitoring of Formaldehyde Vapour and Detection of Ethanol Vapour Using Dehyd rogenase-based Biosensors* Manus J. Dennison, Jennifer M. Hall and Anthony P. F. Turner? Crunfeld Bioterhnology Centre, Cmnfield University, Crurlfield, Bedfordshire, UK MK43 OAL Biosensors capable of directly detecting low levels of formaldehyde and ethanol vapour were constructed. Both biosensors are based on dehydrogenase enzymes which produce reduced nicotinamide adenine dinucleotide as part of the oxidation of formaldehyde and ethanol. The enzymes were immobilized in a reverse micelle medium which did not dehydrate significantly over time, and allowed direct gas-phase monitoring. A screen-printed electrode was used as transducer. Formaldehyde and ethanol vapour partitioned into the reverse micelle media, where it was acted upon by the relevant enzyme.Reduced nicotinamide adenine dinucleotide was oxidized at the working electrode at a potential of 800 mV versus an Ag/AgCl reference electrode. Formaldehyde could be measured over the concentration range 1 ppb-1.3 ppm and ethanol could be detected over the range 50-250 ppm. Keywords: Formaldehyde vapour; ethanol vapour; biosensors; dehydrogenase Introduction Gas-phase sensing has been dominated by non-biological sensors, such as electrochemical, semiconducting and pellister- type sensors. Amperometric and potentiometric gas sensors are generally limited to a narrow range of electroactive gases.' Semiconducting gas sensors suffer from a lack of specificity and have high power consumption demands.There exists a need for gas sensors with low power consumption and which are selective for unreactive gases and vapours. Biosensors have the advantage of high inherent selectivity but to date have been mainly confined to aqueous media and certain organic solvents. Very few biosensors have been demonstrated for direct gas- phase monitoring. Enzymes have been shown to operate in organic solvents and polyphenol oxidase in glycerol formed the basis for a previous gas-phase biosensor.2 Reverse micelles (RMs) provide one method for immobilizing enzymes in organic liquids. RMs are aggregates of surfactant molecules that form when the concen- tration of surfactant in an organic solvent exceeds a critical value.3 Enzyme and water molecules are trapped within RMs to form swollen micelles or micr~emulsions.~ The fact that the enzyme is in an aqueous micro-environment in a bulk organic phase means that the enzyme is less likely to suffer denaturation by the organic liquid.Formaldehyde is an important industrial chemical, being extensively used in the manufacture of resins and is also used as a bactericide and preservative in goods as diverse as toothpaste, washing-up liquid and air fresheners.5 * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March + To whom correspondence should be addressed. 25-29, 1996. The widespread use of formaldehyde in building materials has given rise to reported indoor formaldehyde concentrations of up to 0.37-0.55 ppm in new homes6 and between 0.12 and 1.6 ppm in mobile homes.' Most countries now have guidelines recommending indoor formaldehyde concentrations of not higher than 0.1-0.2 ppm.8 There exist a number of methods concerning the non- biological determination of formaldehyde vapour such as chemical method^.^ gas chromatography,10 semiconductors1 and piezoelectric crystal based sensors.12 These can detect formaldehyde at the ppm and ppb levels; disadvantages, however, can include preconcentration steps, time delays and a lack of selectivity.Formaldehyde dehydrogenase (EC I .2.1.46) from Pseudo- monas putidu catalyses the oxidation of formaldehyde in the presence of its cofactor NAD+: FDH NAD+ + HCHO + H,O-HCOOH + NADH + H+ Formaldehyde dehydrogenase (FDH) will catalyse the dehy- drogenation of formaldehyde and acetaldehyde, but it is inactive towards longer chain aldehydes.The enzyme will also act on butan-1-01, pentan-1-01 and hexan- 1-01, but is completely inert towards methanol and ethano1.l' Under normal conditions, formaldehyde exists as a vapour and this presents an additional problem for biosensing, as most biosensors operate in the liquid phase. To date there has been only one publication concerning a gas-phase biosensor for the direct determination of formaldehyde vapour, 14 although Drae- ger have launched a biosensor-based measurement device. I s Enzymic assays for formaldehyde in the aqueous phase based on spectrophotometric methods, ' 6 fluorimetric methods 17, and amperometric methods19 have been described. GuilbaultI4 described an enzyme-coated piezoelectric crystal detector for the determination of formaldehyde.FDH (EC 1.2.1. 1), which uses both glutathione and NAD+ as a cofactor, was used in conjunction with a piezoelectric crystal. On exposure to formaldehyde vapour, FDH bound formaldehyde, causing a change in frequency at which the piezoelectric crystal oscillated. This biosensor gave a linear response over the range 10 ppb-10 ppm formaldehyde at 50% relative humidity (RH) and the lifetime was 3 or 100 analyses. The fact that ethanol is involved in many manufacturing systems necessitates the determination of ethanol concentra- tions in liquid and gas phases. The permissible legal limit of blood ethanol is 800 mg-1 for driving in the European Union,20 and an attractive method of assessing blood ethanol concentra- tion is the measurement of ethanol in a person's breath.Alcohol dehydrogenase (ADH) (EC 1 . 1 . 1 . 1 ) catalyses the oxidation of alcohols to aldehydes without the requirement for molecular oxygen: ADH CH,CH,OH + NAD' -----+CH,CHO + NADH + H+1770 Analyst, December 1996, Vol. 121 Yeast ADH (E.C. 1.1.1.1) (YADH) oxidizes all primary straight-chain alcohols, but displays poor activity towards secondary and branched-chain alcohols.21 YADH has also been reported to catalyse the oxidation of formaldehyde.22 YADH has a K,n of 21 mmol 1-1 for ethanol.23 Most publications concerning biosensors for monitoring ethanol use either alcohol oxidase (AOX) or ADH as the biological component, with either amperometry or spectro- photometry used as the transduction mechanism.The con- sumption of molecular oxygen or production of hydrogen peroxide due to the AOX-catalysed oxidation of alcohols can be monitored electrochemically,24~25 while the production of NADH due to the ADH-catalysed oxidation of alcohols can be monitored electrochemically2~~27 or spectrophotometrically . Matuszewski and Meyerhoff28 reported a biosensor based on the detection of hydrogen peroxide, produced by the action of AOX on ethanol vapour which had partitioned into an aqueous s y s tem. Mitsubayashi et ul.29 described an enzyme electrode using AOX immobilized in an acrylamide membrane and retained at the surface of a Clark-type oxygen electrode. The enzyme electrode was kept hydrated by means of a circulating buffer system connected to a reservoir.Park et ~ 1 . 3 0 described an enzyme electrode using ADH/ NAD+ immobilized in hydroxyethylcellulose for monitoring ethanol vapour. The NADH produced was monitored amper- ometrically at 650 mV versus Ag/AgCl. Before use, the enzyme electrode was activated by dipping in buffer. The main problem with using enzyme electrodes for monitoring alcohol vapour is water loss, which seriously affects the enzyme performance. One other related problem is the stability of the enzymes. ADH and AOX are both reported to have poor stability, water having a significant positive effect on the initial enzyme activity, but having a significant detrimental effect on long-term enzyme stability.31 There are a variety of methods which can be used to detect ethanol vapour, such as conductimetric methods,32 electro- chemical fuel cells,33 infrared absorption34 and gas chromato- graphy.35 Experimental Reagents Alcohol dehydrogenase (EC 1.1.1.l), 240 U mg-1 solid, from baker’s yeast, formaldehyde dehydrogenase (EC 1.2.1.46), 5.4 U mg- 1 solid, from Pseudomonus putida, nicotinamide adenine dinucleotide (oxidized form), nicotinamide adenine dinucleo- tide (reduced form) and Tween 80 (polyoxyethylene sorbitan monooleate) were purchased from Sigma (Poole, Dorset, UK). BDH (Poole, Dorset UK) suppled silicone oil. Acheson Colloids (Plymouth, UK) supplied carbon polymer thick-film ink and silver/silver chloride ink. Electrode Construction Electrodes consisted of three layers, each of which was screen printed (Fig. 1). The first was a layer of conducting silver/ silver chloride ink which was printed as two tracks per electrode on a PVC sheet.A highly conducting graphite polymer ink was printed on top of a portion of one silver/silver chloride conducting track at the opposite end to the electrical contact area. This provided a carbon working electrode. The final layer was an insulating ink which defined the electrical contacts, and an area opposite the contacts, where electrochemistry was performed. This electrochemical area consisted of a carbon working electrode side by side with a silver/silver chloride electrode. One of the problems encountered with reverse micelle gels was due to the fact that, because of the hydrophobicity of the silicone oil, the reverse micelle gel would spread beyond the defined area of the electrode.The insulation ink used to define the electrodes area is highly hydrophobic, thus encouraging the spreading of the silicone oil. This spreading produced a very thin gel with a large surface area, encouraging the evaporation of water from the gel. In this case it was necessary to build a ‘well’ around the electrode layer to prevent the enzyme/micelle medium moving on the electrode. This was accomplished by painting a layer of Tipp-Ex correcting fluid around the electrode area, taking care to avoid the working or reference electrode surface. Biosensor Construction ADH was immobilized in RM medium by dissolving Tween 80 (1% v/v) in silicone oil and then adding a solution of ADH (3400 U per 100 pl) and NAD+ (5 mg per 100 pl) dissolved in 0.1 moll-1 sodium phosphate buffer with 5 mmoll-l KCl, pH 8.0 (20% v/v).The buffer: Tween 80 : silicone ratio was set at 20 : 1 : 79 (v/v). This mixture was then vigorously shaken to mix the components thoroughly. This RM medium (6 pl) was then deposited on the electrode area of a screen-printed electrode. FDH was immobilized in RM medium exactly as described for ADH, except that the concentration of FDH was 29 U per 100 1.11. Gas Rig A gas rig capable of generating phenol, ethanol or formaldehyde vapour at different RH was constructed (Fig. 2). Formaldehyde permeation tubes or an ethanol diffusion vial (Vici Metronics., Santa Clara, CA, USA) which permeate formaldehyde or ethanol respectively, at a rate which is temperature dependent, were sealed in an air-tight glass U-tube and immersed in an oil- bath at a constant temperature. Low-humidity air was then passed through the U-tube over the permeation tubes/diffusion vial at a controlled flow rate.This yielded low-humidity air containing the relevant vapour, which was then mixed with air that had been humidified by passing it through a Drechsel bottle, containing water, generat- ing air containing the vapour at the required concentration and RH. Two three-way valves were used so that the low RH input air could be switched between clean air and air containing vapour without affecting humidity or flow rate. All interconnecting tubing on the gas rig was short PTFE-lined tubing (Aldrich, Gillingham, Dorset, UK). A thermostated glass chamber with a volume of 25 ml was used as the biosensor test chamber.Four outlets in the test chamber allowed for placement of a temperature and relative 11 1 . . 1 I-+------ . insulation ~ayer Fig. 1 Diagram of screen-printed electrode construction.Analyst, December 1996, Vol. 121 1771 humidity probe (Vaisala HM34 relative humidity and temper- ature meter, RS Components, Corby, UK) the biosensor and an inlet and an outlet for vapour/air. All work was carried out in a fume-hood. Calibration of Gas Rig Two methods were used to calibrate the permeation rate of ethanol vapour: (1) measuring the permeation rate by monitor- ing liquid ethanol loss gravimetrically over time; and (2) measuring the permeation rate by using Draeger tubes. The permeation rate of formaldehyde vapour was calculated using Draeger tubes to measure directly the concentration of formaldehyde being emitted at point B in the gas rig (Fig.2). Apparatus and measurement All electrochemical measurements were carried out using an Autolab Pstat 10 electrochemical analyser (EcoChemie, Utrecht, The Netherlands). Procedure A schematic diagram of the gas rig is shown in Fig. 2. The valves were set initially so that no vapour, but only air of the same humidity as the test vapour, was flowing through the test chamber. The biosensor was connected to the leads from the Autolab using a pair of microclips and inserted into the test chamber. The biosensor was made secure by means of a threaded cap, which had a hole in the centre to allow for passage of the connecting leads. A potential of +800 mV was applied between the working electrode and the combined counter and reference electrode (CC + RE), and after the current had reached a steady-state background level, the valve system was switched so that test vapour was flowing past the biosensor. After a set exposure time, the valves were switched again so that only air was flowing past the biosensor.The response was evaluated by calculating the difference between the baseline current and the amperometric response. The RH, flow rates and temperature were monitored regularly. As the biosensors were designed to be disposable, each biosensor being exposed only once. High Low hmldlty hmldtty a if I I Te s t Chamb e r containin9 biosensor & temp. & Hum, meter 1 0 Manometer F lowmet er I\ / 3-way valve b One-way valve Glass beads for heat exchange PTFE llned 1 tublng Phenol Fig.2 concentrations of vapours. Arrows refer to direction of air flow. Schematic diagram of gas rig constructed for generating different Results and Discussion Determination of Operating Potential NADH has previously been directly oxidized at a platinum electrode at 600 mV36 and at a gold electrode at 750 mV37 versus Ag/AgCl. Experimental work demonstrated that in- creases in anodic current over the potential range 700-1000 mV were observable on addition of NADH to aqueous solutions or to silicone oil-based reverse micelles. An operating potential of 800 mV versus Ag/AgCl was chosen. Formaldehyde Biosensor Exposure to formaldehyde vapour after the background current had reached a steady-state value resulted in large increases in anodic current.Fig. 3 shows a typical amperometric response of the formaldehyde biosensor on exposure to formaldehyde vapour. Exposure to organic solvents (chloroform, acetone and methanol) produced only negligible responses. Steady-state currents were reached 2-10 min after initial exposure to formaldehyde vapour. Calibration Curve for Formaldehyde Biosensor Fig. 4 shows a calibration curve for the formaldehyde biosensor. Steady-state currents were taken as a measure of biosensor response. The biosensor shows a linear response to formaldehyde vapour over the concentration range 1.3 ppb-1.2 ppm with a slope of 0.412 nA ppb-1. Amperometry of ADHlNAD+ in Reverse Micelles Fig. 5 shows the amperometric response of an ADH/NAD+ biosensor on exposure to ethanol vapour. The biosensor shows ri I 0.5 0.4 a 5 0.2 ;=' Om3 a, 0 0.1 0.0 0 400 800 1200 1600 Time/s Fig.3 Amperometric response of formaldehyde biosensor on exposure to 1.23 ppm formaldehyde. Poised potential, 800 mV versus Ag/AgCl. Conditions: 25 "C, 55% RH. Arrows indicate period of exposure. 700 I 1 0 400 800 1200 Formaldehyde concentration (ppb) Fig. 4 800 mV versus Ag/AgCl. Conditions: 25 "C, 50-57% RH. Calibration curve for formaldehyde biosensor. Poised potential,1772 Analyst, December 1996, Vol. 121 a a good response to ethanol vapour after approximately 8 min. Exposure to organic solvents (chloroform and acetone) pro- duced only negligible responses. T Calibration Curve for Ethanol Biosensor Fig. 6 shows a calibration curve, steady state responses to ethanol vapour being used as a measure of biosensor response to ethanol.The response appears to be linear up to approximately 250 ppm ethanol. Saturation of the biosensor response seems to occur above ethanol concentrations of 250 ppm. This ethanol vapour concentration (equivalent to 0.004-0.02 mmol 1-1) is much less than the K, of ADH for ethanol. The fact that saturation occurs at ethanol vapour levels less than the reported K, values seems to indicate that the concentration of ethanol in the biosensor gel is much higher than that of the ethanol vapour, if indeed the biosensor response is limited by the K , of ADH. It would seem that ethanol concentrates in the gel phase. The fact that the Henry's constant for ethanol is approxi- mately 2000 suggests that ethanol vapour would partition into the liquid phase up to 2000 times more concentrated.This does indeed seem to the case, since when the ethanol vapour concentration (0.004-0.02 mmol 1-1) is multiplied by the Henry's constant, a value of 8-40 mmol l-1 is obtained. These values are close to the reported values of K, for ethanol (16-21 mmol 1-I), which would explain the saturation of the biosensor response at these concentrations. Stability of Biosensors The stability of the biosensor can be divided into individual components: enzyme stability and water retention ability of the 0.5 i \I i ' l ' l ' l ' l ' 0 500 1000 1500 2000 Tim e/s Fig. 5 pprn ethanol. Conditions as in Fig. 2. Amperometric response of ethanol biosensor on exposure to 92 s a, 0 Q 1.5 E 0 1.0 .- L c E E 2 0.5 a, a 0.0 0 250 500 750 1000 1250 Ethanol concentration (ppm) Calibration curve for ethanol biosensor. Poised potential, 800 mV Fig.6 versus Ag/AgCl. Conditions: 25 "C, 50-57% RH. gel. Enzyme stability in the gel was investigated by measuring the response of biosensors prepared from gels stored at 4 and 25 "C. There was no decrease in activity up to 60 h when the gel was stored at 4 "C, no decrease in activity was observed in the gel stored at 25 "C for up to 24 h and FDH was stable for at least 12 h at room temperature. The stability of the biosensor under operating conditions (25 "C, 50% RH), as opposed to storage conditions, was measured by ageing the biosensors and then exposing them to ethanol vapour. As ADH was stable in the gel (when sealed to prevent water loss) for more than 24 h, a decline in response indicated water loss.The issue of water activity is crucial for gas-phase biosensors. The prevention of water loss from gas-phase biosensors is a topic which has concerned virtually all workers using gas-phase biosensors. In reverse micelles, water loss seems to be prevented primarily by the large proportion of silicone oil which acts as a barrier to evaporation. The water loss, as a function of time, of reverse micelles and a buffer control was measured gravimetrically. After 90 min RMs lost only approximately 5% of their water content, yet the biosensor activity declined by approximately 75%. The activity of biosensors prepared with RMs seems to be very sensitive to the water content of the RMs.The stability of YADH in buffer is reported to be poor: 20% activity losses were reported after 50 min at 17.5 0C.38 The stability of YADH in RMs would be expected to be comparable, as the enzyme is in an aqueous micro-environment. Lee and Biellmann39 reported that RMs formed with ADH from T. brockii were more stable in non-ionic surfactant-based RMs (such as Tween 80) than in anionic/cationic surfactant- based RMs. They reported that ADH in RMs based on non-ionic surfactants lost approximately 20% of its activity after 4 h at 15 "C. Lee and Biellma~m~~ also drew similar conclusions from a study of horse liver ADH in non-ionic surfactant-based RMs. They reported a 50% decrease in horse liver ADH activity over 4 d. The results presented here appear to agree with the reported poor stability of YADH.The fact that no activity loss seemed to occur after 24 h at 25 "C with this system seemed to be contrary to the very poor reported stability of YADH.38 However, this discrepancy can be explained if an excess of YADH was present in the RMs at the start ('enzyme loading') and activity losses were only noticed when the concentration of active enzyme fell below a critical level. Previous studies indicate that this water content might be near the optimum for ADH activity. Lee and Biellmann40 optimized RMs with horse liver ADH with a water content of 10% v/v. Kawakami et ~ 1 . ~ ~ found that ADH activity was maximum at a water content of 20%, while Larsson et al.42 found that a water content of approximately 12% v/v gave maximum activity for horse liver ADH in RMs.These studies indicate that between 10 and 20% v/v water content is optimum for ADH activity. Conclusion Most hydrophilic solvents are very hostile environments for The use of RMs provided an alternative method of immobilizing enzymes in organic liquids and as an immobiliza- tion medium for FDH proved successful. RMs appear to combine the advantages of organic liquids (prevention of water evaporation) and aqueous buffer (full enzyme activity), and appear suitable for gas-phase sensing. A biosensor produced with FDH and NAD+ immobilized in RM medium had a linear response to formaldehyde vapour of 0.4 12 nA ppb- I over the formaldehyde vapour concentration range of 1.2-1.2 ppm. The limit of detection was not thoroughly investigated, but a formaldehyde concentration of 1.2 ppb could be detected.This compares very favourably with the onlyAnalyst, December 1996, Vol. 121 1773 biosensor reported for formaldehyde vapour, by Guilbault.14 He detected formaldehyde with a detection limit of 10 ppb, while enzymic methods could detect formaldehyde typically in the ppb range, with one assay reporting a detection limit of 120 ppt v/v. As a demonstration of formaldehyde sensing, this biosensor has been successful. However, further work needs to be carried out to improve the stability of the enzyme, which is poor. When steady-state amperometric responses were used as a measure of ADH biosensor response, a linear response was obtained up to approximately 250 ppm ethanol vapour.This ability of the RMs to concentrate the vapour is ideal for sensing low levels of ethanol, but is not suitable for sensing the high levels of ethanol vapour routinely encountered in everyday applications (100-1 000 ppm). Alternatively, the ADH bio- sensor could be used for measuring ethanol vapour, if the exposure times are short, i.e., the concentration of ethanol partitioned in the gel phase is very low and ADH is not substrate saturated. Enzyme stability remains a major problem owing to the poor stability of ADH. Improvements in enzyme purifica- tion and stabilization would greatly enhance further develop- ment of practical ethanol vapour biosensors. If the problem of biosensor stability (i.e., enzyme stability and gel stability) could be overcome, then this formaldehyde biosensor could successfully compete with conventional meth- ods of formaldehyde detection. Its small size compared with the conventional techniques which have to include pumps makes it portable.This formaldehyde biosensor compares well with conventional portable techniques. The limit of detection of this biosensor has not been determined yet, but could possibly be much lower than 1.2 ppb. If exposure times were increased then even lower levels might be determined. References 1 2 3 4 5 6 7 8 9 10 11 Hobbs, B. S., Tantram, A. D. S., and Chan-Henry, R., in Techniques and Mechanisms in Gas Sensing, ed. Moseley, P. T., Norris, J. 0. W., and Williams, D. E., Adam Hilger, Bristol, 1991, pp. 161-181. Dennison, M. J., Hall, J., and Tumer, A.P. F., Anal. Chem., 1995,67, 3922. Fendler, J. H., in Membrane Mimetic Chemistry, Wiley, New York, Zulauf, M., and Eicke, H. F., J . Phys. Chem., 1979, 83, 480. Bardana, E. J., and Montanaro, A., Ann. Allergy, 1991, 66, 441. Environmental Protection Agency, Fed. Regist., 1984, 49, 21 870. Main, D. M., and Hogan, T. J., J . Occup. Med., 1983, 25, 896. Larsen, A., Jentoft, N. A., and Greibrok, K. T., Sci. Total Environ., 1992,120, 261. Noble, J . S., Strang, C. R., and Michael, P. R., Am. Znd. Hyg. Assoc. J., 1993, 54, 723. Beresnev, A. N., Stankov, I. N., Lelikov, Y. A., Yarova, V. A., and Omekhin, A. A., J . Anal. Chem., 1993, 48, 272. Daza, L., Dassy, S., and Delman, B., Sens. Actuators B., 1993, 10, 99. 1982, pp. 48-77. 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 Fatibello-Filho, O., Suleiman, A.A., and Guilbault, G. G., Talanta, 1991, 35, 541. Ogushi, S., Ando, M., and Tsuru, D., Agrir. Biol. Chem., 1986, 50, 2503. Guilbault, G. G., Anal. Chem., 1983, 55, 1682. Biosens. Bioelectron., 1994, 9, v. Ho, M. H., and Samanifar, M., Anal. Chim. Acta, 1988, 215, 249. Lazrus, A. L., Fong, K. L., and Lind, J. A., Anal. Chem., 1988, 60, 1074. Weng, J. L., and Ho, M. H., Anal. Lett., 1990, 23, 2155. Weng, J. L., Ho, M. H., and Nonidez, W. K., Anal. Chim. Acta, 1990, 233, 59. Royal Automobile Club, Camping and Caravaning in Europe, RAC Publishing, London, 1993. Sund, H., and Theorell, H., in The Enzymes, ed. Boyer, P. D., Lardy, H., and Myrback, K., Academic Press, New York, pp.25-83. Kuwabata, S., Nishida, K., and Yoneyama, H., Chem. Lett., 1994, 3, 407. Green, D. W., Sun, H. W., and Plapp, B. V., J. Biol. Chem., 1993,268, 7792. Kunnecke, W., and Schmid, R. D., J . Biotechnol., 1990, 14, 127. Vbradi, M., and Adbnyi, N., Analyst, 1994, 119, 1843. Mizutani, F., Yabuki, S., and Tatsuo, K., Sens. Actuators B, 1993, Wang, J., Romero, E. G., and Reviejo, A. J., J. Electroanal. Chem., 1993,353, 113. Matuszewski, W., and Meyerhoff, M. E., Anal. Chim. Acta., 1991, 248, 379. Mitsubayashi, K., Yokoyama, K., Takeuchi, T., and Karube, I., Anal. Chem., 1994, 66, 3297. Park, J. K., Yee, H. J., and Kim, S. T., Biosens. Bioelectron., 1995,10, 587. Pavaresh, F., Robert, H., Thomas, D., and Legoy, M. D., Biotechnol. Bioeng., 1992, 39, 467. Maekawa, T., Tamaki, J., Miura, N., Yamazoe, N., and Matsushima, S., Sens. Actuators B , 1992, 9, 63. Criddle, W. J., Jones, T. P., and Neame, M. J. H., Meas. Control, 1984, 17, 107. Jones, A. W., Beylich, K. M., Bjomeboe, A., Ingum, J., and Morland, J., Clin. Chem., 1992, 38, 743. Phillips, M., and Greenberg, J., Anal. Biochem., 1987, 163, 165. Gotoh, M., and Karube, I., Anal. Lett., 1994, 27, 273. Miyamoto, S., Murakami, T., Saito, A., and Kimura, J., Biosens. Bioelectron., 199 1, 6, 563. Sarcar, S., Jain, J. K., and Maitra, A., Biotechnol. Bioeng., 1992, 39, 474. Lee, K. M., and Biellmann, J. F., New J . Chem., 1987, 11, 775. Lee, K. M., and Biellmann, J. F., FEBS Lett., 1987, 223, 33. Kawakami, K., Abe, T., and Yoshida, T., Enzyme Microb. Bio- technol., 1992, 14, 371. Larsson, K. M., Aldercreut, P., and Mattiasson, B., Eur. J . Biochem., 1987,166, 157. Kazandijian, R. Z., Dordick, J. S., and Klibanov, A. M., Biotechnol. Bioeng., 1986, 28, 417. Paper 61038541 Received June 3,1996 Accepted July 30, 1996 13-14, 574.

ISSN:0003-2654    

DOI:10.1039/AN9962101769

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 ( I 775-1 778) 1775 Chromogenic Reagents* Mark Dolman, Andrew J. Mason, K. R. A. Samankumara Sandanayake, Andrew Sheridan, Alastair F. Sholl and Ian 0. Sutherland? Department of Chemistry, University of Liverpool, PO Box 147, Liverpool, UK L69 3BX A number of compounds are described which are highly selective chromogenic reagents for cations of biological importance including Li+, Na+, K+, and Ca2+. These reagents are based upon phenolic cryptands with an azophenol chromophore incorporated into their structures, metal complexation is accompanied by deprotonation of the phenolic group and a substantial change in the wavelength of maximum light absorption. The cryptand structures are generally prepared by a novel synthetic route involving reaction of a diaza crown ether with 2,6-bisbromomethylanisole although in one case, for K+ selection, a bridged calix[4]arene is used.Two of the new phenolic cryptands are also highly sensitive reagents for Pb2+. A further variation in the structure to give a phenolic cryptand analogous to the well known solvatochromic Reichardt's dye reagent provides a reagent which is highly selective for Na+ and a response that depends upon solvent polarity. Keywords: Chromogenic; cryptands; cations; lithium; sodium; potassium; calcium; lead; solvatochromic Introduction The classical studies' of Pedersen, Lehn, Cram and co- workers1-6 have shown how highly selective macrocyclic ligands for cations of physiological interest could be designed on the basis of cation-ligand fit and pre-organization of the cation binding site.The requirement for rapid analytical procedures for cations in biological samples has led to the development of TSEs based upon natural and synthetic ion carriers.7,s The development of analytical reagents with a selective optical response has been rather slower, although cation responsive dyes have been known for many years and early work has been reviewed.9.10 Work at Liverpool has concentrated upon the design and synthesis of cation sensitive dyes that could be suitable for use in a simple optical fibre system for cation sensing as developed by Alder and Moss and their co-workers.' 1912 This imposes a requirement that the cation sensitive dye must be water insoluble in addition to the more obvious requirements of good spectroscopic response to the selected cation in the physiological range of concentration (see Table 1) and adequate selectivity to avoid interference from other cations in the sample.Results and Discussion In early work at LiverpoolllJ2 it was shown that ionizable chromoionophores based upon azophenol dyes incorporated into crown ethers 114 and other selective ionophores could be developed into chromogenic reagents for potassium such as z 1 5 and 3,'6 the latter has adequate selectivity for the measurement * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29, 1996. + To whom correspondence should be addressed. of potassium concentrations in serum ( = 5 mmol 1-1) in the presence of much higher concentrations of sodium (up to 150 mmol 1-1).At that point in our work our attention returned to cryptands following the discovery of a very simple synthesis of phenolic cryptandsl7 in a single step from readily prepared diaza crown ethers. This procedure (see Scheme 1) gives good yields of phenolic cryptands, 4, and simply involves heating an appropriate diaza crown ether with 2,6-bisbromomethylanisole in refluxing acetonitrile. The mechanism of the reaction (see Scheme 2) probably involves an intramolecularly templated cyclization reaction and highly effective intramolecular cataly- sis of the subsequent demethylation. The reaction has been found to be successful for a number of diaza crown ethers with 15-21 membered macrocycles (see Scheme 1) and provides a basis for preparation of the cation selective dyes described below.The phenolic cryptands, 4, prepared as summarized in Scheme 1, are readily converted into the corresponding aryldiazophenol dyes, 5 , by coupling with an appropriate diazonium salt (Scheme 3). The dyes, 5, are generally lipophilic compounds which react in a two phase system (CH2C12-H20 or CHC13-H20) with alkali metal cations to give the cryptate salts, 6, as summarized in Scheme 3. The formation of the salt, 6, is Table 1 Concentrations of cations in biological samples* Concentration/mmol 1- Cation Extracellular Intracellular Urine Na+ 150-1 35 1 8-6 220-120 K+ 5-3.5 = 200 80-35 Mg2+ 0.8-0.45 3.2-1.6 8.3-2.5 Ca2+ 1.2-1 .o = 0.05 3.6-0.7 Li+ 1 S - 0 7 * From ref. 13. + In patients undergoing lithium therapy. 1 2 X = OCHp(CH,OCH,)&H,O 31776 Analyst, December 1996, Vol.121 accompanied by a very significant change of absorption in the visible region of the spectrum. As might be expected this process is highly cation selective and the nature of the preferred cation M+ depends upon ring size and the nature of the groups X. The first two compounds of this type to be examined, 7 and 8, were prepared18 from commercially available 1,7-diaza- 15-crown-5 and 1,1 O-diaza- 18-crown-6 in good overall yield. The first of these, 7, which has the smaller cavity proved to be remarkably selective for lithium and values of log& for the equilibrium summarized in Scheme 3 are recorded in Table 2. The reagent, 7, shows a 104 fold preference for Li+ as compared with Na+ and shows virtually no response to K+, Mg2+, and Ca2+ at concentrations up to 1 mol 1-l.The second compound, 8, with a larger cavity shows, not unexpectedly, a reversal of the Li+-Na+ selectivity and shows good selectivity for Na+ as compared with Li+, K+, and Mg2+, unfortunately selectivity for Na+ as compared with Ca2+ is less satisfactory even though formation of the Ca2+ complex involves extraction of a chloride counter ion from the aqueous into the organic phase. A slight structural modification was introduced1’ by the fusion of a benzene ring to the crown ether macrocycle as in the new reagents, 9 and 10. These reagents were expected to show higher selectivity, because they have increased structural rigidity, and also to be more lipophilic than 7 and 8 in accordance with the general requirements of an optical fibre sensing system.The performance of both compounds was up to expectation. The first compound, 9, proved to be totally insoluble in the aqueous phase at pH 7 and above whereas the earlier compound, 7, had been slightly soluble in the aqueous phase at pH 7, and to show virtually no response to any cation other than lithium. The second compound, 10, showed greatly enhanced selectivity as compared with 8 for Na+ against Li+, K+, and, most importantly, Ca2+. In fact these two compounds appear to be the most selective chromogenic reagents that are known for Li+ and Na+, are readily prepared, and have ideal solubility characteristics for use in optical fibre sensors. The introduction20 of a further modification of reagent, 8, by moving the position of one nitrogen atom in the 18-membered macrocycle as in 11 gave a surprising result.Although the cavity size in 11 might be expected to be similar to that of 8, the Na+/Ca2+ selectivity of the two compounds is very different as shown by the values of logK, listed in Table 2. The reagent, 11, has some potential for use in optical fibre sensors for Ca2+ but the value of logK, for Ca2+ shows dependence upon the nature of the anion and the result of adding thiocyanate anions (as LiSCN) to the aqueous phase are recorded. This is not unexpected since a counter anion must be extracted into the aqueous phase to maintain charge neutrality and the thiocyanate anion is more lipophilic than the chloride anion. The develop- ment of a satisfactory chromogenic reagent of the phenolic cryptand type for Ca2+ clearly requires the insertion of a second ionizable group into the reagent, 11, to give a neutral Ca2+ salt in the organic phase.In a later investigation21 it was shown that the reagents, 8 and 11, show a sensitivity to Pb2+ cations that is many orders of magnitude greater than that for Ca2+. This suggests that these two compounds may be suitable for estimation of sub- micromolar concentrations of Pb2+ if a suitable sensing technique can be found to offset the limitations imposed on the use of absorption spectroscopy by the relatively modest values for extinction coefficients (E,,, = 104) for azophenol chromo- phores. The ionizable chromogenic reagents, 7-11, are conventional reagents in that the absorption spectrum of both the reagent and its metal complex are not highly solvent dependent.The development of new cation selective reagents in which the optical response depends upon the medium in which complexa- tion occurs offers interesting possibilities and requires the incorporation of a solvatocbromic chromophore into the reagent. The best known chromophore of this type is that of the betaine, 12, discovered by Dimroth, Reichardt and co-work- ers,22-25 which has been used as the basis for the well known ET(30) scale of solvent polarity. The ET(30) value of a solvent is the magnitude of the energy (in kcal mol-I) associated with the charge transfer band which is responsible for the colour of the dye, 12. The position and intensity of this band depends upon solvent polarity (solvatochromism) and the surface charge density of added cations (halochromism). Charge separation in 12 is reduced in the excited state so that the shifts are hypsochromic (to shorter wavelength) with increasing solvent polarity or in the presence of metal cations of increasing surface charge density (negative halochromism) as in the sequence of physiologically important cations K+, Na+, Li+, Ca2+, and Mg2+.A range of structural modifications of the reagent, 12, have been reported including the crown ether modified dyes, 13.26327 These dyes, 13, show enhanced negative halochromism as compared with 12 and show some cation selectivity in the n X Yield Yo e 1 CH2CH20CH2CH, 75 4 f 3 CH2CHZ Scheme 1 Examples of phenolic cryptand synthesis. Br I + 4a Scheme 2 Mechanism for formation of phenolic cryptand, 4a.Analyst, December 1996, Vol. 121 1777 formation of complexes in a single solvent phase.However, in two-phase systems such as water-dichloromethane and water- chloroform, although the dyes remain in the organic phase, cation extraction from the aqueous phase is accompanied by a pronounced decrease in absorbance which is only slightly cation selective. This effect may be a consequence of the residual hydration shell of the extracted cation since it appears to be absent from cation complexation in a single-phase organic solvent. The dyes, 13, do not show adequate selectivity for sensing cations in biological samples but it was evident that incorporation of the solvatochromic dye into a highly selective phenolic cryptand structure might give a product with the required properties.A betaine of the required type, 14, has been prepared28 as an NaCl adduct 15 by the synthesis outlined in Scheme 4. Deprotonation of the derived hydrochloride salt, 16, to give the betaine, 14, required the use of tetramethylammonium hydrox- T Y M+ I NGN 5 Y=H, Me,0Me,N02 6 Scheme 3 Preparation of arylazophenol dyes from phenolic cryptands. 7 n = l 8 n = 2 9 n = l 10 n = 2 ide in propan-2-01. The NaCl adduct, 15, showed interesting solvatochromic properties rather similar to those of the classic reagent, 12, but with rather smaller hypsochromic shifts than might be expected for bulky solvents presumably due to hindered solvation of the phenolate oxygen due to interference by the rather rigid cryptate system.The absorption maximum of the salt 15 in acetonitrile (h,,, = 544 nm, E = 3050) is similar to that of the crown ether derivatives, 13, in the same solvent but the spectrum of the betaine, 14, in propan-2-01 (A, = 666 nm, E ~ 3 0 0 0 ) shows a significant shift to longer wavelength as compared with the spectra of the dyes 13 (A,,, 532-537 nm)26,27 in the same solvent [ET(30) 48.4 kcal mol-l] presumably as a result of the phenolate oxygen remaining in part in the environment analogous to tetrahydrofuran [ET(30) 37.4 kcal mol-'122-25 which is provided by the cryptand system in close contact with one face of the phenolate system. The ET values calculated from the absorption spectra of the salt 15 [ET(15)] in a wide range of sterically unhindered solvents gave a moderately good linear correlation with ET(30) values (Y = 0.903) suggesting that the solvatochromic effect has similar magnitudes for solvents of this type for both 12 and 15.However, solvents which are evidently too bulky to solvate the cavity efficiently deviate significantly from this correlation. The absorption spectrum of the betaine, 14, has only been examined in a limited range of solvents and its solvatochromism appears to be significantly less than that of the salt 15. The almost colourless salt, 16, reacts readily with sodium chloride in a two-phase system (water-dichloromethane) to form the purple sodium salt, 15 (A,,, = 570 nm). This reaction is highly cation specific and values of the extraction coefficient K , have been determined at pH 7-8 (Table 2).Under these conditions the salt 15 gives no measurable response to other cations of physiological importance but at pH 9 the response to Ph Ph I Ph 0' 12 n = 2, 3, 4 16 13 14 11 Ph Table 2 Summary of cation selectivity for chromoionophores 3,7-11, and 16 Log K,* f 0.2 Cation 3 7 8 9 10 11 16 Li+ -7.0 -9.6 -7.2 -9.5 -9.3 Na+ -9.7 = -11 -6.6 -5.8 -8.5 -6.3 K+ -6.6 -9.3 -9.6 -9.7 < -11.3 Ca2+ -7.9 -9.4 -5.6 (-5.4*) -10.6 pb2+ -t -t -1.2 -$ -0.3 -t -* * Averaged over determinations in the range pH 7.0-10.5 for the two phase system CH2Cl2-H20 (TRIS-HCl buffer). No entry indicates that there is little or no response. Using metal chloride salts. + In the presence of 0.5 moll-' LiSCN. * Not examined. y-42 iii _._c Ph N+ Ph 15 Scheme 4 Synthesis of betaine adduct 15.Reagents and conditions: i, HNO?, NaN02, CHCl,-H20; ii, H2-Pd-C, MeOH; iii, 2,4,6-triphenyl- pyrylium fluoroborate, NaOAc, MeOH; extraction into 3 mol 1-' HCI followed by neutralization (NaHC03), extraction into CH2CI2, and washing with aqueous NaCl.1778 Analyst, December 1996, Vol. 121 1 mol 1-1 LiCl (at 536 nm) is similar in magnitude to that of 0.001 mol 1-1 NaCl although there is still no measurable response to KCl, MgC12 and CaC12. At pH 10.5 the response to 1 rnol 1-1 CaC12 (at 474 nm) is similar to that of 5 X 10-5 rnol I-' NaCl and the response to 1 moll-' KC1 (at 570 nm) is similar to that of 10-5 rnol 1-1 NaC1, in the latter case the response may be due to traces of NaCl in the KC1 (Aldrich, 99.99% purity) since it occurs at the same wavelength as the absorbance of the NaCl adduct, 15.From these results the relative selectivities, based upon aqueous solutions of the chloride salts, for Na+ as compared with other cations of physiological importance, are Na+/Li+ 103, Na+/K+ > 105, Na+/ Ca2+ = 2 X 104 and Na+/Mg2+ too large to be measured. Conclusions The results reported above are summarized in Table 2, which shows values of K, for the five cations Li+, Na+, K+, Ca2+ and Pb2+ for the bridged calix[4]arene, 3, and the phenolic cryptand derivatives, 7-11 and 16. These results, together with results for cryptand derivatives reported by other groups,29-32 indicate the very considerable potential of this type of chromoionophore for use in cation sensors. We thank the Science and Engineering Research Council (UK) for their support.References 1 2 8 9 10 Pedersen, C. J., J . Am. Chem. Soc., 1967, 89, 7017. Dietrich, B., Lehn, J.-M., and Sauvage, J.-P., Tetrahedron Lett., 1969, 2885,2889. Cram, D. J., Angew. Chem., lnt. Ed. Engl., 1986, 25, 1039. Lehn, J.-M., Angew. Chem., Int. Ed. Engl., 1988, 27, 89. Cram, D. J., Angew. Chem., Int. Ed. Engl., 1988, 27, 1009. Pedersen, C. J., Angew. Chem., Int. Ed. Engl., 1988, 27, 1021. Medical and Biological Applications of Analytical Devices, ed. Koryta, J., Wiley, Chichester, 1980. Oesch, U., Ammon, D., Pham, H. V., Wuthier, U., Zund, R., and Simon, W., J . Chem. Soc., Faraday Trans. I , 1986,82, 1179. Takagi, M., and Ueno, K., Top. Curr. Chem., 1984, 121, 39. Lohr, H. G., and Vogtle, F., Acc. Chem. Res., 1985, 18, 65.11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Alder, 3. F., Ashworth, D. C., Moss, R. E., and Sutherland, I. O., Analyst, 1987, 112, 1191. Moss, R. E., and Sutherland, I. O., Anal. Proc., 1989,25, 272. Parker, D., in Crown Compounds. Towards Future Applications, ed. Cooper, S. R., VCH, New York, 1992, p. 53. Kaneda, T., Sugihara, Y., Kamiya, H., and Misumi, S., Tetrahedron Lett., 1981, 22, 4407. Danks, J. P., and Sutherland, I. O., J . Inclusion Phenom. Mol. Recog. Chem., 1992,12,223. King, A. M., Moore, C. P., Sandanayake, K. R. A. S., and Sutherland, I. O., J . Chem. Soc., Chem. Commun., 1992, 582. Sholl, A. F., and Sutherland, I. O., J . Chem. Sac., Chem. Commun., 1992, 1252. Sholl, A. F., and Sutherland, 1. O., J . Chem. Soc., Chem. Commun., 1992, 1716. Sandanayake, K. R. A. S., and Sutherland, I. O., Tetrahedron Lett., 1993,34, 3165. Mason, A. J., and Sutherland, I. O., .I. Chem. Soc., Chem. Commun., 1994, 1131. Mason, A., Sheridan, A., Sutherland, I.O., and Vincent, A., J . Chem. Soc., Chem. Commun., 1994, 2627. Reichardt, C., Angew. Chem., Int. Ed. Engl., 1965, 4, 29. Dimroth, K., Reichardt, C., Siepmann, T., and Bohlmann, F., Annalen, 1963, 661, 1 . Reichardt, C., Chem. Soc. Rev., 1992, 21, 147. Reichardt, C., Asharin-Fard, S., and Schafer, G., Chem. Ber., 1993, 126, 143. Reichardt, C., Asharin-Fard, S., and Schafer, G., Angew. Chem., Int. Ed. Engl., 1991, 30, 558. Reichardt, C., Asharin-Fard, S., and Schafer, G., Annalen, 1993, 23. Dolman, M., and Sutherland, I. O., J . Chem. Soc., Chem. Commun.. 1993, 1793. Helgeson, R. C., Czech, B. P., Chapoteau, E., Gebauer, C. R., Kumar, A., and Cram, D. J., J. Am. Chem. Soc., 1989, 111,6339. Czech, B. P., Chapoteau, E., Zazulak, W., Gebauer, C. R., and Kumar, A., Anal. Chim. Acta, 1990, 241, 127. Czech, B. P., Chapoteau, E., Chimenti, M. Z., Zazulak, W., Gebauer, C. R., and Kumar, A., Anal. Chim. Acta, 1992, 263, 159. Zazulak, W., Chapoteau, E., Czech, B. P., and Kumar, A., J . Org. Chem., 1992, 57,6720. Paper 6103436E Received May 16,1996 Accepted May 24,1996

ISSN:0003-2654    

DOI:10.1039/AN9962101775

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 ( I 779-1 788) 1779 Micro-optical Ring Electrode: Development of a Novel Electrode for Photoelectrochemistry* Gaelle I. Pennaruna, Colin Boxalla,f and Danny O’Hareb a Centre for Photochemistry, Department of Chemistry, University of Central Lancashire, Preston, UK PRI 2HE b Department of Pharmacy, University of Brighton, Brighton UK BN2 4GJ The design of a novel photoelectrochemical sensor, the micro-optical ring electrode (MORE), is described. Based on a thin-ring microelectrode and using a fibre-optic light guide as the insulating material interior to the ring, the MORE is capable of delivering light directly to the region of electrochemical measurement and can therefore be used to conduct microelectrochemical studies of systems with complex photochemistry.A novel fabrication procedure is described, involving the coating of commercially available fibre optics (radius 1.25 X 10-4 m) with a 600 nm layer of gold, so allowing exploitation of the electroanalytical advantages peculiar to thin-ring microelectrodes. The dark electrochemistry of the thin-ring microelectrode is characterized by use of cyclic voltammetry and chronoamperometry and found to agree with previously published theoretical results. Preliminary exploration of the photoelectrochemical response of the MORE is reported, achieved via the interrogation of the photoelectrochemically active phenothiazine dye methylene blue (MB+). Photocurrent signals obtained during cyclic voltammetric and chronoamperometric studies of MB+, conducted with the MORE under illuminated conditions and in the absence of any deliberately added reducing agent, are attributed to the formation and subsequent detection of 3MB+ within the diffusion layer of the microring electrode.The data demonstrate that the use of the MORE for the direct electrochemical detection of photogenerated species with lifetimes of c 9 x 10-5 s is possible. The electrochemistry of 3MB+ over the applied potential range from -0.4 to +1.0 V versus SCE is elucidated and discussed in the context of the behaviour of photoexcited MB+ in the presence of the deliberately added reducing agent Fe2+. Keywords: Ring microelectrode; photoelectrochemistiy y; optical electrode; fibre optic; methylene blue Introduction Microelectrode techniques offer several major advantages over traditional electroanalytical techniques.These have been de- scribed in detail in several recent authoritative reviews1,* and can be summarized as follows: (i) enhanced current density, leading to enhanced mass transport; (ii) smaller cell time constant, leading to improved temporal resolution, shorter measurement times and improved faradaic to non-faradaic current ratio facilitating the investigation of rapid, classically ‘reversible’ electron transfer and very fast coupled homoge- neous reactions (fast-scan microelectrode voltammetry has been used to interrogate the mechanistic details of processes possessed of kinetics approaching the nanosecond time scale);’ * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March + ‘To whom correspondence should be addressed.25-29, 1996. (iii) low current, leading to reduced ohmic polarization, permitting measurements in hitherto electrochemically intracta- ble situations, eg., highly resistive media and solutions without deliberately added supporting electrolyte (the low current also greatly reduces analyte consumption); and (iv) small physical size, allowing measurements in small sample volumes, provid- ing good spatial resolution and permitting virtually non- invasive measurements in living tissue. Applications of micro- electrodes range from the study of electrode kinetics4 to in sitzi electroanalytical determinations of the chemistry of the living brain.5 Similarly, photoelectrochemistry is an expanding research front owing to its range of applications in, among others, the areas of solar energy conversion,6-9 organic and inorganic pollution abatement,’() non-linear optics’ and the study of photoactive drugs for the treatment of cancer.12 Many of these photoelectrochemical processes involve the photogeneration of electrochemically active, short-lived intermediate species, the properties of which, it is reasonable to say in the light of the above, may be readily interrogated by use of microelectrode techniques.This paper describes the construction and establishment of the feasibility of application of a novel design of electrode, the micro-optical ring electrode (MORE). Based on a thin-ring microelectrode design and using the insulating material interior to the ring electrode as a light guide, the MORE is a device capable of delivering light via a fibre optic directly to the region of electrochemical measurement, so allowing microelectro- chemical study of the mechanism and kinetics of systems with complex photochemistry.Ring microelectrodes offer additional advantages over the more commonly used, and easily con- structed, inlaid disc design since the extra edge results in higher mass transport rate constants, leading to higher current densities, which allow the study of faster reactions and improved temporal resolution. Further, the reduction in the geometrical area leads to additional improvements in the faradaic to non-faradaic current ratio. Reliable numerical solutions now exist for ring mi~roelectrodes,13,~4 suggesting the possibility of their calibrationless use within a MORE config- uration.The reduced ohmic drop and non-faradaic current at microelectrodes allow experiments to be conducted in the absence of a supporting electrolyte, the modelling of which is highly advanced.lJ5 Many photochemical systems can only be studied in media of low relative permittivity. Microelectrode measurements have been reported in solutions of 1,2-dichlo- robenzene,16 toluene17 and even supercritical C02.18 The MORE will therefore greatly increase the number of systems amenable to photoelectrochemical interrogation. It should be noted that many optical electrode designs already exist, the two general types of which are shown in Fig. 1; however, neither of these designs is without problems. For instance, if light is shone through solution at the electrode [Fig.l(a)], then the solution has to be transparent enough to allow the passage of light to the electrode. This means that only1780 Analyst, Decemhei- 1996, Vol. 121 a small fraction of light can be absorbed close to the electrode surface. High absorption close to the electrode surface is desirable, however, so as to maximize the current obtained from any photogenerated intermediates with short lifetimes. One possible solution to this problem is offered by the optically transparent thin-layer electrode19 (OTTLE), wherein the use of a thin-layer cell configuration provides for greater intimacy between the working electrode and incident photons. However, all thin-layer cell designs have an inherent problem derived from the presence of an uncompensated solution resistance, limiting their use in mechanistic analysis by cyclic voltam- metry.Two recent innovations in the application of photoelectrodes incorporating through-solution illumination are the photo- electrochemical microscope (PECM) and the use of channel microband photoelectrodes. With the former, pho- toelectrochemical imaging of the spatial variability of electro- chemical processes at a macroscopic electrode surfxe is accomplished by the rastering of a small spot of light across the surface while the total electrochemical response of the entire sample is monitored;20-25 however, this technique is restricted to semiconducting surfaces. With the latter, an array of microband electrodes is illuminated from above,20 allowing the utilization of many, but not all, of the advantages offered by microelectrodes.When light is shone through an electrode [Fig. l(b)], the solution can contain much higher concentrations of absorbing species so that a large part of the light can be absorbed in the diffusion layer of the electrode. Such an optically transparent electrode27 (OTE), which in practice is semi-transparent, may be fabricated either by the sputtering of a thin film of metal (usually gold or platinum) or the pyrolytic deposition of a layer of quasi-metallic semiconductor (n-type Sn02) on to either a quartz microscope slide or the polished end of a quartz rod light guide. A recent innovation in semi-transparent electrode design is the photolithographic generation of arrays of individually addressable tin-doped indium oxide microdisc electrodes, affording the exploitation of many of the advantages offered by microelectrodes.28 A variation on the semi-transparent elec- trode theme is provided by photoelectrodes that employ attenuated total reflectance (ATR) as a mechanism by which to accomplish photoexcitation of solution species; unfortunately, the absorption of light within such a configuration will be restricted to a thin solution volume directly adjacent to the electrode surface, with a concomitant reduction in the size of the photoelectrochemical signal.Conventional semi-transparent electrode assemblies have two intrinsic problems: (i) the presence of a metal/semicon- ductor layer over the end/face of the light guide attenuates the intensity of light emerging into solution (the band gap of semiconductor materials imposes an upper limit on the energies of light transmittable through the electrode surface itself); and (ii) electrode surfaces become poisoned and must be polished Fig.1 Schemes of the two most common photoelectrode configurations: ( a ) through-electrolyte illumination; and (6) through electrode illumina- tion. periodically but polishing a semi-transparent electrode would remove the thin surface film and render the electrode useless. Both disadvantages may be obviated by use of an optical disc- ring electrode c~nfiguration~~ such as that found in the MORE. Light is shone through the central disc and photogenerated product detected at the ring. Repeated polishing of this assembly is possible with no reduction in performance and no surface layer exists to cause light attenuation.The MORE also overcomes the need for high light intensity through the use of a light guide. In contrast to the photoelec- trochemical microscope (PECM), where high-powered lasers may cause thermal perturbation,30-32 we have found that photoexcitation may be achieved by use of a 200 W Xe-Hg arc lamp, resulting in a much less intense light beam. Consequently, thermal perturbation is insignificant. This, combined with the relative insensitivity of microelectrodes to bulk convection, facilitates modelling, especially in the presence of supporting electrolyte where migration is negligible. In addition to the advantages associated with microelectrodes described above, the MORE has an advantage over macroscopic ATR-based electrodes, illuminated channel microband electrodes, semi- transparent microelectrode array devices and the PECM in that it allows access to hitherto physically inaccessible media.Two recently published reports present data obtained with a device of similar design to the MORE. The first report, from Cohen and Weber,33 concerns the use of a commercially available gold-coated fibre optic in the photoelectrochemical detection of catalase. Through the use of an Ru(bipy)32+ sensitizer, the device photochemically generates and electro- chemically detects H202. The electrochemical signal is attenu- ated in the presence of catalase, for which H202 is a substrate. The same group of workers have also used their device as a sensor for Ru(bipy)32+ via the detection of its electrogenerated chemiluminescence.The second report, from Casillas et aZ.,34 concerns the use of a commercially available gold-coated fibre optic within a scanning electrochemical microscope configura- tion (SECM), to study photo-oxidative processes on a Ti02 macro-electrode with a resolution of 0.2 mm. However, the data described in both reports were obtained using: (i) a laser light source, rendering the experiment liable to the effects of thermal convection at the ring electrode-solution interface; and (ii) a thick-ring electrode configuration in which a/b < 0.91, where a and b are the interior and exterior radii (in metres) of the ring, respectively. The design and construction of the MORE obviates difficulties associated with the former by use of a 200 W Xe-Hg arc lamp as illumination source (see above), and it has several advantages over the electrode in the latter by virtue of employing a thin-ring configuration.The use of a thin-ring configuration, where a/h > 0.91,3' offers two additional advantages to those described above.13~3~~37 The first is that the consequent high material flux to the electrode surface both facilitates the detection of short- lived intermediates generated in the solution phase and means that electron transfer will be rate determining over a wider potential range than observed on conventional ring electrodes. l 3 The latter effect ultimately results in an enhanced ring collection efficiency for the capture of electroactive product photogenerated by a concentric optical disc.Indeed, it has been shown that, under conditions of diffusion control, there are singularities at the ring edges and the current density there asymptotically approaches infinity.13 It is therefore expected that the collection efficiency of a thin-ring-based device will be greater than that of a thick-ring-based device of the same internal ring diameter owing to the greater proximity of external ring edge to the region of photogeneration; this is an especially important consideration for the efficient electrochemical detec- tion of short-lived photogenerated species. The second advantage is that all of the extant semi-analytical mathematical solutions for the behaviour of conventional ringAnalyst, December 1996, Vol.121 178 1 electrodes in the dark are for thin rings;37 use of a thin-ring MORE therefore precludes the necessity for mathematical characterization of the electrode response in the dark, an important consideration when dealing with electrochemically complex systems. A significant disadvantage of using thin-ring MOREs is that fibre optics coated with thin metal layers of the order of 10-6-10-5 m thick are not available ‘off the shelf’ and must therefore be fabricated in-house. Many users of microelectrodes have experienced electrode failure; this derives from two general sources: (i) material failure, where one or more of the constituent materials of the electrode assembly have proved to be incompatible with the demands of the experimental environment; and (ii) failure of the electrode-body seal, often due to insufficient cleaning of the electrode components during fabrication.Therefore, we have developed a novel assembly procedure for the reliable manu- facture of MOREs. The objective of this work was to develop a new design of photoelectrode that simultaneously obviates the difficulties associated with, and is more generally applicable than, the devices described above. Indeed, it is ultimately intended that the MORE will be both of generic analytical use and the instrument of choice when investigating new photoelectro- chemical systems: the former derives from the inherent selectivity offered by the ability to tune the energy of the incident light to the specific absorbance wavelength of photoelectroactive species in complex mixtures; the latter obtains from the versatility of the MORE in being able simultaneously to stimulate a photoelectrochemical response from a system and to detect photogenerated reaction products or intermediates.Thus, to demonstrate the feasibility of the latter, this paper will present data describing what we believe to be the first recorded direct electrochemical detection of the short-lived photo-excited triplet state of the well characterized, photo- chemically active phenothiazine dye methylene blue (MB+). The former will be considered in a subsequent paper. Experimental Reagents Methylene blue dichloride (analytical-reagent grade) was purchased from Merck (Poole, Dorset, UK) and used without further purification. All other chemicals were of analytical- reagent grade or better and used as received.Distilled water, produced in a laboratory-made still, was further purified in a de- ionization system (E Pure Model 04642, BarnsteadDherrno- dyne, Dubuque, IA, USA) to a resistivity of 1.8 X lo5 SZ m. Prior to electrochemical analysis, unless stated otherwise, all reagent solutions were purged for 15 min with nitrogen (white spot grade, BOC, Guildford, Surrey, UK) to remove dissolved oxygen. Unless specified otherwise, the supporting electrolyte was 50 mol m-3 potassium dihydrogenphosphate buffer (pH 8). MORE Design and Fabrication Procedure A schematic illustration of the MORE is shown in Fig. 2. Commercially available 2.5 x m (core + cladding) Tip Shank ______ Insulating sheath __~_.___~.._____ -Light in, using auartz core as 7- waveguide Conductive coating, forming ring electrode at tip extending up shaft to establish electrical contact Fig.2 Schematic diagram of the tip cross-section of the MORE. diameter fibre optics (Model FC-2UV, Newport/Micro-Con- trole, Newbury, Berkshire, UK) were stripped to the cladding and washed with sonication for 15 min each in chloroform, acetone, ethanol (all of AnalaR grade, Merck) and triply de- ionized water. Between each washing stage, and after excess solvent had been removed with a lens cloth, the fibres were dried for 15 min at 383 K. The fibres were then washed in 5 kmol m-3 HN03 (prepared from AnalaR-grade acid, BDH) for 1 h, followed by a final wash in triply de-ionized water. The fibres were then oven dried at 383 K for 30 min or until required.Gold deposition was accomplished as follows. The cleaned fibres were mounted on a rotatilt armature within an Edwards (Crawley, Sussex, UK) Auto306 thermal evaporator. Prior to gold deposition, the fibres were further cleaned within the evaporator by use of an rf air plasma at 0.133 N m-2. The fibres were then set rotating at a rate of 20 rpm while gold (Alfa, Johnson Matthey, Royston, Hertfordshire, UK) was thermally evaporated from a molybdenum boat (Agar Scientific, Cam- bridge, UK) (evaporation current = 50 A), and allowed to deposit on the fibres until an even coating of gold with a thickness of up to 600 nm (as measured with an Edwards FTM5 film thickness monitor) had been achieved. (Note. Further improvements in gold-substrate adhesion can be achieved through the use of a recently reported silanizing te~hnique.~~) Electrical contact to the gold coating was achieved by use of silver wire secured by Ag-loaded epoxy resin (2400 Circuit Works conductive epoxy from RS Components, Stockport, Cheshire, UK).The coated fibres were potted in epoxy resin (Araldite CY 1300 and HY 1300, Ciba-Geigy Polymers, Cam- bridge, UK) selected for low water permeability and excellent dielectric properties. As 90% of the diffusion field of a disc microelectrode is confined to within seven electrode radii of the centre of the disc at the steady state,39 and as the diffusion field of a ring microelectrode approximates that of a disc at long times,39 the insulating sheath of the MORE must be at least 7.5 X 10-4 m thick to avoid problems associated with diffusion from behind the plane of the electrode surface.This was accomplished by potting the fibres within nylon pipette tips. On curing, the entire assembly was cut with a diamond saw to expose a thin-ring electrode concentric with the light guide. The electrodes were polished using diamond pastes of successively decreasing grain sizes (range 9 x 10-6-0.1 X 10-6 m, from Marcon, Hitchin, Hertfordshire, UK) spread over a Microcloth polishing cloth (No. 40-7212, Beuhler, Lake Bluff, IL, USA). The electrodes were sonicated in ethanol for 15 min between each polishing step. Finally, the electrodes were polished with a slurry of 15 nm A1203 in a 10 mol m--3 solution of alkaline NaCN, and sonicated in water so as to remove from the ring any embedded y-alumina particles which could otherwise act in an electrocatalytic fa~hion.~O The electrodes were then stored in a dust-free environment until required. The final polishing step was repeated before each use to expose a fresh surface.Out of an initial batch of ten ring microelectrodes fabricated in this way, eight were found to be usable, demonstrating good agreement between the measured dark electrochemical re- sponse and theoretical predictions based on electrode geometry (see below). Equipment and Optical Arrangement A BAS-CV37 Voltammograph (Bioanalytical Systems, West Lafayette, IN, USA) interfaced through a Strawberry Tree ACJR- 12-8 data acquisition card (Adept Scientific, Letchworth, Hertfordshire, UK) to a 486-DX66 desktop computer (Akhter Computers, Burnley, UK) was used for all voltammetric and chronoamperometric measurements. Data acquisition was con- trolled by use of Strawberry Tree QuickLog PC software (Adept Scientific).Cyclic voltammetric and chronoamperometric data1782 Anulyst, December 1996, Vol. 121 was imported from QuickLog into Microsoft Excel for manipulation prior to plotting. A commercial saturated calomel electrode (SCE) (EIL, Chertsey, Surrey, UK) was used as the reference, and a platinum wire (2.5 X 10-4 m diameter, 99.99+% purity, Advent, Halesworth, Suffolk, UK) as the counter electrode during all of the experiments reported here. The potentials cited in this study are all given versus the SCE. Illumination was provided by a 200 W Xe-Hg arc lamp (Ealing Electro-Optics, Watford, Hertfordshire, UK).The output of the lamp was focused through an Ealing Electro- Optics fibre-optic light-guide adapter with coupler-fused quartz lens (Ealing Electro-Optics) on to one end of an FC-2UV fibre- optic light guide (Newport/Micro-Control, Newbury, Berk- shire. UK). The other end of the fibre-optic light guide was coupled to the proximal end of the fibre optic that formed the optical disc of the MORE by use of a laboratory-made optical linkage. Variations in the photocurrent between experiments are attributable to variations in the efficiency of the laboratory- made optical linkage between the fibre-optic light guide and the MORE. As the aim of our work was to demonstrate the applicability of the MORE in qualitative mechanistic interro- gation, it was not necessary to measure the intensity of light emerging from the surface of the optical disc into solution during each experiment; however, a knowledge of light intensity at the MORE surface will be imperative if the device is to be used quantitatively. Experimental Procedures Unless specified otherwise, each individual cyclic voltammo- gram and chronoamperogram was recorded on a freshly cleaned electrode (see above).All (photo-)electrochemical experiments were conducted with the electrochemical cell placed within a laboratory-made Faraday cage, which also served to exclude ambient room light from the cell during experiments in the dark. The complete experimental arrangement is shown schemat- ically in Fig. 3. All solutions were degassed prior to the experiments (see above).All experiments were conducted at room temperature (291 f 2 K). Results and Discussion Chronoamperometric Characterization of the Ring Electrode in the Dark As noted by Szabo,36 the derivation of analytical expressions for the time dependence of the current to microelectrodes is a challenging problem because the diffusion equation must be solved subject to mixed boundary conditions, i.e., the flux is specified at one part of a surface and the concentration at another. These difficulties are exacerbated with ring mic- roelectrodes because a micro-ring of finite thickness does not exhibit a uniform current density over its entire surface. Indeed, as mentioned above, it has been shown that, under conditions of Fig. 3 Schematic diagram of interface via AC-JR p h \ 486 desktop computer the MORE apparatus.diffusion control, there are singularities at the ring edges where the current density asymptotically approaches infinity. 13 Ring microelectrodes are thus mathematically less tractable than macro- or hemispherical electrodes, and have consequently been much less extensively analysed than disc microelectrodes, especially under conditions of departure from diffusion con- trol. A unified analytical treatment of the voltammetric behaviour of ring microelectrodes has yet to be established, although Tallman and co-workers4',42 have computed linear-sweep and square-wave voltammograms of reversible systems at micro- rings via a convolution of the transient current in response to a large potential step and the appropriate potential scan function.Nonetheless, the availability of simple analytical expressions is preferred as it greatly accelerates the analysis of experimental data. The main contributions to this problem have been the semi-analytical approximations for the steady-state current at a microring electrode, iR,ss (A), derived by Szabo,36 Fleischmann and co-workers,43,44 Phillips and Stone13 and Symanski and Bruckenstein35 after a treatment of the capacitance of a circular annulus by Smythe.45 The latter treatment, although it is in itself an approximation, has usually been regarded as being the most accurate13,35,36,44 and yields the following expression for the steady-state diffusion current at a thin-ring microelectrode: iR,+ = nFDClo (1) where n is the number of electrons per electroactive species oxidized or reduced, F is the Faraday constant, D is the diffusion coefficient (m2 s-I), C is the concentration of the electroactive species (mol m-3) and 10 is given by n2(u + h ) and a thin ring is defined as having u/b > 0.9 1.The treatments of Szabo and Fleischmann et al. both assume that the surface of the micro-ring is uniformly accessible, although exact numerical analysis of the surface current density distribution shows that this is not so? Phillips and Stone made no assumptions in their treatment, except that the thickness of the ring is very much less than the over-all ring radius, resulting in an expression for the diffusion-controlled current that agrees with the approximate result of Smythe.The treatment of Szabo also derives an asymptotic expression for the long-time behaviour of the diffusion-limited current, iR,D(t) (A), at a micro-ring electrode, given by ( 3 ) where t is time elapsed (s) and. under conditions where (h - a)/ b << 1, lo is given by n2(u + h) Szabo also described an empirical expression for lo: n2(u + b) ( 5 ) Eqn. ( 5 ) was devised in an attempt to describe ring electrodes with arbitrary thickness. It can be seen that when a = 0,Analyst, December 1996, Vol. 121 1783 substitution of eqn. ( 5 ) into eqn. (3) generates an expression that, as t -+ a, reduces to the equation describing the steady- state current at a disc microelectrode of radius b. Fig. 4 shows the full chronoamperogram for a typical MORE up to 28 s after the application of the potential step, compared with the current-time behaviour that might be expected from eqn.(3). The theoretical and experimental curves are in excellent agreement over the entire time period, with no evidence of convective mass transport disrupting the predicted behaviour. The time taken to reach the steady-state current is of the order l*/D, where 1 is some characteristic dimension of the electrode35 (m); in the case of a ring microelectrode, 1 is taken as being the inner radius of the ring, a.35 The time is calculated as being 25 s for the MORE employed in Fig. 4. The current at that time is -197 nA, which compares very well with the steady-state limiting current of -200 nA obtained from the cyclic voltammogram shown in Fig. 5 and the steady-state currents of - 174, - 176 and - 176 nA derived from eqn.(1) in conjunction with the expressions for lo given by Szabo [eqns. (4) and ( 5 ) ] and Symanski and Bruckenstein and Smythe [eqn. (2)], respectively. The scatter in the last three values is due to the different working assumptions made in the treatments (see above). A ring microelectrode with the dimensions of the MOREs used in this study should show a sigmoidal waveform for voltammetry performed at slow scan rates. A typical sigmoidal voltammogram for a 1 mV s-1 scan rate is shown in Fig. 5 and is in general agreement with the generic linear voltammetric behaviour of reversible systems at thin-ring microelectrodes recorded at low sweep rates predicted by Kalpathy et al.41 All of the approximate treatments described earlier specify that their predictions are based on the use of a thin ring, the most generous definition of a thin ring being that of Symanski and Bruckenstein, i.e., a/b > 0.91 ; the criteria for thinness described 03 I -loo/ experimental, I - ' 0 ° 1 -800 0 5 10 15 20 25 30 Time/s Fig.4 Comparison of the experimental chronoamperogram recorded at a MORE with that predicted by the theory of Smythe45 and Szabo.36 The solution was 0.01 mol mP3 K3Fe(CN)6 in aqueous buffer solution. The applied potential was stepped to -0.3 V vel-ius SCE from floati solution was deoxygenated as described in the text. 50 '1 g. The 0: p -50; g -100: L- L 5 -150f -200: - 2 5 0 ; . . . . , I l . l r , , . . , . . . , , . , . , , , , , . , , , , . , . . , , -0.6 -0.4 4 .2 0 0.2 0.4 0.6 0.8 VoltageN Fig. 5 Steady-state voltammogram recorded using a MORE at a 1 mV s-l scan rate. The starting potential was +0.9 V versus SCE and the initial scan was in the direction of decreasing positive potential. All other experimental conditions were as those described in Fig. 4. in the other reports are all more stringent, often reducing to the requirement that (b - a)/a asymptotically approaches zero.' 37.76 The work of Cohen and Weber33 and Casillas et al.34 employed ring microelectrodes with a/b values of 0.86 and 0.68, respectively, effectively denying them access to the expressions described by and derived from eqns. (1)-(4). However, the MOREs used in this study all have the important advantage over the devices of Cohen and Weber and Cosillas et al.that they have a/b values > 0.995; they may therefore be considered to be thin-ring microelectrode devices with access to the expressions described above. Characterization of the Behaviour of the Probe Compound on Microelectrodes Initial characterization of the photoelectrochemical response of the MORE was conducted via the interrogation of the well characterized, photoelectrochemically active phenothiazine dye methylene blue. Methylene blue was selected for a number of reasons: its dark electrochemistry has been extensively studied on a wide range of electrode materials including mercury,46-48 platinum,49 carbon50 and gold,s1,52 in addition to electrode surfaces that have been modified with Nafions3-55 and sul- fur;56-61 the photochemistry of MB+ has also been widely investigated; indeed, the triplet state of methylene blue was the first species to be observed by nanosecond laser flash photolysis.6' Although the photoelectrochemistry of MB+ has been less extensively studied than its photochemistry and dark electrochemistry , the p hotoelec troc hemistry of thionine (Th), the closest relative to MB+ in the phenothiazine family of dyes, is well known, primarily owing to its role as a photosensitizer in photogalvanic cells,63.64 and much may be inferred about the behaviour of MB+ from those data.With reference to the last point, it should be pointed out that the lifetime of the photoexcited triplet state of MB+ is greater than that of triplet thionine (being 90 X 10-6 and 50 X 10-6 s, respectivelyh').rendering 3MB+ more readily detectable in a direct electro- chemical fashion at the MORE than 3Th. However, photoproducts with still longer lifetimes may be obtained on illumination of MB+ in the presence of a suitable sacrificial electron donor such as Fe*+. In direct analogy to the photosensitized reaction between Th and Fe2+ employed in photogalvanic cells, the reaction between MB+ and Fe2+ may be written as MB+ (aq) + H+ (aq) + 2Fe2+ (aq) % (6) producing Fe3+ and leucomethylene blue (LMB). Using the rate constant for the back-reaction between Fe3+ and t h i ~ n i n e ~ ~ and assuming a photogenerated Fe3+ concentration of I X mol m-3 (the highest concentration of Fe*+ used in the present study), we estimate the lifetime of LMB to be > 3 s. Therefore, the loss of LMB in its passage from the optical disc to the ring of the MORE is insignificant.The photochemical properties of MB+ are summarized in Table 1. Fig. 6 shows what is widely believed to be the sequence of reactions during the heterogeneous electrochemical reduction 2Fe3+ (aq) + LMB (as) Table 1 Photochemical and electrochemical properties of MB+ (adapted from ref. 12), where h is the wavelength of light and E the absorption coefficient h,,,(abs)/nm 565 E" (singlet ground state)/V* -0.2 Emdx/rn2 mol-1 8 160 Eo (first excited singlet state)/V' Eo (first excited triplet state)/V' 1.63 1.24 * All potentials i w s u s SCE.1784 Analyst, December 1996, Vol. 121 300 - p 200- 2 100- g 0: 3 0 -100- -200 - -300- of MB+ to LMB,12,46-s2366 i.e., it follows an ECE process.The protonation reaction has been observed to be very fast or virtually complete in nearly neutral or acidic media; conse- quently, the ECE process appears EE over a wide range of scan rate~.6~ Although extensively studied with a wide variety of electrode systems, comparatively few reports have appeared concerned with the electrochemistry of MB+ at microelec- trodes,66,6* the most recent being a study of the behaviour of methylene blue at carbon fibre microcylinder electrodes.@ To our knowledge, the study of the electrode reaction mechanism at gold microelectrodes has not been reported. Therefore, what follows is a summary of findings made in these laboratories concerning the mechanism of the electrochemistry of MB+ at gold microelectrodes. The elucidation of these conclusions will be described in more detail in a subsequent paper.Fig. 7 shows the cyclic voltammogram of MB+ recorded in the dark using a MORE at a potential sweep rate, Y, of 0.05 V s-1 in aqueous solution buffered at pH 8. The cathodic and anodic potential limits of the sweep are -0.36 and +0.1 V, respectively. Under these conditions, one cathodic and one anodic peak are seen, which shall henceforth be referred to as REDl and 0x1; E,,c and E,,,, the cathodic and anodic peak potentials, respectively, have values of -0.27 and -0.22 V, respectively. The reduction reaction associated with REDl may be represented by the over-all process 2MB+(aq)+ 2H+(aq)+ 3e- -+ LMB/LMB+' (s) (7) where LMB/LMB+* is a mixed-valence salt of the radical cation LMB+* and the expected product LMB.OX1 derives from the reverse of this process: LMB/LMB+' (s)+ 2MB+ (aq)+ 2H+ (aq)+ 3e-- (8) If the cathodic limit of the potential sweep is extended to -0.9 V, two new peaks may be observed in the cyclic voltammogram: a broad cathodic peak between -0.4 and -0.8 V, which will henceforth be referred to as RED2, and an anodic peak, 0 x 2 , for which = -0.14 V. After reaching a maximum value, the current of OX2 decays proportionally to t"2, characteristic of a mass transport-controlled process at an electrically insulating layer. RED2 is associated with the reduction of the LMB/LMB+* mixed-valence salt to generate LMB: LMB/LMB+' (s) + e- + 2LMB (s) (9) OX2 corresponds to the LMB + MB+ stripping peak. Such a reduction would be expected to be accompanied by a change in the crystal phase of the surface layer and therefore be electrochemically irreversible; this conclusion is borne out by the broadness of RED2.It is sufficient at this point to realise that the electrochemical data and reaction mechanism concerning the reduction of ground-state MB+ shown in Table 1 and Fig. 6, respectively, are oversimplified and that the material presented above is more appropriate for an understanding of the following photo- electrochemical data. Cyclic Voltammetry of Methylene Blue on the MORE in the Presence of Illumination Before proceeding to discuss the cyclic voltammetric behaviour of MB+ at the MORE in the presence of illumination, two further points concerning the electrochemistry of MB+ must be addressed. The first is that evidence exists to suggest that MB+ may be oxidized as well as reduced, to generate, in the first instance, the 2+ radical cation shown in Fig. 6.Bauldreay and Archer64 reported a formal potential for MB+ oxidation on SnO2 electrodes of 0.994 V versus SCE in solutions of 0.1 mol m-3 MB+ at pH 1, the reaction being quasi-reversible insofar as the radical cation could be re-reduced on the reverse-going sweep. Karyakin et al.70 also observed the oxidation of MB+ at glassy carbon electrodes in aqueous solution at pH 9.1, the reaction onset potential being about 1 V versus an Ag/AgCl reference electrode in 1 kmol m--3 KC1 solution. However, they did not observe the re-reduction of the radical cation as Bauldreay and Archer did; instead, they reported that electro- generation of the radical cation in some way facilitates the formation of a surface layer of poly(methy1ene blue) at glassy carbon electrodes.The results presented below suggest that the photo-assisted oxidation of MB+ is also possible on gold electrodes. The second of the points to be discussed is that, as can be seen from Table I , the redox characteristics of MB+ in the singlet and Methylene Blue, MB+ + delocalised -> dipole -> dirner 1 400 1 A I H H Fig. 6 MB'. A simplified representation of the redox reactions available to 0 light -4001 r 3 I , , r . . I , * I . . 1 ' . . , I I . . 9 , . . . . , . . , . I , .. 1 . . I . I , , I , 1 -0.4 -0.3 -0.2 -0.1 0 0.1 Voltage# Fig. 7 Dark and light cyclic voltammograms, recorded at a scan rate of 0.05 V s-I using a MORE, of 1 mol m--3 MB+ in water containing 50 mol m--3 phosphate buffer (pH 8).The starting potential and anodic limit was +0.1 V versus SCE, the cathodic limit was -0.36 V versus SCE and the initial scan was in the direction of decreasing positive potential in each case. The solution was deoxygenated as described in the text.Analyst, December 1996, Vol. 121 1785 triplet excited states differ markedly from those in the ground state. In aqueous solutions of pH 8, electrons may be injected into the lowest unoccupied molecular orbital (LUMO) of the singlet ground state of MB+ at electrode potentials more cathodic than -0.2 V versus SCE (see Fig. 7 and, in an oversimplified representation, Table 1) and, in aqueous solu- tions of pH 9, may be removed from the highest occupied molecular orbital (HOMO) of the singlet ground state of MB+ at electrode potentials more anodic than approximately 1 V versus SCE (see ref.70 and Table 1). It should be noted that there will be some displacement in the energies of the molecular orbitals of MB+ in the photoexcited triplet state with respect to the values adopted in the singlet ground state owing to the differences in electron distribution between the two states. This may explain the differences between the electrode potential for the oxidation of ground state MB+ reported by Karyakin et al.70 and the EO for photoexcited 3MB+ in Table 1. Therefore, care must be taken when using the E* for MB+ in the ground state (-0.2 V versus SCE), corresponding to electron injection/ removal to/from the ground-state HOMO, as a measure of the formal potential of the same molecular orbital in the first triplet excited state.The first excited singlet state of MB+ has a lifetime of only 330-390 ps,7' making direct electrochemical detection of lMB+ unlikely on the time-scale of the experiments reported here. Consequently, we shall say no more about it save that it decays to form either ground-state singlet MB+ or the first excited triplet state of MB+. However, as 3MB+ has a lifetime of about 90 x 10-6 s,65 it may be directly detectable at the MORE and, therefore, some consideration of its excited-state electro- chemistry is necessary. As can be seen from Table 1, owing to the excitation of an electron from the ground state HOMO to LUMO, it is possible to reduce the first excited triplet state of MB+ by injection of an electron into what was the ground-state HOMO at electrode potentials more negative than 1.24 V versus SCE; at potentials more negative than about -0.2 V versus SCE, it is thermody- namically possible to reduce 3MB+ by electron injection into either the former ground-state HOMO or LUMO.Similarly, it is possible to oxidize 3MB+ by removing the photoexcited electron from what was the ground-state LUMO at potentials more positive than about -0.2 V versus SCE; at potentials more positive than 1.24 V, it is thermodynamically possible to oxidize 3MB+ by the removal of electrons from either the HOMO or LUMO. The availability to MB+ of these additional reactions in the potential range -0.2 to 1.24 V versus SCE upon photoexcitation defines much of the photoelectrochemistry of methylene blue.Fig. 7 shows the cyclic voltammetric behaviour of MB+ in pH 8 aqueous solution in the absence (see above) and presence of illumination, recorded using the MORE. On illumination with white light, the cyclic vol tammetric characteristics of MB+ during the anodic-going reverse sweep are unchanged with respect to those in the dark; it is during the forward-going cathodic sweep that significant changes in behaviour are observed. Specifically, photoanodic and photocathodic currents are observed at applied electrode potentials more positive and negative than -0.25 V, respectively. This would appear to be readily explained. As discussed above, MB+ has been photoexcited to 3MB+, a more readily reducible species; thus the enhancement in the cathodic current at potentials more negative than -0.25 V might be explained by the photoassisted reduction reaction (10) However, if this was the case, a peak corresponding to OX2 would be expected on the return sweep. No such peak is observed.Further, photoassisted reduction of MB+ by eqn. (10) does not explain the photoanodic current observed under 'MB+(aq)+ H+(aq)+ 2e- + LMB (s) illumination at applied potentials more positive than -0.25 V. As noted above, it is conceivable for photoexcited 3MB+ to be oxidized and/or reduced at an electrode with an applied potential in the range from about -0.2 to 1.24 V. The photoanodic current observed at applied potentials more positive than -0.25 V may therefore be due to the following oxidation: (1 1) The photoelectrochemically generated MB2+ accumulated in solution is then the origin of the photocathodic current seen at applied potentials more negative than -0.25 V as it, and photoexcited 3MB+, are reduced to LMB/ LMB+* according to 2MB2+ (aq) + 2H+ (aq) + 5e - -+ LMB/LMB+- ( 5 ) (12) 2'MB+ (aq) + 2H+ (aq) + 3- + LMB/LMB+.(a) (13) 3MB+ (aq) + MB2+ (aq) + e- or: MB2+ (aq) + 3MB+ (aq) + 2H+ (aq) + 4e- -+ LMB/LMB+* (14) However, there are problems with this explanation: the MB*+ photoelectrochemically generated at applied electrode poten- tials more positive than -0.25 V might well be re-reduced at the same electrode at applied potentials more negative than about - 1 V. Further, while 3MB+ may be oxidized at potentials more positive than -0.25 V, it too may be reduced at the same electrode at applied potentials more negative than 1.24 V.The re-reduction of MB2+ would essentially short-circuit the photoelectrochemistry, while the simultaneous oxidation and reduction of 3MB+ at the same electrode will produce equal, but opposite, photoanodic and photocathodic currents, resulting in no net photocurrent being detected at the MORE. That we do see a photoanodic signal at applied potentials more positive than -0.25 V implies that the reduction of MB2+ and 3MB+ must be inhibited in some manner. One potential source of this inhibition is the inverse Marcus a f f e ~ t . 7 ~ Marcus states that the electrochemical rate constant for a cathodic process, k,, obeys the following approximate proportionality: [A + F ( E - E*')I2 where h is the reorganization energy (J mol-I), E is the applied electrode potential (V) and Ee' the formal electrode potential of the reduction process of interest.Considering the size of the MB+ ion and the diffuse nature of its charge, arising from extensive delocalization, a h value of approximately 0.3 eV (28.9 kJ mol-1) is appropriate (c$, 0.3 eV for the similarly sized methyl viologen73j. Taking E" for the reduction of MB2+ and 3MB+ to be about 1 and 1.24 V, respectively, calculations show that, at applied electrode potentials more negative than +0.2 V, both processes are so inhibited by the inverse Marcus effect as to be kinetically forbidden. Thus, at applied potentials more negative than +0.2 V, "MB+ may be oxidized to MB2+ without complications associated with reductive competition or back- reactions, so giving rise to the photoanodic signal seen at applied potentials more positive than -0.25 V in Fig.7. Chronoamperometry of Methylene Blue on the MORE Dark and light chronoamperograms of MB+ were recorded with the MORE to confirm the presence of a photoanodic current at applied potentials in the range from -0.25 to +0.2 V. Fig. 8 shows the chronoamperometric currents recorded by the MORE from a 0.1 mol m-3 aqueous solution of MB+ in the dark and when illuminated. The potential was stepped from open circuit1786 Analyst, December 1996, Vol. 121 to +0.1 V versus SCE. The current obtained when the potential step was performed in the dark was due to double layer charging and/or the reduction of adventitious impurities derived from the preparation of MB+ used.The current signal obtained when an identical experiment was performed under illuminated condi- tions was significantly different from that obtained in the dark and was due to the oxidation of photogenerated 3MB+ to MB2+ as described by eqn. ( I l ) , confirming the existence of a photoanodic process at +0.1 V. The MORE may therefore be said to be capable of the direct electrochemical detection of photogenerated species with lifetimes of less than 0.1 ms. Cyclic Voltammetry of the Methylene Blue-Fe2+ System using the MORE Photoproducts with longer lifetimes may be obtained on illumination of MB+ in the presence of a suitable sacrificial electron donor such as Fe2+ producing Fe3+ and LMB in accordance with eqn.(6). As an analogous system, that of Th and Fez+, is used as a photogalvanic cell reaction, it was decided to investigate the MB+-Fe2+ system as a model sensitizer/ charge scavenger photosystem. The Fe2+/3+ couple has an Ee of +0.534 V versus SCE.74 Fez+ is therefore capable of reducing photoelectrochemically gen- erated 3MB+ and MB2+ to generate Fe3+ and, ultimately, LMB in the case of the former and MB+ in the case of the latter. The regenerated MB+ may then be photoexcited to 3MB+, where- upon it can react with Fez+ to produce LMB. As discussed above, LMB photogenerated in this fashion will have a lifetime of at least 3 s under the experimental conditions employed in our study and so should be readily detected at the ring electrode of the MORE.Fig. 9 shows the cyclic voltammetric behaviour of the MB+ -Fe2+ system in pH 8 aqueous solution in the absence and presence of illumination, recorded using the MORE. The cathodic and anodic limits of the potential sweep are -0.36 and +0.1 V, respectively, not only for the reasons discussed above, but also because Fe2+ is electrochemically inert over this range. As can be seen from Fig. 9, the cyclic voltammogram of the MB+-Fe2+ system recorded with the MORE in the dark broadly matches that recorded in the absence of Fe2+ (Fig. 7). On illumination with white light, the cyclic voltam- metric characteristics of the MB+- Fe2+ system during the anodic-going reverse sweep are distinctly different from those in the observed in the dark. Specifically, a photoanodic enhancement in the peak corresponding to OX1 is observed, 6 I 3 :h -3 I 30 0 5 10 15 20 25 T ime/s Fig.8 Dark and light chronoamperograms, recorded using a MORE, of 1 mol m--3 MB+ in water containing 50 mol m--3 phosphate buffer (pH 8). In both cases, the applied potential was stepped to +0.1 V versus SCE from floating. The solution was deoxygenated as described in the text. although closer inspection reveals that the observed photocur- rent peak actually occurs at a potential 0.01 V more cathodic than the peak potential of OX1 in the dark, indicating the presence of a new anodic peak, here designated 0x0, derived from the oxidation of some photogenerated species; indeed, OX1 may be seen as a shoulder on 0x0. The photogenerated species responsible for OX0 may derive from one or both of two sources.The first of these two sources has already been discussed: photogenerated LMB. However, as discussed above, the peak corresponding to the re-oxidation of LMB to MB+, referred to as 0x2, occurs at -0.14 V versus SCE, 0.06 V more anodic than the potential of the photoanodic enhancement of OX1 seen in Fig. 9. There are two possible explanations for this observation. The first derives from the origin of the LMB being reduced. In the experiment in Fig, 9, LMB is being photo- generated in solution and diffuses to a gold ring electrode covered with a conducting layer of the LMB/LMB+* mixed- valence salt where it may be readily re-oxidized. Alternatively, LMB may be electrogenerated in the form of an insulating layer over the surface of the gold microelectrode as a result of the solid-state transformation described by eqn.(9). Re-oxidation of LMB at the gold electrode surface is then inhibited by the presence of the overlying LMB molecules; this insulating layer hinders the diffusive transport of counter ions to the electrode surface, resulting in overpotential for the oxidation of LMB that is absent when LMB is photogenerated as a solution species. Overpotentials for the re-oxidation of LMB insulating layers have also been observed on S-modified gold macro- electrodes .5637 The second explanation for the photoanodic enhancement observed in Fig. 9 occurring at a potential 0.06 V more cathodic- than the potential of OX2 requires a more detailed consideration of the mechanism by which LMB is photogenerated.By analogy with Th, the photogeneration of LMB from the MB+- Fez+ system follows the following steps:63 MB+ (aq) hv\ ~ M B + (aq) (16) 3MB+ (aq) + Fez+ (aq) -+ Fe3+ (aq) + MB. (aq) (17) 2LMB+. (aq) -+ LMB (aq) + MB+ (aq) + H+ (aq) (19) which, as an over-all process, may be written as eqn. (6). There- fore, the second possible source of OX0 may be the oxidation of LMB+*, photochemically generated via eqn. (18). It is not possible to establish from the information available whether the oxidation of LMB or LMB+. is responsible for peak 0x0; however, that OX0 derives from the products of a homogeneous photochemical reaction is unambiguous. 200 3 -2004.. ' " " ' " " " ' " " ' ~ " ' " " " " ' " " ' " " ' ' ' ' " ~ -0.3 -0.2 -0.1 0 0.1 Vol tageAl Fig.9 Dark and light cyclic voltammograms, recorded at a scan rate of 0.05 V s-I using a MORE, of a 1 mol m--3 MB+ in water containing 50 mol m--3 phosphate buffer and 1 mol m-3 FeCI3 (pH 8). The starting potential and anodic limit was +0.1 V versus SCE, the cathodic limit was -0.36 V versus SCE and the initial scan was in the direction of decreasing positive potential in each case. The solution was deoxygenated as described in the text.Analyst, December 1996, Vol. 121 1787 Inspection of the current signal recorded during the negative- going forward sweep of Fig. 9 reveals that the photoanodic and photocathodic responses observed at applied potentials more positive and negative than -0.25 V, respectively, from solutions of MB+ in the absence of Fez+ (Fig.7) are absent. The lack of the photoanodic signal, derived from the oxidation of 3MB+ in the absence of Fez+, during the forward sweep in Fig. 9 indicates that nearly all photogenerated 3MB+ reacts with Fez+ before it reaches the ring electrode of the MORE. It should be emphasized that there may indeed be a small photoanodic response at potentials more anodic than -0.25 V in Fig. 9, but that it is beyond the resolution of the experiment. Nonetheless, homogeneous reaction of 3MB+ reduces the rate of the heterogeneous electrochemical generation of MB2+ at the ring electrode, interrupting the mechanism by which the photo- cathodic response observed at applied potentials more cathodic than -0.25 V was generated in Fig. 7. The absence of a photocathodic response at potentials more cathodic than -0.25 V in Fig.9 would seem to corroborate this analysis. However, there would appear to be a small photoanodic response at potentials more cathodic than about -0.24 V in Fig. 9; this is almost certainly due to the local depletion in MB+ concentration close to the electrode surface arising from the reactions described by eqns. (1 6)-( 19). One last point concerning the form of the cyclic voltammo- gram in Fig. 9 needs to be addressed. Fe3+ is generated as a result of the reactions described by eqns. (16)-( 19), and might be expected to give rise to a reduction current at applied electrode potentials more cathodic than about 0.534 V versus SCE; indeed, such a photocathodic response would nullify the photoanodic response described by 0x0.Experiments with Th/ Fe2+ photogalvanic cells63 indicate that, at pH 1, the presence of a layer of thionine at an electrode surface imposes a 0.3 V cathodic overpotential on the reduction of Fe3+, selectively blocking the reaction. Experiments in these laboratories con- firmed that a similar effect is observed with MB+ -coated electrodes. However, the data in Fig. 9 were obtained at pH 8, where Fe3+ precipitates as the ill-defined ‘Fe(OH)3’ spe- cies.74775 It is this precipitation that prevents the re-reduction of Fe3+ at the ring electrode of the MORE and allows the detection of photocurrents from the MB+-Fe2+ system to occur without interference originating from the electrochemistry of the Fe3+ ion. Chronoamperometry of the Methylene Blue-Fe2+ System Using the MORE Fig.10 shows the chronoamperometric currents recorded by the MORE from a 0.1 mol m-3 aqueous solution of MB+ in the presence of a range of concentrations of Fe2+ when illuminated. As in Fig. 8, the potential was stepped from open circuit to +O. 1 V versus SCE. The current signal obtained in the Time/s Fig. 10 Photochronoamperograms, recorded using a MORE, of 0.1 mol m--3 MB+ in water containing SO mol m-3 phosphate buffer at pH 8 and A, 0; B, 10-2; C, 10-1, D, 1; and E, 10 molm-3 FeC13. In each case, the applied potential was stepped to +0.1 V versus SCE from floating. The solutions were deoxygenated as described in the text. absence of Fez+ was due to the oxidation of photogenerated 3MB+ to MB2+ as described by eqn. (1 1). However, as the concentration of Fe2+ ions was increased from 0.01 to 10 molm-3, a concomitant increase in the size of the photo- chronoamperometric response was observed, which may be attributed to the oxidation at the MORE of increasing amounts of solution phase LMB or LMB+*, photogenerated in ac- cordance with the reaction scheme described by eqns.(16)- (19). Conclusions This study has shown that the MORE is a powerful technique for qualitative mechanistic studies of photochemical systems with complex electrochemistry, such as methylene blue. Elec- trochemical studies of MB+ performed with the MORE in the absence of any deliberately added reducing agent in the dark and under illumination allow the determination of photocurrent signals directly attributable to the formation, and subsequent detection, of 3MB+ within the diffusion layer of the illuminated MORE.Analogous studies in the presence of the reducing agent Fe2+ allow the electrochemistry of solution phase LMB or LMB+’ to be investigated and provide important insights into the differences between the homogeneous and heterogeneous electrochemistry of methylene blue. The results presented in this paper were obtained using relatively large MOREs (fibre-optic radius = 1.25 X m, ring thickness = 600 nm) and demonstrate that we can easily detect electroactive species with lifetimes of < 9 X 10-5 s, suggesting the study of photochemically important species such as ‘ 0 2 in aqueous and biological systems (lifetime 4.2 X 10-6 s) may be feasible with smaller microelectrodes. The MORE both obviates many of the difficulties associated with, and is more generally applicable than, existing photo- electrode devices.It is expected that, with further instrument development, it will be possible to use the MORE quantitatively as well as qualitatively. Indeed, by use of a monochromator, it may be that the energy of incident light could be tuned to the specific absorbance wavelength of photoelectroactive species in complex mixtures, providing a high level of selectivity. The authors thank the University of Central Lancashire for a bursary, providing a postgraduate research studentship for G.I.P., and the Royal Society for an equipment grant. They also thank Dr. Christopher G. Phillips (Imperial College, London) for helpful discussions regarding the nature of mass transport at ring microelectrodes, the Electron Microscopy Facility at the University of Brighton for help with the manufacture of the MOREs and one of the referees for drawing their attention to ref.32. References Oldham, K. B., in Microelectrodes: Theory and Applications, NATO ASI Ser. E: Appl. Sci., ed. Montenegro, M. I., Queiros, M. A., and Daschbach, J. L., Kluwer, Dordrecht, 1991, vol. 197, pp. 33-50. Phillips, C. G., and Jansons, K. M., Proc. R. Soc. London, Sev. A, 1990,428,431. Andrieux, C. P., Hapiot, P., and SavCant, J. M., Electroanalysis, 1990, 2, 183. Justice, J. B., Voltammetry in the Neurosciences, Humana Press, Clifton, NJ, 1987. Bond, A. M., Oldham, K. B., and Zoski, C. G., J. Electroanal. Chem., 1988, 235, 71. Hoffman, M. R., Martin, S.T., Choi, Y., and Bahnemann, D. W., Chem. Rev., 1995,95, 69. Linsebigler, A. L., Lu, G., and Yates, J. T., Jr., Chem. Rev., 1995,95, 73s. Kamat, P. V., Chem. Kev., 1993, 93, 267.1788 Analyst, December 1996, Vol. 121 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Hagfeldt, A., and Gratzel, M., Chem. Rev., 1995, 95, 49. Proceedings of the Symposium on Water Purification by Photo- catalytic, Photoelectrochemical and Electrochemical Processes, eds. Rose, T. L., Rudd, E., Murphy, O., and Conway, B. E., Electro- chemical Society, Pennington, NJ, 1994, pp. 236-37 1. Wang, Y., Acc. Chem. Res., 1991, 24, 133. Tuite, E. M., and Kelly, J. M., J . Photochem. Photohiol., B, 1993,21, 103. Phillips, C. G., and Stone, H.A., J . EEectroanal. Chem., 1995, 396, 277. Zhuang, Q.-K., and Chen, H.-Y., Electroanalysis, 1994, 6, 485. Myland, J. C., and Oldham, K. B., J . Electroanal. Chem., 1993,347, 49. Birkin, M. V., Bulhoes, L. 0. S., and Bard, A. J., J . Am. Chem. Soc., 1993, 115, 201. Howell, J. O., and Wightman, R. M., J . Phys. Chem., 1984, 88, 3915. Philips, M. E., Deakin, M. R., Novotny, M. V., and Wightman, R. M., J . Phys. Chem., 1987, 91, 3934. DeAngelis, T. P., and Heineman, W. R., J . Chem. Educ., 1976, 53, 594. Kozlowski, M., Smyrl, W. H., Atanasoska, Lj., and Atanasoski, R., Electrochim. Acta, 1989, 34, 1763. Tyler, P. S., Kozlowski, M. R., Smyrl, W. H., and Atanasoski, R. T., J . Electroanal. Chem., 1987, 237, 295. Kozlowski, M. R., Tyler, P. S., Smyrl, W. H., and Atanasoski, R.T., Electrochim. Acta, 1988, 194, 505. Carlsson, P., Holmstrom, E., Uosaki, K., and Kita, H., Appl. Phys. Lett., 1988, 53, 965. Kucernak, A. R. J., Peat, R., and Williams, D. E., J.Electrochern. Soc., 1991, 138, 1645. Eriksson, S., Carlsson, P., Holmstrom, B., and Uosaki, K., J. Appl. Phys., 1991, 69, 2324. Compton, R. G., Eklund, J. C., and Nei, L., J . Electroanal. Chem., 1995, 381, 87. Albery, W. J., Archer, M. D., and Egdell, R. G., J . Electroanal. Chem., 1977,82, 199. Bartlett, P. N., and Deards, P., presented at Electrochem '94, Edinburgh, September 1994. Albery, W. J., Bartlett, P. N., Lithgow, A. M., Riefkohl, J., Romero, L., and Souto, F. A., J . Chem. SOC., Faraday Trans. I , 1985, 81, 2647. Williams, D. E., Kucernak, A. R. J., and Peat, R., Electrochim.Acta, 1993, 38, 57. Kucernak, A. R. J., Peat, R., and Williams, D. E., Electrochim. Acta, 1993, 38, 71. Hutton, R., and Williams, D. E., Anal. Chem., 1995, 67, 280. Cohen, C. B., and Weber, S. G., Anal. Chem., 1993,65, 169. Casillas, N., James, P., and Smyrl, W. H., J . Electrochem. Soc., 1995, 142, L16. Symanski, J. S., and Bruckenstein, S., J . Electrochem. Soc., 1988, 135, 1985. Szabo, A., J . Phys. Chem., 1987,91, 3108. Kuhn, L. S., Weber, A., and Weber, S. G., Anal. Chem., 1990, 62, 1631. Kriger, M. S., Cook, K. D., and Ramsey, R. S., Anal. Chem., 1995,67, 385. Wightman, R. M., and Wipf, D. O., in Electroanalytical Chemistry, ed. Bard, A. J., Marcel Dekker, New York, 1989, vol. 15, pp. 267- 3.51. Zak, J., and Kuwana, T., .J. Electroanal. Chem., 1983, 150, 645. Kalapathy, U., Tallman, D. E., and Hagen, S., J . Electroanal. Chem., 1992, 325, 65. Tallman, D. E., Anal. Chem., 1994, 66, 557. 43 44 45 46 47 48 49 so 51 52 53 54 5.5 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 Fleischmann, M., Bandyopadhyay, S., and Pons, S., J . Phys. Chem., 1985,89, 5537. Fleischmann, M., and Pons, S., J . Electroanal. Chem., 1987, 222, 107. Smythe, W. R., J . Appl. Phys., 195 1 , 22, 1499. Wopschall, R. H., and Sham, I., Anal. Chem., 1967, 39, 1527. Papeschi, G., Costa, M., and Bordi, S., J . Electrochem. Soc., 1981, 128, 1518. Svetlicic, V., Tomaic, J., Zutic, V., and Chevalet, J., J . Electroanal. Chem., 1983, 146,71. Svetlicic, V., Zutic, V., Clavilier, J., and Chevalet, J., J . Electroanal. Chem., 1985, 195, 307. Sagara, T., and Niki, K., Langmuir, 1993, 9, 831. Zutic, V., Svetlicic, V., Clavilier, J., and Chevalet, J., J . Electroanal. Chem., 1987,219, 183. Shez, E. I., and Corn, R. M., Electrochim. Acta, 1993, 38, 1619. Kuwabata, S., Nakamura, J., and Yoneyama, H., J . Electroanal. Chem., 1989, 261, 363. Guadalupe, A. R., Liu, K. E., and AbruAa, H. D., Electrochim Actu, 1991, 36, 881. John, S. A., and Ramaraj, R., J . Chem Soc., Faraday Trans., 1994. 90, 1241. Svetlicic, V., Zutic, V., Clavilier, J., and Chevalet, J., J . Electroanal. Cltem., 1987, 233, 199. Clavilier, J., Svetlicic, V., Zutic, V., Ruscic, B., and Chevalet, J., J . Electroanal. Chem., 1988, 250, 427. Lezna, R. O., de Tacconi, N. R., Hahn, F., and Ariva, A. J., J . Electroanal. Chem., 1991, 306, 259. Earner, B. J., and Corn, R. M., Langmuir, 1990, 6, 1023. Naujok, R. R., Duevel, R. V., and Corn, R. M., Langmuir, 1993, 9, 1771. Svetlicic, V., Clavilier, J., Zutic, V., Chevalet, J., and Elachi, K., J. Electroanal. Chem., 1993, 344, 145. Danzinger, R. M., Bar-Eli, K. H., and Weiss, K., .I. Phys. Chem., 1967, 71, 2633. Albery, W. J., and Foulds, A. W., J . Photochem., 1979, 10, 41, and- references cited therein. Bauldreay, J., and Archer, M. D., Electrochim. Acta, 1983, 28, 1515. Faure, J., Bonneau, R., and Joussot-Dubien, J., Photochem. Photo- hid., 1967, 6, 331. Unwin, P. R., and Bard, A. J., Anal. Chem., 1992, 64, 113. Murthy, A. S. N., and Reddy, K. S., J . Chem. Soc., Faraday Trans. 1. 1984,80, 2745. Chen, X., Zhuang, J., and He, P., .I. Electroanal. Chem., 1989, 271, 257. Ju, H., Zhou, J., Cai, C., and Chen, H., Electroanalysis, 1995, 7, 1165. Karyakin, A. A., Strakhova, A. K., Karyakina, E. E., Varfolomeyev, S. D., and Yatsimirsky, A. K., Bioelectrochem. Bioenerg., 1993, 32, 35. Beddard, G. S., Kelly, J. M., and van der Putten, W. J. M., J . Chem. Soc., Chem. Commun., 1990, 1346. Marcus, R., Annu. Rev. Phys. Chem., 1964, 15, 155. Duonghong, D., Ramsden, J., and Gridtzel, M., J . Am. Chem. Soc., 1982,104,2977. Kelsall, G. H., and Williams, R. A., J . Electrochem. Soc., 1991, 138, 931. Zhao, Z., Boxall, C., and Kelsall, G. H., Colloids Surf, 1993, 73, 145. Paper 6103942A Received June 5,1996 Accepted September 5, I996

ISSN:0003-2654    

DOI:10.1039/AN9962101779

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 ( I 789-1 793) 1789 Electrochemical Studies of Zinc in Zinc-Insulin Solution* Rui M. Barbosaa,b Luis M. Rosarioa,c Christopher M. A. Brettd,? and Ana Maria Oliveira Brettd,? a Centro de Neurocitncias de Coimbra, Universidade de Coimbra, 3049 Coimbra, Portugal Universidade de Coimbra, 3049 Coimbra, Portugal Universidade de Coimbra, 3049 Coimbra, Portugal d Departamento de Quimica, Faculdade de Citncias e Tecnologia, Universidade de Coimbra, 3049 Coimbra, Portugal Laboratbrio de Me'todos Instrumentais de AncElise, Faculdade de Farmucia, Departamento de Bioquimica, Faculdade de Citncias e Tecnologia, The electrochemical determination of zinc arising from zinc-insulin complexes was investigated and it was demonstrated that zinc in zinc-insulin solution can be measured in the presence of dissolved oxygen by square-wave anodic stripping voltammetry (SWASV) at mercury thin-film electrodes on glassy carbon disc minielectrode and cylindrical carbon fibre microelectrode substrates. Reoxidation signals arise from complexed zinc at low insulin concentrations (< 100 nmol 1-*) and from labile zinc at higher concentrations; the latter can be quantified through linear calibration curves.Batch injection analysis with SWASV was successfully tested for the determination of zinc in zinc-insulin solutions in small sample volumes. Since intracellularly stored insulin exists in the form of a zinc-insulin complex, these techniques are very promising for the indirect study of insulin release from pancreatic P-cells.Keywords: Zinc; insulin; square-wave anodic stripping voltammetry; microelectrodes; batch injection analysis Introduction The polypeptide hormone insulin is synthesized by the pancreatic 0-cells. It is stored within these cells in vesicles and released into the extracellular fluid in response to a rise in blood glucose levels. ,2 Impaired insulin release is an early and key defect in type 2 diabetes.3 The insulin monomer consists of two chains, A and B, which are linked by two disulfide bridges; a third disulfide bridge links two parts of the A chain, (Fig. 1).4,5 In the P-cell vesicles insulin exists in a crystalline form, mainly of insulin hexamers encapsulating two zinc ions.6 Glucose or other stimuli in the extracellular fluid cause these vesicles to move towards the cell membrane and fuse with it, releasing the insulin complex by exocytosis.On release, owing to the decrease in concentration, the complex rapidly transforms into the monomer form plus free zinc ions. There is much interest in being able to follow this process, or at least part of it, in real time. Since zinc ions are released together with insulin during exocytosis, the instanta- neous concentration of extracellular zinc close to the cell membrane may reach a few micromolar during stimulated secretion.7.8 Insulin is most often determined using highly sensitive bioassay methods such as radioimmunoassay9 or ELISA. * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29, 1996. + To whom correspondence should be addressed.However, these methods are laborious, do not have any spatial resolution and do not permit the determination of insulin concentrations in real time. Spatial resolution can be achieved through the use of microprobes of sufficiently small dimen- sions: voltammetry has been used for in vivo or in vitro measurements of neurotransmitters and hormone release from brain slices or single cells using carbon fibre microelec- trodes.11-14 The known electrochemistry of insulin has been dominated by the study of the reduction/oxidation behaviour of disulfide bridges and associated adsorption phenomena. 15-1 8 Further, although the studies have been undertaken in the negative potential range, preferably on mercury in order to increase the negative potential limit, they should be carried out in the presence of oxygen in order to mimic the real situation in vitro or in vivo.Hence some widely used methods for trace determination, such as adsorptive stripping voltammetry, could not readily be employed. However, potentiometric stripping analysis, with oxygen as oxidant of pre-adsorbed reduced insulin species, has recently been investigated.19 There are two ways to approach the monitoring of insulin concentrations. One involves monitoring insulin itself14.1Y and the other, not previously explored to our knowledge, consists in measuring the zinc concentration using square-wave anodic stripping voltammetry (SWASV). SWASV offers a rapid and sensitive method for trace metal determinations in the presence of oxygen, particularly when used at mercury thin-film electrodes on solid substrates such as carbon.These electrodes, if correctly prepared, have good mechanical stability and lifetime.20 In this work, three approaches to the determination of zinc associated with insulin were investigated, employing mercury thin-film electrodes at carbon substrates. First, studies on a conventionally sized glassy carbon substrate were carried out. Second, carbon fibre microelectrodes were used with a view to insertion in the extracellular fluid. Finally, the batch injection technique for the analysis of small sample volumes injected into an inert electrolyte was tested.21-23 Experimental Bovine pancreas insulin was purchased from Sigma (I 5500, zinc content approximately 0.5%). A stock standard solution of insulin was prepared in a similar way to the prccedure of Cox and Gray.24 The insulin was dissolved in 0.1 moll-' NaOH and then diluted to give an insulin concentration of 10-3 mol I-' in 0.01 mol 1-1 NaOH. This solution was divided into 0.5 ml aliquots and stored at -20 "C until used. The specified zinc content was checked by AAS, which gave 0.49%.I790 Analyst, December 1996, Vol.121 All other reagents were of analytical-reagent grade and solutions were made with Milli-Q ultrapure water of resistivity > 18 Mi2 cm (Millipore Intertech, Bedford, MA, USA). The electrolyte employed was 0.15 mol 1-1 physiological phosp- hate-buffered saline (PBS) of pH 7.4, containing 8.0 g of NaC1, 0.2 g of KCI, 1.44 g of Na2HP04 and 0.24 g of KH2P04 per litre of solution. All experiments were conducted at room tem- perature (22-23 "C).Mercury thin film electrodes were pre-formed on a glassy carbon disc electrode substrate (diameter 1 mm) in a glass shroud and a carbon fibre cylinder microelectrode substrate (diameter 8 pm) in a glass capillary, made in-house with the aid of a pipette puller.25 Experiments with microelectrodes were performed within a Faraday cage. Batch injection experiments were carried out using a mercury thin-film electrode pre-formed on glassy carbon disc electrode substrate (diameter 5 mm) in a Kel-F sheath. Injection was performed according to Brett et a1.21322 using a Rainin (Woburn, MA, USA) EDP-Plus 100 programmable electronic micro- pipette at a dispension rate of 22.7 pl s-'; the internal diameter of the micropipette tip was 0.47 mm.The cell design has been described previously.22 The mercury thin-film electrodes were formed from a solution of moll-1 Hg2+ in 0.1 moll-' KN03-5 mmoll-1 HN03 by electrodeposition at - 1 .O V for 5 min. Experiments were carried out in a minicell of capacity 2 cm3 with incorporated stirrer, using an Ag/AgCl (3 mol 1-1 KC1) reference electrode and a Pt wire auxiliary electrode. A PC-controlled EG&G PAR273 potentiostat (EG&G Prin- ceton Applied Research, Princeton, NJ, USA) with appropriate software was employed. Some preliminary cyclic voltammo- grams were obtained using a Metrohm (Herisau, Switzerland) Model 663VA stand with a hanging mercury drop electrode (HMDE) and a Cypress OMNI potentiostat (Cypress Systems, Lawrence, KS, USA).Results and Discussion The insulin molecule contains three disulfide bridges (Fig. 1) and the reduction electrochemistry of the insulin molecule concerns the reduction of these groups. The insulin solution employed in this study contained approximately 0.43 zinc ions per molecule of insulin. This ratio is greater than the 0.33 required to form the 2Zn-insulin crystalline hexamer structure. However, more zinc can associate with the insulin hexamer,26 but the bonding is weaker and these zinc ions are therefore more labile. Fig. 2(a) shows a cyclic voltammogram at a hanging mercury drop electrode of a deoxygenated zinc-containing insulin solution at a concentration for which the hexamer form is predominant. Adsorption occurs through two of the sulfide bridges, resulting in mercuric insulin thiolate, designated In(SHgS)2, or mercurous insulin thiolate, h~(sHgHgS)~, de- pending on the potential, the former at potentials greater than 0 V and the latter at potentials of about 0.3 V.I9 The first of the two reduction peaks Fig.2(a) can be ascribed to reduction and breakage of adsorbed disulfide bonds15 involving four protons and four electrons to form sulfhydryl groups, which can be written as III(SH~H~S)~ + 4H+ + 4e- -+ In(S€-ISH)2 + 4Hg and which leads to insulin denaturation. Partial reoxidation of these sulfhydryl groups can occur so long as they are in the correct relative position on the mercury surface, as seen in the anodic scan. Fig. 2(b) shows cyclic voltammograms in zinc chloride solutions recorded under the same conditions. Compar- ing the voltammograms in Fig.2(a) and (b), it can be deduced that the second reduction peak for the insulin solution in Fig. 2(a) is related to the reduction of the complexed Zn2+ and the anodic peak at -0.85 V on sweep reversal to zinc re- oxidation. The differences in the peak potentials in the two voltammograms reflect the different environments of the zinc ions, free or complexed. A cyclic voltammogram of Zn2+ after addition of insulin (not shown) led to a large decrease in the oxidation peak of zinc previously reduced in the mercury. A possible explanation is that insulin molecules block the surface to access from the V , " . ' " " " . , , ' " 1 1 1 1 1 1 , , , -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 -1.4 -0.2 -0.4 -0.6 -0.8 -1.0 -1.2 -1.4 EIV Fig. 2 Cyclic voltammograms on HMDE in 0.15 mol 1-* PBS (pH 7.4); scan rate, 0.1 V s-1.(a) 2.5 x 10-5 niol 1-1 zinc-insulin, first and ninth (steady-state) scans; and (6) ZnC12, concentrations 6 , 12 and 18 pmol l-I. A chain Fig. 1 complex. Reproduced with permission from ref. 4, p. 34. Structure of insulin entities. (a) Primary structure; and (b) structure of the 2Zn-insulin hexamer; the zinc ions are in the centre of theAnalyst, December 1996, Vol. 121 1791 outside bulk solution, the zinc measured coming almost entirely either from labile zinc within the zinc-insulin complex or from free zinc in bulk solution which is able to pass through the adsorbed insulin layer, Hence electrochemical investigations in a restricted potential zone round -0.8 to - 1.3 V reflect the amount of labile free and complexed zinc.Cyclic voltammetry shows that this is significantly less than that predicted from the bulk insulin concentration. In order to lower the detection limit as much as possible and permit a large insulin concentration range to be probed, SWASV was employed. This has the additional advantage that experiments can be performed in the presence of oxygen as occurs in natural systems, All results presented below were obtained at mercury thin film electrodes (MTFE) on carbon substrates, either glassy carbon or carbon fibre. A deposition time, t&p, of 120 s was usually employed. In the square-wave stripping scan a square wave amplitude, h, of SO mV, a square-wave frequency& of 100 Hz and a scan increment of 2 mV were employed, which correspond to an effective scan rate of 200 mV s-1.The effect, in general terms, of changing oxygen concentra- tion on the square-wave peaks in SWASV experiments is predicted to be zero provided that the deposition time is sufficiently long in order to reduce all oxygen in the vicinity of the electrode surface during the preconcentration step and that the square wave scan is carried out sufficiently rapidly such that there is no time for more oxygen to diffuse to the electrode surface. For the experimental conditions and square-wave parameters used in this work, it was found that a 10 s deposition time is necessary to ensure these conditions, much less than the 120 s used in most of the work described in this paper. However, even for deposition times less than 10 s and for the potential range studied, the effect on zinc peak heights is minimal, the influence of oxygen being manifested by changes in the baseline signal.The viability of performing SWASV of zinc in zinc-insulin solution is shown by the results in Fig. 3, with the zinc re- oxidation peak appearing at exactly the potential expected in simple matrices. Zinc was preconcentrated at the MTFE from insulin solutions of increasing concentration using a 120 s deposition time. A linear calibration curve was obtained following the equation I , = 0.422 + 0.966[insulin] ( r = 0.998), where the current is expressed in FA. Fig. 4 shows a series of voltammograms obtained by S WASV, registered at lower insulin concentrations than those 4.0 1 / I 2.5 1 111 \\I 1.0 ' I I I I 1 -1.3 -1.2 -1.1 -1.0 -0.9 -0.8 EIV Fig.3 SWASV of zinc from zinc-insulin at MTFE on a glassy carbon substrate (diameter 1 mm) in 0.15 moll-' PBS. [Insulin] : 0.5, 1.0,2.0 and 5.0 ymol 1- ; peak height increases with increasing concentration; tdep = 120 s with stimng. SW parameters: frequency, f = 100 Hz; amplitude, h = SO mV; scan increment = 2 mV. in Fig. 3. Successive additions of insulin were made starting at the lowest value. The re-oxidation peak of zinc from the zinc- insulin complex in the cyclic voltammogram in Fig. 2(a) was at -0.9 V; the -0.9 V peak in Fig. 4 can therefore be identified as being due to the same process. At concentrations in the submicromolar range, insulin should exist predominantly in the monomeric form and the zinc ions should therefore be free to move by diffusion and migration.However, adsorption of the insulin molecule on the electrode surface will cause a local increase in concentration in its vicinity and, indeed, may permit the formation of the 2Zn-insulin complex on the surface. Understanding Fig. 4 can be aided by consideration of the peak current versus insulin concentration plots in Fig. 5(a) for the peaks at -0.9 V and - 1.1 V. The peak current at -0.9 V, which we previously associated with complexed zinc, increases from zero to a maximum value at approximately 100 nmol 1-1 insulin and then becomes smaller, eventually disappearing for concentrations higher than 700 nmol 1 - I . The peak current at - 1. I V, which we can associate with free zinc, starts appearing when the concentration of insulin is increased above 100 nmol 1-1 and then increases linearly in magnitude.Fig. S(b) shows the integral of the peak current of the -0.9 V peak versus 2.1 2*2 1 I I I I I , .. ..._._.. t -1.4 -1.3 -1.2 -1.1 -1.0 -0.9 -0.8 -0.7 1.5 ' EIV Fig. 4 SWASV of zinc from zinc-insulin at MTFE on a glassy carbon substrate (diameter 1 mm) in 0.15 mol 1-1 PBS. [Insulin]: A, 5.0; B, 20.0; C, 100; D, 292; and E, 520 nmol 1-1. Experimental conditions as in Fig. 3. 0.6 - 0.5 - 0.4 - 2 0.3 - 0.2 - 0.1 - \ 0 0 0.0 1 I I I I I 0 100 200 300 400 500 600 1 .o 0.8 2 w 0.6 w .* d 2 0.4 0 0.2 0.0 z- 0 100 200 300 400 500 600 700 [1nsulin]/nrnol1-' 2 800 Fig. 5 (a) Plots of Zp versus [insulin]: (e) peak at -0.9 V; and (0) peak at - 1.1 V. (b) Plot of area under Zp versus [insulin] curve versus [insulin] for -0.9 V peak (normalized units of charge).1792 Analyst, December 1996, Vol. 121 [insulin] curve depicted in Fig.5(a), which is proportional to the charge transferred. As can be readily seen, this follows the form of an isotherm and saturation occurs by 700 nmol l-1. Analysis of the form of the isotherm, assuming monolayer formation, shows a higher fractional coverage than that predicted by the Langmuir isotherm, suggesting attractive interaction between the adsorbed insulin entities. We can develop a model based on these results as follows. Insulin adsorbs on the surface as a complex with zinc, the extent of coverage, at low concentration, being concentration depend- ent. The zinc ions within the complex are reduced in the SWASV preconcentration step and then re-oxidized and released to bulk solution during the square-wave scan.On increasing the insulin concentration, there is further adsorption of insulin on the electrode surface and these freshly adsorbed complexes provide more zinc, which can be reduced. Even- tually, the surface becomes saturated with complex and the re- oxidation signal at -0.9 V diminishes to zero. By this time, zinc from bulk solution, where the amount of free zinc is already appreciable, and which can reach the electrode through gaps between adsorbed complexes, is increasing to give large signals. If this explanation is correct, then the integral of the variation of peak current with concentration in Fig. S(a) should have the form of an isotherm.This was indeed the case. Since it is intended to probe extracellular fluid directly, further studies were directed towards the utilization of small electrodes and small sample volumes. Experiments were performed at mercury thin films on cylindrical carbon fibre microelectrodes. The influence of accumulation time on the peak height for zinc re-oxidation, see (Fig. 6), was investigated for a zinc-insulin concentration of 1 pmol 1- 1. Maximum response, corresponding to saturation, was reached at an accumulation time of 15 min. A deposition time of 120 s was chosen, which is well within the linear region of this curve and permits easier comparisons between results obtained at macro- and microelectrodes. Using a deposition time of 120 s [Fig. 7(a)], the insulin concentration was varied and the calibration curve in Fig.7(b) was obtained. The form of this curve is reproducible and the linear region (0.1-3.55 pmol 1-1) can be readily used for concentration determination: I,, = 3.65 1 + 3.638Linsulinl (Y = 0.9995), where current is expressed in nA. Here there is evidence of saturation effects followed by some sort of overriding inhibition process, which is probably the formation of a multilayer adsorbate. As is well known, the concentration gradient at microelectrodes is much higher than that at macroelectrodes, which leads to significantly higher current densities, and it may be that this is responsible for multilayer adsorption. This unusual signal diminution appears similar to that obtained at macroelectrodes (Fig. 5), but in fact refers to different peaks: at macroelectrodes it refers to the peak at 25.0 20.0 -2 I5.O -.10.0 +. 0.0 I I I 1 0.0 5.0 10.0 15.0 20.0 25.0 30.0 tdep/min Plot of peak current, I,, versus t&p from SWASV of 1.0 pmol 1-1 Fig. 6 . . _._- ....y..l:.. h A T C C -_ ,-.n.-l.n_ +':I...-- m,.l.m+-ntP /,4:n-P+P+- Q I,-. IPnmth 150 pm). Experimental conditions as in Fig. 3. -0.9 V. Using microelectrodes we were not able to reach sufficiently low insulin concentrations to see the peak at -0.9 V. The detection limit obtained for Zn-insulin solution with respect to the peak at - 1.1 V is 100 nmol l-1 similar to the value obtained with the minielectrode [Fig. 5(a)]. The batch injection technique, which permits the analysis of small sample volumes (6100 pl) by anodic stripping voltam- metry,23 was investigated as an alternative to micro- electrodes. Typical results for various insulin concentrations are given in Fig.8. As can be seen, the zinc re-oxidation signals appear as expected but the sensitivity is lower than with microelectrodes. Additionally, the background square-wave scan is less flat as the applied potential becomes more positive compared with the other experimental configurations described above. Nevertheless, it is an interesting alternative for situations in which it is desired to analyse small sample volumes rapidly. Finally, the minielectrode (diameter 1 mm) was applied to determine the zinc concentration in the extracellular fluid of an insulinoma cell line (data not shown). This preliminary experiment showed differences between the zinc signal before and after stimulation of insulin release, which augurs well for application of the electrochemical approach to in situ investi- gations of pancreatic (3-cells.Conclusions It has been demonstrated that zinc in zinc-insulin can be measured in oxygen-containing solution by S WASV at mercury thin-film electrodes. The signal obtained results from com- plexed zinc at low insulin concentrations (< 100 nmol l-1) and from labile zinc at higher concentrations. Microelectrodes and 30.0 d 3 25.0 * 20.0 10.0 I5'O i I 1 I 1 I -1.3 -1.2 -1.1 -1.0 -0.9 -0.8 EIV 20.0 15.0 ... 2 10.0 *a 5.0 0.0 0.0 5.0 10.0 15.0 20.0 25.0 [~nsulin]/pmo~ I-' Fig. 7 (a) SWASV of a solution of zinc-insulin at MTFE on a carbon fibre substrate (diameter 8 pm; length 150 pm).Experimental conditions as in Fig. 3. [Insulin]: 0.10, 0.59, 1.58 and 3.55 pmol 1-l; peak height increases with increasing concentration. (b) Plot of peak current, I p , versus [insulin] from data in (a).Analyst, December 1996, Vol. 121 1793 0.140 r 0.125 d E - 0.120 -. 0.115 0.1 10 0.105 ' I L I I I -1.3 -1.2 -1.1 -1.0 -0.9 -0.8 EIV Fig. 8 Batch injection analysis of insulin at MTFE on a glassy carbon substrate (diameter 5 mm). [Insulin]: 5.0, 10.0, 50.0 and 100.0 pmol 1-I; peak height increases with increasing concentration. Injections of 100 pl of solution. Other experimental conditions as in Fig. 3. batch injection analysis can be employed for the determination of zinc in small sample volumes. This technique is very promising for application to the determination of zinc arising from the zinc-insulin complex released into the extracellular fluid during exocytosis from pancreatic @-cells.References Hedeskov, C. J., Physiot. Rev., 1980, 60, 442. Orci, L., Vasssali, J. D., and Perrelet, A., Sci. Am., 1988, 260, 85. Efendic, S., Khan, A., and Ostenson, C. G., Diahete Metah., 1994,20, 81. Derewenda, U., Dercwenda, Z. S., Dodson, G. G., and Hubbard, R. E., in Insulin, ed. Cuatrecasas, P., and Jacobs, S., Springer, Berlin, 1990, ch. 2. Howell S. I., and Tyhurst, M., in The Secretory Process, ed. Poisener, A. M., and Trifarii, J. M., Elsevier, Amsterdam, 1982, vol. 1, ch. 4. 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 2s 26 Blundell, T., Dodson, G., Hodgkin, D., and Mercola, D., in Advances in Protein Chemistry, ed.Anfinsen, C. B., Jr., Edsall, J. T., and Richards, F. M., Academic Press, New York, 1972, pp. 279402. Ferrer, R., Soria, B., Dawson, C. M., Atwater, I., and Rojas, E., Am. J . Physiol., 1984, 246, C520. Perez-Armendariz, E., Atwater, I., and Rojas, E., Biophys. J., 1985, 48, 741. Hales C. N., and Randle, P. J., Biochem. J., 1963, 88, 137. Kekow, J., Ulrichs, K., Muller-Ruchholtz, W., and Gross, W. L., Diabetes, 1988, 37, 321. Millar, J., Slamford, J. A., Kruk, Z. L., and Wightman, R. M., Eur. J . Pharm., 1985,109, 341. Leszczyszyn, D. J., Jankowski, J. A., Viveros, 0. H., Diliberto, E. J., Near J. A., and Wightman, R. M., J . Neurochem., 1991, 56, 1855. 0' Neill, R. D., Analyst, 1994, 119, 767. Huang, L., Shen, H., Atkinson, M. A., and Kennedy, R. T., Proc. Natl. Acad. Sci. USA, 1995,92,9608. Stankovich, M. T., and Bard, A. J., J . Electroanal. Chem., 1977,85, 173. Trijueque, J., Sanz, C., Monlebn, C., and Vicente, F., J . Electroanal. Chem., 1988, 251, 173. Trijueque, J., and Vicente, F., An. Quim., 1990, 86, 538. Trijueque, J . , Vicente, F., Martinez, F., and Vera, J., Port. Electrochim. Acta, 199 1, 9, 399. Honeychurch, M. J., and Ridd, M. J., Electroanalysis, 1996, 8, 49. Wojciechowyki, M., and Balcerzak, J., Anal. Chem., 1990, 62, 1325. Brett, C. M. A., Oliveira Brett, A. M., and Mitoseriu, L. C., Anal. Chem., 1994,66, 3145. Brett, C. M. A., Oliveira Brett, A. M., and Mitoseriu, L. C., Electroanalysis, 1995, 7, 225. Brett, C. M. A., Oliveira Brett, A. M., and Tugulea, L., Anal. Chim. Acta, 1996, 322, 151. Cox, J. A., and Gray, T. J., Anal. Chenz., 1989, 61, 2462. Stamford, J. A., Palij, P., Davidson, C., Jorm, C. M., and Phillips, P. E. M., in Neuromethods, ed. Boulton, A,, Baker, G., and Adams, R. N., Humana Press, Clifton, NJ, 1995, 27, pp. 81-1 16. Blundell, T., Dodson, G., Hodgkin, D., and Mercola, D., in Advances in Protein Chemistry, ed. Anfinsen, C. B., Jr., Edsall, J. T., and Richards, F. M., Academic Press, New York, 1972, p. 325. Paper 6103650C Received May 28, 1996 Accepted July 25, 1996

ISSN:0003-2654    

DOI:10.1039/AN9962101789

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. I21 (179.5-1799) 1795 Titrations With Electrogenerated Halogens in the Diffusion Layer of an lnterdigitated Microelectrode Array* DuSan Bustin, Stanislav Jursa and Peter TomEik Department of Analytical Chemistry, Slovak Technical University, Radlinskiho 9, 812 37 Bratislava, Slovak Republic A technique based on diffusion layer titration was developed for the iodimetric determination of low concentrations of thiosulfate and the bromimetric determination of allyl alcohol. The diffusion layer titrations utilize chemical reactions proceeding quantitatively only in the close vicinity of the electrode. One set of segments of an interdigitated array (IDA) microelectrode serves for galvanostatic anodic generation of titrant (iodine or bromine) and the second set, immersed in the diffusion layer of the generator, detects its unreacted flux.The detector (collector) is potentiostated to the potential of the limiting diffusion current of iodine/bromine cathodic reduction. The diffusion layer titration curves (collector current versus generator current plots) measured ‘point by point’ or by slowly scanning the generator current, show very good reproducibility. Since no bulk phase chemical reaction actually proceeds, the experiment can be repeated extensively in the same solution. The sensitivity of this method is 1424 yA 1 mol-1 and the determination limit is 6 X 10-7 mol 1-1 for thiosulfate determination. The substantially lower sensitivity compared with the rotating ring-disc diffusion layer titration is compensated for by the possibility of a many-fold decrease in the sample volume for IDA microelectrode diffusion layer titration.The technique was applied to the trace determination of thiosulfate in analytical-reagent grade potassium iodide. The thiosulfate content found was slightly lower than that specified by the manufacturer. A sensitivity of 486 yA 1 mol-l and a determination limit of 2 x 10-5 mol 1-1 were found for the bromimetric determination of allyl alcohol. Analytically favourable titration curves with negligible curvature around the end-point were obtained in IDA diffusion layer titrations, in contrast to the rotating ring-disc electrode measurement where considerable curvature was encountered owing to the lower rate of the bromination of allyl alcohol.The absence of almost any curvature in IDA experiments is explained by a significantly slower allyl alcohol flux compared with convective diffusion at a rotating electrode. The slower flux results in more time for the titration reaction to proceed close to equilibrium. Keywords: Diffusion layer titration; iodimetry ; hromimetry; thiosulfate; allyl alcohol; interdigitated microelectrode arrays Introduction Diffusion layer titrations are based on the chemical reactions proceeding quantitatively in the close vicinity of the electrode. The extent of the reactions in the bulk phase is negligible or * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29, 1966. zero. The electrode has the function of (i) a generator of titrant,1,2 (ii) a generator of titrand3,4 or (iii) a generator of redox ~ a t a l y s t , ~ - ~ enabling the titration reaction to proceed at a sufficiently high rate.The type (i) diffusion layer titration, which is dealt with in this paper, can be regarded as analogous to coulometric titration with internal generation of titrant. Similarly to coulometric titration, this diffusion layer titration also requires, in addition to the titrant-generating electrode, another independent electrode which would indicate (e.g., amperometrically) the extent of titration reaction monitoring the concentration of titrant. In the special case of diffusion layer ti tration, the amperometric indication electrode should be located close to the electrode generating titrant to ensure a high detection ability, i.e., titrant collection efficiency.If the titration reaction proceeds quantita- tively and very fast, the indication electrode, the collector, actually compares the generation flux of titrant with the flux of titrand towards the electrode surface. Diffusion layer titrations of type (i) were studied for the first time using a rotating ring-disc electrode,1.2 where the flux of titrand is determined by the hydrodynamic conditions of the experiment while the generation rate of the titrant is controlled by the disc current. The upstream disc is always used as the generator and the downstream ring as the collector (ampero- metric indicator) of the unreacted titrant. Arsenic(m) trans- ported to the vicinity of the rotating disc by convective diffusion was titrated with electrogenerated bromine. 1 Similarly to rotating electrodes, microelectrodes also exhibit a time-independent flux, although the reason is different: a predominance of non-linear over linear diffusion.This is the reason why the diffusion layer of a microelectrode can also be a suitable place where the titration reaction with electro- generated titrant proceeds. The collector should be placed in the close vicinity of the generator microelectrode in such a way as to be immersed in its diffusion layer. The segments of an interdigitated array (IDA) microelectrode have the optimum geometry to fulfil the above-mentioned conditions. One set of the segments is used for the generation of titrant and the second set for the detection (collection) of its unreacted flux.Using the IDA microelectrode a simple and sensitive analytical microtechnique was developed for the determination of some electroinactive species capable of quantitative and fast reaction with the generator electrode product. Two diffusion layer titrations are described in this paper: determination of low concentrations of thiosulfate with electrogenerated iodine and determination of allyl alcohol with electrogenerated bromine. Experimental Reagents All chemicals were used as received from commercial sources. Triply distilled water with KMn04 was used to prepare sample solutions.1796 Analyst, December 1996, Vol. 121 Interdigitated Array Electrode A thin-film microsystem based on a planar interdigitated array of electrodes was used (Fig. 1). The strip microsystem was realized on alumina-boron-silica glass, with dimensions 15 X 3 X 0.6 mm.The widths of both the microelectrodes and the gap between them were 5 pm. Platinum thin films were deposited by rf sputtering and for the patterning of the IDA structure photolithography a ‘lift-off’ technique was used.8 The whole system was passivated by a polyimide protective film; there was only a 600 X 600 pm window over the IDA microelectrodes, and the Pt contacts were opened. Instrumentation and Procedure A Model 366A bi-potentiostat (EG&G Princeton Applied Research, Princeton, NJ, USA) was used to carry out electro- chemical measurements. The generator system was operated in galvanostatic mode anodically producing iodine from 0.1 moll- 1 KI or bromine from 0.18 moll-’ KBr and 1.13 X 10-3 mol 1-1 HC104 (to suppress possible hypobromite formation).A stock standard solution of 0.05 moll-’ Na2S203, used for the preparation of standard samples was standardized iodimet- rically. A stock standard solution of 0.05 mol 1-1 ally1 alcohol for standard sample preparation was standardized by indirect bromatometry . Results and Discussion Generation of Titrant and Its Detection by Collector Titrant (iodinekromine) was generated using galvanostatic polarization of the generator set of IDA segments where the oxidation of iodidebromide proceeded with a high generation efficiency over a broad range of current density from the solution specified under Experimental. The collector was potentiostated to the cathodic limiting diffusion current value of the titrant (-0.1 V vei-sus SCE for iodine and 0.1 V versus SCE for bromine).Its detection is then based on an identical redox process as its generation (in the opposite direction): 13- + 2e 31- (1) Brl + 2e G 2Br- (2) For this case the collection efficiency is measurable as the slope of the collector current (Zcoll) versus generator current (Z,,,) dependence. The expected linearity of the Zcoll versus Zgen plot was proved for both iodine and bromine systems. The slopes were 0.854 and 0.840 for iodine and bromine, respectively. These values are substantially higher than collection effi- ciencies encountered in the case of rotating ring-disc elec- trodes.’ The Icoll versus Zgen dependences can be seen in Fig 2 for iodine transfer. The generation current varied in the range 0-200 nA in the ‘point by point’ measurement or in a very slow Fig.1 electrodes (not to scale). Layout of the thin-film microsystem with the horizontal IDA current scan 1 nA s-I . Higher scans ( > 20 nA s-I) led to non- linearity of the Zcoll versus Zgen plot and to its hysteresis, with a significant difference for the increasing and decreasing genera- tion current scans. It can be seen that the cathodic limiting current of iodine reduction on the collector is almost zero for generator currents <90 nA in the case of the IDA 1 microelectrode. As will be shown later, this value represents the blank in the diffusion layer titration with iodine. The Model 366A bi-potentiostat permits measurement of the potential of the anodically polarized generator set of micro- segments.Polarization curves constructed in this way are shown in Fig. 2, curves b. Their shapes indicate that more than one species undergo anodic oxidation at the generator. The limiting current of the oxidation proceeding at lower anodic potentials compared with the oxidation of iodide (e.g., curve bl) equals the ‘blank’ of iodide at corresponding electrode (e.g., curve al). The value of the ‘blank’ cannot be influenced by the usual procedures of electrode treatment (e.g., by repeated anodic and cathodic polarization). It differs considerably for individual IDA microelectrodes (Fig. 2). The product of the oxidation taking place at potentials <0.3 V is not transferred to the collector (or at least it does not give any signal). These anodic oxidations are probably connected with some surface redox processes of the materials used for IDA fabrication.As shown in Fig. 3, both IDA microelectrodes examined exhibited almost the same blank value (70 nA) when the generation and collection of bromine were studied. It equals the limiting current of the oxidation proceeding at lower potentials compared with the potential of oxidation of bromide from 0.18 moll-1 KBr and 1.13 X 10-3 moll-’ HC104. The product of this oxidation is either not reduced at the collector polarized to 0.1 V or it is not transferred to the collector at all. Its value is reproduced well using the same KBr solution. Diffusion Layer Titration of Thiosulfate With Electrochemically Generated Iodine The transfer of iodine from the generator to the collector set of IDA segments is influenced and consequently the shape of the Zcoll versus Zgen plot changes substantially if thiosulfate is added to the solution.Thiosulfate is electrochemically inactive at the potential of anodic generation of iodine. As the chemical reaction of iodine with thiosulfate is rapid and quantitative in 0.1 mol 1-1 KI solution, the surface concentration of iodine at the generator will remain zero until the current flowing through the IDA generator is such that the flux of outgoing iodine is larger than the flux of incoming thiosulfate. As the generator I I 40 I w 0.3 0 2 Q 0.1 Lu” $ 0.2 3 5 0.0 0 50 100 150 200 I,en/nA Fig. 2 IDA collector versus generator currents plot (curves a) measured ‘point by point’ for three IDA microelectrodes. Curves b, anodic galvanostatic polarization curves of generators of the three IDA micro- electrodes.Solution, 0.05 mol I-’ KI (Aldrich ACS reagent).Analyst, December 1996, Vol. 121 1797 5 Lu" current increases, the iodine spreads across the insulating gaps until it reaches the collector segments. Then the collector current increases from zero, since the collector is set at the potential of the limiting diffusion reduction current of iodine. The Zcoll versus I,,, plot influenced by thiosulfate concentration can be utilized for its determination based on the reaction: ..-zd , I I The Zcoll versus Zgen plot (Fig, 4) is in fact a coulometric diffusion layer titration curve with the IDA generator producing titrant and the IDA collector serving as an amperometric detector. The titration curves generated for different thiosulfate concentrations are depicted in Fig.4. Comparing these titration curves with those obtained at a rotating ring disc electrode, it can be seen that there is no significant curvature of the IDA curves around the point of intersection. The curvature of the ring disc titration curve has been explainedlJ as a consequence of an insufficiently narrow ring (ring width around 1 mm). The absence of any significant curvature of the IDA titration curves can be understood if we realize that the IDA collector microband is about 200 times narrower than the ring. The extrapolation of the linear portion of the increasing part of the titration curve to the collector residual current gives an end-point value of the generator current ZgenE.As no chemical reaction actually takes place in the bulk phase of the sample, the ' titration' can be extensively repeated in the same solution with excellent reproducibility of the Zcoll versus Zgen plot. The RSD of 20 consecutive runs of the same solution of 8 X 10-5 moll-' ~n t L- a2--a3/ -rv I 20 > 5, I i Fig. 3 IDA collector versus generator currents plot (curves a) measured 'point by point' for two IDA microelectrodes. Curves b, anodic galvano- static polarization curves of generators of the two IDA microelectrodes. Solution, 0.18 mol 1- KBr and 1.13 X 10-3 rnol 1-' HC104. / / / 1 I f f f I I 0 50 Id0 160 200 250 360 IgenInA Fig. 4 Diffusion layer titration curves of thiosulfate by electrogenerated iodine using IDA 3 microelectrode.Bulk concentration of thiosulfate: (1) 2.66 x (2) 4.91 X lops; (3) 7.02 X and (4) 9.13 X rnol 1- I. Na2S203 is 1.4% for Z g e n ~ and 2.1 % for the slope of the rising part of the titration curve. ZgenE is a linear function of thiosulfate concentration for any of the IDA microelectrodes, as can be seen in Fig. 5. It fits the equation: (4) The blank is 90 nA for IDA 1, 27 nA for IDA 2 and 3 nA for IDA 3, whereas the slope is virtually the same, 1424 pA 1 mol-l, for all the IDA microelectrodes. The precision and accuracy of the analysis for IDA 3 are given by the RSDs of its slope, 1.4%, and intercept, 10.2%. The intercept can be regarded as the blank and can be subtracted from the experimental ZgenE. Its value corresponds to approximately 2 X 10-6 mol 1-1 thiosulfate, and the determination limit is estimated to be about 6 X 10-7 mol 1-1 Na2S203.Under our experimental conditions (sample volume 10 ml), it represents 1 pg of Na2S203. The value of the slope of eqn. (4) represents the sensitivity of the method. The sensitivity given by Bruckenstein and Johnson' for analogous ring-disc diffusion layer titration of As"' with iodine is over 5 X 105 pA 1 mol-1, i.e., more than 350 times higher. This is a consequence of the substantially higher flux rate of the titrand in the case of the rotating ring-disc electrode. Correspondingly there is a substantially lower minimum analyte concentration. A 1000-fold decrease in the sample volume compared with the typical volume for hydrodynamic voltam- metry would not cause any serious experimental problems.This would more than compensate for the lower sensitivity of IDA diffusion layer titration in practical applications. Seven standard samples of 11-62 pg of Na2S203 were analysed and the results are summarized in Table 1. Statistic- ally, the mean of parallel determinations does not differ from the known thiosulfate values for all standard samples. The confidence interval is acceptable for the concentration level of thiosulfate determined. The relatively low determination limit enables the technique to be applied to the trace determination of thiosulfate in analytical-reagent grade potassium iodide. A 0.5 mol 1-I solution of four chemicals products in triply distilled water with KMn04 was analysed. The analyte content was at least 50 times higher than the determination limit.The IDA 3 microelectrode was applied for diffusion layer titrations. The standard additions method was applied for evaluation of the end-point value of the generator current. The results are summarized in Table 2. The thiosulfate contents (the sum of species reacting with iodine) were found to be slightly lower than those specified by the manufacturers. ZgenE = blank + slope X c - - I 3 Concentration11 0-5 rnol dm-3 Fig. 5 Relationship between the end-point generator current value ZgenE and thiosulfate bulk concentration: A , IDA 1 (solid line, ZgenE = 90 X 10-9 + 1424 X 10-6c); H, IDA 2 (solid line, Zger,E = 27 X 1 O V + 1424 X 10-6c); and a, IDA 3 (solid line, ZgenE = 3 X + 1424 X 10-k).1798 Analyst, December 1996, Vol.121 Diffusion Layer Titration of Allyl Alcohol With Electrochemically Generated Bromine The dynamic character of diffusion layer titration requires, in addition a high equilibrium constant, also a high rate of the titration reaction. Addition reactions of halogens with unsatu- rated organic compounds proceed relatively slowly and are therefore often subject to kinetic studies. The rate constant for the bromination of allyl alcohol: CH2=CHCH2OH + Br2 -+ BrCHzBrCHCH20H (5) was determined by a low-concentration potentiometric tech- nique10 and using a rotating ring-disc electrode.11 Because of the low value of the rate constant, the ring current versus disc current plot (the titration curve) exhibits a more pronounced curvature than that resulting from the insufficiently narrow ring.The evaluation of the curvature of the ring-disc curves allowed the calculation of the second-order rate constant as Table 1 Results of diffusion layer titration of Na2S203 with elec- trogenerated iodine in standard samples, base electrolyte 0.1 mol I-* KI (Aldrich ACS reagent) Confidence interval for 95% probability Na2S203 takedpg 11.2 18.1 25.9 39.7 45.3 57.4 62.6 Na2S203 found/pg 11.7 18.9 24.9 38.2 46.9 59.4 64.7 No. of analyses 5 5 5 5 5 5 5 Pg 0.5 0.8 1 .o 1.5 1.6 2.0 2.1 Relative % 4.5 4.4 3.8 3.8 3.5 3.5 3.4 80 I 0 50 I00 150 200 I,en/nA Fig. 6 Diffusion layer titration curves of allyl alcohol with elec- trogenerated bromine using IDA 3 microelectrode. Bulk concentration of allyl alcohol: 1,0.81 x 10-4; 2, 1.72 X and 4, 3.52 X moll- 1 .3, 2.51 X ~ ~~ ~~~~~ 2 X lo5 1 mol-l s-1, which is in good agreement with the low- concentration potentiometric results. IDA diffusion layer titration curves measured for different amounts of allyl alcohol, shown in Fig. 6, show virtually no curvature around the intercept of their linear parts. The lack of curvature can be explained by the above-mentioned fact that the titrand flux rate towards the generator is much lower for microelectrodes than for a typical rotating disc. The slower flux makes more time available, in which also slower reactions can proceed close to equilibrium. From the point of view of the accuracy and precision of analytical determinations, this represents a considerable advantage over the rotating ring-disc technique.The absence of curvature allowed the evaluation of the Zcoll versus Zgen plot as a titration curve similar to that in thiosulfate Concentration/l 0-4 mol dmV3 Fig. 7 Relationship between the end-point generator current lgenE and allyl alcohol bulk concentration: A , IDA 2; and 0, IDA 3 (solid line, ZgenE = 70 x 10-9 + 486 x io-6c). Table 3 Results of diffusion layer titration of allyl alcohol with electrogenerated bromine in standard samples (base electrolyte 0.18 moi 1-1 KBr and 1.13 x 10-3 moll-l HC104) Confidence interval for 95% probability Allyl alcohol Allyl alcohol No. of taken/pg found/pg analyses pg Relative % 45.1 43.0 5 2.1 4.7 78.9 82.4 5 3.5 4.4 11 1.6 107.1 5 4.5 4.0 142.3 147.7 5 5.4 3.8 195.1 187.9 5 7.2 3.7 Table 2 Results of diffusion layer titration of Na2S203 with electrogenerated iodine in commercial KI chemicals of analytical purity (0.5 mol 1-l solution in water redistilled with KMn04) Sample KI (analytical-reagent grade) KI (analytical-reagent grade) KI (analytical-reagent grade) KI (ACS reagent) (Lachema, Bmo, Czech Republic) (Lachema, Bmo, Czech Republic) (unknown origin) (Aldrich, Milwaukee, WI, USA) Na2S203 specified by Na2S203 manufacturer/ found/ lo-" g g-' 10-6 g g- 50 32 50 45 52 - Confidence interval for 95% probability No.of 1 analyses 10-6 g 8-1 Relative % 5 3.6 11.2 5 5.1 11.4 5 5.2 10.0Analyst, December 1996, Vol. 121 1799 analysis. There was no problem with finding the intercept of the increasing part with the collector residual current (IgenE). The repeatability of titration runs in the same solution was tested using 1 X 10-4 mol 1-1 allyl alcohol.The RSD of ZgenE for 15 consecutive runs was 2.1%. ZgenE was found to be linearly dependent on the allyl alcohol concentration, as shown in Fig. 7. The intercept of the linear dependence [(eqn. (4)] is 70 nA and the slope of the calibration plot is 486 pA 1 mol- for all of the IDA microelectrodes tested. The precision and accuracy of the analysis are given by the RSD of its intercept, 4.2%, which is actually the blank of the titration. Its value corresponds to approximately 1.44 X mol 1-1 allyl alcohol. The determination limit is estimated to be 2 X 10-5 mol 1-l allyl alcohol, representing 10 pg of analyte in a typical sample volume of 10 ml. Five standard samples containing allyl alcohol in the range 45-195 pg were analysed and the results are summarized in Table 3. The mean of parallel determinations does not differ from the known values for all the standard samples. The confidence interval is acceptable for the concentration level of analyte determined. We gratefully acknowledge the assistance of Doc. Dr. V. TvaroZek and Mr. I. Novotnq of the Department of Mic- roelectronics, Slovak Technical University, Bratislava, in the fabrication of IDA microelectrodes. This research was sup- ported by VEGA, the Grant Agency of Science (Slovak Republic). References 1 2 3 4 5 6 7 8 9 10 11 Bruckenstein, S., and Johnson, D. C., Anal. Chem., 1964, 36, 2186. Albery, W. J., Bruckenstein, S., and Johnson, D. C . , Trans. Faraday Soc., 1966, 62, 1938. Kemula, W., and Michalski, M., Rocz. Chem., 1936, 16, 533. Orleman, E. F., and Kolthoff, I. M., J . Am. Chem. Soc., 1942, 64, 1044. Bustin, D., and Mocdc, J., Talanta, 1973, 20, 1185 and 1191. HofbauerovB, H., MocBk, J., and Bustin, D., Chem. Pap., 1974, 26, 609. Bustin, D., and Rievaj, M., Chem. Pap., 1987, 41, 227. Tvaroiek, V., Ti Tien, H., Novotn?, I., Hianik, T., Dlugopolsky, J., Ziegler, W., LeitmannovB-Ottova, A., JakaboviE, J., RehBCek, V., and Uhlfir, M., Sens. Actuators B , 1994, 18-19, 597. Albery, W. J., and Hitchman, M. L., Ring-Disc Electrodes, Clarendon Press, Oxford, 197 1. Bell, R. P., and Atkinson, J. R., J . Chem. Soc., 1963, 3260. Albery, W. J., Hitchman, M. L., and Ulstrup, J., Trans. Faraday Soc., 1969,65, 1101. Paper 6102295 B Received April 2, 1996 Accepted July 16, 1996

ISSN:0003-2654    

DOI:10.1039/AN9962101795

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 (1801-1804) 1801 Electrochemistry of the Nitroprusside Ion. From Mechanistic Studies to Electrochemical Analysis* Helena M. Carapucaa, Joao E. J. Simaoat and Arnold G. Foggb t7 Chemistry Department, Loughborough University, Loughborough, Leicestershirc, UK LEI1 3TU Department of Chemistry, University of Aveiro, 381 0 Aveiro, Portugal The electrochemical reduction of the nitroprusside ion (NP) was examined by differential-pulse polarography and square-wave voltammetry. The effect of pH on the electrochemical behaviour of NP was studied and experimental evidence for the occurrence of adsorption of the products of the first one-electron reduction of NP is shown. Owing to the adsorption of these products, which are the reactants in the second reduction step of NP, the peak current of the second process is highly enhanced compared with that of the first process, especially at low concentration levels (< 1 X 10-5 rnol dm-3).For acidic solutions the enhancement is considerably higher than at pH 3 7. A surface reaction leading to the regeneration of the reactants of the second reduction process can be advanced as a possible explanation. Square-wave cathodic stripping voltammetry with adsorptive accumulation was used for the determination of NP. The detection limit depends on the square wave parameters and can be as low as 2.3 nmol dm-3 for a delay time of 60 s. Keywords: Nitropi-irsside ion; electrochemical reduction; adsorption; square-wave voltammetry; cathodic stripping voltammetry Introduction The reduction of the pentacyanonitrosylferrate(r1) or nitro- prusside ion (NP), [Fe(CN)SN0]2-, has been studied for more than four decades by chemical, electrochemical and radio1 ytic techniques.1.2 In aqueous solution the electrochemical reduction of NP is an over-all four-electron, three-proton process which can occur as three reduction steps. Both of the first two steps correspond to 1 -electron reversible transfers whereas the third step involves the irreversible consumption of two or three electrons, depending on pH. The first one-electron reduction is independent of pH and produces the ion [Fe(CN)5N0]3-, in which the NO+ ligand has been reduced to NO. The ion [Fe(CN)SN0]3- can lose the axial cyanide ligand, giving [Fe(CN)4NO]2- in a pH-dependent chemical step which, in alkaline solution, is the rate-limiting step.' The ion [Fe(CN)4NOl2- is the reactant of the second reduction process in alkaline and neutral solutions.In acidic solution the primary one-electron reduction product may be protonated on the nitrosyl ligand, producing [Fe(CN)SNOH]2-, as proposed by Masek and Maslova. They suggested that this protonated ion would then be the reactant of the second reduction. Hence the second and third reduction steps are pH dependent. Further, in neutral and alkaline solutions, the reduction intermediates adsorb on the mercury electrode.3 * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29, 1996. To whom correspondence should be addressed. In acidic solution, cyclic voltammograms of 1 X 10-?-1 X 10-4 rnol dm-3 NP solutions show one or two cathodic peaks in the anodic scan, which seem to be associated with the second reduction p r o ~ e s s .~ The occurrence of these inverted peaks may be due to some regeneration process leading to the formation of the reactants of the second reduction step of NP. This process involving adsorbed species was assumed to be a comproportio- nation reaction between NP and the product of the second reduction step.4 Sodium nitroprusside is a hypotensive drug, whose rapid physiological activity is related to its electrophilic character and also to its redox behaviour. NP can be used as a source of nitric oxide in body tissues and it is in fact the NO molecule that is responsible for the smooth muscle relaxation process.5 Hence not only has the chemistry and the redox chemistry of NP been a subject of general interest but also its analytical determination in biological fluids has engaged the attention of researchers.Electrochemical methods have been developed as an alternative to the UV spectrophotometric determination of NP,G some in order to improve the analysis at therapeutic levels (ng cm-?). Leeuwenkamp et al.7 reported a differential pulse polarographic (DPP) method for the determination of NP in acidic solution (HC104, 1 rnol dm-3). Pirzad et al.8 developed a differential- pulse cathodic stripping voltammetric method based either on the accumulation of NP at a poly-L- lysine modified hanging mercury drop electrode or on the accumulation of a copper-NP species at an unmodified hanging mercury drop electrode. In this work, we studied the electrochemical behaviour of NP in aqueous solutions of different pH and NP concentrations.The adsorption characteristics of the reduction process at low concentrations were studied. The application of square-wave voltammetry to the determination of NP at the 10-9 rnol dm-3 concentration level was tested. Experimental Reagents and Solutions Sodium nitroprusside dihydrate (Panreac, Barcelona, Spain), orthophosphoric acid (Riedel-de Haen, Hannover, Germany), boric acid (Panreac), acetic acid (Merck, Darmstadt, Germany), sodium perchlorate monohydrate (Merck), sodium hydroxide (Merck) and Triton X- 100 (Merck) were of analytical-reagent grade and were used as received. Stock standard solutions of nitroprusside (about 1 X 10-3 rnol dm-3) were prepared weekly and were stored in a flask wrapped in aluminium foil.Phosphoric acid (with a formal concentration of 0.1 mol dm-3) and Britton-Robinson buffer (0.08 mol dm-3 each of orthophosphoric acid, boric acid and acetic acid) were used as electrolytes, the pH being adjusted with 1 rnol dm-3 NaOH. In some experiments the ionic strength was adjusted to 0.8 rnol dm-3 with sodium perchlorate. De- ionized water (obtained from a Milli-Q system; Millipore- Waters, Milford, MA, USA) was used for preparing all solutions.1802 Aiialyst, December. 1996, Vol. 121 Instrumentation and Procedure Electrochemical experiments were performed using a BAS I OOB/W electrochemical analyser (Bioanalytical Systems, West Lafayette, IN, USA) connected to a PAR 303A mercury electrode stand (EG&G Princeton Applied Research, Princeton, NJ, USA) in the HMDE or SMDE mode (medium drop size; drop area, 0.01 6 cm2).The auxiliary electrode was a Pt wire and the reference electrode was Ag/AgCl (saturated KC1). pH measurements were made with a Russell CWL/S7 combined pH-reference electrode (Russell pH, Fife, UK) and an Anatron pH300 pH meter (Anatron Instruments, Matosinhos, Portugal). All solutions were deoxygenated with nitrogen for 4 min prior to the electrochemical measurements, All experiments were carried out at room temperature. The voltammetric cell was wrapped in aluminium foil to prevent light degradation of NP. In the cathodic stripping voltammetric experiments, four replicate voltammograms were recorded for each solution.After each addition of a volume of nitroprusside, the solution in the voltammetric cell was deoxygenated for 30 s. Results and Discussion General Electrochemical Behaviour of Nitroprusside In alkaline solution, differential pulse polarograms of NP solutions of different pH and NP concentration (Fig. 1) show that at the 1 X 10-3 mol dm-3 concentration level the second 4- V I I 4 2 0.0 -0.26 -0.52 -0.78 -1.0 -1.3 0.0 -0.30 -0.60 -0.90 -1.2 3 A L 0.0 -0.26 -0.52 -0.78 -1.0 -1.3 0.0 -0.30 -0.60 -0.90 -1.2 PotentialN Fig. 1 Differential-pulse polarograms of NP in Britton-Robinson buffer (with 0.8 rnol din- NaC104) at different pH and NP concentrations: (a) pH 9.2, 1.6 X 10-3 mol dm-3 NP; (h) pH 9.2, 4.0 X 10-6 mol dm-3 NP; (c) pH 3.0, 1.6 X rnol dm-3 NP; and (d) pH 3.0, 3.6 X 10-6 mol dm-3 NP. Scan rate, 4 mV s I ; drop time, 0.5 s; pulse amplitude, 50 mV.”“’~’-- -”? t \ > ’1 , A [Nitroprusside]/mol dm-3 Fig. 2 Ip”/Ip1 ratio I.”I’SUS NP concentration for DPP measurements at different pH (A, pH 3; B, pH 4; C, pH 7; and D, pH 9). Experimental conditions as in Fig. 1. peak, at about -0.58 V, is lower than the first peak, at about -0.32 V. This is due to the occurrence of a rate-limiting chemical step, the release of CN- from the primary one- electron reduction product of NP. ’ However, for lower concentrations of NP [(Fig. 1 (b)] the second peak is higher than the first, which can be accounted for by the occurrence of adsorption of the reactant of the second reduction step, the [Fe(CN)4N0l2- ion.Leeuwenkamp3 also showed that the reactant of the second reduction of NP was adsorbed at the mercury electrode in neutral and alkaline solutions. In acidic solutions [Fig. 1 ( c ) and (41, the second peak is always higher than the first. For a 1.6 X 10-3 rnol dm-3 solution the second peak also has an unusual shape, which suggests that the process occurring at these potentials must be complicated. In fact, the corresponding cyclic voltammograms (not shown) display cathodic peaks in the anodic scan, occurring at the same potential as the second cathodic process. This was previously interpreted as being due to some regenera- tion process involving adsorbed ~ p e c i e s , ~ presumably a com- proportionation between the adsorbed product of the second reduction step and the NP that diffuses towards the electrode 0.0 -0.3 -0.6 0.0 0.3 0.6 PotentialN Fig.3 Square-wave voltammograms of 2.3 X rnol dm-3 NP solutions (Britton-Robinson buffers) at different pH and at two square- wave frequencies, ( a ) 15 and (h) 250 Hz. Curves A, pH 3.0 and curves B, pH 7.6. Square-wave amplitude, 50 mV; AE = 10 mV; Ei, = 0 V; t d = 0 s; and teq 2 s. .8 L -6.0 I I 0 -0.4 -0.8 PotentialN Fig. 4 Plots of the individual forward (F) and reverse (R) square-wave currents of 2.3 x 10-5 mol dm-3 NP solutions at different pH and at two square-wave frequencies: (a) pH 7.6,f = 15 Hz; (h) pH 7.6, f = 250 Hz; (c.) pH 3.0,f = 15 Hz; and (d) pH 3.0,f = 250 Hz. Experimental conditions as in Fig. 3.Analyst, December 1996, Vol. 121 1803 surface, thus regenerating in situ the reactant of the second reduction process.Moreover, for pH < 6, the cyclic voltammo- grams show two cathodic peaks for the second reduction process; these peaks were assigned to the reduction of both [Fe(CN)4N0]2- and [Fe(CN)SNOH]2-. Hence it was assumed that, for pH < 6, there could be two reactants for the second reduction process.4 With the 3.6 X 10-6 mol dm-3 solution of pH 3.0 [Fig. 1 (41 the current of the second peak is substantially higher than that of the first peak, and this enhancement is much more significant than at pH 9.2. A plot of the DPP peak current ratio ZP)/Z&I) against the nitroprusside concentration for several pH values shows all these variations more clearly (Fig. 2). For all pH values the ratio Zp(ll)/Z&l) increases as concentration decreases.In alkaline and neutral solutions this increase can be explained by assuming accumulation by adsorption of [Fe CN)4N0]2-. In acidic alkaline and neutral solutions. Two reasons can be advanced to account for this change: (i) either the protonated ion, presumed to be formed in acidic solution, [Fe(CN)SNOH]2-, adsorbs to a greater extent than [Fe(CN)4N0]2-, or (ii) a chemical reaction occurs which regenerates the reactant(s) of the second process, this reaction being pH dependent. As the ZP(l1)/Ip(l) is favoured at low concentrations, the regeneration process seems to involve adsorbed species. solutions, the enhancement of ZP(I1)/Zp(I 5 is much higher than in Square-wave Voltammetry of Nitroprusside The square-wave voltammograms in Fig.3 show how the behaviour of the system in acidic and slightly alkaline solutions can be exploited using two different square-wave frequencies. At lower frequencies (f = 15 Hz) the second process is favoured at pH 3.0 compared with pH 7.6. This agrees with the DPP results described above. At higher frequencies (f = 250 Hz) the opposite occurs: the second process is favoured at pH 7.6. Plots of the individual forward and reverse square-wave currents corresponding to the voltammograms in Fig. 3 show different patterns for different pH values and frequencies (Fig. 4). At pH 7.6 the second process has both forward (reduction) and reverse (oxidation) peaks, as is expected for a reversible process. However, at pH 3.0 the reverse peak is absent or strongly diminished and, at low frequency, the current of the second forward peak is strongly favoured relatively to the first.Hence, at pH 3.0 it may be assumed that some chemical process takes place leading to the consumption of the product of the second reduction, thus regenerating the reactant of this same reaction. In accordance with the results shown in Fig. 2, this regeneration process could be induced by protons. Mo et aZ.9 reported that the complex Fe(N0)2+, in which the nitrosyl ligand has a formal charge of zero, as is the case for the reactant of the second reduction process of NP, is reduced in a one-electron step to give Fe(NO)+. In acidic solution the peak current of the reduction of Fe(N0)2+ was substantially en- hanced for 1 X 10-5 mol dm-’ solutions when slow scan rates were used. A catalytic surface process, based on the dis- proportionation of Fe(NO)+ and the regeneration of the reactant Fe(N0)2+, was proposed as the mechanism.It seems possible that the kind of reaction proposed by Mo et al.9 for the complex Fe(N0)2+ may occur for NP in acidic solution. A possible reaction scheme could be as follows: [Fe(CN)SN0]2- + le += [Fe(CN)5N0]3- + H+ + [Fe(CN)5NOH]2- -+ [Fe(CN)SNOH]2-,ads, + le + first reduction step at about protonation of the one- [ Fe (CN) NO] - -0.3 V [ Fe(CN)5NOH]2- electron reduction product [Fe(CN)5NOH12-(ads) second reduction step at T [Fe(CN)SNOH]3-,ad,, about -0.6 V I 2 H+ + 3 [Fe(CN)5NOH]3-,ads) + 2 [Fe(CN)SNOH]2-,,ds) + [Fe(CN)SNH20H] - The last reaction involves disproportionation of [Fe(CN)S- NOH]3-, occurring only for the adsorbed species at the surface of the mercury electrode.This step would be induced by H+ and would regenerate in situ the reactant of the second process, [Fe(CN)5NOH]2-. However, this regeneration reaction would be effective only under the conditions of relatively low scan rate [Nitr~prusside]/lO-~mol dm-3 Fig. 5 Current of the second square-wave peak versus NP concentration at pH 3 (phosphoric acid solution). Experimental conditions: Ei, = -320 mV; td = 20 s; tes = 2 s; square-wave amplitude; 50 mV; f = 250 Hz; and AE = 10mV. PotentialN Fig. 6 Square-wave voltammograms of a 3.2 X 10-6 mol dm-3 NP solution at pH 3. A, Without Triton X-100; B, with 3 X lO-4%; and C , with 6.5 x lo-% Triton X-100 present in solution. Other experimental conditions as in Fig.5. Table 1 Calibration data* relative to square-wave cathodic stripping voltammetric determinations of NP at two different pH values. Experimental conditions: Ei, = -320 mV; t d = 20 s; res = 2 s; square-wave amplitude, 50 mV;f = 250 Hz; and AE = 10 mV S lope/A os/A LOD/ pH InterceptIA oi/A mol-1 dm3 mol-1 dm3 r (n = 4) mol dm-3 3.0 8.4 x 10-8 3.3 X 10-8 2.34 0.05 0.999 7.0 X 10-8 7.6 2.2 X 10-8 7.4 X 10-8 3.7 0.1 0.999 9.8 X * LOD is the detection limit (for the 99.5% confidence level), ( ~ i and (J, are the standard deviations of the intercept and slope, respectively, r is the correlation coefficient and n is the number of data points.1804 Analyst, December 1996, Vol. 121 Table 2 Calibration data* relative to square-wave cathodic slripping voltammetric determinations of NP at pH 3.Experimental conditions: E,, = -320 mV; td = 60 s in stirred solution; t,, = 2 s; square-wave amplitude, SO mV: and AE = 10 inV Slope/A o,/A LOD/ f/Hz InterceptIA o,/A mol-I dm3 mol-’ dm’ I’ (12) moldm 3 250 5.2 X lo-* 0.4 X 10V 28.2 0.4 0.999 (17) 2.3 X 10 80 -4.0 x 10-9 2.8 x 10-9 9.6 0.2 0.998 (12) 3.2 X 10-9 * See Table 1 , experiments (as occurs in DPP measurements or at low square- wave frequencies). For higher scan rates (or frequencies), the effect of the accumulation of the reactant of the second process by adsorption would prevail. This matter is being further investigated. Adsorption Characteristics As the reactant(s) of the second reduction process adsorb(s) on the mercury electrode, one can produce a preconcentration step by waiting for a given delay time at a potential between the first and the second reduction processes. Nitroprusside itself (at potentials more negative than the first reduction process) adsorbs very weakly.* At the 10-7 mol dm-7 concentration level, using an initial potential (E,,,) of -320 mV and a delay time (td) of 20 s , a linear relationship between the current of the second square-wave peak and the NP concentration was observed for both pH values studied using a square-wave frequency of 250 Hz.In the data treatment a total accumulation time (tact) is used instead of the delay time and is defined as t,,, = t d + teq + IE, - E,nl/(AEfl, where E, is the peak potential, feq is the equilibration time between the initial delay time and the application of the potential scan and AE is the square-wave step height.At pH 7.6 the sensitivity (or the slope) is higher than that at pH 3.0 (see Table 1) as expected because of the high frequency used. For pH 3.0, in spite of the lower sensitivity, the detection limit is better and the background is ‘cleaner.’ Hence, taking these facts into consideration and also that the adsorption characteristics of NP in acidic solution had not yet been studied, we chose pH 3.0 for the following analytical study. Fig. 5 shows the results of a calibration experiment at pH 3.0 over a wide concentration range. Only at lower concentrations is the peak current proportional to the NP concentration. This trend is consistent with a process limited by adsorption of the reactant.From the values of the point where the curve in Fig. 5 leaves the straight line, a maximum value of the product C a can be calculated as 7.05 X 10-6 mol dm- 3 ~1’2. This value can be taken as indicative of the saturation of the electrode surface. The addition of a small amount of Triton X-100 to an NP solution hinders the second reduction process (see Fig. 6), meaning that the strong surfactant prevents the adsorption of the intermediates of the NP reduction process which are responsible for the second peak. Further evidence of adsorption is that the peak current depends on the delay time before the potential scan (and consequently it depends on the accumulation time). Plots of the peak current versus the square root of t,,, for a given NP concentration (not illustrated) are similar in shape to those of Zp versus C for a given accumulation time, shown in Fig.5 ; there is linearity only for lower accumulation times and then deviation from linearity occurs. Hence, for a C c value lower than 7.05 X 10-6 mol dm-3 plots of I , versus a for several NP concentrations are always linear. A plot of the slopes of those I,, versus < representations against NP concentration is also linear (correlation coefficient 0.9994). These results show that Z, is proportional to the product C c provided that C or t,,, is not large enough for saturation of the surface to occur. This is the expected dependence for diffusion- controlled adsorption of the reactant during the delay time prior to its reduction. Further, plots of Zp/<t& versus frequency are linear (the slope of the relationship l o g ( I , , / c ) vei-sus lo@ was 0.96).The peak potential is independent of the NP concentration, of the delay time and of the initial potential, but shifts to more negative values as f increases. In fact, the relationship E, versus logfis linear with a slope of -60 mV per decade. Once again all these dependences are in accord with the reduction of an adsorbed species.10 Square-wave Cathodic Stripping Voltammetric Determination of Nitroprusside Based on the over-all increase in the height of the second DPP peak for acidic solutions Leeuwenkamp et al.7 developed a DPP method for the determination of NP at low concentrations. The detection limit that they reported was 7 X 10-9 rnol dm-3. As square-wave voltammetry is a fast and sensitive technique, it was our intention to test its application to the determination of NP.It was found that, at pH 3, the reactant(s) of the second reduction process adsorb(s) at the mercury electrode and NP can be determined easily at the 1 X 10-7 mol dm-3 level using a delay time of 20 s. In order to improve the detection limits, a larger t d associated with forced convection may be used. Hence, for a delay time of 60 s in stirred solution a linear calibration was obtained from 2.3 X 10-9 to at least 30 X 10-9 rnol dm-3 for both frequencies used (see Table 2). We have not tested the determination of NP in biological samples where the presence of surfactants (e.g., in serum) would have the same effect as Triton X-100 on the second reduction peak of the adsorbed NP intermediate. Therefore, some pre-treatment of such samples would almost certainly be required. References 1 2 3 4 5 6 7 8 9 10 Masek, J., and Maslova, E., Collect Czech. Chrm. Commun., 1974, 39, 2141. Glidewell, C., and Johnson, 1. L., Inorg. Chim. Ada, 1987, 132, 145. Leeuwenkamp, 0. R., PhD Thesis, University of Leiden, 1985. CarapuGa, €1. M., Simao, J. E. J., and Fogg, A. G . , Port. Electrochim. Actu, 1995, 13, 349. Butler, A. R., and Williams, D. L. H., Chem. Soc. Rev., 1993, 233. Baaske, D. M., Smith, M. D., Kamatz, N., and Carter, J. E., J. Chromatop., 1974, 212, 339. Leeuwenkamp, 0. K., van der Mark, E. J., Jousma, H., van Bennekoni, W. P., and Bult, A., And. Chinz. Acla, 1984, 166, 129. Pirzad, R., Moreira, J. C., Rangel, A. 0. S. S., Alonso, K. M., Edmonds, T. E., and Fogg, A. G., Rnulyst, 1994, 119, 963. Mo, J., Jin, D., Pewi, J., and Zhao, Z.. Anal. Sci., 1990, 6, 251. Webber. A., Shah, M., and Osteryoung, J., And. Chinz. Actu, 1984. 157, 1. Paper 6/03 776C Received May 30,1996 Accepted July 24, 1996

ISSN:0003-2654    

DOI:10.1039/AN9962101801

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 (1805-1810) 1805 Hot-wire Electrodes: Voltammetry Above the Boiling Point* Peter Griindler, Andreas Kirbs and Tadesse Zerihun Fachhereich Chemie, Universitat Rostock, Buchhinderstr. 9, D- I805 I Rostock, Germany Heated wires in a special symmetrical arrangement were used for electrochemical experiments in a thin, hot, near-electrode solution layer. By application of a series of short-time heat pulsing sequences with current sampling and synchronized polarization steps, voltammetric curves above the boiling point were recorded. Methods for determining and controlling the actual temperature are presented. The temperature dependence of quantities such as diffusion coefficient or electrode potential can be determined in a convenient manner. Examples are given for different redox systems, including hexacyanoferrate(1ii)-(u), Fe3+-Fe2+, bromate, oxalic acid and formaldehyde.Keywords: Hot-wire rlectrochemistry; microelectrodes; pulse idtanzmetry ; heated electrodes Introduction It is very simple to perform electrochemical measurements at a heated wire. Both ends of a thin metallic wire are coupled to an electric power source and the wire heated this way is connected to the appropriate input of a potentiostat and placed in an electrochemical cell to form the working electrode. The heating current must be an alternating current, because a direct current would bring about a voltage drop that might cause undesired dc polarization. It is surprising that the electrochemical application of such a simple arrangement has rarely been reported.l.2 On the other hand, it is well known that, with electrodes heated in situ, many interesting phenomena can be studied.Thus, e.g., the rate of slow redox reactions can be increased, transport processes are enhanced and the signal-to-noise ratio of electrochemical measurements can be improved by thermal modulation. The most promising feature of a heated electrode surface is that only a thin solution layer becomes hot and molecules outside this layer are not exposed to high temperatures. Thus, no decomposition or outgassing in the bulk volume occurs, even with delicate or volatile components. Studies with heated electrodes have been done mostly with laser-illuminated electrode surfaces,3-1*, but cumbersome in- strumentation is necessary.An electrically heated mic- roelectrode could be used for the same purpose in a much simpler and cheaper way. It should be possible even to construct tiny heated electrochemical detectors for flow stream applica- tion. Such detectors could be useful, e.g., for smooth reactions of inert organic substances at the outflow of a chromatographic column. Several problems with heated microelectrodes had to be solved. Without special precautions, a strong ac distortion of the electrochemical signal would occur, caused by the high ac heating current. The latter will generate an ac voltage drop along * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29. 1996. the wire length, and the ac signal built up in this way will find a pathway through the solution towards the counter electrode.If the frequency of the heating ac current is high enough, this distortion may be suppressed by a low pass in the signal pathway, but the signal itself might be deformed in this way. It was shown earlier13 that this undesirable effect can be overcome by arranging the heated electrode in a symmetrical configura- tion, where the connection to the working electrode input of the potentiostat is located between two equal halves of the heated wire. The disturbing ac voltage signals appear with equal magnitude, but with opposite sign at both sides of the electrode wire. As a result, their effect on the measuring circuit is nearly completely cancelled out. Very low-magnitude electrolysis signal currents can be recorded at an electrode which is at the same time exposed to a high-magnitude, high-frequency ac current.With the symmetrical arrangement, two ways of working with heated electrodes were found: (a) permanent heating with continuous signal recording; and (b) pulse heating with signal current sampling at short time intervals in a pulsing sequence. In method (a), a stationary temperature at the electrode surface is established. The experiment is not far from common practice and commercial instruments can be used with some modifica- tions. It has to be considered, however, that the surface temperature of the electrode must stay well below the boiling- point of the solution. Method (b) is presented here for the first time. It allows electrochemical experiments to be performed far above the boiling point without causing the solution to boil.The resulting sampled voltammograms have an appearance as if they had been recorded in a homogeneous, very hot solution. Exgerimen t a1 Instrumentation and Materials The electrodes were made of platinum wire 25 pm in diameter (Goodfellow, Bad Nauheim, Germany). Equal lengths were mounted in an assembly of printed board material as described earlier.14 Reference and counter electrodes were a saturated calomel electrode (Radiometer, Copenhagen, Denmark) and a platinum foil electrode, respectively. The working electrode assembly was heated by a laboratory- made sine-wave power generator that was connected to the assembly via a high-frequency transformer. The heating current frequency used was 100 kHz throughout.In experiments with permanent heating, the electrode ar- rangement was connected to a general-purpose electrochemical instrument (AUTOLAB, ECO-Chemie, Utrecht, The Nether- lands). In experiments with pulsing sequences, a laboratory-made potentiostat was combined with a personal computer via a commercial analogue-digital input-output interface card (Type DT-2801A, Data Translation, Marlboro, MA, USA). Time control and measurement and recording were performed by means of a computer program written in ASYST.1806 Analyst, December 1996, Vol. 121 All chemicals were of analytical-reagent grade (Merck, Darmstadt, Germany) and were used as received. Solutions were prepared with water obtained from a Milli-Q system (Millipore-Waters, Milford, MA, USA).Procedures Determination of electrode temperature and temperature profiles The following two methods for the experimental determination of the electrode temperature were used. In the ‘resistance’ method, the electrode assembly was heated in pure water with a precisely stabilized direct current of well known magnitude. The dc voltage drop at the wire was recorded by a calibrated precision oscilloscope. The resulting voltage- time diagrams reflect the change in resistance during the heating process. Taking into account the temperature coefficient of the specific resistivity of platinum, every measured voltage value corresponds to one temperature value of the electrode metal. The temperature inside the wire was assumed to be homoge- neous. By careful measurements with low-magnitude currents it was ensured that the conductivity of the solvent did not give any detectable contribution to the results. In the ‘potentiometric’ method, the open-circuit potential of a reversible redox couple was measured during heating.The procedure has been described earlier. l4 As an alternative to the described experimental detennina- tion, both the electrode temperature and the temperature profiles inside the solution were calculated by means of a digital simulation analogous to the well known ‘box method’ com- monly used to solve diffusion problems in electrochemistry.” The solution around the wire was divided symbolically into i individual cylindrical shells (boxes) with a thickness h in the region of 1 pm. The amount of heat energy transferred from box to box inside the solution during n individual time steps At of 2.5 ps was calculated using the law of heat conduction, whereas the heat energy transferred to the electrode wire itself (Le., the ‘box 0’) was considered as joulean heating of a metallic cylinder.As a result, temperature profiles for each time step were calculated. One point in these profiles is the actual electrode temperature, i.e., the temperature at zero distance from the electrode surface. Most of the calculations were carried out by means of a general purpose spreadsheet program (Microsoft, Excel) where a vertical column contains the equations describing the actual new temperature for every box i. Every spreadsheet cell in a horizontal row represents a new time step n. An alternative numerical calculation method gave equal results in less computer time.16 The constants used were the heat conductivity coefficients of the materials (platinum and aqueous solution), the specific heat values and the densities. This calculation approach is so far the only way to obtain information about the temperature distribution on the solution side.The agreement of one of the calculated box temperature values, namely that of the electrode itself, with the measured electrode temperature was considered to be a test for the accuracy of the calculation. Sampling voltammetry with pulsed heating In Fig. 1, the procedure is illustrated schematically. A series of heating pulses were applied, each of them not longer than 0.1 s, and each pulse was followed by a much longer relaxation period.During the short heating intervals, the temperature rises as shown, but decreases rapidly after heating has stopped. Before the next pulse starts, the electrode has cooled to room temperature. Synchronously with the heating cycle, the polar- ization potential was varied stepwise. This was mostly done as shown, i.e., in the form of a staircase, but in other experiments also with rectangular voltage pulses (analogous to normal-pulse voltammetry) and with different waveforms. As a result of the sharp temperature and voltage jumps, the actual electrolysis current changes in a characteristic manner. Typically, near the end of every heat pulse the current change tends to cease, very similarly to the curves found in chronoamperometry. In this instance, for a sampling period in the range of microseconds ‘current samples’ (indicated by the black squares in Fig.1) were taken repeatedly. The individual samples were used to compose a complete voltammogram. In practice, several hundred samples were taken in one experiment to form a voltammetric curve. Every point in such a ‘heat-sampled voltammogram’ is measured exactly at the same time delay after the corresponding heat pulse has started. This means that every individual measurement is made when the electrode has reached exactly the same temperature as in the previous heat pulse. The resulting voltammogram appears to be recorded in a solution with constant, increased temperature. This temperature can even be higher than the boiling-point, because the heated solution layer has no chance to form vapour bubbles during the short heating period. In this case, electrochemical experiments are performed in a metastable, over-heated state.Voltammetry at a permanently heated electrode The procedure has been described in detail in a previous paper. 14 Results and Discussion Temperature When the wire temperature values measured following the described experimental procedures were compared with the calculated values, a maximum deviation of +11.9% was found. This deviation is attributed mainly to the uncertainties in the potentiometric method. The maximum deviation between results of the resistance method and the calculated values was less than +5%. The reasonable agreement of measured and calculated electrode surface temperatures allows the conclusion that the calculated temperature profiles on the solution side give a realistic description.Temperature profiles were calculated for different heating current magnitudes and for a time interval up to 0.1 s after heating had started. Heat pulses Temperature Current Voltammogram I I I Time Fig. 1 Pulsing scheme (simplified) for creating a voltammogram with current sampling at the end of each heat pulse. Sampling periods are indicated by black squares.Analyst, December 1996, Vol. 121 1807 The method of digital simulation was used also to calculate concentration profiles for potentiostatic and galvanostatic electrolyses in the temperature regime of interest. An important result of the calculations was that in every case the thickness of the temperature dissipation layer is much higher than that of any diffusion layer which might develop in aqueous solution.Every concentration change, either accumulation or depletion, pro- ceeds inside an enveloping heated layer. Examples for the time dependence of the heated wire temperature are given in Figs. 2 and 3. With prolonged heating, the temperature rises rapidly up to a maximum that is followed by the establishment of a stationary temperature. The actual temperature in the rising part until about 0.1 s was found to be completely independent of electrode orientation, whereas the temperature values following this period depended on the y 100 2 3 80 60 3 a, Q 8) ' O I 20 '6 2 4 6 8 10 Ti me/s Fig. 2 Long-time temperature change of a platinum wire of 2 cm X 25 pm diameter when heated in water with a 100 kHz ac current.Heating current magnitudes, given in root-mean-square values (from top to bottom): 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, and 0.1 A. 160 T 120 1401 A 0.0 0.1 0.2 0.3 0.4 0.5 Time/s Fig. 3 Parameters as in Fig. 2. Short-time temperature change of a horizontal platinum wire. position, i.e., differences appeared between horizontally and vertically arranged electrodes. This behaviour is explained by the influence of convection caused by density gradients. Convection starts with some delay owing to mass inertia. It cools the electrode wire until a stationary state has been reached. This means that in experiments with permanent heating, the effect of convection has to be considered. The numerical calculation used here gave agreement only for the time period covering the mentioned rising part of the curves.It can be concluded that there exists, after heating has started, a time interval of approximately 0.1 s where convection does not have any substantial effect. This 'convection-free' interval was utilized for sampling voltammetry with heating pulses. Heating and cooling periods connected with a heating pulse of 0.1 s are shown in Fig. 3. It is also very important for application in sampling voltammetry that the electrode reaches its original temperature within less than 1 s after heating has stopped. Temperature values at the end of a 0.1 s heating pulse are given in Table 1. Sampling Voltammetry With Pulsed Heating In Fig. 4, a sampled voltammogram taken according to the procedure illustrated in Fig. 1 is given.Three of the curves correspond to temperatures above the boiling-point. Never- theless, they appear as smooth lines, as if they had been recorded in a homogeneous hot solution. Such measurements are commonly made in a pressurized vessel, with the con- sequence of destroying unstable substances. The voltammograms are composed of a large number of individual current samples. Every sample is part of a Cottrell- type response following the synchronous pulsing of heat energy and potential, as illustrated schematically in Fig. 1. The reasons for this response are the same as with ordinary chron- oamperometry, i.e., the growth of a diffusion layer in quiescent solution, caused by a stepwise potential variation. It should not be forgotten, however, that the thermo-e.m.f. contributes to the over-all potential due to the temperature coefficient of standard redox potentials.Fig. 4 shows the effects of gradually increasing temperature on a reversible redox couple. First, a shift of half-wave potential in the negative direction is found. This is in agreement with the generally accepted value of the temperature coefficient of the ferrocyanide/ hexacyanoferrate(i1)-hexacyanoferrate(II1) stan- dard potential. It was reported by Olivier et a1.17 to be -1.6 X 10-3 V K-1; in previous experiments14 we found a value of -1.56 X 10-3 V K-1. Obviously, heated electrodes offer a convenient way to determine redox entropies. Fig. 4 also shows the increase in diffusion current with temperature.With the studied reversible couple, this increase can be explained only by the enhancement of diffusional transport. As stated above, the procedure applied does not cause detectable additional convection during individual sampling periods. The increase in diffusion current must be due to increased values of the diffusion coefficient. It was found that the diffusion current increased by about 1% K-I. This is in agreement with the rule of thumb where this temperature dependence is expected to be 1-2% K-1. Some uncertainty arises, however, from the fact that the temperature in the diffusion layer is fairly uniform, but not completely constant. Nevertheless, the method should be useful for the estimation of diffusion coefficients at increased temperature. Table 1 Surface temperature (rounded values) of a heated 25 pm platinum wire at the end of a 0.1 s heating pulse Heating current/A,, 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Temperature/'C 22 30 37 50 67 90 116 140 170 2001808 Analyst, December 1996, Vol.121 Curves with only a moderate increased temperature, in contrast to that without heating, show a pure sigmoidal shape. This is explained by a gradual transition from partially linear diffusion to completely cylindrical for the electrode dimensions applied in the experiments. With increasing diffusion coeffi- cient, the Nernst diffusion layer becomes wider and thereby more curved, with the consequence of increased diffusional transport towards the surface of a cylindrical wire. At a given scan rate, the transition from linear to cylindrical diffusion is marked by a tendency to obtain sigmoidal instead of peak- shaped curves.This is also true for sampled voltammograms, where the scan rate is not referred to a continuous potential variation, but to stepwise changes. In Fig. 5 , the method described above was applied to the iron(m)-iron(rr) couple, which is considered to have some degree of irreversibility at room temperature. The heat effect here results in some enhancement of kinetics. In Figs. 6 and 7, two typical irreversible redox reactions are considered, i.e., the reduction of bromate in alkaline solution and the oxidation of oxalic acid in sulfuric acid. Both reaction mechanisms are complicated, and the temperature increase acts primarily on individual reaction steps. It is not the intention to explain the temperature effect in full detail, but merely to note that the electrolysis current is drastically increased.In all experiments with heated electrodes, a linear relation- ship between limiting diffusion current and concentration was found for the whole temperature range studied. Generally, the residual current did not vary very much with temperature. Thereby, the signal-to-noise ratio in voltammetry is improved even with reversible couples where the increase in the reaction rate does not play any role. One of the most difficult problems with heated electrodes is that the usable potential region becomes narrower when the temperature increases. With platinum, both the cathodic and the anodic current rises are shifted, mainly owing to enhancement Potentiall V "9 i -IS0 Fig.4 Sampled voltammograms of 5 X mol 1-I hexacyanofer- rate@)-hexacyanoferrate(ri1) (in 0.1 mol 1-I KC1) at a 25 pm Pt wire electrode (symmetrical arrangement consisting of two wires, each 5 mm in length) heated with an ac current of 100 kHz. Heating pulse, 0.1 s; interval between pulses, 1 c; staircase mode with potential steps of 0.01 V. Heating currents: A, 0: B, 0.3; C, 0.4; D, 0.5; E, 0.6; F, 0.7; G, 0.8; and H, 0.9 A,,,. of their reaction rates. This effect can be seen in the present figures. Additionally, the electrode metal tends to become less inert and less noble with increase in temperature. With platinum, there appear more or less expressed oxide peaks depending on the solution composition. So far, no useful alternative to platinum as an electrode material could be found.Other noble metals are not generally more stable. Glassy carbon fibres are destroyed rapidly when polarized in a heated state.18 ---ae- 5 E 3 g 150 I00 50 IJ Fig. 5 Sampled voltammograms of 5 X 10-3 mol 1-1 iron(n)--iron(rn) sulfate (in 0.1 mol 1-1 H2S04) at a 25 pm Pt wire electrode heated with different ac currents at 100 kHz. Heating currents: A, no heating; and B-J, varied from 0.1 to 0.9 A in steps of 0.1 A,,,,,. Other parameters as in Fig. 4. Potentiall V -1.2 -0.8 -0.4 0 20 0 4 -20 L 0 4 0 -60 -80 -1 00 -120 Fig. 6 Sampled voltammograms of 5 X 10-3 mol 1-1 KBr03 in 0.1 me1 1-1 KOH at a 25 pm Pt wire electrode heated with an ac current of 100 kHz. Heating pulse, 0.1 s; interval between pulse, 1 s; staircase mode with potential steps of 0.01 V.Heating currents; A-H, varied from 0.1 to 0.8 A in steps of 0.1 A,,,,,.Analyst, December 1996, Vol. 121 1809 4 300 4 E a 2 250 200 150 100 J 20 15 10 5b 2 5 5 -5 ' Fig. 8 mol 1-' formaldehyde in 0.01 mol 1-' sulfuric acid at a 2.5 pm platinum wire electrode (symmetric array, oriented vertically). Solid line, electrode heated continuously with 0.453 A,,,, 100 kHz ac; dotted line, no heating. Cyclic voltammogram of 10 Fig. 7 mol 1-1 oxalic acid in 0.1 mol 1-1 H2S04 at a 25 pm Pt wire electrode heated with ac current of 100 kHz. Heating pulse, 0.1 s; interval between pulses, 1 s; staircase mode with potential steps of 0.01 V. Heating currents: A, no heating; and B-J, varied from 0.1 to 0.9 A in steps of 0.1 A,,,,.Sampled voltammogram of 5 X 10 Voltammetry at a Permanently Heated Electrode Permanent electrode heating has been applied sucessfully in the stripping determination of lead. Low concentrations of Pb2f were deposited anodically as Pb02 at a heated platinum wire at 77 O C , and then stripped cathodically.lg In this case, the convection caused by thermal gradients proved to be a useful substitute for mechanical stirring during the deposition process. It was found that an increased electrode surface temperature drastically enhanced the deposition, probably by lowering the nucleation overvoltage. The temperature effect on the cathodic stripping step was less emphasized. With permanently heated electrodes, continuous potential ramps with arbitrary variation of scan rate can be applied.This may be useful if complex reaction mechanisms are under study, where single reaction steps can be manipulated by variation in the temperature level and scan rate. As an example, in Fig. 8 a cyclic voltammogram of formaldehyde in sulfuric acid is shown. On going from room temperature to 79 "C, the impact on a single reaction step is considerable. This provides an opportunity to study the redox reactions of unstable substances as a function of temperature. It should be kept in mind that even with permanently heated electrodes, delicate substances remain unaffected by heat until the molecules reach the hot reaction layer. Another example where permanent heating was useful is the reduction of dissolved oxygen in aqueous solution.~4 Conclusion The methods presented are suitable for analysing solutions of unstable or volatile substances at increased temperature. The temperature dependences of important quantities such as electrode potential, diffusion coefficient and reaction rate can be determined in a convenient manner.With pulsing sequences, electrochemical experiments above the boiling point can be performed. As a future application of pulse heating, electrochemistry in supercritical fluids will be possible. Another future application of very short pulses of extreme heating current magnitudes could be a special variant of thermal modulation voltammetry with a further improved signal-to-noise ratio. The methods can be extended easily to non aqueous solvents. In non-aqueous medium, higher temperature values can be reached with less energy demand because of the lower thermal conductivities of the solvents.Different forms of heated electrodes have been tested. With thin band structures on an inert support, more robust sensor arrangements suited for flow stream application are envisaged. With smaller electrode dimensions, the temperature remains nearly constant during most of the heating-up period. Hence complete voltammograms with a continous fast voltage sweep are realizable during one individual heating pulse. We are grateful to the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie for financial support. References 1 2 3 4 5 6 7 8 9 10 11 12 13 Gabrielli, C., Keddam, M., and Lizee, J. F., J . Electroanal. Chem., 1983,148, 293. Gabrielli, C., Keddam, M., and Lizee, J. F., J Electroanal. Chrm., 1993, 359, I . Miller, B., J . Electrochem. Soc., 1983, 130, 1639. Valdes, J. L., and Miller, B., .I. Electrochem. SOC., 1988, 135, 2223. Valdes, J. L., and Miller, B., J . Phys. Chem., 1988, 92, 4483. Valdes. J. L., and Miller, B., J. Phys. Chem. , 1988, 92, 525. Rogers, L. C., J . Electroanal. Chem., 1993, 356, 181. Benderskii, V. A., and Velichko, G. I., .I. Elertroanal Chem., 1982, 140, 181. Smalley, J. F., Krishnan, C. V., Goldman, M., and Feldberg, S. W., J. Electroanal. Chern 1988, 248, 255. Smalley, J. F, Macfarquhar, R. A., and Feldberg, S. W., J . Efectro- anal. Chem., 1988, 256, 21. Konovalov, V. V., and Raitsimring, A. M., .I. Electroanal. Cheni.. 1985, 195, 1.51. Benderskii, V. A., and Velichko, J . I., J . Eltwi-oanal. Chem., 1984, 181, 1. Griindler, P., Zerihun, T., Kirbs, A., and Grabow, H., Anal. Chrm. Acta, 1995, 305, 232.1810 Analyst, December 1996, Vol. 121 14 Zerihun, T., and Griindler, P., J . Electroanal. Chem., 1996, 404, Griindler, P., Zerihun, T., Moller, A., and Kirbs, A., J. Electrounul. 243. Chem., 1993,360, 309. 15 Britz, D., Digital Simulation in Electrochemistry, Springer, Berlin, Zerihun, T., and Griindler, P., J . Electroanal. Chem., 1996, in the 1988. press. 16 Frischmuth, K., Visocky, P., and Grundler, P., Znr. J . Eng. Sci., 1996, 34, 523. Paper 61038401 17 Olivier, A,, Merienne, E., Chopard, J.P., and Aaboubi, O., Electro- Received June 3,1996 chim. Acta, 1992, 37, 1945. Accepted August I, 1996 18 19

ISSN:0003-2654    

DOI:10.1039/AN9962101805

出版商:RSC    

年代:1996

数据来源: RSC

摘要:

Analyst, December 1996, Vol. 121 (1811-1815) 181 1 Electrochemical and Thermal Behaviour of Calcium-selective Membranes* Arthur K. Covington" and Eugenia Totub a Department of Chemistry, Bedson Building, University of Newcastle upon Tyne, Newcastle upon Tyne, UK NEl 7RU Institute of Microtechnology, P.O. Box 38-1 60, 72225 Bucharest, Romania An ion-selective membrane based on a polyimide polymeric matrix plasticized with diethylene glycol dibenzoate and containing calcium synthetic ionophores was used to obtain a potentiometric Ca2+ sensor. It gives a linear response over the pCa range 1-5 with a slope of 27.4 mV (pCa)-l. The sensor can be used for the pH range 3-8.5. The calcium-selective membranes were characterized in terms of their surface morphology and the structural parameter, the glass transition temperature.The ac impedance and FTIR analysis of the polyimide-based polymeric membrane were also studied. Keywords: Calcium ion-selective electrodes; ion-selective field eflect transistors; polyimide; polymer plasticizers Introduction Ion-selective electrodes (ISEs) based on ionophore-impreg- nated supported liquid membranes or polymeric membranes [typically plasticized poly(viny1 chloride) (PVC)] are now commonly used in a variety of analyses.l-4 Among these, neutral carrier electrodes are of practical importance in the determination of alkaline earth metal cations. Calcium ions are of special interest in this field owing to their clinical importance. The aim of this work was to obtain and characterize a calcium-selective membrane using polyimide as polymeric support.Quantitative electrochemical characteristics such as slope, detection limit and selectivity coefficients were used for the description of electrode functions5 and the glass transition temperature6 was used to describe the thermal behaviour of the polymeric Ca2+-selective polyimide membrane. In order to determine the main features of the membranes obtained, FTIR spectroscopy analysis, ac impedance measurements and scan- ning electron microscopy were used. Experimental Reagents All chemicals and solutions were of analytical-reagent grade (Merck, Poole, Dorset, UK); membrane components were purchased from Fluka (Buchs, Switzerland). The water used was either distilled once or de-ionized. The ionophores used for calcium were ETH 129 and ETH 1001.The membrane compositions consisted of an electroactive material, viz., the calcium ionophore, plasticizer [either dimethyl phthalate (DMP) or diethyl glycol dibenzoate (DGD)], lipophilic agent [potassium tetrakisb-chlorophenyl)borate] and a polymeric support, viz., polyimide. The polyimide was either encapsulant solution [PI 2555D, from DuPont (Stevenage, Herts, UK)], or polyimide powder [from Aldrich (Gillingham, Dorset, UK) or * Presented at the 6th European Conference on Electroanalysis, Durham, UK, March 25-29th. 1996. Scientific Polymer Products (SPP) (Ontario, NY, USA)], subsequently dissolved in anhydrous dimethylformamide (from Aldrich). Preparation of Ion-selective Membrane For Caz+-selective membranes the ionophore was added to a mixture of a lipophilic agent and plasticizer.This mixture was then added to a polyimide solution.7 The final mixture was then cast on a glass slide* and left to dry in a desiccator for 4-5 d to evaporate the solvent. Construction of Ca2+ ISE A circular section from the selective membrane was cut and mounted in a Philips IS-561 electrode body. The inner solution was a 0.01 mol dm-3 solution of the analyte ion. The electrodes were conditioned for 24 h before use in 0.01 mol dm-3 calcium chloride solution. Calibration A range of standard solutions were prepared in 0.05 mol dm-3 TRIS-HCl buffer (pH 7.2) (from 1.0 down to 10-6 mol dm-3 in calcium chloride). The potential difference reading, on a pH millivoltmeter, was taken 5 min after the electrode was introduced to the stirred solution, to allow time for a steady reading.The lowest concentration solution was measured first to reduce cross-contamination errors between adjacent solu- tions. A calomel electrode was used as the external reference electrode, with saturated KC1 as the bridge solution. All measurements were made at 20 "C in an electrochemical cell as follows: Hg I Hg2C12 I KCl,, 11 test solution I membrane I reference solution I AgCl I Ag The potential difference E (mV) was plotted against -log cca, where cCa is the primary ion concentration (mol dm-3), and the slope was calculated. Selectivity Measurements The selectivity coefficients, kyPot, for the Ca2+ ISE were determined by the separate solutions method. Using 0.1 rnol dm-3 solutions of the primary and interference cation, the selectivity coefficient was calculated from the equation9-10 (1) where i is the primary ion andj is the interferent ion, a , is the activity of solution species x in solution, Ex is the potential of the solution species x , and S = (RT/z,F)lnlO (2) where R is gas constant, T the absolute temperature, F the Faraday constant and z, the charge on species x.k,]POt = (a,/aJ"'/"J) 1 O(EJ-EJS1812 Aiialyst, December 1996, Vol. 121 Thermal Study Thermal analysis curves were recorded with a Perkin-Elmer thermal analysis system at constant heating rates of 2,5, 10 and 20 K/min in a nitrogen atmosphere provided by a continuous gas flow of 60 cm3 min-1. Surface Characterization Scanning electron microscopy (SEM) provides a very conve- nient and simple method for characterizing and investigating the structure of selective membranes.The morphology of the polymer material matrix used for membrane preparation directly affects its permeability. Consequently, we performed SEM analysis in order to have a clear view of the over-all structure of the calcium-selective membranes. Impedance Measurements All ac impedance measurements using a four electrode cell were carried out at room temperature using an IM5 Zahner system. In the four-electrode mode the impedance of two silver/silver chloride reference electrodes is compensated and the impedance of the membrane itself is determined. Silver/silver chloride electrodes prepared by chloridizing silver were used. The soaking solution used in the cell was 0.01 mol dm-3 aqueous CaC12 containing a small amount of LiCl in order to produce a 0.01 mol dmP3 chloride ion concentration to give adequate conductivity to the solution and maintain the stability of the silver/silver chloride reference electrodes used.F TIR Spectroscopic Measurements IR investigations were performed either on a dry selective membrane or after exposure to aqueous solution, in order to establish the influence of water. 1R spectra recorded with a Nicolet (Warwick, UK) FTIR spectrometer were also obtained for the initial polymeric matrix with or without plasticizer. Results and Discussion Electrochemical Study Calibration curves (Fig. 1) were obtained for all the ISEs used, by performing electrochemical measurements (dip tests), and slope values were calculated (Table 1).These values are lower than those predicted by the Nernst equation." For the Ca2+ membranes the slope increased from 18.5 mV (log cCa2+)-1 (no buffer) to 22.8 mV log (log cCa2+)-I in buffered solution (pH 7.2) for the same electrode. The slope values ranged between 160 T .- + c a, 0 c, a 40 " 7 -6 -5 -4 -3 -2 log [Gal Fig. 1 Calibration curves for calcium ISEs based on polyimide polymeric membrane. A , encapsulation polyimide, DGD, ETH 1001; W, polyimide, DMP, ETH 129; a, polyimide, DGD, ETH 129; +, polyimide, DGD (from SPP), ETH 129. 21.4 and 27.4 mV (log cCa2+)-I for the different formulations. A slight decrease in slope with time was observed. For an electrode with an initial slope of 26 mV (log cCa2+)-1, the slope gradually decreased over 4 weeks to 19.6 mV (log cCa2+)-l at pH 7.2.The main electrochemical characteristics of the calcium- selective electrodes are presented in Table 2. The selectivity coefficients, k p t , for two of the calcium electrodes towards different cationic species, j,+, were evaluated by the separate solutions r n e t h ~ d . ~ Both electrodes gave good selectivity for ions of biological interest. The detection limit and the response time for the electrode with a slope of 27.4 mV (log cCa2+)-l (Table 3) were determined according to IUPAC definitions.5 The calcium-selective polyimide membrane soaked in cal- cium chloride solution displayed well resolved bulk impedance semicircles. The bulk resistance decreased with the condition- ing time of membrane. It was concluded that the energy of transfer of calcium ions across the polymer membrane de- creases with increasing conditioning time.According to the ac Table 1 Calibration slopes for Ca2+ ISE Calcium-selective electrode Slope/mV (pCa2+)- I Encapsulation polyimide matrix with DGD plasticizer and Polyimide* (solution in DMF) matrix with DMP plasticizer Polyimide" (solution in DMF) matrix with DGD plasticizer Polyimidef (solution in DMF) matrix with DGD plasticizer PVC matrix with o-NPOE* and ETH 129 l 2 PVC matrix with n-NPOE and ETH 1001 I 3 * o-NPOE = o-nitrophenyl octyl ether. active compound ETH 1001 22.8 and active compound ETH 129 21.4 and active compound ETH 129 27.4 and active compound ETH 129 26.4 29.0 28.5 * Polyimide powder from Aldrich. t Polyimide powder from SPP. Table 2 Electrochemical characteristics of Ca2+ ISE based on polyimide matrix Characteristic Linearity range/mol dm-3 Calibration slope/mV (pli1)- Standard deviation of slope Correlation coefficient ( r ) Detection limit/mol dm-3 Response time/min pH range* Value 1 27.4 0.5 0.9960 6 X 10-5 5 x 10-5-10-1 < 3 3 .O-8.5 * The measurements were made by small additions of 0.1 mol dm-3 HCl or 0.1 rnol d 1 r 3 NaOH solutions.Table 3 Potentiometric selectivity coefficients (klJPut) for ion-selective electrodes (primary ion, i, = Ca2+) ISE Na+ K+ NH4+ Mg2+ ETH 100 1, polyimide matrix, ETH 129, polyimide matrix, ETH 1001, PVC matrix, ETH 129, PVC matrix, DGD membrane -2.15 -3.31 --2.25 -2.30 DGD membrane -2.10 -3.25 -2.21 -3.90 o-NPOE membrane l4 -3.6 -3.7 - -4.2 o-NPOE membrane ' 5 -3.7 -4.0 - -4.9Analyst, December 1996, Vol.121 1813 impedance spectra, both Nyquist and Bode plots (Fig. 2) are consistent with an equivalent circuit for a non-blocking interface16 with the bulk resistance of the membrane under study, Rb, in series with a parallel RcJCdl circuit (Rct is the charge-transfer resistance and Cdl is the double-layer capaci- tance). From impedance data a resistance of 5.7 X lo5 S2 cm was determined for a PI-DGD membrane, compared with 9 X 105 Q cm for a PVC-o-NPOE membrane. The bulk semicircle from the Nyquist diagram is followed by a series of points on the real axis at lower frequencies. No linear section (Warburg impedance contribution) is shown in the impedance spectra. Ion-exchange processes are reversible for the frequency range 10-0.01 Hz; only the bulk semicircle appears.Rapid ion exchange occurs at the surface with no Warburg diffusion, or surface reaction, kinetic limitation. A polyimide selective membrane conditioned in TRIS-HCl buffers of different pH showed on the Nyquist diagrams bulk impedance semicircles with the bulk resistance increasing with increasing pH (Fig. 3). Owing to the presence of nitrogen atoms in the polyimide matrix, this was probably due to the need to 'pump' hydrogen ions into the polymer membrane. Therefore, hydrogen might be an interferent for the polyimide calcium-selective membrane but the electrochemical determinations showed no effect of hydrogen ions on the Ca2+ response. It is suggested that the energy of transfer of ions across the polyimide membrane also increased with increasing pH.Thermal Study Two structural parameters that affect membrane permeation very strongly are the glass transition temperature (T,) and the crystallinity.17 Since the transport proceeds mainly via the amorphous regions, it is very important to know the degree of the crystallinity in the polymer. As consequence, in order to obtain a suitable ion-sensitive membrane, the Tg of the + Phase -A- impedance I 10000 75T \ T 9000 f 301 a Frequency / Hz Fig. 2 129). Bode plot for polyimide calcium membrane (PI, DGD, ETH l 2 O I 01 4 4 0 40 80 120 160 200 :(real)/ k G? Fig. 3 Nyquist plots of polyimide membranes immsersed in (+) pH 6.0, (a) pH 7.0 and ( A ) pH 8.0 TRIS-HC1 buffered solutions (frequency 105-1 0- 1 HZ). polymeric matrix is a major factor.Some thermogravimetric (TG) and differential scanning calorimetric (DSC) experiments were performed on Ca2+-selective polyimide membranes using two plasticizers: diethylene glycol dibenzoate (DGD) and dimethyl phthalate (DMP). For the compositions containing only DGD, the initial decomposition temperature of the polymeric matrix decreased with increasing amount of DGD added. Plasticizer mixtures of DGD and DMP, also used for some polymeric membranes, did not affect the initial decom- position temperature of the polyimide. Compared with the initial polymer, the polyimide, the compositions with DGD had a lower initial decomposition temperature of polyimide (390 "C). If DMP plasticizer was used instead, this value was not modified. The polymeric films obtained with DGD as plasticizer were transparent, flexible and without microstructure characteristics of the surface (wrinkles, voids or holes).The Tg values for different calcium-selective polyimide membranes using DGD as plasticizer ranged from 9 to 90 "C ( Fig. 4). Suitable ion-selective electrodes could be obtained using compositions with Tg between 9 and 15 "C. When using DMP up to a concentration of 15% m/m, in the polymer composition, transparent, flexible and mechanically resistant membranes were obtained. With increasing DMP concentration above 15%, the membrane became slightly opaque and lost their mechanical strength. At 25% DMP, the samples were completely opaque and brittle. From the Tg values of the compositions when the two plasticizers were combined with a total polymer composition of 60%, increasing the proportion of DMP lowered the plasticization effect, and it was possible to incorporate DMP at levels up to 25%.When used alone, the maximum incorporation achieved for DMP was 15%. The Tg values for the different plasticizer concentrations where the two plasticizers were combined in a total amount of 60% are shown in Fig. 5. At temperatures above 400 "C and under non-oxidative conditions, the polymeric matrix decomposed with evolution of 150 T 120 I\ 30 60> 0 0 20 40 60 80 Plasticizer (YO) Glass transition temperature (T,) of polyimide membrane composi- Fig. 4 tions as function of plasticizer (DGD) content. OOT ,* t / 0 20 40 60 80 100 Plasticizer (YO) Fig. 5 mixture. Tg values for different compositions of polyimide with DGD-DMP1814 Analyst, December 1996, Vol.121 carbon monoxide and carbon dioxide, as determined using a coupled gas evolution detection system. Hydrogen evolution occurred above 525 "C. A mobile equilibrium is proposed between the normal and isoimide forms of the repeat units to account for the products. Using thermal analysis in a nitrogen atmosphere with a heating rate of 5 K min-I, the thermal stability, as determined by mass loss, shows that the polyimide membrane partially decomposes, forming a char above 700 "C. The thermograv- imetric results obtained from TG data are presented in Table 4. The water is lost in two stages, which signifies that the water molecules either are adsorbed on the surface or are constitution water.The amount of water first eliminated is larger than that adsorbed from the oven atmosphere, which means that on the surface of the polyimide membrane there exists an important amount of adsorbed water. Surface Characterization Fig. 6(a) shows the top surface of a calcium-selective polyimide membrane after 24 h of conditioning in 0.01 mol dm-3 CaC12 as observed by SEM. The geometry of the pores can be clearly seen. As the 'holes' do not penetrate into the depth of the membrane [Fig. 6(b)], its morphology is in accordance with its Table 4 Mass variation of polyimide film (2Cr500 "C; sample mass 3.174 mg) Temperature range/"C Mass loss/mg Assigned process 25-1 05 -0.0 10 Absorption of H20 105-1 17.8 0.016 Elimination of absorbed H20 117.8-140 0.0 19 H20 IOSS 140-290 0.035 co loss 307.9-396 0.037 CO and C02 loss 396-498 0.165 CO and C02 loss electrochemical behaviour as a calcium selective membrane. There is a difference between the top and the bottom surface membrane morphology [Fig.7 ( a ) and (b)] as the top surface was exposed to the atmosphere, with no contact with any solid surfaces during casting, whereas the bottom surface was formed in contact with the glass slide. The morphology of the membranes presents a consistent composition and character- istics of an asymmetric membrane. The components of the membrane were compatible so no solid inclusions appeared in the bulk of the membrane. After soaking the polyimide membrane for a longer period (72 h) in 0.01 mol dm-3 CaC12 solution, a dramatic change was observed in surface morphology with the appearance of a spongy aspect, However, the electrochemical measurements demonstrated that the membrane still gave a good response to Fig.7 (a) Top surface morphology of a polyimide membrane as observed from SEM micrographs (magnification X1500). (b) SEM image of the bottom surface morphology for the same polyimide membrane (magnifica- tion X1500). fa} Top surface morphology (b) Bottom surface morphology Fig. 6 SEM images of calcium-selective polyimide membrane after Fig. 8 SEM image of top surface for a polyimide membrane after conditioning for 24 h in 0.1 mol dm-3 CaCI2 solution (magnification X exposure for 72 h in 0.01 mol dm-3 CaC12 solution (magnification 1 5 000). X 160).Analyst, December 1996, Vol. 121 1815 calcium ions.The FTIR spectra of the initial polyimide powder and of the membranes obtained (either dry or after exposure to aqueous solution) showed one wide peak at 3330 cm-1, which presumably represented a combined peak of both symmetric and unsymmetric stretching vibrations of the N-H bond due to the polymer n-system. Other IR absorption peaks observed seemed to originate from the stretching vibration of the ArC=N double bond, the skeletal vibration of the benzene ring and the characteristic peaks for 1,4-substituted benzene. The IR spectra of the membrane after exposure to aqueous solution show in the 3000-3200 cm-1 region a strong signal characteristic of imides. It was concluded that the water acts as a plasticizer for polyimide. In the 1600-1800 cm-1 region stronger bands characteristic of OH- occur and, in correlation with other experimental data, it is assumed that water is not only acting as a plasticizer modifying the structure, but also exists adsorbed on the membrane surface, as already observed from the TG data presented above.As a consequence of polyimide plasticization by water, the combination band from 2500 cm-1 is displaced towards 3000 cm-1 because of weak hydrogen bonding, the polyimide being imidized in a greater proportion. Conclusions Although the electrochemical responses for electrodes presen- ted above are sub-Nernstian, their selectivities towards cations of biological interest are reasonable. The results obtained show that polyimide could function as suitable polymeric matrix for calcium ion-selective membranes and would be particularly useful for polyimide encapsulated ISFETs.8 During the last decade, great interest has been directed towards miniaturization of ISEs, and as consequence many types of chemical sensors were developed.ISFETs belong to the modern generation of chemical sensors based on the MOSFET structure. One of the most important challenges that arises during ISFET construction is solving the problem of the adhesion of a thin membrane to the transistor insulator. Polyimide compounds show a good adhesion to integrated circuit surfaces and polyimides are compatible with standard planar silicon technology. Taking into account these consider- ations, permselective membranes which use polyimide as a matrix would seem to be much more compatible than PVC membranes with modern microelectronic technology.Preli- minary tests, performed in an ultrasonic bath, have shown that polyimide calcium-selective films applied to a silicon nitride surface peeled from the surface of the wafer after longer times than PVC membranes. From the IJVg characteristics, the correct behaviour of devices loaded with selective polyimide membranes is obtained. These results demonstrate that a polyimide polymeric matrix could be successfully used for making suitable ISFETs. The experiments performed showed that samples with similar temperature-mass curves and Tg values, often showed com- pletely different electrochemical behaviour. Therefore, thermal analytical characterization (TGA, DSC or TGD) of the membranes is not sufficient by itself.More detailed studies are needed to establish the possible correlation between the electrochemical behaviour of a selective membrane and its thermal characteristics. The compatibility of polyimide with DGD and DMP was evaluated by the DSC method, indicating a limiting value of 40% plasticizer in the polymer composition. In spite of its limited compatibility DMP (15%) can act as a secondary plasticizer enhancing the over-all performance of the compositions. It is well known that an important aspect of trying to develop an alternative membrane matrix to PVC is that not every polymer can hold the large amounts of plasticizer that are necessary for proper functioning as an ion-selective membrane. Therefore, the use of other plasticizers that fulfil the demands of high lipophilicity , no exudation and no crystallization in polyimide membranes, and polyimide membrane-based, cal- cium-sensitive microelectronic devices are the topics of further investigation. We thank M.Odlyha and J. Slater of Birkbeck College, London, for their valuable help with the thermal analyses of the polyimide membranes. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Moody, G. J., Saad, B. B., and Thomas, J. D. R., Sel. Electrode Rev., 1988, 10, 71. Thomas, J. D. R., Anal. Chim. Acta, 1986, 180, 289. Oesch, U., Ammann, D., and Simon, W., Clin. Chem., 1986, 32, 1448. Arnold, M. A., and Meyerhoff, M. E., Crit. Rev. Anal. Chem., 1988, 20, 149. Buck, R. P. and Lindner, E., Pure Appl. Chem., 1994, 12, 2527. Dunn, J. G., and Sharp, J. H., Treatise on Analytical Chemistry. Part I . Thermal Methods, ed. Kolthoff, I. M., and Winefordner, J. D., Wiley-Interscience, New York, 2nd edn. 1993, vol. 13, pp. 224- 229. Totu, E., and Covington, A. K., unpublished results. Cha, G. S., and Brown, R. B., Sens. Actuators, B , 1990, 1, 281. Buck, R. P., and Cosofret, V. V., Pure Appl. Chem., 1993,8, 1849. Umezawa, Y., Umezawa, K., and Sato, H., Pure Appl. Chem., 1995, 3, 507. Craggs, A., Moody, G. J., and Thomas, J. D. R.,J. Chem. Educ., 1974, 51, 541. Gehring, P., Rusterholtz, B., and Simon, W., Chimia, 1989, 43, 377. Kelly, P. M., PhD Thesis, University of Newcastle upon Tyne, UK, 1993. Sokalski, T., Maj-Zurawska, M., and Hulanicki, A., Mikrochim. Acta, 1991, 1, 285. Schafer, U., Ammann, D., Pretsch, E., Oesch, U., and Simon, W., Anal. Chem., 1986, 58, 2282. Bard, A. J., and Faulkner, L. R., Electrochemical Methods, Wiley, New York, 1980. Privalko, V. P., Macromolecules, 1973, 6( 1 l), 11 1. Paper 6103835B Received June 3,1996 Accepted August 21, I996

ISSN:0003-2654    

DOI:10.1039/AN9962101811

出版商:RSC    

年代:1996

数据来源: RSC