摘要:

Studies of the Cathodic Adsorption of Hydrogen and the Anodic Formation of Oxide on Platinum in Perchloric Acid Solutions using Modulated Specular Reflectance Spectroscopy BY A. BEWICK AND A. M. TUXFORD Dept.of Chemistry The University Southampton SO9 5NH Received 28th September 1970 The changes in reflectivity of a platinum electrode in molar perchloric acid accompanying the formation of a layer of oxide or of adsorbed hydrogen were investigated in the wavelength range 200-600 nm using modulated specular reflectance spectroscopy. Two correlation techniques phase sensitive detection or signal averaging were used to achieve a high sensitivity (f2 x The origin of the reflectivity changes is discussed and it is shown that a simple three-phase model accounts tor most of the effects. Two types of adsorbed hydrogen were distinguishable and the implications of the special optical properties of the more strongly bound species are discussed.Optical techniques are being used increasingly for the study in situ of the electrode- solution interphase. Ellipsometry is well established in the electrochemical field as a tool to investigate the growth and the nature of thin layers on electrode surfaces and it has been applied recently to the study of adsorbed 1ayers.l The interface has been probed from the electrode side using attenuated total reflectance with semi- conductor electrodes and with thin metal electrodes deposited on transparent sub- has been developed into a useful technique to study reaction intermediates in the reaction-diffusion layer adjacent to the electrode surface.More recently direct specular reflectance spectroscopy has been applied to electrochemical ~ystems.~-~ This technique has developed directly from that used to study electro-reflectance effects in semiconductors using electric field modulation. Applications have emphasized on the one hand the properties of the metal and on the other the accessibility of electrochemically relevant information for surface phases and adsorbed layers. In view of the increasing use of the specular reflectance methods it is important to establish the nature of the various effects that can give rise to reflectance changes and their relative importance in different systems. Feinleib demonstrated large changes * in the reflectance of gold electrodes in response to changes in their potential. Prostak and Hansen showed that this effect could be consistent with a model in which the optical properties of the metal are perturbed by the changing free electron concentration ; however,1° this treatment does not fit measurements on copper and on silver electrodes.The effect is linear with charge and its sign depends upon the sign of the charge unlike the Franz-Keldysh effect which is a field effect and independent of the sign of the field. They also calculated that the effect could not be due to the formation of a dielectric layer on the electrode or to changes in the refractive index of the double-layer region of the electrolyte (see also Stedman l). This reflectance change with charge is particularly large for a wavelength region in which the optical 114 Transmission spectroscopy through optically transparent electrodes A .BEWICK AND A . M. TUXFORD 115 constants are changing rapidly e.g. for gold there is a very large effect near 550 nm corresponding to the 5d-6s interband transition; the steep edge in the reflectivity- wavelength curve acts like an amplifier characteristic so that a small modulation of the optical properties results in a large change in reflectivity. This has been applied elegantly to interesting electrochemical systems by observing the charge on the electrode resulting from the formation of an oxide layer or from the adsorption I ~~ FIG. 1 .-Block diagram of apparatus. L 200 W lamp ; M monochromator ; P polarizer ; PM photomultiplier ; A amplifier ; 0 oscilloscope ; Pot potentiostat ; WFG wave form generator and linear sweep unit ; Osc sine wave oscillator ; PS phase shifter ; C phase sensitive detector or signal averager.of ionic species. There are two major shortcomings to this method; it is restricted to the use of electrode materials possessing suitable optical properties in the accessible spectral region and it gives electrochemically significant information in quantity but not quality i.e. the optical properties of the metal dominate those of the solution side of the interphase and the latter cannot be measured and analyzed to obtain data on the nature of absorbed layers etc. It is clearly of great importance to refine the reflectance technique to provide a generally applicable method to study the nature and kinetics of formation of films adsorbed layers and intermediates at electrode surfaces. This paper describes some applications of such a technique.EXPERIMENTAL A block diagram of the apparatus is shown in fig. 1. The light source was an Englehard- Hanovia 200 W high pressure Xe-Hg arc. Desired wavelengths were selected with a Bausch and Lomb high intensity monochromator. An E.M.I. 6256 B photomultiplier was used as detector. A sine wave oscillator provided a trigger signal for the Chemical Electronics type RB1 waveform generator which provided the square pulse that was applied to the cell via a Chemical Electronics potentiostat. The current-time transients were monitored on an oscilloscope during the optical measurements. The oscillator also provided a reference 116 MODULATED SPECULAR REFLECTANCE SPECTROSCOPY signal fed via a Brookdeal type 421 phase shifter for the phase-sensitive detector (PSD) (Brookdeal type 41 1).The output of the phase shifter was also used as trigger for the signal averager. The signal averagers used were the Nuclear Measurements 546C the Fabri-Tek 1070 and the Hewlett-Packard model 3721 A correlator. The cell was glass with a Spectrosil window. Parallel-polarized light which allows higher sensitivity was employed for those measurements at wavelengths> 275 nm (the cut-off wavelength of the polarizer) and for most experiments the angle of incidence was 45". The working electrode was a bright platinum sheet 0.5 cm by 0.5 cm and the sub- sidiary electrode was a platinum wire arranged to give uniform current distribution without obstructing the light path. All the potentials quoted are on the normal hydrogen scale. The values of AR/R quoted were those defined by the quantity (R,-R2)/R where R1 and R2 are the values of the reflectivity of the electrode at the two potentials defined by the square wave and R is the average value.With the system described the limiting sensitivity was ARIR = f2x RESULTS AND DISCUSSION Fig. 2 shows the change in reflectance of a platinum electrode in molar perchloric acid as its potential is slowly cycled at a constant rate from a value just anodic to hydrogen evolution up to a potential well into the oxide region. A 30 Hz 150 mV square pulse was superimposed on the slow sweep (added anodically to the quoted bias potential) and the fractional change in reflectance ARIR produced by this square wave modulation was measured using the phase sensitive detector. In the + 2 I ' I I I00 5 00 1000 Bias potential (mV) FIG.2-Change in reflectance with potential at 365 nm of a Pt electrode in perchloric acid as the potential was linearly and cyclically varied at 125 mV/min. Modulation in the form of a 150 mV 30 Hz anodic square wave was superimposed on the sweep and the optical response detected with a P.S.D. oxide region above 550 mV bias AR/R becomes large and positive in the hydrogen adsorption region it is negative and in the double-layer region only small values are obtained. These changes in reflectance cannot be ascribed to the charge effect described by Prostak and Hanson because on their model an increase in the amount of adsorbed hydrogen which is accompanied by an increase in negative charge density on the metal should change R in the same direction as a decrease in the amount of platinum oxide.This conclusion is supported by data on the wavelength A . BEWICK AND A . M. TUXFORD 117 dependence of the reflectance (see later) and together with Parsons’ measurements on copper lo must cast doubts on the applicability of the model. If the changes were due to a Franz-Keldysh effect AR/R would be significant also in the double layer region and not just in potential ranges corresponding to the formation of layers. The values observed for the double-layer region are of the order predicted by Sted- man l 1 using a model based on concentration changes in the interfacial region. It will be shown later that a simple three-phase model fits the results quite well. THE OXIDE REGION Fig. 3 is a plot of AR/R as the electrode is pulsed increasingly anodic into the oxide region from a fixed bias potential of + 670 mV and fig.4 shows the correspond- ing wavelength dependence. A small signal corresponding to the formation of a small amount of material is detectable at about + 700 mV. The signal remains small until about + 1 V ; it then increases smoothly and rapidly to large values and begins to level off at about + 1.5 V ; measurements beyond this potential are obscured amplitude of anodic pulse (mv) FIG. 3.-AR/R for platinum oxide at 425 nm as a function of the anodic pulse amplitude at 30 Hz superimposed on a fixed bias potential of +670 mV. by oxygen evolution. These results are in a close agreement with the ellipsometric measurements of Greef l2 and they confirm his comments on the work of Reddy Genshaw and B ~ c k r i s .~ ~ Although there is clear evidence for the formation of an incomplete layer in the potential range +700 mV to + 1 V there is a rapid change beyond + 1 V. Greef pointed out that these changes are closely paralleled by the charge-potential characteristic. The correlation between charge and AR/R is not complete in the region +700 mV to +1 V (AR/R levels off while the charge is still increasing) and there might be a change in the nature of the layer to one with different optical constants at about + l V. However the interpretation is made difficult by the need to allow for other contributions to the reflectance change at small values of AR/R (Stedman-type double-layer effects and the charge effect of Prostak and Hanson). The sensitivity of the method in the present context is such that AR/R values of about are discernable and this should be compared with the value of 2 x observed at short wavelengths for the oxide layer at + 1.5 V.The wavelength dependence of the reflectance of a thick oxide layer fig. 4 shows a substantial rise in AR/R at shorter wavelengths with a maximum at about 250 nm. Calculations for a simple model were made to determine if this spectrum was due to 2 x 118 MODULATED SPECULAR REFLECTANCE SPECTROSCOPY the metal or to the oxide layer. The dotted line on fig. 4 shows the computed values of AR/R for a three layer metal-oxide-solution model in which the optical constants for the solution and also those for the oxide layer (these latter values were those quoted by Reddy et a1.l3) were kept wavelength invariant and only those for the platinum 300- w s! 5 x 200- % Q 100 1 I I I I 79bo 2 5 0 300 350 400 450 wavelength (nm) FIG.4.-Spectral dependence of AR/R. (a) experimental data for platinum oxide using a 30 Hz square wave between $670 mV and 41.47 V ; (b) calculated curve for the three-layer model. - were given a wavelength dependence.14 This is an extreme model since the constants for the oxide probably vary significant1y,l4 but it serves to show the effects due to the metal. An oxide thickness of 0.5 nm corresponding to a potential of 1.5 V was used and the reflectance equations derived by Hansen l5 were employed. Over the range covered both curves have the same form the dip at 310 nm being present 1 I I I I I I I I ' 2 0 0 2 5 0 3 0 0 350 400 4 5 0 5 0 0 5 5 0 6 0 0 wavelength (nm) FIG. 5.-Spectral dependence of AR/R in the double layer region on platinum.30 Hz square wave from + 370 to +470 mV. on each but it was not possible to span the peak at 250 nm due to the lack of data for platinum. The poor fit in terms of absolute magnitude could be due to the simplified optical model the appreciable time required to form and to remove the oxide and to uncertainty in the optical parameters for the oxide. The assignment of the spectrum to the optical properties of platinum is confirmed by the spectrum A. BEWICK A N D A. M. T U X F O R D 119 5 0 0 300 2 00 1 0 1 1 2 00 3 0 0 4 0 0 5 0 0 600 700 8 00 wavelength (nm) FIG. 6.-Calculated electroreflectance spectrum for platinum using the Prostak and Hansen model Angle of incidence 70" and assuming a 1.6 x J modulation of the optical parameters.observed in the double-layer region fig. 5 where no layer is formed on the electrode. Fig. 6 is a calculated curve for platinum using the Prostak and Hansen model based on charge modulation and it is clear that this effect is not important in the present case. Preliminary measurements have been made using signal averaging techniques anodic cathodic period period FIG. 7. The time dependence of the reflectivity change at 435 nm for the formation and removal of platinum oxide at 30 Hz in the potential range +670 mV to + 1.47 V averaged from 512 repeats, 120 MODULATED SPECULAR REFLECTANCE SPECTROSCOPY of the time dependence of the reflectivity as the oxide layer is formed or removed at constant potential fig. 7. Time dependent information can be obtained using the P.S.D.and a variable modulation frequency. However the integration of the time information by the P.S.D. means that the output is one further step removed from the measured parameter and this can easily lead to ambiguities in mechanistic assign- ment. This difficulty is strikingly demonstrated for the hydrogen adsorption on platinum (next section). The characteristic in fig. 7 is simple and shows that the formation and the removal of the major part of the oxide layer has a time scale of about 10ms. Work is continuing on this aspect to obtain information on the growth mechanism of the oxide- + 4 0 t 20- \o 2 0' THE HYDROGEN ADSORPTION REGION A number of measurements were made in the hydrogen adsorption region with potential pulses and the P.S.D. in an analogous manner to that described for the oxide region.Measurements at different wavelengths were made in various ways fixed anodic bias potential with varying cathodic pulses fixed cathodic bias potential with varying anodic pulses and varying bias potential with fixed pulse height. An example is given in fig. 8. In most cases the reflectance-potential characteristics showed changes of slope at potentials corresponding to well-known features on the current-potential curves i.e. to the two peaks on cyclic voltammograms that come spond to different types of adsorbed atomic hydrogen. Arrows on the figure mark the peak potentials for the present system. The wavelength dependence fig. 9 is very similar at shorter wavelengths to that for platinum oxide fig. 4. This indicates that the major factors determining the spectrum in this region are the optical para- - o-o- I 1 I I I 5 0 200 2 5 0 3 60 /3 SO 4 00 -60 ~~~~~ bias potential (mv) FIG.8.-AR/R at 438 nm for platinum in the hydrogen region against bias potential with a 200 mV 30 Hz anodic square-wave modulation. The arrows mark the potentials of the hydrogen adsorption peaks observed by cyclic voltammetry. o-g-0-0 - meters of the platinum substrate and again a simple three-layer model for the system would fit the data using a reasonable value for the optical constants of the adsorbed layer. A closer analysis of the results obtained with the P.S.D. indicated a complex situation (see e.g. the hump in fig. 2) and it was considered desirable to observe directly the time effects using signal averaging to achieve the necessary sensitivity. Fig. 10 and 11 show the time dependence of the reflectivity in response to a A .BEWICK AND A . M. TUXFORD 121 - 2 4 1 2& 250 3 0 0 350. 400 4 5 0 5 0 0 5 5 0 6 0 0 wavelength (nm) FIG. 9.-Spectral dependence of AR/R for platinum pulsed at 30 Hz between different potentials in the hydrogen region 1 f20 to +120mV; 2 +120 to +320mV; 3 +70 to +190mV; 4 +190 to 3-390 mV. A B &Anodic. --&Cathodic -4 period period FIG. 10.-The time dependence of the reflectance change at 435 nm on switching the potential with a 30 Hz square wave from +320 mV to the following potentials (a) +20 mV ; (b) +70mV; (c) +120 mV; (d),+170 mV. Each curve averaged from 512 repeats. 122 MODULATED SPECULAR REFLECTANCE SPECTROSCOPY square-wave potential profile across different parts of the hydrogen region.In the sequence fig. 1Oa-d the region is scanned from a fixed anodic potential on the edge of the hydrogen region + 320 mV. Curve 10a can be analyzed in terms of a square- wave component (dotted line) corresponding to a decrease in R in the cathodic branch as the adsorbed hydrogen layer is formed together with a slow decrease in R on the anodic branch between A and B and a spike at B just as the potential is switched to the cathodic value. As the cathodic potential is reduced load the square wave component is diminished ; it has completely disappeared in 10d and all that remains R t k Anodic .-+-Cathodic -I &Anodic -+-Cathodic -cl FIG. 11.-The time dependence of the reflectance change at 435 nm on switching the potential with a 30 Hz square wave from +70 mV to the following potentials (a) + 120 mV ; (b) + 170 mV ; (c) f220 mV ; (d) +270 mV ; (e) +320 mV ; (f) +370 mV ; (g) +420 mV ; (h) +570 mV.is the slow decrease corresponding to the section A-B in lOa this feature being retained throughout the sequence. As a result 1Od resembles a square wave of opposite phase to that of lOa i.e. the formation of the adsorbed layer leads to an increase in the reflectance. The transients fig. 1 la-h illustrate the corresponding changes when the cathodic potential is fixed and the anodic value is changed. The slowly- A . BEWICK AND A . M. TUXFORD 123 falling component and the spike on the anodic branch gradually develop as the potential pulse is extended to more anodic potentials and the time scale of the fall decreases. In addition the wavelength dependence of the transients was examined.In every case the square-wave component increased in magnitude as the wavelength was shortened and the slow fall and spike became undetectable in the u.-v. region although they were prominent throughout the whole of the visible part of the spectrum. Part of this spectral dependence can be seen from the curves in fig. 9 (P.S.D. output) in which the transition from negative to positive values of AR/R varies with potential. The following deductions can be made from this data. (i) There are two differing types of adsorbed hydrogen which are optically distinguishable and these are formed at potentials in agreement with the established electrochemical data. (ii) The more strongly bonded hydrogen that adsorbed at more anodic potentials leads to an increase in the reflectance in the visible region.(iii) The less strongly bound hydrogen causes a decrease in the reflectance and this change is larger at shorter wavelengths as is also observed for platinum oxide. A mechanism can be put forward to account for the time-dependent features of the transients. We consider fig. lO(a) which corresponds to a full excursion over the hydrogen region. At the end of the cathodic part of the pulse the electrode is covered with a layer composed of both types of adsorbed species and when the potential becomes anodic the weakly bound hydrogen H, is rapidly ionized off and the reflectance increases to the value at A Hgds+ H ++ H,BdS' Fast C During the anodic period the strongly bound hydrogen H is ionized off relatively slowly and this produces the slow decrease in R Hidr + H +.e slow This rate is potential dependent as seen in the sequence ll(c)-(h). When the potential is first switched to the high cathodic value at point B the H is rapidly formed causing an increase in R until the reflectivity change becomes dominated by the formation of H, which leads to an overall decrease in R. These two processes cause the spike at B. The phase-reversed characteristic fig. lO(d) is readily accounted for on the same basis when it is noted that Hw is not formed in the potential range of this experiment. Although the layer of H appears to give rise to a broad band in the visible fig. 9 it is not suggested that this spectrum is caused by the platinum-hydrogen bond. The vibrational spectrum of this bond for the two types of hydrogen chemisorbed onto platinum from the gas phase has been observed in the infra-red by Pliskin and Eischens.16 These authors concluded that their data was accounted for best if Hw and H were adsorbed in either of the two ways i' HW HW Pt Pt Pt or H I Hs.I / -.*.. I / ***.* i" I" * P t - H * P t . Either of these schemes would fit well with the mechanistic scheme described above and also with the special optical behaviour of the layer of H for which AR/R changes sign with wavelength fig. 9. Calculations using the three-layer model show the limiting combinations for the real and imaginary components of the refractive index n and k needed for the adsorbed layer to cause an increase in R. Table 1 shows some 1 24 MODULATED SPECULAR REFLECTANCE SPECTROSCOPY typical values for 400 nm. In general k needs to be quite large and the larger n is the larger k must be ; the thickness of the layer affects the magnitude of the reflectance change but not its sign this being determined solely by n and k.These are the characteristics of an absorbing layer with metal-like properties. The proposed TABLE 1.--CALCULATED VALUES OF ARIR FOR A THREE-LAYER MODEL SHOWING THE EFFECT FOR PLATINUM AT 2.22 - 4.05 i. OF CHANGING THE OPTICAL CONSTANTS OF THE ADSORBED LAYER n2 AND kz KEEPING THOSE WAVELENGTH 400 nm ANGLE OF INCIDENCE 45" PARALLEL POLARIZED LIGHT LAYER THICKNESS 0.1 nn. n2 kt ARIRX 103 1 .o 0 + 0.278 2.0 0 + 0.488 3.0 0 + 2.36 4.0 0 + 5.20 1 .o 1 .o + 4.07 2.0 1 .o + 2.78 3.0 1 .o + 4.47 4.0 1 .o + 7.67 1 .o 2.0 - 0.01 1 2.0 2.0 + 2.35 3 .O 2.0 + 5.22 4.0 2.0 f9.11 1 .o 3.0 - 2.33 2.0 3.0 + 1.02 3.0 3.0 + 4.91 4.0 3.0 + 9.58 1 .o 4.0 - 4.82 2.0 4.0 - 0.855 3.0 4.0 + 3.77 4.0 4.0 +9.16 model for the PtH layer in which the hydrogen atoms are in or almost in the surface of the metal could well be consistent with such properties.Clearly it would be useful to make measurements on a palladium electrode which readily absorbs hydrogen to seek further insight on the nature of this layer.* * Y. C. Chiu and M. A. Genshaw J. Phys. Chern. 1968,72,4325 ; 1969,73,3571. * T. Takamura and H. Yoshida Extended Abstracts 17th C.Z.T.C.E. Meeting (Tokyo 1966) p.260. A. H. Reed and E. Yeager Office of Naval Research Report no. 6 Project No. NR 056-451. W. N. Hansen and A. Prostak Phys. Reu. 1968,174,500. H. B. Mark and B. S. Pons A d . Chem. 1966,38,119. T. Kuwana R. K. Darlington and D.W. Leedy Anal. Chem. 1964,36,2023. J. W. Strojek and T. Kuwana J. Electroanal. Chem. Interfacial Electrochem. 1968,16,471. T . Kuwana and 3. W. Strojek Disc. Faraday SOC. 1968 45 134. J. D. E. McIntyre 135th American Electrochemical Society National Meeting New York May 1969 ; Extended Abstracts no. 232. B. J. Holden and F. G. Ullman 135th American Electrochemical Society National Meeting New York May 1969 extended Abstracts no. 42. D. F. A. Koch and D. E. Scaife J. Electrochem. SOC. 1966,113 302. D. C. Walter Can. J. Chem. 1967,45,807; Anal. Chem. 1967,39,896. * It has been pointed out that the monochromator used in this work produces a large intensity of scattered light at short wavelength. In view of this data below about 250 nm are unreliable. A . BEWICK AND A . M. TUXFORD 125 T. Takamura K. Takamura W. Nippe and E. Yeager J . Electrochem. SOC. 1970,117 626. ' €3. 0. Seraphin Phys. Rev. 1965 140A 1716. M. Cardona K. L. Shaklee and F. H. Pollack Phys. Reg. 1967,154,696. J. Feinleib Phys. Rev. Letters 1966 16 1200. A. Prostak and W. N. Hansen Phys. Rev. 1967,160 600. lo B. J. Parsons Phys. Rev. 1969 182 975. l 1 M. Stedman Chern. Phys. Letters 1968 2,457. R. Greef J. Chem. Phys. 1969,51 3148. l 3 A. K. N. Reddy M. A. Genshaw and J. O'M Bockris J . Chem. Phys. 1968,48.671. l4 values quoted by R. Parsons in a private communication. I6 W. A. Pliskin and P. Eischens 2. phys. Chem. N.F. 1960,24,11. W. N. Hansen J. Opt. SOC. Amer. 1968,58 380.

ISSN:0430-0696    

DOI:10.1039/SF9700400114

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

GENERAL DISCUSSION Prof. B. E. Conway (University of Ottawa) said With reference to the reflectance experiments of McIntyre and Kolb in the H region on platinum it is unclear why there is a separation of the maxima for anodic-going and cathodic-going modulations. In cyclic voltammetry even up to relatively high sweep rates the two (principal) H peaks (a small third peak is always seen in the anodic cycle in highly purified solution) are almost exactly reversible with no separation (<5 mV) of anodic and cathodic potentials provided iR effects are absent. Would this be the reason for the separation observed in the maxima of the optical signals? Would the separation disappear after iR potential correction ? The relation to the main two H peaks normally observed is also unclear in terms of the potentials quoted.The two main peaks apear at ca 0.10 and 0.27 V EH not 0.10 and 0.18 V as mentioned by McIntyre and Kolb. Dr. James McIntyre (Bell Telephone Laboratories) (commimicated) In reply to Conway the apparent 40mV separation of the maxima in fig. 10 is principally an experimental artifact due to the filter time-constant z of the lock-in amplifier (ca. 5 s) and the rate of potential scanning (4 mV s-l). Owing to the low level of the modulating voltage (20 mV r.m.s.) the signal-to-noise ratio in these experiments was very low and the use of a relatively large value of z was required to avoid amplifier saturation by noise. The curve shown in fig. 10 is in fact the average result of 64 scans measured with a Fabri-Tek Model 1072 signal averager. Uncompensated resistance effects are negligibly small at the low scan rate and modulation frequency (43 Hz) employed.The preliminary modulation spectroscopy experiments reported here were primarily designed to illustrate the utility of this method for studying monolayer surface films. There is now substantial evidence that the 1 M HC104 solutions employed (which were not purified by pre-electrolysis) contained trace amounts of CI- (ca. 2 x M) due to use of ordinary reagent-grade acid. Bagotsky and co-workers have shown that adsorption of C1- on platinum alters the energy distri- bution of coadsorbed H atoms without changing the total amount adsorbed. The peak at 0.26V corresponding to the strongly bound form is diminished while a new peak at 0.16 V grows simultaneously. The peak at 0.1 1 V corresponding to weakly-bound hydrogen is shifted to more negative potentials.The peak positions observed in the present work are in close agreement with those reported by Bagotsky. The peak at 0.26 V is more prominent in the cathodic sweep since C1- is desorbed near 0.OV and does not re-adsorb appreciably until potentials more positive than 0.3 V are reached during the anodic sweep. Dr. J. McIntyre (Bell Telephone Lab. N. J.) (partly communicated) said With regard to the studies of hydrogen adsorption on Pt Mignolet's measurements indicated that both the strongly-adsorbed form (Pt+-H-) and the weakly adsorbed form (Pt-H+) were present on the metal surface in an H2 atmosphere at 20°C; ' V. S. Bagotzky Yu. B. Vassilyev J. Weber and J. N. Pirtskhalava J. Electroannl. Cliem. 1970 27 31. J. C. P. Mignolet J.Chinz. Phys. 1957 54 19. 126 GENERAL DISCUSSION 127 evacuation removed only the latter species. Mignolet’s findings were confirmed by field emission microscopy. Infra-red spectroscopy also indicates the existence of two forms of adsorbed hydrogen on alumina-supported platinum. However resistance measurements on Pt films in both gaseous ambients indicate that a single species is predominant at room temperature. This species designated as an s adatom is envisaged as an H atom which is dissolved in the metal surface and dissociated into a proton and a conduction band electron. The s adatoms decrease the resistance of the metal adsorbent by increasing the number of conduction electrons and have a positive dipole moment which decreases the work function. Will’s electrochemical measurements indicated that the two adsorption states observed for poly-crystalline electrodes are associated with two different crystallo- graphic planes.The optical measurements on the adsorbed H layers in Pt presented in this Symposium clearly indicate the existence of two (or more) adsorption states. How- ever our postulate that the “ inverse electro-reflectance effect ” arises from removal of free electrons from the conduction band is inconsistent with the experimental observation 9 that H adsorption decreases the substrate resistance. Whether the optical effects are due to a charge-transfer absorption of the species Pt-Had or to a perturbation of the optical properties of the metal surface layer caused by dissolution of H atoms is not yet clear. Careful measurement of the dielectric constant of the surface layer should help to resolve this problem.An apparent inconsistency exists between the results of our electrochemical modulation spectroscopy (EMS) studies on the Had layer and those of Bewick and Tuxford (BT). Our results at 300nm (cf. fig. 10 of ref. (8)) showed that small anodic voltage pulses (20 mV rms) produce a reflectivity increase in both the strongly and weakly bound Had regions. A subsequent repetition of this experiment at 435 nm yielded the same result ; the only significant change was an attenuation of the amplitude of the signal. These results are in fact in close agreement with those shown by BT in their fig. 8. Their definition of AR/R is opposite in sign to ours.) The sign reversal occurs at the onset of the double-layer region where the “ordinary ” electro-reflectance effect becomes operative.The results shown in fig. 8 of the paper by BT appear not to be consistent with those shown in their fig. 9 where a sign reversal occurs as a function of wavelength in the Had region. Our EMS studies have not revealed such an effect. The use of signal-averaging techniques combined with a potential programme designed to remove any spurious effects caused by adsorption or desorption of solution impurities is required to reveal further details of the hydrogen adsorption mechanism. Concerning the oxide film on platinum Kolb’s reflection spectroscopy measure- ments at an angle of incidence of 70” in 1 M HClO yield a different set of optical constants for the PtO layer formed at EH = 1.5 V than those calculated from our measurements at 45”.This lends further support to our postulate that the film is and aqueous 4* W. J. M. Rootsaert L. L. van Reijen and W. M. H. Sachtler J . Catalysis 1962 1 416. W. A. Pliskin and R. P. Eischens 2. phys. Chem. N.F. 1960,24 11. R. Suhrmann G. Wedler and H. Gentsch 2. phys. Chem. N.F. 1958,17 350. H. Shimizu Electrochim. Acta 1968 13,27. T. Dickenson private communication. J. Horiuti and T. Toya Solid State Surface Science ed. M. Green (Marcel Dekker New York 1969) vol. 1 p. 1. F. G. Will J. Electrochem. SOC. 1965 112 459. J. D. E. McIntyre and D. M. Kolb Disc. Faraday Soc. 1970 J. D. E. McIntyre Adva. Electrochem. Electrochem. Eng. ed. R. H. Muller (Academic Press New York) vol. 9 to be published. 128 GENERAL DISCUSSION anisotropic. At 41 = 70" and A = 300 nm (4.13 eV) the ratio of the mean-square field-strength of the components normal (2) and parallel (x) to the surface for p- polarized radiation at z = 0 is ( E ~ ) / ( E ~ ) = 2.6 whereas at 45" ( E ~ ) / ( E ~ ) = 2.0.The relative interaction of the radiation field with the z-component of the matrix element of the transition dipole moment is thus greater at 70" than at 45". For comparison with the results shown in fig. 5 of the paper by Barrett and Parsons the optical constants of the film are given in the table below for four selected wavelengths. The values for = 70" lie just to the left of the shaded region of their figure. This can be considered good agreement in view of the somewhat different values for n3 k3 and d used in the calculations. Optical Constants of the 0 Layer on Pt (1M HC104 EH = 1.5 V) (nm) (eV) n2 k2 n2 k2 250 4.96 1.70 0.90 1.53 1.60 290 4.28 1.88 1.00 1.78 1.52 350 3.54 2.12 1.07 2.00 1.44 400 3.10 2.25 0.97 2.23 1.16 I ha 41 = 45' 41 = 70' With regard to the use of multiple-reflection cells for specular reflection spectro- scopy the experimental findings of Barrett and Parsons are in accord with theory.For a shot-noise limited detection system the optimum signal-to-noise (S/N) ratio for thin-film detection is obtained when the light intensity is reduced after n reflections to l/e2 = 0.135 of its incident value. For a photon detection system with a constant noise level the S/N ratio is a maximum when R" = l/e = 0.365. Thus for a metal electrode with R = 0.5 a system with two or three reflections yields the maximum sensitivity. For many applications use of an optical cell with a single reflection is most advantageous.Miss M. A. Barrett and Dr. R. Parsons (University of Bristol) (communicated) The derivative of our stationary (reflectivity potential) curves can be taken as approxi- mately equivalent to the differential reflectance curves as fig. 10 of McIntyre and Kolb if the changes are considered instantaneous and reversible. This appears to be approximately so in the hydrogen region provided impurities are absent in the electrolyte. It is then possible to make some comparison by inspection. Drawing on several results not illustrated in our paper reversal of the sign of the differential R would then occur between weakly- and strongly-bound hydrogen under certain conditions becoming negative with increasing wavelength or decreasing angle of incidence.At 45" this appears between 250 and 350 nm while at 60" between 350 and 550 nm. The sign across the double-layer region would also vary according to angle of incidence being positive above 60" and negative below with no wavelength dependence observed. The available data from HC104 showed no difference from HzSOS apart from evidence of slight impurity of C1- to the extent of M or less not sufficient to produce kinetic effects in the hydrogen region. With regard to optical constants of oxide on Pt agreement between those of McIntyre and Kolb and the areas compatible with our optical observations are indeed much better if their values for npt are adopted. Prof. B. E. Conway (University of Ottawa) said In relation to Bewick's comments about the state of H at Pt electrode interfaces a surface-interstitial state for H atoms GENERAL DISCUSSION 129 is reasonable in relation to Mignolet’s observations of negative surface potentials for strongly bound chemisorbed H on Ni and Pt.This was explained in terms of the H being held interstitially within the surface as a particle bearing the expected partial positive charge but with compensating electron charge distributed nearer the “ inter- face ”. If H were adsorbed on Pt atoms with an expected charge distribution such as Ha++Pts- an opposite sign of x would be found. Donation of electrons to the metal d-band is expected on the basis of observed changes of paramagnetic suscepti- bility e.g. for Ni and Pd-H. It is of interest to note that two types of adsorption site will exist on an f.c.c.close-packed surface ; one at trigonal holes where another Pt atom is immediately below and the other at trigonal holes where a Pt atom is not immediately beneath. This will presumably affect the state of H atoms sorbed interstitially within the Pt surface where they will attain maximum multiplicity of interaction with the metal atoms. Dr. R. Parsons (University of Bristol) said Following Conway’s remarks I would mention the evidence presented by Frumkin and Petrii at the CITCE meeting in Detroit in 1968. They were able to analyze the potential drop across the platinum/ sulphuric acid interface into the component due to the ionic double layer and that due to the adsorbed atomic hydrogen. The latter changes sign at approximately +0.18 V. At potentials more positive than this corresponding to the strongly adsorbed hydrogen these results indicate that the Pt-H dipole has its negative end towards the solution.The weakly bound hydrogen on the other hand has the positive end towards the solution. These results agree well with the measurements made at the platinum/gaseous hydrogen interface by Mignolet. On the problem of the oxide film on platinum Miss Barrett has recently made some reflectance measurements during potential sweeps. The wavelength dependence of these clearly indicates a different behaviour between the film below 1.1 V and above this potential. At the higher potentials the slope of the reflectance potential line is more strongly wavelength dependent than that at lower potentials. There is also a marked change in properties at higher potentials.Biegler and Woods showed by coulometric measurements that the relation between amount of oxygen adsorbed and the potential has an inflection at 1.5-1.6 V and reaches a limiting value from about 2.3 V. The limiting charge was originally given as 2.66 times twice the monolayer charge of hydrogen but more recently they have given arguments suggesting that this ratio is closer to 2.0 indicating two oxygen atoms per surface platinum atom. During the last year Dr. W. H. M. Visscher has been studying this region ellipso- metrically. As in the coulometric measurements it is necessary to form the oxide layer at the required high potential and then to lower the potential and remove oxygen bubbles. In this work the measurements were made at 1.5 V but other potentials in this region gave similar results.The value of A found in this way follows a similar potential dependence to the charge passed in reducing the layer. There is an inflection at about 1.5 V and a limiting value is reached at high potentials. Thus the optical measurements essentially confirm the observations of Biegler and Woods. Further experiments were carried out by Dr. Visscher on platinum using acetate buffer solutions. Again the observations of Woods of a limiting value were con- J. C. P. Mignolet Disc. Faraday Soc. 1960 8 105; J. Chim. Phys. 1957 54,19. Mignolet J. Chim. Phys. 1957 54 19. Biegler and Woods J. Electroanal. Chem. 1969 20 73. Biegler Rand and Woods J. Electroanal. Chem. 1971,29,269. Woods J. Electroanal. Chem. 1969,21,457. s4-5 130 GENERAL DISCUSSION firmed. Woods's measurements were confined to the higher potentials.We observe an inflection in the (A potential) plot at about 1.5 V as in sulphuric acid and this has been confirmed by coulometric measurements. Plots of A against + are non-linear for the acetate buffer while they are linear for the sulphuric acid solutions. This suggests that there is a major difference in composition. It is possible to regard the acetate data as comprising t~7o linear regions and to analyze these in terms of a two- layer adsorbed film. These results will be reported in more detail in the near future. Dr. A. Bewick (University of Southampton) said Extreme sensitivity can be obtained in the application of optical methods to electrochemistry by modulating the electrode potential and employing phase-sensitive detection of the optical signal.I would emphasize that there are dangers inherent in the use of the phase-sensitive detection technique. The direct time information in the optical signal is integrated out and can be recovered only in an indirect manner by varying the modulation frequency over a wide range. This is relatively unimportant for kinetically simple situations but it imposes severe limitations on the interpretation of data from more complex systems and will often prevent an unambiguous mechanistic assignment being made. It is preferable to retain the direct time information on the optical signal and to obtain the necessary sensitivity by using some other correlation tech- nique such as signal averaging. It is particularly appropriate in the electrochemical situation to explore the use of other cross-correlation techniques to gain sensitivity and mechanistic insight simultaneously e.g.to generate the cross-correlation function of the optical-time signal with either the current-time or coulomb-time signal from the same system. Dr. R. Parsons (University of Bristol) said The results for oxide formation shown in fig. 3 of the paper by Bewick et al. are possibly misleading. The fixed bias potential of +670 mV seems high so that the oxide will be incompletcly removed from the electrode surface. With a cycle frequency of 30 Hz I believe that a true removal- deposition cycle would not be achieved for such an irreversible process. It is also important to call attention to the risks involved in the use of perchloric acid in this type of experiment. Vasina and Petrii have shown that perchlorate ion adsorbed on platinum is reduced to chloride in the range 0.20-0.55 V.This could be particu- larly important in reflectivity measurements because of their sensitivity to chloride. Dr. A. Bewick and Mr. A. M. Tuxford (University of Southampton) (partly com- municated) We agree that the form of the (reflectivity potential) curve is influenced greatly by the choice of the cathodic bias potential. To retain clarity of presentation the data plotted in fig. 3 were restricted to that for a single bias potential. Our results for other bias potentials are now presented and they show clearly the expected effects due to incomplete removal of the oxide layer during the cathodic period. With regard to the reduction of perchlorate to chloride Vasina and Petrii conclude that the reaction is very slow on platinum (a period of 30 min being required to produce a detectable amount of chloride ion).We believe that our work on the perchloric acid system does not show any effects from such a reduction process. Prof. MI. Fleischmann (University of Southampton) said The papers presented at this Symposium have concentrated on optical studies in the visible/ultra-violet region of the spectrum and such measurements will undoubtedly give valuable information on the nature of the electronic surface states. While this should be a major objective ' Vasina and Petrii Elektuokliiin 1970 6 242. GENERAL DISCUSSION 131 at least for electrochemists the observation of the vibrational frequencies of adsorbed species could well prove equally valuable.In recent years there have been many reports of such measurements by infra-red spectroscopy including a number on relatively simple systems such as of hydrogen on platinum and nicke1.l There has also been a number of reports of related measurements on polar non-metallic surfaces by laser-Raman spectroscopy.2 amplitude of anodic pulse (mV) FIG. l.-AR/R for platinum oxide at 425 nm as a function of the anodic pulse amplitude at 30 Hz for different bias potentials 1 +470 mV ; 2 +570 mV ; 3 +670 mV ; 4 +770 mV. In the course of a recent study of the electrochemical H/D ~eparation,~ where direct measurement of the vibrational frequencies would be particularly relevant we have attempted to determine this frequency for hydrogen adsorbed on cathodically polarized palladium electrodes using a Spex Laser-Raman spectrometer with an argon ion laser.A broad peak at around 1580 cm-' was observed and this peak was absent when the adsorbed and dissolved hydrogen were removed anodically. The conventional method of calculating this frequency using the Morse-Mulliken equation indicates that this line would be expected to occur at around 1652 cm-l in reasonable agreement with the observed value. The observation of this weak band may be thought to be surprising since the adsorbed hydrogen would not be expected to scatter strongly. It is relevant however that electrochemical measure- ments indicate that Pt-H bonds are strongly polarizable and that the adsorbed species vibrates in a region of very high field strengths. These observations demand further verification and my purpose in presenting them here is to enquire whether anyone else is making similar or related measurements ? Prof.E. Yeager (Cleveland) said In reply to Fleischmann the vibrational spectra of adsorbed species on electrode surfaces should indeed yield interesting information R. P. Eischens and A. Pliskin 2. phys. Chem. N.F. 1960 24 11. A. Kant J. Chem. Phys. 1964 41 1872. For reviews of infra-red spectroscopy of adsorbed layers see L. H. Little Infra-Red Spectra of Adsorbed Species (Academic Press New York 1966). M. L. Hare Infra-Red Spectroscopy in Surface Chemistry (Marcel Dekker New York 1967). P. J. Hendra and E. J. Loader Nature 1967 216 789 ; 1968 217 637. P. J. Hendra J. R. Horder and E. J. Loader Chem. Comm. 1970 563. R. Dandapani Ph.D. Thesis (University of Southampton 1969). P. M.Morse Pliys. Rev. 1929 34 57. 132 GENERAL DISCUSSION concerning the nature of the interactions of the adsorbed species with the surface and the structure of the adsorbed layer. I believe that infra-red techniques offer considerable promise for obtaining such data perhaps more promising than Raman techniques even using laser sources. An attempt was made several years ago in my laboratory by Dr. A. H. Reed to examine the infra-red spectra (particularly the C-H stretch) of adsorbed organic molecules at a germanium/electrolyte interface. Internal reflection spectroscopy was used with approximately 100 reflections within a trapezoidal germanium plate [see A. H. Reed and E. Yeager Electrochim. Acta 1970 15 13451. D20 was used as the solvent to reduce interference from solvent infra-red absorption at wave- lengths in the vicinity of the C-H stretch (-3000 cm-l).The C-H stretch was detected at ca. 2920 cm-I in stearic acid and also the surfactant Triton X114 (Rohm and Haas Co.) with the formula CH3 CH3 CH3-C-C-C ' ' -~-(OCH,CH,),OH I H I CH CH3 but the sensitivity was not sufficient for quantitative studies with low-molecular- weight organic molecules because of signal-to-noise problems with the apparatus employed in this work (Perkin-Elmer infra-red spectrophotometer model 621 with scale expander with the light beam modulated by a rotating optical chopper and the electrode maintained at a fixed potential). With more sophisticated signal processing it should be a relatively simple matter to extend the sensitivity of the method to the point that low-molecular-weight adsorbed organic species can be studied quantita- tively.Fourier transform spectroscopy may prove promising for this purpose. The use of a germanium electrode in such studies may be attractive because of its relatively good infra-red transmission properties but presents substantial problems in that any potential modulation or change between the electrode and solution phases appears mostly across the space charge region within the germanium. For this reason a very thin metal film or an infra-red transparent substrate appears to be a more attractive approach. Adequate conductivity can be achieved with metal films sufficiently thin to be infra-red transparent through the use of a vacuum-deposited or sputtered metal grid structure to provide conductivity with the thin film of metal in the open spaces of the grid structures on the infra-red transparent substrate.We plan to resume work on the infra-red spectra of species adsorbed on electrode surfaces using such techniques in the near future. Mr. A. M. Tuxford (University of Southampton) said Modulated specular reflectance spectroscopy is a powerful tool for the detection and identification of intermediates and for measuring their lifetimes. Two systems we have studied indicate the scope of the method. The oxidation of Ag' to Ag" on platinum in nitric acid and the coupled reaction of the Ag" with benzene were studied using a range of modulation frequencies from 3Hz to 15 kHz. The spectrum of Ag" was readily observed and the amplitude of its absorption peak as a function of modulation frequency indicated two relaxation times for the coupled system and a single relaxa- tion time in the absence of the organic substrate.From these rate constants can be calculated in the usual way. In the reduction of C 0 2 in aqueous solution it is not possible using the fastest electrochemical methods to detect directly the intermediate anion radical. Using reflectance spectroscopy at a lead cathode the spectrum of this very short-lived species was readily obtained fig. 1 for a concentration of about GENERAL DISCUSSION 133 - 60 n - 4 0 2 X v n 0 - 2 0 wavelength (nm) FIG. 1 . 4 ~ ) Spectral dependence of AR/R for the formation of the C02-radical ion in 10-1 M tetramethylammonium chloride at a lead cathode. 30 Hz square wave from - 1.2 to - 1.8 V against the silver-silver chloride reference electrode.M sodium formate solution saturated with COz obtained using pulse radiolysis (dose 7500 rad/2 ps pulse) (J. P. Keene Y. Raef and A. J. Swallow in PuZse Radidysis M. Ebert J. P. Keene A. J. Swallow and J. H. Baxendale ed. (Academic Press London 1965) p. 99.) 10-l2 mol cm-2. Therefore this technique provides a means for the study of very fast processes. (6) Absorption spectra of the C0,-radical ion in Dr. B. Cahan (Case Western Reserve University) said Since both Bewick and Parsons have been discussing the questions of the oxides on platinum and of the wavelength dependence of their reflectivities I present the following with the hope that it may help to clarify some points. The figure represents some reflectivity data obtained in HC104 on a sputtered Pt electrode which was contaminated during the sputtering process possibly by the polymerization of traces of pump-oil by the sputtering plasma.Attempts to identify the contaminant have been unsuccessful. No extraneous elements have been detected by emission spectroscopy or by X-ray fluorescence analysis and ATR studies of the surface have shown no i.-r. absorption bands due to organics. Whatever the material may be it is present only in very small quantities has no gross effect on the electrochemical behaviour as evidenced by the voltammetric curves and it is removed from the surface by cycling the potential between 0 and 1.5 V for 2-3 h. The principle optical effect of this impurity is to lower the reflectivity of the base Pt but not that of an oxide-covered surface. This has the fortuitous effect of showing up detail in the electroreflectivity curve that has been unobserved on the clean metal.On a bright Pt electrode the reflectivity shows two segments in the oxide region which are not resolvable because both have a high slope. The effect of the contaminant is to resolve these two segments more clearly as shown in the figure. The mechanism for this is not clearly understood but can be explained as follows. Because the reflectivity of the metal is reduced but not that of the oxide this has the effect of flattening the curve in the region >0.7 V. Since the wavelength dependence of the reflectivity of the Pt is greater in the presence of this contaminant the two segments 134 GENERAL DISCUSSION show up strikingly at different wavelengths a change of sign of slope occurring at 500 nm.We believe that this is strong evidence for the existence of two distinct phases in the growth of oxide on the platinum surface. The first phase occurs at about 0.75 V in t h s electrolyte with a transition to the second phase at 0.9 to 1.0 V. zc! I 0 0.4 0 . 8 1.2 V(NHE) FIG. 1 . A detailed quantitative measurement of the slope of the reflectivity of the oxide on clean platinum should show much the same result. This work was supported by the Office of Naval Research. Prof. Dr. K. J. Vetter (Free University of Berlin Berlin-Dahlern) said I would draw attention to the structure of the oxygen layer on gold and on platinum which follows from recent investigations of the kinetics of the anodic formation and cathodic reduction of these layers.’. We have to distinguish between a “chemisorption layer ” of oxygen ions on the nearly undisturbed metal surface and an “ oxide layer ” of one or two monolayer thicknesses in which a displacement of metal ions and oxygen ions has taken place.Only this assumption could explain the complicated ’ J. W. Schultze and K. J. Vetter Ber. Bunsenges. phys. Chem. 1971,75 in press (gold). K. J. Vetter and J. W. Schultze in preparation (platinum). GENERAL DISCUSSION 135 relations between the anodic and cathodic current densities the potential the degree of coverage the pH-value and a complex time effect. The existence of a chemisorp- tion layer in addition to an oxide layer is also in agreement with the double-layer capacity and its dependencies on the potential and frequency. Prof. B. E. Conway (University of Ottawa) said I would refer to work Dr.Laliberte and I have recently done in applying optical methods to the study of the surface state of Pt anodes at which the Kolbe reaction with acetate is occurring (2CH3COO-+ C2H6 + 2C02 + 2e). Electrochemical work on this reaction has been described previously and reviewed.2 In initial work on Dr. Ord's (University of Waterloo Canada) automatic instrument we observed a marked inflection of the relation .I .2 1 2.04 1.64 1.24 .84 .44 . 0 4 1 -.05 0 0.2 - 0.4 - 0.6 - 7- 2 - 1.0 ' 0.8 - Kl - I. 2 - 1.4 potential (N.H.E.) FIG. 1.-Changes of A and J for oxidation of a Pt electrode surface at which the Kolbe reaction is proceeding. + determined in steady state measurements ; 6A in ellipsometric transient measure- ments by means of cyclic voltammetry (2 M CH,COONa+ 1 M CH,COOH 25OC).between A and t,b as the potential at Pt was driven into the region where the Kolbe reaction proceeds efficiently (> 2.2 V EH). We have now extended this work on our own instrument by the method of off-null intensity transients using the principles discussed by Cahan Brusic and Gen~haw.~ Fig. 1 shows the variation of A and tj with potential in anodic and cathodic transients up to the Kolbe region. Surface layer formation continues beyond the normal range of potentials accessible since O2 B. E. Conway and A. K. Vijh J. Phys. Chem. 1967,71 3637 ; 1967,71 1967 3655. A. K. Vijh and B. E. Conway Chem. Rev. 1967,67,623. V. Brusic M. A. Genshaw and B. D. Cahan AppI. Optics 1970 9 1634. 136 GENERAL DISCUSSION evolution is inlubited.' At ca. 1.6 V a second inflection in the (A potential) relation arises (the first arises at ca.1.1 V) and this may be due to (a) an increasing degree of surface oxidation of Pt (greater O/Pt ratio) and (b) the formation of the previously- proposed dual surface layer composed (at these potentials) of phase-oxide species and co-adsorbed carboxylate and/or methyl radicals. At the high anodic potentials involved the ratio of anodic charge Qa to QH is larger than it is in the normal oxide region in the absence of CH3COO- up to potentials of 1.5 V. The optical studies also confirm that the Kolbe reaction (in aqueous medium) proceeds on the oxidized Pt surface although this is not a requirement for the reaction to occur since Kolbe products are also efficiently generated in anhydrous media.2 The changes of $ suggest significant electro-chemisorption of organic intermediates at the Pt electrode in addition to the presence of surface oxide when the Kolbe reaction is proceeding. A. K. Vijh and B. E. Conway Chem. Rev. 1967,67,623. B. E. Conway and A. K. Vijh 2. anal. Chem. 1967,224,149.

ISSN:0430-0696    

DOI:10.1039/SF9700400126

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Studies of Adsorbed States at the Electrode/ElectroIyte Interface by Specular Reflection of Light BY W. J. PLIETH lnstitut fur Physikalische Chemie der Freien Universitat Berlin Berlin-Dahlem Received 26th October 1970 The reflection of light from an electrode surface was investigated using a multiple-reflection and a single-reflection cell. The single-reflection studies were combined with a modulation of the reflec- tivity of the interface. The potential drop across the electrolytic double layer was modulated. The electro-reflection technique enables one to investigate the adsorbed layer without interference by light absorption in the electrolyte. Results are reported for the adsorption of methylene blue on a platinum electrode. The properties of the electrode/electrolyte phase boundary are connected with its light reflectivity.Some information about the state of the intedace can be obtained by measuring the parameters of the reflected light. The intensity of the light reflected from gold and platinum electrodes and both specular and diffuse reflectivity were investigated. In this paper the results of the specular reflection studies are reported. REFLECTION OF LIGHT FROM AN ELECTRODE The reflection of light from a surface can be characterized by the reflectivity R. The intensity of the incident light is denoted by I and the intensity of the reflected light by I,.. Then the reflectivity R is given by R = rr/ro (1) where R is a function of the wavelength. For an electrode R is also a function of the electrode potential. Adsorption is an important factor in influencing the state of an electrode surface.The effect of adsorption on the reflectivity of an electrode can be divided into two parts. Adsolbed molecules can change the electron states of the electrode surface and the additional amount of the adsorbed substance in the adsorption layer can lead to an increased absorption of the light equal to decreased reflectivity of the surface. Therefore we can distinguish between the reflection of light from the electrode metal atoms RM and the transmittance of the adsorption layer Tad. Then the overall reflectivity is given by R = RMTad. (2) The three quantities are functions of the wavelength ;1 of the incident light of the electrode potential and of the surface coverage 8. In addition to the effects on the interface the measurable intensity of the specularly reflected light is dependent on its absorption in the electrolyte phase.This effect is dependent on the thickness of the electrolyte on the concentration of the absorbing substance and on the wavelength of the incident light. It should be possible to eliminate the absorption in the electrolyte by an extrapolation procedure. For very 137 138 thin electrolyte layers and for small concentrations of an absorbing substance the absorption of light in the electrolyte becomes negligible compared with the decrease in the light intensity during reflection from the electrode surface. The effect of the electrolyte is also negligible in a wavelength range where the electrolyte is non- absorbing. ADSORBED STATES AT ELECTRODE/ELECTROLYTE INTERFACE MULTIPLE SPECULAR REFLECTION For very thin electrolyte layers and for small concentrations of an absorbing substance the overall decrease in the light intensity during a reflection is so small that detection of changes in the intensity becomes difficult.This has led to the development of multiple reflection cells. Descriptions of different multiple reflection cells have been given by Francis and Ellison1 and by Leja Little and Poling.2 A special electrochemical arrangement has been described by Takamura Nippe and Yeager.3 The multiple reflection cell described in this paper was developed in 1968/69 and is described in more detail el~ewhere.~ The principle of the arrangement is shown in fig. 1 and 2. FIG. 1 .-Diagram of the multiple specular reflection cell. Glass plates GP with segments of vacuum-deposited gold films GF were used as working electrodes and as windows for the entrance EN and exit EX of the light beam.Two glass plates resting on a sealing ring SR in two Teflon blocks were separated by a Teflon mask TM. The three parts were pressed together in a metal box. The distance between the two plates was determined by the Teflon mask. The mask also created a channel for the electrolyte E. The gold segments on the glass plates overlapped over a distance of 10 mm. This area was used for the multiple reflection of the incident light. Platinum foils on the Teflon mask carried the current to the working electrodes. The counter electrode and the reference electrode could be placed outside the cell or in the chambers in front of the thin layer arrangement. The angle of the incident light between the light beam in the electrolyte and the electrode was chosen to be 45".The length of eIectrode overlapping remained W. J . PLIETH 139 constant at 10 mm. Only the distance between the two working electrodes affected the number of the reflections; 10-20 reflections could be detected. These numbers of reflection were obtained for distances of 1.0 and 0.5 mm. The distance travelled by the light through the solution was 15 or 30 mm. The absorption by this solution must be negligible or comparable with the decrease of light intensity due to the reflections. The multiple reflection cell was used to measure the change in reflectivity of gold film electrodes following an adsorption. Two identical cells were used. One cell contained an electrolyte absorbing in an appropriate wavelength range and the other cell contained a non-absorbing solution.The two cells could be brought consecu- tively into the light beam of a Zeiss-PMQ-I1 spectrometer. Both cells were kept at the same electrode potential. In this manner the change in reflectivity caused by adsorption of permanganate on the gold electrodes in a wavelength range of 520- 620 nm was investigated. A 0.1 mol l.-l Na,SO solution and permanganate EN I f M EX FIG. 2.-Perspective illustration of the multiple specular reflection cell. concentrations of 1 x to 1 x mol I.-' were used. The electrolyte was kept at a temperature of 25°C and was degassed with nitrogen. The electrolyte streamed continuously through the cell. The potential during the measurements was kept at a value of EH = 875 mV. A shift of the maxima of the spectrum of a solution of permanganate to longer wavelength by about 17 nm was observed.It was concluded that the permanganate was adsorbed on the gold surface. A MODULATION TECHNIQUE FOR SINGLE SPECULAR REFLECTION STUDIES The multiple specular reflection technique has several disadvantages. A method was developed in order to study a single reflection. This method is based on a modulation technique. Several parameters can be modulated to increase the sensitivity to changes in the intensity of the specular reflected light. A modulation 140 ADSORBED STATES AT ELECTRODE/ELECTROLYTE INTERFACE of the amplitude and of the wavelength are possible. The modulation of the reflec- tivity is known from semiconductor investigati~ns.~* This technique is also an ideal method for metal electrodes.In the usual stationary techniqu.e the intensity of the reflected light is given by rr = I,R. Differentiation leads to dl = IodR. (3) (4) One can divide this equation by equation (3) to give dIr/Ir = dRJR. ( 5 ) The measurable quantity a,./& is identical with the required quantity dR/R and no longer includes the value I. of the incident light. The experimental work then becomes much easier. The thickness of the electrolyte and the concentration of a coloured substance can be chosen arbitrarily. The cell construction is no longer restricted to thin-layer cells and can be considerably simplified. Different possibilities exist for the modulation of the reflecti~ity.~~ For an electrodejelectrolyte phase boundary the most important method is the modulation of the potential difference across the double layer (electro-reflection).In semi- conductor investigations the double layer consists of two parts one part in the semiconductor and the other in the electrolyte. The main interest in semiconductor investigations was directed towards the solid state phase. Because of this only the effect of the modulation in the semiconductor part of the double layer was considered. The electrolytic part of the double layer was assumed to be unaffected by the modula- tion. On metal electrodes the double layer is only established in the electrolyte and only the potential drop across the electrolytic double layer is modulated. There- fore when metal electrodes are used specific investigations can be carried out in the double layer. The results are unaffected by light absorption in the electrolyte and are unambiguous in the local coordination.EXPERIMENTAL The arrangement for the performance of modulation experiments is given in fig. 3. A monochromatic light beam (Bausch and Lomb grating monochromator MC) was focused by a lens on the platinum working electrode WE of an electrochemical cell containing also the reference electrode RF and a counter electrode CE. The reflected light was passed through a second lens and detected in a photomultiplier MP (EM1 9526 B). The potential of the working electrode could be controlled by a potentiostat P (Jaissle 300 R). The control FIG. 3.-Block diagram of the modulation spectroscopy arrangement. W. J . PLIETH 141 potential CP (potential sweep Wenking potentiometer SMP 66) was modulated. A sine waveform with a 50 mV amplitude was used for the modulation (sine waveform generator SG Siemens Re1 3W 38c).A modulation frequency of 32 Hz was used for most of the investigations. A modulation of the reflected light intensity due to the modulation of the electrode potential could be detected by the arrangement shown on the right-hand side of fig. 3. The output of the multiplier MP is connected to the input of a lock-in amplifier L (Philips PM 7835/PM 6045). The amplifier compares the modulated multiplier signal with the original modulation signal. The rectified output signal of the lock-in amplifier is then divided by an electronic divider D (Burr-Brown 4029/25) through the dnvoltage of the multiplier. The final result dZ/Z in arbitrary units (output voltage of D) is recorded on an X Y recorder R (Moseley 7000 A) against the electrode potential.Reflectivity changes of 0.01 % were detectable with this electronic arrangement. An increase in sensitivity by about lo2 to lo3 seems possible. The working electrode WE was a platinum plate of 4cm2 cemented to PVC. The electrode was polished with diamond paste and was cleaned in an ultrasonic field before it was used. A 1 N calomel electrode served as a reference electrode RE (EH = +280 mV) and the counter electrode CE was a platinum net electrode. The electrolytic cell was constructed from PVC. Nitrogen was bubbled through the electrolyte for about 30 min. Then the electrolyte at 25"C was streamed continuously through the cell. Several sub- stances were investigated. A 0.5 moll.-l Na2S04+0.01 mol l.-l H2S04 solution was the basic electrolyte.Permanganate Neutral Red (toluylene red) and Methylene Blue were added to the basic electrolyte. The concentrations of the added substances were 1 x mo1L-l ; 0.02 moll.-l acetate buffer was used in preference to H2S04 for the organic dyes. All substances were p.a.-products. The result was independent of the scanning rate of the potential. RESULTS In fig. 4 the results are given for the basic electrolyte. The current i and the relative reflectivity d I / Z = dR/R were plotted against the electrode potential. No modulation in the reflectivity could be detected for the wavelength 400 500 and 600 nm in the potential range covered. The same results were obtained for Neutral Red and for permanganate. The results are different when Methylene Blue is added to the basic electrolyte (fig.5). A modulation of dl is observed for a wavelength of 500 nm in the potential +I*O z o *- - 1-0 - > +200 E * o ;2- Y a -200a - 5 0 0 0 +500 + 1000 G a l h V 1 FIG. 4.-Shows the dependence of the current i and of the relative reflectivity dZ/l= dR/R of a platinum electrode on the potential ; electrode surface 4 cm2 ; 0.5 rnoll.-l Na2S04+0.01 moll.-' HzSO4-solution ; izi scanning rate 10 mV s-I. 142 region of Ecal = - 150 mV and Ecai = + lo00 mV. The first modulation peak was independent of the scanning rates. The second modulation peak disappeared when a slow scanning rate was used and for a negative scanning direction. ADSORBED STATES AT ELECTRODE/ELECTROLYTE INTERFACE + 1.0 7 0 E Y .C) - 1.0 0 > -1.0 +I*O + o 'c1 - 1.0 - \ - 5 0 0 0 + 5 0 0 + 1000 Ecal b V 1 FIG.5.-Shows the dependence of the current i and the relative reflectivity dZ/Z = dRiR of a platinum electrode on the potential ; electrode surface 4 cm2 ; 0.5 moll.-' Na2S04+0.02 moll.-' acetate buffer+0.0001 mol I.-' Methylene Blue ; scanning rates 10 mV s-' (upper curve) and 1 mV s-' (lower curve). The wavelength dependence of the first peak was investigated. The result is shown in fig. 6. The occurrence of a modulation dZ is restricted to a small wave- length region around 550-570nm. A slight dependence of the maxima on the electrode surface was observed. A [nml FIG. 6.-The dependence of the relative reflectivity dl/Z = dR/R on the wavelength ; Ecal = - 150 mV. The modulation d1 is also effected by the potential modulation frequency. The measurements were carried out for 32 Hz 1 and 10 kHz.No modulation dl was observed for 10 kHz. For 1 kHz only a slight effect was observed (fig. 7). No electrochemical effects due to the presence of Methylene Blue were observed in the (i,E) diagram. The equilibrium potential of Methylene Blue/leuco-Methylene Blue in the acetate-buffered solution is equal to Ecal = - 122 mV. Reduction of Methylene Blue seems not to occur below the potential of hydrogen evolution. W. J . PLIETH 143 1 (nm) FIG. 7.-The dependence of the relative reflectivity d l / l = dR/R 011 the modulation frequency ; Ecal = - 150 mV ; modulation frequencies 32 Hz and 1 kHz. DISCUSSION The observed behaviour can be interpreted as changes in the reflectivity of the surface due to adsorption/desorption of the Methylene Blue. In this interpretation the modulation in potential is followed by a modulation in the surface coverage of Methylene Blue.The correlation between potential and surface coverage can only occur within a potential range of weak adsorption. These are the boundary potential zones in the potential region of adsorption. The Methylene Blue is adsorbed on the electrode surface at a potential Ecal x - 150 mV and remains adsorbed until a potential Ecal x + 1000 mV is reached. This is the potential range in which after cathodic polarization neither oxide nor H-atoms are adsorbed on the electrode surface. In accordance with this interpreta- tion light intensity modulation was not observed for Ecal z 1000 mV when anodic polarization and forming of an oxide layer proceeded. At Ecal w - 150 mV the oxide layer is completely reduced and the light intensity modulation can again be observed.The hysteresis in the forward and backward behaviour of the light intensity modulation at the potential Ecal w - 150 mV can be explained in terms of hydrogen coverage. In an anodic potential sweep direction the hydrogen atoms on the electrode delay the Methylene Blue adsorption. The observed dependence of the reflectivity modulation on the wavelength is in this interpretation the difference in the absorption spectrum of adsorbed and non- adsorbed molecules. In a solution with a Methylene Blue concentration of moll.-l the Methylene Blue molecules are present in a bimolecular form.7 The maximum in the inodulation/wavelengths dependence between 500-600 nm could be the maximum difference between the bimolecular and the adsorbed Methylene Blue molecules.Different homogeneous and heterogeneous molecular forms can also be involved in the adsorption and desorption processes. The absence of maxima for wavelengths greater than 600nm could be due to a considerable decrease in sensitivity of the multiplier for wavelengths greater than 600 nm. Limits for the adsorption rate can be estimated from the measurements. The modulation in the light intensity disappeared for a modulation frequency of 1 kHz. A complete desorption and adsorption of a monolayer (I' = mol cm-2) for one cycle of this frequency gives dI'/dt = mol cm-2 s-'. This is an upper limit for the value dr/dt = kadsc = lo-' mol cm-2 s-' (for moll.-l Methylene Blue). Therefore kad < 10 s-l cm. A lower limit of the rate constant of adsorption requires 144 an estimation of the smallest detectable surface concentration. A value of 0.001 r gives kad >0.32 x s-l cm. A greater number of modulation frequencies are necessary for a precise determination of kad. ADSORBED STATES AT ELECTRODE~ELECTROLYTE INTERFACE S. A. Francis and A. H. Ellison J . Opt. SOC. Amer. 1959 49 131. J. Leja L. H. Little and G. W. Poling Trans. Inst. Min. Met. 1963,72,414. T. Takamura K. Takamura W. Nippe and E. Yeager J. Electrochem. Soc. 1970,117,626. W. J. Plieth HubiZitutiunsschrijl (Freie Universitat Berlin 1970). M. Cardona Proc. 9th Int. Conf Phys. Semicond. (1968) 1,365 ; SoZid State Physics suppl. 1 1 (Academic Press New York 1969). B. Lax Pruc. 9th Int. Con$ Phys. Semicond. 1968,1,253. ' K. J. Vetter and J. Bardeleben 2. Elektrochem. 1957 61 135.

ISSN:0430-0696    

DOI:10.1039/SF9700400137

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Spectroscopic Studies of Electrochemical Reactions of Adsorbed Dye Layers BY R. MEMMINO AND F. MOLLFRS Philips Forschungslaboratorium Hamburg GmbH Hamburg 54 Germany Received 14th September 1970 Dye reactions at transparent electrodes were studied by spectroscopic investigations using the attenuated total reflection technique. All experiments were performed by modulating the concentra- tion and consequently the absorption itself (potential modulation). The applicability of this method has been demonstrated by studying the distribution of monomers and dimers in adsorbed dye layers (viologene) or in the vicinity of the electrode. Measurements at higher modulation frequencies have shown that unstable semiquinones in adsorbed dye layers (methylene blue thionine) involved in the oxidation-reduction cycle could be detected.During the last two years we studied electrode processes of dyes at semiconductor electrodes with the main emphasis on the charge and energy transfer between semi- conductor and excited dye molecules. Experiments with Gap Sic ZnO and SnO have shown that such a sensitization process only occurs with dyes being adsorbed on the electrode. These investigations have shown that it is necessary to get more information about the absorption of adsorbed dyes and about reaction products. A well-known optical method to study adsorbed layers is the attenuated total reflection-technique (ATR) as applied mainly for investigations of germanium and silicon surfaces.l In this paper we report about spectroscopic studies of dyes using Sn0,-layers as transparent electrodes as they were first used for similar experiments by Kuwana and coworkers.2* These authors discussed in detail how interference phenomena complicate an evaluation of the spectra.We tried to avoid this problem by a special preparation technique of the Sn02-layers. EXPERIMENTAL For the optical investigations (ATR-technique) we used a single beam equipment. The spectral distribution was measured using a xenon lamp as a light source and a h i s s mono- chromator. The main part in the whole arrangement is the cell (fig. 1). The light enters the ATRelement via a prism glued on the glass substrate. In our investigations we used SnO2 - / I ayer prism / SnO2- s plate S \ cell FIG. 1 .-Electrolytic cell. 145 146 REACTIONS OF ADSORBED DYE LAYERS mainly effect modulation by applying a square-wave potential between the Sn02-electrode and a platinum counter electrode The modulated transmitted light was detected by a multiplier and the corresponding a.c.signal was amplified (lock-in amplifier) and displayed on a recorder. The intensity of the incident light used as reference was measured by chopping the light itself. All experiments were performed under potentiostatic conditions the electrode potential was measured against a Ag/AgCl electrode (in acetonitrile) or against a standard calomel electrode. The transmissive Sn02-layers were produced by spraying an organic solution of SnCI4 on a heated glass ~ubstrate.~ In order to avoid interference effects we made layers varying in thickness over the whole surface area. The dyes used in our experiments were products from different companies (methylene blue thionine from Merck and viologene from Brit.Drug Houses Ltd.). ATR-TECHNIQUE WITH SnOz-LAYERS Using the attenuated total reflection (ATR) method for studying the absorption of adsorbed films it is necessary to obtain a sufficiently large number of reflections at the inter- face. The greatest number of reflections may be achieved if the angle of incidence equals the critical angle & which is given by sin & = n&, where ns and nE are the refractive indices of the denser and rarer medium respectively. In our experiments we used Sn02-films on a glass substrate as shown in fig. 1. In this case the light beam passes the whole glass plate after each reflection so that the highest number of total reflections attainable is determined by the critical angle of the glass/air interface (ns = 1.51).Then only 3 reflections per cm are obtained. In most cases this technique (glass/Sn02-electrolyte) would not be very sensitive. It is only useful if the absorption itself can be modulated e.g. by a periodic oxidation and reduction of an adsorbed layer. another difficult problem arises with the system glass/Sn02-electrolyte. Since the refractive indices of glass and SnOz differ con- siderably interference effects occur by multiple reflections within the Sn02-layer. This is easily proved by measuring the spectral distribution of the light transmitted through an ATR unit which shows a number of well-defined maxima and minima. This side effect is even worse when measuring modulated absorption. Applying e.g. an a.c. voltage across the Sn02/electrolyte interface the density N of free carriers within the Sn02 electrode is modulated.According to Drude’s theory a small change in N will produce also a change in the refractive index. In our case the free carrier density N was about lO2O/crn3. The plasma wavelength A is about 1.8 p approximately as measured with similar Sn02-layers by Groth and K a ~ e r . ~ According to Kuwana’s measurements a charge 4x A/cm2 leads to a variation of the refractive index by Ans = lo-’. This value shifts the interference spectrum by about 2-3 A. Consequently a potential modulation results in a considerable intensity modulation. Measuring the spectral distribution of this effect by using a lock-in technique a type of a sinusoidal variation of the transmitted intensity with wavelength was obtained.Relative values of the intensity change up to 10 % were observed. Such a large side effect makes it almost impossible to study absorption changes during a redox process with sufficient sensitivity. Therefore we tried to prepare Sn02-layers inhomo- geneous in thickness using a special spraying technique as described in the experimental section. In this way we could reduce this intensity modulation down to less than 0.01 %. As discussed by Kuwana and coworkers 2s EVALUATION OF SPECTRA If total reflection occurs at an interface the light penetrates into the rarer medium to a certain depth depending on the angle of incidence. The strength of absorption which may be achieved by total reflection from an absorbing medium can be determined by an exact calculation using the Fresnel equations.For a relatively weak absorption a simpler approxi- mation can be derived as follows. At total reflection the intensity of the light decreases in the rarer medium by dIR(x) = (2/6)I(x)&s (1) R. MEMMING AND F. MOLLERS 147 where I(x) is the intensity within the rarer medium at the distance x from the surface. 6 is the penetration depth of the light. If the rarer medium is absorbing the intensity is also decreased by absorption so that the intensity decrease is given by dl = dIR+dIA = iI(x)dx+-I(x)dx &C(X) cos 4 where E is the extinction coefficient c the concentration of the absorbing species in the solution and 4 the angle of incidence. From eqn (2) the intensity I at the distance x from the interface can be obtained I(x) = I . exp[ -T-J 2x X&C(x) -dx] .ocos 4 From eqn (1) and (3) the fraction of light reflected at the distance x then amounts to dIR = 2 -Io exp[ - 2x .-I “&C(X) -dx]dx. 6 ocos 4 (3) (4) The total intensity of all reflected intensities can be derived by integrating eqn (4) over all I R . Taking into account that the light is attenuated a second time by absorption before it re-enters the electrode and that there is not only one but N reflections one obtains CONSTANT CONCENTRATION WITHIN THE ELECTROLYTE Assuming the concentration c to be independent of x eqn ( 5 ) simplifies to In most cases the second term in the denmoinator is small compared to 1 assuming average values of &( N lo4) and 6 ( N lod4 cm) e.g. for c 4 1 mol/I. the above equation can be simplified to ADSORBED LAYER If the absorbing species is strongly adsorbed on the electrode surface the simpIification in eqn (6) has to be modified.For an adsorbed layer it can be assumed that the concentration c is constant over its thickness d. Neglecting any further absorption within the solution itself one obtains by integrating eqn (5) over the thickness of the adsorbed layer For thin adsorbed layers ( d 4 @-and those are of interest in this paper-the above equation may be reduced to The latter part of the equation is practically identical with eqn (7) only the penetration depth b is exchanged by the thickness d of the adsorbed layer. 148 REACTIONS OF ADSORBED DYE LAYERS PERIODIC MODULATION OF CONCENTRATION Studying redox processes of dyes dissolved in the electrolyte or adsorbed at the surface the concentration can be varied periodically by potential modulation.Using this method for small concentration variations one obtains the corresponding transmission change by differentiating eqn (7) and (9) &/I0 = -i$(N/cos4)dc (dissolved species) dI/Io = - &Ad(N/coS+)dcA (adsorbed species). This method has the advantage that only the periodic intensity variations d l have to be amplified leading to a much higher sensitivity of the apparatus Moreover this method gives the possibility to study intermediate species by performing experiments at different modulation frequencies. An increase of frequency is limited with respect to the above equation however by the boundary condition that the thickness of the diffusion layer is larger than the penetration depth 6 of light d(D/v)>6 with D = diffusion coeff.v = frequency. RESULTS AND DISCUSSION Using the ATR-technique we studied redox processes with various dyes mainly with viologene thionine and methylene blue. In all these cases the dye is reduced in two steps. Viologene is the simplest example because the redox potentials for the reaction steps Ox/Sem (UE = -0.5 V) and Sem/Red (UE = -0.9 V) are very well separated (see fig. 2) i.e. the semiquinone (viologene radical V+) is stable. This makes it possible to study the absorption change during a one-electron-transfer . FIG. 2.-(Current,potential) curve for benzyl-viologene M) in de-aerated acetonitrile at SnOz and R. In fig. 3b the spectral distribution of the transmission change AI/Io is shown as measured by potential modulation in the potential range +0.2-0.7 V. For com- parison we also plotted in rig.3a the absorption spectrum of the semi-reduced viologene *+) as obtained by a standard absorption measurement (normal incidence of light). In the latter case the semi-reduced species was obtained by electrochemica1 reduction of viologene before measuring the absorption. Both spectra (fig. 3a and b) agree very well a fact which demonstrates that the ATR-technique combined with potential modulation gives correct results. It also proves that the relationship (which is an approximation) between intensity and concentration of the absorbing species is correct. According to eqn (10) the relative intensity variation AI/Io is proportional to the extinction coefficient E i.e. the vertical scales in fig. 3a and b are directly comparable. R . MEMMING AND F. MOLLERS 149 c6%- CH IQ-Q -cH~- Q H ~ a) measured by standard photometer - W 350 400 I 500 ,d 550 600 b) measured by potential modulated ATR - technique ,c 600 350 400 500 550 O ~ ~ ~ ’ ~ ~ > 4 A (nm) (b) spectral distribution of the modulated absorbance.FIG. 3 . 4 0 ) Absorption spectrum of semireduced benzyl-viologene in deaerated acetonitrile ; In the following part we give a few examples of (a) how the distribution of mono- mers and dimers in an adsorbed dye-layer or in the vicinity of the electrode may be obtained and (b) how intermediate species (radicals) can be recognized. FORMATION OF DIMERS NEAR THE ELECTRODE The simplest example is again viologene (V2+). This compound is colourless and has its absorption maximum at 260 nm. The semiquinone of viologene (V+) is blue and shows two main absorption bands around 400 and 570 nm.According to fig. 3 each absorption band shows two maxima; their height depends on the concentration of the semiquinone. In the shorter wavelength region e.g. the absorption peak at 380 nm occurs only above a certain concentration. From this fact it may be concluded that this peak is due to the formation of a dimer whereas the peak at 394 nm has to be assigned to the corresponding monomer. The same effect was observed at higher wavelength (dimer at 555 nm monomer at 600 nm). In order to check whether the ratio of dimer to monomer concentrations near the surface differs from that of the bulk we determined the absorption spectra at several modulation frequencies keeping the potential amplitude constant. According to the results obtained at 2 20 and 200 c/s as shown in fig.4a (solid lines) much lower signals are observed if the modulation frequency is increased. This result is reason- able since at higher frequencies only a much lower amount of the dye (V2++e-+Q+) can be expected to be involved in the electrode process (see also eqn (10)). On the other hand if the frequency is increased only those molecules which are close to the electrode can follow the oxidation-reduction cycle. Besides this general decrease of AI/Io the shape of the spectra also varies. At higher modulation frequencies more dimers are visible in the spectra. For a better comparison of the 150 REACTIONS OF ADSORBED DYE LAYERS different spectra we shifted the 2 and 200c/s curves so that the monomer peaks coincide (dotted lines). According to this result the dimer concentration is larger than the monomer concentration in the vicinity of the electrode surface.2oocps A a) I 360 380 400 420 Yi reflections b) I I I I 360 380 400 420 1 (nm) FIG. 4.-Modulated absorbance of semireduced benzyl-viologene in acetonitrile (a) at different modulation frequencies ; (b) for different angles of incidence (dashed curve see text). This result can be proved by a different kind of experiment. We measured again the spectral distribution now for different angles of incidence which also led to different numbers of reflections. In fig. 4b results are shown obtained with angles of 60 and 74" keeping the frequency in both cases constant. It is evident that a relatively large number of dimers occurs if the angle of incidence is increased and thus the penetration depth decreased.According to Harrick * the penetration depth is very sensitive to any change of the angle of incidence especially around the critical angle. Consequently we could expect a decrease of the penetration depth by more than a factor of up to 10 when varying the angle of incidence from 60 to 74". This result indicates that the concentration of dimers increases with decreasing distance from the electrode. It cannot be concluded from this experiment however whether adsorbed molecules are involved in the reduction and oxidation mechanism. We tried to solve this problem by using polarized light. Since adsorbed dye molecules should be oriented at the surface and since the transition moment for the optical transition (singlet-+singlet) is parallel to the molecular plane one should also expect differences in the ratio monomer/dimer for different polarizations of the light.We R . MEMMING AND F . MOLLERS 151 observed indeed such an effect but the accuracy of the corresponding measurements was poor so that this observation is not proved. The [dimer]/[monomer] ratio was mainly studied in the shorter wavelength range. Qualitatively the same results were obtained for the other transition around 550 nm. Finally we worked at rather low modulation frequencies. This is due to a certain inhibition of electrode processes observed with viologene leading to a low yield of charge transfer during one cycle. eD (CH3)zNG ' "N(CH& 3x104 n I rl lo4 CI 2 - W QJ 3d03 500 600 700 500 600 700 1 (M-4 FIG. 5.-(a) Absorption spectrum of methylene blue ( M) in 0.1 N H2S04 ; (b) spectral distribution of modulated absorbance at different frequencies (modulation amplitude AU = 0.6 V).The zero point of modulation t A U + is +0.25V (against SCE); dotted line semiquinone spectrum l+I - U obtained by subtracting curve 1 by 2. INTERMEDIATES I N ADSORBED D Y E LAYERS We here discuss results obtained with dyes which are readily reduced electro- chemically to their leuco-form as e.g. methylene blue and thionene. The semi- quinone of these compounds is normally not stable except at very low pH-values (PH c - l).9 A typical spectrum as obtained with methylene blue at a low modulation frequency (2 c/s) is shown in fig. 5b (solid line 2 c/s). For comparison we also plotted the absorption spectrum of methylene blue in fig. 5a. The spectrum measured by using potential modulation agrees well with the absorption spectrum.The potential modulation was adjusted in this experiment so that leuco-methylene blue 152 REACTIONS OF ADSORBED DYE LAYERS was formed during the cathodic sweep and re-oxidized during the anodic sweep. In the wavelength range considered here only methylene blue itself absorbs whereas the leuco-form does not show any absorption. The AI/l,,-values are here of opposite sign to that of viologene because the concentration of methylene blue decreases in the cathodic phase whereas for viologene the absorbing species (semiquinone) is formed in the cathodic phase. This kind of spectrum was also obtained after having exchanged the dye solution by the pure solvent. This fact proves that adsorbed dye molecules are almost entirely involved in the oxidation-reduction cycle.The purpose of the investigations with these compounds was to check whether it is possible also to observe semiquinones e.g. by increasing the modulation frequency although the life-time of semiquinones is expected to be very short. First we performed some basic measurements with methylene blue in strong alkaline solutions-where still two reduction steps can be recognized in the current potential curve (as indicated by arrows in fig. 6a) ; i.e. where the serniquinone is not completely unstable. Since in this solution a very thick layer of methylene blue is adsorbed -2 1 (nm) FIG. 6 . 4 2 ) Current-potential curve for M methylene in lo-' N NaOH ; (b) transmission change against potential ; (c) Modulated absorbance against wavelength (amplitude A U = 0.6 V).it was possible to measure directly the transmission change as a function of potential even without using the lock-in technique. T h i s is demonstrated in fig. 6b for two different wavelengths. Sweeping the electrode potential from anodic towards cathodic potentials the transmission decreases around - 0.4 V and increases at higher cathodic potentials. Comparing this transmission change with the corre- sponding current-potential curve (fig. 6 4 it is evident that the transmission minimum has to be assigned to the formation of the semiquinone whereas the transmission increase is due to the formation of leuco-m.b. which is equivalent to a decrease of the methylene blue concentration. Performing the same type of experiment by using the lock-in technique which means a certain kind of averaging of the potential dependent absorption positive and negative AI/lo-values are obtained in the spectral distribution (fig.6c). The absorption spectra of methylene blue and semiquinone R . MEMMING AND F. MOLLERS 153 0 2 a' lU u =+ 0.4V 3 4 A (nm) FIG. 7.-Spectral distribution of modulated absorbance with methylene blue M) for potentials f7 Dashed line semiquinone (obtained by subtracting the 0.25 V curve from the curve). different -0.12 v FIG. 8.-(Current potential) curve of adsorbed thionine on SnOz in 0.1 N H2S04 at different sweeping rates. 154 REACTIONS OF ADSORBED DYE LAYERS overlap considerably. At those wavelengths at which negative values occur the absorption of the semiquinone dominates and vice versa. We did not evaluate this spectrum in detail the results presented in fig.6 only demonstrate principally the kind of curves obtained if the spectra of two species overlap. In most cases the semiquinone is not stable. This can be concluded from the current potential curves as measured with thionine at different sweeping rates (fig. 8). According to these curves about 8-10 monolayers of the dye are adsorbed on the electrode. The same result was obtained with methylene blue using a solution of the same pH-value. In agreement with the current potential behaviour no optical absorption by the semiquinone is observed in the absorption spectrum at a lower modulation frequency (2c/s). At higher modulation frequencies however the shape of the absorption curve (AI/Z, against A) varies considerably (see the 20c/s curve in fig.5b). This result indicates that an intermediate species probably the semiquinone is formed which-as out-lined above-should lead to AZ/Zo-values of opposite sign. The spectra of the two species overlap so much and the absorption of methylene blue itself dominates so that only a relative change could be observed. At higher modulation frequencies the AI/Io-values are over the whole wavelength range smaller than for low frequencies because fewer adsorbed molecules are reduced and re- oxidized. In order to be able to evaluate the spectrum of the intermediate species we FIG. 9 . 4 ~ ) Absorption spectrum ofLhionine m) in 0.1 N H2S04 ; (b) spectral distribution of modulated absorbance for different U-values. Dashed line semiquinone (obtained by subtracting the 0.14 V-curve from -0.12 V-curve).R . MEMMING AND F. MOLLERS 155 shifted the 20 c/s curve upwards so that it coincides with the 2 c/s curve at the shorter wavelength side as indicated by an arrow. This procedure seems to be reasonabIe since both curves are paxallel in this range (logarithmic scale!). The spectrum of the intermediate species was then obtained by subtracting the corresponding AI/Io values of both curves from each other (dashed curve A in fig. 56). The same change of spectra was also observed by shifting the potential modulation either more into the anodic or cathodic potential region keeping the modulation amplitude and frequency constant. This is demonstrated for methylene blue in fig. 7 and for thionine in fig. 9b. In these figures we have already shifted the curves so that they coincide in the shorter wavelength region.The spectra of the inter- mediates were obtained again by subtracting the corresponding curves. This fit of the curves by shifting allows only a qualitative determination of the semiquinone spectrum. On the other hand spectra of semiquinones obtained by photochemical investigations are available which may be used for comparison. Lewis and Bigeleisen lo studied the semiquinone formation in solid solutions and found an absorption band at 575 nm for thionine and at 630 nm for methylene blue. These values agree well especially for methylene blue with our maxima. On the other hand flash photolysis measurements have shown that different protonated semiquinones exist.'' l 3 In solutions of a pH<6 Fischer l1 observed e.g.two absorption bands for thionine at 400 and 750 nm. According to Kramer l 2 the semiquinone is formed by the following reaction scheme +hv +H+ +red. ag. . T+ -+T+*+3T+ + 3TH2+ + TH++product (T+* = excited state 3T+ = triplet state red. ag. = reducing agent) i.e. the protona- tion occurs in the triplet state. At higher pH-values the semiquinone absorption disappears (PK- 8). Similar results were obtained with methylene blue. All these authors report that at higher pH-values the unprotonated semiquinone is formed and indicate that the absorption overlaps with that of thionine. According to these results we found only the unprotonated semiquinone even at pHc6. In an electrochemical process a charge transfer would be the primary and a protonation the secondary step +e- . +H+ T+ -+ T -+ TH+.'These considerations lead to the conclusion that only unprotonated semiquinone molecules are adsorbed on the electrode. Thus according to the results presented in this paper this optical method is applicable to studies of adsorbed layers electrode reactions and even of short-lived intermediates. The authors are indebted to Dr. K. H. Beckmann and Dr. U. Kriiger for many stimulating discussions. Thanks are also due to Mr. H. Tolle and Mr. G. Kiirsten for performing the experiments with great care and to Mr. G. Bringmann for preparing the Sn0,-films. K. H. Beckmann and N. J. Harrick Proc. Electrochem. Symp. Opt. Properties of Dielectric Films (Boston 1968). N. Winograd and T. Kuwana Electroanal Chem. Inter$ Electrochem. 1969,23 333. R. Groth E. Kauer and P. C. v. d. Linden 2. Naturforsch. 1962 17a 789. P. Drude 2. Phys. 1900,1 161. K. H. Beckmann unpublished results. ' V. S. Srinivasan and T. Kuwana J. Phys. Chem. 1968,72 1144. 156 REACTIONS OF ADSORBED DYE LAYERS ’ W. M. Schwarz Thesis (Univ. Wisconsin 1961). * N. J. Harrick Internal Reflection Spectroscopy (Interscience Publishers New York 1967). L. Michaelis M. P. Schubert and S. Granick J. Amer. Chem. SOC. 1940,62,204. lo G. N. Lewis and J. Bigeleisen J. Amer. Chem. SOC. 1943 65 2419. l1 H. Fischer 2. phys. Chem. N.F. 1964,43,177. l2 H. E. A. Kramer and A. Maute Ber. Bunsenges. 1968,72,1092. l3 R. Bonneau J. Faure and J. Joussot-Dubien Ber. Bunsenges. 1968,72,263.

ISSN:0430-0696    

DOI:10.1039/SF9700400145

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Application of Internal Reflection Spectroscopy to the Study of Adsorbed Layers at Interfaces BY HARRY B. MARK JR. AND ERIC N. RANDALL Dept. of Chemistry The University of Cincinnati Cincinnati Ohio 45221 U.S.A. Received 1 st September 1970 The application of the technique of internal reflection spectroscopy (IRS) has certain inherent properties which allow the direct observation of the spectra of adsorbed layers at interfaces. Several representative examples of specific research problems related to the study of adsorbed films which have been investigated by IRS are discussed. Included are discussions of adsorbed films on both optically transparent and opaque adsorbents surface functional groups and adsorption on thin metal film electrodes. Also discussed are the principles involved the advantages and limitations of the method and the instrumental experimental and optical problems peculiar to IRS measure- ments.Attenuated total reflectance (ATR) the principle of internal reflectance spectro- scopy has long been used in the measurement of refractive index.l Later the application of IRS as a method for obtaining infra-red spectra of opaque materials was introduced. 2-6 In internal reflection spectroscopy monochromatic light is injected into a crystal of suitable refractive index n which in turn is in contact with the sample medium refractive index = n2 where n,>n2 (fig. 1). If the crystal surface is highly polished and the angle 0 is greater than the critical angle 8 (defined by 0 = sin-' (n2/n1)) total internal reflection will take place at the interface. As shown in fig.I components of the incident and reflected waves combine to form a standing wave on the crystal side of the interface. An E-vector evanescent wave is also produced which penetrates into the sample medium; the amplitude of the evanescent wave decays exponentially with increasing distance into the sample. The range of penetration is around 12/7-12/10.7 IRS Element Sample FIG. 1 .-Attenuated total internal reflection. Energy is transferred to an absorbing sample via the evanescent wave. 157 158 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS The presence of a photo-active species on the sample side of the interface will cause absorption of energy from the evanescent wave and transfer of energy across the interface. This loss of energy to the absorbing species is seen as an attenuation of the reflected beam.8 By increasing the number of reflections at the crystal-sample interface i.e.multiple reflectance spectroscopy (MRS) the resulting attenuation is increased proportionally.8 The attenuation of the light passing through an IRS cell is monitored and recorded just as in a transmission spectrum. There is no shift in frequency between the light in the crystal and the evanescent wave but the resulting internal reflection spectrum shows absorptions at slightly shifted wavelengths from a transmission spectrum. Detailed discussions of this effect and the principles of IRS are given in ref. ((8)-(12)). Fig. 2a and 2b show two typical IRS cell attachments schematically. Such attachments use either mirror or prism systems to redirect the optical light path in the sample compartment of conventional spectrophotometers through the IRS elements as shown.The sample of interest is brought into contact with the reflecting element. Sample I 1 Internal Reflection Element Prlsm Prism - Light Mlrrorr A Somple I 1 Internal Reflection Element L Light Prism Prlsm 8. FIG. 2.-Mdtiple reflection IRS cells. A cell using reflecting prisms and mirrors ; B cell using refracting prisms. IRS techniques seem particularly well-suited for investigations of adsorbed layers at interfaczs. In order to demonstrate the versatility of the method a variety of systems are discussed in some detail. Examples of the IRS spectra of adsorbed films on both optically transparent and opaque surfaces are given and the significance of these spectra with respect to the understanding of adsorptive interactions are dis- cussed.The potential of the technique for the study of the role of adsorption in electrochemical processes is also given. In addition concurrent developments in instrumentation such as rapid scan spectroscopy on-line computer spectrometers and Fourier Transform spectroscopy are discussed as they apply to the special problems of IRS studies of adsorbed films. H. B. MARK JR. AND E. N . RANDALL 159 IRS STUDIES OF ADSORBED FILMS AT INTERFACES The information that can be gained from IRS studies of adsorbed films is multifold. The first and probably most important kind of information desired is to detect the presence of very thin films. Much work has been done with ultra-thin films in the attempt to detect their presence and make identifications (e.g.see ref. (13)-(17)). METHYLENE BLUE O N GLASS Methylene blue is strongly adsorbed on soft glass as a monomolecular layer where a gradual decomposition of the dye occurs. * As this system is ideally suited for study by IRS a few simple experiments were carried out to illustrate the application of IRS to the study of adsorbed films. In these experiments the glass plate on which the methylene blue is adsorbed acts as the reflection element. The methylene blue film was prepared by dipping a fire-polished glass plate internal reflection element A (md FIG. 3.-Visible internal reflection. Spectra of a methylene blue monolayer adsorbed on glass. (IRE) 1 mm thick in this instance into an aqueous solution of methylene blue. The IRE was then rinsed in distilled water and wiped dry with lens tissues.Fig. 3 shows the visible IRS spectra of methylene blue adsorbed on the glass plate using both parallel and perpendicular polarization of the incident light beam. The transmission spectrum of aqueous methylene blue is shown also for comparison. The polarization of the incident beam is defined in terms of the E-field vector orienta- tion with respect to the plane of incidence.I2 (A transmission spectrum of the glass plate prepared in the above manner (not shown) showed no signs of the methylene blue.) 160 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS There are certain features of these spectra which are of special interest. Unlike typical solution or solid matrix IRS spectra which are ordinarily strikingly similar to the transmission spectra (although not exactly the same),8 there is a significant change in the intensities of the absorption peaks at 600 and 660 nm relative to those of the transmission spectrum.In addition there is the appearance of a new peak at 740 nm in the IRS spectrum. Rabinowitch and Epstein postulated that methylene blue exists in neutral solutions in both a monomeric and a dimeric form.18 The dimer form is presumably a loosely bound " sandwich-type " compo~nd.'~ Rabinowitch has assigned the peak at 664 nm in the transmission spectrum to the monomeric form and the peak at 600 nm to the dimer. Further studies of this dye have substantiated this as~ignment.'~ In the solution phase the monomer peak is significantly larger than the dimer peak. 4 O j 2.8 - Y = Perpendicular X = Parallel x 2 = Par. Normal f ZlA FIG.4.-Electric field intensities for total internal reflection at a glass-water interface. No absorption; 8 = 60"; near critical angle. The dimer 600 nm absorption peak in the IRS spectra is relatively enhanced which suggests that the dimer form is in some way preferentially adsorbed at the glass surface. This hypothesis is further supported by the presence of the new peak at 740 nm which is predicted by theory.20 When the dimer is formed the recombina- tion of molecular orbitals results in new empty excited state orbitals of higher and lower energy than the empty excited state orbitals of the monomer. Ths is reflected by a splitting of the absorption peak of the monomer yielding peaks at shorter and longer wavelengths for the dimer. It is possible to obtain further information concerning the nature of the adsorbed film by comparing the spectra obtained by perpendicular and parallel polarized incident light.There is an overall difference in the magnitude of the energy absorbed in the two modes. For parallel polarization the absorbance of light energy is significantly lower. The direction of the electronic transition moment vector for a molecule dictates the plane of polarization of the absorbed radiation.2 l For transitions involving n-orbitals this vector is always in H . B. MARK JR. AND E. N. RANDALL 161 the plane of the molecule. Also perpendicularly polarized incident light in IRS has only a component of the E-vector parallel to the IRE interface (E,,)? while parallel polarization has components both parallel and normal to the interface (Ex and E respectively).8* l2 The relative magnitudes of these E-vector components at the interface are in the order E,>E,>E as shown in fig.4.22 Thus it appears that adsorbed molecules of methylene blue are predominantly oriented at the interface such that only the E-vector components parallel to the interface give rise to electronic spectra i.e. the molecules are adsorbed with the plane of the molecule parallel to the plane of the substrate. This conclusion is suggested by the fact that the overall intensity of the spectrum obtained with parallel polarization is diminished with respect to that obtained using perpendicular polarization. The above conclusions concerning orientation are based on preliminary experimental results and are therefore not conclusive. It will be necessary to obtain more information with respect to the actual thickness of the methylene blue layer (i.e.it must be ascertained whether or not the films being studied are indeed monolayers) the nature of the glass surface etc. before the results can be interpreted absolutely. MONOETHYL OCTADECANEDIOATE ON Ge SURFACES Sharpe has used IRS techniques to study ultra-thin films of monoethyl octa- decanedioate in the infra-red.13 Fig. 5a shows the IRS spectrum of a thick film of the monoethyl ester of octadecanoic acid on a germanium IRE. For the bulk or thick film the orientation of the molecules is head-head tail-tail head-head etc. 6 T IX CONTACT 3000 2600 1800 1600 FIG. 5 .-Infra-red internal reflection spectra of monoethyl octadecanedioate films on germanium. A thick film ; B monomolecular film ; C 3.5 molecules average thickness.S4-6 162 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS through hydrogen bonding with respect to the z - d i r e ~ t i o n . ~ ~ ~ 24 As a consequence one observes a strong hydrogen bonded acid-carbonyl peak at 1710cm-l. The 2800-3000 cm-l i.e. methylenic C-H stretch region is included to show the spectra of non-participating bonding groups remain unchanged. Fig. 5b is the IRS spectrum of an oriented absorbed monomolecular layer of monoethyl octadecanedioate. In this spectrum there is an absence of the 1710 cm-I peak but there is the appearance of a peak of higher energy at approximately 1745 cm-1 which is indicative of the presence of a free acid-carbonyl group. Fig. 5c shows the spectrum of a film of monoethyl octadecanedioate of 3.5 molecules average thickness which had been in contact with the germanium crystal for a period of 25 h.There is still no appreci- able peak at 1710cm-l indicating that the layers are stacked head-tail head-tail etc. this being the orientation expected by the method of dep0siti0n.l~ Moreover a new peak has appeared at 1600 cm-I. Sharpe attributes this peak to the formation of a germanium salt by reaction of the carbonyl group with the surface after chemi- sorption.13 This illustrates that chemisorption bonds can be observed and studied by IRS techniques. ACTIVE CARBON Mattson et al. have performed a number of experiments involving the application of infra-red IRS techniques in an attempt to elucidate the surface chemistry involved in adsorptions on active carbon surfaces.15* 22* 25-28 These studies illustrate unique applications of IRS methods for obtaining the spectrum of the surface of optically opaque materials such as carbon.I5 Transmission studies of powdered carbon samples although attempted in some s t ~ d e s ~ ~ ' 31 are unsatisfactory for studies of surface phenomenon.This is due not only to the problem of their high extinction coefficient but also the high refractive index which brings on the problem of light scattering by the carbon particles. Only some 1 % of the incident energy will be transmitted by a carbon film sample 3.7 r ( ~ thick at a wavelength of 5 p.15 FORWARD SCATTERED- RADIATION )/CARBON MICROCRYSTALS~ I N A TRANSPARENT MATRIX FIG. 6.-Schematic diagram of the forward scattering process obtained when strongly absorbing carbon particles are included in a transparent matrix.Fig. 6 is a schematic diagram of the forward scattering process involved in the trans- mission mode. Transmission measurements made on samples of this type are in fact a complicated form of a diffuse reflectance measurement.15 The use of IRS methods to study absorbing powders circumvents this attenuation-scattering problem H. B. MARK JR. A N D E . N . RANDALL 163 in certain cases.32 In situations where the depth of penetration of the evanescent wave into the supporting medium is small compared to the dimensions of the particles scattering does not appear to present a problem. This conclusion can be attained by first assuming that the propagative nature of the evanescent wave which is not transverse in nature is not a significant contributor to the interaction as there are few particles in the region where the evanescent field is intense.Those particles or portions of particles which are in this region adjacent to the IRS will tend to absorb energy rather than scatter it. The area of interaction includes therefore the points of contact and part of the area of the particles in the neighbourhood of the contact points. By using the equations for a multiphase system developed by Hansen,12 a model system as shown in fig. 7 (which is a four-phase system) can be set up. Calculation of the reflectance [employing a distribution function to describe the radii (hence the surface area) of spherical particles] as a function of the distance perpen- dicular to the interface gives a fairly accurate representation of the observed experi- mental attenuation for such a s y ~ t e n i .~ ~ Thus scattering does not appear to be an important factor. VANESCENT WAVE EFRACTED BYSPHERE ATTENUATED -- SUPPORTING MEDIUM PHASE 3 FIG. 7.-Schematic representation of the model assumed for the calculation of the reflectance of a species adsorbed on spherical particles. For example fig. 8 shows the 1600-1200cm-1 portion of the IR transmission spectrum (top) of p-nitrophenol. This region contains the NO2 and C-0 group bands. The lower spectrum is that obtained for adsorbed p-nitrophenol on active carbon.27 From the fact that the nitro- and phenolic C-0 peaks are essentially unchanged by adsorption on carbon (as shown in fig. 8) it was concluded that the oxygen of the phenol or phenolate C-0 group could not be directly associated with the active carbon surface in any specific bond interaction related to the chernis~rption.~~ Other IRS studies of the spectra of sirface functional groups of active carbon as a function of activation temperature and atmosphere,26 showed that the surface concentration of carbonyl oxygen on the carbon was directly related to the absorption capacities for p-nitrophenol These results lead to the conclusion as shown in fig.9. 164 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS that the mechanism of the adsorption reaction for p-nitrophenol on active carbon was the formation of a donor-acceptor complex between the surface carbonyl oxygen (donor) and the n-system of the p-nitrophenol (acceptor).27* 35* 36 Similar donor-acceptor complex formation has also been proposed by Low et aL3' to explain the i.-r.spectra obtained by transmission methods for the adsorption of ammonia on dehydroxylated germania gel. 1600 I40 0 I 1200 FIG. 8.-1600-1200cm-1 region. The top spectrum shows the NOz group bands and C-0 band of thick film of p-nitrophenol. The lower spectrum shows the same bands of p-nitrophenol adsorbed on active carbon. frequency cm-1 SIMULTANEOUS IRS-ELECTROCHEMICAL STUDIES The application of IRS to the study of electrochemical processes using optically transparent electrodes (OTE) is currently gaining ~ s a g e . ~ ~ - ~ O Both thin metal films on glass 38* 39 and semi-conducting tin-oxide-coated glass 40 have been employed. Both electrode systems have been employed successfully in the investigation of homogeneous chemical reactions involved in electrode processe~,~~ 9 42 but little work has been carried out on studying the role of adsorption in electrode reactions.However the technique appears to have considerable promise. One example is cited below. In order to use a metal film for simultaneous electrochemical-IRS studies one is faced with the problem of obtaining a film of proper thickness. If the film is too thick its own absorbance will preclude spectrophotometric measurements being made on species in contact with it the extinction coefficient of conductors generally being quite large.43 On the other hand if the film is too thin the conductivity decreases to the point that the film is no longer suitable for use as an electrode.44 An isotropic gold or platinum film of about 50-400A in thickness are suitable for these types of studies.39 H.B. MARK J R . A N D E. N. RANDALL t I 1ooa o k 4 frequency cm-l A 165 activation temperature "C B FIG. 9.-A LRS spectra of sugar carbons activated in 1 % 02+99 % N2 at various temperatures; B p-nitrophenol capacities for sugar carbons activated in 1 % 02+99 % N2 as a function of activa- tion temperature. 166 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS Visible IRS spectra of species in contact with such a metal film (deposited on glass) electrode have been shown to be essentially identical to the corresponding transmission ~pectra.~ One particularly interesting result of such investigations was that the presence of the gold not only did not distort spectra but for parallel polarization of the incident light even enhanced the spectrum sensitivity relative to that obtained at the glass substrate only (e.g.see fig. 4 4 9 45 This phenomenon was first observed by Pons et ~ 1 . ~ ~ and quantitatively explained by Prostak et ~ 1 . ~ ~ and Mattson et aL4' A comparison of fig. 11 and 4 illustrates the effect of the metal film (gold) on the magnitude of the parallel components of the E-vector at the solution interface. 400 500 600 400 500 600 wavelength (nm) 4 reflections. FIG. 10.-Internal reflection spectra of eosin Y. 100 g/l. eosin Y 50A thick gold film 0 - 72" With respect to adsorption at electrode interfaces fig. 12 shows the IRS spectra obtained using a gold OTE of 4,7-dimethyl ferroin a highly coloured redox indicator in aqueous solution with 0.1 M Na,SO present as supporting electrolyte. The IRS spectrum for the solution on glass and the transmission spectrum are also included for comparison.Electrochemical studies have shown that at an applied potential of +0.379 V against a saturated calomel electrode (SCE) the species present at the electrode surface to be in the fully reduced (coloured dimethyl ferrion) form (see fig. 12). At 0.940V against SCE which represents a point on the diffusion-limited oxidation wave the surface concentration of the species is all in the colourless (dimethyl ferriin) oxidized form. Because this is not a massive electrolysis experi- ment a transmission OTE spectrum would be virtually unchanged by a potential step from +0.38 to +0.94V. However the IRS spectrum clearly indicates a reduction in the concentration of the reduced form at the electrode surface after the potential step is applied as shown in fig.12. However the absorption at 450nm does not go to zero as expected the~retically.~' (The surface concentration of the reduced form equals zero and the optical window is small compared to the diffusion H . B . MARK JR. AND E . N. RANDALL 167 FIG. 11.- 8.00 7.20.. 6'40- 5.60- 480- r r "V -2.00 -1.60 -1.20 -80 -*40 -00 a024 -18 -34 *50 *66 -82 ZIA HIA = 2.40E-02 ZlA glass/l20A goldlwater; K(3) = 0.0; 0 = 60"; near critical angle Y = perpendicular; X = parallel X ; Z = par. normal -Electric field intensities for total internal reflection at a glass-gold film-water interface 400 450 500 550 600 650 wavelength (nm) FIG. 12.-Spectra of 12.5 mM 4,7-dimethyl ferroin solution and supporting electrolyte - spectrum of supporting electrolyte only -- Conditions for internal reflection spectra 50 A thick gold film parallel polarization 8 = 72" 4 reflections ; conditions for transmission spectrum cell thickness = 55 luu 168 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS layer thickness.'* 39- 40) This observation suggests that an adsorbed layer of the reduced complex exists on the electrode surface which is electro-inactive at the applied potential resulting in a diffusion limited oxidation of the bulk complex.39 Calcula- tions show that this adsorbed layer corresponds roughly to a monolayer of the reduced form.39 Thus it appears that the electron transfer process is taking place through this adsorbed layer.49 This conclusion is based on limited experimental evidence but it does serve to illustrate the power of simultaneous IRS-electrochemical tech- niques in the study of adsorption in electrode processes.Because the IRS spectra obtained through a metal film are transmission-like they may be useful in identifying reactive intermediates in many types of surface reactions such as catalysis reactions surface oxide formation,8* 3 9 9 etc. INSTRUMENTAL CONSIDERATIONS IN IRS STUDIES OF ADSORBED FILMS In all the cases discussed above with the exception of the IRS studies of carbon surfaces the absorbed film of interest had intense spectral bands and little or no energy loss to the lower refractive index media (air or water) in the wavelength region of interest. Thus conventional spectrophotometers could be used in these studies. However in future studies most systems will involve adsorbed films with medium to low absorptivity which will be in contact with supporting media which may absorb many times the light energy absorbed by the film.In other words the spectra of the films will consist of a small signal on top of a massive interfering signal from the supporting media. For such systems conventional instrumentation is not possible. This part of the paper discusses both rapid scan spectrometry and signal averaging techniques that axe now being developed to handle these problems and shows that the technique of Fourier Transform spectroscopy 1-55 is ideally suited for development of surface studies. PM T REF 13 L4 B 8 SAMPLE PMT -------_I FIG. 13.-Schematic diagram of the optical system of a rapid scan spectrophotometer. B beam- splitter ; C Cassegrainian collector ; G diffraction grating ; L2-L4 focusing lenses ; M vibrating mirror ; PMT photomultiplier tube ; Sz S2 slits.The first developments in instrumentation for IRS studies (with optically transparent electrodes) was the rapid scan spectrophotometer developed by Strojek Gruver and Kuwana. Fig. 13 and 14 show that basic optical and electronic arrange- ments respectively of this instrument. The essential features of this instrument are the vibrating mirror M which is electronically controlled to oscillate at a constant velocity with no " dead time " a beam-splitter B and two matched photomultiplier tubes (PMT). The outputs of the PMT are converted for direct absorbance readings by the differential log ratio amplifier. This unit is capable of taking over 2000 spectra H.B . MARK JR. AND E. N. RANDALL 169 per second over the entire visible wavelength region (the same principles can be used in any region however) and storing these spectra on magnetic tape for computerized data reduction etc. By means of simple signal averaging of these repetitive experi- ments signals which represent of the overall signal can be resolved. This rapid scan technique is also valuable for the study of rapid time-dependent processes at the interface and has been employed successfully in the study of homogeneous intermediates in electrode processes. In principle it could also be used to study heterogeneous processes as well. I A WAVE GEN. I J MIRROR bias DRIVING AMPS m - 4 RG. 14.-Schematic representation of the electronic components of a rapid scan spectrophotomder. The rapid scan approach has also been taken in this laboratory in the design of an IR ~pectrometer.~~ However a circular filter wheel (CVF) is employed to obtain the rapid wavelength scan (100 spectrals) as shown in fig.15. Also the sensitivity (or signal-to-background ratio) has been improved by about an order of magnitude by including modulation of the light signal by a chopper system and phase selective detection of the modulated signal as shown in fig. 15. All operations of this unit (modulation wavelength scan signal averaging data processing etc.) are controlled and carried out by a small in-line dedicated computer. Although this device has been employed for some external reflectance studies of organic films it has not yet been applied to IRS interface studies. Such problems are now under investigation.One recent development which offers increased advantages with respect to sen- sitivity rejection of background and resolution is the so-called Fourier Transform spectrometer.58* 5 9 Even though the principles and hardware are considerably different from those of ordinary dispersion instruments the spectra obtained from each are identical in form. Fig. 16 shows a schematic diagram of a Fourier Trans- form spectrometer. At the heart of the system is a Michelson interferometer. Polychromatic light passes through the interferometer and sample to a detector. With suitable optical components the ultra-violet visible and infra-red regions of the spectrum are all accessible. One mirror of the interferometer moves at a constant velocity and an interferogram is produced that contains information about the intensity and frequency of the light.The original polychromatic light and the inter- ferogram are related as a Fourier pair. By subjecting the interferogram to Fourier analysis one is able to extract an intensity-frequency plot. Detailed descriptions of Fourier Transform spectroscopy and its applications can be found in ref. (51)-(55) and (60). Fourier analysis while straightforward has been tedious and time-consuming. 170 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS This is the chief reason that Fourier Transform spectroscopy has only recently been adopted as a practical research tool. The data can be treated by analogue devices to give directly an intensity-frequency plot but the advent of the mini-computer has made digital data reduction practical and economical.By digitizing the interferogram 1 - - -_. CHOPPED LIGHT SIGNAL FROM IRS CELL OPTOMECHANICAL SCAN HEAD --.. - -_--__ - CONTROL AND SIGNAL SEPPING PROCESSING CONSOLE COMMANDS tVF REFERENCE POSITION SIGNAL ECfERNAL 1 'IGNAL . A.0 CONVERTER -______ INTERFACE ELECTRONlCS t I CHWPED LIGHT SIGNAL (SANE CK)PPER FOR CONTROLS INDICATORS SPECTROMETER SOURCE I 1 - AND REFERENCE LIGHT) t INPUT/OUTWT UNIT ASR-33 KEYBOARD PRINTER osclLLoscoPE DISPLAY W E R TAE- J FIG. 15.-Schematic representation of a computer monitored rapid scan infra-red spectrometer. and having a computer perform the Fourier analysis this method gains a versatility unattained by other methods. Because the resulting spectrum is in digital form in the computer it can be manipulated to give any of a number of desired types of output by writing the proper software.For instance full-scale expansion of a single band or elimination of interfering spectra might be obtained on command if the proper In terferometer Source Sample Compartment ,P Detector FIG. 16.-Schematic diagram of the optical system of a Fourier Transform spectrometer (FTS). instructions have been programmed into the computer. Because the instrument has a short scan period (on the order of seconds) multiple scanning with an inherent improvement of the signal-to-noise (S/N) ratio due to averaging of random signals becomes feasible. Thus the advantages of Fourier Transform spectroscopy are numerous. Either a significant improvement in the S/N ratio (with multiple scanning) or rapid response with an equivalent S/N ratio (with a small number of scans) can be obtained.A constant resolution of 0.5 cm-' throughout the i.-r. spectral range is currently avail- H . B . MARK J R . A N D E . N . RANDALL 171 able with improvements expected in the future (the resolution being a function of the distance travelled by the moving mirror of the interferometer). The amount of radiant energy passing through a sample is comparatively large due to the fact that filters gratings slits etc. are not required by a Fourier Transform Spectrometer. As an example of the power of Fourier Transform techniques fig. 17 shows a comparison of the ordinary Nujol mull transmission spectrum of histidine hydro- chloride and a Fourier Transform spectrum (using a Block Engineering FTS-14 spectrometer) of histidine hydrochloride dissolved in water in a spectral region where water itself has absorbed over 95 % of the incident energy using a 0.5 mm cell.No ordinate scale expansion was used. This spectrum was obtained using 256 repetitive scans.61 We are presently constructing both a visible and j.-r. region Fourier Trans- form system for IRS studies using OTE. We believe that it will be possible to obtain the spectra for species in the electrical double layer and adsorbed on the electrode during electrolysis and perhaps even to obtain spectra of intermediates in the heterogeneous electron process. This unit will also be employed for surface studies in heterogeneous catalytic processes and adsorption reactions. 1550 1400 1200 lOOv I ! 950 A 1 1 1 1 1 1 1 1 1 1 1 10 1400 1200 1000 B 950 FIG. 17.-Comparison of FTS and conventional spectra of histidine hydrochloride.A In water 0.05 mm cell (FTS) ; B in a Nujol mull (conventional). This research was supported in part by the National Science Foundation Grant No. GP-9307 and the Federal Water Quality Administration Grant No. 16020 ELH. In addition one of us (E. N. R.) is grateful for support received through an educational paid-leave-of-absence provided by Corning Glass Works Corning New York. B. J. Simmons US. Patent 2885923 (May 12 1959). J. Fahrenfort Spectrochim. Acta 1961 17 698. N. J. Harrick Phys. Reu. Letters 1960 4; 224. N. J. Harrick Ann. N. Y. Acad. Sci. 1963 101 928. P A. Fluornoy J. Phys. Chem. 1963,39 3156. W. N. Hansen Anal. Chem. 1963,35,765. ' W. N. Hansen T. Kuwana and R. A. Osteryoung Anal. Chem. 1966,38,1810.* N. J. Harrick Internal Reflection Spectroscopy (Interscience Publishers Inc. New York. 1967). M. Born and E. Wolf Principles of Optics 3rd ed. (Pergamon Press Oxford England 1965). lo P. Drude The Theory of Optics (Longmans Green New York 1920). 172 INTERNAL REFLECTION SPECTROSCOPY OF ADSORBED LAYERS J. R. Reitz and F. J. Milford Foundations of Electromagnetic Theory (Addison-Wesley Publish- ing Co. Reading Massachusetts 1967). l2 W. N. Hansen J. Opt. SOC. Amer. 1968,58 380. I3 L. H. Sharpe Proc. Chem. SOC. 1961,461. l4 L. H. Sharpe Bell. Lab. Record 1962,40,62. J. S . Mattson H. B. Mark Jr. and W. J. Weber Jr. Anal Chem. 1969,41 355. l6 P. A. Fluornoy and W. J. SchaRers Spectrochim. Acta 1966,22 5 . I' P. A. Fluornoy Spectrochim. Acta 1966,22 5. E. Rabinowitch and L. P.Epstein J . Arner. Chem. Soc. 1941,63 69. K. Bergrnann and C. T. O'Konski J. Phys. Chem. 1963,67,2169. 2o G. S. Levinson W. T. Simpson and W. Curtis J. Amer. Chem. SOC. 1957,79,4314. 21 R. Daudel R. Lefebure and C. Moser Quantum Chemistry (Interscience Publishers Inc. New York 1959). 22 J. S. Mattson H. B. Mark Jr. and W. J. Weber Jr. Proc. 20th Annual Symp. Spectroscopy (Plenum Press New York 1970) in press. 23 A. Muller Proc. Roy. SOC. A 1927,114,542. 24 A. Muller Proc. Roy. SOC. A 1928,120,437. 25 J. S . Mattson Doctoral Thesis (The University of Michigan 1970). 26 J S Mattson L Lee H B Mark Jr. and W. J. Weber Jr. J. Colloid Interface Sci. 1970 27 J. S. Mattson H. B. Mark Jr. M. D. Malbin W. J. Weber Jr. and J. C. Crittenden J. Colloid 28 J. S. Mattson and H. B. Mark Jr. J. Colloid Interface Sci.1969,31 131. 29 V. A. Garten D E Weiss and J. B. Willis Austral. J. Chem. 1957 10 295. 30 J. K. Brown J. Chem. Soc. 1955,744. 31 R. A. Friedel and J. A. Queiser Anal. Chem. 1956,28,22. 32 N. J. Harrick and N. H. Riederman Spectrochim. Acta 1965,21,2135. 33 J. S. Mattson H. B. Mark Jr. M. C. MacDonald Jr and C. E. Schutt Interaction of the 34 E. S. Gould Mechanism a d Structure in Organic Chemistry (Holt Rinehart and Winston 35 R. S. Drago R. J. Niedzielski and R. L. Middaugh J. Amer. Chem. SOC.. 1964,86,1694. 36 R. W. Coughlin and F. S. Ezra Environ. Sci. Tech. 1968,1,291. 37 M. J. D. Low and K. Matsushita J. Phys. Chem. 1969,73,908. 38 H. B. Mark Jr. and B. S. Pons Anal. Chem. 1966,38,119. 39 A. S. Prostak Doctoral Thesis (The University of Michigan 1969). 40 W.N. h e n R. A. Osteryoung and T. Kuwana J. Amer. Chem. SOC. 1966,88 1062. 41 N. Winograd and T. Kuwana J. Electrounal. Chem. 1969,23,333. 42 A. Prostak H. B. Mark Jr. and W. N. Hansen J. Phys. Chem. 1968,72,2576. 43 0. S. Heavens Optical Properties of Thin Solid Films (Dover Publications Inc. New York 44 B. S. Pons J. S. Mattson L. 0. Winstrom and H. B. Mark Jr. Anal. Chem. 1967,39,685. 45 W. N. Hansen ISA Trans. 1965,4,263. 46 A. Prostak and W. N. Hansen Phys. Rev. 1976,160,600. 47 J. S. Mattson H. C. MacDonald Jr. and H. B. Mark Jr. private communcation. 48 J. R. Dyer Applications of Absorption Spectroscopy of Organic Comporrndr (Prentice-Hall Inc. 49 A. S. Prostak and H. B. Mark Jr. private communication. 50 W. W. Wendlandt and H. G. Hecht Refictance Spectroscopy (Interscience Publishers Inc.51 M. J. D. Low J. Chem. Ed. 1970,47 A163. 52 M. J. D. Low J. Chem. Ed. 1970,47 A255 53 M. J. D. Low J. Chem. Ed. 1970,47 A415. s4 M. J. D. Low Anal. Chem. 1969,41,97A. s5 L. Mertz Transformations in Optics (John Wiley & Sons Inc. New York 1965). 56 J. W. Strojek G. A. Gruver and T. Kuwana Anal. Chem. 1969,41,482. 57 J. S. Mattson H. B. Mark Jr. C. E. Schutt and A. Prostak Environ. Sci. Technol. in press. 58 M. J. D. Low Spectrochim. Acta 1966,22,396. s9 W. J. Hurley J. Chem. Ed. 1966,43,236. 6o G. A. Vanasse and H. Sakai Progress in Optics vol. VI E. Wolf ed. (John Wiley & Sons 61 Applications Manual Digilab (Division of Block Engineering Iac. 1969). 33,284. Interface Sci. 1969 31 116. Evanescent Field of IRS with Particles (in preparation. New York 1959). 1965). Englewood Cliffs N.J. 1965). New York 1966). Inc. New York 1967).

ISSN:0430-0696    

DOI:10.1039/SF9700400157

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

GENERAL DISCUSSION Dr. R. Memming (Hamburg) said I would ask Plieth why does the maximum of the relative reflectivity in fig. 6 occur at 570 mp and not at the absorption maximum of methylene blue (600 for dimer and 660 for monomer) ? If he interprets his results as a shift of the absorption curve due to adsorption then one should also expect relative reflectivity values of opposite sign at longer wavelengths. According to his measurements however a change of sign was not observed. Dr. W. J. Plieth (Free University Berlin) said In reply to Memming the electro- modulation technique is used to measure the changes in the reflectivity AR of an electrode occurring along with changes in the electrode potential A&. To answer his question one has to summarize the different possibilities of influencing the reflec- tivity by the potential.A potential change can influence the reflectivity of the electrode RM in eqn (2) by increasing or decreasing the number of electrons on the surface. It can alter the value of Tad by altering the absorption spectrum of the adsorbed molecules by a change in the structure and charge of the adsorption layer or by a potential-dependent adsorption or desorption. Finally a potential change can result in a reduction and an oxidation in the diffuse layer by altering the adsorption in the homogeneous region before the electrode. Now one can compare the possibilities with the experimental results. The overall potential change in the modulation was 50 mV. Changes in RM for A& = 50mV are much smaller than the observed effects. A reduction/oxidation in the diffuse layer can be excluded on the basis of the following arguments.No reduction nor oxidation currents are observed at the peak potentials. The dependence of the height of the - 150 mV peak on the modulation frequency v should be proportional to Jv (if a diffusion controlled process is assumed) and greater effects at v = 1000 Hz should be observed. Finally the wavelength-dependence of the AR-values should show maxima identical to those in the transmission spectrum of the Methylene Blue solution. Therefore the observed effects especially for the -15OmV peak can only be explained by changes in Tad due to a change in the structure and charge of the adsorption layer or due to a potential-dependent desorption adsorption. Then the wavelength dependence of AR is the difference between the spectrum of the adsorbed Methylene Blue molecules (perhaps in a dimer form) and the spectrum of the adsorbed or desorbed molecules formed in the cathodic potential direction.No modulation effects are observed for wavelengths over 600 nm. This can be explained by the considerable decrease in sensitivity in this region. In this wave- length range absorption in the homogeneous electrolyte increases and the sensitivity of the multiplier decreases. The result is a decrease in the signal-to-noise ratio no longer allowing the detection of an AR-signal. Dr. R. Parsons (University of Bristol) said I would ask Memming and Mollers what is the limit of life-time at which a semi-quinone can be detected by their method? Can the rate constants of the elementary steps in the redox reactions be determined? Also can they determine the dissociation constants of the semi-quinone and to what extent are parameters of this type modified by adsorption of the reactant ? I wonder if this method can compete with photochemical methods for the determination of such parameters.17 3 174 GENERAL DISCUSSION Dr. R. Memming (Hamburg) said In reply to Parsons the lower limit of life-time at which a semi-quinone can be detected is determined by two factors (i) It is useless to increase the modulation frequency above few kHz because of the RC-value of the system. (ii) The lowest detectable value of A l / l o is limited by the free carrier adsorp- tion within the SnO which is also modulated (in our case (AI/Io)min- Accord- ing to these conditions the lowest detectable life-time is roughly It is in principle possible to determine the rate constants of the elementary steps using a memory display.We have such an apparatus and further measurements are under way. Photochemical methods as e.g. photo-flash measurements are more sensitive (by a factor of 10-100) as far as the life is concerned. In electrochemical reactions however quite often adsorbed species are involved. In such a case the modulation spectroscopy may be more applicable. According to our results the properties of the semi-quinone seem to be somewhat different in the adsorbed state in particular its life is larger than in solutions. We are attempting to prove this by performing photochemical measurements in a multiple reflection system. s. Dr. R. Memming (Hamburg) said In the absorption spectrum (fig.3) which Mark and Randall obtained for a methylene blue monolayer by internal reflection spectroscopy three peaks are visible at 660 for the monomer and at 600 and 740 mp for the dimer. According to the theory two transitions are expected for a dimer but the transition of lower energy is forbidden. Do they interpret the maximum at 740 mp to be such a transition? As far as I know such a distinct long wavelength maximum has not been found in absorption measurements of homogeneous dye solutions. Prof. B. E. Conway (University of Ottawa) said I would mention an interesting example of the optical observation of an organic intermediate chemisorbed on Pt. Tn recent work with MacDougall and Kozlowska in our laboratory we have studied - .8 -. 6 ” .4 - .2 .- .2 .4 .6 FIG. 1 .-Cyclic voltammetry curve for oxidation and reduction of electrochemisorbed CH3CN species at Pt in aq.1 N H2S04 ; 25°C. Oxide formation and reduction is suppressed and main peak is due to CH3CN oxidation and reduction. - 1 N aq. H2S04 ; - - - 1 N aq. H2S04+ M CH3CN. the electrochemical behaviour of acetonitrile at Pt over a wide range of concentrations in aqueous solution. At Pt a strongly bound chemisorbed species is formed which is reversibly oxidized and reduced in the double-layer region (0.4-0.65 V EH). At 0.4 V the species is initially electrochemisorbed from solution with a cathodic transient GENERAL DISCUSSION 175 current (contrast methanol which electrochemisorbs with an anodic transient current due to dissociative adsorption and oxidation of H atoms). Oxide formation is suppressed and cathodic and anodic reduction and oxidation peaks are observed in the double- layer region.The peak currents are excellently linear in sweep-rate indicating that the oxidation involves a surface species. Only partial blocking of H chemisorption arises even at appreciable " coverage " by CH3CN. Fig. 2 shows the corresponding optical behaviour in terms of dA transients. The oxide blocking is discerned together with an appreciable change of A in the double-layer region in both the cathodic and anodic directions of sweep. Further studies on this adsorbed intermediate are in progress by measurements of relative reflectance changes over a range of wave-lengths into the u.-v. V (N.H.E.) Cyclic voltammetry curves are shown in fig. 1. + 1.3 +0.5 9 = 70.75"; A = 6328 A FIG.2.-Ellipsometric behaviour in terms of changes of A in electrochemical oxidation and reduction of chemisorbed acetonitrile corresponding to fig. 1. . Prof. W. N. Hansen (Utah State University) said Concerning the relative intensities of the I and 11 ATR spectra of methylene blue as shown in fig. 3 to predict (AAll /AAL) for an isotropic layer it is necessary to take into account the fact that ( ~ 9 ~ ) ~ is con- tinuous across a boundary where (E2)1l is not. In fact for the normal component of the field (E2)Ilz it is (n2+k2)2(E2)11z that is continuous. This will decrease the expected intensity of the 11 spectrum. Near the critical angle the theoretical ratio (AAlllAA.) can be readily calculated using eqn (7) and (8) of Paper 2 of this sym- posium. Taking n1 = 1.5 for the incident glass phase n3 = 1 for air and n2 = 1.5 k2 = 0.5 as a rough guess for the methylene blue film in its intense region 1 S 2 n:nS - (z)oc = (n; + kg)2 - (1.52 +0.52)2 = 0'36* There will be some ( E 2 ) ~ ~ component which will increase this ratio somewhat.The relative intensities of fig. 3 may not be far from those expected for an isotropic layer. Dr. Memming (Hamburg) said In the absorption spectrum (fig. 3) of Mark and Randall's paper which you obtained for a methylene blue monolayer by internal reflection spectroscopy three peaks are visible at 660 nm for the monomer and at 600 and 740 nm for the dimer. According to the theory two transitions are expected for a dimer but the transition of lower energy is forbidden. Do they interprete the 176 GENERAL DISCUSSION maximum at 740 nm to be such a transition? As far as I know such a very distinct long wavelength-maximum has never been found in absorption measurements of homogeneous dye solutions. Prof. H. B. Mark and Prof. E. N. Randall (University of Cincinnati) said In reply to Memming we can only state that the cited peak at 740 nm is real as it appeared in all of the methylene blue ATR spectra that we studied. However the assignment of this peak is open and deserves further attention.

ISSN:0430-0696    

DOI:10.1039/SF9700400173

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Mechanism of Film Growth and Passivation of Iron as Indicated by Transient Ellipsometry BY J. O’M. Bockris M. Genshaw * and V. Brusic -f Electrochemistry Laboratory Dept. of Chemistry University of Pennsylvania Philadelphia Pennsylvania Received 18th November 1970 A method has been devised which utilizes intensity measurements at different offset angles of the analyser to obtain A and $J. The method is valid for examining changes in A and (J so long as (a) the Jr changes are small compared with the A changes ; (b) the electrode surface can be obtained in a sufficiently reproducible state for three successive measurements under differing optical conditions. In examinations of the growth of thin films on metals in the electrochemical situation the rate of dissolution of the metal in the early stage of film formation may be much greater than that of film formation.The surface of the electrode may roughen and this may give rise to significant changes of measured A which are then misinterpreted in terms of iilm formation. A model for this situation is proposed. If the metal surface has sites in which the dissolution rate e c x 10 times that for the rest of the surface and if their concentration is high (e-g. 1OI2 cm-2) the model predicts significant changes of A due to roughening could arise from dissolution of the metal for minutes at current densities in the micro-ampere region. Qualitative evidence is aiso in favour of this hypothesis. An initial phase oxide probably Fe(OH)* grows before the current-potential peak in the passivation of iron (pH-8.4; borate buffer).It grows initially by a 1-d mechanism followed by a mechanism involving place exchange of Fe and 0. The phase oxide thickness at the current-potential peak is about 1 monolayer. At the current-potential peak Fe(OH)2 is converted to a Fez03. The con- version continues without significant change in oxide thickness for 0.2-0.4 V +ve to the peak. The cause of passivation is the sealing off of the Fe surface by the phase-oxide Fe(OH)2. The “passive layer ” which grows at higher anodic potentials is Fe203. The present work is not consistent with a dissolution-precipitation mechanism for the mechanism of the initial growth of the phase-oxide because changes in A are observed from 0.01 s. It is also not consistent with the importance of adsorbed 0 as the principal observable entity on the surface before the current peak.In spite of the importance of blocking of the half-crystal position in growth the coincidence of the attainment of the monolayer of the phase-oxide along with the peak of the current-potential relation is inconsistent with initial kink site blocking as a major passivating mechanism. The formation of oxide and other films on metals has been often studied by electro- chemical means e.g. by coulometry. One type of oxide film has a special interest viz. that formed in the so-called “ Wagner passivity ”. It is characterized by a thickness of 1-20 monolayers by its great protective power to the substrate and by a particularly characteristic course of the (current potential) curve observed when after a certain rate of dissolution of a metal is reached passivity sets in and the current is reduced by several orders of magnitude.Research on the structure of the films present on metals at potentials more anodic than the “ passivation potential” has been extensive and there is some agreement concerning its chemical constitution. Studies on the electrochemical kinetics of the formation and the relations between rate and the structure which provokes the formation of a film of singular properties are less advanced. Electrochemical * present address Ames Co. Div. Miles. Lab. Elkhart Indiana. t present address IBM Inc. Thorn. J. Watson Res. Centre Yorktown Heights N.Y. 177 178 MECHANISM OF FILM GROWTH BY TRANSIENT ELLIPSOMETRY measurements alone may not yield sufficient information. Ellipsometry with possibility of providing values of sub- and mono-layer thicknesses refractive index absorption coefficient seems appropriate for investigations of passivity.Thus the light used may be in the visible range so that passage through the solution will not lead to absorption. However classical ellipsometry is not directly applicable to studies of thc passi- vating potential region. Thus measurements should record change in about 0.01 - 0.1 s. Further the great sensitivity in the phase-shift parameter A means that one must be cautious since local variations in the position of the metal surface (in the 1-10 A region) accompanying the dissolution of the metal before the passive film is formed can be confused with film formatioa8 In the present paper solutions of these difficulties are proposed and electro- chemical and (simultaneous) ellipsometric measurements are reported which lead to a formulation of the film growth and passivation mechanisms.INTENSITY TRANSIENTS IN ELLIPSOMETRY It is difficult to adjust to extinction the prisms of an ellipsometer within 0.01 s. One qualitative method first used by Bockris Reddy and Devanathan,' is to record the intensity of the light at a given prism offset during film formation and growth. Brusic Genshaw and Cahan made this method quantitative basing their work on the equations of Archer (1) where A and P are arbitrary settings of the analyzer and polarizer and A. and Po are the corresponding extinction settings. It follows that and If the instrument settings P and A are held constant then the intensity varies due to changes in Po and A . caused by film growth etc.If (P-PO)inilial = 45" ; (A-Ao)initial = 0 and small changes in @ are expected then (3) shows that the sensitivity to changes in A (i.e. P) is maximal. The linearity is also greatest. $ does change but the error introduced by the assumption that it does not can be calculated. who showed that Z = Zm,,(sin2 ( A -Ao)+ sin 2Ao sin 2A sin2 (P-Po)] dI/dA = Zmax(sin 2(A-Ao)+2 sin 2Ao cos 2A sin2 (P-Po)) &/dP = Imax(Sin 2Ao sin 2A sin 2(P-Po)). (2) (3) Now if A = 0" eqn (1) becomes I00 = I,, sin2(-A,) = I,,,, sin2 Ao and if A = go" hence 1900 = Lax(COS2 A019 Io/1900 = tan2 A . = tan2 $. ($ = -A0 = A:; A = 90°-2P6. Hence two transients at A = 0" and 90" give rc/ as a function of time.' When dZ/dA = 0 eqn (2) gives tan 2AODtimum = cos 2(P-P0) tan 2A,. Thus the optimum value of A is a function of the offset-angle P-Po for a given Ao.The optimal initial offsets can only be found experimentally. Fig. 1 shows A,, as J . O'M. BOCKRIS M. GENSHAW AND V. BRUSIC 179 Errors Errors in A caused by changes in It is about 1 % i.e. 0.1" in A. a function of P-Po. in d$ determination are ca. 0.02" i.e. 2 4 %. $ can be calculated from the above equations. The optimum is (P-Po) 2 28' and A N 22.5 for Fe.* ( P - 4 FIG. 1 .-( X ) The dependence of the optimum setting of A on the offset angle (P-Po) ; (0) the variation of sensitivity at the optimum setting of A with the offset angle. 400 200 C > -.20c E -400 -6OC -8OC -!OOC t 6) FIG. 2.-Variation of A + and potential with time for the passivation of iron at 4 mA/cm2. A Rudolph ellipsometer was used with a quarter-wave plate at 45" positioned before the reflecting surface.A 2 W tungsten arc point-source lamp was used. An interference filter (bandwidth 50 A) gives light at 5460 a ; nsoln = 1.338. The angle of incidence was 68.05'. A Motorola MRD 300 photo-transistor and operational amplifier was used as a detector. Intensity and potential transients were recorded on a Sanborn 322 instrument. A typical result is given in fig. 2. Thus a hand-designed ellipsometer can easily be connected to a transient instrument. However it is only feasible to use it if d A B d ~ and where the surface is highly reproducible so that the 3 transients with A at 0,22 and 90" can be assumed to relate to the same surface. * The curves of fig. 1 refer to a given experimental extinction value of A i.e. the one shown refers to Fe.180 MECHANISM OF FILM GROWTH BY TRANSIENT ELLIPSOMETRY EXPERIMENTAL POSSIBLE EFFECTS OF ROUGHENING OF SUBSTRATES ON ELLIPSOMETRIC RESULTS A three-electrode cell used was air-tight had a Teflon and KELF body and quartz windows. The counter electrode a Pd foil was separated from the main compartment by a frit ; the working electrode was a polycrystalline Fe rod polished before being assembled in the cell and positioned on the ellipsometric table. The solution was boric acid and borax pH 8.5 prepared and purified by pre-electrolysis in an N2 atmosphere for 17-20 h before being introduced into the cell. The optical reference state was obtained by reduction at - 740 mV against the N.H.E. A mercury switch enabled the fast application of a preset potential or current. Potentiostatic oxidation at varying potentials between - 740 and + 900 mV was applied from the reference state (single-step oxidation).After the steady state had been reached. the electrode was reduced. Alternatively an oxidizing potential was applied in a series of steps of about 50-100 mV and only decreased at the end of the experiment. Double-layer measurements were made galvanostatically. Those used in the pre-film region were carried out at -550 mV (NHE scale) to avoid pseudo-capacitance effects. RESULTS The optical reference state was pure i.e. on the basis that the nM = 3.24 and kM = 3.98 obtained from ellipsometry on the pure metal are in excellent agreement with the results of other authors who have prepared pure surfaces. -10 -16 -21 y)I1 (+660) -22 (4060) -2. -1 0 dV FIG. 3.-The (A 4) relations obtained in various types of experiments (a) on poorly polished electrodes potentiostatic oxidation steps increasing by 50 mV ; (b) as (a) but only carried out in passive region (negligible dissolution of metal) ; (c) well-polished single-step experiments (i.e.electrode not repeatedly dissolved). J . O'M. BOCKRIS N. GENSHAW AND V. BRUSIC 181 In single step oxidation after reduction these parameters were re-obtained i.e. the film is completely reducible. In the stair-case approach the parameters were not re-attained after reduction. The effect increased with stirring and with decrease of initial polishing. Fig. 3 shows the great difference between (A,$) relations obtained in the stair-case manner those carried out in the passive region (no roughening) and those obtained by the single step method.There is a radical difference in the shape of the (A,$) plot when roughening occurs. The final value after reduction differs greatly from that observed in situations where roughening is greatly reduced. - - - - - 140 120 100 -80 -60 'E; n -.. 3- 40 20 I I I I I -700 - 600 - 500 - 400 potential (mV) against NHE FIG. 4.-Capacity (hence relative surface area) against current density for an electrode oxidized and then reduced. After the reduction follows an oxidation ; dissolution made in the active region ; the capacity is increased compared to the value before oxidation. Fig. 4 shows that the real surface area changes little during oxidation-reduction But if the oxidation is carried out in the active region a large increase in cycles. capacitance (area) measured after film reduction occurred.DISCUSSION An explanation of the present results can be given in terms of derma-sorption. But it is then unclear why the effect should increase with a degree of stirring which would increase the rate of metal dissolution. Consider fig. 5 as a model of the dissolving situation. With the assumption of the additivity of polarizations the Clausius-Mossotti relation is * 2 * 2 (4) 4 n -nsoln n:'+ 2 =' n,*2+2 * Here n,* is the complex refractive index observed according to the model for which the (A,$) changes observed in the stair-case situations were due to roughening; nsop is the real refractive index of the solution and nt is the complex refractive index of the metal (see fig. 5). Thus,13 as q changes nz changes. We now assume that irregular dissolution occurs over the surface i.e.faceting takes place. Alter- natively and additionally the mechanical polishing probably leads to local de- formation and hence an irregular rate of dissolution. Thus even on single crystals an irregular dissolution rate may occur near steps and irregularities particularly 182 under conditions of limiting self-diffusion. In order to test the consistency of this concept of the origin of a q and hence an nt (and therefore A and +) variation independently of oxide film growth we assume (cf. ref. (6)) that (iDfss) at a constant MECHANISM OF FILM GROWTH B Y TRANSIENT ELLIPSOMETRY .d optical smooth surface .-- optical rough surface a ) a') 1 b ' ) . I . - - I I v --I L - - - - FIG. 5.-Dissolution of areas of locally-higher activity causes roughening (perhaps from critical facet having h.i.p.).It is assumed that there is dissolution along the entire surface. overpotential for a high index is greater than that for a low index plane.6 It is consistent with the evidence to equate iDiss h . i . p . = 10 iDiss It is further assumed that the h.i.p. are randomly distributed and that there are 10 l 2 " more active " sites cm-2 (e.g. crystallites containing high index planes) the dissolution takes place at time (min) FIG. 6.-Calculated change of n k and L with time of dissolution according to model (b) of fig. 5. 10 pA cm-2 and entirely traverses these h.i.p. At a given current density [lo x is the dissolution rate. 60/2 x 105](60/8) = 2 x lo-* cm3 min-' Hence the rate of growth of each pit is (2x = J . O'M. BOCKRIS M.GENSHAW AND V. BRUSIC 183 cm3 min-' = vo. Thus after t min the volume is vat or the depth (uot)+ ( 5 ) 2 x cmmin-'. Now qt = [(vat)* - luoNtl/(uot)* where N is the number of crystallites cm-2 of high index plane or qt = 1 -"vat%. Substitution of these values into (4) gives n and using this value allows one to calculate the corresponding values of A and Y. The results are shown in fig. 6 7 and (8). The magnitudes derived show that there may be a masking of the changes due to film formation. 2c 115 I IC I05 I / A .a I t / -9 I I I 10 20 30 L 29 ?0 c" s 27 time (min) FIG. 7.-Change of A and 4 with time of dissolution model (b). EXPERIMENTAL GROWTH OF A N INITIAL FILM BEFORE PASSIVATION The same manual and transient ellipsometry was used. Continuous recording of the intensity transients from 0.01 s was used.nFe and kFe were known. The electrode and cell were as before; the 99.998 % polycrystalline Fe was polished. The solution and optical reference state were as before. Potentiostatic oxidation and/or reduction was carried out by a fast single-step application of a pre-set desired potential ; A and t,b and Q (mC/cm2) were recorded. After reaching a steady state A and t,b were also determined manually and the electrode was reduced. In the galvanostatic reduction a Hg-relay enabled the potential to be disconnected and a pre-set current to be applied. The oxidation current was varied from 8 ~ A c m - ~ to 4 mA cmW2. The reduction current was 30 pA cm-2. A and t,b were continually recorded as were the millicoulombs passed through the systems. 184 MECHANISM OF FILM GROWTH BY TRANSIENT ELLIPSOMETRY 25 20 I I5 I IC 105 1 I I 6 27 28 29 +<"I FIG.8 . 4 6 +) for progressive dissolution model (b); cf. the experimental results of fig. 3. (Parenthesis gives time in min and thickness in A). FIG. potential mV against NHE 9.Variation of A and JI with potential in steady-state and transient ( x ,0.8 ; 0,0.32 ; rnA cm-*) oxidation ; and (i V)relation in steady-state oxidation. A 0.08 J . O’M. BOCKRIS M. GENSHAW AND V. BRUSIC 185 RESULTS Fig. 9 shows A and $ values obtained in two ways. One set was obtained potentiostatically (with return to the optical reference state i.e. complete reduction) before a further experiment was made at a new potential. On the same figure A and $ values obtained in the galvanostatic transient mode are given.A changes immediately as the current passes. ~ changes only after the peak. 1 .o .a r - 4 5 .6 4 .4 .2 time (s) FIG. 10.-Variation of L(A) ; i (mA cm-2) and Q (mC m r 2 ) during the initial state of oxidation. In the (Ay t,b) relations for the potentiostatic steady-state values three regions of slope exist. In the kinetic measurements (fig. 10 1 l) the data refer to the first region where the film properties are evidently constant and hence A may be taken as pro- portional to thickness. One finds that (L)v = At (O.O<L< -0.1 s) (L) = B+Clog t (O.O5<L<-12OO s). These relations are shown in fig. 10 and 11. In fig. 12 galvanostatic transient el!ipsometry is illustrated. A changes can be detected after 0.01 s. log t (s) FIG. 11 .-(L, log t ) variation during potentiostatic oxidation at 0 - 540 ; X -490 in unstirred and at - 540 mV in stirred solution D .186 MECHANISM OF FILM GROWTH BY TRANSIENT ELLIPSOMETRY DISCUSSION It is now possible to deduce the nature of the pre-passive layer. Thus (i) in the past peak regions (more positive than about -480 mv) the (A II/) plots (potentio- static steady-state various potentials) show two slopes (fig. 3); the (A t ) galvano- static transients (fig. 12) shows also two slopes. During reduction these two slopes 7 c -“OI I 1 I I I I I I ] 1 2 3 4 5 time (s) FIG. 12.Variation of A + and Y with time of galvanostatic oxidation in unstirred solution (0.8 nQ m-2). are evident in the millicoulomb-time plot and can be converted into equivalent coulombs. The ratio of coulombs in the upper region is 1 :2. Hence in the passive region ferric oxide is being reduced.(ii) Corresponding plots in the pre-passive region show only one slope henceferrous oxide is being reduced. (iii) The A changes are far greater than those observed (Paik Gensbaw and Bockris 9 in ellipsometric measurements of adsorption. Thus in the pre-passive region a ferrous phase-oxide is being formed. It must therefore be FeO or Fe(OH),. TABLE EVALUATION OF THE FILM THICKNESS AT E = E’ layers of layers of la ers of HOFeOH formed roughness mol Fez+ /cmz Fe(OH)2 formed HOF& forpled if each Fo atom in‘ factor expt. Q ( p = 3.41) (ionic radn) metal ream 1 3.63 to 6.2 2.1 to 3.6 1.6 to 2.8 1.3 to 2.25 1.2 3.25 to 5.1 1.7 to 2.9 1.2 to 1.9 1 to 1.6 1.4 2.6 to 4.5 1.5 to 2.5 0.9 to 1.5 0.7 to 1.2 experimentally determined 0.7-1.2 mC/cm2 The thickness of the film at the peak can be evaluated from cathodic coulometry.There is a corresponding deposition of hydrogen for which one has to make a correc- tion. Knowing this (<0.5 mC cm-2) one m y obtain roughness factors (deter- mined by galvanostatic transients (see fig. 4)). However various assumptions are possible. Should the density of this film be a bulk value? Should one accept lattice dimensions or calculate the thickness on the basis of ionic radii ? Table 1 shows the J . O’M. BOCKRIS M. GENSHAW AND V. BRUSIC 187 result of various assumptions. The most probable result is that of the last column with roughness factor 1.4 i.e. there is present about a monolayer of a phase-oxide at the peak. In respect to the mechanism of growth of the phase-oxide there is growth from 0.01 s.and no effect of stirring. Hence no dissolution-precipitation model is possible. Further in the initial stage (i.e. up to ca. 0.1 s cf. fig. 12) dL/dt = con- stant. Hence one can reject mechanisms of the type because they imply a decrease of dL/dt with time at constant potential. The need for a model consistent with dL/dt = constant suggests a 1-d imensional growth of discrete centres. Instantaneous initial nucleation and growth outwards are unlikely. Thus one may picture the initial growth mechanism as being along the edges of screw dislocations. The edge of the step would move around the point at which the dislocation emerges. The observed thickness would be proportional to the length of the emergent dislocation and the total number of these active regions.After one rotation of the vector the surface on which further rotation could occur is now an oxide. There will be an increased difficulty in the supply of Fe2+ to react with OH- and form a further rotating edge. Let there be lo1 l-10l2 dislocations cm-2 (heavily mechanically polished surface). Only a small fraction is likely to be active at low overpotential.1° Suppose that this fraction is 1 % and that the average length of an active dislocation is cm. The area of rotation is then 7 ~ ( l O - ~ ) ~ and the total volume cm-2 at the end of the growth is lo9 h (10-5)2 = using h = 3.3 A (unit cell of 3.3~4). The area covered is Nnr2 = 30 % at the time of change of mechanism (i.e. after 0.1 s. when ciL/dt changes slope). In the mechanism of growth after 0.1 s the two salient experimental characteristics are (i) at constant potential (cf.fig. 11) (L& = A +B log t ; (ii) (dL/dt) = f( V ) at constant total i (cf. fig. 9 11). One also notes that calculations of the expected rate of dissolution of Fe at the relevant potentials and time show that the film must be growing laterally and that the growth rate is much less than that directly calculated from the current i.e. most of the current is used for iron disolution. Place exchange of 0 with the surface as an initial mechanism for oxide growth on metals was suggested by Lanyon and Trapnell. However when simply applied it is not consistent with the results obtained here. We therefore propose l2 Fe + OH-$FeOHAd + e- (6) (7) HOAds - Fe *Fe- OH(be1ow surface) = Fe(-OH) with possible simultaneous dissolution by the reaction FeOHAds+FeOHd+isWlved + e- Fe( - OH) + OH- + HO-Fe(-OH) + e- r.d.s.HO-Fe(-OH) +Fe(OH) (phase-oxide) in the r.d.s. under Tempkin conditions. Hence From (7) defihnldt = k3COH-@Fe(-OH) exp [-k!f(etota,/RT] [exp(BFv/RT)I eFe(-OH) = k2f(°FeOH. Ads)* 188 MECHANISM OF FILM GROWTH BY TRANSIENT ELLIPSOMETRY From (6)y eFeOHAds = klCOH-(1- $total) exp [-f(eT)/RTl exp [Fv/RT1* (12) (13) (11) and (12) in (10) give where Ototai = Or = eFeOH + eFe(OH)2 phase. &eOH is used for Fe dissolution and hence is assumed to be < 1 consequently dO,,,,/dt = k,k,k,c&-(1- OT) exp [ - 3pf(B,)/RT] exp [3pVF/RT] eTheFdOH)2. phase' Neglecting the pre-exponential in the Tempkin region then dO,/dt = k4 exp (3FVIRT) exp (-+f(B,)/RT). (14) Iff(&) = re integration gives (aO/a log t)" = 2.3 x 2RT/3r or (aO/aV) = F/r.One finds that (OFe<OH)2 = 1 if L = 4 . 5 4 with a roughness factor of 1 and r = 3.3 kcal mol-' ; dL/dV = 32 .$/V compared with the experimental value of 48 A/V. An easy place exchange (cf. eqn 7) would not be expected after the first 1-2 layers. EXPERIMENTAL MECHANISM OF PASSIVATION Galvanostatic and potentiostatic coulometry galvanostatic transient ellipsometry and potentiostatic (steady-state) ellipsometry were used. Corresponding measurements were made during reduction. RESULTS In fig. 12 an example of simultaneous ellipsometry and coulometry is given. The various divisions in the (A time) relations during reduction correspond to those for coulometry. DISCUSSION The film formed before the current peak is a ferrous film and after the peak it becomes increasingly ferric.This makes it difficult to obtain L k and n from the ellipsometric data. Thus in introducing cou€ometry into this calculation it is necessary to make an assumption concerning the nature of the oxide present other- wise one cannot obtain the thickness. The evidence from coulometry is that at sufficiently +ve potentials the film is Fe,O,. In the pre-peak region it is likely to be Fe(OH)2. In fig. 14 ellipsometric data thus evaluated are plotted as solid lines. In the change-over region the oxides are assumed to be homogeneously mixed. The thickness seems to remain roughly constant over 300-400 mV region this may be regarded as the region in which the film is being oxidized to a higher state. k values pass through a maximum near the peak but there are no spectroscopic data for Fe(OH)2 or Fe20 in the relevant wave length.One might interpret the k rise as reflecting an increase in electronic conductance (Fe(OH)2+Fe20 = Fe304 + H20). J . O'M. BOCKRIS M. GENSHAW AND V. BRUSIC 189 Before oxidation Oxidation at+860mV Reduction at-740 mV 10.4 5-2 2.6 FIG. 13.-Simultaneous coulometry and transient ellipsometry. - 7'8--Q (rnC)- -(surface area - 1.25 cm2) The refractive indexes of fig. 14 show limiting values of about 1.4 and 2.7 for the pre-peak and post-peak oxides. But can such values be compared with those of bulk Fe(OH) and Fe,03? For films of > 10 A such a comparison is reasonable to within 10 % (in confirmation of Fe,O,.) For films of lesser thicknesses one can -60 0 - / //I/' /5' -A!. . .. - * * -40 &(a - I k l -600 -400 -200 0 200 400 600 potential mV against NHE FIG.14.-The solution of the ellipsometric results using coulometry and assuming Fe(OH)2 below the peak and Fe203 above it. obtain some idea of the expected deviation of the observed refractive index from the bulk value by finding the change in n brought about by a change in p from the Clausius-Massotti equation (see fig. 15). If the thin surface oxide (clumps?) differs from that of the bulk in density by ca. 20 % (an estimate) the calculated increase of n would be about 26 %. The expected value for Fe(OH)2 on this basis is 1.45 (see fig. 15). 190 MECHANISM OF FILM GROWTH BY TRANSIENT ELLIPSOMETRY FIG. 15.-The (refractive index density) plot from the Ciausius-Mossotti equation. Hence after integration Thus i.e. iDiss+O when t is large.Also as from (16) and (17) jDiss/iff = k2/k3COH- i.e. (&f)+O. Thus beyond the peak dissolution stops and growth of Fe(OH) ceases. Both coulometry and ellipsometry indicate that the growth process +ve to this peak is increasingly in terms of Fe203 first by conversion of Fe(OH)2 (cf. fig. (14)) and then by increasing of the thickness of Fe203 outwards from the electrode. Thus dissolution is blocked increasingly by the phase-oxide Fe(OH)2 growing J . O’M. BOCKRIS M. GENSHAW AND V . BRUSIC 191 in clumps which seals the surface until at the peak the current begins to decrease. Thereafter Fe,O is formed and grows to a thickness of about 40A at a potential (NHE scale) of ca. 700 mV. Earlier views,13 in which the material before the peak is adsorbed 0 are incon- sistent with the present ellipsometric data which indicates a phase oxide.If blocking of kink sites by adsorbed 02- were the mechanism one would not need a monolayer for passivation.14 The stress given by some workers on the importance of dproperties to the passivation process does not necessarily justify an assumption of adsorbed 0 as the entity present before the current potential peak. The dependence of the rate of electrode reactions producing adsorbed oxygen (which then controls the phase- oxide formation) on the d properties of transition metals has been proved.16 ’ J. O’M Bockris A. K. Reddy M. A. V. Devanathan Proc. Roy. SOC. A 1964,279,327. ’ R. J. Archer J. Opt. SOC. 1962,52,970. V. Brusic M. Genshaw and B. Cahan J. Opt. Soc. 1970,9,1634. see tables comparing data in V. Brusic Thesis (University of Pennsylvania 1970).A. Despic and J. O’M Bockris J. Chem. Phys. 1960,32,389. A. Damjanovic T. H. V. Setty and J. O’M Bockris J. Electrochem. SOC. 1966,113,429. A. E. C. Contract 7405 erg- 48 Oct. (1965). R. Tronstadt Thesis (Trondheim 1931). W. Paik M. A. Genshaw and J. O’M. Bwkris J. Phys. Chem. 1970,74,4266. ’ R. H. Muller and J. R. Mowat Report. Lawrence Radiation Lab. (University of California) ; lo H. Kita M. Enyo and J. O’M. Bockris Can. J. Chem. 1961,39,1670. l 1 A. H. Lanyon and B. M. W. Trapnell Proc. Roy. SOC. A 1955,227,387. H. Wroblowa V. Brusic and J. OM. Bockris J. Phys. Chem. in press. l3 V. Brusic M. Genshaw and J O’M. Bockris Surface Sci. in press. l4 H. H. Uhlig and Z. A. Faroulis J. Electrochem. SOC. 1964,111 13 ; F. Mansfield and H. H. l5 R. Frankenthall J. Electrochem. SOC. 1969,116,580. l6 J. O’M. Bwkris A. Damjanovic and R. Mannan J. Electroanalyt. Chem. 1968,18 349. Uhlig J. Electrochem. SOC. 1968,114,900.

ISSN:0430-0696    

DOI:10.1039/SF9700400177

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

Ellipsometry Techniques and Their Application to Polymer Adsorption ROBERT R. SIXOMBERG LESLIE E. SMITH AND FRANK L. MCCRACKIN National Bureau of Standards Institute for Materials Research Washington D.C. 20234 Received 26th October 1970 The application of ellipsometry to the study of the adsorption of polymers at interfaces is discussed. The correction of certain instrumental errors such as imperfect components and the use of solutions containing optically active materials are described. Results for the measurement of the extension of polystyrene poly(ethy1ene o-phthalate) and certain proteins present in blood adsorbed on surfaces are given and interpreted in terms of changes in molecular conformation that occur during the adsorption process. The differences observed are ascribed to the interaction energies between the different polymers and surfaces.Ellipsometry is a valuable technique for the study of adsorbed layers at interfaces. The great sensitivity of the state of polarization of the reflected light to very thin films on a surface allows measurements to be made on films of thicknesses of only a few A. The ability to make measurements of films in situ in liquid or gaseous environments makes it particularly useful for adsorption studies. In this paper we discuss certain aspects of the technique and the results of its application to an investi- gation of the adsorption of polymer molecules from solution. TECHNIQUES CORRECTION OF INSTRUMENTAL ERRORS As the technique of ellipsometry requires no calibration procedures but permits calculations to be made from first principles any errors due either to imperfect components or to improper procedures are especially important as there is no standard for comparison.In an alignment procedure previously described the compensator is removed from the ellipsometer and the polarizer and analyzer prisms are adjusted for extinction of the light reflected from a surface. Ideally extinction is only possible if the axis of transmission of the polarizer is in the plane of incidence ; otherwise the reflected light is elliptically polarized and cannot be completely extinguished by the analyzer. This fixes the polarizer in the plane of incidence and the analyzer at 90" from the plane of incidence. In practice the azimuths of the polarizer and analyzer often do not differ by exactly 90" when the light has been extinguished.This effect has been explained based upon an analysis assuming an imperfect polarizer and a perfect analyzer.' Following the same line of analysis but using Mueller matrices to facilitate computation results were obtained for the case of both imperfect polarizer and analyzer. Experimental results from alignment on various surfaces allowed the calculation of errors in the respective prisms. The calculated prism errors were different from the different surfaces indicating a fundamental error in the assumption that the effect was due to imperfect polarizing prisms. 192 R . R . STROMBERG L . E . SMITH AND F . L . MCCRACKIN 193 In recent experiments moreover we have apparently eliminated this effect by slightly tilting the interference filter used between the light source and polarizer.This implies that the effect may be due to parasitic reflections between these optical elements. By tilting the filter this unwanted reflection was deviated from the optic axis so it did not reach the detector. Even after the effect was eliminated for most alignment surfaces a particular form of pyrolytic carbon continued to show an error. This is contrary to the notion that parasitic beams are totally responsible for the effect and therefore the problem is still not completely resolved. The values of A the relative phase difference and Y the relative amplitude reduc- tion for a surface may be calculated from any set of azimuths of the polarizer analyzer and compensator that produce extinction of the light when the constants of the compensator are known. However a method using measurements made in two zones that corrects for imperfections in the compensator has been given.2 Although not previously mentioned this method also corrects for small angles of tilt of the specimen.For measurements of a surface in vacuum or under liquids the birefringence of the windows of the cell holding the sample causes errors. A method to determine a correction factor for these errors from measurements of a sample both in and out of the cell has been given.2 OPTICALLY ACTIVE SOLUTIONS The effect of circular birefringence (optical activity) in the liquid in the cell was analyzed since such solutions were used in some of the experiments that are reported in this paper. In this case the light from the polarizer passes through the wave plate through a liquid with circular birefringence 6 is reflected from the surface passes through liquid of biregringence 6' and is extinguished by the analyzer.The components are described by Mueller matrices. The matrix of the first liquid is for small birefringence * 6. in place of 6. with the compensator set at 45" are The matrix for the second liquid is the same with 6' Proceeding as in ref. (2) the conditions for extinction of the light sin 2Y cos (A+2P) = 0 (1) 6' cos 2Y + sin 2Y sin (A + 2P) - cos 2" + 6 sin 2P + 6' sin 2Y sin (A + 2P)' tan2A = where P and A are the azimuths of the polarizer and analyzer respectively zone 2 defined in ref. (2) let For 2P = -A-90 (3) A = Y+E (4) and to first order in e 6 and S' we obtain P = A -$(A+90) A = Y - $8 cos A sin 2Y - 38'. * This approximation is to first order in 6 and the entire treatment is expected to be accurate to 0.01" if the birefringence is less than 1".s4-7 194 ELLIPSOMETRY TECHNIQUES AND POLYMER ADSORPTION Similarly for zone 4 P = +(-A+90) A = -Y-@ cos A sin 2Y-+6'. When these values are combined by the method of $5 of ref. (2) the terms in 6 and 6' cancel. Therefore small circular birefringence in the liquid in the cell does not cause an error in the calculated values of A and Y when readings in the two zones with the compensator set at 45" are combined. This can be also shown to apply for the zones with the compensator set at -45". INHOMOGENEOUS FILMS While the solutions of the Drude equations yield thicknesses based upon a film of homogeneous refractive index many adsorbed films may be inhomogeneous normal to the surface.The optical effect of a film on reflected polarized light depends on the difference between the real refractive index of the film and that of the solution (n-n,). Since the effect will also depend on the thickness of the film it may be argued that the effect of a film of infinitesimal thickness dt can be expressed as (n-nJdt. The effect of a film of varying index of refraction would then be given by the integral of this quantity over the entire film. As this effect is also given by an equivalent homogeneous film of index nh and thickness th these effects are equated to give (9) By approximating the inhomogeneous film as a succession of thin homogeneous films of slightly differing refractive index McCrackin and Colson calculated A and Y values for three different distributions.They then calculated the thickness and refractive index of an equivalent homogeneous film and found good agreement between the two sides of eqn (9). This lends some credence to an admittedly heuristic derivation of eqn. (9). It was then possible to compare the thickness derived from the homogeneous film model with the root-mean-square average thickness of the inhomogeneous film defined as / (n - ns) dt = (n - n,)t,. t& = 1 a (n - n,)t2 dt/ 1 Q (n - ns) dt. 0 0 The value for th was larger than trms by a factor of 1.5 for an exponential distribution and larger by a factor of 1.75 for a linear one. APPLICATION TO POLYMER ADSORPTION The process of polymer adsorption is relatively complicated compared with small molecule adsorption. In addition to the usual competitive interactions the many conformations available to a polymer molecule on adsorption makes the determina- tion of the changes in conformation during adsorption one of the most important problems to be resolved.It is now generally accepted that a flexible macromolecule will be adsorbed as first proposed by Jenckel and RumbackY5 insequences (trains) of segments attached to a surface separated by loops of segments which extend into the solution. This would permit " monolayer " adsorption with adsorbance values far in excess of those that would be obtained for monolayer coverage by monomer molecules. Of considerable interest is a quantitative description of the arrangement of the adsorbed trains and unattached loops. R . R . STROMBERG L. E. SMITH AND F. L . MCCRACKIN 195 It is apparent that measurement of the extension of an adsorbed chain normal to the surface must be made in situ with the surface remaining immersed in the polymer solution to prevent the loops from collapsing.Ellipsometry is one of the few techniques that appears to be applicable to the problem. The segments of polymer attached to the surface and loops extending into the solution together with solvent is considered as a " film " over the surface. The quantity of polymer in this film is relatively low ; therefore the principal experimental limitation has been the small refractive index difference between this film and the solution. However by proper choice of polymer and solvent accurate ellipsometric measurements of the thickness and concentration of the film have been made. EXPERIMENTAL The measurements were carried out in glass cells with optically flat strain-free windows attached at each end so that light entered and left at normal incidence.The optical constants of the surface were determined with the surface under the appropriate solvent. The solvent was then exchanged for polymer solution without exposing the surface to air and ellipso- metric measurements were made as a function of time on the identical locations on which the optical constants of the surface had been obtained. The polymers studied include a flexible high molecular weight polymer polystyrene ; a relatively low molecular weight polyester poly(ethy1ene o-phthlalate) ; and some proteins believed to be important for the coagulation of blood. The polystyrene was prepared by anionic polymerization and further fractionated to yield samples with very low molecular weight distributions.The molecular weights of the polystyrene samples studied ranged from 26 000 to 3 300 OOO. Some of the polystyrene samples were obtained from an outside source * while other polystyrene samples and the polyester were prepared in this laboratory. The proteins were obtained commercially and used as supplied. The metal surfaces used for these adsorption studies included steel gauge blocks and chromium gold or copper that had been electrodeposited on such gauge blocks. Chrome surfaces cut from commercial ferrotype plates were also extensively used. Measurements on the polystyrene and polyester systems were made using the 5461A line from a mercury arc lamp. A He-Ne gas laser operating at 6328A was used for the protein adsorption measurements since its higher intensity permitted studies on low reflectance surfaces.The values of the thickness and refractive index for the adsorbing film are calculated from the Drude equations for a homogeneous film model with discrete boundaries. The computer programme by McCrackin was used for these calculations. RESULTS AND DISCUSSION Polystyrene was adsorbed on the chrome ferrotype surface from cyclohexane at approximately the theta point a condition at which polymer-polymer interaction is to a first approximation equal to polymer-solvent interaction. DiMarzio and McCrackin have shown that the segment distribution normal to a surface of an adsorbed molecule under theta conditions can be represented by an exponential function except for a high segment density which extends for a short distance from the surface.As described earlier root-mean-square (rms) values have been calculated from the homogeneous film thickness assuming an exponential distribution. The use of an exponential function with a small step close to the surface does not appre- ciably affect the factor used to convert the measured value for an equivalent homo- geneous film to an rms average. * We acknowledge the assistance of Dr. H. W. McCormick of the Dow Chemical Company for providing these samples. 196 ELLIPSOMETRY TECHNIQUES AND POLYMER ADSORPTION Typical results * for the rms thickness as a function of time are shown in fig. 1 for a 5.4 x lo5 molecular weight fraction of polystyrene at a given solution concentra- tion. The vertical lines give a measure of the uncertainty of the measurement for that point.The adsorbed film thickness increases with time until a final plateau time min FIG. 1.-Root-mean-squaxe extension of adsorbed polystyrene (m.w. = 5.4 x lo5) against time. Adsorbed from cyclohexane solution concentration of 1 .O mg/ml on chrome ferrotype surface at 34°C. The different symbols represent different runs and the open and closed points represent separate measurements made on the same set of slides. time min FIG. 2.-Adsorbance of polystyrene (m.w. = 1.3 x lo6) on chrome ferrotype against time. Adsorbed from cyclohexane solution concentration of 0.55 mglml. The different symbols represent different runs and the open and closed points represent separate measurements made on the same set of slides. R. R. STROMBERG L.E. SMITH AND F. L. MCCRACKIN 197 value is reached. The results are typical for all solid metallic surfaces electro- deposited chromium gold and silver chrome ferrotype and steel. The highest molecular weight sample 3.3 x lo6 showed a maximum in the (extension time) curve. The shape of the curve shown in fig. 1 was the same for the remaining mole- cular weight samples. The time required to reach an equilibrium value was dependent on the solution concentration and the molecular weight of the polymer. From the measured values of the refractive index and thickness and an indepen- dently determined value for the change in refractive index with polymer concentration dnldc the adsorbance (amount adsorbed per unit area) can be calculated. A typical curve is shown in fig. 2. The adsorbance increases with time in a manner similar to the increase of thickness with time until a final plateau value is attained.This behaviour was typical for the polystyrene for all molecular weight samples and for all solid metal surfaces studied. Within experimental error the refractive index of the adsorbing film remains constant during the adsorption period. concentration mg/cm FIG. 3.-Root-mean-square thickness and adsorbance of polystyrene (m.w. = 5.4 x lo5) against solution concentration. Adsorbed from cyclohexane at 34°C on chrome ferrotype plate. These results are interpreted to indicate that the polystyrene molecule is initially adsorbed in a relatively flattened configuration. As additional polymer molecules are adsorbed some of the attached segments of the previously adsorbed molecules desorb resulting in larger loops and increased film thickness until a plateau thickness is established.Fig. 3 shows a typical curve for the plateau value of the extension of the adsorbed polystyrene film as a function of solution concentration. For the molecular weight range studied the extension increased with increasing molecular weight. These data are interpreted to indicate that the polymer molecule is adsorbed in a relativeIy flat conformation at low solution concentrations in a manner identical to that which occurs at short adsorption times. As solution concentration increases bringing about an increased competition for adsorption sites the loop size extending into the solution increases resulting in larger film thicknesses until a plateau value is attained. Studies of the same polystyrene samples adsorbed under identical conditions on a liquid mercury surface and of a relatively low-molecular-weight polyester s4-7* I98 ELLIPSOMETRY TECHNIQUES AND POLYMER ADSORPTION adsorbed on steel and chrome surfaces lo were also carried out.In both cases no change in the thickness was observed as a function of time although adsorption occurred during the time period studied. This is shown in fig. 4 for polystyrene adsorbed on the mercury surface. In these cases the refractive index of the adsorbing films increased with time until a plateau value was attained. 100 4 500 8 5 rn 4 200 g !OO .- 0 I T T x x Y Q Y Y I I t - - - FIG. 4.-Root-mean-square extension of adsorbed polystyrene (m.w. = 1.3 x lo6) on liquid mercury against time. Adsorbed from cyclohexane solution concentration of 2.26 mg/ml at 34°C.It appears that for these conditions the polymer attains its final conformation relatively early in the adsorption period and retains this conformation as additional adsorption occurs. The conformation is relatively flat and the early arrivals do not change into more extended conformations as surface pcpulation increases. Rather " holes " in the adsorbed layer are filled with later arrivals and the concentration of polymer in the adsorbed layer increases with time until a plateau value is attained. 0;I ;5 'p 2; 3OflO6 Iooo t 800 4 cn rn ' 600 ." % s rA 400 1 * O O i 0.4 0.8 I .2 1.6 2.0~ to3 0 (m.w.)+ FIG. 5.-Root-mean-square extension of polystyrene adsorbed on chrome ferrotype and mercury surfaces against square root of molecular weight.For both polystyrene on liquid mercury and the polar polyester on the solid metals it is probable that the interaction energies involved are higher than for the polystyrene on the solid metal surfaces. It is to be expected therefore that the adsorbed molecule is more closely attached to the surface. This is shown most clearly in fig. 5 in which R. R . STROMBERG L. E. SMITH AND F. L. MCCRACKIN 199 0 I I 1 I I I\ I 4 0 60 120 180 240 300 360 420 480 540 time min FIG. 6.-Adsorbance of fibrinogen on chrome ferrotype at 37°C. Adsorbed from phosphate buffer at pH 7.4. Fibrinogen 2.5mg/ml. The two curves represent different runs. the thickness of the adsorbed layer is shown as a function of the square-root of the molecular weight. The linear behaviour finally achieving a plateau at high-molecular- weight values also has been developed theoretically by Hoeve.ll By this treatment the independence of thickness with molecular weight for polystyrene on liquid mercury is attributed to the higher interaction energies and is expected to be identical with the other curve at the lower molecular weight values.As a class the globular proteins are polar macromolecules with specific shapes in solution that are far from the random coil of the flexible non-polar polymers discussed above. Both these factors tend to make their adsorption behaviour rather different from the polystyrene. For example fig. 6 shows the adsorbance on chrome of fibrinogen a protein with a molecular weight of 3 . 4 ~ lo5 that is important in blood clotting. There is some increase with time but not the marked increase typicaliof 6oo 3 0 0 60 120 180 240 300 360 420 480 540 time min Fro.7.-Extension of fibrinogen on chrome ferrotype at 37°C. Fibrinogen 2.5 mg/ml. 200 ELLIPSOMETRY TECHNIQUES AND POLYMER ADSORPTION polystyrene of about this molecular weight. The thickness moreover shows almcst no change over this time period (fig. 7) giving no evidence of any molecular rearrange- ment in this system again in contrast to the polystyrene. The thickness is that of a homogeneous film uncorrected for any distribution. A second blood protein prothrombin with a molecular weight of 63 000 shows very similar adsorbance and extension curves with time on this chromium surface. Study of the adsorption of these blood proteins is currently being extended to various lower energy surfaces.On polyethylene for example prothrombin adsor- bance curves similar to those found on chrome are obtained. The extension from the polyethylene surface is much larger however indicating a more diffuse adsorbed layer with fewer points of attachment. This perhaps implies less distortion of the protein upon adsorption on this surface as well as easier desorption. This work was supported in part by the U.S. Army Research Office Durham. ' F. L. McCrackin E. Passaglia R. R. Stromberg and H. L. Steinberg J. Res. Nat. Bur. Stand. A 1963 67 363. F. L. McCrackin J . Opt. Sue. Amer. 1970,60 56. W. A. Shurcliff Polarized Light (Harvard Univ. Press Cambridge Mass. 1962). F. L. McCrackin and J. Colson Ellipsometry in the Measurement of Surfaces and Thin Films ed. E. Passaglia R. R. Stromberg and J. Kruger (Nat. Bur. Stand.) Misc. Publ. 256 1964 p. 61). E. Jenckel and B. Rumback 2 Efektrochem. 1951,55,612. F. L. McCrackin A Fortran Program for Analysis of Ellipsometer Measurements (N.B.S. Tech. Note 479 1969). ' E. A. DiMarzio and F. L. McCrackin J. Chem. Phys. 1965,43,539. * R. R. Stromberg D. J. Tutas and E. Passaglia J. Phys. Chem. 1965,43 539. lo P. Peyser D. J. Tutas and R. R. Stromberg J. Polymer Sci. A-1,1967,5,651. '' C. A. Hoeve J. Chem. Phys. 1965,43,3007 ; 1966,44,1505 ; Polymer Preprints (Amer. Chem . R. R. Stromberg and L. E. Smith J. Phys. Chem. 1967,71,2470. Soc. Meeting Chicago Illinois 1970).

ISSN:0430-0696    

DOI:10.1039/SF9700400192

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

GENERAL DISCUSSION Dr. W. Paik and Prof. J. O’M. Bockris (University of Pennsylvania) (communicated) For the examination and detection of a conducting oxide film ellipsometry has hitherto had to be combined with e.g. coulometry in order that the three film properties thickness refractive index and absorption coefficient be measured. There is a certain inconvenience and a great increase in time needed in having to carry out this second set of parallel measurements. We have recently shown that an additional measurement of the reflectance of the surface enables us to avoid the auxiliary non-optical measurements. Changes of A and + can be measured by record ing the intensity of the reflected light at specific settings of the quarter wave plate (as described in the paper of Bockris et al.); in addition one measures the change in reJEectance during film formation.We have shown that it is then possible to obtain uniquely the three film parameters of a given film without the need for auxiliary non-optical measurements. Dr. R. Parsons (University of Bristol) said I would ask whether Bockris et al. have any other evidence for the mechanism proposed in their eqn (6)-(8). It seems improbable that a place exchange process of the type described in eqn (7) should be in equilibrium. Similarly what evidence is there that kinetic equations approximately of the type proposed by Temkin should be valid for processes of this type especially with a single-valued heterogeneity coefficient. Finally the reasons for assuming that the major constituent of the surface is the product of the rate-determining step are by no means clear.Prof. J. O’M. Bockris (University of Pennsylvania) (communicated) I think that Parsons has not fully understood the mechanism proposed for the oxide growth in the prepassive region. It is proposed that a Temkin step is involved in the equilibrium of OH- in solution and the adsorbed OH on the surface and this cannot be objectionable. The adsorbed OH- discharge which is the rate-determining step occurs also on empty M sites (although with a bound 0 in the second layer beneath them) and it is then after this Temkin-affected step that a fast place-exchange occurs by means of which the adsorbed OH aggregates to the edge of the oxide clump. The origin of the Temkin kinetics would only break down if the oxide clumps become too frequent i.e. round the top of the (i,V) relation.It is after 80 % coverage that use of Temkin kinetics is dropped. In further reply to Parsons the evidence for the mechanisms proposed in eqn (6)-(8) in our paper is the exclusive character of the consistency with experiment which the present model gives. Many other models were applied in the work with Genshaw and Brusic * in some mathematical detail to the growth rate but none of them provided the degree of numerical consistence between model and data which the present proposal gives. Thus few mechanisms of oxide growth are consistent with a logarithmic growth law. The most obvious alternate is a rate-determining chemisorption step in Temkin kinetics however ellipsometry indicates that it is a phase oxide which is growing (in clumps therefore) before the peak of the (i,V) Temkin Zhur.Fiz. Khim. 1941 15 296. V. Brusic Thesis (University of Pennsylvania 1971). 201 202 GENERAL DISCUSSION curve. The place exchange path for the initial oxidation phase growth in a metal was first suggested by Lanyon and Trapnell.' That place exchange plays an import- ant part in the growth of the passive oxide was proposed by Cohn,2 though in a different form to that given here. That the rate-determining step in the sequence is the discharge of OH- onto Fe atoms in which the exchange suggested by Lanyon and Trapnell has occurred seems physically consistent for here the second bond between Fe and 0 will be weaker than the first. Prof. Dr. K. J. Vetter (Free University of Berlin Berlin-Dahlem) said The assumption of a primary inner Fe(OH) film by Bockris et al.is quite unusual. So far the widely accepted model of the passive film has been an inner layer of Fe304 (magnetite) covered by an outer layer of y-Fe,O, e.g. ref. (3)-(8). This structure of the passive film is in agreement with all properties of the passive iron so far known to me. If they postulate such a substantial modification of the film model it seems essential first to prove their ellipsometric values by using the common model confirmed in so many cases. In my opinion the model has to be critically analyzed. 3t 170 m I d Y not 1 m 2 potential V[mV] against NHE FIG. 1. A. H. Lanyon and B. M. W. Trapnell Proc. Roy. SOC. A 1955,227,387. N. Nagayama and M. Cohn J. Electrochem. Soc. 1966 111 512; M. Sato and M. Cohn J. Electrochem. SOC. 1964 111 512. K. J. Vetter 2.Elektrochem. 1958 62 642. H. Gohr and E. Lange Naturwiss 1956,43 12. K. E. Heusler K. G. Weil and K. F. Bonhoeffer 2. phys. Chem. N.F. 1958,15 149. K. E. Heusler J. Electrochem. Soc. 1963,110,703. J. Kruger J. Electrochem. Soc. 1963 110 654. ' M. Nagayama and M. Cohen J. Electrochem. Soc. 1962,109,781 ; 1963,110,670. GENERAL DISCUSSION 203 Fig. 1 represents values of the layer thickness and the refractive index N = n(l -iK) of the film using the Fe304/y-Fe203 model. The equivalent volumes of Fe(OH) are 13.22 cm3/val (p = 3.41 g/cm3) of Fe304 5.58 cm3/val (p = 5.18 g/cm3) of y-Fe203 (Maghemite) 5.42 cm3/val (p = 4.907 g/cm3) and of a-Fe,O 5.06 cm3/val (p = 5.25g/cm3). The film thicknesses given in their fig. 14 which result from coulometric measurements are converted in that the equivalent volumes of Fe,O or y-Fe203 are used.The result is a continuous increase of the film thick- ness and the disappearance of the peculiar range where the thickness is constant. If the ellipsometric data of the potentiostatic steady state given in their fig. 9 are used other values of components n and IC of the refractive index are obtained according to calculations made in collaboration with Dr. F. Gorn. These values are also repre- sented in fig. 1.. The peculiar values of n (approximately no of the electrolyte) at negative potentials disappear. The new value if approximately that of bulk Fe304 (n = 2.47 I). The adsorptive index is also approximately the known bulk value of Fe304 ( K = 0.227 l). When reaching the Flade potential VF = 0.58- 0.059 pH the adsorptive index K decreases owing to the lower electronic conductivity of y-Fe203.The value n does not change as much as expected. There is therefore no reason for a modification of the accepted model particularly because the Fe(OH),/ Fe203-model does not match the ellipsometric and coulometric data. Further unpublished investigations by Dr. J. W. Schultze in our laboratory show that the passive film on iron does not include any hydrogen whatsoever. By applica- tion of a tracer method using tritium (HTO) it is proved that the passive film only adsorbs a monolayer of water as for the active state of metals. The amount of tritium on the film is exchanged within short periods of time in natural water and does not depend on the film thickness the potential and the pH-value. This result again refutes the existence of a Fe(OH) film.The reaction 4Fe(OH) + Fe304 + Fe + 4H20 has a negative free reaction energy AG = -7.8 kcal/mol. That means that the activity of metallic iron (MeZ++ze-) dissolved in the hydroxide Fe(OH) is higher than unity. A direct formation of an oxide or a hydroxide on the surface of a metal without the formation of a dissolved intermediate product in the electrolyte is possible only if the activity of the metal in the reaction product is smaller than ~ n i t y . ~ - ~ That means that the reaction product is not saturated with the metal. Bulk Fe(OH), however is supersaturated with metallic iron according to the referred negative AG value. A similar problem is the dissolution of salt in a solvent. Only an unsaturated or a saturated solution can be formed but never a supersaturated one.For thermodynamic reasons a divalent iron hydroxide or oxide therefore cannot be formed directly at the interface between metal and oxide. Fe304 is the onlyequilibrium oxide that exists at room temperature. Dr. B. Caban (Case Western Reserve University) said Regarding Vetter’s question as to whether the passivation could not involve an Fe304-F203 sandwich I would make the following comments. The ellipsometric galvanostatic transients showed that three distinct time regions were apparent (see fig. 2 12 and 14). The first and the third were strictly linear and the third region extrapolated back to zero at t = 0 within the accuracy of the measurements. Coulometry showed that only two St. Loria and C. Zakrzewski Anz. Krakau Akad. A 1910,284 ; St. Loria Proc.Acad. Amsterdam 1912 14 972 from Gmelins Hdb. anorg. Chem. Verlag. Chemie. 1932 8. Aufl. Bd. 59B p. 50-51. ’ K. J. Vetter 2. Elektrochem. 1962 66 577. K. J. Vetter J. Electrochem. SOC. 1963 110 597. K. J. Vetter Ber. Bunsenges. phys. Chem. 1965 69 589. 204 GENERAL DISCUSSION monolayers of oxide were formed during the first time segment. It was possible to fit the first and third segments with the assumption of only one material in each. The material formed during the first segment could not have been Fe304 or Fe203 because of the low value for k obtained from the computations. The only way the third (passive) segment could have been fitted with a sandwich structure would have been if the ratio of the two thicknesses had remained constant. Since the thickness was only two monolayers at the start this would have required one monolayer of Fe304 and one of Fe203.A consideration of the structure and the unit cell size of the two materials ruled out this possibility. When we assumed that oxide (1) formed starting from t = 0 and that during the second time region this was transformed with first order kinetics to the oxide (2) together with a simultaneous formation of further oxide (2) on the already converted areas with (2) being the only species in the third region the number of electrons required per iron atom for (1) and (2) was always in the ratio 1 1.5 (or 2 3) except at extremely low c.d. where no passivation occurred. This implies one divalent oxide (or hydroxide) at first which is later converted to one trivalent oxide without a sandwich structure being possible.The coulometry involved in the three regions was also informative. Miss V . Brusic (University of Pennsylvania) (communicated) in reply to Vetter although neither ellipsometry nor coulometry alone could give a direct evidence about the film composition our results are better interpreted in terms of the formation of Fe(OH)2 and its complete transformation into Fe203 than by the formation of a duplex film Fe304-+Fe304 I Fe20, for a number of reasons. (1) Optical results such as the plot of A against $ (fig. 3 of the paper) show three regions of approximately linear dependence. The first and third region can be extrapolated to the origin (corresponding to film free state of the surface) indicating that a single film with varying thickness is present. The model suggested by Vetter has an inconsistency in the third region of potential the film thickness recalculated by him increases with potential as if due to the increase of Fe,03 thickness only above the layer of Fe304 which stays unchanged.Thus the ratio of Fe203/Fe304 changes continuously and one can expect that the index of such film changes too. However in the third region in Vetter’s calculation as well as in ours the index of refraction is constant. One might argue that the constancy of the index of refraction is due to the simultaneous growth of both layers of the film in a constant ratio. Such mechanism is unlikely TABLE 1 .-NUMBER OF MILLICOULOMBS DETERMINED DURING GALVANOSTATIC REDUCTION OF THE FILM AND FILM COMPOSITION AS A FUNCTION OF POTENTIAL. potential mC/cm2 mC/cm2 mC/cmz Fe(0Hh Fe203* % Fez03 mV (NHE) total 1st move 2nd move mCz/mCl mC/cm2 mol mC/cmzmol in the film -440 1 0 1 co 1 5 ~ 1 0 - ~ 0 0 0 -340 1.57 0.13 1.44 1 1 1.18 5 .9 ~ 1 0 - ~ 0.39 6.5xlO-’O 10 -140 2.60 0.73 1.87 2.56 0.41 2 . 2 ~ lo-’ 2.19 3 . 6 ~ 62 t-60 3.34 1.11 2.23 2 0 0 3.34 5 . 5 7 ~ 100 + 620 +620 5.0 1.72 3.28 1.91 0 0 5.0 8.34xlO-’ 100 1st 2nd * Assuming that the reduction proceeds according to Fe3+ +Fe2+ +Fe the number of milli- coulombs for the reduction of Fe20s was calculated as 3 x (mC/cm2)l,t move or when the ratio of mC2/mC1 was -2 a total number of millicoulombs was taken to be due to the reduction of tri- valent iron (Fe203). In the transition region the total number of millicoulombs exceeds that for Fe203 ; the difference is assumed to be due to the reduction of the originally present di-valent iron (Fe(0HM.GENERAL DISCUSSION 205 and is difficult to justify in the present results; e.g. if the mechanism were indeed the formation of the first film (first region) the growth of the second film over it (second region) followed by simultaneous growth of the two in the third one would expect that such a mechanism is dependent on film thickness or field; the close similarity of the results obtained in galvanostatic (fast) transients and potentiostatic (" steady state ") measurements indicates however that the process is similar in nature both in short- and long-time experiment and is preferentially potential dependent (fig. 9 in the paper). (2) Coulornetric results When the electrode was oxidized in the prepassive region only a single wave is observed during galvanostatic reduction.4 lo0 - 80 - 60 - 40 - 20 - looh 0 I I / -600-400 -200 0 200 400 600 800 Potential mV NHF FIG. 1.-% Fez03 in the anodic film on iron as a function of potential. I I I I 1 L (Results of galvanostatic reduction are close to the results of potentionstatic reduction corrected for hydrogen thus giving lower limit of thickness.) Reduction from higher potentials gives two waves whose ratio changes in the so-called " transition region " and becomes constant above +0.060 V having a value of 1 2 in the third region (see the attached table). This suggests a reduction process of Fe3+ to Fe2+ to Fe as it would be in Fe203 without the presence of Fe30,. Assuming that the amount of Fe203 in the film is proportional to the number of coulombs of the first wave the film composition is calculated as given in the table and fig.1. ( 3 ) When coulometry was used to calculate the film thickness in the pre-passive region it was assumed at first that either of the following compositions was possible FeO Fe(OH), Fe30, Fe203 and FeO(0H). The calculated index of refraction however irrespective of what film was assumed was always less than 2 (limits of n when Fe,O was assumed were 1.5 and 1.9). Thus the assumption of Fe(OH) being present gave the least discrepancy between the obtained and expected value of n. (4) The model is not entirely in disagreement with other reported data e.g. Foley et al.' have found by electron diffraction in the active potential region both Fe30 and FeO(OH) depending on the crystallographic orientation of iron. Thus hydrogen is present in the prepassive film.Although Fe(OH) was not detected it could be argued that the unstable prepassive film may change during the preparation of the specimen for electron diffraction analysis ; it may easily oxidize or even lose the water during the exposure to the electron beam. The latter is shown by Arm- strong for In(OH), which transforms into In,03. In the passive layer neglecting C. L. Foley J. Kruger and C. J. Bechtoldt J. Electrochem. Soc. 1697,114 994. R. D. Armstrong A. B. Suttie and H. R. Thirsk Electrochem. Acta. 1968 13 1. 206 GENERAL DISCUSSION the possible changes in the film the results of electron diffraction suggest the presence of Fe,O with some possibility of the masked presence of Fe304. The results of the experiments using Mossbauer effect conducted in situ do not confirm the presence of any Fe304.(5) The model of one film being transformed into another is also checked by the analysis of the galvanostatic results given in fig. 2 of the paper or schematically on fig. 2 here. The experimentally observed ratio of t2 to t the j time i tl t2 FIG. 2.-Schematic representation of the change of A with time of a typical galvanostatic transient. inflection points in delta-time curves is close to the calculated value for the assumed model. It was assumed that (i) in the first region of (A t ) curve (schematically given in fig. 2) the entire number of millicoulombs is used for the formation of an oxide and Me + H,O+ Me-oxide + nle diV/dt = iT/nl (ii) In the second region the first oxide is transformed into higher oxide which can also be formed independently Me-oxide +Me-oxide + (n - n,)e- (3) and Me + H20-+Me-oxide + n2e-.(4) The conversion current of eqn (3) is assumed to be a linear function of time and to be equal to the total current at tl and zero at t = t2. Thus Formation of the second film is described by ic = iT(t2 - t ) / ( t 2 - t,). dN2 i ~ - i c o n v + konv dN2 d N 2 dt n2 n2 - n 1 dtdirect dtconverted' dt n 2 n2@2 - tl> (n2 - n 1 W z - t l ) - (5) +---- - -~ -- - dN2 i iT(t2-t) + iT(t2 - t) - = - W. O'Grady private communication. (Work conducted as a part of Ph.D. Thesis at Univ. of Penn. Philadelphia). GENERAL DISCUSSION 207 Integrating over t where t varies between t2 and tl At t = tZ which when introduced into the above equation finally gives N2 = idt2n2 %In1 = (tl + t2)/2t,.The ratio of the oxidized state of the metal cation in the oxide should be according to the model proportional to the inflection times t and t2. The ratio t2/tl has been averaged from 14 individual values to be 1.89 i.e. t2 = 1.89t1. Thus n2/nl = 1.445 (experimental). In comparison with some of the possible theoretical ratios Fe(OH) -+FeO(OH) Fe(OH),-+Fe,O Fe304-+ Fe203 Fe(OH) -+ Fe,O FeO(OH)-+Fe203 n2/n = 1. n2/nl = 312 = 1.5 n2/nl = 312 = 1.5 n2/n = 918 = 1.125 n2/n = 816 = 1.333 The experimental ratio most closely supports the ideas of transformation of some divalent iron compound into a tri-valent compound i.e. the transformation of Fe(OH) into Fe203 seems to be very probable. Dr. M. A. Genshaw (Elkhart Indiana) said In reply to Vetter we prefer to interpret our results in terms of a film transformation mechanism rather than the formation of a duplex film for a number of reasons.One is based on optical con- siderations alone. In the plot of $ against A obtained during the formation of the film three regions of approximately linear dependence are observed. From optical calculations it is easily shown that the growth of a single film with constant optical properties produces an approximately straight line in a ($ A) plot (optical properties assumed constant) with the line originating at the point corresponding to the bare surface. This is illustrated in fig. 1. This may be compared to fig. 3 in our paper. The first and third region of linearity extrapolate to the origin (this is much clearer in the original when each experiment is plotted separately).This suggests a single film of varying thickness is present in regions (1) and (3). A line indicates the type of plot expected if an initial film of low refractive index (Fe(OH),) is transformed to a film of higher refractive index. The formation of a double film is also illustrated. Only two regions of linearity exist in this plot and the second region does not extra- polate to the origin. To obtain the experimental plot it would be necessary to have the first film form and the second film grow over it followed by simultaneous growth of both films in a constant ratio. This seems unlikely. A second argument arises from coulometry. In the prepassive region only a single reduction wave is observed. For the passive film two reduction waves are observed with one coinciding with that of the prepassive film.The ratio of the two waves increases with increasing potential to a ratio of 4 in the third region. This suggests a reduction process of Fe3+ to Fe2+ to Fe. A reduction of Fe3+ to Fe%+ to Fe would give rise to a ratio of &. For the reduction of a duplex film to maintain a constant ratio of coulombs the films must maintain a constant ratio of thicknesses. Another argument is that the refractive index calculated when coulometry is used to calculate thc thickness is always low (less than 2) irrespective of what film is assumed. The only iron oxide which has a low refractive index is Fe(OH),. This then is 208 GENERAL DISCUSSION suggested to be the film initially formed. Arguments based on thermodynamic potentials may be used only to show that the formation of certain films is impossible.For films formed under kinetic control as are the passive films the potential of formation and inflections need not have thermodynamic significance. 29. I 29.1 L 28.1 28.9 I I I I I I24 I22 120 I18 I I6 A" FIG. 1.-Theoretical (G A) plots for Fe(OH)2 (n = 1.5-Oo.075i) FeJOs (n = 2.39-0.2493') and Fe203 (n-t-2.88-0.378i). A film transition and double film are illustrated. Points are indicated at 10 A intervals. Prof. J. O'M. Bockris (University of Pennsylvania) (communicated) I have nothing to add to the presentation of Genshaw and Cahan respectively in answering Vetter's point. The ellipsometric measurements are not consistent with the sandwich structure for passive iron. I think this shows well the increased resolution possible in passivity problems by using the ellipsometric approach.Prof. M. J. Dignam (University of Toronto) said Since the topic of surface rough- ness has been raised by Genshaw and earlier by Hayfield I shall now make a contribution of a general nature which relates to the influence of roughness on reflec- tance measurements from metal surfaces and hence to the interpretation of such measurements. The work which I shall outline briefly was dolie by Mr. M. Moskovits and myself and is concerned entirely with optical effects brought about by adsorption from the gas phase. The situation should not be markedly different however for electrode +electrolyte systems. We have developed a model which predicts the effect of gas adsorption on the reflectivity and transmissivity of evaporated metal films and have tested the predictions of this model against experimental data.The principle distinguishing features of the model arise from the explicit inclusion of roughness parameters. These have been incorporated on the assumption that for normal incident or s-polarized radiation the influence of the metal bumps may be approximated by treating them as metal spheres. The effect of these spheres on s-polarized radiation is to introduce a resonance absorption phenomenon the central GENERAL DISCUSSION 209 frequency of which depends on the packing density of the spheres while the breadth of the absorption band depends to a good approximation on the mean radius of the spheres. The absorption phenomenon arises principally from the plasma resonance in the spheres with interband transitions also making a contribution.The radius of the spheres affects the damping factor for the electron plasma and hence the breadth of the absorption band by influencing the effective mean free path of the conduction electrons. This resonance is not easily detected directly but is made apparent on observing changes in reflection from or transmission through a metal film possessing such a roughness region following adsorption of gas molecules. Increasing the packing density of spheres shifts the resonance frequency to lower frequencies. Molecules which intereact weakly with the metal (physical adsorption) produce a similar though reduced shift. The model has been extended to cover strong interaction (chemi- sorption) by including the effect of electron transfer between the metal and the adsorbate.The expressions derived for the change in reflectivity and transmissivity on adsorption as a function of wavelength involve in total four wavelength independent parameters which reduce to three for physical adsorption. Two of these the mean radius of the bumps and the volume fraction of the roughness region occupied by metal are roughness parameters while the other two the product polarizability x coverage and an electron transfer parameter are adsorption parameters. The apparatus designed for this study consists of an ultra-high vacuum system incorporating a quartz crystal piezoelectric microbalance facilities for silver deposi- tion and for gas handling and a dual beam spectrophotometer which can be used in both a reflectance and transmittance mode. FIG. 1.-Transmittance data for methanol adsorbed on a 500 silver film.The lines represent the computed curve the points the data. Changes in transmittance and reflectance of evaporated silver films following adsorption of oxygen ethanol diethyl ether and carbon dioxide have been made and compared with microbalance data. The results while at variance with previous optical models confirm all the features predicted by the present model and yield 210 GENERAL DISCUSSION values for the roughness parameters and coverages in good agreement with indepen- dent estimates. Thus the parameters representing the volume fraction of the roughness region occupied by metal was a value close to 0.3 for all the adsorbates while the parameter representing the mean radius of the bumps takes on values in the neighbourhood of lOA in good agreement with estimates made from electron scattering measurements.'.Again for physically adsorbed species (ether and CO,) coverages calculated from the appropriate molecular polarizabilities and the values for the optical parameter repre- senting polarizability x coverage agree with those measured directly using the micro- balance to within 10 to 30 %. wavelength nm FIG. 2.-Reflectance and transmittance data for oxygen adsorbed on a 500 A silver film. The lines represent computed curves the points the data. The quality of fit achieved using our model is demonstrated in fig. 1 and 2. The methanol data were fitted using only three parameters (i.e. the electron transfer parameter was found to be zero as expected). The oxygen data require all four parameters the value for the electron transfer parameter indicating that about 1.5 electrons are lost from the metal per O2 molecule adsorbed.The principle of the deriva- tion is however straight forward. The expression for the polarizability of a metal sphere is combined with that for the adsorbed species using the Lorentz-Lorenz equation for the local field to calculate the effective complex refractive index of the roughness region. In calculating the polarizability of the metal spheres the bulk optical constants were used as modified by two effects the first being the change in the effective mean free path of the conduction electrons arising from collisions with the surface of a metal sphere and the second the change in the plasma frequency H. F. Bennett R. L. Peck D. K. Burge and J. M. Bennett J .Appl. Phys. 1969 40 3351. H. Raether in The Structure and Chemistry of Solid Surfaces ; ed. G. A. Samorjai (John Wiley and Sons Inc. 1969) p. 10. The equations which we have derived are complex. GENERAL DISCUSSION 21 1 arising from the change in the concentrate of conduction electrons in the roughness layer due to electron transfer between the metal and adsorbate. Extension of the model to cover the electro+electrolyte system and the effect of electrode potential variation is straightforward and is currently being undertaken. The model is also being extended to cover p-polarized light and hence ellipsometric measurements. The latter problem is not as trivial as might first appear since the roughness layer must be treated as an optically anisotropic one. Finally I would point out the real possibility that the extremely thin anodic films formed on certain metals (e.g.the anodic oxide film formed on platinum) are not in fact good conductors (i.e. do not have complex refractive indices in the visible region of the spectrum) but rather that this apparent result is an artifact arising from the optical properties of the roughness layer. Miss M. A. Barrett and Dr. R. Parsons (University of Bristol) (communicated) The treatment of rough surfaces by Dignam is interesting. The problem of a rigorous treatment of adsorbed species on metal has generally been sidestepped and linearity of optical changes has been assumed between zero and full coverage. With regard to anodic films on platinum if the roughness of the metal were to play a major role it is difficult to see how the various optical changes observed and the various deter- minations of film index come as near agreement as they do when based on evaporated film (McIntyre and Kolb Visscher) on bright rolled and on diamond dust polished surfaces (our results).The question also arises whether the Lorentz-Lorenz equation for local field can be accurate in a film of such small thicknesses as are being dealt with here. Dr. W. E. OeGrady and Prof. J. O’M. Bockris (University of Pennsylvania) (communicated) We have recently carried out Mossbauer measurements on passive films and these are relevant to the interpretation of the ellipsometric work carried out by Bockris Genshaw and Brusic. We have not yet completed the evaluation of our Mossbauer measurements and there are difficulties in the interpretation which are concerned with effect of fine particle size.However it is clear at the present time that there is no indication whatsoever of characteristic chemical shifts which belong to an Fe,O layer. A duplex film that which Vetter supports is certainly not consistent with our Mossbauer measurements. Dr. Th. F. Tadros (Plant Protection Ltd. Jealott’s Hill Res. Station Berks) said With regard to the paper by Stromberg Smith and McCrackin in polymer adsorp- tion one must consider the diffusion of the high-molecular-weight molecules to the solid/solution interface. For adsorption on ferrotype metals the surface is rough and one would expect pores to be present so that time is required for molecules with relatively high molecular weight to diffuse to the solid/solution interface.Thus in this case both the amount adsorbed and thickness of adsorbed layer will be expected to increase with time. With relatively low-molecular-weight polystyrene on steel and chrome surfaces the diffusion of the molecules into the porous structure is rapid and no change of thickness of adsorbed layer with time will be observed. On the other hand on a smooth surface such as mercury the relatively high molecular weight polystyrene can still reach the interface quickly and the adsorption of the first few molecules takes place rapidly reaching their final conformation in a very short period. These molecules will not cover the whole surface but will leave “ gaps ” or “ holes ” in between them. The next molecules arriving at the interface need time to diffuse into these “ gaps ” or “ holes ” between already adsorbed molecules.212 GENERAL DISCUSSION This explains the increase in amount of adsorption with time. However as the molecules reach their final conformat ion quickly on the mercury/solution interface no change in thickness with time is observed. The increase in thickness of adsorbed layer with increase in molecular weight (fig. 5 of their paper) can be due to steric hinderance preventing complete penetration of the molecules in the pores. As the molecular weight gets higher the molecules will adsorb at the outer surface only and this explains the plateau reached in the curve of fig. 5. With a smooth surface as for mercury this would not be the case and no change in thickness with molecular weight would be expected. Dr. R. R. Stromberg (Nat.Bur. Stand. Washington) said In reply to Tadros our analysis of the ellipsometric results presumes a solid substrate with discrete boundaries. For this reason experiments of the type described in our paper cannot ascertain with certainty whether the time effects observed for adsorption could be ascribed to some degree to diffusion into pores of the substrate as suggested by Tadros. Part of the difficulty lies in the inability to predict the optical effects of the complex mixed film of oxide and polymer that would result from such a process. However unpublished work on the rates of adsorption of radio-labelled polystyrene on chrome ferrotype plate does argue against a diffusion mechanism. First the time required to reach the equilibrium adsorbance value was the same for samples with molecular weights of 76,000 and 1 x 106 adsorbed from cyclohexane solution.The diffusion process would be expected to be slower for the higher molecular weight. Secondly these same polymers adsorbed from benzene solution reached their equilibrium adsorbance much sooner than from cyclohexane. If anything diffusion in benzene solution should be slower than in cyclohexane because of the more extended conformation in the good solvent. Dr. R. Parsons (University of Bristol) said With regard to the paper by Stromberg et a/. is it possible to detect ellipsometrically the rotation of a large molecule in the absence of any other change of configuration? Dr. R. Stromberg (Nat. Bur. Stand. Washington) said In reply to Parsons it is not possible by a single ellipsometric measurement to determine the optical rotation of an adsorbed molecule.We have attempted to analyze optically active films and do not know whether this would be possible for some combination of measurements. Prof. M. J. Dignam (University of Toronto) said I would briefly comment on the alignment procedure referred to in the paper by Stromberg et a/. I have published a paper,2 with Moskovits in which this alignment procedure is discussed. We have shown theoretically that for an anisotropic reflector extinction is not in general achieved for crossed polarizers no compensator and one of the polarizers aligned with its axis in the plane of incidence. Furthermore we have demonstrated the effect experimentally with data for an aluminium surface polished preferentially in one direction. Extinction for crossed polarizers could be achieved only if the polish- ing direction was either parallel or perpendicular to the plane of incidence. Dr. R. Stromberg (Nat. Bur. Stand. Washington) said In reply to Dignan the alignment procedure referred to in our paper assumes an optically isotropic substrate. We have not analyzed the case of an anisotropic reflector. W. H. Grant and R. R. Stromberg Nat. Bur. Stand. Washington; see also preprints Division of Polymer Chemistry Anier. Chem. SOC. Meeting Chicago September 1970. M. J. Dignam and M. Moskovits Appl. Optics 1970 9 1868.

ISSN:0430-0696    

DOI:10.1039/SF9700400201

出版商:RSC    

年代:1970

数据来源: RSC

摘要:

AUTHOR INDEX * Barrett M. A. 49 72,98 128,211. Bewick A. 49,114 130. Bockris J. O’M. 85,177 201,208 211. Brusic V. 177,203. Cahan B. D. 36 46 49 50 52 61 85 90 133 Conway B. E. 95 126 135 174. Dignam M. J. 47,208,212 den Engelsen D. 45. Fleischmann M. 130. Genshaw M. A. 87,177,207. Hausen W. N. 27 175. Hayfleld P. C. S. 7. Horkans J. 36,49 52 61. Kolb D. M. 99. Kuhn A. T. 86. McCrackin F L. 192. McIntyre J. D. E. 46 50 55 61 99 126. 203. Mark H. B. Jr. 157 176. Memming R. 145 173 174 175. Meyer F. 17. Miillers F. 145. O’Grady W. E. 211. Paik W. 85 201. Parsons R. 49,72,85,98,128,129,130,173,201 Plieth W. J. 45 89 137 173. Randall E. N. 157,176. Smith L. E. 192. Spamaay M. J. 17 50. Stedman M. 48 64 85 86 88. Stromberg R. R. 192 212. Tadros Th. F. 211. Tuxford A. M. 114 130 132. Vetter K. J. 134 202. Yeager E. 36,49 52 54 61 91 131. 211,212. * The references in heavy type indicate papers submitted for discussion.

ISSN:0430-0696    

DOI:10.1039/SF9700400213

出版商:RSC    

年代:1970

数据来源: RSC