Electrosorption of iodide on platinum - American Chemical Society

Feb 13, 1990 - AnX-ray standing wave is generated whenever coherently related incident and reflected X-ray plane waves interfere. In this study, we us...
1 downloads 0 Views 2MB Size
8280

J . Phys. Chem. 1990, 94, 8280-8288

Electrosorption of Iodide on Platinum: Packing Density and Potential-Dependent Distributional Changes Observed in Situ with X-ray Standing Waves G . M. Bommarito, J. H. White, and H. D. Abruiia* Department of Chemistry, Baker Laboratory, Cornell L'nioersity, Ithaca, New York 14853 (Received: February 13, 1990)

We describe the X-ray standing wave technique for directly measuring, in situ, the ionic concentration profile at an electrode/electrolyte interface on the angstrom scale. An X-ray standing wave is generated whenever coherently related incident and reflected X-ray plane waves interfere. In this study, we use X-ray standing wave generated during Bragg reflection from a platinum/carbon layered synthetic microstructure to investigate the electrosorption of iodide on platinum. In the presence of I O KMsodium iodide in solution, the electrosorptionisotherm for iodide showed similarities to that for the Pt( 1 1 I ) , T system reported previously. In addition, potential-dependentdistributional changes (normal to the electrode/electrolyte interface) of iodide in the diffuse layer were inferred from the observed potential-dependent changes in the fluorescence-detected X-ray standing waves. We attempt to relate features in the aforementioned isotherm to the observed distributional rearrangements in the diffuse layer.

Introduction The reactivity of an electrochemical interface is extensively determined by its composition and structure. Electron transfer, adsorption, and other processes of fundamental importance are governed by the structural and compositional properties of the interface. The need for a better understanding of structure/ function relationships has motivated considerable experimental and theoretical activity in this field. With the advent of ultrahigh-vacuum (UHV) and surface spectroscopic techniques, in particular low-energy electron diffraction (LEED) and Auger electron spectroscopy, a great deal of structural and compositional information has been gained on the solid/solution interface of emersed electrodes.' By their very nature, however, these studies suffer from the loss of the solution side of the interface, a significant component of the system, determining mass transport and providing for electroneutrality. In addition, potential stability of the system is questionable upon emersion, resulting in ambiguous variations in the activity and concentrations of interfacial species. As a result, in situ investigations of the structure of liquid/solid interfaces are clearly necessary . Because of their intensity, collimation, polarization, and great penetrative power, X-rays, from a synchrotron source are an ideal structural probe of condensed phases. The recent application of extended X-ray absorption fine structure (EXAFS) and surface diffraction techniques has yielded new insights into the electrode/electrolyte interfacial structure, at the atomic level.* These studies have focused primarily on the characterization of the compact part of the double layer and in many cases on adsorbates binding specifically to the electrode surface. The lack of studies involving the direct investigation of the extended part of these systems (i.e., the diffuse layer) is caused largely by the formidable obstacles encountered in such studies: ( I ) the nature of the diffuse layer is that of a liquid, best described in terms of average distribution functions rather than a well-defined structure, as is the case for the compact layer: ( 2 ) techniques probing the solution side of the interface must, in most instances, be capable of detecting characteristic scattered radiation originating from minute amounts of an ionic component. Given these two considerations, it is not surprising that direct structural studies of the double layer as a whole are virtually nonexistent. ( I ) (a) Hubbard. A. T. Arc. Chem. Res. 1980, 13, 177. (b) Homa, A. S.; Yeager, E.; Cahan, B. D. J . Elecrrounof. Chem. 1983,150, 181. (c) Wagner, F. T.; ROSS,P. N . J . Electrounul. Chem. 1983, 150, 141. (2) (a) Abruiia, H. D.; White, J. H.; Albarelli, M. J.; Bommarito, G. M.; Bedzyk, M. J.; McMillan, M. J . Phys. Chem. 1988, 92, 7045. (b) Samant. M . G.:Toney, M . F.; Borges, G . L.: Blum, L.; Melroy, 0. R. J . Phys. Chem. 1988. 92. 220.

0022-3654 I 9 0 12094-828O$O2.511 .~ ,I O

In this paper, we demonstrate the utility of the X-ray standing wave (XSW) method,3 using synchrotron radiation, as an in situ technique yielding an absolute measurement, on the angstrom scale, of the double-layer distribution for iodide in a supporting electrolyte contacting the platinum surface of a platinum/carbon layered synthetic microstructure (LSM) (40.8-%, d spacing). The adsorption of iodide on polycrystalline and single-crystal Pt electrodes has been widely investigated using LEED, Auger, EXAFS and voltammetric techniques. It is generally accepted that immersion of a Pt( 11 1) surface into aqueous iodide (or HI) solutions results in the formation of an ordered adlayer of iodine atoms. Furthermore, the Pt( 11 1 ) / I system possesses a rich potential-dependent coverage isotherm, which has been characterized by Auger spectroscopy for the emersed case4 and in situ by X-ray absorption spectroscopy (XAS).S Its features are explained in terms of potential-dependent structural and distributional changes. The aim in this work is to study in situ, via the XSW method using LSMs, the potential dependence of structural changes in the distribution of iodine species and to relate features of the aforementioned isotherm to observations from this measurement.

Theoretical Background XSWs are generated when coherently related incident and reflected plane waves interfere (Figure 1 ). Conventionally, XSWs are generated by dynamical Bragg reflection from perfect single crystals (e.g., Si, Ge, or GaAs).6 Because the standing wave periodicity is determined by the d spacing of the generating substrate, this technique is a precise tool (*I% of the d spacing) for measuring bond lengths between adsorbate atoms and surface/bulk lattice positions, over the range 1-4 %, (Le., typical crystallograhic d spacings). In this work, the interest lies in studying structural changes for ionic distributions with characteristic decay lengths ranging from a few tens to hundreds of angstroms. Standing waves generated from perfect single crystals are too short to effectively measure these extended distributions. Thus, an alternative type of substrate must be used to create XSW with long periods. Furthermore, this experiment requires the substrate's surface to be metallic, since it is also the working electrode. Both of these constraints are met by employing LSMs as Bragg diffracting structures, (3) (a) Batterman, B. W. Phys. Reo. 1964, 133, A759. (b) Batterman, B. W . Phys. Reo. Lett. 1969, 22, 703. (4) Lu, F.; Salaita, G. N.; Baltruschat, H.; Hubbard, A. T. J . Electrounul. Chem. 1987, 222, 305. ( 5 ) White, J . H.; Abruiia, H . D. J . Phys. Chem. 1988, 92, 7131. (6) (a) Cowan, P. L.; Golovchenko, J . A,; Robbins, M. F. Phys. Reu. Lett. 1980.44, 1680. (b) Golovchenko, J. A,; Patel, J. R.; Kaplan, D. R.;Cowan, P. L.; Bedzyk, M. J. Phys., Rea Lett. 1982. 49, 560. (c) Bedzyk, M. J.; Materlik. G. Phys. Rec. B 1985, 31, 41 I O .

0 1990 American Chemical Society

The Journal of Physical Chemistry, Vol. 94, No. 21, 1990 8281

Electrosorption of Iodide on Platinum

.

0

,

-I--,--,

,

,

,

,

,

,

,

,

2

m r[,

0

x

-

26

24 I

125 laver pairs

I

1

28

a

.

30

I

Figure 1. Illustration of the X-ray standing wave generated by the interference of the coherently related incident and Bragg diffracted plane waves for a platinum/carbon LSM. The solid lines indicate the antinodes of the incident and diffracted waves, the bold lines point out the antinodes of the standing wave, and the dashed lines mark the position of the diffraction planes in the multilayer.

LSMs are artificial, depth-periodic structures, prepared by alternating layers of high and low electron density elements, thus creating a superlattice structure with diffraction planes centered in the high electron density layers. LSMs are especially attractive in the present context because they have (1) large and controllable d spacings (20-200 A) resulting in the generation of long-period XSW and (2) high Bragg reflectivity (SO-SO%) and (3) can be prepared with specified surface and bulk compositions. The XSW technique employing LSMs has been applied to the study of Langmuir-Blodgett films7 where the distance between a heavyatom label in the film and the substrate's surface can range from tens to hundreds of angstroms. In the study described here we employed a Pt/C LSM with a d spacing of 40.8 A and Pt as the sur face materia 1. Figure 1 illustrates the formation of a standing wave field by the interference of an incident plane wave with wavevector k0 and a specularly reflected plane wave described by kR, during Bragg reflection from an LSM. The standing wave electric field is described by

Figure 2. Angular dependence of the reflectivity R (solid line) and the relative phase u (dashed line) at the solid/vacuum interface for an emersed Pt/C LSM.

C- Electrode

I

Electrolyte

j

Figure 3. Schematic of the model used to describe the iodide distribution formed at the electrode surface.

where d is the LSM characteristic d spacing. The angular dependence of eq l is contained within the variables R ( 8 ) and u ( 8 ) which correspond respectively to the intesity and the phase of the Bragg reflected wave relative to the incident wave. The calculated reflectivity*R ( 8 ) and phase u ( 8 ) at the solid/vacuum interface for a Pt/C LSM (d = 40.8 A) are shown in Figure 2. An important observation to be made from this figure is that rotation of the LSM across the strong Bragg reflection causes a gradual

change of T in the relative phase at a fixed point in the multilayer. This phase change causes the antinodes of the standing wave field to move inward by one-half of a d spacing, from an initial position halfway between the diffracting planes to a final position coincident with them. A single layer of heavy atoms positioned above the LSM surface will witness, as the standing wave moves inward, the passage of an antinode and/or a node in a unique sequence, characteristic of the distance separating the multilayer surface and the overlayer. Since the photoelectric effect for core electrons is directly proportional, in the dipole approximation? to the electric field intensity a t the center of an atom, the emission yield (i.e., the fluorescence yield) from the atoms in this layer will be uniquely modulated as a function of 8. The phase of this modulation is a measurement of the mean position ( 2 ) (so-called coherent position) of the atom overlayer, normal to the diffracting planes of the multilayer, where the (z) scale is modulo of the d spacing. The amplitude of the modulation is a measurement of the width ( z 2 ) ' I 2(coherent fraction) of the distribution of heavy atoms in the overlayer. Coherent position and coherent fraction are model-independent quantities, relevant only when describing symmetrical atomic distributions centered around a given ( 2 ) position.I0 For an extended (i.e., a distribution with width >d), unsymmetrical distribution N ( z ) (Figure 3), coherent position and coherent fraction are inadequate in characterizing the 8 dependence of the standing wave fluorescence yield. To calculate a yield in this case, a model for N ( z ) must be chosen and the standing wave electric

(7) (a) Bedzyk, M. J.; Bilderback, D. H.; Bommarito, G. M.; Caffrey, M.; Schildkraut, J. S. Science 1988,241, 1788. (b) Bedzyk, M. J.; Bommarito, G. M.; Schildkraut, J. S. Phys. Reo. Lett. 1989, 62, 1376. ( 8 ) (a) Parratt, L. G. Phys. Reo. 1954, 95, 359. (b) Bommarito, G. M. M.S. Thesis, Cornell University, 1987.

(9) Wagenfeld, H. Phys. Reo. 1966, 144, 216. ( I O ) Coherent position and coherent fraction could also be defined so as to describe a system consisting of a series of symmetrical distributions. (See: Vlieg, E.; Fischer, A. E. M. J.; Van der Veen, J. F.; Dev, B. N.; Materlik, G. Surf. Sci. 1986, 178, 36.)

I(e,z)= If0

+ fRI2 = IEOl2[1 + R + 2R'I2 COS (U - ~ T Q z ) (1) ]

where f0.R

= E 0 . R exp(i[ot - 2 ( k p - kzz)11

(2)

are the incident and reflected plane waves if their respective wavevectors k0 and kR lie in the x-z plane with the z axis normal to the LSM surface. Q = k0 - kR is the momentum transfer with a magnitude given by

IQI = Q =

1/d

(3)

Bommarito et al.

8282 The Journal of Physical Chemisfry, Vol. 94, No. 21, 1990

field intensity f((3.z)must be averaged over the entire distribution:

1.25

y ( e ) = Jmi(e,z) 0 ~ ( z dz )

1.00

(4)

In analyzing the XSW data from this experiment, we have chosen the simple model depicted in Figure 3 to describe the iodide distribution. This model consists of three basic components: (1) a step at the electrode’s surface extending over a few angstroms, describing the specifically adsorbed iodine atoms, ( 2 ) an exponential tail with a characteristic decay length K , extending out into solution from the iodine adlayer, portraying iodide anions attracted to the electrode surface (the diffuse layer), and (3) a second step, with a width equal to the thickness of the solution layer, depicting bulk iodide. Analytically, this model can be expressed as follows:

0.73 0.50

1.25 1.oo

0.75

where Nadis the concentration of iodine atoms specifically adsorbed,lad is the adlayer thickness, Ndirr is the initial concentration of iodide in the diffuse layer, K is the decay length of this diffuse is the iodide bulk concentration, and fWl is the thickness layer, Nbulk of the solution layer. The choice of an exponential decay to model the diffuse layer was motivated by its mathematical simplicity as well as its frequent usage in simple theoretical descriptions of the electrical double layer.” The standing wave yield can now be calculated by using the distribution N(z) defined above ( 5 ) in the integral (4). Computationally, Nd$fand Nbulkwere expressed as fractional values of Nadand the distribution was normalized by using the condition

0.50 1.25 1.0@

0.75

0.50

Yo, = LldN(z) dz where Yoe is the measured off-Bragg fluorescence yield, a value proportional to the total number of I/I- species present in the solution layer tWl. As a result, the model has three free parameters: and the adlayer thickness tad, the fractional quantity Ndiff/Nad, the decay length K . The remaining parameters are held constant; is known, and the thickness tSolcan be dethe ratio Nbulk/Nad termined experimentally by using the measured X-ray reflectivity. Figure 4 illustrates the effects of varying tad, K , and the ratio Ndiff/Nad on the xsw fluorescence yield. In these caku~ations, X-rays with incident energy of E., = 6.0 keV reflect from the surface of a Pt/C LSM ( d = 40.8 A) in contact with a solution layer 5.24 pm thick, containing a 10 pM bulk concentration of iodide, and encapsulated by a 6 pm polypropylene film. When the model ( 5 ) is simplified to

the standing wave yield can be expressed as (using eq 4 and 6)

where zo is the position of the step’s center, T represents the electric field magnitude of the incident wave at zo, and R and u correspond to the intensity and phase of the Bragg reflected wave relative to the incident wave at zo. Equation 8 describes the yield from an adsorbed layer centered at z = zo with a z-projected concentration Nadlarge enough to neglect contributions to the total XSW yield from the bulk (i.e., Nbulk/Nad is d),sin (aQtad)/(aQtad)changes sign, resulting in the subtraction of the interference term in eq 8. In this case, an interesting effect can be observed: for a particular value of tad, the interference term assumes a value approximately equal to -R, resulting in a nearly linear XSW yield (Figure 4a, dash-dot curve). The effect of including the diffuse layer along with the adsorbed layer in the model ( 5 ) corresponds to a superposition of a randomlike component to the coherent XSW yield from the adlayer (provided that the thickness fad is narrow with respect to the substrate’s d spacing), in a ratio proportional to the population in each layer, with the number of atoms in the diffuse layer being controlled by the decay length K and the initial concentration Ndin (Figure 4b,c). What is important to note is the sensitivity of the XSW technique to these distributional changes. Adding a diffuse layer with a falloff length of only 10 8, produces a dramatic change in both the amplitude and the phase of the calculated signal (Figure 4c). Furthermore, appreciable differences are seen when the diffuse layer population is changed by varying Ndiff(Figure 4b). Thus, in principle, the XSW method is extremely sensitive

Electrosorption of Iodide on Platinum

-

The Journal of Physical Chemistry, Vol. 94, No. 21. 1990 8283

onochromator

Y

c

6

I'iKLSM

A

0

I

,

I

1.o

I

I

0.5

I

I

0.0

l

l

1

1

0.5

I

I

I

J

0.0

Potential, Volts vs Ag/AgCl

s

Figure 5. (a, top) Experimental arrangement. (b, bottom) Goniometer and electrochemical cell assembly used in the in situ X-ray standing wave experiment.

to subtle changes in the atomic/ionic distribution formed at the electrode/electrolyte interface.

Experimental Aspects X-ray Source. The experiment was carried out at the Cornell High Energy Synchrotron Source (CHESS), under parasitic conditions ( 5 GeV, 50-60 mA), using the B beam line and a tunable multipurpose monochromator (MPM) outfitted with double Si( 1 I 1) crystals. The MPM assembly resided inside B cave along with the experimental setup. The incident beam had a height and width of 0.2 and 4 mm, respectively. The vertical and horizontal divergences where 0.1 and 2 m a d , respectively. An X-ray energy of 6.0 keV was selected to excite iodine L fluorescence. The incident X-ray flux was IO9 photons/s. General Experimental Arrangement. The experimental arrangement is shown in Figure Sa. White beam radiation enters the experimental hutch through a thin beryllium window and passes through the double-crystal Si(l11) monochromator. The monochromator resolution was -0.3 eV at 6.0 keV. The emerging beam is collimated in the plane of dispersion by motor-controlled Huber slits. A high degree of collimation is needed due to the geometric constraints of this grazing incidence experiment and to ensure that, to a good approximation, the incident beam is a plane wave. The helium-filled ion chambers labeled monitor and detector measure the intensities of the incident and diffracted beams, respectively, the absolute reflectivity being given by their ratio. A half-slit (not shown) prevents any of the primary beam frm entering the detector ion chamber. A Princeton Gamma Tech Si(Li) solid-state detector was used, in conjunction with an EG&G Ortec Model 673 spectroscopy amplifier and LeCroy ADC histogramming memory modules, to measure the characteristic I L fluorescence yield. The detector provided an energy resolution of approximately 150 eV at 6.0 keV.

Figure 6. Voltammetricprofiles in 0.1 M H,SO, of (A) polycrystalline Pt electrode (100 mV/s), (B) Pt/C LSM (100 mV/s), (C) clean, wellordered Pt( 11 1) electrode prepared by flame annealing followed by quenching in PDW" (50 mV/s), and (D) same electrode as in (C) after I O oxygen adsorption-desorption cycles between + I .20 and -0.28 V at

50 mV/s.

which was more than adequate to separate the fluorescence signals of interest from those of other materials in the system. The detector was placed at a glancing angle to the multilayer's surface, behind horizontal slits used to minimize the background signal. Electrochemical Cell. The electrochemical cell (Figure 5b), contained inside an aluminum housing (not shown), consisted of a cylindrical Teflon body with feedthroughs for electrolytes and electrode connections. This all-Teflon construction allowed cleaning of the cell in chromic acid overnight, followed by a 3 0 " n wash in hot concentrated 1:l H N 0 3 / H 2 S 0 4 solution and a thorough rinse in pyrolitically distilled water (PDW). The filling and rinsing of the cell with electrolyte were accomplished with a pressurized glass vessel through the fluid feedthroughs. A thin-layer of solution (-3-6 pm thick) was trapped between the electrode, and a 6 pm thick polypropylene film which was held in place by a Teflon ring. Whenever the applied potential was adjusted, the film was distended by introducing additional electrolyte, so as to allow for adsorption (desorption) from (into) bulk solution. Removal of the electrolyte solution restored the thin-layer configuration. The aluminum housing was constantly flushed with helium to minimize oxygen diffusion into the solution and scattering. The entire cell, with aluminum housing, was mounted on a Huber 410 one-circle goniometer stage fitted with a 20:l gear reducer which allowed for an angular resolution better than 10 prad. The applied potential was controlled with a Princeton Applied Research Model 173 potentiostat and Model 175 universal programmer. Potentials are reported against a silver/silver chloride reference electrode which was separated from the thin-layer cell by a porous Vycor plug. Sample Preparation and Experimental Procedure. A platinum/carbon LSM of dimension I S mm X 20 mm was obtained from Ovonic Synthetic Materials Co. (Troy, MI). The microstructure was prepared by magnetron sputtering deposition in vacuo. The LSM had a d spacing of 40.8 8, and consisted of 200 layer pairs of C ( 1 3.06 A) and Pt (27.74 A), deposited on a 0.01 5 in. thick Si( 1 1 1 ) substrate, with Pt as the outermost layer. Solutions were prepared with ultrapure reagents (Alfa) and PDW. Prior to introduction of electrolyte into the cell, the solution was degassed with high-purity helium for at least 2 h. Electrode surface pretreatment consisted of oxidation-reduction cycles at a sweep rate of 20 mV/s in the pure supporting electrolyte (0.1 M sodium sulfate at pH = 6.7, phosphate buffer), just prior to the introduction of iodide into the thin-layer cavity. This procedure produced voltammetry in sulfate supporting electrolyte which showed only one pronounced (weakly bound) hydrogen

8284

Bommarito et al.

The Journal of Physical Chemistry, Vol. 94, No. 21, 1990

0

1-

6[ +

250 Layers E=6.0 keV (in-situ R )

TABLE I: Reflectivity Measurements" potential, peak solution layer transmission V , vs Ag/AgCI reflectivity

,_,

LSM d=408 A PL/C

/

i

t '(

, )I

-0.90 -0.45 -0.10 0.15 0.30 0.40 0.49

!

0.142 f 0.003 0.1 12 f 0.003 0.082 f 0.002 0.081 f 0.002 0.132 f 0.003 0.089 f 0.002 0.067 f 0.002

0.582 0.500 0.407 0.404 0.557 0.428 0.358

solution layer thickness, A 30310 38 887 so 377 50815 32 856 47 586 57 655

The polypropylene layer thickness is 63 500 A,the polypropylene layer transmission is 0.614, and the emersed Pt/C LSM reflectivity is 0.509 f 0.003. 24

26

28

30

0 (mrad) Figure 7. Angular dependence of the measured absolute reflectivity (filled circles) for a Pt/C LSM under a solution layer 5.04 pm thick, encapsulated by a 6.35-rm polypropylene film. The dashed line represents the theoretical prediction for the reflectivity when the interfacial roughness of the substrate is neglected. The dash-dot line depicts a calculation attempting to fit the experimental data by increasing the solution layer thickness (to 7.73 r m ) while continuing to neglect the interfacial roughness. The solid line represents the best fit to the data and corresponds to the reflectivity of an LSM under 5.04 pm of solution and with a n rms interfacial roughness of 5.2 A.

adsorption peak (Figure 6B) characteristic of a clean, well-ordered Pt( 1 1 1 ) electrode (Figure 6C) which has been subjected to a few oxygen adsorption-desorption (OAD) cycles (Figure 6D), resulting in a Pt( 1 1 1 ) surface with nearly randomly distributed monatomic steps.'* The surface chemistry and electrochemistry of these multilayers are currently under investigation. After pretreatment, the electrode was exposed to a solution of 0.1 M N a 2 S 0 4(pH = 6.7) containing 1 X M NaI by rapid rinsing of the cell's cavity with the solution. Initially, the electrode was allowed to remain in contact with the solution for 30 min at an applied potential of -0.1 V. (Adsorption at open circuit produced a rest potential of +0.16 V.) Prior to making a measurement, the applied potential was adjusted to the desired value at a rate of 20 mV/s, followed by an equilibration time of about 30 min with the cell in the distended condition. The order in which the potentials studied were chosen was random in order to remove bias from the data. We have previously observed that changes in the applied potential produced essentially reversible changes in the surface concentration of iodine on Pt(l1 The XSW fluorescence yield measurement consisted of monitoring the characteristic IL lines as a function of the angle of incidence (8)of the beam with respect to the sample. The Bragg reflectivity was measured simultaneously. For each scan, the energy-dispersed spectrum at a given angular position was recorded into 256 channels of the histogramming memory module. A typical scan consisted of 64 points over a total angular range of 6 mrad about the first-order Bragg reflection and took approximately 20 min to complete.

Data Analysis The background-subtracted IL fluorescence yield was extracted from each spectrum (in energy dispersed form) by fitting to a linear combination of Gaussians on a quadratic background. The extracted yields from each angular interval in a given scan were then normalized to the live time, to take into account dead time effects, and combined to give the fluorescence yield versus angle profiles presented in Figure 8. The reflection R ( 8 ) and transmission T ( 8 )intensity ratios were calculated next, using a computational scheme based on the framework for stratified interfaces developed by Parratt.s The ( I 2) (a) Aberdam, D.; Durand, R.;Faure, R.;El-Omar, F. Sur/. Sci. 1986, 171, 303. (b) Wagner, F. T.; ROSS,P.N . J . Electroanal. Chem. 1983, 150, 141.

basic requirement of this method is the continuity of the tangential components of the incident and reflected electric field vectors at each interface in the multilayer. These boundary conditions are used to determine the Fresnel coefficients at each slab interface, which in turn are employed in a recursive mode to calculate the coefficients T(8) and R(8) for a given (z) position. This stratified medium formalism allows for the inclusion of the solution and polypropylene layers along with the multilayer in a single model, having the advantage of being exact by taking both refraction and adsorption properly into account. The reflectivity for the bare (ex situ) LSM was fitted to R(8) computed at the solid/vacuum interface, using a nonlinear least-squares routine with interfacial roughness as the only free parameter. The solution layer thicknesses for each potential studied were calculated next, by comparing the reflectivity measured for the bare multilayer to the reflectivity measured in situ. Finally, the experimental fluorescence yields were x2 fitted to the theoretical yields calculated from integral (4) using the model (5) with t,d, Ndlff/Nad, and K as free parameters. For the XSW data of the first potential only (-0.1 V), the Pt surface layer thickness was also allowed to change from its proposed value of 27.74 A to account for possible surface erosion as a result of repeated cycling during the sample's pretreatment. The theoretical yields were also corrected for two geometrical factors. The first took into account the changes in the sample volume illuminated by the finite height of the beam as 8 was increased. The second evaluated the increase in fluorescence as 8 was increased due to the finite solid angle subtended by the solid-state detector.

Results and Discussion Reflectivity Measurements. Reflectivity measurements are valuable in characterizing certain structural properties of the substrate. The measured absolute reflectivity, shown in Figure 7, depends on the thickness of the solution layer covering the LSM and the thickness of the polypropylene film encapsulating this solution layer, according to the simple relation Rabsolute

-- Rsolid/vacuum TpolyproTsoiution

(9)

where Rso,,dd:vacuum refers to the reflectivity of the emersed multilayer (Figure 2) and where Tpolypro and T,ol,,,o,are transmission coefficients expressed as T = exp(-(2t/sin 8)pl

(10)

representing the attenuation to both the incident and reflected beams by the polypropylene film and the solution layer, respect , o24.5 n cm-', tively. p is the linear absorption coefficient ( ~ s o ~ u = ppolypro = 10.5 cm-' at 6.0 keV), t is the thickness of the polypropylene or solution layer, and 2t/sin 8 is the effective thickness traveled by the incident and reflected X-rays at a given angle of incidence 8. For a polypropylene film 6.25 pm thick and a solution layer 5.0 pm thick, X-rays incident at the B r a g angle (27.4 m a d ) traverse 0.1 1 mm of film and 0.36 mm of solution before reaching the LSM surface, losing -60% of their intensity in the process. This illustrates the tremendous penetrative power of X-rays, even at this relatively low energy. The definition for the absolute reflectivity given above (eq 9) can be used to calculate the transmission coefficient TSolution and

Electrosorption of Iodide on Platinum

The Journal of Physical Chemistry, Vol. 94, No. 21, 1990 8285

TABLE 11: Oualitative Analvsis of Iodide L Fluorescence Yields'

potential, V, vs Ag/AgCI

normalized off-Bragg yield

-0.90 -0.45 -0. I O 0.15 0.30 0.40 0.49

0.07 f 0.04 0.42 f 0.02 1 .O i 0.03 1.38 f 0.03 1.20 f 0.04 1.37 f 0.04 1.44 f 0.03

normalized background slope -0.036 f 0.005 -0.040 f 0.003 +0.025 f 0.001 -0.001 f 0.001 -0.016 f 0.001 -0.017 f 0.001 -0.009 f 0.001

angular position of fluorescence peak, mrad 27.88 27.78 27.58 27.48 27.98

f f f f f

modulation amplitude, % of off-Bragg

0.06 0.06 0.006 0.06 0.06

20 f 16f 25 l5f 16f

*

1 1 1 1 1

"The background slope for a random distribution is -0.037, and the off-Bragg yield was corrected for variations in the solution layer thickness with potential. the solution layer thickness, since the reflectivities Rabaolute and RJolidlvacuum are experimentally measured quantities, and the transmission coefficient T lyprocan be accurately calculated since the film thickness is well-rnown (6.25 km). Table I presents the results of this calculation by listing the solution layer thicknesses for each potential studied. The variation in these values is to be expected, since the thin-layer cavity is established by withdrawing the excess electrolyte manually, using a syringe. The absolute reflectivity also depends on the interfacial roughness of the LSM. The net effect of interfacial roughness is to enhance the transmission (penetration) of X-rays through the multilayer. This results not only in a decrease of reflectivity but also in a sharpening of the reflection profile, as illustrated in Figure 7, since the multilayer's reflection width A 8 / 8 is inversely proportional to the effective number of layers participating in Bragg d i f f r a ~ t i 0 n . l ~In other words, interfacial roughness reduces this bandwith A€)/€) because, by increasing the transmitted amplitude of the incident X-rays, it allows more layers in the LSM to partake in diffraction. The interfacial roughness was modeled by multiplying each of the Fresnel coefficients used in the recursion formula to calculate reflectivity, by a Debye-Waller term of the form e x p ( ( - h ~sin

8)/XIz

0.10

(11)

where u is the rms interfacial roughness. In this approach, a Gaussian distribution describes the roughness as a function of depth about the average of the roughness profile. The LSM interfacial roughness was determined to be 5.2 f 0.5 A and did not vary as a function of the applied potential. We point out that this value of 5.2 A does not necessarily describe accurately the surface roughness of the multilayer, since the Bragg reflectivity profile is dominated by the bulk structure and is not very sensitive to the surface condition. Reflectivity measurements in the total external reflection region (Le., around the critical angle) are more suited to a determination of the surface roughness, since in this regime the reflectivity is extremely sensitive to the surface structure because the incident X-ray wave penetrates only a few angstroms into the multilayer. Thus, the interfacial roughness determined for the multilayer used in this experiment should be used only as a lower limit of its surface roughness. Finally, it is important to note that although the measured peak reflectivity (at the polypropylene/vacuum interface) is only -lo%, at the solid/solution interface this peak reflectivity is -50%. Thus, the strength of interaction between the incident and reflected beams is more intense in the solution region proximal to the electrode surface, and the standing wave electric field (eq 1) shows the largest modulations here. As a result, the greatest variations in phase and amplitude will be observed in the fluorescence yield profiles for species accumulated close to the surface, which is the region of interest. Standing Wave and Off-Bragg Yield Measurements. The IL fluorescence yield as a function of the angle of incidence 8, for each of the potentials studied, is shown in Figure 8. As we discussed previously, changes in both the phase and amplitude of the standing wave signal are indicative of distributional changes (13) Bilderback, D. H. Nucl. Instrum. Methods 1983, 208, 251.

(a)

0

0

/,eB:, 25.0

R

0

00

, " I

,

L,

27.5

0.00

30.0

8 (mrad) Figure 8. (a) Angular dependence of the experimental I L fluorescence for the various applied potentials studied. (b) Experimental reflectivity at 6.0 keV for the polypropylene/solution/Pt/C LSM system.

in the direction normal to the substrate surface. An advantage of the standing wave method lies in the ability to attain qualitative structural information about the system from a simple inspection of the phase and amplitude for each fluorescence profile. Thus, we begin by describing each standing wave profile in terms of background slope, peak position, and modulation amplitude, to glean an initial picture of the I-/I distributional changes as a function of applied potential. Referring to Figure 8, we observe different values of the offBragg fluorescence yield for the various applied potentials investigated. These differences correspond to changes in the total amount of I-/I species sampled by the incident and reflected beams. We observe the lowest off-Bragg yield at an applied potential of -0.9 V (vs Ag/AgCI). At this potential, all iodine adatoms are reductively desorbed from the electrode surface, and the measured fluorescence yield originates essentially from bulk iodide anions. (The bulk iodide concentration is 10 k M . ) Due to the low yield measured in this condition, we will neglect bulk contributions to the overall fluorescence yields in the discussion of the other potentials studied. As the applied potential is made progressively more positive, the off-Bragg yield grows, indicating a substantial increase (-100-fold) in the amount of I-/I species present in excess of the bulk iodide concentration. The off-Bragg yield will be discussed in more detail later in this section. The background slope, the relative off-Bragg yield on the low and high angle sides of the Bragg reflection, also shows a marked potential dependence (Table 11). At the more negative potentials

8286

The Journal of Physical Chemistry, Vol. 94, No. 21, 1990

Bommarito et al.

TABLE 111: Model Parameters (See Figure 3) Determined by Nonlinear Least-Squares Fitting of the Iodide L Fluorescence Yield Profiles’ adlayer fractional diffuse layer normalized

potential, V , vs Ag/AgCI -0.90 -0.45 -0. IO 0.15 0.30 0.40 0.49

thickness

concentration

t(ad),

N(diff)/N(ad)

10.0 f 10.5 f 11.4 f 10.3 f 10.8 f

A

1.7 1.8 2.0 1.6 1.8

0.095 f 0.005 0.039 f 0.003 0.0081 f 0.0021 0.0053 f 0.0016 0.020 f 0.002

decay length k, A

50 f 115 f 588 f 824 f 214 f

7 18 80 123 35

adlayer coverage 0.07 f 0.04 0.42 f 0.02 1 .oo f 0.02 1.26 f 0.02 1.19 f 0.03 1.25 f 0.02 1.32 f 0.02

“The thickness of the topmost Pt layer in the LSM was determined to be 16.0 f 0.5 A. (-0.9 and -0.45 V) this slope is negative, and its magnitude is characteristic of a random distribution of species in the solution layer. In this case, the off-Bragg fluorescence yield is proportional only to the illuminated volume of solution, which in turn varies as l/sin 9. Most notably, for an applied potential of -0.1 V the background slope becomes positive. For all of the remaining potentials (+0.15 to +0.49 V) there is a negative slope of varying magnitude, but which never approaches the value expected for the random case. In addition, the peak in each fluorescence yield curve does not occur at the same angular position (Figure 8) but shifts smoothly from a position on the high side of the Bragg angle (e,) for -0.1 V to a position approximately equivalent to 9, at applied potentials of +0.3 and +0.4 V, and finally back to an angle greater than 9, at +0.49 V. Furthermore, the amplitude of the modulation in each of the fluorescence yield profiles changes with potential, and despite the fact that the differences are subtle and a trend is less apparent, it is clear that the largest modulation amplitude is observed at +0.3 V, while the smallest modulation occurs for an applied potential of +0.49 V (neglecting data at -0.9 and -0.45 V). All of the trends observed above can be understood in terms of changes in the distribution of iodine/iodide species at the electrode/electrolyte interface. We simplify such a distribution by assuming a primitive model (Figure 3) subdivided into three portions: specifically adsorbed iodine atoms, iodide anions in a diffuse layer with a characteristic decay length, and bulk solution iodide. The overall distribution is then defined in terms of the concentration of species in each one of these regions. Each of the three pieces in this model makes a distinct contribution to the overall standing wave fluorescence yield. As we discussed previously, the standing wave fluorescence yield originating from a narrow adlayer at the electrode surface will possess the richest phase information and will exhibit the largest amplitude modulation (Figure 4a), because ( I ) due to the strong interaction between the incident and reflected waves occurring at the solid/solution interface, this adlayer experiences the largest modulations in the standing wave electric field, and (2) since the spread of the adlayer is much smaller than the standing wave periodicity, smearing of the modulation in the fluorescence yield is reduced to a minimum. I n contrast, the standing wave fluorescence yield emanating from a diffuse layer with a large decay constant will lack phase and amplitude information (Figure 4c). This is a consequence of the standing wave periodicity becoming smaller than the decay length of the diffuse layer, resulting in a washing out of the interference effects in the standing wave signal. A diffuse layer with a small decay length, compared to the standing wave period, will produce a yield that is a compromise between the two limiting cases just described. Thus, for the type of distribution studied here, we can qualitatively think of the overall standing wave fluorescence yield as a superposition of a coherent yield from the ordered, crystallinelike arrangement of iodine in the adlayer and a random yield from the disordered, liquidlike arrangement of iodide in the diffuse layer and bulk. The ratio in which these yields are added is roughly dictated by the concentration of species in each component of the distribution. The background slope is representative of the ratio in which these random and coherent components of the standing wave yield are added. For a strictly random distribution we expect this slope to vary as 1 /sin c); deviations from this dependence point to the

presence of a coherent component in the distribution (Le., an adlayer). Furthermore, the magnitude of such deviations is an indication of the preponderance of this coherent phase. Thus, it is clear that (see Table 11), for all applied potentials studied with the exception of -0.9 and -0.45 V, there is a strongly bound, coherent layer of iodine on the electrode surface. This conclusion is also supported by the considerable increase in the magnitude of the off-Bragg yield when increasingly positive potentials are applied. In addition, the suppression of oxide formation on the platinum surface of the LSM observed during cyclic voltammetry also indicates strong adsorption of iodide. Furthermore, the rapid attainment of a condition of essentially zero current flow at all but the most positive potentials studied is indicative of the establishment of an adsorption-desorption equilibrium and the formation of a double layer with some definite distribution of species. Given that an adlayer of iodine exists, what is perhaps most surprising is the observation of small modulation amplitudes in the standing wave profiles. After all, if the overall distribution of iodide was comprised of an adlayer at the surface and bulk solution, we would expect these amplitudes to be large (- 100%) and sharp (Figure 4a). We conclude from this observation that the iodide distribution also contains a diffuse tail of considerable length, extending out from the electrode surface. In particular, the standing wave yield at -0.1 V, which shows the highest degree of coherence, depicts an iodide distribution consisting of an adlayer and a narrow diffuse layer. As the potential is made more positive this diffuse layer seems to extend further out from the surface, progressively transforming the standing wave yield into one more characteristic of a random distribution. However, even for applied potentials where this effect is most noticeable (+0.3 and +0.4 V), the yield retains some coherent information and is never completely random. Since the contribution from the diffuse layer to the overall yield is notable, the total amount of iodide contained in this layer must be significant in comparison to the surface concentration of iodine adatoms. Interestingly, the diffuse layer appears to stop expanding and starts shrinking back at potentials past +0.4 V, as evidenced by the standing wave yield at +0.49 V where we observe a nearly horizontal background. The variation in the angular position of the fluorescence peak maximum is consistent with the qualitative picture given above. Recalling that the standing wave electric field moves inward by half of a d spacing in going from the low to the high angle side of the Bragg reflection, the movement of the fluorescence peak to lower angles as the potential is increased indicates that, on average, the interfacial iodide density is moving away from the electrode surface. Furthermore, the return of the peak maximum to the high angle side of e, for a potential of +0.49 V points to a contraction of this iodide density. To more precisely quantify the qualitative results presented above, we have x 2 fitted the data to theoretical yields based on the model defined in eq 5. The results are presented Table 111 and Figure 9. The adlayer thickness tad is, statistically, the least meaningful parameter in the analysis. Since the observed emission is a core phenomenon, a perfectly planar adlayer on an atomically smooth electrode surface could be theoretically modeled as a 6 function. Thus. the finite thickness tad evaluated here is primarily a result

The Journal of Physical Chemistry, Vol. 94, No. 21, 1990 8281

Electrosorption of Iodide on Platinum 0.10

I

1000

T

E

-2

125-

e I

025

0.00 0.25 Potential (V vs Ag/AgCl)

0.50 9 ,

Figure 9. Plot of (a) the diffuse layer decay length K (filled circles), the fractional concentration Ndim/Nad (open circles), and (b) the normalized adlayer coverage (filled diamonds) as a function of applied electrode potential. The inset plots the normalized adlayer coverage at the extreme negative potentials studied.

of the rms surface roughness of the platinum surface. More significant are the variations observed in the two remaining free parameters: the fractional concentration Ndiff/Nad (where Ndinrefers to initial number concentration of iodide in the diffuse layer and Nadis the number concentration of adsorbed iodine adatoms), and the decay length K. The potential dependence of these two parameters is shown in Figure 9a, where we observe the decay length increasing smoothly from -0.1 to +0.4 V, followed by an abrupt decrease at the highest potential investigated. In contrast, we note exactly the opposite behavior for the fractional concentration. What is most remarkable is the magnitude of the variations in these parameters with potential, especially in light of the fact that iodide is present in solution in a ratio of 1 to 10000 relative to the sulfate dianion in the supporting electrolyte. Thus, we are able to observe tremendous distributional rearrangements of a very small amount of iodide, over considerable length scales, with angstrom resolution. It should be noted that the observed changes in the decay lengths of the diffuse layer are opposite to those expected from electrostatic arguments, from which we would anticipate a decrease in the decay length as the surface charge becomes more positive. However, we would also expect that the very large excess of the dianion in the supporting electrolyte, in addition to being largely responsible for charge screening effects, will respond much more rapidly to surface charge changes than the iodide ion. Furthermore, primitive double-layer models that neglect interion interactions are invalidated by the high ionic strength of our system. Figure 10 shows two typical theoretical fits and points out the noticeable differences produced in the fluorescence yield by two different iodide distributions. Referring to this figure, if the diffuse layer had been a negligible component of the iodide distribution, the observed fluorescence yield would have followed the theoretical yield calculated for an adlayer alone. On the other hand, if the bulk or diffuse layer concentration was too great, or the adsorbed layer was too thick or not present at all, the fluorescence data would have looked like the yield calculated for a random distribution. The best fit of the observed yield, in both cases, is a compromise between these two extreme cases. The changes in adlayer normalized coverage with applied potential are also listed in Table 111 (see also Figure 9b). These values were obtained by taking the off-Bragg yield at each potential and subtracting the contribution due to the diffuse layer, as determined from the fitting parameters K and Ndiff/Nad. The

c

24

--------,/

26

p 28

,LU

30

8 (mrad) Figure 10. Theoretical fits (solid lines) at (a) +0.3 V and (b) -0.1 V. Also shown, on both plots, are the theoretical yield predicted for an iodine adlayer (10 A thick) alone (dashed lines) and the yield expected from a random distribution of iodide in solution (dash-dot lines).

Potential (Vvs Ag/AgCl) Figure 11. Comparison of iodide electrosorption isotherms recently reported in the literature. Plotted data are taken from this work (diamonds), ref 4 (open circles), ref 5 (closed circles), and ref 13 (squares).

variations in the normalized coverage are analogous to pa'cking densities observed on Pt( 1 1 1) from dilute NaI solution by in situ X-ray absorption spectroscopy5and on emersed Pt( 11 1) by Auger electron spectroscopy4 (Figure 11). In these cases, these changes are attributed to structural changes in the adsorbed iodine adlayer. In particular, referring to Figure 11, a relative packing density increase of approximately 20% is observed between -0.1 and +O.O V. This difference is ascribed to a structural transformation from a ( d 3 X d 3 ) R 3 0 ° (0, = 1/3) iodine adlattice to the higher density (v'7Xv'7)R19.1° (0, = 3/7) adlattice. The maximum coverage achievable by closest packing of iodine (2.15-A van der Waals radius) corresponds to a packing density 0, = 0.44. In contrast, similar coverage measurements performed at a polycrystalline Pt electrode show that a saturation coverage of iodine (0, = 0.44) is attained for potentials as negative as -0.3 V.I4 Furthermore, (14) Rodriguez, J. F.; Bravo, B. G.;Mebrahtu, T.;Soriaga, M. P. Inorg. Chem. 1987, 26, 2760.

8288

The Journal of Physical Chemistry, Vol. 94, No. 21, 1990

. iT

-N

z

"F 0.0 1

r i

Figure 12. Illustration of the changes in the iodine/iodide distribution (eq 5 ) as a function of the applied potential as determined from x 2 fitting of the 1 L fluorescence yields (Figure 8).

Auger spectroscopic studies of iodine adsorbed on a Pt stepped surface (Pt(s)[6( 1 1 I)X(l1 I ) ] ) from a beam of I2 vaporI5 reveal that a Pt(s)[6(1 1 I)X( I 1 1)](3XI)-I adlattice is formed, with a packing density Oi = 0.49. At elevated temperatures (1 25 "C), the (v'3Xv'3)R30°-I adlattice is also observed, but a structure similar in packing density to the (v'7Xd7)Rl9.l0-I was never seen on this stepped surface. Since the changes observed in the normalized coverage in this experiment are in excellent agreement with the changes in the packing density measured on a Pt( 11 1) electrode surface rather than on polycrystalline Pt or a stepped Pt( 11 1) surface, we believe that a similar structural transition takes place for the iodine adlayer formed at the Pt surface of the Pt/C LSM employed in this work. To support this conclusion, we also point out that the Pt(1 11)-(v'7Xv'7)R19.1°-I surface was found to be remarkably hydrophobic: unable to retain measurable amounts of ionic species, whereas the Pt( 1 1 1)(v'3Xv'3)R30°-I surface was observed to be hydrophilic, showing normal retention of ions and solution films. The same trend is evident (Figure 9) in this work by observing the changes in the fractional concentrations Ndif/Nad, where we note an abrupt decrease in this parameter (indicating a decrease in the ability of the iodine-covered Pt surface to retain ionic species) in correspondence with the increase in normalized coverage. I n addition, these observations and the voltammetric measurements mentioned previously indicate that the Pt surface of these LSMs appear to have a morphology more characteristic of a Pt( I 1 1) surface than a polycrystalline Pt surface or a miscut Pt( 1 1 1 ) surface. In view of the conclusions drawn from analysis of the standing wave and off-Bragg yield measurements, we depict the potential dependent 1/1- distributional changes in Figure 12. From this (15) Solomun, T.;Wieckowski, A.; Rosasco, S.D.; Hubbard, A. T. Surf, Sci. 1984, 147, 241

Bommarito et al. quantitative picture, we note a marked accumulation of iodide in the diffuse layer, weakly associated with the adsorbed iodine, when the Pt surface is not saturated by iodine adatoms (-0.1 V). Perhaps, this striking association of iodide with the iodine adlattice is driven by the hydrophilic character of the unsaturated Pt surface. Furthermore, the increase in the adsorbed iodine packing density to saturation coverage is accompanied by an abrupt decrease in the concentration of this accumulated iodide. Thus, we suggest that the potential-dependent structural transformation in the iodine adlattice could be viewed as a phase transition, in which iodide anions in the liquidlike arrangement of the diffuse layer are incorporated into the crystallinelike structure of the iodine adlattice, resulting in a saturated Pt surface, possibly hydrophobic, and the concomitant decrease in the concentration of iodide in the diffuse layer, associated with the adsorbed iodine. The observations and conclusions made in this study point to several of the strengths and weaknesses of the standing wave technique and its employment in the study of electrochemical interfaces. Perhaps, the biggest drawback consists of the fact that this method only yields structural information in the direction normal to the substrate surface. Thus, to gather a complete structural picture, we must combine this type of measurement with a technique yielding accurate information of the in-plane structure. To this extent we are currently pursuing in situ grazing incidence X-ray diffraction measurements for this same system. In this work, we were able to follow, in situ, distributional changes for a minute amount of an ionic content, with angstroms resolution in the direction normal to the surface. The model chosen to describe this distribution was primitive; more complex models, using liquid-pair distribution functions, might describe the diffuse layer more rigorously. However, to discern unambiguously the value of each of the many parameters used to define these types of distributions, a collection of standing wave measurements, over higher order Bragg reflections as well as the total external reflection regime, must be available for analysis. In this regard, we are carrying out standing wave experiments, of this and similar systems,I6 in the total external reflection region.

Conclusion We have employed fluorescence-detected X-ray standing waves to describe, in situ and with angstroms resolution, the potential-dependent density and distributional changes of iodide at the Pt surface of a Pt/C LSM with a 40-A d spacing. We explain the observed variations using a simple model consisting of an adsorbed adlayer and an exponentially decaying diffuse layer. We suggest that the differences in the adlayer packing density could correspond to a phase transition. This study shows the feasibility of examining structural changes at the electrode/electrolyte interface using the X-ray standing wave technique. Acknowledgment. This work was supported by the Materials Chemistry Initiative of the National Science Foundation and by the Office of Naval Research. H.D.A. is the recipient of a Presidential Young Investigator Award (1984-1989) and a SIoan Foundation Fellow (1987-1991). The authors are grateful for the contributions made by M. J. Bedzyk, D. Acevedo, and J. F. Rodriguez. (16) Bedzyk, M. J.; Bommarito, G.M.; Caffrey, M.; Penner, T. Science 1990, 248, 52. (17) Clavilier, J. J . Electroanal. Chem. 1980, 107, 21 1.