Optical properties of (AgI)x-(Ag2O-2B2O3)1-x glasses

: Applied

Surface Science 65/M

(1993) 302-307

applied surface science

North-Holland

Optical A. Arena

Received

properties

30 June IYY2: accepted

The optical properties

‘, A.M. Mezzasalma

for publication

of (AgI),-(Ag~O~~lB?OI),

by means of spectrophotometric elements

of (AgI),-(Ag20-2B,O,),

‘, G. Galli “, R. Giorgi

techniques

2 October

_-x

‘I, S. Patane

within the investigated

‘I and G. Saitta ”

~, samples with II = 2 and H valuer ranging from 0.U to 11.6. are lnvcrtigatrd

and reflection

electron

useful information

energy loss spectroscopy.

The

used in conjunction

relative

concentration

with ion sputtering

to be gained ahout the considerable

diffusion

of the

in order

10

of the silver ion\

solids.

1. Introduction

The possible applications as second species conductors in batteries as in fuel cells, together with their capabilities for studies on mass transport in solids, have enhanced the interest on the “superionic” materials. Besides the superionic crystals such as the AgI [1,2] or the mixed-phase superionic polymers as PEO (poly ethylene oxide) and its complexes with suitable salts [3] and others [4], the network glasses doped with metalhalide salts have been the subject of many investigations in the last years, owing to their characteristic room-temperature high ionic conductivity (about IO-’ S/cm). Inorganic fast ion conductivity glasses are generally obtained from an oxide glass former (such as SiO:, PzO, and B,O,) and a network modifier such as Ag,O that together with the glass former can dissolve suitable amounts of the metal-halide salts (e.g. AgI). In (AgI),-(Ag,O-nB,O,), ~., glasses, B20, acts as a glass former; the addition of silver oxide increases the coordination number of some boron atoms, enhancing the n-dependent ratio, between the population of the BzO, and the B,O, units, allowing the Agl salts to be dissolved into the OlhY-~332/93/$0h.O0

glasses

lYY2

in the samples has been tested by means of XPS. This technique.

remove the first surface layers, have allowed

..,

z

1903

Elsrvier

glassy matrix. The main interest of the studies on the electrical conductivity of these materials has been about the dynamics of the silver cations. that have been acknowledged as being rcsponsiblc of the electric current transport [5]. Up to now. the microscopic origin of the high conductivity of these materials has been classified, but some questions remain still unresolved, in particular with regard to the role played in the time and compositional stability of the glasses by the “mobile” (of AgI origin) and the “immobile” (01 Ag,O origin) cations. The question of the origin and the amount of the migration of ions throughout the glasses, has been largely discussed. In particular, it is not clear if all cations present in the matrix are equally mobile with a tempcraturcdependent mobility (strong elcctrolytc model [(,I). or whether only a temperature-dcpendcnt fraction of the existing cations is mobile (weak clcctrolytc model [7]). In this paper. the strong dift‘usion of the silver ions towards the external surfaces of some (AgI),-(Ag,O-?B,O:), , glasses with x = 0.0, _I-= 0.3 and .I- = 0.6 has hccn cvidented by means of two high surface sensitivity techniques: reflection electron energy loss spectroscopy (REELS) and optical rctlectivity mea-

Science Publishers B.V. All rights rr\ervcd

A. Arena et al. / Optical properties

surements. In order to gain information about the relative concentration of the components of the glasses going from the first surface layer to the bulk, X-ray photoelectron spectroscopy (XPS) measurements have been performed after the samples have been exposed to argon ion bombardment.

of (AgI)~-(Ag,O-2B,O.,)t

features and the secondary electron can be represented as follows:

N(q,

E) =B

REELS measurements were carried out using a VG UHV chamber equipped with a VG LEG61 electron gun, under a base pressure of approximately 7 X lo-“’ mbar. The counting rate was a few hundred counts per second with a beam current of about 10 mA. The primary energies used range from 0.4 to 2.0 keV. The scattered electrons were collected over a 6” solid acceptance angle by a VG CLAM100 single-channel hemispherical analyser set for a constant pass energy of 10 eV. XPS measurements were performed using an ESCALAB MkII VG system, at a base pressure of 1O-9 mbar. The Al Ka line at 1486.6 eV was employed as excitation source. Counts were collected at intervals of 0.5 eV with a pass energy of 50 eV, giving an effective resolution of 1.2 eV. The samples were sputtered using Ar ion bombardment, carried out at 3 keV accelerating voltage. During this operation, the dynamical pressure was of the order of 5 X lo-’ mbar. The low-energy optical behaviour of the samples was investigated by means of optical reflectivity measurements performed over an energy interval ranging from 1.1 to 6.5 eV, using a Perkin-Elmer mod. Lambda2 UV-VIS-NIR spectrophotometer equipped with two reflectance accessories.

3. Results and discussion In the frame of dielectric theory [g-11], the energy distribution of the electrons scattered from a solid under electron beam irradiation, neglecting the zero loss peak, the multiple scattering

background,

Im ( E(9,

E) i -1

+S 2. Experimental

303

~ glasses

111;1:>

E)+J

where the scattering parameters 4 and E are the electron momentum and the electron energy transfer, respectively. The first term on the righthand side of the above expression is termed bulk loss function, since it describes the single energy loss events within the bulk of the material, while the second term accounts for the surface losses. Apart from the two scale factors B and S, depending on the geometry and on the nature of the sample [121, the response of the medium to the electron probe is completely characterized by ~(q, E), a function that, in the dipole approximation scheme and in the limit of low momentum exchanges, coincides with the optically defined complex dielectric function. Expression (1) states that an electron probe as well as a photon beam can in principle be used to investigate the optical behaviour of a medium. The existence of low-cost and high-performance electron technology, able to investigate a wide energy range overcoming at the same time most of the difficulties proper to the conventional optical methods, have stimulated considerable attention towards the use of REELS as a tool for determining the dielectric response of solids. Several techniques have been proposed in order to extract the bulk single scattering distribution from the REELS spectra [13-221. Most of these techniques involve either the use of a step-by-step elimination of the spectral features other than those of interest or a numerical best fitting of the whole experimental data. However, a general procedure able to separate into single contributions the electron energy loss spectra, does not yet exist. The question is not trivial, since the problem of deconvolving the REELS spectra that usually appear in the valence energy region as superposition of unresolved structures, is complicated further because of the not straightforward

the theoretical energy loss was assumed to be the sum of a number of arbitrary positions, intensities and widths, bell-shape structures, representing the zero loss peak and the other main loss features, superimposed on a featureless secondary electron background. The elastically scattered electron peak was modelled by a gaussian function. The secondary electron background was dcscribed by means of an exponential law suggested by Van Attekum [21]: B(E‘)=cl{l-hexp[-(E-c)/d]].

(2) The bulk and surface single and multiple scattcring distributions were assumed to consist of a sum of Drude-type functions described by the following expression:

0.41

-.

0

.

5

.

10

.

’

15

Energy

-

’

-

20

Loss

m.

25

m .

30

r

35

]

40

(eV)

Fig. I. Details of the REELS spectral analysis: reflection electron energy loss spectra of the (A&-(Ag>O-2Bz0,), , samples recorded at a primary energy of 2.0 keV (dotted line): the best fit (solid line); the single xattering distribution (dashed line); the inelastic background (dot-dashed line): (a) for x = 0.0; (b) for x = 0.3; (c) for x = 0.6.

dependence of the bulk and surface loss profiles upon the geometric and instrumental factors. Fig. 1 shows the reflection electron energy loss spectra recorded bombarding with a 2.0 keV electron beam the (Ag,O-2B203), (Agl),,.3-(Ag20and (AgI),,,,-(Ag20-2B203)(,.4 sam2B203)0.7 ples. The approach we choose in order to isolate the bulk loss function from the experimental spectra, involves the use of a least-squares analysis which minimizes the sum of the squared differences behveen the recorded spectrum and a theoretical estimation of the energy loss intensity. In correspondence with each experimental point,

where the coefficients A,, E, and f; rcprcsent the intensity, energy position and half-width at half-maximum of each structure. The floating parameters of the reconstructed spectrum include the zero loss peak and the main hulk plasmons intensities, positions and widths, together with the background parameters and the intcnsitics which account for double-scattering excitations. Only the energy positions of the first-order surface structures were allowed to vary. with the constraint that they be comprised between E,/fi and E,,, E,, being the position of the bulk counterparts, while the intensities of the surface and bulk scattering were assumed to bc equal, apart from a scale factor. At each iteration the reconstructed energy loss distribution has been corrected using the spectrometer response weight functions R,, and R,, obtained by integrating upon the spectrometer acceptance angle 0,, the bulk and surface single scattering distribution per unitary solid angle. Assuming that H is the inelastic scattering angle and defining 19, as the ratio between the energy loss dE and twice the primary energy E,, the weight functions assume the following expressions

A. Arena et al. / Optical properties of (AgI),-(Ag,O-2B,O,),

where g(B, (Y, f3,) is the function which accounts for the dependence of the surface loss profile upon the incidence angle (Y[22]. Once the bulk single scattering distributions have been extracted from the REELS spectra, they are normalized by using the low-energy complex dielectric function as deduced from optical measurements, and then submitted to a conventional Kramers-Kronig analysis. The analysis was carried out iteratively, following the procedure suggested by Ohno [1.51, until a good agreement between the REELS and optically derived complex dielectric function was achieved. The complex dielectric functions derived for the three (AgIl,-(Ag,O-2B,O,), _-x samples are reported in fig. 2. Fig. 3 shows the low-energy complex dielectric function as deduced by REELS for the sample with x = 0.0. The same picture reports as a comparison the E, and Ed functions of the same sample, as deduced by means of optical reflectivity measurements. Taking into account the fact that the energy resolution typical of the optical analysis is at least two order of magnitude greater with respect to REELS, the shape and peak positions obtained from electron energy loss spectroscopy seem to reproduce quite well the optical results. The electron energy loss analysis revealed the presence in the bulk loss function of all the investigated samples of a group of structures centred below 4 eV. The peaks found by the best fit approximately at 3.7 and 3.4 eV, respectively, can be identified as the first bulk and surface collective excitations characteristic of silver [23,24]. The collective character of the bulk peak is confirmed by the behaviour of both the optically and REELS derived real and imaginary part of the complex dielectric function, which intersect at about 3.4 eV, as is shown in figs. 2 and 3. The heights of the 3.4 and 3.7 eV peaks relative to the zero loss peak are quite independent from the x values; this suggests that independently of the AgI nominal concentration, the migration of silver ions from the bulk creates small silver-rich islands at the surface [25]. This result agrees with the experimental indications given by Carini et al. [26] about the presence in the AgI doped glasses, of metallic halide clusters in the form of mi-

305

x glasses

--__.-mm__

,

01.. 0

5

10

15

Energy

20

Loss

25

30

35

40

(eV)

Fig. 2. Real part (solid line) and imaginary part (dashed line) of the REELS derived complex dielectric function of the (AgI),-(Ag,O-2B,O,), ~~‘i sample: for (a) x = 0.0; (b) for X = 0.3: (c) for x = 0.6.

crodomains weakly bonded to the glassy network and interconnected among them 1271. It has been suggested [28,29] that it is just this interconnection between the silver-rich domains that is the mechanism which provides the conductivity pathway, through the jump-diffusion of the silver cations. The presence of silver in great concentration detected by REELS also at the surface of the binary AgZO-2B,O, sample demonstrates that together with the AgI originated cations, also the Ag,O originated silver ions are quite free to migrate within the glassy matrix. This result is in agreement with the conclusions reached using the NMR analysis by Martin et al. 1301, that at room temperature, at least two types of mobile silver ions exist: one stationary on the time scale of the

NMR, and the other responsible for fast ionic diffusion at low temperature. No evidence of the silver plasma resonance located at 7.5 eV is found in the complex dielectric functions of the investigated samples. Probably this plasma excitation is shifted at higher energies, as a consequence of the presence of a strong interband transition between 5 and 8 eV. According to the complex dielectric function behaviour, the sharp increase in the experimental energy loss intensity above 10.5 eV, can be interpreted as due to a second plasma resonance. Finally, the migration of silver ions throughout the (Agl),-(Ag,O-2B,O,), ox samples, evidenced by the electron energy loss measurements, had been investigated also by means of XPS measurements, performed repeatedly after the samples had been exposed for 60, 120, 240 and 400 s to the argon ion bombardment. In fig. 4 are reported the wide XPS spectra for the three samples recorded after 120 s or argon beam irradiation. Approximately at 286 eV, the C 1s photoelectron peak can be noted; it is reduced but not completely eliminated after the ion bombardment: the presence of carbon as the principle contaminant, can probably be related to the delicate thermal treatment the samples have suffered during their preparation. The experimental spec-

s

12

t”.“““““““““““““j

1

2

II,

3 Energy

4 Loss

5

6

7

(eV)

Fig. 3. Low-energy real part (solid line) and imaginary part (dashed line) of the REELS derived complex dielectric function of the Ag,O-2B,O, sample. At the right top of the picture, are reported the E, and Ed functions of the same sample as deduced by optical methods.

a)

0 I‘ AI 3d

60;

100

200

300 Binding

400 Energy

I

500

600

700

(eV)

Fig. 4. XPS wide spectra of the (Agl), -(AgzO-2B:C),), samples: (a) for I = 0.0; (b) for .Y= 0.3; Cc) for x = 0.6.

I

tra have been fitted using an exponential law for the inelastic background distribution and Voight line shape for the photoelectron peaks. The analysis of the Auger parameter, determined by using the binding energy of the Agjd,,, photoelectron peak and the kinetic energy at which the AgM,VV Auger transition was detected in the X-ray-excited Auger electron spectra performed by us, confirmed the presence at the surface of the unsputtered samples of metallic silver. The Ag3d peaks seem to shift towards lower binding energy with increasing irradiation time. This can probably be explained assuming that the metallic silver detected in the first surface layers of the unsputtered samples, has been eliminated by the argon ion sputtering, thus the scans performed deeper in the sample. detect only Ag,O instead of metallic silver. The areas under the B Is peak

0.30

0.20 E T 2 0.10

0.00 0

100

200

300

Exposition

Time

400

500

bed

Fig. 5. Silver-to-boron concentration ratio. as a function of the exposure time to argon ion bombardment: x = 0.0 ( + 1; s = 0.3

(a ). .Y= 0.6(0). and the Ag3d peaks, have been normalized using the atomic sensitivity factors [31] and their ratios have been used to check the variations of the elemental relative concentrations with depth. The silver-to-boron concentration ratio as a function of the irradiation time, is reported in fig. 5. The results indicate that in all three samples the silver content with respect to boron decreases as the exposure time to the ion flux increases, confirming the existence of thin silver surface layers also in the binary Ag,O-2B,O, sample.

References [I] L.W. Strock, Z. Phys. Chem. Abt. A 325 (1934) 441. [2] H. Wiedersich and S. Geller, The Chemistry of Extended Defects in Non-Metallic Solids (North-Holland, Amsterdam, 1970). [3] J.J. Fontanella, M.C. Wintersgill. J.P. Calame and C.G. Andeen, Solid State Ionics B 8 (1983) 333. [4] M.J. Rice and W.L. Roth. J. Solid State Chem. 4 (1972) 294. [S] G. Chiodelli. A. Magistris, M. Villa and J. Bjorkstam. J. Non-Cryst. Solids 51 (1982) 143.

[h] O.L. Anderson and D.A. Stuart, J. Am. Ceram. Sot. 37 (1954) c73. [7] D. Ravaine and J.L. Souquet. Phys. Chem. Glasses IX (19771 27. [X] H. Frohlich and H. Pelrer. Proc. Phys. Sot. London A 6X (19551 52s. [Y] P. Nozieres and D. Pines. Phys. Rev. I13 (19591 1254. [lo] R.H. Ritchie, Phys. Rev. 106 (19.57) 874. [I I] R.11. Ritchie, Phya. Rev. I14 (19591 644. [12] G. Chiarello. E. Colavita, M. De Crescenzi and S. Nannarone, Phys. Rev. B 29 (1984) 4X7X. [13] G. Mondio. F. Neri. M. Stocker. T. Janssens. J.R. Castro and K. Wandelt, Surf. Sci. 251/252 (1991) 243. (141 J.C. Ingram, K.W. Nebesny and J. Pemberton, Appl. Surf. Sci. 44 (I 990) 279. [IS] Y. Ohno, Phys. Rev. B 39 (1989) X209. [I61 Ch. Kleint and M. Funke, Appl. Phys. A 46 (1988) 137. [ 171 F. Yubero. J.M. Sanz, E. Elizade and L. Galan. Surf. Sci. 237 ( 1990) 173. [IX] G. Marletta. F. Iacona, G. Mondio, F. Neri, S. Patane and A. Arena, Thin Solid Films 207 (1992) 313. [ IY] Ch. Kleint and S. Mahmoud Abd El Halim, Surf. Sci. 247 (1991) 37s. [20] S. Tougaard and I. Chorkendorff. Solid State Common. 57 (19X6) 77. [2l] P.M.Th. Van Attekum and J.M. Troostcr. Phys. Rev. B IX (197X) 6570. [22] R.F. Egerton, Electron Energy-Loss Spectroscopy in the Electron Microscope (Plenum. New York, 1986). (231 J.C. Ingram, K.W. Nebesny and J.E. Pemberton, Appl. Surf. Sci. 44 (I9YOl 293. [24] F. Yubcro, J.M. Snnz. E. EliLada and L. Galan. Surf. Sci. 251/252 (I991) 296. [25] S.W. Martin. Solid State Ionics 51 (1992) 19. [2h] G. Carini, M. Cutroni. M. Federico, G. Galli and G. Tripodo, Phya. Rev. B 320 (1984) 7219, [27] J.P. Malugani and R. Mercier, Solid State Ionics I3 (1984) 293. [2X] L. Borgjesson, L.M. Tore], U. Dohlbarg and W.S. Howells, Phys. Rev. B 39 (1989) 3404. (291 G. Dalba. P. Fontana, A. Fontana, F. Rocca and E. Burattini. Solid State lonics 2X-30 (19Xx) 713. [3(l] J.W. Martin, H.J. Bisshoff. M. Nali, J. Ross and D. Brinkman. Solid State Ionics IX/19 (19X6) 295. [3l] C.D. Wagner. W.M. Riggs, L.E. Davis, J.E. Moulder and G.E. Muilenberg. Handbook of X-Ray Photoelectron Spectroscopy (Pcrkin-Elmer. Physical Electronics Division, Eden Prairie. MN. 1978).