Quantitative aspects of Ion Scattering Spectroscopy (ISS)

Surface Science 47 (1975) 222-233 0 North-Holland Publishing Company

QUANTITATIVE

ASPECTS OF ION SCATTERING

SPECTROSCOPY (ISS) * H.

NIEHUS and E. BAUER

Physikalisches Institut, Technische Universittit Clausthal, Clausthal, Germany

The application

of ion scattering spectroscopy (1%) as a tool for the analysis of the of solids has been hampered to date by the need of special equipment and the lack of understanding of the interaction processes between ions and surfaces. The purpose of this paper is to examine what information can be extracted from the experimental data and to show that ISS can be easily combined with AES and LEED in the same UHV system. The commercial cylindrical Mirror Analyzer (CMA) for the AES measurements was simply switched in situ to ISS by reversing the polarity of the deflection voltage. The surface studied was a W (110) surface either clean or covered with an electronegative (oxygen) or an electropositive (beryllium) adsorbate in the coverage range from 0 to 1 monolayer. He+ scattering in the energy range from 300 to 1000 eV shows that ISS at present is better suited as a tool for surface structure analysis than for the determination of the species and quantity of adsorbed atoms.

surface composition

1. Introduction

In ion scattering spectropscopy a surface is bombarded by a beam of ions with mass M, , energy E, and intensity IO (fig. 1a) and the energy distribution of the ions scattered into a scattering angle 9 is analysed. If there are several kinds of atoms on the surface the energy distribution consists of several peaks i, k, . . . (fig. 1b). These peaks can be characterized by their energy position Ei, their height Zi (or area 1: respectively), their half width aEi and other features characteristic for the peak shape such as the degree of low energy tailing. The purpose of this paper is to examine what information can be extracted from these quantities. As bombarding ion He+ was chosen because for this ion the single collision model [l] should be most applicable. The surface studied was a W (110) surface, either clean or covered with an electronegative (oxygen) or an electropositive (beryllium) adsorbate in the coverage range from 0 to 1 monolayer. The scattering angle was fixed at 68”. The experimentally determined quantities of the peak position Ei, half width A.L?iand peak height Zi depend on the following parameters: * Work supported

by the Deutsche

Forschungsgemeinschaft.

H. Niehus, E. BauerlQuantitative

aspects of ISS

I

223

I (E)

I, ,E, ,AE;

I,, E”

a)

bl

Fig. 1. (a) Scheme of ion scattering experiment. For explanation see text. (b) Energy distribution of ions scattered by two different surface atoms with massesMzi and kfzk.

Ii = Ii(jo, Ml ,M,j, 0, ei;J/, \o,Ni(z)>Pi(Z),**.I.

(3)

The dependence on the quantities left of the semicolon follows already from the binary collision model, that on the other quantities is unknown. These quantities are the glancing angle of incidence J/, azimuth of incidence rp, the atomic distribution normal to the surface N&C), the probability for scattering without neutralization Pi(z), the fractional coverage Oi and other parameters characterizing the surface structure and inelastic processes. If ion scattering spectroscopy is to be used as a quantitative tool for chemical and structural analysis the influence of the quantities to the right of the semicolon has to be studied. This can be done by comparing the experimental results with the predictions of the binary collision model [I] which leads to the following equations for energy position and peak intensity:

Ei=Ep

~~~M2i~2 -

(cosa+[(

2,”

-sin28]li2)

2,

MzilM~ 21~(4)

It should be noted, that a linear relationship between scattered intensity and coverage is obtained only if the differential scattering cross section do/da and the neutral-

224

H. Niehus,

E. BauerlQuantitative

aspects of ISS

ization factor Pi are independent of coverage. Furthermore it is implicitly assumed that only the visible atoms contribute to the peaks. It will be seen that these assumptions are not generally valid.

2. Experimental

set-up and procedure

Fig. 2 shows schematically the ultrahigh vacuum system used. The components essential for ion scattering spectroscopy are the ion gun, the specimen manipulator, the energy analyzer and the slit manipulator defining the scattering angle 6. The differentially pumped ion gun is of the electrostatic type and produces an ion beam with a half width of less than 1 eV for ion energies between 300 eV and 600 eV. Typical ion current densities used are 3 X 10e8 A/cm2. The ion beam was aligned with deflection plates and focussed onto the specimen with an electrostatic lens to form a spot with 2 mm diameter. The specimen manipulator allowed variation of angle of incidence and azimuth and translation in three orthogonal directions. The scattered beam was energyanalyzed with a commercial cylindrical mirror analyzer which was also used for Auger electron spectroscopy [2]. The rotatable slit was used only occasionally to check whether or not there was significant scattering out of the plane of incidence or in the backward direction. As these contributions were found to be small, most of the

Auger-

Analy.?er

(CMA)

Mefol mporotor

Fig. 2. Scheme of the AES-1% the AES and ISS equipment.

spectrometer.

Dashed lines indicate

parts in a plane 10 cm above

H. Niehus, E. BauerjQuantitative

aspects of Xi’

225

measurements were made without slit. This made switching back and forth between ion scattering spectroscopy and Auger electron spectroscopy very easy. The crystal could be exposed to well defined doses of oxygen via a leak valve, a tungsten oxide source and a Be evaporator. The surface could be characterized in situ by LEED and secondary ion mass spectrometry in addition to Auger electron spectroscopy. The base pressure of the system was 1 X lo-lo torr, during the ion scattering spectroscopy experiments 1 X lo-* torr He. The performance of the system is illustrated in fig. 3 which shows the energy distribution of the He+ beam scattered from the clean W surface. The background signal is less than 10 counts/set over the whole energy range from 0 eV to 300 eV.

3. Results 3.1~ Peak positions The peak positions according to the binary collision model are characteristic for the mass of the scattering atoms. For our experimental conditions we obtain from theory the values of the peak positions relative to the primary energy shown in table 1. The experimental values are listed here for two different angles of incidence.

He*+W(llO) ED = 5OOeV

d

Fig. 3. Energy

I

I

I

100

200

300

spectrum

I

cocl

of He+ scattered

q

68’

\ 500

energy

(eV) -

from a clean W (110) surface.

226

H. Niehus,

E. BauerlQuantitative

aspects of ISS

Table 1 Relative energies Ei/Ep of the WISS, 01s~ and Be18S peaks; experimental data and data calculated from the binary collision model [BCM, eq. (4)] for He+ scattered into an angle fi = 68”

EiIEp W

3

Be

0.710 0.735 0.780

0.537 0.620

._._.~ BCM (talc) J, = 40” (exp) $ = 18” (exp)

0.986 0.964 0.966

_.-.-

These values are independent of azimuth cp,coverage 0 and primary energy Ep. The independence on azimuth and primary energy is illustrated in fig. 4 for half a monolayer(ML) of oxygen on tungsten. Here the relative peak positions of W and 0 are plotted versus primary energy for two different azimuths and for $ = 40”. Shown is also the oxygen peak position for $ = 18”. The peakshift towards higher energies with decreasing angle of incidence becomes significant below $ = 25” and can be attributed to an increasing contribution of multiple scattering. The results show that the binary collision model predicts the peak positions only approximately, even at relatively large angles of incidence. The fact that the peak position is independent of coverage (below 1 monolayer), of azimuth and of primary energy however allows a calibration for a given J/ and therefore mass identification. 5 ML 0 on W (110) 1.0 1

.98.96-

I

.9L.92-

w

+

4.

u

0

+ 0

”

y

He*-

1 .90a

z w

3

.aa-

b

b

e

-

t

W (1101

= 68’

qJ = 4o”

.86.81-

lo1 (4

. az-

I+) (0) in [llll

eo.70.76-

I” [1101 dIrectIon dIrectIon

qJ = Ia0

_x-------r

.7L-

.72-

200

LOO

600

800 EP

1000

1200

-

Fig. 4. Energy position E/E of WISS and 01~s for He+ scattering covered by 0.5 ML oxygen Por two different azimuths.

from a W (110) surface

H. Niehus, E. Bauerl Quantitative aspects of ISS

227

2. Peak half-width and low energy tail Mass identification in the presence of several masses on the surface is of course only possible if the peaks are sufficiently well resolved. Thus the peak half-width becomes a decisive quantity and - when a small peak occurs on the low energy side of a large peak - also the low energy tail of the large peak. These quantities are determined by the contributions of multiple scattering - which at low resolution gives a broadening towards the high energy side -, of inelastic scattering, which causes the tail towards the low energy side -, and of thermal vibrations. Increasing the primary energy leads indeed to an increase of the low energy tail of the W peak which can be attributed to an increasing contribution of He+ ions scattered from the second and deeper layers. In fig. 5 the half-width of the W and 0 peaks are plotted versus oxygen coverage. The projection of the ion trajectory on the crystal surface lies in the [ 1 lo] direction ((p = 0”). The oxygen half-width is independent of coverage, while the half-width of the W peak increases. This can be explained as follows: with increasing oxygen coverage the number of He+ ions scattered by the topmost W layer decreases and the W signal is caused increasingly by ions which had to pass through the adsorption layer thereby losing some energy in inelastic processes. Thus new peaks at lower energies increase and the superposition of the original peak and the new ones results in a broadening of the W1~s peak. The ions scattered by oxygen atoms do not have to penetrate a layer, so that the half-width of the oxygen peak does not increase with coverage.

Oxygen

A

on WI1101

A

A

AO

A

A

,JW

-?

He*-W

(110)

Ep : 350 eV

~68’

4J = LO0

10 I I

I

I

.2

.L

.6 Oxygen

I

.8 coverage

I

1.0 (ML) -

Fig. 5. Half-width AL?of the WISS and 01~s peak as a function of the oxygen coverage (9 = 0”).

H. Niehus, E. BauerlQuantitative

228

aspects of ISS

3. Peak height The factors determing the peak height according to the binary collision model have been pointed out before: coverage, scattering cross section and neutralization. Fig. 6 shows the W and 0 peak heights as a function of oxygen coverage. The oxygen coverage is given in numbers of tungsten oxide doses which were deposited at room temperature followed by annealing at 1250 K. This method leads to the same surface structures as exposure to oxygen gas and has the advantage that the sticking coefficient is coverage independent up to 1 monolayer [S]. A further amount of tungsten oxide results in the formation of three-dimensional oxide which has a lower binding energy and desorbs completely at the annealing temperature of 1250 K. As can be seen from fig. 6 the oxygen AES signal increases linearly from 0 to 1 monolayer. In contrast to this, neither the W nor the 0 1% signal is linear in this range. The variation of the oxygen signal with coverage may be attributed to an increase of the neutralization probability with coverage. Such an interpretation is suggested by the variation of the work function with coverage: it is constant up to about $ monolayer and then rises rapidly up to 1 monolayer [6]. A more electronegative surface should give rise to higher neutralization. The rapid decrease of the tungsten peak height can only partly be attributed to the shadowing effect of the oxygen atoms From the linear extrapolation of the initial slope it could be concluded, that one oxygen atom would have to screen five W atoms if the initial slope were just due to w -

oxide

on

W 1110~. annealed ,/-

‘AES

He’-

W 1110)

Ep = 350

2

9

= 68O

JI

:LOO

WI,,

eV

I4 x 10

123~56789 number

of

doses

deposIted_

Fig. 6. AES and ISS peak heights as a function of the number of deposited W-oxide doses; annealing temperature: 1250 K (q = 0”).

H. Niehus, E. BauerjQuantitative

aspects of ISS

229

shadowing. On the other hand the neutralization probability was just assumed to increase with coverage while the W signal decreases less with increasing coverage. measurements in the coverage region from 0 to 0.5 monolayer have also been carried out when oxygen was brought onto the crystal surface in form of oxygen gas. In this case one oxygen atom had to screen initially 4 W atoms. This value is independent of azimuth. The simple geometrical shadowing effect can neither explain the high value of 4-5 W atoms nor that there is no azimuth dependence. Thus additional factors influence the ISS peak heights. This is even brought out more clearly by the results shown in fig. 7 for adsorbed Be which is electropositive. The sharp break of the Be AES signal at 6 doses signals the completion of the first monolayer. Therefore the extrapolation of the initial slope of the W signal indicates that one Be atom initially shadows two W atoms which appears reasonable on the basis of geometrical considerations. With increasing coverage the shadowing effect per atom decreases causing a slower decrease of the W signal similar to the case of oxygen adsorption. The Be signal, however, shows a rather abnormal behaviour as function of coverage: it decreases beyond a coverage of about 0.6 monolayers. The LEED data show that at this coverage a considerable structural rearrangement of the adsorbed layer occurs.

Be on W(llO)

18

t

He’-+WillO) Ep = 350 eV a

= 6ao

qJ = LO0

2

i

i number

5

of doses

6 deposited

-

Fig. 7. AES and ISS peak heights as a function of the number of deposited Be doses; 1 dose corresponds to 0.16 ML (rp= 0”).

H. Niehus, E. BauerjQuantitative

230

aspects of ISS

The two examples show that there does not always exist a linear relationship between coverage and ISS signal as it was found by Smith et al. [3] for Au-Ni alloys and by Taglauer and Heiland [4] for sulfur on Ni in the submonolayer region. Under some conditions the signal may not only not increase with coverage but even decrease. The dependence of the peak heights on the primary energy is shown in fig. 8. The shape ofthe curve of clean W, which is the result of a competition between scattering cross section and surface neutralization efficiency, is similar to that found by Smith et al. [3] for Cu (100). Measurements were carried out with various oxygen coverages and show that the shape of the oxygen peak remains constant, only the absolute value of the peak height changes with coverage as shown before. The W signal however is strongly modified by the presence of oxygen: it is reduced considerably in strength and its primary energy dependence is changed. This is understandable, because at low energies the oxygen atoms shield very effectively the W atoms while with increasing energy incident and scattered ions increasingly penetrate the adsorption layer thus leading to an increase of the W signal. This is even more evident in fig. 9, in which only the W peak heights are plotted for various coverages as a function of primary energy. Curves were taken starting from the high energy side as well as from the low energy side with very low current densities in order to exclude oxygen desorption or rearrangement of the surface. All curves obtained with oxygen coverage show the same behaviour independent of

L-

-M---d

2OIs5(.5ML oxygen on W(1101)

L

r

200

I

'LOO

1

600

I

I

I

800 lOal 1200 pr,mory energy l&l A

lLO0

Fig. 8. Peak heights of WISS and 01s~ for a clean and a 0.5 ML oxygen covered W (I 10) surface. Scattering angle 19= 68”. Angle of incidence J/ = 40”. The data are not corrected for the energy dependent analyzer transmission (9 = 0”).

H. Niehus, E, BauerlQuantitative

8 ii 7-

W,,,

amplitude

(arbitrary

Oxygen

units)

aspects

on

of ISS

231

WI1101

(4

W,,

I.SML

0 on W il:O)f

I .9 ML 0 on W1110))

t clean i 1ML

T 300

1

UJO

I

500

I

600

I

700

,

800 primary

I

0 on

W (110))

W (1lOf)

,

900 1000 energy / eV -

Fig. 9. Peak heights of W~SS for a clean, a 0.5 ML, a 0.9 ML and a 1 ML covered W (110) surface; ~9= 68”; $ = 40”. The data are not corrected for the energy dependent analyzer transmission

(lp= 0”). azimuth: they are nearly constant up to about 450 eV and increase at higher energies. This observation can be explained by assuming that the topmost layer at monolayer coverage consists of oxygen atoms only and that the He* backscattering is due only to this layer for primary energies below about 450 eV while at higher energies also the second and perhaps deeper layers contribute. An alternative explanation‘, which assumes both oxygen and tungsten atoms in the topmost layer at monolayer coverage and energy-dependent shadowing of the W atoms by the 0 atoms, is unlikely on the basis of the change of work function with oxygen coverage \5]. It can also be excluded by KS at large incidence and scattering angles J/, 6, for in this case shadowing of atoms by atoms in the same layer is ne~i~ble. Fig. 10 shows the results of such an experiment with $ = 76’) 8 = 152’ together with some of the data of fig. 9. Although there is more scatter in the points due to the considerably lower signal to noise ratio, the same energy dependence as at the smaller $, 8 values of fig. 9 is apparent. Therefore it is concluded that He’ ion scattering at least in the system oxygen on tungsten is sensitive only to the topmost layer at primary energies below about 500 eV, while at higher energies the contribution of the second and possible deeper layers rapidly increases reaching a saturation at about 1100 eV. This has important consequences for surface structure analysis: the complex LEED pattern produced by a full monolayer of oxygen on the W (110) surface can

H. Niehus, E. BauerjQuantitative

232

161

Oxygen

1i

on

aspects of ISS

WITlO) He+-

1L

1x1 i+i

L;; 2 12-

W (110) 8 = 152’

+=

3 L .@ 2 IO-

(~Ib)io)

9: /

W,s,

1.5 ML 0

76’

8 = 68’ LO0

on WlllO))

6L 2

1

x

1 x20

_ T 300

r I LOO 500 primary

x A

p/Wlss

[lML

0 on

W(llOJJ

/ I 600 energy

I 700 (eV1

I 600 -

Fig. 10. Peak heights of W~SS for a clean, a 0.5 ML and a 1 ML oxygen covered W (110) surface for.two different scattering angles 191= 152” and 92 = 68”. The data have been corrected for the energy dependent transmission of the energy analyzer. Data for 0 = 152” are normalized to those for 9 = 68” by multiplying with the ratio of ion currents obtained at the two angles with 600 eV ions from the clean surface (p = 0”).

be explained by a reconstructed surface, i.e. a surface in which 0 and W are present in the top layer, or an unreconstructed surface containing only 0 atoms in the top layer. At present no other surface tool has enough surface sensitivity to decide between these two models. Scattering of He+ ions with less than about 500 eV energy however allows a clear distinction in favour to the nonreconstructed surface. Thus surface structure analysis by ion scattering cannot only be done by angular dependence studies [7] but also by energy dependence studies.

4. Summary (1) The energy of scattered He+ ions can be described reasonably well by the single collision model, independent of primary energy, azimuth and coverage. Agreement between model and experiment decreases with decreasing glancing angle of incidence. (2) The half-widths of peaks due to adsorbed atoms are independent of azimuth,

H. Niehus, E. BauerlQuantitative

233

aspects of ISS

primary energy and coverage, that of the substrate increases with coverage and primary energy. (3) The he&$ of peaks due to adsorbed atoms is in general not a linear function of the coverage, neither does that of the substrate decrease in a simple manner with coverage. Shadowing considerations and coverage dependent neutralization can not explain all the observed effects. Structure and inelastic effects may have to be taken into account. (4) He ion scattering samples not only the topmost layer, but also the second and possibly deeper layers, depending upon primary energy. For oxygen on tungsten only the first layer contributes up to about 500 eV primary energy. (5) On the basis of these results it must be concluded that ISS at present is better suited as a tool for surface structure analysis than for the determination of the species and quantity of adsorbed atoms. These quantities can be obtained much easier with AES. The experimental arrangement used in this work allows a convenient combination of the two techniques which in combination with LEED should prove very useful in surface composition and structure analysis.

References [l] [ 21 [3] [4] [ 51 [6] [7]

See for example: E.P.Th.M. Suurmeijer and A.L. Beers, Surface Sci. 43 (1974) A. Niehus and E. Bauer, Rev. Sci. Instr., to be published. D.P. Smith, Surface Sci. 25 (1 971) 171. E. Taglauer and W. Heiland, Appl. Phys. Letters 24 (1974) 437. T. Engel and E. Bauer, in preparation. T. Engel, H. Niehus and E. Bauer, in preparation. W. Heiland and E. Taglauer, J. Vacuum Sci. Technol. 9 (1972) 620.

309.