Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Crater Edge Profiling of Ni/Cr Sandwich

Multilayer Thin Films by Scanning Auger


Microscopy (SAM)

A. Zalar
Institute for Electronics and Vacuum Techniques, Teslova 30, Yu-6001 Ljubljana, Yugoslavia
S. Hofmann
Max Planck Institut fur Metallforschung, Institut fur Werkstoffwissenschaften, Seestrasse 92, D-7000 Stuttgart 1, FRG

The principle of the technique of crater edge profiling is described as an alternative method to conventional
sputter profiling. The predictions for the lateral composition profile assuminga Gaussian intensity distributionof
the primary ion beam are tested for a multilayer sandwich structure of Ni/Cr layers of 11.5 nm single layer
thicknesses after Ar+ ion sputtering through 20 layers and scanning Auger microscopy with a 10 p m diameter
electron beam. Due to the small angle of 442” of the slope of the crater formed by ion sputtering, a magnification
factor (lateral variation/depth variation) of 5 x lo3is obtained.

INTRODUCTION distribution, it can be checked whether the condition


AzI/z s is met. Assuming a Gaussian ion intensity
profile, resulting in z ( x ) = zo exp (-x2/2v2)
the opti-
In depth profiling by AES combined with argon ion
mum case is an Auger analysis at the centre (Fig. 1,
bombardment,”2 a flat-bottomed crater within the
electron beam at A). If dI is the FWHM (‘diameter’) of
analysed area is of decisive importance with respect to
the primary ion beam and d, that of the electron beam, a
optimum depth resolution44 This condition requires a
simple calculation gives approximately:
beam of primary ions with uniform intensity over this
area. Any deviation from uniformity will cause a
degradation of the depth resolution.’-8 A still unsolved
problem is the determination of the magnitude of this
instrumental effect of ion beam inhomogeneity
compared to that of the other effects which contribute to where x is the lateral shift of the electron beam from the
a limited depth resolution, such as: Auger electron maximum intensity (centre) of the ion beam. For exam-
escape depth, original surface roughness, crystal orien- ple, d I= 2 mm and x = 0.5 mm give AzI/z = 5% for a
tation and defects, sputtering induced surface roughen- 100 p m electron beam diameter, whereas AzI/z < 1%
ing, ion mixing, preferential sputtering, surface and bulk is obtained for a mismatch x s 2 mm with an electron
diffusion e t ~ . ~ ’ ~ beam diameter of only 5 p m . This demonstrates the
With the exception of orientation effects, the relative advantage of the small electron beam diarileter used in
depth resolution Az/z defined by the broadening Az of scanning Auger miscroscopy (SAM).’”” Another
~ - ~ the sputtered depth z ,
a rectangular p r ~ f i l e ~ ’upon advantage is the applicability of an alternative method to
generally decreases with because the absolute obtain depth distribution profiles: the so-called crater
contribution is either independent of depth (Az/z z - ) edge profiling10”2 first described by Taylor et It is
or increases with the square root of the depth analogous to the conventional angle-lapping method
(Az/z oc 2-l”). However, the relative depth resolution used in electron microprobe analysis.
caused by primary ion beam inhomogeneity, AzI/z, is Since a static primary ion beam produces a crater of a
constant because AzI is ro ortional to the primary ion shape which is similar to the ion current density dis-
flux and therefore to z.’~ &r this reason, AzI becomes tribution in the x-y plane, a crater with a more or less
the limiting contribution to the total depth resolution for Gaussian profile is generally ~ b t a i n e d If. ~the electron
large z. With respect to the order of magnitude of beam is scanned from the bottom of the crater to the
measured Az value^,^ AzI/z G lo-’ can be tolerated up edge (Fig. l),the depth analysed gradually decreases.
to a sputtered depth z of about 100nm. For larger The important difference from the conventional angle-
depths (up to 1 p m or beyond) AzI/z should be at least lapping technique is that the angle with respect to the
one order of magnitude lower, i.e. SW3.This value original surface can be made several orders of magni-
corresponds to the required ion beam homogeneity. It is tude smaller in the crater edge technique.
usually met by rastering a focussed ion beam over a large The following shows how crater edge profiling is
performed and how the slope of the crater can be
With a static ion beam, and knowledge of the electron evaluated by a multilayer sandwich sample of Ni/Cr of
beam diameter and the shape of the ion current density known layer thickne~ses.’~.’~
ccc-0142-2421/80/0002-0183 $02.00
@ Heyden & Son Ltd, 1980 SURFACE AND INTERFACE ANALYSIS, VOL. 2, NO. 5, 1980 183
A. ZALAR AND S . HOFMANN

RESULTS

Figure 1 shows the principle. Assuming a Gaussian


shape of the primary ion beam, the cross section of the
crater should have a similar shape. The relation between
the depth scale ( 2 ) and the lateral scale (x) defined by the
movement of the electron beam in the x direction ('line
scan') is given by the slope angle a ;

tan a
az(x)
=--- -Ad (2)
ax Ax
NI Cr NI Cr NI Cr NI
The maximum slope for a Gaussian profile as indicated (b)
in Fig. 1 is given by:
20
tan a, = 0.6 - (3)
X,

where zo is the sputter depth at the centre and x, is the x


value corresponding to the standard deviation u of the
Gaussian function. It is recognized that the slope
remains almost constant around its maximum at x = u.
The deviation is within s 10% of tan a, from x = 0 . 7 ~
to 1 . 3 5 ~(see Fig. 1). In the experimental set up the
angular relationships between the normal to the sample NI Cr NI Cr NI Cr Ni
surface, and the ion and electron beams in the y-z plane H
must also be considered. However, in this case they can 100p m
be neglected because their directions are perpendicular Figure 2. Auger spectroscopicelemental mapping of a section of
to the x axis. the Ni/Cr multilayer structure shown schematically in Fig. 1.
As an example, crater edge profiling was performed
on a multilayer sandwich sample consisting of alternat-
ing Ni and Cr layers of thickness 11.5 nm, which were simultaneous observation of the negative peak intensity
evaporated on to an optical polished flat glass sub- of the Cr (529 eV) or the Ni (848 eV) Auger signal, the
strate.13 The thickness was controlled during evapora- elemental maps of these elements were obtained as
tion by a quartz microbalance. Sputtering was done with shown in Figs. 2(a) and (b).
a static beam of 1keV Ar' ions from a 5 keV gun in a The striation structure gives a qualitative picture of
commercial SAM apparatus (Physical Electronics the radial elemental distribution with a curvature in the
Industries-now Perkin Elmer Physical Electronics direction of the centre of the sputtering crater which is
Division-Model 545). When the glass surface was outside the upper left of the figures. The intensity
reached in the centre area of the primary ion beam, the modulation is difficult to quantify because of the some-
sputtering was stopped. By rastering the 1 0 p m what arbitrary contrast settings and relatively high noise
diameter electron beam over a limited sample area and (dependent on the sweep rate). However, a line scan
performed through the centre of Figs 2 (a) and (b)gives a
more quantitative result of the revealed radial dis-
tribution of the elements.
The line scans for the Ni and Cr signals are depicted in
Fig. 3, which resembles the conventional depth profile of

I
a multilayer s t r ~ c t u r e . With
' ~ the known layer thickness,
h tan a = d
the crater profiling conditions assumed in Fig. 1 can be
Ax
fa
U
L I NI i
Cr I
I
d
NI
b Cr
z
1.0 I I I I
0 0.5 1.0 1.5 2.0
Normalized distance, x / u
Figure 1. Principle of crater edge profiling: normalized crater
depth z/zoas a function of the normalized distance x / u from the
centre if the assumed profile has a Gaussian shape with standard H
100 p m
deviation U. Optimum depth resolution is obtained with an elec-
tron beam at A; B indicates the scanning position used for crater Figure 3. Line scans for the Cr (529 eV) and Ni (848 eV) Auger
edge profiling. a, is the maximum slope angle given by Eqn (3). peak-to-peak height signal across a selected region of the crater
The Ni/Cr multilayer structure profiled is schematically shown. from layers 18 to 10 (see Fig. 5).

184 SURFACE AND INTERFACE ANALYSIS, VOL. 2, NO. 5, 1980 @ Heyden & Son Ltd, 1980
CRATER EDGE PROFILING OF Ni/Cr SANDWICH MULTILAYER THIN FILMS

1 4 . . . .

0 50 100 150
Sputtering tme ts (min)

Figure 5. Conventional AES depth profile of the Ni/Cr multilayer


sample performed with 1 keV Arf ions (from Ref. 13). To be
compared with Fig. 3.

x (pm)

Figure 4. Evaluation of the shape of the crater using the position inherent depth resolution of the method is limited by the
of the maxima and minima of the Cr and Ni signalsfrom Fig. 3 and primary electron beam diameter with respect to the x
the known single layer thickness, d = 11.5 nm.
scale, i.e. a diameter of 10 p m corresponds to 1.5 nm in
depth. This is substantially smaller than the depth
resolution expected in the region from the 10th to the
checked, as seen in Fig. 4. Here the position of the 18th layer shown in Figs 3 and 4.
maxima and minima of the Cr and Ni signals on the x A comparison with Fig. 5, which is a sputtering profile
axis (Fig. 3) are plotted against a constant layer thick- of a similar layer given in an earlier p ~ b l i c a t i o nshows
’~ a
ness, giving a picture of the crater edge. A constant slope remarkably good agreement with respect to the shape of
is obtained over a wide range. the profile between layers 10 and 18. In this region, the
It should be mentioned that the constant layer thick- depth resolution from the sputtering profile was found to
ness applies only if the sputtering rates for both elements be about l o % , i.e. 14 nm. From the general evaluation
are the same, as is the case for Ar’ sputtering of Cr and scheme for the profile of a sandwich layer of thickness
~ i13.14
. From the known scanning display magnification d l s it can be estimated that limited plateaus (in a
(250x), the x scale is calibrated, giving an apparent
mathematical sense) instead of the distinct maxima
layer thickness of about 6 0 p m on the x scale. This (recognized in Fig. 3) should occur if A z / d < O S . That
means Az is at least 6 nm as seen from the shape of the
corresponds to the ‘real’ 11.5 nm on the z scale. There-
profile, i.e. measured depth resolution is not limited by
fore, we can calculate the ‘magnification’ factor accord-
instrumental factors but by the various additional
ing to Eqn (2)
contribution^.^'^.'
11.5 x 1 0 - ~
tan a = = 1.93 x lop4
60 x CONCLUSION
This means that the angle a is as small as 1.23”x In the case presented, the crater profile could be checked
or 44.3“. by the known layer structure of the sample. Generally,
the primary ion intensity distribution in the x direction
has to be measured, e.g., by a Faraday cup.4tan cy can be
DISCUSSION determined from Eqn (2) with z =it. In principle, a
change in the sputtering rate i can be taken into
The results in Figs 3 and 4 are in agreement with the account. Of course, depth resolution in crater edge
predictions following from Fig. 1:whereas tan CY is fairly profiling will also depend on the additional Az contri-
constant in the middle regions, the angle CY is smaller at butions as discussed elsewhere.5373x
the left and right hand sides, with deviations in the The advantages compared to conventional depth
expected direction if a Gaussian profile is assumed. This profiling are:
corresponds to a reasonable primary ion beam diameter
(FWHM = 2.35 a ) of x, = 1.7 mm for zo = 230 nm (a) The depth profile can be obtained at ease after
according to Eqn. (3). sputtering with almost no limitation in data acquisition
The increasing peak intensities from left to right (Fig. time over the whole area of the crater. Thus, possible
3) may be due to an increasing contribution of the lateral inhomogeneities of the elemental in depth dis-
backscattered electrons if the underlying metal layer tribution can be studied (the question of the ‘typical’
becomes thicker. Furthermore, the acceptance of the depth profile representative of a sample is of great
analyser may change along the x axis so that a direct importance in practical applications).
evaluation of the relative peak heights in terms of depth (b) In the related technique of spherical erosion by
r e s ~ l u t i o n ~appears
~ ” ~ not to be justified here. The ‘ball rate ring''^^'^ the crater is better defined. However,

@ Heyden & Son Ltd, 1980 SURFACE AND INTERFACE ANALYSIS, VOL. 2, NO. 5, 1980 185
A. ZALAR AND S. HOFMANN

the absolute depth resolution is rather restricted due The most outstanding feature of crater edge depth
to the mechanical roughening. Compared to sput- profiling is the possibility of obtaining large lateral
tering, crater formation by ball bearing appears only magnification factors (>lo3, due to a small crater edge
useful if a large depth range of several microns is of angle, < 5 ’ ) using conventional ( S 10 k m electron beam
interest. diameter) scanning Auger microscopy.

REFERENCES

1. J. W. Coburn, J. Vac. Sci. Technol. 13, 1037 (1976). 11. R. Browning and M. Prutton, Phys. Technol. 10,259 (1979).
2. S. Hofmann, Appl. Phys. 9 , 5 9 (1976). 12. N. J. Taylor, J. S. Johannessen and W. E. Spicer, Appl. Phys.
3. H. W. Werner, Acta Necfron. 19, 53 (1976). Lett. 29, 497 (1976).
4. H. J. Mathieu and D. Landolt, Le Vide-Les Couches Minces, 13. S.Hofmann, J. ErleweinandA.Zalar, ThinSolidfilms43.275
Spectrometrie Auger, No. Special, 273 (March 1979). (1977).
5. S. Hofmann, Le Vide-Les Couches Minces, Spectrometrie 14. S. Hofmann and A. Zalar, Thin Solid Films 60, 201 (1979).
Auger, No. Special, 259 (March 1979). 15. S. Hofmann, Proceedings of the 7th lnfernational Vacuum
6. E. Stumpe, H. Oechsner and H. Schoof, Appl. Phys. 20, 55 Congress and 3rd International Conference on Solid Sur-
(1979). faces, Vienna, 1977, Vol. 111, p. 2613. Berger, Vienna (1977).
7. S. Hofmann, Appl. Phys. 13, 205 (1977). 16. V. Thompson, H. E. Hintermann and L. Chollet, Surf. Technol.
8. S. Hofmann, i n Wilson and Wilson’s ComprehensiveAnaly- 8, 421 (1979).
tical Chemistry, edited by G . Svehla, Vol. IX, p. 89. Elsevier, 17. J. M. Walls, D. D. Hall and D. E. Sykes, Surf. InferfaceAnal. 1,
Amsterdam (1979). 204 (1979) .
9. R. E. Honig and C. W. Magee, Proceedings of the26fhAnnual
Conference on Mass Spectrometry and Allied Topics. p. 207. Received 26 March 1980; accepted 6 June 1980
St. Louis (1978).
10. A. van Oostrom, Surf. Sci. 89, 615 (1979). @ Heyden & Son Ltd, 1980

186 SURFACE AND INTERFACE ANALYSIS, VOL. 2, NO. 5, 1980 @ Heyden & Son Ltd, 1980

You might also like