Aluminium nitride layers investigated by slow positrons

Aluminium nitride layers investigated by slow positrons

•.:.::i:~:~:~:~:'.:~:~.~: ~~:~?..?:~:~:~:~:~:!.:~:~.:'~:~:?:,'4:~::::.. :::: i~:i~i~:~i~:~!~:::~:; !~;...

226KB Sizes 0 Downloads 39 Views

•.:.::i:~:~:~:~:'.:~:~.~: ~~:~?..?:~:~:~:~:~:!.:~:~.:'~:~:?:,'4:~::::.. :::: i~:i~i~:~i~:~!~:::~:; !~;.....................................................~:::~!!i!i~

surface s c i e n c e ELSEVIER

Applied Surface Science 85 (1995) 325-328

Aluminium nitride layers investigated by slow positrons A.S. Saleh a, P.C. Rice-Evans a,,, S.J. Bull b a Physics Department, Royal Holloway, University of London, Egham, Surrey TW20 OEX, UK b AEA Industrial Technology, B522 HarwellLaboratory, Oxfordshire, 0 X l l ORA, UK Received 4 May 1994

Abstract The structure and uniformity of aluminium nitride (AIN) layers produced by the sputter-ion plating coating technique has been assessed with the positron beam apparatus, Tacitus. A positron stopping protrde accounting for different densities was applied to layered structures of A1N/AI and TiN/AIN. Values for the Doppler line-height parameter are attributed to the variation of the coating processes. The thicknesses of the layers were determined by applying the diffusion model.

1. Introduction Aluminium nitride is an attractive material due to its combination of thermal and mechanical properties. Aluminium nitride films are used as a substrate in microelectronics industry and their combination of good corrosion and wear characteristics make them of great interest in general engineering industry [1,2]. The uniformity of film composition as well as the nature of the interface with the substrate are parameters affected by coating processes. The presence of voids or defects at the interface leads to poor adhesion and a thin metallic interlayer deposited prior to coating is a well established technique to improve adhesion. The important feature of positrons is their sensitivity to vacancy-type defects [3]. Essentially, vacancies are potential wells that trap slow positrons and the proportion of positrons being trapped may be revealed by an analysis of the energies of the annihilation photons. With a beam of controllable energies,

* Corresponding author. Fax: + 44 784 472 794. Elsevier Science B.V.

SSD! 0 1 6 9 - 4 3 3 2 ( 9 4 ) 0 0 3 5 3 - X

the positrons may be implanted at various depths beneath the surface and thus be used to probe the composition and interfaces in multilayer systems.

2. Experimental The two samples used in this study were prepared at Harwell by the sputter-ion plating coating technique on 1.25 cm radius stainless steel discs. The substrates were heated to 150°C under vacuum for outgassing, then a DC glow discharge used to sputter aluminium from the cylindrical cathode and this then deposited on the discs hanging inside a cylindrical volume. For the first sample aluminium was deposited before nitrogen was admitted into the coating system to form A1N, thus introducing an aluminium interlayer. In the second sample, the A1N film is sandwiched in between a titanium nitride film and a substrate as in Fig. 1. The samples reach 500°C during nitride deposition. The samples were mounted in the Royal Holloway variable energy positron beam providing energies from 1 eV to 25 keV. A G e detector placed just

A.S. Saleh et al. /Applied Surface Science 85 (1995) 325-328






Fig. 1. Schematic representation of the samples structures. ~, 0.46

E behind the sample chamber was used to detect the annihilation photons. The Doppler broadening of the annihilation peak has been characterized by the S parameter, defined as the area of a central region of the photoelectric peak divided by the total area of the peak. The S parameter is sensitive to electron momentum distribution at the site of annihilation. The diffusion model accounting for different densities and considering positron diffusion between layers was applied to the data and has led to good fits. The results of the measured S parameter as a function of positron energy for the first sample are shown in Fig. 2, two layers are clearly recognisable from the two distinc-

~, 0.45 5")




5 10 15 20 POSITRON ENERGY (keY)


Fig. 3. The experimental and fitted S values as a function of incident positron energy for the second sample.

tive S parameters. Fig. 3 for the second sample, reveals two layers that are thicker than the first. 0.48

3. Diffusion model analysis


The measured lineshape parameter S at a given energy consists of contributions of annihilations in each layer, in the bulk and at the surface. In our analysis we have assumed that positrons may diffuse through the interfaces. The positron implantation profile P ( z , E) is given by [4,5]


~ o.45 0.44

P(z, E )



0.42 i





5 i0 Positron



15 Energy


20 (keY)

= _ daz exp _


where the mean range of positrons for a given energy is i


Fig. 2. The experimental and fitted S values as a function of incident positron energy for the first sample.




7/ ( E ) " .


iV is the gamma function, p is the layer density, m and n are constants taken as 2.0 and 1.6, and the

A.S. Saleh et al. /Applied Surface Science85 (1995) 325-328


parameter ~i is determined by requiring the positron transmission to be continuous and given by

Table 1 Summary of fitted S parameter values and layer thicknesses for the first sample (A1N/AI) and the second sample (TiN/AIN)

~i=Xi( 1 -

Sample No.


Pi+Pi) 1 '



Thickness X (/~)

0.4385 0.5451 -

0.4002 0.4002 0.4002

6100 6900 -

0.4219 0.4772 -

0.4050 0.4050 0.4050

9300 10800 -

First sample where x is the thickness of the layer. Thermalized positrons diffuse in random directions, and the fraction trapped and annihilated at a surface can be written as

[ (z Xl l Js(E)=foXlp(z,E)exp- --~1 ]ldZ.


The fraction of positrons diffusing forwards from the first layer is given by J f l ( E ) ----fXjolp(z '

E) exp



and from the second layer Jf2(E) = £1 P ( z, E) exp




A1N layer AI layer Stainless steel

Second sample TiN layer A1N layer Stainless steel


Lp =

foP(Z, E)e-z/Lop dz.


L~p is the scattering length and Sep is the characteristic S value for epithermals. Applying this model to the measured data has led to the fits (solid lines) on Figs. 2 and 3.

The fraction of positrons diffusing backwards from the substrate to the second layer is defined as

Jb(E)=/x~P(z,E)exp[-{z-x:l] t-~b ] ] dz,

4. Results and discussion


and that from the second layer to the first is

[ (Z-Xl l

Jb2(E)=fx P(z,E) exp -t--~2 ]]dz.


Due to the large thicknesses of the layers compared to the diffusion lengths, positrons crossing a complete layer and annihilating in the next is not significant. Thus the lineshape parameter at a given energy may be expressed as S ( E ) =JsSs + (P1 - J ~ - J f l +Jb2)Sl

+( Pe -Jfe + Jb)S2 + ( Pb --Jb + Jf2)Sb, (9) where P1, P2 and Pb are the fractions stopping in the first layer, second layer and the substrate, respectively. Incorporating the epithermal positrons at very low energy results in the measured lineshape parameter equal to [6]


=Jep + (1 - J e p ) S ( E ) ,


The results in Figs. 2 and 3 illustrate the different S parameters attached to each material. The positron diffusion lengths in AI and stainless steel were determined by measuring the S parameter for each material individually; the values were 1300 ,~ with an S parameter of 0.5300 and 950 A, with an S parameter of 0.4018, respectively, and these are employed in the starting values in fitting these results. The fitted values of the layer thicknesses, S parameters of different layers and the diffusion lengths are summarized in Table 1. The S value for the A1 layer (0.5451) is higher than the value of bulk A1 (0.5300), which can be explained in terms of the presence of open volume defects and the spaces between grains. The layers in the second sample are thicker than the first, 9300 ,~ TiN layer and 10800 A AIN layer. The S parameter of the A1N layer (0.4772) appeared to be higher than the value in the first sample, this could be explained by the variation of the defect density created by thermal expansion mismatch between A1N and TiN and AIN and stainless steel.


A.S. Saleh et al. /Applied Surface Science 85 (1995) 325-328

The thermal expansion coefficient of A1N (5.1 X 10 - 6 K - 1) is much lower than that of TiN (8 X 10 - 6

K - l ) or stainless steel (16 X 10-6 K - l ) . On cooling from the deposition temperature (500°C) the nitride coatings will thus be put into a state of compressive residual stress. In the case of the first sample the presence of the AI interlayer can lead to some stress relief since it may yield to reduce the thermal stresses, whereas this cannot occur for the second sample. The thermal stress will force the columnar units of the coating together which can lead to plastic deformation, fracture and defect creation within the individual columns. The results generated here support the assertion that the magnitude of the residual stress in the coating is related to the defect density. An alternative explanation is the grain size of AIN nucleated on TiN is smaller than that for AIN on stainless steel. Further work is necessary to determine which of these factors is significant.

5. Conclusion We have presented an application of the diffusion model to a multilayer structure taking into account positron diffusion through the layers with no trapping at the interfaces. The thicknesses of the layers

have been determined. The effects of the sputtering ions and the substrate temperature on AIN defect concentration are revealed, showing that positron beam spectroscopy is potentially a valuable technique for monitoring the production processes of these films.

Acknowledgement The authors whish to thank the EPSRC for support.

References [1] E. Ruiz, S. Alvarez and P. Alemany, Phys. Rev. B 49 (1994) 7115. [2] J.M.E. Harper, J.J. Ctlomo and H.T.G. Hentzell, Appl. Phys. Lett. 43 (1983) 547. [3] P.J. Schultz and K.G. Lynn, Rev. Mod. Phys. 60 (1988) 701. [4] A. Vehanen, K. Saarinen, P. Hautoj~irvi and H. Huomo, Phys. Rev. B 35 (1987) 4606. [5] A. van Veen, H. Schut, R.A.J. de Vries, R.A. Hakvoort and M.R. IJpma, in: Positron Beams for Solids and Surfaces, Eds. P.J. Schultz, G.R. Massoumi and P.J. Simpson (Am. Phys. Soc., New York, 1990) p. 171. [6] D.T. Britton, P.C. Rice-Evans and J.H. Evans, Phil. Mag. Lett. 57 (1988) 165.