Complementary scattered and recoiled ion data from ToF-E heavy ion elastic recoil detection analysis

Boom Interactions with Materials 8 Atoms

ELSEVIER

Nuclear Instruments and Methods in Physics Research B 136-138 (1998) 669-673

Complementary

scattered and recoiled ion data from ToF-E heavy ion elastic recoil detection analysis

P.N. Johnston a**,M. El Bouanani a, W.B. Stannard ‘, I.F. Bubb a, D.D. Cohen b, N. Dytlewski b, R. Siegele b ’ Deparimenr of Applied Physics, Royal Melbourne Institute qf’ Technology, GPO Box 2476 V. Melbourne 3001. Ausfrdiu h Austruliun Nuclear Science and Tdmology Organization. PMB 1. Menui 2234, Austrulirr

Abstract The advantage of Time of Flight and Energy (ToF-LJ Heavy Ion Elastic Recoil Detection Analysis (HIERDA) over Rutherford Backscattering (RBS) analysis is its mass and energy dispersive capabilities. RBS cannot separate signals from very heavy elements; the limitation is a result of physical principles of the two-body collision process. The mass resolution of ToF-E HIERDA deteriorates for very heavy elements; the limitation is related to the poor energy resolution of Si detectors for heavy ions. Very high ion energies improve the energy resolution and consequently the mass resolution of HIERDA, but significantly lower the cross section. Usually the energy spectra from ToF-E HIERDA data are used to extract depth profiles. In this work, the benefits of using the time spectra of both the recoiled and the scattered ions are discussed. The advantages are: (i) improved depth resolution by using the time spectra. and (ii) although the energy of very heavy recoils decreases with increasing mass. the energies of scattered ions increase. Time and energy projections of spectra. which have overlapping signals provide complementary data for analysis. 0 1998 Elsevier Science B.V.

1. Introduction

Heavy Ion Elastic Recoil Detection Analysis (HIERDA), using either the mass Time of Flight and Energy (ToF-~5’) or the nuclear charge dispersive AE-E (Energy Loss and Energy) detection systems, has been extensively studied during the last few years. The exceptional and unambiguous multi-elemental depth profiling capabilities of HIERDA are due mainly to its dispersive and si-*Corresponding author. Tel.: +61 3 9660 2138: fax: +61 3 9660 3837; e-mail: [email protected]. 0168-583X/98/$19.00 0 1998 Elsevier Science B.V. All rights reserved PIISO168-583X(97)00779-9

multaneous multi-elemental detection characteristics. The potential of HIERDA has been demonstrated through studies of material science problems, which are intractable with other Ion Beam Analysis (IBA) techniques including RBS. Those studies used high energy (0.2-2 A MeV) heavy ions by groups in Sweden [l], Germany [2], Canada [3], France [4], Finland [5] and Australia [6]. Despite the effort devoted to optimising HIERDA, there is not yet a mature methodology for the analysis of HIERDA spectra to derive depth profiles. This is due to: (i) uncertainties in the understanding and modelling of stopping power,

straggling and multiple scattering of heavy ions, and (ii) the detector energy response instability during ion irradiation [7] and (iii) the poor energy resolution of Si detectors for heavy ions. Current limitations of HIERDA are not a result of physical principles but are mainly related to the mass or nuclear charge separation performance of the detection system. In the case of the ToF-E detection telescope, this limitation is mainly due to the energy resolution of the Si detector. In a previous study, the mass resolution of ToF-E HIERDA [8] was determined, showing an improvement with increasing energy and lighter recoils due to the energy response of the Si detector. For ToF-E detection systems, several data processing methodologies have been developed and used to simulate (ToF, E) data [9] or extract elemental energy spectra [lo] as a first step in the estimation of depth profiles. Time spectra have been largely ignored. Also the forward-scattered spectrum is not used even though it contains valuable information. This work demonstrates that: (i) instead of the ToF detector, the Si detector should be used as the dispersive instrument. The time spectra provide better depth profiling information, improving the depth resolution of HIERDA, and (ii) some difficulties in extracting elemental energy spectra of the overlapping signals from heavy recoils (Pt-Bi) can be overcome by using the complementary time spectra of the recoiled and forward-scattered ions.

2. Time spectra and depth resolution In IBA studies, energy spectra are usually measured and used for depth profiling. Time spectra are used in heavy ion RBS for improved depth resolution but are converted to energy spectra before data analysis [11,12]. This is possible because the mass of the scattering ion is well known. The ToF detector in HIERDA is normally only used for mass dispersive purposes, while the energy spectra are used to calculate the depth profile. The time spectra have not been used because isotopic separation is usually not achievable for hea-

ions and consequently the timeenergy VY conversion is not possible. Conversion of time spectra to energy spectra is not a necessity but simply a conventional method. Time spectra have not been used because the ToF (unlike the energy) is neither an absolute quantity nor an intrinsic physical parameter. The ToF depends on the flight length (L), which is decided to suit a particular experiment. In addition, modelling of the ion-matter interactions such as energy loss. straggling, multiple scattering as well as kinematic factors are defined in terms of energy. There is no need to reformulate the models of ion-matter interactions in terms of the ToF parameter. Experimental time spectra can be directly used for depth profiling. This removes the limitation in performing the time to energy conversion for unresolved isotopes. The depth profiles are extracted using a simulation where each isotopic contribution is calculated independently. The simulated time spectrum is the sum of these isotopic contributions. This is very advantageous for HIERDA as the depth resolution is significantly improved and ion damage induced changes in the Si detector pulse height do not affect the analysis. The energy resolution of Si detectors over the wide range of energies and masses used in HIERDA is not well known. On the contrary, the time resolution can be approximated by a constant. This simplifies the simulation of time spectra. Fig. 1 shows the gain in depth resolution when using time spectra instead of energy spectra for a SrBiZTa,Oq ferroelectric sample. The energy resolution of the Si detector for Bi, Pt and Ta recoils detected at 45” from 97.5 MeV ‘?‘I incident ions is so poor that it is difficult to distinguish the high energy edges of the different elements. On the contrary, the shape of the time spectrum shows well defined high energy edges, demonstrating superior resolution of the ToF detector. A comparison between the energy resolutions of the Si detector and the ToF detector for IL71ions is presented in Fig. 2. The energy resolution of the Si detector is estimated using Amsel’s formula [ 131 and experimental data [14]. Amsel’s empirical formula is AEs, = a + hELi’.

(1)

P. N. Johnston et al. I Nucl. Ins~r. and MeA

611

in Ph_w. Rex B 136-138 (1998) 669-673

1000

0

m

=200

LL

0

s

iii? rn

100

(6

0.6

0 Fig. I. (a) Projected energy spectrum of the overlapped Ta. Pt and Bi recoils detected at 45” from 97.5 MeV ‘?‘I incident ions. and (b) projected time spectrum of the same recoils. Details of the measurements on the sample of nominal structure SrBilTa209/PtiTi0$3i are given in [6].

41

0

t

0

20

40

60

80

Iodine Energy (MeV) where u and h are constants for a given atomic mass number. The energy E is expressed in keV. From Hinrichsen’s compilation [14] the mass dependence of the parameters a and b is estimated to be

Fig. 2. (a) Energy resolutions for both ToF and Si detectors of “‘I ions. (b) The ratio of energy resolutions of Si detector and ToF detector for ‘?‘t ion for L of 0.5 and I m.

AEsi = (108.08 - 3.38M) + (0.51M - 585)E”!

contributions for scattered (2)

where M is the atomic mass of the detected ion. This expression is a fit to limited experimental data ranging from ‘“C to *‘Br ions with energies below 25 MeV. It is an extrapolation for ‘?‘I ions which may have large uncertainty. A preliminary measurement of the energy resolution for 58 MeV lz71 ions gives 3200 keV including geometrical factors. This suggests that the expression (2) underestimates the energy resolution for lz71 ions at 58 MeV. The energy resolution of the ToF detector can be derived from kinetic energy of the ion AET+ = 2E At/t,

and sample roughness. IL71 ions in this work.

The energy resolution of the ToF detector for lz71ions (Fig. 2(b)) is lower than the energy resolution of the Si detector and is at least a factor of 2 better at high energies and a factor of 10 better below 20 MeV for L of 0.5 m. Fig. 2(b) shows that the energy resolution of the ToF detector can be improved by a longer L at the expense of the solid angle. Figs. 1 and 2 show that the use of time spectra in HIERDA for depth profiling gives better depth resolution than the use of Si detector energy spectra.

3. Complementary (3)

where t is the ToF of the ion. At is the total time resolution of the ToF detector including the intrinsic time resolution, electronic noise, geometrical

At is 400 ps

spectra of scattered and recoiled

ions Data processing and the depth profile extraction in ToF--E HIERDA have been done in two main

612

P.N. Johnston et al. I Nucl. Instr. und Mrth. in Phys. Rex B 136-138 (1998) 669-673

ways: (i) the experimental data are compared with two-dimensional (ToF, E) simulations in an attempt to extract elemental depth profiles [9]. This has the advantage of avoiding the combination of correlated parameters and associated uncertainties, and (ii) experimental data are transformed to (M, E) space allowing an extraction of the elemental energy spectra for depth profiling. In the case of unresolved signals, the isotopic contributions from overlapped signals are extracted by fitting multiple Gaussians to the energy-gated parts of the projected mass spectrum [lo]. Recently, a statistical method based on isotopic spectral decomposition using a predefined absolute mass calibration and a semiempirical mass broadening model has also been evaluated [ 151. The depth profile is normally calculated using the energy spectrum of the recoil ions. The spectrum of forward-scattered ions is ignored. The (ToF, E) simulation software [9] which simulates the complete (ToF, E) spectrum of recoils and the forward scattered ions is expected to generate more accurate depth profiles. The mass resolution can be improved by increasing incident ion energy. It has been shown [224] using a AE-E telescope, that an isotopic mass separation of 1 u in the mass 60 region can be achieved when using 240400 MeV “‘I, “‘Xe or “‘Au beams. The optimum incident ion mass is much heavier than the target mass, but for the case of very heavy materials, e.g. Bi, this is not achievable. The energy of the incident beam can be increased but the recoil cross section decreases and the depth resolution may also be degraded. For species like Ta, Pt and Bi, existing ToF-E detectors cannot resolve the masses and achieve depth resolution approaching 10 nm, so we abandon the strategy of complete resolved signals. It is proposed to use the time spectra of both the forward scattered and the recoiled ions as they contain valuable complementary information. In fact, by looking carefully through the kinematics governing the recoiling and scattering processes, one can see as shown in Fig. 3 that the mass dependencies of the energy transfer to the recoil and the scattering energies follow opposite trends. This is true in the case where the mass of the incident ion is lighter than the analysed target masses.

60

50

recoil

40 150

175

Target

200

Atomic

225

25;

Number

Fig. 3. Surface energy of scattered (dashed line) and recoiled (bold line) ions from 95 MeV 12’1 ion irradiation as a function of target mass for 45” scattering angle.

This important property means that the overlapping isotopic contributions in the unresolved signal from the recoiled ions are completely different from the overlapped contributions to the scattering signal originating from the same recoils. This is shown more clearly in Fig. 4, where the scattered lz71 from the Bi at the SBT sample surface is well defined whereas its contribution to the recoil spectrum is overlapping with Ta and Pt. The simulation of both the scattered and recoiled data, even though unresolved, from heavy ions adds extra information for the depth profile extraction. The optimum simulated structure for scattered and recoiled ions differs marginally because of inaccuracies in stopping powers. However it is important to recognise that the overlapping signals from heavy recoils result in a decrease of the sensitivity of HIERDA. In fact, when signals from two elements are overlapping and when the contribution of one of them is about the same order of magnitude as the statistical error of the remaining contribution from the other element, it is impossible to extract individual isotopic components using exclusively the ToF-E recoil spectrometry data. Better mass separation is the ideal solution but the required energy resolution is not achievable with the Si detectors. The use of a gas ionisation chamber detector may significantly improve the mass separation of heavy re-

P.N. Johnston et al. I Nucl. Instr. nnd Meth. in Phlx

Rrs. B 136-138

(1998) 669473

673

profiling and reduces current limitations when dealing with very heavy ions. Acknowledgements

This work is supported by the Australian Institute of Nuclear Science and Engineering (AINSE). Recoiled

Bi,Pt,Ta

1000

1500

References

500

2000

111H.J.

(b) -

.*

. 2500

3000

Time Channel Fig. 4. Superposition of experimental and simulated time spectra for (a) overlapped Ta, Pt and Bi recoils (b) forward-scattered ‘?‘I from Ta. Pt and Bi.

coils but for complete isotopic separation, a better detection approach is needed.

4. Conclusions We have demonstrated that in HIERDA using ToF-E telescope: The high stability of the ToF signal which is independent of the mass, atomic number and energy of the detected ions has significant advantages over the use of energy signal from Si detectors. Energy spectra need not be used for depth profiling. Time spectra of overlapping signals from heavy recoils are a better choice for good depth resolution and improved depth profiling. The simulation of the complementary scattered and recoiled ion time spectra improves depth

Whitlow. E. Johansson. P.A. Ingemarsson, S. Hogmark. Nucl. Instr. and Meth. B 63 (1992) 445. 121W. Assmann. H. Huber. Ch. Steinhausen, M. Dobler, H. Gluckler, A. Weidinger. Nucl. Instr. and Meth. B 89 (1994) 131. [31 R. Siegele. H.K. Haugen. J.A. Davies. J.S. Forster. H.R. Andrews. J. Appl. Phys. 76 (1994) 4524. [41 J.P. Stoquert. G. Guillaume. M. Hage-Ah. J.J. Grob. C. Ganter. P. Siffert. Nucl. Instr. and Meth. B 44 (1989) 184. [51 J. Jokinen. J. Keinonen, P. Tikkanen. A. Kuronen. T. Ahlgren. K. Nordlund. Nucl. Instr. Meth. B 119 (1996) 533. PI W.B. Stannard, P.N. Johnston, S.R. Walker, I.F. Bubb. J.F. Scott, D.D. Cohen. N. Dytlewski. J.W. Martin, Integ. Ferroelectrics 9 ( 1995) 243. M. El Bouanani. W.B. Stannard. I.F. [71 P.N. Johnston. Bubb. D.D. Cohen, N. Dytlewski. R. Siegele, Vaccum 48 (12).in press. PI M. Huh. M. El Bouanani. L. Persson, H.J. Whitlow. M. Andersson. C. Zaring. M. Gstling. P.N. Johnston, S.R. Walker, I.F. Bubb. D.D. Cohen. N. Dytlewski. Nucl. Instr. and Meth. B 101 (1995) 2633266. M. El Bouanani, W.B. Stannard. I.F. [91 P.N. Johnston. Bubb. P. Jonsson. Y. Zhang, H.J. Whitlow, in: J.L. Duggan and I.L. Morgan (Eds.). Proceedings of the Fourteenth International Conference on the use of Accelerators in Research and Industry, Denton. Texas, 6-9 November 1996. American Institute of Physics Conference Proceedings 392. AIP Press, New York, p. 715. [lOI M. Huh. L. Persson. M. El Bouanani, H.J. Whitlow. M. Andersson. M. &tling. N. Lundberg, C. Zaring, K. Georgsson. P.N. Johnston. S.R. Walker, J. Appl. Phys. 77 (1995) 2435. [Ill R. Groleau. S.C. Gujrathi. J.P. Martin. Nucl. Instr. and Meth. 218 (1993) Il. [121 J.P. Thomas, M. Fallavier. D. Ramdane. N. Chevarier. A. Chevarier. Nucl. Instr. and Meth. 218 (1993) 115. iI31 G. Amsel. C. Cohen, A. L’hoir. in: 0. Meyer. G. Linker, F. Kappeler (Eds.). Ion Beam Surface Layer Analysis. vol. 2. Plenum Press, New York, 1976. S.C. Gujrathi. L. [I41 P.F. Hinrichsen, D.W. Hetherington, Cliche, Nucl. Instr. and Meth. B 45 (1990) 275. u51 Leif Persson. Ph.D. Thesis. ISBN 91-628-I 717-5, Lund University. Sweden,