- Email: [email protected]
Use of the 16O(3He, α)15O reaction for studying oxygen-containing thin films
100
Nuclear Instruments and Methods in Physics Research B45 (1990) 100-104 North-Holland
USE OF THE ‘60(3He, a)“0 THIN FILMS + F. ABEL, G. AMSEL,
REACTION
E. d’ARTEMARE,
FOR STUDYING
C. ORTEGA,
OXYGEN-CONTAINING
J. SIFJKA and G. VIZKELETHY
++
Groupe de Physique des Solides, Universitk Paris VII, Tour 23, 2, place Jussieu, 75251 Paris Cedex OS, France
The quantitative analysis of I60 in near-surface regions of samples is usually performed with the 160(d,p)170 * reaction, at rather low energies ( < 900 kev). In many IBA facilities this technique cannot be used because of the absence of adequate biological shielding against the neutrons produced by the deuteron beam. On the other hand I60 depth profiling is often carried out with the resonance near 3.05 MeV of 160(o, (u)mO. With this technique the overall oxygen content is difficult to measure with precision on high-Z targets like high-T, superconductors, due to the strong RBS background under the 160 peak. Moreover, o-energies above 3 MeV are not available in many facilities. In this article a method is presented using the ‘60(3He, o.)150 reaction for the background-free determination of oxygen. The cross section of this reaction, which was remeasured at tilab= 90 o using thin SiO, /Si targets, presents at this angle a peak near 2.4 MeV and a drastic reduction below 2 MeV. The overlapping peaks from 160(3He,p)‘sF, which present no cutoff at lower energies, were eliminated from the spectra by using a detector with a very thin depletion zone. The advantages of this reaction are: very low neutron yields; background-free 160 detection, hence high precision, even on oxygenated substrates, as counts from deeper regions are drastically reduced by the cross-section cutoff, possibility to carry out simultaneous 3He RBS measurements, an asset for in situ analysis; reduced bean-target interaction effects as compared to hydrogen beams which might induce reduction phenomena at high temperatures.
1. Introduction 3He-induced
nuclear
reactions
are of relatively
less
common use for nuclear reaction analysis than those induced by protons or deuterons. The main reason seems to be historical. In fact 3He particles require in most cases higher bombarding energies than their lighter competitors for inducing reactions with yields suitable for analysis. A widely used exception is D + 3 He --, p or (Y with a maximum yield near 700 keV [l]. Hence, with the most widespread 2 MV single-ended Van de Graaffs which equipped many laboratories until the senventies the use of such reactions was out of reach. Since the eighties the accelerator population of the IBA community underwent a striking change: most of the 2 MV Van de Graaffs have been upgraded to 2.5 MV or more; a large number of small tandem machines where set up which allow the physicists to use routinely He beams with energies from 3 to 5 MeV or more. This increase of energy available from small machines leads for example to the extended use of the 3.05 MeV narrow resonance of 160( (Y,a)160. However, it seems that relatively little advantage was taken of the higher He beam energies available for using 3He-induced nuclear reactions. The latter are employed mainly for charged-particle activation analysis with relatively large accelerators. Note, + Work supported by GDR 86 of CNRS. ++ Present address: Central Research Institute for Physics, P.O. Box 49, H-1525 Budapest, Hungary. 0168-583X/90/$03.50 (North-Holland)
0 Elsevier Science Publishers B.V.
however, that Gossett [2] used 3He-induced reactions to measure 160 and 12C, but without remeasuring the corresponding cross sections. We recently undertook the exploration of the use of 3He beams for NRA. The latter have some advantages among which we may quote: - Much lower neutron yields at medium energies than those from deuteron beams, thus extending the range of nuclear reactions within reach in laboratories which are not protected with heavy biological antineutron shielding, a most common situation. - The possibility to carry out simultaneous 3He RBS and NRA measurements. The mass and depth resolution are only moderately reduced in 3He RBS as compared to 4He RBS, while p- or d-RBS may be used in very special cases only. Such simultaneous measurements are a great asset in many experiments, the complementary features of NRA and RBS being taken advantage of. They are moreover absolutely crucial for in situ measurements of physico-chemical processes, when for example the kinetics of thin-film solid-state reactions at high temperature, such as oxidations or reductions, phase changes, etc., are recorded in real time. In view of the efficient use of 3He RBS, the classical simulation program RUMP [3] has been modified so as to accept 3He particles. - Chemical beam-target interaction effects are reduced as compared to hydrogen beams which might induce reduction phenomena during in situ experiments at high temperatures.
F. Abel et al. / Use of the ‘60(3He, aj’%
In this paper we present our first results on 160 quantitative analysis with the ‘60(3He, u)t50 reaction (Q = 4.9140 MeV), this reaction presenting some other additional advantages over the more traditional r60(d,p)170* rea ct i on of widespread use at rather low energies, typically Ed < 900 keV, or with respect to the 160( (Y 01)160 resonant RBS technique. These advantages will bl discussed in detail. The cross section of the 160(3He, a)150 reaction presents a strong peak near 2.4 MeV, with the yield decreasing fast and monotonically towards zero at energies below the wide resonance. This reaction has been extensively studied by Bromley et al. [4]. At 8,, = 90° the reaction is nearly cut off below 2 MeV, while this is not the case for the parallel 160(3He,p)18F reaction or for 160(’ He, or)150 at larger angles. Such a cutoff may be of great advantage when measuring the content of oxide fibs deposited on oxygenated backings, a common situation for thin-film high-T, superconductors. Thus 160(3He, aji70 has a depth discrimination capability with respect to 1aO(d,p)170* which may be most welcome, as the cross section of the latter reaction does not present the corresponding features.
2. Experimental The cross section curve of the 160(3He, or)“0 reaction was remeasured at Blat.,= 90 o from 1600 to 2600 keV. u-particles were detected with a low-resistivity semiconductor detector with a thin enough depletion zone so as to achieve full ~~~~ation between u-particles and protons from the parallel i’O( 3He,p)‘8F reactions or other interfering reactions like *2C(3He, p)14N. Such a discrimination may be obtained without resorting to a dE/dx type detector provided the pulse-shaping time constant r is short enough [5]. We used an ORTEC 527 amplifier with r = 0.5 us, in conjunction with 500 G cm Intertechnique ion-implanted detectors with a nominal depletion zone of 50 pm under 20 V bias. A 10 pm thick aluminium foil was used to stop the scattered 3He beam particles. Ahuninium was preferred here to Mylar in view of the in situ experiments at high temperature [6] to be carried out with our setup. o-p discrimination was obtained by reducing the detector bias until the counts in the valley between the proton high-energy tail and the OLlow-energy tail were negligible even for thick oxygenated targets. Depending on the detector used, this voltage ranged from 10 to 20 V, the maximum proton pulse height being equivalent to about 1.8 MeV, the corresponding depletion zones were around 45 urn thick. The detector fitted with a 13 mm diameter collimator was placed at 65 mm from the beam impact point. In this geometry used for high detection efficiency, the
reactionfor studyingU-containingthinfihs
101
kinematic energy spread of the detected a-particles is large 171; the shape of the ol-spectrum was hence discarded and only its integral was considered. The thin oxygen target used for the measurements was a - 230 A thick SiO, layer on silicon. No appreciable a-yield from the silicon itself was observed. The precise 160 content of the SiO, layer was checked with the r60(d, p)l’O * reaction, by comparison with a classical Ta,Os absolute 160 standard [8]. The target was tilted to 45O off the beam. Carbon deposition at the 2 mm diameter beam impact point was minimized by operating an internal liquid-nitrogen cold trap during the measurements. The impact point was changed several tunes and the consistency of the yield from overlapping energy regions was checked. No change in the target thickness was evidenced during an experiment for the 3He+ currents used, ranging from 300 to 600 nA. The yields were recorded with 100-200 uC per point according to the energy region. The experiments were carried out with the 2.5 MV HVEE single-ended Van de Graaff of our laboratory. Particular care was taken for the energy calibration of the accelerator generating voltmeter. The energy scale was defined just before the excitation curve measurement by recording with a 2 in. X 2 in. BGO y-ray detector at @,, = O” the classical relevant 27Al(p, y) reference resonances and two strong narrow resonances at higher energies from 24Mg(p, y), at 2010 f 10 and 2400 k 10 keV. Thin MgO targets, sputter-deposited on tantalum were used. Although this energy scale reflects the uncertainties as quoted in the literature, on the “Mgt’p, y) resonances [9] the position of the measured cross-section peak is reproducible with better precision with respect to these easily recorded resonances. Moreover, it seems, according to our energy calibration measurements that the best estimate of the real position of the “2010 keV” resonance of 24Mg(p, v) is 2002 keV. The peak region of the excitation curve was recorded by using an automatic electrostatic-energy scanning device designed in our laboratory [lo].
3. Results Fig. 1 shows the excitation curve of the I6 0( 3He, ,)I5 0 reaction at 8,, = 90 ‘. The peak of the cross section is at 2370 keV, or more precisely 31 keV below the “2400 kev” resonance of “Mg(p, y). The maximum cross section is close to 3.4 mb/sr. The data are available on floppy disk. These values of E, and o, max should be compared to those reported by Bromley et al. [4]: 2420 keV and 4.4. mb/sr. The discrepancy is most probably due to calibration problems in this pioneer experiment; in particular, Bromley et al. could use only one calibration energy at 1450 keV. I. NRA
F. Abel et al. / Use of the ‘60(3He, a)150 reaction for studying O-containing thin films
102
t;
?:
10x
5000
..*
*.***
10500
*.....
.a* **
2000
Energy
[ keV]
.**... 1 i 2500
Fig. 1. Excitation curve of the 16q3He, a)“0 reaction at Blab= 90”, for a SiO, target containing 1OOX1O’5 ‘60/cm2, tilted to 45O of the beam, Q = 31.4~ 10W3 sr. The vertical scale may be converted into a cross-section scale in mb/sr by the multiplication factor 1.8 x 10m4. The energies E1 and E, correspond to the z4Mg(p, v) narrow resonances mentioned in the text.
We have thoroughly investigated neutron flux yields when using 3He beam: low neutron fluxes during such measurements are one of their main practical advantages, as stated above. The measurements were carried out with the help of the Nuclear Safety Department of the French Atomic Energy Commission; the results are given in mRem/h equivalent. Let us recall that the standard permissible rates for 2000 hours of work per year are: 2.5 mRem/h for directly involved workers, 0.75 mRem/h for indirectly involved workers and 0.25 mRem/h for the public. With a 2.6 MeV 3He+ beam and - 200 nA current on target we got the following typical results: - At 90 cm from the center of the deflecting magnet, at the side of the deflection: 4.5 mRem/h. Note that our magnet operates since 21 years and the walls of its chambers are thus heavily l2 C-contaminated. - At 45 cm from a bulk carbon target: nearly isotropitally 4 mRem/h. The background with beam measured with a clean tantalum target was - 0.7 mRem/h, while for bulk Al,O, and SiO, targets the rates were only slightly higher. This confirms that at 2.6 MeV bombarding energy the main source of neutrons is the 12C(3He,n)140 reaction, which has a threshold at 1.45 MeV. _ At 40 cm from the beam energy control slit system and from the main collimator, perpendicularly to the beam direction: 4 mRem/h. In conclusion, in normal experimental conditions, the main sources of neutrons are the magnet and the slit-collimator assembly. 3 m away from these units, i.e. one storey above or below the facility, the rate is lower than around 0.5-l mRem/h, i.e. might be tolerated even for the public for 500-1000 hours per year.
Let us note that with - 400 n4 beam on target the yields from the magnet are not much higher, as higher beam currents on target are mainly obtained by improving the focusing: the number of particles lost in the magnet does not increase in proportion to the on-target beam currents. It is clear that these results provide only a typical example, as each machine may behave differently. Preliminary neutron flux measurements are required for each particular facility and a neutron survey detector should be permanently installed near unprotected IBA facilities, chiefly for the case when higher neutron fluxes would be generated either from particular targets or at higher energies.
4. Discussion While having the marked advantages outlined above, for I60 determinations this reaction has the main disadvantage of presenting no plateau in its cross section. Hence the measurements must always be corrected for cross-section variations with depth and the use of a NRA spectrum simulation program is required if good accuracy is aimed at. We systematically used the program SENRAS [7]. Let us first consider the simple case when we want to measure the thickness, i.e. the overall oxygen content, of a thin oxide layer on an oxygen-free backing. This may be performed ideally with the l6 O(d, p)i’O * reaction at 900 keV except if, as stated above, no biological antineutron shield is available; the *H beam induces chemical changes in the target and/or simultaneous RBS is a must, like for in situ kinetic studies. Moreover other nuclei may produce (d,p) interference effects absent in 3He-induced NRA: a typical case is that of 160 determinations with the (d, p) reaction in presence of 14N, which may be far from trivial. Fig. 2 shows the calculated yield as a function of thickness for the typical case of CuO films, tilted to 45”, for various fixed bombarding energies E,. A saturation effect shows up: as the thickness of the oxide layer increases, the exit energy decreases and the corresponding cross section decreases monotonically for large enough thicknesses. In the case considered, the best bombarding energy is slightly above the resonance energy E,, the counts being nearly proportional to thicknesses up to - 1500 X 1015 at/cm2 of CuO. Above - 2500 x 1015 at./cm* the deviation from linearity is too strong to be compensated by a simple correction: the back region of the films contributes much less to the counts than the front region. We reach here the limits of simple 160 content measurements with this technique. The situation is similar to that which shows up in the PIXE technique, where the cross section decreases also monotonically with energy.
F. Abel et al. / Use of the t60( ‘He, a) “0 reactzonfor studying O-containing thin films
103
ioooo ~ ---
6000
$
Eo=2200 keV Eo=2350 keV Eo=2400 keV
. . ‘-
..
,:.
6000
$ 2 4ooa 2000
C 1000
Thickness
3000 2000 [ 10’s atom/c%‘]
5000
Fig. 2. Counts for - 50 KC beam dose in a typical geometry from CuO films of increasing thicknesses (expressed in overall at/cm*) for various beam energies I& The target is tilted to 45 o off the beam.
Another
important
determinations, surements
situation
like oxygen
arises
uptake
at high temperature
I
I
”
2 1
3
a
!
0.00
0.20
0.40 x in
0.80
1.00
CuOx
Fig. 4. Signal-to-noise ratio as a function of x for CUO, films of various Cu contents deposited on a thick SiO, backing, for E,, = E,. The ratio improves rapidly when the thickness along the beam increases: the effect of changing the tilt from 4.5 ’ to 60° (i.e. multiplying the thickness by 6) is illustrated for a 1250 X 10” Cu/cm’ film.
for stoichiometry
or loss kinetics
mea-
and in situ. Fig. 3 shows
the results of calculations for the typical case of CuO, films, containing 1250 X 1015 Cu/cm2. The corresponding in situ experiment with simultaneous 3He RBS is described in ref. [6]. It appears from this figure that the best choice for getting a response as linear as possible in x for films with formula CuO, is a bombarding energy such that E, is reached by the slo~ng-doe 3He particles just in the middle of the film for x = 1, in this case 2440 keV. Let us emphasize here that for very thick oxide films (or for bulk oxides) the a-yield is thickness-independent and depends only on the stoichiometry and on the various stopping powers involved. In this case stoichiometries may be directly compared, the measure-
6000
go;;; ~~~1
CIrp
1250~10~~
Cu/cma
Fig. 3. Counts in the same conditions as in fig, 2 for CnO, films of various compositions. The optimum energy E, is such that for a film with x = 1, En is reached in the middle of the f&n, here E,, = 2440 keV.
ments being backgro~~d-free (except from possible interference from some other light nucleus). We consider now the important case when the oxygen concentration must be measured in a film deposited on an oxygenated backing. If the energy loss of the 3He beam in the film of interest is large enough, the contribution from the oxygen in the backing will be small and the signal-to-noise ratio favourable: the cutoff effect is then put to benefit. Fig. 4 shows the calculated signalto-noise ratio as a function of x for CuOX films with various Cu contents deposited on thick SiO,; bombarding is at peak energy E, = 2370 keV which seems here nearly optimal, the targets being tilted to 45 O. A typical case is that of a film containing 1250 x 1015 Cu/cm2, for which the signal-to-noise ratio is 1.4 for x = 1. This ratio may be improved by further tilting the target: for a tilt of 60 o we get the ratio 2.7. For thicker films much higher signal-to-noise ratios may be reached. Finally we discuss how the use of the I6 O( 3He, ,)I5 0 reaction compares with RBS and in particular with the use of the 3.05 MeV resonance of 160(01, a)i60. If pure oxides are considered, variations of oxygen content may be evidenced in RBS by observing the plateau heights corresponding to a heavy component. This is illustrated by the results on CuO, presented in ref. [6]. However, for high-Z materials like YBa.,$u,O,_, the variations of the plateau height with oxygen content, i.e. 6, are small and this technique is not sensitive enough, as the stopping power per oxygen atom as compared to the overall stopping power is small. It turns out however, that such materials are ideally analyzed with 160( 3He, a)i’O as the corrections discussed above decrease when the overall stopping power variations with oxygen content decrease. Thus the two techniques are complementary: the less favourable is RBS, the better is 160{3He, t1)“O. I. NRA
104
F. Abel et al. / Use of the 160f 3He, CT)‘~Oreaction for studying O-~~n~~‘~~ng thin films
The resonant RBS detection of I60 may be well suited for samples such that the background under the oxygen peak is small enough. As soon as the oxygen peak sits on a large background like for thin oxides on heavy backings or thick oxides of high-2 elements, the 160(a, 01)160 resonance lacks sensitivy due to a small signal-to-noise ratio. Typically this ratio is - 0.25 with resonant RI% for thin YBaCuO films on yttria-stabilized zirconia and - 0.1 for thick YBaCuO films, while it may reach easily 3 or more with the r60( 3He, a)“0 reaction, for well chosen film ticknesses.
4. Conclusion The 160(3He, cw)“O reaction may present marked advantages for 160 dete~nations with respect to the reaction and to the ‘%(a, (u)r70 reso%(d,p)r’O* nant backscattering technique. As discussed, the choice between these three methods depends on the experimental conditions and requirements and also on the type of IBA facility used for the measurements: maximum available He energy, existence of a biological antineutron shield. The results on signal-to-noise ratios presented here may be improved if, in addition to their integral, the &ape of the a-spectra is also analyzed. This will be attempted by optimizing the detection conditions so as to improve energy resolution, without losing too much in counting rate, an essential parameter for in situ measurements. At 8 = 90 O, real depth profiling of 160 with this reaction is difficult due to geometric kinematic-energy spread, except when very small solid angles and absorber-free o-detection is used, like by Gossett [Z]; however, such a technique is only suita-
ble for films deposited on very light backings, such as carbon. Depth profiling will nevertheless be attempted at large detection angles 8, accepting a cross section which is lower and which does not exhibit a “cutoff” effect at B,, = 90 O, but where the kinematic spread is much reduced. We thank 0. Jaoul for providing us with thin MgO targets and H. Quechon for the measurements related to biological hazard due to neutron emission induced by 3He beams.
References III D. Dieumegard, D. Dubreuil and G. Am&, Nucl. Ins&. and Meth. 166 (1979) 431. 121 C.R. Gossett, Nucl. Instr. and M&b. 218 (1983) 149. 131 L.R. Doolittle, Nucl. Instr. and Meth. B9 (1985) 344 and B15 (1986) 227. [41 D.A. Bromley, J.A. Kuehner and E. Almquist, Nucl. Phys. 13 (1959) 1. r51 G. Amsel, J.P. Nadai, E. d’Artemare, D. David, E. Girard and J. Mot&in, Nucl. Instr. and Meth. 92 (1971) 481. 161 F. Abel, G. Amsel, C. Ortega, J. Siejka and G. Vizkelethy, to be published. 171 G. Vizkelethy, these Proceediugs (Ninth ht. Conf. on Ion Beam Analysis, Kingston, Ontario, Canada, 1989) Nucl. Instr. and Meth. B45 (1990) 1. [81 G. Amsel, J.P. Nadai, C. Ortega, S. Rigo and J. Siejka, Nucl. Instr. and Meth. 149 (1978) 705. 191 P.M. Endt and C. Van der Leun, Nucl. Phys. A214 (1973) 130. 1101 G. Amsel, E. d’Artetnare and E. Girard, Nucl. Instr. and Me& 205 (1983) 5.
Nuclear Instruments and Methods in Physics Research B45 (1990) 100-104 North-Holland
USE OF THE ‘60(3He, a)“0 THIN FILMS + F. ABEL, G. AMSEL,
REACTION
E. d’ARTEMARE,
FOR STUDYING
C. ORTEGA,
OXYGEN-CONTAINING
J. SIFJKA and G. VIZKELETHY
++
Groupe de Physique des Solides, Universitk Paris VII, Tour 23, 2, place Jussieu, 75251 Paris Cedex OS, France
The quantitative analysis of I60 in near-surface regions of samples is usually performed with the 160(d,p)170 * reaction, at rather low energies ( < 900 kev). In many IBA facilities this technique cannot be used because of the absence of adequate biological shielding against the neutrons produced by the deuteron beam. On the other hand I60 depth profiling is often carried out with the resonance near 3.05 MeV of 160(o, (u)mO. With this technique the overall oxygen content is difficult to measure with precision on high-Z targets like high-T, superconductors, due to the strong RBS background under the 160 peak. Moreover, o-energies above 3 MeV are not available in many facilities. In this article a method is presented using the ‘60(3He, o.)150 reaction for the background-free determination of oxygen. The cross section of this reaction, which was remeasured at tilab= 90 o using thin SiO, /Si targets, presents at this angle a peak near 2.4 MeV and a drastic reduction below 2 MeV. The overlapping peaks from 160(3He,p)‘sF, which present no cutoff at lower energies, were eliminated from the spectra by using a detector with a very thin depletion zone. The advantages of this reaction are: very low neutron yields; background-free 160 detection, hence high precision, even on oxygenated substrates, as counts from deeper regions are drastically reduced by the cross-section cutoff, possibility to carry out simultaneous 3He RBS measurements, an asset for in situ analysis; reduced bean-target interaction effects as compared to hydrogen beams which might induce reduction phenomena at high temperatures.
1. Introduction 3He-induced
nuclear
reactions
are of relatively
less
common use for nuclear reaction analysis than those induced by protons or deuterons. The main reason seems to be historical. In fact 3He particles require in most cases higher bombarding energies than their lighter competitors for inducing reactions with yields suitable for analysis. A widely used exception is D + 3 He --, p or (Y with a maximum yield near 700 keV [l]. Hence, with the most widespread 2 MV single-ended Van de Graaffs which equipped many laboratories until the senventies the use of such reactions was out of reach. Since the eighties the accelerator population of the IBA community underwent a striking change: most of the 2 MV Van de Graaffs have been upgraded to 2.5 MV or more; a large number of small tandem machines where set up which allow the physicists to use routinely He beams with energies from 3 to 5 MeV or more. This increase of energy available from small machines leads for example to the extended use of the 3.05 MeV narrow resonance of 160( (Y,a)160. However, it seems that relatively little advantage was taken of the higher He beam energies available for using 3He-induced nuclear reactions. The latter are employed mainly for charged-particle activation analysis with relatively large accelerators. Note, + Work supported by GDR 86 of CNRS. ++ Present address: Central Research Institute for Physics, P.O. Box 49, H-1525 Budapest, Hungary. 0168-583X/90/$03.50 (North-Holland)
0 Elsevier Science Publishers B.V.
however, that Gossett [2] used 3He-induced reactions to measure 160 and 12C, but without remeasuring the corresponding cross sections. We recently undertook the exploration of the use of 3He beams for NRA. The latter have some advantages among which we may quote: - Much lower neutron yields at medium energies than those from deuteron beams, thus extending the range of nuclear reactions within reach in laboratories which are not protected with heavy biological antineutron shielding, a most common situation. - The possibility to carry out simultaneous 3He RBS and NRA measurements. The mass and depth resolution are only moderately reduced in 3He RBS as compared to 4He RBS, while p- or d-RBS may be used in very special cases only. Such simultaneous measurements are a great asset in many experiments, the complementary features of NRA and RBS being taken advantage of. They are moreover absolutely crucial for in situ measurements of physico-chemical processes, when for example the kinetics of thin-film solid-state reactions at high temperature, such as oxidations or reductions, phase changes, etc., are recorded in real time. In view of the efficient use of 3He RBS, the classical simulation program RUMP [3] has been modified so as to accept 3He particles. - Chemical beam-target interaction effects are reduced as compared to hydrogen beams which might induce reduction phenomena during in situ experiments at high temperatures.
F. Abel et al. / Use of the ‘60(3He, aj’%
In this paper we present our first results on 160 quantitative analysis with the ‘60(3He, u)t50 reaction (Q = 4.9140 MeV), this reaction presenting some other additional advantages over the more traditional r60(d,p)170* rea ct i on of widespread use at rather low energies, typically Ed < 900 keV, or with respect to the 160( (Y 01)160 resonant RBS technique. These advantages will bl discussed in detail. The cross section of the 160(3He, a)150 reaction presents a strong peak near 2.4 MeV, with the yield decreasing fast and monotonically towards zero at energies below the wide resonance. This reaction has been extensively studied by Bromley et al. [4]. At 8,, = 90° the reaction is nearly cut off below 2 MeV, while this is not the case for the parallel 160(3He,p)18F reaction or for 160(’ He, or)150 at larger angles. Such a cutoff may be of great advantage when measuring the content of oxide fibs deposited on oxygenated backings, a common situation for thin-film high-T, superconductors. Thus 160(3He, aji70 has a depth discrimination capability with respect to 1aO(d,p)170* which may be most welcome, as the cross section of the latter reaction does not present the corresponding features.
2. Experimental The cross section curve of the 160(3He, or)“0 reaction was remeasured at Blat.,= 90 o from 1600 to 2600 keV. u-particles were detected with a low-resistivity semiconductor detector with a thin enough depletion zone so as to achieve full ~~~~ation between u-particles and protons from the parallel i’O( 3He,p)‘8F reactions or other interfering reactions like *2C(3He, p)14N. Such a discrimination may be obtained without resorting to a dE/dx type detector provided the pulse-shaping time constant r is short enough [5]. We used an ORTEC 527 amplifier with r = 0.5 us, in conjunction with 500 G cm Intertechnique ion-implanted detectors with a nominal depletion zone of 50 pm under 20 V bias. A 10 pm thick aluminium foil was used to stop the scattered 3He beam particles. Ahuninium was preferred here to Mylar in view of the in situ experiments at high temperature [6] to be carried out with our setup. o-p discrimination was obtained by reducing the detector bias until the counts in the valley between the proton high-energy tail and the OLlow-energy tail were negligible even for thick oxygenated targets. Depending on the detector used, this voltage ranged from 10 to 20 V, the maximum proton pulse height being equivalent to about 1.8 MeV, the corresponding depletion zones were around 45 urn thick. The detector fitted with a 13 mm diameter collimator was placed at 65 mm from the beam impact point. In this geometry used for high detection efficiency, the
reactionfor studyingU-containingthinfihs
101
kinematic energy spread of the detected a-particles is large 171; the shape of the ol-spectrum was hence discarded and only its integral was considered. The thin oxygen target used for the measurements was a - 230 A thick SiO, layer on silicon. No appreciable a-yield from the silicon itself was observed. The precise 160 content of the SiO, layer was checked with the r60(d, p)l’O * reaction, by comparison with a classical Ta,Os absolute 160 standard [8]. The target was tilted to 45O off the beam. Carbon deposition at the 2 mm diameter beam impact point was minimized by operating an internal liquid-nitrogen cold trap during the measurements. The impact point was changed several tunes and the consistency of the yield from overlapping energy regions was checked. No change in the target thickness was evidenced during an experiment for the 3He+ currents used, ranging from 300 to 600 nA. The yields were recorded with 100-200 uC per point according to the energy region. The experiments were carried out with the 2.5 MV HVEE single-ended Van de Graaff of our laboratory. Particular care was taken for the energy calibration of the accelerator generating voltmeter. The energy scale was defined just before the excitation curve measurement by recording with a 2 in. X 2 in. BGO y-ray detector at @,, = O” the classical relevant 27Al(p, y) reference resonances and two strong narrow resonances at higher energies from 24Mg(p, y), at 2010 f 10 and 2400 k 10 keV. Thin MgO targets, sputter-deposited on tantalum were used. Although this energy scale reflects the uncertainties as quoted in the literature, on the “Mgt’p, y) resonances [9] the position of the measured cross-section peak is reproducible with better precision with respect to these easily recorded resonances. Moreover, it seems, according to our energy calibration measurements that the best estimate of the real position of the “2010 keV” resonance of 24Mg(p, v) is 2002 keV. The peak region of the excitation curve was recorded by using an automatic electrostatic-energy scanning device designed in our laboratory [lo].
3. Results Fig. 1 shows the excitation curve of the I6 0( 3He, ,)I5 0 reaction at 8,, = 90 ‘. The peak of the cross section is at 2370 keV, or more precisely 31 keV below the “2400 kev” resonance of “Mg(p, y). The maximum cross section is close to 3.4 mb/sr. The data are available on floppy disk. These values of E, and o, max should be compared to those reported by Bromley et al. [4]: 2420 keV and 4.4. mb/sr. The discrepancy is most probably due to calibration problems in this pioneer experiment; in particular, Bromley et al. could use only one calibration energy at 1450 keV. I. NRA
F. Abel et al. / Use of the ‘60(3He, a)150 reaction for studying O-containing thin films
102
t;
?:
10x
5000
..*
*.***
10500
*.....
.a* **
2000
Energy
[ keV]
.**... 1 i 2500
Fig. 1. Excitation curve of the 16q3He, a)“0 reaction at Blab= 90”, for a SiO, target containing 1OOX1O’5 ‘60/cm2, tilted to 45O of the beam, Q = 31.4~ 10W3 sr. The vertical scale may be converted into a cross-section scale in mb/sr by the multiplication factor 1.8 x 10m4. The energies E1 and E, correspond to the z4Mg(p, v) narrow resonances mentioned in the text.
We have thoroughly investigated neutron flux yields when using 3He beam: low neutron fluxes during such measurements are one of their main practical advantages, as stated above. The measurements were carried out with the help of the Nuclear Safety Department of the French Atomic Energy Commission; the results are given in mRem/h equivalent. Let us recall that the standard permissible rates for 2000 hours of work per year are: 2.5 mRem/h for directly involved workers, 0.75 mRem/h for indirectly involved workers and 0.25 mRem/h for the public. With a 2.6 MeV 3He+ beam and - 200 nA current on target we got the following typical results: - At 90 cm from the center of the deflecting magnet, at the side of the deflection: 4.5 mRem/h. Note that our magnet operates since 21 years and the walls of its chambers are thus heavily l2 C-contaminated. - At 45 cm from a bulk carbon target: nearly isotropitally 4 mRem/h. The background with beam measured with a clean tantalum target was - 0.7 mRem/h, while for bulk Al,O, and SiO, targets the rates were only slightly higher. This confirms that at 2.6 MeV bombarding energy the main source of neutrons is the 12C(3He,n)140 reaction, which has a threshold at 1.45 MeV. _ At 40 cm from the beam energy control slit system and from the main collimator, perpendicularly to the beam direction: 4 mRem/h. In conclusion, in normal experimental conditions, the main sources of neutrons are the magnet and the slit-collimator assembly. 3 m away from these units, i.e. one storey above or below the facility, the rate is lower than around 0.5-l mRem/h, i.e. might be tolerated even for the public for 500-1000 hours per year.
Let us note that with - 400 n4 beam on target the yields from the magnet are not much higher, as higher beam currents on target are mainly obtained by improving the focusing: the number of particles lost in the magnet does not increase in proportion to the on-target beam currents. It is clear that these results provide only a typical example, as each machine may behave differently. Preliminary neutron flux measurements are required for each particular facility and a neutron survey detector should be permanently installed near unprotected IBA facilities, chiefly for the case when higher neutron fluxes would be generated either from particular targets or at higher energies.
4. Discussion While having the marked advantages outlined above, for I60 determinations this reaction has the main disadvantage of presenting no plateau in its cross section. Hence the measurements must always be corrected for cross-section variations with depth and the use of a NRA spectrum simulation program is required if good accuracy is aimed at. We systematically used the program SENRAS [7]. Let us first consider the simple case when we want to measure the thickness, i.e. the overall oxygen content, of a thin oxide layer on an oxygen-free backing. This may be performed ideally with the l6 O(d, p)i’O * reaction at 900 keV except if, as stated above, no biological antineutron shield is available; the *H beam induces chemical changes in the target and/or simultaneous RBS is a must, like for in situ kinetic studies. Moreover other nuclei may produce (d,p) interference effects absent in 3He-induced NRA: a typical case is that of 160 determinations with the (d, p) reaction in presence of 14N, which may be far from trivial. Fig. 2 shows the calculated yield as a function of thickness for the typical case of CuO films, tilted to 45”, for various fixed bombarding energies E,. A saturation effect shows up: as the thickness of the oxide layer increases, the exit energy decreases and the corresponding cross section decreases monotonically for large enough thicknesses. In the case considered, the best bombarding energy is slightly above the resonance energy E,, the counts being nearly proportional to thicknesses up to - 1500 X 1015 at/cm2 of CuO. Above - 2500 x 1015 at./cm* the deviation from linearity is too strong to be compensated by a simple correction: the back region of the films contributes much less to the counts than the front region. We reach here the limits of simple 160 content measurements with this technique. The situation is similar to that which shows up in the PIXE technique, where the cross section decreases also monotonically with energy.
F. Abel et al. / Use of the t60( ‘He, a) “0 reactzonfor studying O-containing thin films
103
ioooo ~ ---
6000
$
Eo=2200 keV Eo=2350 keV Eo=2400 keV
. . ‘-
..
,:.
6000
$ 2 4ooa 2000
C 1000
Thickness
3000 2000 [ 10’s atom/c%‘]
5000
Fig. 2. Counts for - 50 KC beam dose in a typical geometry from CuO films of increasing thicknesses (expressed in overall at/cm*) for various beam energies I& The target is tilted to 45 o off the beam.
Another
important
determinations, surements
situation
like oxygen
arises
uptake
at high temperature
I
I
”
2 1
3
a
!
0.00
0.20
0.40 x in
0.80
1.00
CuOx
Fig. 4. Signal-to-noise ratio as a function of x for CUO, films of various Cu contents deposited on a thick SiO, backing, for E,, = E,. The ratio improves rapidly when the thickness along the beam increases: the effect of changing the tilt from 4.5 ’ to 60° (i.e. multiplying the thickness by 6) is illustrated for a 1250 X 10” Cu/cm’ film.
for stoichiometry
or loss kinetics
mea-
and in situ. Fig. 3 shows
the results of calculations for the typical case of CuO, films, containing 1250 X 1015 Cu/cm2. The corresponding in situ experiment with simultaneous 3He RBS is described in ref. [6]. It appears from this figure that the best choice for getting a response as linear as possible in x for films with formula CuO, is a bombarding energy such that E, is reached by the slo~ng-doe 3He particles just in the middle of the film for x = 1, in this case 2440 keV. Let us emphasize here that for very thick oxide films (or for bulk oxides) the a-yield is thickness-independent and depends only on the stoichiometry and on the various stopping powers involved. In this case stoichiometries may be directly compared, the measure-
6000
go;;; ~~~1
CIrp
1250~10~~
Cu/cma
Fig. 3. Counts in the same conditions as in fig, 2 for CnO, films of various compositions. The optimum energy E, is such that for a film with x = 1, En is reached in the middle of the f&n, here E,, = 2440 keV.
ments being backgro~~d-free (except from possible interference from some other light nucleus). We consider now the important case when the oxygen concentration must be measured in a film deposited on an oxygenated backing. If the energy loss of the 3He beam in the film of interest is large enough, the contribution from the oxygen in the backing will be small and the signal-to-noise ratio favourable: the cutoff effect is then put to benefit. Fig. 4 shows the calculated signalto-noise ratio as a function of x for CuOX films with various Cu contents deposited on thick SiO,; bombarding is at peak energy E, = 2370 keV which seems here nearly optimal, the targets being tilted to 45 O. A typical case is that of a film containing 1250 x 1015 Cu/cm2, for which the signal-to-noise ratio is 1.4 for x = 1. This ratio may be improved by further tilting the target: for a tilt of 60 o we get the ratio 2.7. For thicker films much higher signal-to-noise ratios may be reached. Finally we discuss how the use of the I6 O( 3He, ,)I5 0 reaction compares with RBS and in particular with the use of the 3.05 MeV resonance of 160(01, a)i60. If pure oxides are considered, variations of oxygen content may be evidenced in RBS by observing the plateau heights corresponding to a heavy component. This is illustrated by the results on CuO, presented in ref. [6]. However, for high-Z materials like YBa.,$u,O,_, the variations of the plateau height with oxygen content, i.e. 6, are small and this technique is not sensitive enough, as the stopping power per oxygen atom as compared to the overall stopping power is small. It turns out however, that such materials are ideally analyzed with 160( 3He, a)i’O as the corrections discussed above decrease when the overall stopping power variations with oxygen content decrease. Thus the two techniques are complementary: the less favourable is RBS, the better is 160{3He, t1)“O. I. NRA
104
F. Abel et al. / Use of the 160f 3He, CT)‘~Oreaction for studying O-~~n~~‘~~ng thin films
The resonant RBS detection of I60 may be well suited for samples such that the background under the oxygen peak is small enough. As soon as the oxygen peak sits on a large background like for thin oxides on heavy backings or thick oxides of high-2 elements, the 160(a, 01)160 resonance lacks sensitivy due to a small signal-to-noise ratio. Typically this ratio is - 0.25 with resonant RI% for thin YBaCuO films on yttria-stabilized zirconia and - 0.1 for thick YBaCuO films, while it may reach easily 3 or more with the r60( 3He, a)“0 reaction, for well chosen film ticknesses.
4. Conclusion The 160(3He, cw)“O reaction may present marked advantages for 160 dete~nations with respect to the reaction and to the ‘%(a, (u)r70 reso%(d,p)r’O* nant backscattering technique. As discussed, the choice between these three methods depends on the experimental conditions and requirements and also on the type of IBA facility used for the measurements: maximum available He energy, existence of a biological antineutron shield. The results on signal-to-noise ratios presented here may be improved if, in addition to their integral, the &ape of the a-spectra is also analyzed. This will be attempted by optimizing the detection conditions so as to improve energy resolution, without losing too much in counting rate, an essential parameter for in situ measurements. At 8 = 90 O, real depth profiling of 160 with this reaction is difficult due to geometric kinematic-energy spread, except when very small solid angles and absorber-free o-detection is used, like by Gossett [Z]; however, such a technique is only suita-
ble for films deposited on very light backings, such as carbon. Depth profiling will nevertheless be attempted at large detection angles 8, accepting a cross section which is lower and which does not exhibit a “cutoff” effect at B,, = 90 O, but where the kinematic spread is much reduced. We thank 0. Jaoul for providing us with thin MgO targets and H. Quechon for the measurements related to biological hazard due to neutron emission induced by 3He beams.
References III D. Dieumegard, D. Dubreuil and G. Am&, Nucl. Ins&. and Meth. 166 (1979) 431. 121 C.R. Gossett, Nucl. Instr. and M&b. 218 (1983) 149. 131 L.R. Doolittle, Nucl. Instr. and Meth. B9 (1985) 344 and B15 (1986) 227. [41 D.A. Bromley, J.A. Kuehner and E. Almquist, Nucl. Phys. 13 (1959) 1. r51 G. Amsel, J.P. Nadai, E. d’Artemare, D. David, E. Girard and J. Mot&in, Nucl. Instr. and Meth. 92 (1971) 481. 161 F. Abel, G. Amsel, C. Ortega, J. Siejka and G. Vizkelethy, to be published. 171 G. Vizkelethy, these Proceediugs (Ninth ht. Conf. on Ion Beam Analysis, Kingston, Ontario, Canada, 1989) Nucl. Instr. and Meth. B45 (1990) 1. [81 G. Amsel, J.P. Nadai, C. Ortega, S. Rigo and J. Siejka, Nucl. Instr. and Meth. 149 (1978) 705. 191 P.M. Endt and C. Van der Leun, Nucl. Phys. A214 (1973) 130. 1101 G. Amsel, E. d’Artetnare and E. Girard, Nucl. Instr. and Me& 205 (1983) 5.