"Total" Ion Beam Analysis – 3D imaging of complex samples using MeV ion beams



Download 99.18 Kb.
Date conversion20.10.2016
Size99.18 Kb.
"Total" Ion Beam Analysis – 3D imaging of complex samples using MeV ion beams

© C.Jeynes, 3rd April 2012

University of Surrey Ion Beam Centre, Guildford GU2 7XH, England

An article in the Ion Beam Methods Chapter of the Wiley Characterisation of Materials (2nd edition) on-line book


Introduction


In this Chapter the synergy between a number of closely related techniques for thin film depth profiling are described; they all use ion beams from MV accelerators as probes. These include the nuclear methods: RBS, EBS, ERD, NRA (and see Particle Scattering in the Common Methods Chapter). But they can also include PIXE (see Atomic Excitations in the Common Methods Chapter). See Table 1 for the expansion of the acronyms and references to the list of the detailed articles on individual techniques: this article will not describe the techniques themselves but will concentrate specifically on the synergisms available. I will use acronyms for complementary techniques freely: a Glossary for these can be found in the Introduction to this Chapter (Ion Beam Methods).

"Total-IBA" is operating when multiple IBA techniques are being handled self-consistently to obtain more information than the sum of that available from each technique handled separately [1]. We will show that the sum of the whole is far more than the sum of the parts, to the extent that large new classes of samples become tractable and powerful new types of characterisation become feasible: the various IBA techniques are in fact strongly complementary. Indeed, we believe that even chemical tomography is feasible with these new techniques.

The alert reader will object that we are here only stating the obvious: it is easy to find examples showing that this complementarity has always been recognised. For example, Feldman et al presented a paper combining He-RBS and He-PIXE to the first Ion Beam Analysis Conference nearly forty years ago in 1973 [2]. The Abstract (not available electronically) is informative for us :-

Anodic oxide films on GaAs have been studied by the combined use of He back-scattering [sic] and He-induced X-rays. Back-scattering is hampered by the lack of mass resolution between Ga and As. X-ray analysis has excellent mass resolution but poor depth resolution. This poor depth resolution is overcome by increasing the effective thickness of the films by entering at grazing angles and making use of the property that the He-induced X-ray cross-sections fall steeply with decreasing energy. This technique and the methods of data analysis are discussed in detail. The anodic oxide films are found to be deficient in As within 200Å of the surface and to have a Ga:As ratio of approximately 1:1 for the rest of the oxide. On heating to 650°C most of the As diffuses out of the films.

This early use of RBS/PIXE is exemplary, and includes an explicit awareness of the strengths and weaknesses of each technique. We shall underline these below, and show why it is only recently that the idea of using the IBA techniques self-consistently has been picked up and made usable by the analytical community.

We will first briefly survey the individual techniques, particularly with 3D chemical imaging in mind. This survey will overlap surprisingly little with the separate articles treating each techniques. We will then show why the nuclear and atomic communities have pursued largely separate courses over the last 40 years. We will describe the recent advances that have made Total-IBA possible. Finally we will show the extraordinary power of the new technique. We expect that IBA computed tomography (IBA-CT) over sample sizes ~20 m with deep sub-micron voxel sizes will become available in the next five years or so. (Provided civilisation survives the current crises.)


IBA Techniques not considered as " Total-IBA "


Total-IBA envisages an MeV ion beam striking a sample surrounded by detectors for all the various reaction products: backscattered (RBS and EBS), forward scattered ("RBS" or off-axis STIM) and forward recoiled (ERD) particles; particles from nuclear reactions (NRA), and photons from both nuclear (PIGE) and atomic (PIXE) excitation.

LEIS and MEIS are both RBS techniques, but they are low energy and usually applied to surface science problems involving a few nm at most. Total-IBA is applicable to thin film applications involving surface layers (or samples) up to ~20 m thick. LEIS and MEIS are complex and are usually used alone (or with other surface science instrumentation – XPS, LEED etc).

Dynamic (depth profiling) SIMS is a destructive (sputtering) technique using entirely different methods which we will not consider here. However depth information from SIMS is commensurate with Total-IBA information, and in principle (and practice too [3]) could be incorporated. Note here that Ellipsometry in the Optical Imaging and Spectroscopy chapter is also potentially (and practically [4]) a commensurate technique, as are polymer diffusion studies ([5] [6]; see Small Angle Neutron Scattering in the Neutron Techniques chapter) and protein crystallography ([7]; see Single-Crystal X-Ray Structure Determination in the X-Ray Techniques chapter).

AMS, IBIC, He-Ion- and Field-Ion-Microscopy, and the Radiation Damage studies are quite different types of characterisation or materials modification applications.


Strengths and Weaknesses of Single IBA Techniques


All the analysis methods with particle resultants (RBS, EBS, ERD, NRA) are intrinsically sensitive to depth directly through the energy loss of the ion beam in the sample, since the detected particle energy is always analysed. (There are exceptions for NRA in many cases either where there is little depth information intrinsically in the signal or where it is degraded by range foils or the kinematical broadening of large detectors.).

However, none of the analysis methods with photon resultants (PIXE, PIGE) are usually very sensitive to depth. The exceptions in PIXE are where closely adjacent matrix elements bring the absorption edges into play, and in PIGE where resonant nuclear reaction allow depth profiling by stepping the beam energy.


Elastic (Rutherford and non-Rutherford) backscattering


RBS is the simplest technique. The yield is proportional to the square of the atomic number and inversely proportional to the square of the scattering particle energy, and the kinematical factor for a head-on scattering event of a nucleus mass M1 on a target nucleus mass M2 is {(M2-M1)/(M2+M1)}2 (see Eqs.1,3 in Elastic Backscattering of Ions for Compositional Analysis). Thus light element signals are at a lower energy than (and therefore superimposed on) heavy element signals. This means that the sensitivity for light elements is low both because the absolute cross-section is low (1 barn/sr at 170° scattering angle for 1 MeV 4He on Si) and also because the light element signal frequently has a heavy element background.

The great strength of RBS is that because the cross-sections are accurately calculated with a Coulomb potential, the spectra have absolutely traceable quantitation [8] [9]. Because the charge solid-angle product (see Eq.8 in Elastic Backscattering of Ions for Compositional Analysis) can be readily determined from the spectral data and the material stopping power, and because in one important case (He RBS of Si) the stopping power can be accurately determined from a certified standard material [10] we expect that 1% traceable accuracy (that is, 2% at 95% confidence)

For EBS there may be greatly enhanced cross-sections many times Rutherford which frequently are highly useful to improve the sensitivity to light elements [11] [12]. On the other hand, protons on Al or alphas on Si (for example) have complicated cross-section functions with very many sharp resonances but average cross-sections close to Rutherford. In these cases EBS is distinctly unhelpful, being very complicated but just as insensitive as RBS.

Elastic Recoil Detection and Nuclear Reactions


ERD and NRA are good Total-IBA techniques when used with light particle beams for which other signals (RBS/EBS and PIXE) are also available. He-ERD is a very useful beam for determining H depth profiles in a variety of materials, and NRA is indispensible for sensitivity to certain light atoms (Z<14) in certain contexts. Both of these are standard techniques in use for decades. The limitation of both light ion ERD and NRA is that they are usually sensitive to only one isotope present in the sample, so that they both need a separate calibration and the analysis of the sample depends on complementing with other techniques.

Particle-Induced X-ray Emission


PIXE is not usually considered to be sensitive to depth since the detected X-rays at any particular energy have been integrated along the whole beam path. Most published PIXE work has been either on samples effectively homogeneous in depth, or on samples where the depth profile is not important, or on very simple layered samples whose structure is already known. Indeed, one major PIXE code (GUPIX [13]) does not currently support the analysis of diffusion (or any other sort of complex) profiles. This is not due to a basic limitation of the code: it is just the way it has been implemented – the authors took the view that such complexity was of no interest!

However, because PIXE does not have direct depth information does not mean that there is no depth information folded into the signal. SEM-EDS and XRF are analytical methods comparable to PIXE (using respectively electrons and X-rays as the probe beam: see the ATOMIC EXCITATIONS article) and commercial software packages for both SEM-EDS and XRF are commonly used routinely to give "layer thicknesses". And in surface science angle-resolved XPS is also used routinely to give ultra-high depth resolution. "Differential" PIXE using beam energy ([14] [15] [16]) or beam geometry (that is, an "angle-resolved" method [ref.1] [17]) variation has been recognised and used for a very long time. Ahlberg [18] showed that a single measurement using these equivalent effects had sensitivity to the depth distribution of an element through the yield ratio within characteristic line groups. But the sensitivity is limited, as he showed rather elegantly.

On the other hand, the sensitivity of PIXE is enormous. For 3 MeV protons on Si the K-shell production cross-section is about 87 barns/sr. The RBS cross-section for this beam on Si (170° scattering) is 7 mb/sr, four orders of magnitude smaller!

Barriers to Synergy


If Total-IBA is so wonderfully powerful, and has been recognised as such for nearly 40 years, why make so much fuss about it now? Why isn't everyone doing it? There are essentially three main reasons for this (two good and one bad) and a few minor reasons (we are here talking mostly about the synergy between the particle and the photon techniques). The position is that, from a practical point of view, Total-IBA has only been routinely feasible in the last five years or so.

Barrier 0: Separation of Nuclear and Atomic Communities


The most obvious point is that atomic (PIXE) and nuclear (RBS/EBS/ERD/NRA) processes are entirely different, and have completely different formulations. Their descriptions (see the appropriate individual articles) use different physics and involve a completely different literature. And technical problems in both the atomic and nuclear physics communities persist until today. Neither field was mature enough until quite recently to admit a usable synthesis between them

Barrier I: Code Limitations


Total-IBA is only needed for relatively complex samples. For such samples the particle scattering codes must take many second order effects into account. These are described in detail in the recent IAEA-sponsored intercomparison and review of particle-scattering codes [19], with only two codes recognised as "new generation", that is, able to model all these effects. These are the DataFurnace [20] and SIMNRA [21] codes, which are both only just over a decade old. Recent reviews are available (respectively [22] and [23]). The fitting accuracy available with these new codes is extraordinary, especially compared to what was considered acceptable forty years ago (see Fig.14 in Elastic Backscattering of Ions for Compositional Analysis: this is Fig.1 in [ref.19]; the other Figures in the EBS article are also impressive).

We should note parenthetically that the knowledge of stopping powers of the probing beam in the materials being analysed is critical to interpreting particle scattering spectra, and the semi-empirical database of these data regularises a massive experimental effort, much of which is fairly recent. Modern knowledge is far more accurate than was available a generation ago (again, see the EBS article for more details, and [24]).

Lastly, and crucially, integrated codes allowing analysts to use Total-IBA routinely have only recently become available. OMDAQ [25] (also see [ref.6]) has been available for over 15 years, but it is mainly a microbeam PIXE code with only rather simple facilities for fitting particle spectra. IBAlab [26] does allow simulation of Total-IBA analyses, which is an important step. But DataFurnace, one of the "new generation" particle-scattering codes, only acquired a PIXE module in 2006 [27]. It is clear, from all the published Total-IBA work so far, that good fitting of the particle spectra is essential to be able to make full use of Total-IBA synergies. Moreover, self-consistent fitting of multiple spectra is crucial to Total-IBA applications.

Barrier II: EBS cross-sections


Table 1 in the Elastic Backscattering of Ions for Compositional Analysis article shows that the boundary between RBS and EBS is exceeded for 2 MeV proton beams onto all atoms lighter than Fe (see also [ref.10]). This is where the beam energy is so high that the Rutherford approximation of the Coulomb interaction of point charges breaks down, and a proper quantum mechanical treatment is needed for the interaction. The EBS article explains that in the last century, ion beam analysts were generally forced to use empirical scattering cross-section functions to make use of EBS. These are very frequently a strong function of scattering angle, so that typically everyone measured their own data.

However, 2 MeV (and higher energy) proton beams are typical for PIXE analysis (see the Particle-Induced X-ray Emission article), so that to interpret particle spectra collected simultaneously with X-ray data was often difficult. Much data did exist of course, but its quality was always suspect even where the scattering angle was appropriate. Therefore, there was a strong disincentive for using Total-IBA techniques on the grounds that the particle data were too often intractable.


Barrier III: Low statistics particle spectra


The other perceived barrier to doing PIXE and particle scattering simultaneously was that, as mentioned above, the X-ray production cross-sections for proton beams hugely exceed the particle scattering cross-sections. Thus, for microbeam (imaging) PIXE applications, where the beam current is usually far smaller than for normal RBS, the particle spectra typically had very few counts.

The temptation is to consider that the amount of information in a spectrum is proportional to the number of counts in it. Of course, this is entirely false! Figure 1 demonstrates this quantitatively in the case of a mixed silicide. If this were a Total IBA analysis, the particle spectra would be needed to determine the layer structure of the sample so that the PIXE spectrum could be properly quantified. Different layer structures can give vastly different absorption behaviour, which is what is needed to interpret the relative line intensities. Even extremely noisy spectra can give qualitatively correct layer structures, with quite accurate thicknesses. For such mixed silicides with adjacent elements the absorption corrections can be very large.


Other Barriers


Because of the other barriers to Total-IBA, and in particular because of the unavailability of codes able to handle multiple spectra self-consistently, ion beam analysts frequently did not put multiple detectors into their measurement systems. There was another justification for this too: that the problems amenable to PIXE and those amenable to particle scattering tended to be in distinct classes.

So, for whatever reason, IBA has historically been split into the "PIXE" and the "RBS" camps, roughly speaking. Of course, many labs did both, but not usually together. It is telling that only very recently a "Total-IBA" paper from a major lab in a high impact journal was published, but where the techniques reported did not include self-consistent treatment of the data ([28]: this is despite the fact that the same lab did previously report such a self-consistent treatment on some of the same samples [29]!)! Forty years after Feldman's paper there is again recognition that our historic methods are heavily limited.


TOTAL-IBA: Synergies between IBA methods


Much work has been published recently using powerful methods involving synergies between various of these IBA techniques used together self-consistently. We shall survey these applications fairly systematically, leading towards the goal of 3D elemental and chemical imaging.

RBS is good for heavy elements in a light matrix and typically the mass resolution is not very good, so that only fairly simple things can be said about fairly simple samples. On the other hand, PIXE on its own cannot compete on price against the almost equivalent XRF (there is even an explicit comparison of PIXE with XRF showing their near-equivalence [30]). But putting these techniques together allows the strengths of the one to compensate for the weaknesses of the other so that the combination is extraordinarily powerful.

In the previous section we surveyed the reasons for these synergies to have been avoided until now: here we survey existing Total-IBA examples – that is, examples of the synergistic use of multiple IBA methods.

Ambiguity


Of course, the underlying reason for using multiple spectra self-consistently is that individual spectra are always more or less ambiguous. Trivially, in Fig.1 the spectra do not identify Fe and Co since the mass resolution is too poor and in any case in RBS there is always a mass-depth ambiguity: we know these metals are present since they were used to make the samples! But the combined use of PIXE would unequivocally identify the metals, and a self-consistent data treatment of the particle and photon spectra would confirm that the PIXE line intensities were consistent with the numbers of atoms counted by RBS. A similar but much more complex example (requiring quantitative as well as qualitative analysis) is the measurement of the functionalisation of carbon nanotubes, where the catalyst signal must be accounted for to extract the desired light element signal [31].

Multiple detectors or, equivalently, multiple beam incidence angles are regularly used in RBS to identify surface signals and relieve the worst mass-depth ambiguities. So Fig.17 in the EBS article shows a zirconia/silica multilayer sample whose spectrum is unequivocal only because the analysis had data from two beam incidence angles [32]. Fig.16 in the EBS article shows a complex case where two detectors and two incidence angles are used, and all the information is used to obtain unequivocal information about the samples. In both of these examples it was essential to impose chemical constraints on the data to interpret them unequivocally, as discussed long ago by Butler (1990) [33].

There are many ways of overcoming the ambiguity of any particular measurement, and these are discussed in depth in the review by Jeynes et al (2003) [34]. We show examples below of "Total-IBA", that is, the use of multiple techniques analysed self-consistently.

Synergy Examples I: RBS/ERD


Fig.1 in the EBS article shows a case where it was essential to take the hydrogen content of a sample into account, even though it was not the required measurand, because the uncertainty of the final result needed to be as small as possible. With He RBS one needs only to tilt the sample such that the H-recoils can escape and be detected to obtain simultaneous H-ERD data. One "invisible" element can of course be inferred from spectral data, but this depends on accurate knowledge of the energy loss (which in any case is very low for H), and usually only rough information can be extracted in the absence of a direct signal.

Synergy Examples II: RBS/EBS/ERD/NRA


Figure 2 shows a complex analysis applied to an important case. the Joint European Torus (JET) is a long-running tokamak experiment where many detailed analyses are needed as a function of position of the fusion vessel linings. The distribution of the light elements (and the heavy contaminants) is of great interest, and this analysis using data from multiple beams can be done entirely automatically using simulated annealing [35] on the whole (very large) dataset.

In this example of Total-IBA the heavy element profiles are given by both He-RBS and H-RBS, where the H-RBS has greater information depth but lower sensitivity. O and Be profiles are given by EBS (with a significant cross-section enhancement over Rutherford), and the H and D profiles are obtained by He-ERD (using a range foil) with the D profile being obtained independently by 3He-NRA.


Synergy Examples III: RBS/EBS/PIXE


A fully self-consistent and convenient PIXE/BS analysis code based on the DataFurnace [ref#35] and DATTPIXE [36] codes was introduced in 2006 [37]. This was used to analyse Niépce's heliograph of 1827 [38], a 19th century reproduction of Frans Hals' La Bohémienne [ref#29], oxidation of carbon nanotubes [ref#30] and photovoltaic and ferroelectric materials [39] [40] [41], and so-called "Darwin glass" samples (see Figs. 4 & 5, discussed below). In all these cases the PIXE signal was crucial in quantifying signals that were either trace elements with no significant particle scattering signal, or elements inextricable from others in the particle scattering signal. In all cases the samples had more or less complex layering, so that the PIXE signals could not be quantified without the depth information in the particle scattering spectra. For example, the important heliograph of 1827, as the "first photograph", is a priceless record in the history of photography now in the collection of the Louvre museum. It suffers from corrosion and the conservators wanted to know what exactly was the nature of the damage. The surface is a Pb/Sn alloy, and the questions are : Is the corrosion oxidation? If so which species is oxidising? What is the thickness of the modified layer? RBS is able to give the Pb/Sn ratio at the surface, PIXE can give the total Pb and Sn content, and EBS gives the O (and C!) profile. Using Total-IBA with the external beam that is standard at the Louvre (see the recent review [42]) it is clear that the tin is oxidising, and the depth of the corroded layer is also determined. This information could not easily be obtained by IBA without self-consistent data analysis, although this particular case is simple enough that iterative methods would have worked [ref#28]. Note that these samples are relatively rough, and in principle the particle scattering data can even quantify the roughness [43].

A further example is shown in Figure 3 of a significantly harder case which required both differential PIXE and high energy resolution PIXE (HR-PIXE) as well as the particle scattering spectra. This work aimed to evaluate the use of SrTiO3 as a temperature compensator for MgTiO3 films used for filters or oscillators in telecomms devices. A thin layer of SrTiO3 is deposited on the Pt electrode and under the final MgTiO3 film, and the analytical question is: what happens to it? There is too little Sr for the particle spectra to have any sensitivity for it, but it is clearly visible in PIXE. To find out where the Sr is, differential PIXE used 250, 325, 700, 1000 and 1960 keV H beams at glancing exit geometries. For these low energy beams there is no usable cross-section for the Sr K X-rays, and the L lines must be used instead (not shown in Fig.3). But Sr L lies between the strong Si K and K lines, and with the normal EDX detectors the Sr L signal must be inferred with a subtraction of the interfering Si signal that assumes knowledge of the Si K/K ratio. In this work the Si K/K ratio was measured directly using a high energy resolution EDS (energy dispersive) X-ray detector of a microcalorimeter design based on superconducting transition-edge sensors. Then, using the Si K/K ratio measured directly by HR-PIXE, the Sr L signal could be extracted from the differential PIXE spectra, and the depth profile obtained.

There is a further problem that complicates this analysis and is also quantified by HR PIXE: there is a Pt radiative Auger emission (RAE) satellite peak at about the same energy as the Sr L signal. Auger electrons originate in a radiation-less process: these RAE satellites are "forbidden" transitions requiring both photons and "Auger" electrons to be emitted [44]. Detailed use of PIXE data requires a detailed knowledge of PIXE physics, which is not always available despite the "maturity" of X-ray physics! The recent availability of HR-EDS X-ray detectors has underlined this problem.

Synergy Examples IV: RBS/PIXE/MeV-SIMS


Figure 4 shows a remarkable image from an unexpected Total-IBA application using SIMS with an MeV ion beam (MeV-SIMS). Where regular SIMS using a keV ion beam generates sputtered ions by a nuclear displacement process, MeV-SIMS sputtering is due to the electronic energy deposition, and therefore occurs appreciably only for insulating samples. It has already been demonstrated that MeV-SIMS produces a significantly higher proportion of high molecular weight sputtered ions than does keV-SIMS, even using molecular primary ion beams such as C60 [45]. And of course, MeV beams can be used in an external beam analysis, that is, in atmosphere. Since ion beams can be readily focussed it seems that not only is high spacial resolution chemical imaging in atmosphere feasible, but it can also be combined with simultaneous complementary methods than can quantify or otherwise complete the data. MALDI (matrix-assisted laser desorption ionisation) is a powerful and popular technique with similar capabilities, also used at ambient pressure, but it is without either spacial resolution or the complementary information like the PIXE that naturally accompanies MeV-SIMS.

Chemical information, that is, information about the chemical state of the elements present in the sample, is in principle readily available with X-ray techniques. But the measurement of chemical shifts requires an energy resolution at or below 1 eV. This has historically been available only with electron spectrometers (XPS, AES, EELS) or with wavelength-dispersive (WDS) X-ray spectrometry. But third-generation HR-EDS X-ray detectors are expected to also achieve energy resolution comparable to WDS in the near future [46]. We have already cited exciting HR-EDS-PIXE using such (first-generation) detectors; there is no reason not to expect a much more powerful capability to emerge.


TOTAL-IBA: Tomography


Tomography is a 3D imaging technique based on taking a series of slices of a sample. With computed tomography (CT), physical slices are not taken, but a series of images are obtained (by any technique) and subsequently reconstructed by calculation. X-ray tomography (XR-CT) images by radiography, that is, the contrast mechanism is from differential absorption due to density variation. XR-CT is long established, with Cormack & Hounsfield taking the 1979 Nobel prize in medicine. STIM-CT is an almost equivalent (and solved) problem [47].

There has been very significant recent interest in tomographic methods sensitive to the chemistry (stoichiometry) of the sample under investigation using both microfocussed confocal XRF and XRF-CT [48] [49]. We should comment that synchrotron XRF is not essential to this: there have also been serious reports of XRF-CT on desktop tools [50]. There has long been interest in XRF-CT: for example, Brunetti & Golosio in 2001 [51] published an open code [52] capable of this using Hogan’s 1991 algorithm [53]. Great strides have also been made towards a PIXE-CT: see the summary in a recent review [54].

In principle, where radiation damage is an issue, XRF CT is preferable to confocal XRF for obtaining 3D information about samples, since confocal techniques throw away all information not from the confocal plane and therefore a full analysis of the sample must take much longer by confocal methods. Similarly, we believe that IBA CT will be found preferable to XRF-CT because the particle scattering that must accompany the PIXE signal carries much information not available in XRF.

The best sy-XRF-CT spacial resolution so far reported (200 nm, determined by the spot size of the X-ray nanobeam) is by Silversmit et al [55]. This used optimal stepping (pixel size) of 100 nm and 180° scan in 4.5° steps per sinogram. The largest dimension in the sample (a cometary fragment from the Stardust space mission) was 2 m and the measurement took 26 hours beam time. Synchrotron XRF has many very powerful variants. For example, for some samples XANES is very powerful for chemical speciation tomography (see for example Blute et al [56]), but de Jonge & Vogt comment that “this approach to chemically resolved tomography is not generally applicable due to the need for a strong, unambiguous XANES feature for contrast” [57].

There are (at least) two big problems in XRF-CT, one algorithmic and one practical. Fully quantitative XRF-CT will require self-absorption to be properly taken into account. In the XRF community this has been approximated, but the discretised real-space reconstruction algorithm (DISRA [58]), permitting a more exact treatment, is currently used only in the ion microprobe community.

Also, as hinted above, the classical tomography algorithms require much data and many slices, and, even with a relatively non-destructive X-ray primary beam, the radiation sensitivity of the sample dominates the applicability of the method. Moreover, the sy-XRF facilities all depend on moving the sample through the beam, which is a) slow and b) gives extra mechanical problems where very high spacial resolution is the aim. Therefore the IBA methods still appear very interesting for the following reasons: a) a scanning microbeam (or nanobeam) is easy, both being much faster and also avoiding the mechanical problems of moving the sample, b) the spacial resolution is dominated by the beam size, and a 100 nm proton beam for analytical purposes appears to be entirely feasible [59], c) there are more IBA microbeam facilities in the world than there are synchrotrons, and they could become as common as SEMs if desktop IBA tools based on small cryogenic synchrotrons become commercially successful, d) many of the advances happening at the synchrotron facilites are equally applicable to IBA facilities (fast detectors for example [60]).

However, new algorithms incorporating particle scattering data may well be orders of magnitude faster than the classical tomography algorithms, and in addition it may be that an order of magnitude higher spacial resolution is possible, given by the energy resolution of the particle spectra (for samples where 180° rotation is feasible). An interesting example is an analysis of a 20 m geological sample [61]. Figure 4 shows PIXE maps for one orientation of the sample, together with various BS spectra from various parts of the sample. Figure 5 shows the spacial distribution of one principal component of the PIXE data cube in Fig.4, together with the PIXE and BS spectra for that component. But the depth profile can be obtained explicitly for the PIXE/BS data, using Total-IBA, so that from the projection at just one angle the entire 3D elemental map of the sample can be reconstructed, combining the principal components appropriately for each pixel. Many problems remain before the algorithm for this reconstruction can be specified in detail, an algorithm which in any case is quite different from the classical CT algorithms. If it exists, this algorithm will be hugely more efficient. And it seems very likely that it exists.

We have shown that IBA-CT (that is, using the BS signals as well as the PIXE signals) may already be achievable, and should be orders of magnitude more efficient (and therefore much faster!) than pure XRF CT (or PIXE-CT) since a single slice already has very substantial depth information from the particle spectrum. This is important since tomography is rather slow, and its importance is increased since it seems that beam damage severely limits the use of a pure PIXE-CT for important classes of samples [62]. This is also true for XRF-CT [ref#41]. In principle, using the depth information available explicitly in IBA-CT (from the particle signals) must be quicker than unfolding the depth information available only implicitly (and at much lower depth resolution) in the PIXE signals.


Conclusions


In the last five or six years there has been a step-function change in the capability of ion beam analysis, as we have found out how to put the various IBA techniques together in a powerful synergy. The most critical step has been the incorporation of PIXE into Total-IBA, since PIXE and the particle methods complement each other perfectly, and have historically been the most completely separated. Where one has excellent depth resolution the other has excellent elemental discrimination, and a Total-IBA analysis inherits the strengths of both. And since ion beams are readily focussed, Total-IBA mapping (in 3D because of the depth sensitivity) is straightforward and should also have a powerful tomographic capability.

Analysts today should always include both particle and photon detectors in their target chambers, and users should expect them to do so!






Figure 1: Noisy spectra can contain crucial information!

1.5MeV He RBS spectra of a mixed Fe:Co silicide on a Si substrate are simulated from an initial structure (top right), given various charge solid-angle products (left hand column). These simulated spectra are treated as data, fitted (lines) and hence inverted back to depth profiles, with ±1 uncertainties given by a Bayesian analysis (right hand column). Modified from Figs.1&2 of Barradas et al, 2000 [63]. (Reproduced from Fig.27 of Jeynes et al, 2011 [64].)



F
igure 2: "Total-IBA" of JET samples using sequentially collected RBS, EBS, ERD, NRA data.


Top left: 2 MeV 4He RBS; Top right: 2.45 MeV 1H EBS; Bottom left: 2 MeV 4He ERD with inset of 2.3 MeV 3He NRA. Bottom right shows the depth profile obtained through a self-consistent treatment of all the data. (Reproduced from Fig.2 of Alves et al, 2010 [65])

Figure 3: RBS/EBS/PIXE analysis of MgTiO3/Sr TiO3/ bilayer on Pt/Si

Data from Fig.2 of Reis et al, NIMB 268, 2010, 1980-5 [66], with extracted depth profile (bottom, right) reproduced from Fig.4 of Reis et al, XRS 40, 2011, 153-156 [67]






Figure 4: Images of a grid obtained by Total-IBA using 10 MeV O4+

Ta La PIXE and RBS show the metal grid with shadowing contrast depending of the detector position, and MeV-SIMS signals (from the insulating carbon tape) for two different molecular weight fragments. Reproduced from Fig.3 of Jones et al, NIMB 268, 2010, 1714-1717 [68]






Figure 5: "Total-IBA" of an inclusion in a Darwin Glass.

Above: selected PIXE maps showing distribution of Si, Fe, Cu; Centre: BS spectra at varying energies of the resin region showing the 12C(p,p0)12C resonance at 1734 keV; Below: BS spectra at 1.9 MeV for three areas marked on the Si PIXE map (above, left). (See Bailey et al, 2009 [ref#]). After Fig.1 of [ref#1]. Reproduced from Fig.28 of Jeynes, Webb & Lohstroh, 2011 [ref#47]



Figure 6: Principal component decomposition of the data cube of Fig.3.

One component from the principal component decomposition of the data cube of Fig.3 using AXSIA (see Doyle et al, 2006 [69]). This component is one of the several Si-rich components. Reproduced from Fig.2 of [ref#1].



sdfsdfs


Table 1: Glossary, with names of other articles in the Ion Beam Methods Chapter

Term

Expansion

Reference

IBA

Ion Beam Analysis

Introduction

Total-IBA

(RBS, EBS, ERD, NRA, PIXE, MeV SIMS; using MeV ion (micro)-beams for elemental (and chemical) depth profile (including 3D) information.

This article

RBS

Rutherford backscattering spectrometry

Elastic Backscattering of Ions for Compositional Analysis

EBS

Elastic (non-Rutherford) backscattering spectrometry

Elastic Backscattering of Ions for Compositional Analysis

ERD

Elastic recoil detection

Elastic Recoil Detection Analysis

NRA

Nuclear reaction analysis

Nuclear Reaction Analysis and Particle-Induced Gamma Emission

PIXE

Particle-induced X-ray emission

Particle-Induced X-ray Emission

PIGE

Particle-induced gamma-ray emission

Nuclear Reaction Analysis and Particle-Induced Gamma Emission

Channelling

Well-collimated beam aligned with crystallographic axes in single crystals for crystalline defects and lattice location of impurities.

Medium Energy Ion Scattering

SIMS

Secondary Ion Mass Spectrometry (with a keV ion beam) and also MeV-SIMS

Secondary Ion Mass Spectrometry

LEIS

Low Energy Ion Scattering

Low Energy Ion Scattering

MEIS

Medium Energy Ion Scattering

Medium Energy Ion Scattering

IBIC

Ion-Beam-Induced Charge

Ion-Beam-Induced Charge

AMS

Accelerator Mass Spectrometry

Accelerator Mass Spectrometry


References


1 C.Jeynes, M.J.Bailey, N.J.Bright, M.E.Christopher, G.W.Grime, B.N.Jones, V.V.Palitsin, R.P.Webb, "Total IBA" – where are we? Nucl. Instr. Methods B, 271 (2012) 107-118

2 L.C. Feldman, J. M. Poate, F. Ermanis and B. Schwartz, Combined use of He backscattering and He-induced X-rays in study of anodically grown oxide films on GaAs, Thin Solid Films 19 (1973) 81-89

3 Barradas, N.P., Likonen, J., Alves, E., Alves, L.C., Coad, P., Hakola, A., Widdowson, A, Integration of SIMS into a general purpose IBA data analysis code, AIP Conference Proceedings 1336 (2011) 281-285

4 Barradas N P, Keddie J L and Sackin R, Bayesian inference analysis of ellipsometry data Phys. Rev. E 59 (1999) 6138–51

5 Payne, R. S.; Clough, A. S.; Murphy, P.; Mills, P. J., Use of the d(3He,p)4He reaction to study polymer diffusion in polymer melts Nucl. Instrum. Methods B, 42 (1989) 130–134

6 Thompson RL, Hardman SJ, Hutchings LR, Narrainen AP, Dalgliesh RM, pH-Controlled Polymer Surface Segregation, Langmuir 25(5) (2009) 3184-3188

7 Elspeth F. Garman, Geoffrey W. Grime, Elemental analysis of proteins by microPIXE, Progress in Biophysics and Molecular Biology 89 (2005) 173–205

8 G.Boudreault, C.Jeynes, E.Wendler, A.Nejim, R.P.Webb, U.Wätjen, Accurate RBS measurement of ion implant doses in a silicon, Surf.Interface Anal., 33 (2002) 478-486

9 C.Jeynes, N.Peng, N.P.Barradas, R.M.Gwilliam, Quality assurance in an implantation laboratory by high accuracy RBS, Nucl. Instr. Methods B, 249 (2006) 482–485

10 N.P. Barradas, K. Arstila, G. Battistig, M. Bianconi, N. Dytlewski, C. Jeynes, E. Kótai, G. Lulli, M. Mayer, E. Rauhala, E. Szilágyi and M. Thompson, "Summary of 'IAEA intercomparison of IBA software' ", Nucl. Instr. Methods B, 266 (2008) 1338-1342

11 A.F.Gurbich, Evaluated differential cross-sections for IBA, Nucl. Instrum. Methods B, 268 (2010) 1703-1710

12 D.Abriola, N.P.Barradas, I.Bogdanović-Radović, M.Chiari, A.F.Gurbich, C.Jeynes, M.Kokkoris, M.Mayer, A.R.Ramos, L.Shi, I.Vickridge, Development of a reference database for Ion Beam Analysis and future perspectives, Nucl. Instrum. Methods B, 269 (2011) 2011-2016

13 J.L.Campbell, N.I.Boyd, N.Grassi, P.Bonnick, J.A.Maxwell, The Guelph PIXE software package IV, Nucl. Instrum. Methods B, 268 (2010) 3356–3363

14 M.E Fedi, M Chiari, L Giuntini, F Lucarelli, P.A Mandò, The differential PIXE set-up at the Van de Graaff laboratory in Florence, Nucl. Instrum. Methods B, 189 (2002) 56-61

15 G. Demortier, J.L. Ruvalcaba-Sil, Differential PIXE analysis of Mesoamerican jewelry items, Nucl. Instrum. Methods B, 118 (1996) 352-358

16 F. William Reuter and Harold P. Smith, Full range solution for the measurement of thin film surface densities with proton excited x rays, J. Appl. Phys. 43 (1972) 4228-4232

17 W. Pabst, Ion-induced X-ray spectroscopy as a method to determine depth distribution of trace elements, Nucl. Instr. and Meth. 124 (1975) 143-147

18 M.Ahlberg, Simple depth profile determination by proton-induced X-ray emission, Nucl. Instr. and Meth. 131 (1975) 381-384

19 N.P.Barradas, K.Arstila, G.Battistig, M.Bianconi, N.Dytlewski, C.Jeynes, E.Kótai, G.Lulli, M.Mayer, E.Rauhala, E.Szilágyi, M.Thompson, IAEA intercomparison of IBA software, Nucl. Instr. Methods B, 262 (2007) 281–303

20 N.P.Barradas, C.Jeynes, R.P.Webb, Simulated annealing analysis of RBS data, Appl. Phys. Lett. 71 (1997) 291–293

21 M.Mayer, SIMNRA, a simulation program for the analysis of NRA, RBS and ERDA, AIP Conference Proceedings 475 (1999) 541-544

22 N.P.Barradas, C.Jeynes, Advanced physics and algorithms in the IBA DataFurnace, Nucl. Instr. Methods B, 266 (2008) 1875-1879

23 M. Mayer, W. Eckstein, H. Langhuth, F. Schiettekatte, U. von Toussaint, Computer simulation of ion beam analysis: Possibilities and limitations, Nucl. Instr. Methods B, 269 (2011) 3006-3013

24 James F. Ziegler, M.D.Ziegler, J.P.Biersack, SRIM – The stopping and range of ions in matter (2010), Nucl. Instrum. Methods B, 268 (2010) 1818-1823

25 G.W. Grime, The “ Q factor” method: quantitative microPIXE analysis using RBS normalisation, Nucl. Instr. Methods B, 109-110 (1996) 170-174

26 I. Orlić, Shijun Zhou, J.L. Sanchez, F. Watt, S.M Tang, Virtual PIXE and RBS laboratory, Nucl. Instr. Methods B, 150 (1999) 83-89

27 C.Pascual-Izarra, M.A.Reis, N.P.Barradas, Simultaneous PIXE and RBS data analysis using Bayesian inference with the DataFurnace code, Nucl. Instr. Methods B, 249 (2006) 780-783

28 L.de Viguerie, L.Beck, J.Salomon, L.Pichon, Ph.Walter, Composition of Renaissance paint layers: simultaneous PIXE and BS, Anal. Chem., 81 (2009) 7960-7966

29 L.Beck, C.Jeynes, N.P.Barradas, Characterization of paint layers by simultaneous self-consistent fitting of RBS/PIXE spectra using simulated annealing, Nucl. Instr. Methods B, 266 (2008) 1871-1874

30 Giulia Calzolai, Massimo Chiari, Franco Lucarelli, Federico Mazzei, Silvia Nava, Paolo Prati, Gianluigi Valli, Roberta Vecchi, PIXE and XRF analysis of particulate matter samples: an inter-laboratory comparison, Nucl. Instr. Methods B, 266 (2008) 2401–2404

31 J.C.G. Jeynes, C. Jeynes, K.J. Kirkby, M. Rümmeli, S.R.P. Silva, RBS/EBS/PIXE measurement of single-walled carbon nanotube modification by nitric acid purification treatment, Nucl. Instr. Methods B, 266 (2008) 1569–1573

32 C. Jeynes, N. P. Barradas, H. Rafla-Yuan, B. P. Hichwa, R. Close, Accurate depth profiling of complex optical coatings, Surf. Interface Anal. 30 (2000) 237–242

33 J.W.Butler, Criteria for validity of Rutherford scatter analysis, Nucl. Instr. Methods B, 45 (1990) 160–5

34 C Jeynes, N P Barradas, P K Marriott, G Boudreault, M Jenkin, E Wendler and R P Webb, Elemental thin film depth profiles by ion beam analysis using simulated annealing—a new tool, J. Phys. D: Appl. Phys. 36 (2003) R97–R126

35 N.P.Barradas, C.Jeynes and R.P.Webb, Simulated annealing analysis of Rutherford backscattering data, Appl. Phys. Lett. 71 (1997) 291–3

36 M.A. Reis, L.C. Alves, A.P. Jesus, Matrix effects correction for quantitative TTPIXE analysis, Nucl. Instr. Methods B, 109-110 (1996) 134-138

37 C. Pascual-Izarra, M.A. Reis, N.P. Barradas, Simultaneous PIXE and RBS data analysis using Bayesian inference with the DataFurnace code, Nucl. Instr. Methods B, 249(1-2) (2006) 780-783

38 C. Pascual-Izarra, N. P. Barradas, M. A. Reis, C. Jeynes, M. Menu, B. Lavedrine, J. J. Ezrati, S. Röhrs, Towards truly simultaneous PIXE and RBS analysis of layered objects in cultural heritage, Nucl. Instr. Methods B, 261(2007)426-429

39 V. Corregidor, P. C. Chaves, M. A. Reis, C. Pascual-Izarra, E Alves and N. P Barradas, Combination of IBA techniques for composition analysis of GaIn AsSb films, Mat. Sci. Forum, 514-516 (2006) 1603-1607

40 Chris Jeynes, G.Zoppi, I.Forbes, M.J.Bailey, N.Peng, Characterisation of thin film chalcogenide PV materials using MeV ion beam analysis, IEEE Proceedings of Supergen Conference, Nanjing, April 2009, DOI:10.1109/SUPERGEN.2009.5348162

41 Reis MA, Barradas NP, Pascual-Lzarra C, Chaves PC, Ramos AR, Alves E, Gonzalez-Aguilar G, Costa MEV, Salvado IMM, Holistic RBS-PIXE data reanalysis of SBT thin film samples, Nucl. Instr. Methods B, 261 (2007) 439-442

42 L.Giuntini, A review of external microbeams for ion beam analyses, Anal.Bioanal.Chem. 401 (2011) 785-793

43 S L Molodtsov, A F Gurbich, C Jeynes, Accurate ion beam analysis in the presence of surface roughness, J. Phys. D: Appl. Phys. 41 (2008) 205303 (7pp)

44 F.Bloch, Double Electron Transitions in X-Ray Spectra, Phys. Rev. 48 (1935) 187-192

45 Y.Wakamatsu, H.Yamada, S.Ninomiya, B.N.Jones, T.Seki, T.Aoki, R.P.Webb, J.Matsuo, Biomolecular Emission by Swift Heavy Ion Bombardment, Ion Implantation Technology (AIP: CP1321, 2010, edited by J.Matsuo, M. Kase, T. Aoki, and T. Seki) 233-236

46 T. Jach, J.N.Ullom, W.T.Elam, The microcalorimeter X-ray detector: A true paradigm shift in X-ray spectroscopy, Eur. Phys. J. Special Topics 169, 237–242 (2009)

47 T.Satoh, M.Oikawa, T.Kamiya, Three-dimensional measurement of elemental distribution in minute samples by combination of in-air micro-PIXE and STIM, Nucl. Instr. Methods B, 267  (2009)  2125-2127

48 Janssens K, De Nolf W, Van Der Snickt G, Vincze L, Vekemans B, Terzano R, Brenker FE, Recent trends in quantitative aspects of microscopic X-ray fluorescence analysis. Trends Anal. Chem. 29(6) (2010) 464–478

49 Ursula Elisabeth, Adriane Fittschen & Gerald Falkenberg, Confocal MXRF in environmental applications, Anal. Bioanal. Chem. 400 (2011) 1743–1750

50 Brian M. Patterson, John Campbell and George J. Havrilla, Integrating 3D images using laboratory-based micro X-ray computed tomography and confocal X-ray fluorescence techniques, X-Ray Spectrom. 39 (2010) 184–190

51 A. Brunetti, B. Golosio, Software for X-ray fluorescence and scattering tomographic reconstruction, Computer Physics Communications 141 (2001) 412–425

52 http://cpc.cs.qub.ac.uk/summaries/ADPH

53 J.P. Hogan, R.A. Gonsalves, A.S. Krieger, Fluorescent computer tomography: a model for correction of X-ray absorption, IEEE Trans. Nucl. Sci. 38 (1991) 1721–1727

54 C.G.Ryan, PIXE and the nuclear microprobe: Tools for quantitative imaging of complex natural materials, Nucl. Instr. Methods B, 269 (2011) 2151-2162

55 Geert Silversmit, Bart Vekemans, Frank E. Brenker, Sylvia Schmitz, Manfred Burghammer, Christian Riekel, and Laszlo Vincze, X-ray Fluorescence Nanotomography on Cometary Matter from Comet 81P/Wild2 Returned by Stardust, Anal. Chem. 81 (2009) 6107–6112

56 N.K. Blute, D.J. Brabander, H.F. Hemond, S.R. Sutton, M.G. Newville, M.L. Rivers, Arsenic sequestration by ferric iron plaque on cattail roots, Environ Sci Technol, 38 (2004) 6074–6077

57 Martin D de Jonge, Stefan Vogt, Hard X-ray fluorescence tomography — an emerging tool for structural visualization, Current Opinion in Structural Biology, 20 (2010), 606–614

58 Sakellariou A, Cholewa M, Saint A, Legge GJF, An accurate reconstruction algorithm for tomography experiments that involve complex probe–sample interactions, Meas Sci Technol. 8 (1997) 746-758

59 D.N.Jamieson, New generation nuclear microprobe systems, Nucl Instrum Methods B, 181 (2001) 1-11

60 Ryan CG, Siddons DP, Moorhead G, Kirkham R, Dunn PA, Dragone A, De Geronimo G, Large detector array and real-time processing and elemental image projection of X-ray and proton microprobe fluorescence data, Nucl Instrum Methods B, 260 (2007) 1-7

61 M.J.Bailey, K.T.Howard, K.J.Kirkby, C.Jeynes, Characterisation of inhomogeneous inclusions in Darwin glass using IBA, Nucl. Instr. Methods B, 267 (2009) 2219–2224

62 T.Andrea, M.Rothermel, R.Werner, T.Butz, T.Reinert, Limited angle STIM and PIXE tomography of single cells, Nucl. Instr. Methods B, 268 (2010) 1884–1888

63 N.P.Barradas, C.Jeynes, M.Jenkin, P.K.Marriott, Bayesian error analysis of Rutherford backscattering spectra, Thin Solid Films, 343–344 (1999) 31–4

64 Chris Jeynes, Roger P. Webb, Annika Lohstroh, Ion Beam Analysis: A Century of Exploiting the Electronic and Nuclear Structure of the Atom for Materials Characterisation, Reviews of Accelerator Science and Technology, 4 (2011) 41–82

65 L.C. Alves, E. Alves, N.P. Barradas, R. Mateus, P. Carvalho, J.P. Coad, A.M. Widdowson, J. Likonen, S. Koivuranta, Erosion and re-deposition processes in JET tiles studied with ion beams, Nucl. Instrum. Methods B 268 (2010) 1991–1996

66 M.A. Reis, L.C. Alves, N.P. Barradas, P.C. Chaves, B. Nunes, A. Taborda, K.P. Surendran, A. Wu, P.M. Vilarinho, E. Alves, High Resolution and Differential PIXE combined with RBS, EBS and AFM analysis of magnesium titanate (MgTiO3) multilayer structures, Nucl. Instr. Methods B, 268 (2010) 1980–1985

67 M. A. Reis, N. P. Barradas, P. C. Chaves and A. Taborda, PIXE analysis of multilayer targets, X-Ray Spectrom., 40 (2011) 153–156

68 B.N.Jones, V.Palitsin, R.P.Webb, Surface analysis with high energy time-of-flight secondary ion mass spectrometry measured in parallel with PIXE and RBS, Nucl. Instr. Methods B, 268 (2010) 1714-1717

69 B.L.Doyle, P.P.Provencio, P.G.Kotula, A.J.Antolak, C.G.Ryan, J.L.Campbell, K.Barrett, PIXE-quantified AXSIA: Elemental mapping by multivariate spectral analysis, Nucl. Instr. Methods B, 249 (2006) 828-832




The database is protected by copyright ©ininet.org 2016
send message

    Main page