Article pubs.acs.org/ac
Reduction of the SIMS Matrix Effect Using the Storing Matter Technique: A Case Study on Ti in Different Matrices B. Kasel and T. Wirtz* Centre de Recherche Public - Gabriel Lipppmann, 41, rue du Brill, L-4422 Belvaux, Luxembourg S Supporting Information *
ABSTRACT: Matrix effects in secondary ion mass spectrometry render quantitative analysis difficult. In this paper, we report on the quantitative potential of the so-called storing matter technique. On the basis of a case study focusing on Ti sputtered from five different chemical environments (Ti, TiB2,TiC, TiN and TiO 2), we demonstrate for the first time that the SIMS matrix effect can be avoided in a multitude of matrices by this novel approach. The effect of the collector material (Au, Ag, Cu, and Ta) on the overall efficiency of the storing matter process is investigated. In addition, the influence of oxygen on the obtained useful yields is exposed. Finally, the performance of the storing matter technique is compared to standard SIMS analysis. To limit the sputtered flux during the deposition and determine the size of the deposit, a circular aperture is placed in front of the collector. A constant rotation of the collector generates a ring shaped deposit. The adjustment of the rotation speed allows the tuning of the density of the deposit. More importantly, the rotation of the collector enables us to perform depth profiling by transforming the depth information in the sample into lateral information on the collector. This technique has been applied to a high-dose boron implant in silicon,16 where the high boron concentration leads to matrix effects in SIMS, which are successfully eliminated by means of the storing matter technique. Becker et al.17 have applied the technique to polymer samples and have identified characteristic peaks of PMMA, PVC, and PS in storing matter mass spectra, that is, mass spectra recorded on a storing matter deposit of the polymers. The dilute deposit of sputtered polymers on a metal collector such as silver promotes cationization and exhibits fragmentation patterns that are different from standard SIMS analysis. In this study, we investigate the quantitative capabilities of the storing matter technique by applying it to a set of samples containing Ti (Ti, TiB2, TiC, TiN, and TiO2). The storing matter analysis is performed using different collector materials (Ag, Au, Ta, and Cu). To study the effect of oxygen on the storing matter analysis, the analysis step is performed with and without oxygen flooding for Ti. We compare the useful yields obtained for the storing matter analysis UYStomat with the useful yields for conventional SIMS measurements UYSIMS for the Ti+ ion.
S
econdary ion mass spectrometry is the most prominent analysis technique used to determine dopant profiles in semiconductors because of its excellent detection limit and depth resolution.1−4 Variations of the ionization probability of the detected species with the chemical environment in the sample, which are called matrix effects, constitute a major hurdle for quantitative analysis.5−7 A typical example of this phenomenon is the enhancement of a positive signal by oxygen present in the matrix, a fact that is exploited by flooding a sample with oxygen to increase the secondary ion signal.8−10 A concept generally used to describe the sensitivity is the useful yield of SIMS, UYSIMS. It is defined as the ratio of the number of detected ions to the number of atoms sputtered from the sample. The useful yield is proportional to the ionization probability and is a clear indicator for matrix effects. The storing matter technique was developed to circumvent the matrix effect by decoupling the sputtering of the sample and the analysis of ions.11−13 The sample is first bombarded with a focused ion beam, and the sputtered particles are collected on a well-defined collector surface. The conditions are chosen in such a way that the resulting deposit is sufficiently dilute not to give rise to matrix effects. Next, the deposit on the collector is analyzed using conventional SIMS analysis. Because the deposit is in the submonolayer range, the chemical environment is defined by the collector. Now the SIMS useful yield is independent of the initial sample composition, thus removing the matrix effect. The quantitative capabilities of the technique have been demonstrated in the case of Si− in pure Si and SiO2.14 In the same study, the authors demonstrate that the collector material can be optimized in terms of work function to achieve optimal ionization of the deposited matter during SIMS analysis in accordance with the electron tunnelling model.15 © 2014 American Chemical Society
Received: September 24, 2013 Accepted: March 12, 2014 Published: March 12, 2014 3750
dx.doi.org/10.1021/ac4030472 | Anal. Chem. 2014, 86, 3750−3755
Analytical Chemistry
■
Article
described in previous work.18 The SIMS conditions used are summarized in Table 1. Because the deposit was very dilute,
EXPERIMENTAL DETAILS First of all, 1 in. Si wafers were cleaned in deionized water, acetone, and ethanol prior to introduction into the vacuum system (in the lower range of 10−8 mbar). These wafers were the substrates for the collectors used in the following procedure. After the collectors were introduced into the instrument, they stayed under vacuum until after the final SIMS analysis was finished. Figure 1 depicts the three major
Table 1. SIMS and Storing Matter Analysis Conditions
care had to be taken not to sputter it away too quickly. A compromise between signal intensity and number of data points had to be taken. Here, we used a rather gentle bombardment of 1 keV Cs+ with a current of 2−3 nA rastered over 500 μm × 500 μm. The list of analyzed ions was limited to Ti+, Cs+, and CsM+, where M represents the collector element. The matrix signal was recorded to allow the correction of instrument transmission variations by normalizing with respect to the equilibrium matrix signal. In a second series of experiments, the effect of oxygen on the storing matter analysis was studied by sputtering pure Ti on the 4 kinds of collectors. The subsequent analysis step was then performed first under the same conditions as before and a second time with oxygen flooding at a pressure of 2 × 10−5 mbar. As a reference, SIMS analysis was also performed on the initial titanium samples to explore any existing matrix effects. In this case, the SIMS analysis used an O2+ or Cs+ beam at 8 or 4 keV impact energy, respectively. The conditions were chosen to produce deep craters, which are easily measurable by surface profilometry to determine the number of sputtered Ti atoms and, hence, the UYSIMS. In contrast to the storing matter analysis, we also monitored the CsTi+ cluster, which is known to suffer less from matrix effects, in the case of Cs + bombardment.19−22 To calculate UYSIMS(x) for element x, we considered the following formula, which is valid for a homogeneous sample:
Figure 1. Diagram showing the three major steps of the storing matter process.
steps of the storing matter process. In a dedicated chamber of the storing matter instrument, metal pellets (Ag, Au, Ta, and Cu) were then evaporated onto the wafers using an electron beam evaporation system (Figure 1a). The nominal thickness of the coating was chosen to be around 50 nm as measured with a quartz microbalance to ensure a homogeneous film. The collector was then introduced in the sputter-deposition chamber and positioned 3 mm above and facing the samples to be sputtered. An aperture with a diameter of 500 μm was placed as close as possible to the collector without touching it. This way, the sputtered flux was limited to a 500 μm broad area on the collector. The rotation of the collector around its central axis resulted in a 500 μm broad ring-shaped deposit near the edge of the collector. This way, the area covered by each deposit was large enough to allow several SIMS analyses later on. The different titanium-based samples were commercially available sputter targets with a purity of at least 99.5% (99.99% in the case of Ti). As indicated in Figure 1b, the titanium samples were then successively sputtered with a raster-scanned 10 keV Ar+ beam with an intensity of 203 nA at 45° to the sample normal. This resulted in arc-shaped deposits close to the edges of the collector. To ensure identical sputter-deposition conditions for each sample, the sample holder was moved laterally to access the different samples, whereas the collector stage and Ar+ beam stayed at the same position. Before each deposit, the samples were presputtered during 1.5 h to ensure that any surface oxides or contaminations were removed. After the sputter-deposition, the collectors were transferred to a Cameca SC Ultra for analysis with a vacuum suitcase
UYSIMS(x) =
I (x ) c(x) ·A ·vsput,SIMS
where I(x) is the mean signal intensity of species x in the equilibrium regime, c(x) is the atomic concentration of species x, A is the analyzed area, and vsput,SIMS is the sputtering rate in SIMS, which was determined by means of stylus profilometry. The density of the materials used was determined experimentally by weighing and measuring the sample volumes. This was necessary because of the granular structure of the sputtering targets, which caused the measured densities to deviate severely from tabulated values. The disadvantage of this was that an additional error was introduced into the results. In storing matter, we also defined the useful yield as the number of detected ions divided by the number of sputtered atoms from the initial sample,
UYStomat =
Ndet Nsput
However, there are a few subtleties that need to be clarified. For storing matter, the initial lateral position on the sample of a sputter-deposited atom is less well-defined than in SIMS because of the absence of elaborate optical or electronic gating, 3751
dx.doi.org/10.1021/ac4030472 | Anal. Chem. 2014, 86, 3750−3755
Analytical Chemistry
Article
as in SIMS. Because of the angular distribution of the sputtered matter, which comprises ions as well as neutrals, and despite the collection aperture placed between the sample and the collector, an atom sputter-deposited on the collector may originate from any position on the bombarded area, although with a distribution peaked around the center of the crater. To take this into account, we considered the entire volume sputtered during deposition. This way, it actually sufficed to determine the partial sputter yield for Ti under argon bombardment Y, once for each sample to derive the number of sputtered Ti atoms. We use the Ar+ current i, the total sputter-deposition time tsput, and the sputter yield of Ti to calculate the total number of sputtered Ti atoms Y·(i/e)·tsput (e is the elementary charge). Furthermore, the rotation of the collector during sputter-deposition with a constant speed homogeneously distributed the sputtered matter over an arc with a total angle of θtot and radius of R. The width of the deposit was defined by the size of the aperture diameter, w. Finally, the subsequent SIMS analysis was performed on the deposit with an analyzed area, A. This means that the number of initially sputtered Ti atoms on area A corresponds to
Figure 2. SIMS useful yield for titanium.
main source of error is the depth measurements, which have a relative standard deviation between 0.7% for TiO2 and 12.8% for Ti. The standard deviation of the depth of the crater was determined on the basis of six profilometry measurements. The relative standard deviation of the signal intensity was determined to be
