Nonlinearities in depth profiling nanometer layers

Abstract

An analysis is made of the sputter depth profiling of ultrathin silicon dioxide layers on silicon to evaluate the variation in the sputtering rate in the first few nanometers. Such changes in sputtering rate are important for the development of the analysis of nanoparticles. Cs+ ions are chosen as an example of a metal ion popular in secondary ion mass spectrometry (SIMS) studies that provide excellent depth resolution. It is found that, if it is assumed that the signal is linear with oxygen content, the sputtering rate falls rapidly by a factor of 4.8, with an exponential decay near 1.2 nm when using 600 eV Cs+ ions at 60° incidence angle. The interface may be described by the integral of the response function of Dowsett et al. developed for SIMS depth profiling of delta layers with λu=0.5 nm, λd=0.7 nm, and σ=0.4 nm, showing the excellent depth resolution. However, if published data for the nonlinearity of the signal with oxygen content are used, the rapid change is still seen but with an initial sputtering rate that is reduced from the above 4.8 to 3.5 times that at equilibrium.

Analytical and numerical depth resolution functions in sputter profiling

“…(8)) is extremely useful when the apparent location of interfaces in the measured profile of thin layers with increasing thickness have to be compared with the true layer thickness to find the change of the sputtering rate with depth. Such a case was reported by Seah et al [50] for SIMS depth profiles of SiO 2 layers with different thickness obtained by bombardment with 600 eV Cs + ions. These authors employed the UDS response function (Eq.…”

“…(3)) which could not yield the physically relevant interface positions (see Section 2.3.1), and they finally assumed the location of the interface at 50% of the plateau intensity (see p. 1219 of ref. [50]). In contrast, the MRI model gives the correct DRF position and discloses the true location of the interface with respect to the 50% interface position (and additionally with respect to the DRF or delta layer profile peak position).…”

“…Applying the original UDS response function, Wittmaack et al [47,48] tried to answer the question whether the delta layer location is at the centroid or at the maximum of the DRF, and opted in favor of the maximum. In contrast, Seah et al [50] opted for the centroid of the DRF. With the original UDS function these authors referred to, neither of them is right (see e.g.…”

“…Because of the similarity of the upslope parameter u (UDS) with the partial resolution function of the information depth parameter in the MRI model, the latter has been frequently taken as one or two atomic monolayers when applied to SIMS depth profiles [22,33], or has been ignored because of the much higher values of w and [34][35][36]. Indeed, most SIMS measurements are fitted with UDS and small u values between zero and two monolayers [16,45,[47][48][49][50][51][52]. In contrast to the MRI model, the exponential upslope function in the UDS response function is directly folded with a Gaussian function and works like a correction of the latter at lower intensities.…”

“…The physical meaning of the empirical upslope parameter u is still not clear [47][48][49][50][51][52] and will be discussed below in Section 2.3.2. …”