Artifacts in AES microanalysis for semiconductor applications

Abstract

Auger electron spectroscopy analyses of submicron features on semiconductor surfaces are routinely accompanied by analytical artifacts such as sample degradation and background contributions arising from electron beam scattering. Submicron analyses are commonly carried out at electron beam densities in excess of 1 A cm −2 and are especially damaging to silicon oxides. The evolution of oxide reduction is observed both as a loss of oxygen versus beam exposure and in a complementary growth of Si LVV and Si KLL elemental peaks. The O KLL signal intensity from a 1 µm 2 area of thermally grown oxide is found to decrease by 22% after exposure to a rastered 20 kV/10 nA beam for 10 min. Another aspect of submicron analysis is the contribution to survey spectra that originates when surrounding material is excited by backscattered electrons. Background contributions may dominate AES spectra even when the sample is flat and the probing beam is smaller than the feature of interest. Tungsten damascene contacts provide a useful platform for investigating this phenomenon in the absence of topography. Spectra have been collected from tungsten contacts of various sizes and the target and background contributions quantified. When a 0.25 µm diameter tungsten contact is probed with a narrow 20 kV beam, the W MNN signal intensity is determined to be only 70% of that emitted from a large tungsten structure. Target signal reduction coincides with increased signal contributions from the surrounding oxide.

Physical characterization of thin‐film solar cells

“…Several excellent reviews of the two experimental methods are available. [75][76][77][78][79][80] Although the physical basis for surface sensitivity, a short (1-5 nm) electron inelastic mean free path, is the same for both methods, there are several important differences. AES typically uses electron beam excitation, and thus the study of insulators, although not impossible, is difficult.…”

“…Examples include the graphitization of organic substances such as those used in dye-sensitized/TiO 2 solar cells 77 and the electron-beam-induced reduction of metal oxides such as TiO 2 and ZnO. 77,79,83 Electron-beam damage has been observed in electron-beam-induced current (EBIC) experiments on CdTe. 84 Thin stainless steel or polyimide substrates popular with PV manufacturers using 'roll-to-roll' processes can exacerbate electron-beam damage effects (A. Swartzlander, unpublished work).…”

Growth and trends in Auger-electron spectroscopy and x-ray photoelectron spectroscopy for surface analysis

“…60 Tougaard 61 showed from model calculations that the same Cu 2p 3/2 XPS intensity could be observed for a submonolayer of Cu on Au, for a Cu-Au alloy extending to a depth of 5 nm on Au, for a Cu layer at a depth between 2 and 3 nm in an Au matrix, and for a 2.5 nm Au overlayer on a Cu substrate. 66,67 For AES, signals can also be generated by backscattered electrons, and the analysis area is generally much greater than naively expected from the full width at half maximum intensity of the incident beam. 61 Tougaard has developed software and procedures with which portions of a measured AES or XPS spectrum can be analyzed to give information on the sample morphology as well as the surface composition.…”

“…If the hypothetical Cu 2p 3/2 peak intensity had been analyzed to yield a Cu composition, the uncertainty in the Cu quantification could be up to several hundred percent. [67][68][69] Useful information on the presence and distribution of multiple surface phases on the sample surface can be obtained from methods of multivariate image analysis. 7 The spectrum of inelastically scattered electrons extending up to about 50 eV below the peak of interest is analyzed with information on the differential inelastic-scattering cross section as a function of energy loss.…”