Papers

AVS1997 Session AS-TuM: Depth Profiling and Analysis

Tuesday, October 21, 1997 8:20 AM in Room J2

Tuesday Morning

Start	Invited?	Item
8:20 AM		AS-TuM-1 The Dependence of the Depth Resolution on the Structure of the Specimen. M. Menyhard, A. Sulyok (Research Institute for Technical Physics, Hungary) The sputter depth profiling techniques are an important segment of the thin film analytical methods, and due to recent developments they can be applied for analysis of multilayer structures (thickness down to 1 nm), deeply buried thin layer (thickness around 2-3 monolayers ) etc. The term depth resolution has been introduced, based on the concept of Gaussian broadening) for the quantification of the method. In fact it can be used for characterization of the depth profiling device mainly because of the not well understood physics of the depth profiling. In this contribution we will report on the result of an extensive depth profiling study. Various thin film structures (interface between thick layers, imbedded, 1-2 nm thick, layers, multilayers with thickness of 2 and 3 nm) were produced of amorphous Si and Ge. The quality of the specimens were checked by X ray diffraction and XTEM. Depth profiling was carried out by using various (Xe, Ar, Ne and He) ions with energy in the range 0.2-2 keV. It turned out that the depth resolutions determined according to the standard were similar for the various projectiles at various energies. On the other hand the depth profiles recorded at multilayers were drastically different; e.g. using He ions the multilayer structure 2 nm can be hardly resolved, while it is well visible by using Ar. We will show that the surface roughening can not be blamed for this difference but the depth dependence of the ion mixing. We will also show that the analytical model can not describe these results but the dynamic TRIM code provides a reasonable results. Based on these results recommendations on the usage of the depth resolution will be also given.
8:40 AM		AS-TuM-2 Determination and Application of the Depth Resolution Function in Sputter Profiling with SIMS and AES S. Hofmann, J. Schubert (National Research Institute for Metals, Japan) The depth resolution function describes the relation between the measured depth profile and the true in depth distribution of composition. In high resolution depth profiling, blurring of the profile often is markedly asymmetric and needs several parameters for its precise description. The depth resolution function can be experimentally determined with appropriate reference materials, e.g. samples containing delta layers or sharp interfaces between two media, and it can be theoretically predicted. Recently, a semi-empirical model was developed based on the most fundamental physical mechanisms of atomic mixing, surface roughness and information depth (MRI-model) ¹. These three parameters precisely describe the resolution function which was tested using results obtained on several structures and with different methods, e.g. AES profiles of GaAs/AlAs multilayer interfaces and delta layers, and on SIMS profiles of B in Si and of SiO₂/Ta₂O₅ multilayers. It is shown how the depth resolution function is applied to reconstruct in depth distributions of composition of the measured profiles, and how in AES depth profiling either the depth scale can be determined knowing the lectron escape depths or vice versa. Furthermore, information about thin layer thickness, roughness development and graded composition at interfaces are readily obtained. Preferential sputtering can be reasonably well approximated with the MRI - model. More severe limitations are encountered by radiation enhanced segregation, compound formation and other non-linear effects. ^ext S. Hofmann, Surf. Interface Anal. 21 (1994) 673
9:00 AM		AS-TuM-3 SIMS Depth Profiling of Ultra-Shallow Boron Implants in Silicon: Use of Low Energy Oxygen Primary Ion Bombardment and Oxygen Flooding of the Sample Surface M.S. Denker, C.W. Magee (Evans East) As the junction depths for semiconductor devices has decreased, the surface analysis techniques used to characterize them have evolved to meet this challenge. The area of the depth profile near the surface which has been previously ignored or deemed unimportant is now the site of considerable attention. Much of the dose of the dopant species is contained in this near- surface region, and it is therefore crucial to accurately profile this region. One area of ultra-shallow junction technology which is currently receiving much attention is the implantation of boron into silicon. Implant energies as low as 125 eV are being used. This poses quite a challenge to ion sputtering techniques such as SIMS (Secondary Ion Mass Spectrometry), in which primary ion beams of energy in the keV range have been typically used to sputter depth profile implanted species. This talk will present SIMS data of these ultra-shallow boron implants, using oxygen primary ion energies of 2keV down to 0.8keV. In addition, an oxygen leak flooding the analysis area is used to produce a depth profile in which the sensitivity of the boron is nearly constant throughout the analytical depth.¹ Without the oxygen flood, the very different sensitivities of the boron in the native oxide and in the Si would cause significant quantification errors of the boron dose residing in the native oxide. Finally, a comparison of this technique for ultra-shallow profiling with that of normal incidence bombardment with no oxygen flooding will also be discussed. ¹1. C.W. Magee, J.R. Shallenberger, M.S. Denker, D.F. Downey, and L. Meloni, Proc. Fourth Internat. Workshop on the Measurement, Characterization, and Modeling of Ultra-Shallow Doping Profiles in Semiconductors, RTP, NC (1997)
9:20 AM		AS-TuM-4 Ultra Shallow Depth Profiling by TOF-SIMS K. Iltgen, O. Brox (University of Muenster, Germany); E. Niehuis (Ion-Tof GmbH, Germany); A. Benninghoven (University of Muenster, Germany) Quantitative SIMS analysis of ultra shallow doping profiles requires high depth resolution and has to consider changes in the erosion rate and in the ionization probability during establishing a sputter equilibrium. As we have shown in previous works optimized depth resolution in the 0.5 nm range is provided by SF₅⁺-sputtering. We have now studied the influence of the sputter parameters on the changes in erosion rate and ionization probability in the transient region. We will report in particular on results for low energy B-implants in Si. All experiments were carried out in a Dual Beam TOF-SIMS providing high mass resolution at full transmission and a parallel detection of all mass lines. The Dual Beam Mode allows a free choice of the sputter parameters regarding ion species (Cs⁺, noblge gas ions, O₂⁺, SF₅⁺, ...), ion energy and angle of incidence. At any stage of the profiling process - in particular in the transient region - the surface oxidation state can be described by the lattice valency G⁺ of the Si surface atoms. This lattice valency can be determined from the intensity ratios of the positive secondary ions Si₂O_n⁺ (n = 0 .... 4). We have determined the erosion rate and the ionization probability for B as a function of the Si surface oxidation state. Furthermore, we have studied the influence of the sputter parameters (ion species, ion energy and angle of incidence) on the width of the transient region and investigated the influence of oxygen flooding. By ¹⁸O₂-exposure the interface of the native oxide to the silicon substrate has been detected.
9:40 AM		AS-TuM-5 SIMS Measurements of Ultra-Shallow Junctions Comparing Normal and Oblique Angle Primary Ion Beam Bombardment G.R. Mount, S.P. Smith, C.J. Hitzman (Charles Evans & Associates) As devices approach deep submicron dimensions, gate lengths become shorter and short channel effects become important considerations. Short channel effects can be minimized by reducing the source/drain implant energy so that lateral straggle does not further reduce the effective channel length. Accurate and reproducible measurement of these implants before and after annealing is important for equipment qualification and for the calibration of process modeling software. Secondary Ion Mass Spectrometry (SIMS) is well established as an analytical tool for the measurement of as implanted and annealed dopant profiles. Using typical primary ion beams and energies results in an equilibration zone of up to 20 nm where accurate quantification is difficult. Much of the ultra-shallow implant dose and profile shape is within 20 nm of the surface. Lowering the primary beam energy to 1keV or less minimizes but does not eliminate the equilibration zone. Secondary ion yields can be made constant to within a few tenths of a nanometer of the surface by using either oxygen leak or a normal incidence primary ion beam. Both methods result in the formation of an oxide at the surface minimizing any surface ion yield transients. Recently there has been some discussion about the relative merrits of each method ¹ and this paper directly compares both methods using the same samples and the same instrument. Ultra-shallow boron implants are measured using oblique angle oxygen leak and normal incidence. Results are compared for ion yield, sputter rate, stability, depth resolution, surface transients and dosimetry. The relative merrits of each technique will be discussed. ¹The Third International Symposium on Ultra-Shallow Junctions, 1997, in press
10:00 AM		AS-TuM-6 Progress and Challenges in Depth Profiling with Angle Resolved XPS B.J. Tyler (Montana State University) Calculating depth profiles from angle resolved x-ray photoelectron spectroscopy (ARXPS) data is of interest in a wide range of fields. Unfortunately, inverting ARXPS data to obtain depth profiles is a mathematically unstable problem. As a result, the answer obtained will be strongly influenced by the algorithm used in the calculation and by the constraints and assumptions used in formulating the algorithm. When comparing algorithms, there are several key issues. These include stability of the algorithm to random and systematic errors, resolving power of the algorithm, and the influence constraints and assumptions on the performance of the algorithm. Common assumptions that are used when formulating algorithms include that the sample is perfectly smooth, that the attenuation length is uniform throughout the sampling depth, that the detector has 0° solid angle of acceptance, and that the atomic density of the sample is uniform through the sampled depth. All of these assumptions are false for most real samples. In this presentation, several algorithms that have been presented in the literature will be discussed and compared. Performance of the algorithms on both synthetic data sets and real world samples will be presented.
10:40 AM		AS-TuM-8 X-ray Photoelectron Spectroscopy, Depth Profiling, and Elemental Imaging of the Effects of Humidity and Temperature on the Stability of a Metal/Polyimide Interface. D.R. Jung, T. Long (Science Applications International Corp.); B. Ibidunni (Sheldahl Corp.) The stability of a metal/polyimide interface exposed to high humidity and temperature has been studied. The multilayer metallization fabricated for IC packaging consist of an adhesive chromium layer, evaporated copper, electroplated copper, nickel and gold on polyimide. The metallization was encapsulated with a solder mask, and solder balls were attached to the gold surface. After exposure to 100% rh and 125 C for 96 hours, the metallization delaminated from the polyimide. XPS analysis with conventional and small area spectra, Ar ion sputter depth profiling, and imaging of specific elements is presented for these failure surfaces. These results unambiguously show that failure occurred on the polymer side of the metal/polyimide interface. Massive copper migration into the polyimide was observed along with Pb. Cu migrated deeper than Pb. Zn was detected at the delaminated interface also. The roles of these migrating species in the cohesive failure of the polyimide will be discussed.
11:00 AM		AS-TuM-9 Tungsten Silicide Composition Analysis by RBS, AES, and XPS C.A. Bradbury, D.K. Fillmore (Micron Technology, Inc.) While Rutherford backscattering spectroscopy (RBS) is the accepted method for determining the composition of tungsten silicide films, the system is expensive and therefore not widely available. Further, the depth resolution of RBS is of the order of 150 Å, which does not allow for the compositional analysis of thinner films. Chemical information is also not possible. These are critical issues as the geometries of ULSI devices become smaller. With special care applied to data acquisition and reduction, both Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) provide excellent compositional agreement with RBS. Additionally, AES and XPS provide chemical information about thin layers, which is unavailable from the RBS analysis. Tungsten silicide films deposited on silicon with both chemical and vapor deposition processes were analyzed after anneal. RBS, AES, and XPS were used to analyze the film composition. Silicon-to-tungsten ratios were comparable among the techniques. In addition, surface oxide thickness and interfacial layer compositon were determined using AES and XPS.
11:20 AM		AS-TuM-10 The Use of the Navy Oxygen Cleaner for Bearing Steels N.H. Turner, R.L. Mowery (Naval Research Laboratory) The Navy Oxygen Cleaner (NOC), a basic inorganic solution, has been evaluated in the cleaning of several types of steels used for precision bearings, i.e., 440C, 52100, and M50. Steel coupons with simulated contamination by either a light machine oil or a grease and control specimens were ultrasonically cleaned with NOC at 60 to 65°C for periods of less than 90 minutes. The composition in the surface region was determined with the use of regular and angular resolved XPS, depth profiling, and FTIR. The results indicated that the native oxide layer thickness and the residual carbon on the contaminated coupons were reduced compared to as received coupons cleaned with 2-propanol. The analyses also found on the surface a small amount of silicon as a silicate type species after the NOC cleaning. The thickness of the silicate layer is less than 3 nm, and this species may be preventing any observable corrosion and pitting of the steels during cleaning. Specimens stored in a dry atmosphere also showed little if any differences after several months compared to freshly cleaned samples. Details about the surface chemical species and their depth distribution in the surface region will be discussed.
11:40 AM		AS-TuM-11 An Electron Spectroscopic Study of the Effects of Annealing on the Pd/SiC Schottky Structure L. Chen, G.W. Hunter, P.G. Neudeck (NASA Lewis Research Center); D. Knight (Cortez III-NASA Lewis Research Center) Palladium-based Schottky diode gas sensors using silicon carbide (SiC) as the semiconductor have demonstrated very high sensitivity to hydrogen and hydrocarbon gases at high temperatures. Surface and interface stability are critical for the development of stable Pd and SiC based gas Schottky sensors operable for long-term at high temperatures. The simplest Pd-based SiC Schottky diode system is Pd directly deposited on SiC (Pd/SiC). Previous studies of an annealed Pd/SiC interface by AES indicated that annealing at 425 °C significantly promoted interfacial diffusion and chemical reactions with the major interfacial product being palladium silicides, Pd_xSi. From these studies, the chemical state(s) of the interfacial products are qualitatively known and correlated to the relative concentrations of Pd and Si at the interface region. However, quantitative measurements of the distribution of the interfacial products throughout this interface were not achievable in the AES studies. In this work, we investigate the interface of an annealed Pd/SiC sample composed of 400 Å of Pd magnetron-sputter deposited onto 6H-SiC substrate by profiling the interface by ESCA and noble gas sputtering techniques. The chemical states of the interface products at various depths will be quantitatively studied and correlated to the relative concentrations of Pd and Si as a function of the depth. The configuration of the interface of the annealed sample will be compared with that of an as-deposited sample. These results will be compared with our previous AES data and related to the sensors performance.