SIMS2015 Session MI1-MoM: Microelectronics

Isotopically enriched films of silicon have recently become of interest for emerging technologies in microelectronics that rely on spin-dependent properties of electrons, i.e., spintronics and quantum computing. The reason is that the elimination of the odd isotope ²⁹Si that carries a nuclear spin can greatly reduce the interaction of the substrate with the spin state of the carriers and thereby increase their coherence time. We have produced isotopically purified films of ²⁸Si on silicon substrates with a custom-made hyperthermal energy ion beam system that features a Penning gas ion source and magnetic mass separation. Silicon ions were produced from silane with a natural Si isotopic composition. Secondary Ion Mass Spectrometry (SIMS) was employed to determine the isotopic purity of the deposited films. A large-geometry magnetic sector SIMS instrument was used for these measurements. A resolving power of 6000 was required to cleanly separate the extremely low ²⁹Si abundance in the isotopically purified films from the much larger ²⁸SiH signal. In one silicon film the average measured ratios over the depth of this film were 1.3x10^-6 for ²⁹Si/²⁸Si and 0. 3x10^-6 for ³⁰Si/²⁸Si. Thus the isotopic purity of ²⁸Si is 99.9998 % compared with its natural abundance of 92.23 %, an enrichment level that is more than adequate for the proposed applications and believed to be the isotopically purest silicon ever reported. Remaining challenges are to ensure epitaxial deposition of the film, reduce chemical impurities and increase the deposition area to a size on which test structures can be produced.

The reduction of dimensions of advanced III-V devices with 3-D architecture requires the development of advanced characterization methods to provide information on the chemical composition. For this, the depth and lateral resolution become critical for accurate determination of heteroepitaxial interfaces.

Secondary ion mass spectrometry (SIMS) has excellent depth resolution and sensitivity and is widely used in microelectronics for characterizing the abruptness of interfaces and the chemical composition of layers. III-V heterostructures of 200 nm in width selectively grown on non-planar Si wafers by MOCVD were studied. It was recently demonstrated that SIMS protocol using an averaged analysis for such structures is promising [1,2]. The developed SIMS protocol allows the abruptness of interfaces in patterned structures to be studied and to be correlated with photoluminescence measurements (PL).

In this work we discuss how to improve this type of profile as good depth resolution is critical for analyzing InGaAs quantum wells a few nanometers in thickness. Low energy ion beam sputtering is used to obtain a high depth resolution, but reducing the primary energy too much leads to an increase of surface roughness. For example InGaAs is known to form ripples under oxygen bombardment. The best depth resolution is found using a parallel orientation of the ion beam to the III-V trenches but roughness is still formed decreasing the depth resolution and causing variations in ion yields and sputter rate [3]. Here we use sample rotation in a ToF-SIMS to reduce formation of surface roughness and we compare the results obtained on planar and patterned samples correlated with AFM measurements of sputtered craters. The effect of incidence angle is also investigated using pretilted samples in a magnetic SIMS and compared with AFM measurements. The developed SIMS protocols can be implemented for the characterization of future III-V devices monolithically integrated on Si.

Acknowledgements: This work was supported by the French "Recherches Technologiques de Base" Program and was performed on the Nano Characterization Platform (PFNC) of the CEA Grenoble.

[1] V. Gorbenko, M. Veillerot, A. Grenier, G. Audoit, W. Hourani, E. Martinez, R. Cipro, M. Martin, S. David, X. Bao, F. Bassani, T. Baron, J.P. Barnes, physica status solidi (RRL) - Rapid Research Letters, 9 (2015) 202-205.

[2] W. Vandervorst, A. Schulze, A.K. Kambham, J. Mody, M. Gilbert, P. Eyben, physica status solidi ©, 11 (2014) 121-129.

[3] P.F.A. Alkemade, Z.X. Jiang, Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures, 19 (2001) 1699.

Seawater desalination is becoming a promising alternative to meet increasing water demand by municipal and industrial users, inevitably leading to increased boron concentration in processed water, typically 0.3-1.5 mg/L, by the reverse osmosis (RO) membrane process [1]. The recommended quality is 50 ng/L or lower for high purity water (HPW) for Si semiconductor processes [2]. The presence of excess boron in HPW might cause poor P-N junction and lower the production yield of silicon devices (SD) [3]. The effect of boron concentration in HPW on electrical properties of SD has not been systematically studied yet.

Herein, we analyzed boron contaminant in SD using secondary ion mass spectrometry (SIMS) and inductively coupled plasma-mass spectrometry (ICP-MS) as well as electrical properties of relevant SD using electrical parameter analyzers. The aim was to identify the spatial distribution and concentration of boron in SD as well as to investigate boron effect on SD electrical properties. D ifferent boron concentration solutions (50-500 ng/L) were used to simulate HPW for cleaning Si wafer at wet-cleaning step. The cleaned wafers were immediately subjected to silicon dioxide (100 nm thickness) deposition by plasma-enhanced chemical vapor deposition. High voltage capacitor (HVC) devices as model SD were fabricated by lithography and etching in the remaining half wafers. Finally, the electrical properties of HVC were measured by following the standard protocols.

The localization of boron and some metal species that may affect SD electrical property was found at the interface between silicon dioxide layer and Si substrate in HVC by SIMS depth profile. Boron concentration in HVC and HPW determined by ICP-MS reveals that the amount of boron in HVC is proportional to the boron concentration in HPW. At 125 and 500 ng/L boron concentrations, the breakdown voltage, threshold voltage, and flat band voltage were all significantly decreased. In contrast, the electrical properties exhibit slightly different at boron concentration lesser than 125 ng/L, which is 2.5 times of the recommended value. The combined results of SIMS, ICP-MS, and electrical property measurements successfully delineate the pronounced effect of boron threshold concentration on SD under our fabricating conditions.

[1] Redondo, J. A. Desalination 2001, 138, 29.

[2] SEMI F63-0213 - Guide for Ultrapure Water Used in Semiconductor Processing.

[3] Yagi, Y.; Hayashi, F.; Uchitomi, Y. EVALUATION OF BORON BEHAVIOR IN ULTRAPURE WATER MANUFACTURING SYSTEM; Semiconductor Pure Water & Chemicals Conference: Sunnyvale, 1994.

Semiconductor structures are becoming smaller in every dimension every year. Analyses of this type of devices must be performed using techniques capable both of nm lateral resolution, sub-nm depth resolution and ppm sensitivity, which is hardly achievable. Newly emerged technique TAP (Tomographic Atom Probe) is, probably, the most promising but still has to overcome certain existing limitations. At the same time, traditional dynamic SIMS could be employed if there is a way separate secondary ion signal, coming from a few-nm large device or the mono-atomic layer structure from the signals, coming from the rest of the matrix material. This could be possible if there were mass peaks in the mass-spectra which could be identified as originating only from the 3D device area or layer of interest. Method based on such an approach became known under the name of Self Focusing SIMS. This approach, very promising for dynamic SIMS, still needs to be further developed via building database of the ‘fingerprints’ for each device structure and used analytical conditions of SIMS experiment.There is also a demand for high depth resolution SIMS analysis of very thin mono-atomic (or few-atomic) layers such as Graphene. It is expected that the very low impact energy sputtering will allow approaching the required depth resolution limit. However, up to now mono-atomic layer position cannot be accurately determined. The Self Focusing SIMS methodology can be employed in such case looking for the part of the mass-spectra emitted exclusively from layer of interest.

ToF-SIMS is routinely used for depth profiling of microelectronic materials but interpretation of results can sometimes be hampered by matrix effects. Plasma Profiling Time of Flight Mass Spectrometry (PP-TOFMS) provides direct measurement of the elemental composition of materials as a function of depth, with nanometre resolution and the capability to measure both thin and thick layers [1]. It consists in a pulsed radio frequency glow discharge plasma source fed with pure Ar and created under a pulsed RF potential coupled to a time of flight mass spectrometer (TOFMS). The ultra-fast detection and quasi-simultaneous acquisition of all mass ions of the TOFMS fits well with the fast erosion rate of the high density and low energy plasma source. Furthermore the separation between sputtering and ionisation processes makes this technique much less matrix dependent compared to SIMS. In addition, the orthogonal TOFMS configuration allows for temporal monitoring of the transient signals generated in the pulsed plasma. This is all the more important as signals are largely enhanced in the plasma extinction phase (once RF is turned off) in the so-called afterglow region. Ion signals are then generated through Penning Ionisation by Ar metastables.

Examples in microelectronics and nanotechnology will be presented such as magnetic layers for 3D sensors, Pt-doped Ni-silicides for advanced contacts, and TiN layers for CMOS. PP-TOFMS profiles compare well with ToF-SIMS depth profiles. Although depth resolution and sensitivity are less in the PP-TOFMS profiles both techniques were able to determine the composition, detect contamination, measuring doping level, and characterize diffusion as a function of annealing. X-ray fluorescence (XRF) and X-ray reflectivity (XRR) were used to provide thickness and composition information for comparison.

Acknowledgements: This work was supported by the French "Recherches Technologiques de Base" Program and was performed on the Nano Characterization Platform (PFNC) of the CEA Grenoble and at HORIBA Jobin Yvon.

[1] A. Tempez, S. Legendre, P. Chapon, Depth profile analysis by plasma profiling time of flight mass

Spectrometry, Nucl. Instr. Meth. B, 332, 351-354 (2014). http://dx.doi.org/10.1016/j.nimb.2014.02.094

Introduction of SiGe –either as a raised source/ drain strain element or as SiGe channel– in advanced CMOS technology has enabled significant enhancement of transistor performance [1, 2]. Quantitative analysis of Si_1-xGe_x composition by means of SIMS depth profiling has been demonstrated with good accuracy using either Cs⁺ or O₂⁺ primary beams [3]. Here we explore quantitative analysis of (in-situ) B-doped SiGe, using extremely low (down to 125 Ev) impact energy O₂⁺ primary beams.

Quantitative analysis of [B] in In-Situ Boron Doped (ISBD) SiGe using reactive low energy O₂⁺ ion sputtering is complicated due to large variations in both sputter and ionization yield with Ge-content [3, 4]. We present an extended multi-standard calibration protocol based on multiple B-implanted epitaxial Si_1-xGe_x standards on with constant [Ge] ranging from 20 to 90 at.%, with the implanted dose confined within the SiGe layer. This approach allows for explicit correction of both SiGe sputter yield and B⁺ and Ge⁺ yield variations over a wide range of [Ge]%, with Si_1-xGe_x composition and thickness in excellent agreement with X-Ray Diffraction (XRD). Boron sensitivity was observed to be dependent on [Ge] content, with pronounced effects of O₂⁺ impact energy in the 300-750 Ev energy range. Generally, we observe reduced dependence of both sputter and ion yield on Ge% at lower O₂⁺ impact energy. This is attributed to higher O-retention in Si_1-xGe_x, leading to smaller variations in steady state O-concentrations in SiGe. This is particularly useful to extend the quantification protocol towards Ge-rich SiGe ([Ge] > 50 at.%).

In summary, we have demonstrated quantitative analysis of (in-situ) doped SiGe in terms of Si_1-xGe_x composition, thickness and doping, employing low energy O₂⁺ primary beams. This enables to achieve sub-nm depth resolution (B decay length better than 7Å/decade) in ultrathin epitaxially grown B-doped SiGe /Si superlattice structures, comparable to best depth resolution reported in idealized d-B structures in Si [5, 6].

1. K. Cheng et al., 2012 IEEE International Electron Devices Meeting (IEDM), vol., no., pp.18.1.1,18.1.4, 2012

2. P. Hashemi et al., 2014 IEEE International Electron Devices Meeting (IEDM), vol., no., pp.16.1.1,16.1.4, 2014

3. C. Huyghebaert, T. Conard, B. Brijs, W. Vandervorst, Appl. Surf. Sci. 2004, 231–232, 708.

4. Z. Zhu, P. Ronsheim, A. Turansky, M. Hatzistergos, A. Madan, T. Pinto, J. Holt, Surf. Interface Anal. 43(1-2) (2011) 657-660.

5. A. Merkulov et al., J. Vac. Sci. Technol. B 28 (1), 2010, C1C48.

6. M. Tomita et al., J. Vac. Sci. Technol. B 2 (4), 2008, 1844.