Elsevier

Thin Solid Films

Volume 557, 30 April 2014, Pages 222-226
Thin Solid Films

Improved depth resolution of secondary ion mass spectrometry profiles in diamond: A quantitative analysis of the delta-doping

https://doi.org/10.1016/j.tsf.2013.10.076Get rights and content

Highlights

  • A double boron-delta-doped diamond sample was grown by homoepitaxy.

  • Delta layers are too thin and require to deconvolve the chemical depth profile.

  • The depth resolution function is extracted from a 13C-delta-doped diamond sample.

  • We confirm the conformal diamond growth layer by layer during the delta-doping.

Abstract

In this work, we used the depth resolution function (DRF) of the secondary ion mass spectrometry (SIMS) to deconvolve the boron depth profile of nanometer-thin embedded diamond layers. Thanks to an isotopic change within a thin layer, where carbon-12 (12C) and carbon-13 (13C) are substituted, the DRF was evaluated by a self-consistent algorithm. In a second step, this DRF was used to deconvolve the boron depth profile of a double delta-doped diamond analyzed under the same ion beam condition. The expected position, thickness, and boron concentration of the embedded layers were confirmed. This technique has enhanced the SIMS performance, and the depth resolution reached the nanometer range. Interface widths of boron-doped diamond multilayers were resolved well below 1 nm/decade over a large doping range, from 3 × 1016 cm 3 to 1.2 × 1021 cm 3, and confirmed a conformal growth layer by layer.

Introduction

The development of diamond growth technology has largely improved the fabrication of homo and heterostructures with abrupt interfaces such as superlattices and quantum wells [1]. Consequently, the request for a very accurate characterization has become more demanding even though the analysis of such structures is difficult and sometimes a challenge of its own (nanometer scale, low concentration of light atoms, hard material, and so on). Secondary ion mass spectrometry (SIMS) is commonly used to obtain depth profiles of dopants over many orders of magnitude in concentration. However, below 100 nm in thickness, SIMS induced ion mixing is no longer negligible; it affects strongly the depth profile measurements by broadening and distortion, so that the raw SIMS profile differs from the dopant profile, up to the point where thickness values and atom peak concentrations in multilayer stacks become erroneous. Other alternative and promising techniques like atom probe tomography [2] are not yet so commonly available, and in fact not yet demonstrated on the diamond material.

This work is dedicated to the potentiality of SIMS applied to the characterization of nanoscale diamond embedded heterogeneous structures. Diamond has several excellent properties, in most cases superior to those of other semiconductors, e.g., Si and SiC. Actually, two types of application require the availability of very thin layers (boron or nitrogen-doped) in the range of nanometer thickness, the so-called “delta structures” [3], [4], [5], as well as the possibility to characterize such ultrathin epilayers. These applications are related to high breakdown voltage/high temperature electronic devices [6] aimed at the development of next-generation high power devices, but also to colour centers, e.g., NV centers in diamond [3], [7], a very active research field of photonics and spintronics, more in line with the optical properties of diamond.

Technically, during a SIMS analysis, the experimental depth profile is the convolution of the dopant depth profile and of the depth resolution function (DRF) [8]. Evaluation of this DRF (which depends on the probed atom) is a key issue in nm-range secondary ion mass spectrometry. Deconvolution analysis using such a DRF provides accurate measurements on abrupt dopant depth profiles over many orders of magnitude in concentration. The best tool to estimate quantitatively the influence of ion mixing during the SIMS analysis is the local isotopic substitution (or “isotopically pure growth”). This has already been demonstrated with silicon superlattices (28Si/30Si) [9]. The atomic substitution by an isotope is the best approach to extract the experimental response, i.e. the DRF, because it introduces only a negligible difference in mass (same recoiling effect) and ionization threshold as well as no additional crystalline strain (same lattice parameter). Once the DRF expression is known for carbon in diamond, we can apply this function to determine a genuine dopant depth profile for nitrogen, or boron, or phosphorus.

However, the requirements to record an accurate DRF are stringent. The embedded layer has to be in the same thickness range as the lattice parameter. The fabrication of such structure requires strict conditions such as flat interface, no chemical diffusion in the matter, and a single crystalline substrate [10]. Diamond epitaxial multilayer stacks fulfill these requirements.

Section snippets

Diamond sample growth

Two diamond single crystalline samples were grown in this study. A first sample, composed of a synchronized boron- and carbon-13-doped layer, was used to extract the DRF from the 13C signal intensity. Furthermore, the fitting process was applied on the boron profile, in order to qualify the possibility to deconvolve the boron concentration and the layer thickness. The second sample was constituted of a double boron-doped delta layers in order to analyse the growth uniformity and the interface

SIMS profile fitting

Several authors have reported that a SIMS profile can be modelled by convolving the genuine atom profile with the SIMS depth resolution function, a response which depends on instrumental and fundamental aspects as well (convolution model). In the 90s, Dowsett et al. [10] have demonstrated that for delta-doped layers characterized by few atomic layers and hence below the SIMS resolution, an excellent approximation of depth resolution function (DRF) can be obtained by convolving a double

Discussion

The deconvolved boron profile of the first sample was found to be 5 to 7 times sharper: the initially measured 1.5 nm/decade rising edge became 0.3 nm/decade (see Fig. 6). The best fit seemed to justify the presence of a finite thickness for the interface located between boron-doped and intrinsic diamond layer. Nevertheless, such interface thickness was composed of two points only. The sampling (Δz = 0.9 nm) was not rich enough to really conclude about an exponential or linear dependence on the

Conclusion

By a local isotope enrichment of diamond, we were able to extract the instrument response of the SIMS and to characterize the incorporation of both carbon and boron atoms. These treatments allowed to increase the SIMS resolution, in order to subtract the broadening and the distortion induced by ion-mixing and to reach the nanometer-range. This procedure yielded a more reliable characterization by SIMS of nanometer thin diamond embedded layers containing specific impurities over a wide range of

Acknowledgments

The financial support of Agence Nationale de la Recherche under contract ANR08-BLAN-0195 and la Région Rhône-Alpes for the bourse de mobilité Explora'Doc is gratefully acknowledged. This work was also supported by the Strategic International Collaborative Research Project from the Japan Science and Technology Agency, Japan and Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science, Japan (No. 23360143).

References (18)

  • H. Watanabe et al.

    Diam. Relat. Mater.

    (2011)
  • A. Fiori et al.

    Diam. Relat. Mater.

    (2012)
  • H. Umezawa et al.

    Diam. Relat. Mater.

    (2012)
  • Y. Shimizu et al.

    Appl. Surf. Sci.

    (2008)
  • M. Kamo et al.

    J. Cryst. Growth

    (1983)
  • J.F. Ziegler et al.

    Nucl. Instrum. Meth. Phys. Res. Bull.

    (2010)
  • D. Araujo et al.

    Diam. Relat. Mater.

    (2010)
  • Y. Shimizu et al.

    J. Appl. Phys.

    (2013)
  • K. Ohno et al.

    Appl. Phys. Lett.

    (2012)
There are more references available in the full text version of this article.

Cited by (24)

  • Hopping conduction of non metallic heavily doped delta-layers in CVD diamond

    2021, Diamond and Related Materials
    Citation Excerpt :

    Thin (with thickness of order several crystal lattice constants) heavily doped layers (so called delta-layers) in chemically vapor deposited (CVD) diamond were obtained in many laboratories [1–8] and believed to be a promising tool to increase the charge carrier mobility for various electronic applications, e.g. for the creation of CVD diamond-based high-speed field-effect transistors [9,10].

  • High resolution boron content profilometry at δ-doping epitaxial diamond interfaces by CTEM

    2018, Applied Surface Science
    Citation Excerpt :

    Fig. 5 shows SIMS profile the sample, dark cross in the left axis has been used to highlight obtained value of boron doping content x = 1.6·1021 at/cm−3, dashed line has been used to guide the eyes on this value, evidencing a minor discrepancy between the SIMS measured value of 1021 at/cm−3 and the calculated one. Fig. 5 also evidences a discrepancy in the measured doped layer thickness value (this is a knew and common artifact induced by SIMS operation procedure [33]). The different boron content obtained, however, can be due to the thickness of doped layers studied.

  • Investigation of boron incorporation in delta doped diamond layers by secondary ion mass spectrometry

    2018, Thin Solid Films
    Citation Excerpt :

    The problem of deconvolution of the delta layer profile is non-trivial [30–33] and cannot be solved without knowledge of the DRF or additional a priori assumptions. In this paper, we employ a widely used approach to the reconstruction of the DRF profile [24,34,35]. We assume that the boron concentration profile in the delta layer is rectangular.

  • Analysis of boron in diamond with UV photoluminescence

    2017, Carbon
    Citation Excerpt :

    The optical, physical and electronic properties of a diamond are influenced by its impurities and defects that hence require characterization and analysis [1–3]. Among many and diverse analytical methods established to determine qualitatively or quantitatively these impurities and defects, both destructive and non-destructive techniques have been developed to characterize not only natural diamonds but also those from synthesis at great pressure and from deposition of a chemical vapor (CVD) [4–12]. In particular, photoluminescence (PL) emission upon excitation with visible or ultraviolet light provides a sensitive and non-destructive technique for the optical characterization of impurities and defects in diamonds [13].

  • Lithium implanted into diamond: Regular trends and anomalies

    2016, Surface and Coatings Technology
    Citation Excerpt :

    As compared with the calculated profile, the SIMS profile of lithium has a more extended tail into the depth of the crystal. These distinctions can be explained at least in two ways: а) ion channelling to a greater depth is not taken into account in the calculation by the Monte Carlo method [48] and due to the reduction of energy losses of ions because of accumulation of radiation damage with the build-up of the fluence [49]; each of these factors extends the distribution tail; b) in the SIMS method, the ion sputtering of material is accompanied with the knocking material atoms (including impurity atoms) into the depth – the ion beam pushes impurity atoms (especially such light ones as 7Li) into the depth as long as ion sputtering continues [50]. This distorts the depth distribution of impurity in areas where its concentration decreases, and worsens the resolution of the SIMS method.

View all citing articles on Scopus
View full text