Diamond thin films: a twenty-first century material. Part 2: a new hope

Abstract

Nearly a quarter of a century ago, we wrote a review paper about the very new technology of chemical vapour deposition (CVD) of diamond thin films. We now update this review and bring the story up to date by describing the progress made-or not made-over the intervening years. Back in the 1990s and early 2000s, there was enormous excitement about the plethora of applications that were suddenly possible now that diamonds could be fabricated in the form of thin films. Diamond was hailed as the ultimate semiconductor, and it was believed that the few remaining problems would be quickly solved, leading to a new 'diamond age' of electronics. In reality, however, difficulty in making large-area diamond wafers and the elusiveness of a useful n-type dopant slowed progress substantially. Unsurprisingly, over the following decade, the enthusiasm and funding for diamonds faded, while competing materials forged ahead. But in the early 2010s, several new game-changing applications for diamonds were discovered, such as electrochemical electrodes, the nitrogen-vacancy (NV) centre defect that promised room-temperature quantum computers, and methods to grow large single-crystal gemstone-quality diamonds. These led to a resurgence in diamond research and a new hope that diamond might finally live up to its promise.This article is part of the theme issue 'Science into the next millennium: 25 years on'.

Keywords: chemical vapour deposition; diamond; review.

Figure 1.

Diagram of a LAPD system. Redrawn from [20].

Figure 2.

(a) Schematic diagram of a DAAM system, redrawn from [22]. MW power is coupled into a rectangular waveguide and is divided equally between 16 applicators (180 W each), which transport the power to the sources within the chamber using coaxial cables. One of the applicators is shown in more detail on the left; the power passes through a matched load into a tuned antenna protected by a quartz coating, igniting a plasma around each source protruding inside the low-pressure reactor. (b) View of the 16 coaxial ignited plasma sources plasma sources inside the DAAM reactor arranged in a 4 × 4 matrix. Photo reproduced with permission from [22].

Figure 3.

Schematic illustrations of components of a SWP reactor, (a) a slot antenna, (b) a MW launcher with five slots, (c) a large-area MW antenna composed of four launchers and (d) large-area SWP reactor. Figure based upon diagram in [23].

Figure 4.

An example of a top-hat MWCVD reactor (Seki model SDS6380). This reactor has power supplies up to 5 kW (at 2.45 GHz) and is capable of depositing polycrystalline diamond films, SCD diamond plates or SCD gemstones at high rates. The chamber is water-cooled, with a substrate platen capable of accommodating samples up to 50 mm in diameter. Photo courtesy of Philippe Bergonzo, Seki Diamond Systems [13].

Figure 5.

Series of elementary steps by which an incident CH₃ radical can insert into a C−C dimer bond on the C(100):H (2 × 1) surface. The respective minima (normal font) and transition state (in italics) energies are in kJ mol⁻¹ relative to that of structure 1. The dots indicate the location of a radical group, i.e. a ‘dangling bond’ with an unbonded electron. Redrawn using data in [29].

Figure 6.

3D projections of the surfaces generated by a 3D kMC model representing the different types of diamond growth resulting from different process conditions. More details of these diamond film varieties can be found in §5. (a) UNCD, with no crystal faces visible as a result of continual random renucleation. (b) Nanocrystalline diamond (NCD), with some small terraces appearing (i.e. nanocrystallites) as a result of a layer-by-layer growth coupled with renucleation. (c) Microcrystalline diamond (MCD), where layer-by-layer growth dominates leading to large micron-sized faceted crystallites. (d) SCD, where growth is now entirely by layer-by-layer growth. The layers have been coloured to make it easy to distinguish individual layers. Reproduced with permission from [42] under the Creative Commons CC BY 3.0 licence.

Figure 7.

SEM image showing macroscopic step-bunching on the surface of an SCD sample grown using nitrogen-containing gas mixtures. Reprinted with permission from [51].

Figure 8.

Schematic representation of various types of CVD diamond film deposited on to a substrate (grey). The colours yellow, brown and black are used to allow the reader to discriminate between different diamond crystallites (grains) and their growth evolution. (a) SCD film grown epitaxially on an SCD substrate. (b) MCD columnar growth from randomly located nuclei, where the slowest growth face determines the overall film texture, in this case (100). (c) Highly oriented textured MCD obtained following special nucleation procedures, such as BEN (see §7). (d) Faceted NCD, which is really just thin MCD with high nucleation density. (e) ‘Cauliflower’ (or ‘ballas’) NCD before it has coalesced into a continuous film. (f) Cauliflower NCD film. (g) UNCD. Computer simulations of some of these diamond-film types were shown previously in figure 6.

Figure 9.

Freestanding unpolished large-area SCD wafer synthesised by heteroepitaxy on Ir/YSZ/Si(001). The thickness of the disc is 1.6 ± 0.25 mm and its weight is 155 carat. Reproduced from [74] under the Creative Commons Attribution 4.0 License.

Figure 10.

Conductivity (σ) at 300 K as a function of boron content measured by secondary ion mass spectrometry (SIMS) for various boron-doped diamond samples. Different conductivity mechanisms operate at different B concentrations, as shown by the labelled regions on the plot. The curve predicted by the Seto model of conductivity is plotted as a solid line. Figure reproduced from [134] with permission and colourised. For [B] > 1 × 10²¹ cm⁻³ and temperatures less than 10 K (not shown) the conduction mechanism switches to superconductivity via Cooper pairs.

Figure 11.

The three most common dopant types (e.g. phosphorus, nitrogen and boron) in diamond with the associated energies within the band gap; VB and CB stand for the valence band and conduction band, respectively. The B acceptor level is measured from the top of the VB, while the P and N donor levels are measured from the bottom of the CB.

Figure 12.

Schematic diagram of the NV defect centre in the diamond unit cell.

Figure 13.

Examples of fancy-coloured CVD diamond gemstones. Photo reproduced with permission of Infi Advanced Materials Co., Ltd [176].

Figure 14.

Procedure to produce clones and tiled clones. Figure redrawn based upon diagram in [206].

Figure 15.

The potential windows of several different electrochemical electrodes. Redrawn using data from [210].

Figure 16.

An example of the sensitivity and selectivity of BDD electrodes used for electrochemical trace analysis in water. (a) Differential pulse voltammograms recorded for different concentrations of dopamine (DA) (1–9 = 0.0–5.0 × 10⁻⁶ M) in the presence of a chemically similar analyte (3.0 × 10⁻⁵ M uric acid). Inset: current density j (mA cm⁻²) versus concentration of dopamine (mM) showing a linear response with concentration. (b,c) SEM images of ‘black diamond’, nanostructured BDD needles on the surface of an electrode, with electrochemically active area 220 times larger than that of a flat diamond electrode. Figures reprinted under CC BY 4.0 licence from [211]. (Published by The Royal Society of Chemistry).

Figure 17.

Different vertical device structures for diamond diodes. (a) Vertical SD using only p-type diamond. (b) Pseudo-vertical SD that uses only p-type diamond but no back contact. (c) PIN diode composed of a sandwich of p⁺-type diamond, an undoped intrinsic diamond layer and an n-type diamond layer, with Ohmic contacts on either side. (d) Schottky PN diode. Figure reproduced from [265] and modified under the Creative Commons Attribution 3.0 licence.

Figure 18.

A typical design for a diamond lateral deep-depletion MOSFET-style device. The type-1b diamond substrate (usually HPHT) has a p-type BDD layer (green, labelled 'P') deposited homoepitaxially on to it. The source and drain electrodes are composed of heavily BDD (grey, labelled P⁺) capped with a highly conducting metal such as Au (gold), with a thin Ti layer (blue) at the interface allowing a good electrical contact to be formed with the underlying diamond. These are patterned using standard photolithography and dry etching methods. The Au gate electrode is electrically isolated from the BDD layer by a gate oxide (grey), in this case made from insulating Al₂O₃. Image reproduced and modified from [265]. Creative Commons Attribution 3.0 licence.

Figure 19.

Schematic cross-sectional view of a diamond-based MOSFET device utilizing the 2DHG conduction channel in H-terminated diamond protected by a MoO₃ capping layer, based on the design proposed in [156]. S, G and D refer to source, gate and drain, respectively.

Figure 20.

A simplistic representation of quantum entanglement of two photons. (a) Two independent photons with opposite spin (depicted as red and blue) emitted from different NV centres approach each other inside a fibre-optic waveguide. (b) When they meet, their wavefunctions entangle, and they both share the same superposition wavefunction—i.e. the photons are now both red and blue simultaneously (depicted as purple). (c) When they separate, both photons remain entangled in the same shared quantum state. (d) If one of the photons is observed, the superposition wavefunction collapses, and the spin of that photon is fixed at that instant as being either up or down (red or blue, in this case red). At that same instant, the spin of the other photon is also known (blue), even though it was never observed and might now be thousands of kilometres from its partner.