Skip to content

Influence of growth-sector structure on electrical characteristics of Schottky and ohmic contacts to boron-doped HPHT diamond

FieldValue
JournalSemiconductor Physics Quantum Electronics & Optoelectronics
Published2026-06-22
AuthorsYa. Ya. Kudryk, V. V. Strelchuk, D.M. Maziar, A.E. Belyaev, Т.В. Коваленко
DOI10.15407/spqeo29.02.162

Growth-sector structure produced different metal/diamond transport behavior across one [001]-oriented boron-doped HPHT diamond wafer. Uncompensated boron reached 1.25 × 10¹⁹ cm⁻³ in the most doped sector, and the sector-to-sector difference could reach one order of magnitude. Ti/Pt/Au contacts in the (001) and (113) sectors showed lower specific contact resistance than contacts in (111) at comparable boron concentrations.

A [001]-oriented cut from one HPHT crystal retains a common macroscopic orientation. Its lateral regions can still originate from different growth faces, so each region has its own boron incorporation and defect formation history. The principal sectors studied were (111), (113), and (001).

The μ-FTIR histograms gave these characteristic uncompensated boron concentrations:

Sample(111)(113)(001)
A5.5 × 10¹⁸ cm⁻³1.95 × 10¹⁸ cm⁻³0.75 × 10¹⁸ cm⁻³
B1.25 × 10¹⁹ cm⁻³7.8 × 10¹⁸ cm⁻³0.8 × 10¹⁸ cm⁻³

These were modal values, not uniform levels within each sector. Sample A had the highest concentration in (111), followed by (113) and (001). Sample B followed the same order, with much higher concentrations in (111) and (113). Structures on sector boundaries were excluded from sector statistics.

Boron-doped single crystals were grown by the temperature-gradient method in a Fe-Al-B-C system at 6.5 GPa and 1380 to 1420 °C. The controlled boron content of the growth medium ranged from 2 × 10⁻⁴ to 10⁻² at.%.

Two [001]-oriented substrates, each about 500 μm thick, were cut from separate bulk crystals. Mechanical polishing produced a surface RMS roughness of about 0.3 nm for contact formation.

Uncompensated boron was mapped in transmission by μ-FTIR with a 50 μm step and a 100 × 100 μm² aperture. Spectra covered 650 to 4000 cm⁻¹ at 4 cm⁻¹ resolution. Concentrations came from boron absorption bands at 2800, 2450, and 1290 cm⁻¹.

Before metallization, the authors cleaned the samples for 30 min in HCl:HNO3 = 3:1. They then rinsed for 3 min in deionized water, boiled for 30 min in H2SO4:HNO3 = 3:2, rinsed again for 3 min, washed in isopropanol for 5 min, and dried in nitrogen.

Optical lithography used a 2.7 μm positive photoresist and a 405 nm laser writer. Ti/Pt/Au ohmic stacks had layer thicknesses of 38/75/50 nm and received 600 °C rapid thermal annealing for 60 s. Ni/Pt/Au Schottky stacks had thicknesses of 75/75/50 nm and were annealed at 250 °C for 60 s.

The arrays contained 150 planar Schottky diodes on sample A and 75 on sample B. Equal-area diodes with a 25 μm barrier-contact radius were used for leakage statistics. Linear TLM patterns used 300 μm strips and spacings from 10 to 50 μm. All I-V characteristics were measured at room temperature.

The Schottky arrays showed two transport regimes. In the lower-doped regions, good diodes had reverse currents of about 10⁻¹² to 3 × 10⁻¹¹ A at a 2 V reverse bias. A second population reached 10⁻⁶ A and carried excess current at low forward bias.

The high-leakage response was consistent with a localized shunting channel rather than a smooth doping effect. The authors associated this behavior with killer defects, commonly attributed to dislocations decorated with impurity atoms. This was an electrical interpretation rather than a direct measurement of each defect.

The quality yield did not follow the boron-concentration order. For sample A, defect-free yields were 60% in (111), 54% in (113), and 67% in (001). The sector counts were 40, 13, and 15, with 24, 7, and 10 classified as defect-free.

Earlier work cited by the authors linked lower dislocation density with uniform mosaic-wafer devices and breakdown fields above 3 MV cm⁻¹. This study did not report breakdown fields.

Above approximately 9 × 10¹⁸ cm⁻³, reverse current became strongly dependent on uncompensated boron concentration. The most heavily doped regions became nearly ohmic, and many structures showed shunting or tunneling current.

The authors interpreted the change with the Padovani-Stratton framework. At room temperature, kT/q was 0.026 eV, and the thermionic-field-emission crossover corresponded to Na = 8 × 10¹⁸ cm⁻³. More acceptors narrow the depletion region and raise tunneling probability.

For sample B, the authors assigned thermionic emission to (001), thermionic-field emission to (113), and field emission to (111). Their characteristic concentrations were 0.8 × 10¹⁸, 7.8 × 10¹⁸, and 1.25 × 10¹⁹ cm⁻³, respectively. Rectification was strongest in (001) and most suppressed in (111).

The Ti/Pt/Au specific contact resistance decreased as uncompensated boron concentration increased. The trend is consistent with a narrower interfacial barrier and greater tunneling-assisted injection. This parameter averages over the contact area, so isolated shunting defects are unlikely to explain the sector offset.

The decrease was stronger than the concentration scaling cited from Kupka and Anderson, which predicts an approximately order-of-magnitude resistance reduction for an order-of-magnitude doping increase. Concentration was therefore important but incomplete.

At comparable uncompensated boron concentrations, contacts in (001) and (113) generally had lower specific contact resistance than contacts in (111). The same ordering appeared in dedicated TLM measurements and values estimated from Schottky forward branches. The authors proposed sector-dependent compensation, defect density, or near-surface state as possible causes, but the data did not identify the microscopic origin.

The study covered two substrates and room-temperature transport in one Ti/Pt/Au ohmic stack and one Ni/Pt/Au Schottky stack. Sector-boundary devices were omitted from sector statistics. The data show sector trends, but do not separate compensation, dislocation structure, and interface chemistry as the cause of the ohmic-resistance offset.

For Schottky design, a [001] wafer orientation cannot replace a local boron map. Below about 9 × 10¹⁸ cm⁻³, leakage and yield depend chiefly on local shunting defects. Above that level, high doping can erase useful rectification. Layouts should screen for shunting defects and avoid the heavily doped (111) sector when rectification is required.

For ohmic contacts, record sector affiliation alongside the measured uncompensated boron concentration. Equal nominal doping does not guarantee equal resistance on one HPHT wafer. Contact screening should include local μ-FTIR mapping, dedicated TLM structures, and tests on the final metal stack and annealing process.

The paper’s material requirement is single-crystal boron-doped HPHT diamond for Schottky diodes and Ti/Pt/Au ohmic contacts, with local boron mapping before metallization. 6CCVD’s diamond materials catalog lists boron-doped diamond, single-crystal and polycrystalline diamond grown by MPCVD, plates and wafers up to 125 mm, and metallization with Au, Pt, and Ti. Because the paper used HPHT growth, engineering support should confirm contact behavior and annealing for the selected MPCVD substrate.

View Original Abstract

Influence of growth-sector structure on electrical characteristics of Schottky diodes and ohmic contacts fabricated on boron-doped high-pressure high-temperature (HPHT) growth diamond was investigated by correlating electrical measurements with μ-FTIR mapping of uncompensated boron concentration. The studied substrates exhibited pronounced multisectoral inhomogeneity of electrically active boron with systematic variation between the principal growth sectors. The Schottky structures revealed two characteristic transport regimes. In the lower-doped regions, the reverse current was mainly controlled by localized shunting defects, whereas in the higher-doped regions it increased strongly with uncompensated boron concentration and the contacts gradually changed their behavior from rectifying to nearly ohmic one, consistent with increasing contribution of tunneling-assisted transport. The specific contact resistance of the Ti/Pt/Au ohmic contacts decreased with the boron concentration and exhibited clear dependence on growth-sector affiliation. At comparable uncompensated boron concentrations, the contacts formed in the (001) and (113) sectors exhibited lower specific contact resistance than those formed in the (111) sector. The results demonstrate that the growth-sector structure of the boron-doped HPHT diamond influences not only spatial distribution of electrically active boron, but also transport characteristics of metal/diamond contacts therefore representing an important technological factor in fabrication and optimization of diamond-based electronic devices.