Heat extraction is often essential to ensuring efficient performance of semiconductor devices and requires minimizing the thermal resistance between the functional semiconductor layers and any heat sink. This paper reports epitaxial growth of N-polar GaN films on polycrystalline diamond substrates of high thermal conductivity with metal-organic vapor phase epitaxy by using a SixC layer formed during deposition of polycrystalline diamond on a silicon substrate. The SixC layer acts to provide the necessary structure ordering information for the formation of a single crystal GaN film at the wafer scale. It is shown that a three-dimensional island (3D) growth process removes hexagonal defects that are induced by the non-single crystal nature of the SixC layer. It is also shown that intensive 3D growth and the introduction of convex curvature of the substrate can be deployed to reduce tensile stress in the GaN epitaxy to enable the growth of crack-free layer up to a thickness of 1.1m. The twist and tilt can be as low as 0.65 and 0.39 respectively, values broadly comparable with GaN grown on Si substrates with a similar structure. ; Description of data All the data used in this article takes the form measurable quantity (dependent variable) versus parameter varied under operator control (independent variable) and is recorded in tabulated form in spreadsheets. Apart from one data set referred to in a “non-quantitative way”, all the data used was presented in the above article in graphical form indicated as figures 1b, 2, 3 etc. Figure 1a is a sketch and therefore does not contain measurable data. The data-containing graphs are: Figure 1b in file Fig_1b_curvature: The bow of both kinds of PD substrates measured by moving a digitally-controlled stylus across the sample surface over a distance of 5 mm and measuring the vertical deflection of the stylus every few microns. The PDA substrate was found to have a convex bow (~1.5 μm over 5 mm) whilst PDB had a concave bow (~6 μm over 5 mm), where convex and concave are defined in the article. Figure 2 in file Jiang_Fig_2_xps_data: X-ray photo-electron spectroscopy (XPS): intensity of emitted x-ray photon versus binding energy. These data reveal the nature of the chemical bonds between the elements found on the surface or immediate sub-surface layer of the samples under consideration. This in turn provides information about the chemical nature of the surface on which the III-Nitride layers were grown. In describing the data in this graph, reference is made to energy dispersive x-ray (EDX) measurements that confirm the presence of both silicon and carbon on the surface of the nominally polycrystalline diamond growth substrates, both in large quantities. Taken together with the chemical bonding information obtained from the XPS measurements (presented in figure 2) it was inferred that almost all the silicon present was bonded chemically to carbon, with some silicon bonded to oxygen. Figure 3 in file Jiang_Fig_3_reflectivity: These data were collected via in-situ measurements of the intensity of a visible laser beam (typically red light) from the surface of the III-Nitride layers as they were grown using a standard commercially available tool (specified in the article). The independent variable is time. Figure 4 in file Jiang_Fig_4_Surface_MOR: The data are images formed by the secondary electron emission, taken by digital camera during inspection of the surface of the grown III-Nitride layer by scanning electron microscopy. The data sets in the form of the intensity reaching each pixel in a two dimensional array of such pixels in the digital camera. Figure 5 in file Jiang_Fig_5_comparison_of_XRD_3D_GaN(0002)_and_GaN(103): These data are the intensity of a diffracted x-ray beam measured by rocking samples 3 and 4 (described in the article, especially in Table 1) when illuminated by a beam of x-rays under conditions that reveal the twist in the III-Nitride layers [beam alignment along crystallographic direction GaN(10-13)] and the tilt [ beam alignment along GaN(0002)]. Figure 6 in file Jiang_Fig_6_E2H Raman_comparison_3D_growth_time: Raman spectra of Sample 3 and Sample 4 in the vicinity of the E2H phonon mode for two samples to reveal the impact of the duration of the 3-diminsional growth mode on the strain in the GaN epitaxial films. Figure 7 in file Jiang_Fig_7_Raman_comparison of growth on bowed substrates: Raman spectra in the vicinity of the E2H phonon mode for (a) Sample 5 grown on a concave PDB substrate and (b) Sample 6 grown on a convex PDA substrate to reveal the impact on the strain in the GaN epitaxial films of the growing the III-Nitride layers on bowed substrate. ; Apart from the photographs the data processing involved only plotting measured dependent variable values against the corresponding measured independent variable values. Where applied, curve fitting performed by using the polynomial curve fitting tool available in Excel. Otherwise the data were not processed in any way. ; Two kinds of PD substrate (hereafter called PDA and PDB) from different suppliers [Element Six Ltd, Diamond Materials GmbH] were used in the work, both supplied with their initial Si(111) growth substrate intact. In each case the Si(111) was removed by immersion in hot KOH or/and isotropic Si etching solution (a mixture of HNO3 and HF). A distinguishing feature of the two types of substrate after removal of the Si(111) substrate is their curvature, as shown in Figure 1a. The curvature measured by a stylus technique (Dektak Stylus profiler), shown in Figure 1b, occurs as a result of residual stress in the PD layers [9]. PDA substrates exhibited a convex bow (~1.5 micron over 5 mm) while PDB substrates had concave bow (~6 micron over 5 mm). X-ray photoelectron spectroscopy (XPS) measurements were performed under sub-contract at the University of Cardiff. Energy dispersive x-ray spectroscopy (EDX) was performed a JEOL SEM6480LV electron microscope at the University of Bath. Surface reflections were recorded during epitaxial growth of all samples using a LAYTEC EpiSense system. X-ray diffraction (XRD) measurements were performed using a BEDE D1 system. Raman spectroscopy was performed using a Renishaw inVia systemusing a laser source of wavelength = 530 nm). The curvature of the substrates after epitaxial growth was measured with a Dektak stylus profiler. All data collection was performed using the software supplied by the equipment manufacturers available at the time of purchase of the equipment. All equipment was operated under the standard conditions recommended by the equipment manufacturers and all measurements were performed at room temperature, except the surface reflection measurements performed during epitaxial growth, which were performed at the temperatures specified in the manuscript. ; Sample preparation apart from the growth of the polycrystalline diamond substrates (which were supplied commercial manufacturers named in the manuscript) and epitaxial growth, was carried out in David Bullett Nanofabrication Facility at the Univesity of Bath.