Imaging Contrast of Diamond Nanoparticle Structures on Different Supporting Films
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Molecules |
| Published | 2026-09-22 |
| Authors | Yuying Yang, Feng Jiang, Yueyun Li, Jie Zhang, Tianhui Wu |
| DOI | 10.3390/molecules31193368 |
Graphene oxide raised the contrast-to-noise ratio of 8 nm nanodiamond HRTEM images from 1.3 to 3.9, a factor of 3.0 over amorphous carbon. It also raised {111} FFT detection from 35% to 85% for the same particle size.
Measured contrast advantage
Section titled “Measured contrast advantage”The 8 nm CNR was 3.9 ± 0.6 on GO and 1.3 ± 0.4 on amorphous carbon. Each condition included 20 particles, and the support comparison gave p < 0.001. The {111} detection rate rose from 35% on carbon to 85% on GO for these 8 nm particles.
GO also produced sharper interfaces. The 8 nm edge intensity drop was 69 ± 8% on GO and 35 ± 9% on carbon. Its {111} lattice visibility was 0.42 ± 0.06 on GO and 0.19 ± 0.03 on carbon.
For 6 nm particles, GO gave CNR 2.6 ± 0.5 and detection rate 60%. Carbon gave CNR 0.6 ± 0.3 and detection rate 15%. The authors interpret this result as a shift in the detectable size limit toward smaller particles.
Specimen preparation
Section titled “Specimen preparation”Explosion-produced diamond powders spanned 5 to 10 nm. The quantitative comparison used nominal particle sizes of 6, 8, and 10 nm.
Graphene oxide was prepared by a modified Hummers method. Graphite powder of 1.0 g was stirred in an ice bath before 5.0 g of KMnO4 was added slowly. The suspension remained below 20 °C during addition and was then stirred at 20 °C ± 2 °C for about 48 h. A 0.02 mg/mL suspension was ultrasonicated for 2 h to obtain single-layer or few-layer sheets. One drop was dried on a 400-mesh TEM grid before the nanodiamond was added.
Selected-area electron diffraction showed a sixfold pattern consistent with a hexagonal lattice. The inner 010 reflections were stronger than the outer 110 reflections, consistent with monolayer GO. Imaging regions were predominantly single-layer, as confirmed by diffraction and low, uniform HRTEM backgrounds. Overlapping GO sheets had a 14.0° torsion angle, but particles were imaged mainly on single-layer regions.
GO’s oxygen-containing groups made its surface hydrophilic and enabled uniform aqueous nanodiamond dispersion.
TEM conditions and quantification
Section titled “TEM conditions and quantification”HRTEM imaging used an FEI Tecnai G2 F20 S-TWIN microscope at 200 kV without aberration correction. The microscope had a 0.24 nm point resolution and a 0.14 nm lattice information limit. No objective aperture was used.
CNR was calculated from the intensity difference between the particle center and adjacent support film. The difference was divided by the support film’s intensity standard deviation. The particle-center region of interest was a 3 to 5 nm diameter circle. Line profiles measured edge intensity from the particle center to its boundary. Lattice visibility used the difference between peak and valley fringe intensities divided by their sum. FFT patterns were computed from 5 nm × 5 nm regions containing each particle. Two observers independently scored the presence of {111} reflections.
Each support and size combination had 20 analyzed particles. CNR, edge drop, and lattice visibility used two-tailed t-tests with Welch’s correction. Detection rates used a two-tailed Fisher’s exact test, and p < 0.05 was the significance threshold. All analyses used mean ± SD, except detection-rate uncertainty, which was reported as the standard error of the proportion.
No filtering, denoising, or background subtraction was applied to the HRTEM images. Only global brightness and contrast changes were made for publication. Original images are supplied in the supplementary figures.
Why support choice changes interpretation
Section titled “Why support choice changes interpretation”Film-free mounting caused 6, 8, and 10 nm particles to cluster over grid holes. Overlapping particles produced coarse interference fringes and obscured individual edges. One overlap showed a lamella twin as Moiré fringes from overlapping {111} lattices. Other marked boundaries were co-axis {111} twins viewed along <110>. Their rotation was 70.52° around <111>, consistent with a Σ3 coincidence-site lattice.
A continuous amorphous carbon film was about 15 nm thick. Its diffuse scattering exceeded the weak scattering from particles below 10 nm. The 6, 8, and 10 nm particles provided no analyzable crystallographic information along characteristic zone axes. Contrast reappeared as particle diameter increased because the projected particle potential became denser. The GO layer was 1.4 Å thick, so it scattered far less material than the carbon film. The authors attribute the improvement to lower background scattering and phase noise.
The figures label underfocus as Δf ≈ −56 nm, Scherzer defocus as −66 nm, and overfocus as −76 nm. Scherzer defocus produced positive phase contrast and the clearest structural detail. Contrast and sharpness changed sharply with focus for every particle size. Over- and underfocus obscured lattice detail.
Size-dependent quantitative results
Section titled “Size-dependent quantitative results”Values are mean ± SD, except {111} detection, whose uncertainty is the standard error of the proportion. Each row contains 20 particles.
| Support | Particle size (nm) | CNR | {111} lattice visibility | {111} FFT detection (%) |
|---|---|---|---|---|
| Carbon | 6 | 0.6 ± 0.3 | 0.10 ± 0.03 | 15 ± 8 |
| Carbon | 8 | 1.3 ± 0.4 | 0.19 ± 0.05 | 35 ± 11 |
| Carbon | 10 | 1.8 ± 0.5 | 0.29 ± 0.03 | 70 ± 10 |
| GO | 6 | 2.6 ± 0.5 | 0.30 ± 0.04 | 60 ± 11 |
| GO | 8 | 3.9 ± 0.6 | 0.42 ± 0.06 | 85 ± 8 |
| GO | 10 | 5.2 ± 0.7 | 0.42 ± 0.06 | 95 ± 5 |
For 10 nm particles, CNR reached 5.2 ± 0.7 on GO and 1.8 ± 0.5 on carbon. The corresponding detection rates were 95% and 70%, and edge intensity drops were 80 ± 7% and 44 ± 10%. GO’s lattice visibility increased from 0.30 ± 0.04 at 6 nm to 0.42 ± 0.06 at 8 and 10 nm. Carbon rose from 0.10 ± 0.03 to 0.19 ± 0.05 and 0.29 ± 0.03 over the same sizes.
Fitted slopes were 0.0532 nm⁻¹ for GO and 0.0499 nm⁻¹ for carbon, with R² values of 0.999 and 0.996. All support comparisons had p < 0.01, except the 6 nm detection rate, which had p < 0.05.
Beam stability and practical workflow
Section titled “Beam stability and practical workflow”GO can undergo electron-beam-induced deoxygenation at 200 kV. The authors cite prior in situ work rather than measuring long exposure directly. It reports stable morphology without noticeable shrinkage or hole formation during the first 10 to 20 s.
Images here used electron flux below 100 e⁻ Å⁻² s⁻¹ and exposure times of 1 to 2 s. The accumulated dose was below 200 e⁻ Å⁻² per image. Focusing and aberration tuning typically required less than 10 s of exposure on the region of interest. The authors argue that partial reduction preserves the single-layer morphology and low projected mass. Thus, they recommend low-dose acquisition within the stable time window for routine imaging.
Graphene is more chemically stable and structurally uniform, but its CVD growth and transfer are slower and less accessible. GO can be drop-cast from an aqueous dispersion without a transfer step. They present GO as the practical choice for routine nanoparticle HRTEM, not for every beam-sensitive experiment.
Limits and characterization guidance
Section titled “Limits and characterization guidance”The study compared one GO preparation route with one conventional amorphous carbon film. All tested particles came from explosion-produced powder and had nominal diameters of 6, 8, or 10 nm. The results do not establish transfer to other diamond sources, morphologies, or support thicknesses. Each size and support condition had 20 particles, so the data cover a narrow experimental set.
The paper gives carbon CNR for 8 nm particles as 1.3 ± 0.3 in one passage and 1.3 ± 0.4 in Table 1. It reports the 8 nm detection-rate gain as 2.5× in the discussion and 2.4× in the conclusion. Table 1 values and raw detection rates are used here. The paper does not report how nominal size populations were separated before imaging.
For diamond nanoparticle characterization, support film choice should be treated as part of the measurement. GO was the better support in this comparison for recovering {111} structure from sub-10 nm particles. The authors identify graphene as the more stable alternative for prolonged beam exposure. At 8 nm, the support choice changed CNR from 1.3 ± 0.4 to 3.9 ± 0.6 and detection from 35% to 85%.
View Original Abstract
Electron microscopy is essential for nanoparticle research because it provides the critical structural data of particle size, morphology, crystallography, and elemental composition that are essential for tailoring and for predicting nanoparticles’ optical, electronic, catalytic, and mechanical properties. The weak scattering from nanoparticles results in low contrast in electron microscope images. Therefore, observing and interpreting such images requires special care to avoid erroneous conclusions. This study investigates the imaging contrast of diamond structures on different supporting films. When the nanodiamond is overlapped with a carbon film, particle edge details cannot be explained, and the overlapping interference lattices produce a pattern of interference fringes that is much coarser than the original pattern. Single graphene oxide (GO) sheets, as the ideal support film, are highly electron-transparent for the study of nanoparticles. As a low-background and high-transparency support to identify the dominant structural information of the nanodiamond, GO is superior to carbon support film. Quantitative contrast analysis reveals that GO film improves the contrast-to-noise ratio (CNR) by a factor of 3.0 for 8 nm particles compared with conventional amorphous carbon films. The {111} lattice-fringe detection rate increases from 35% on carbon film to 85% on GO for the same particle size. These results establish GO as a quantitatively superior support film for high-resolution electron microscopy (HRTEM) imaging of sub-10 nm nanoparticles.
Single Crystal Cvd Diamond Plate
View Specs →
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →