Before the Nucleus: How Graphite Becomes Diamond

Graphite turning into diamond under high pressure and high temperature is a textbook phase transition - the common ground beneath both the formation of natural diamond and every synthesis technique humans have built. Yet what happens in the very first instant has been argued over for decades.

A collaboration between Jilin University, Qufu Normal University and Zhejiang University has now published Scale-dependent pathways of the graphite-to-diamond transition in Nature Communications, breaking the process into three stages: perturbation, nucleation and growth. Co-first authors are Wang Di, a doctoral student at Jilin University, and Dr. Zhao Bo of Qufu Normal University; corresponding authors are Professor Liu Xiaobing, Professor Li Quan and Academician Ma Yanming.

1. Why the argument lasted so long

Two reasons, both hard:

1. Nucleation is extraordinarily brief. The early structure spans only a few atomic layers and exists for picoseconds - almost impossible to preserve and identify experimentally;
2. The nucleation barrier is high. Conventional molecular dynamics cannot cover the full path within a feasible computation time.

So most earlier simulations took a shortcut: they pre-planted a diamond nucleus in the model, or introduced defects by hand. That skips the hardest stretch and starts the calculation from "a nucleus already exists." What gets skipped is precisely spontaneous nucleation.

2. Method: no preset answer, let the system find its own path

The team combined two tools:

• Well-tempered metadynamics (WTMetaD), which applies a bias potential to accelerate rare events into an observable time window;
• The ReaxFF reactive force field, which allows chemical bonds to form and break dynamically during the simulation.

The critical choice was the collective variable. Rather than something directional such as distance to a particular crystal plane, the team used the average coordination number of carbon atoms - simply how many neighbours each carbon has. It does not presuppose whether the product will be cubic or hexagonal diamond.

The system therefore finds its own route instead of evolving along a direction chosen by the researcher. That is what made spontaneous nucleation observable at all.

3. Three stages: the nucleus is an outcome, not a starting point

The simulated sequence runs against intuition.

The first step is not a diamond nucleus, but local buckling of graphite layers. Carbon atoms in adjacent layers are pulled closer and transient sp³ bonds form between layers. Those bonds break and re-form repeatedly, sitting in dynamic equilibrium - clusters that gather and disperse.

As the fluctuation accumulates, locally formed interlayer sp³ bonds connect into a network, the stacking sequence adjusts (the familiar ABA-to-ABC shift), and the structure finally solidifies into a diamond-like nucleus capable of further growth.

In one sentence: a persistent nucleus is the accumulated result of interlayer bond fluctuation, not the starting point of the transition. Which is exactly why simulations that pre-plant a nucleus never see the real first step.

4. Experiment: freeze the reaction and catch it

To catch this fleeting process in an electron microscope, the intermediate state has to live long enough to be seen. The team chose two clever precursors:

• Multi-walled carbon nanotubes (MWCNT): the tubular geometry acts as a constraint, limiting large radial deformation and "freezing" early structures - ideal for observing nucleation;
• Highly oriented pyrolytic graphite (HOPG): open space, well suited to observing later growth and grain-boundary adjustment.

The microscopy did capture diamond nuclei of 2–6 nm embedded in a disordered carbon matrix.

One detail matters more than the rest: nuclei below 4 nm carry pronounced lattice expansion strain, and only relax toward bulk diamond parameters as they grow. That is precisely the signature of a critical nucleus size - below it the nucleus cannot hold; above it, stable growth begins.

The lattice spacing makes the process legible: graphite's (002) spacing of about 3.32 Å in the measured precursor versus diamond's (111) spacing of about 2.04–2.07 Å. The layers are squeezed by nearly a third.

5. Two growth modes, and where the old dispute came from

Once the nucleus exists, two routes operate at the same time:

• Interlayer growth: new layers stack between existing sheets;
• Intralayer growth: the crystal extends outward within the plane of a sheet.

Which one dominates depends on the spatial scale of the precursor:

• when the sheet is confined in-plane, intralayer growth is suppressed and the system becomes interlayer-dominated;
• when sheets are long with few stacked layers, interlayer expansion is limited and the intralayer path dominates.

This explains why earlier studies contradicted each other: each model and precursor sat in a different scale window, so each saw a different path. Much of what looked like a dispute was a difference in scale.

The final stage holds one more step that is easy to overlook: grain-boundary adjustment. The paper documents dislocations, twin boundaries (TB), stacking faults (SF) and the formation and reorganisation of high-angle interfaces at roughly 43°, 39°, 37° and 25°. This is how diamond stitches together many small, misoriented crystalline regions into one body.

6. What this means, and what it does not

The value is in the method and in precursor design. If the path is set by scale and confinement, then the reverse also holds: by choosing the geometric dimensions and confinement conditions of the precursor, one can steer nucleation and growth, and thereby design diamond's microstructure. That has real guidance value for synthesis.

But it is not a breakthrough in low-pressure or low-energy synthesis. The work offers mechanistic understanding and a design framework, not a new industrial route, and the paper says so explicitly. This point has to be made clearly, or it will be misread as "diamond can now be made cheaply."

In one line

What this research fills in is the blank of the first few nanometres: diamond does not appear all at once - it is negotiated, bond by bond, out of interlayer fluctuations. And what it eventually becomes depends on how much room it has to grow.

About huicetools: a professional manufacturer of diamond tools, founded in 2006. With years of hands-on experience, we sell factory-direct at highly competitive prices. Our product line covers diamond segment, diamond wire saw, diamond saw blades, diamond grinding blocks, diamond drill bits and a full range of diamond tools for construction & Stone applications.

(Diamond Wall Saw Blade with Arix Segment made by Huicetools)

Disclaimer: This article is adapted and rewritten from publicly available reports on the Chinese Internet and open scientific literature, provided for industry reference only. Experimental methods, simulation parameters, observations and author details cited herein come from the original paper and public reports; specific data and conclusions rest with the original publication. We make no representation as to their completeness, accuracy or timeliness, and this article does not constitute investment or procurement advice. For republication or citation, please contact huicetools and credit the original source.