Two-Dimensional Materials and Van der Waals Stacking

The layered crystal: strong within, weak between

Back in the bonding rung you learned that not every bond in a crystal has to be the same kind or the same strength. Graphite is the classic split personality. Inside each flat sheet, every carbon locks to three neighbours through strong, directional covalent bonds — a rigid honeycomb net. But between one sheet and the next there is only the feeble, non-directional van der Waals attraction, the weak stickiness that even neutral atoms feel. Because covalent bonds are directional and van der Waals is not, the crystal ends up wildly lopsided: thousands of times stiffer in-plane than out. That extreme structural anisotropy is the whole personality of a layered crystal, and the graphite structure is its poster child. It is also why a pencil writes — you are shearing off stacks of sheets onto the paper.

Put numbers on the lopsidedness. Within a graphite sheet the carbon–carbon bond is 1.42 angstrom; between sheets the layers sit 3.35 angstrom apart. The in-plane bond is a genuine covalent bond worth several eV; the interlayer glue is worth only a small fraction of that. The wide, weakly-bonded space between sheets even has a name — the van der Waals gap. And here is the leap that launches this whole guide: if the gap is so weak, why not peel off just one sheet? Do it, and you are left with something remarkable — a crystal that is a full, strong covalent net in two directions but only one atom thick in the third. That is a two-dimensional material.

LAYERED CRYSTAL  (graphite, side view)          bond regime

  layer A   C=C=C=C=C=C=C=C=C=C   <-- strong covalent, in-plane  d(C-C)=1.42 A
              . . . . . . . .      ~3.35 A van der Waals GAP (weak)
  layer B     C=C=C=C=C=C=C=C=C   <-- shifted: AB (Bernal) stacking
              . . . . . . . .      weak gap
  layer A   C=C=C=C=C=C=C=C=C=C

  strong WITHIN a sheet  vs  weak BETWEEN sheets  ->  extreme anisotropy
  peel off ONE sheet  ->  a 2-D crystal, one atom thick

  MoS2 monolayer is a 3-plane SANDWICH (not one plane):
      S  S  S  S      chalcogen
      Mo   Mo   Mo    metal        one layer ~ 6.5 A thick
      S  S  S  S      chalcogen

A layered crystal seen edge-on: strong covalent bonds run within each flat sheet, while only a weak van der Waals gap separates one sheet from the next — the source of extreme structural anisotropy. Peel off a single sheet and you have a two-dimensional crystal. Note a dichalcogenide monolayer is itself a three-plane chalcogen–metal–chalcogen sandwich, not a single atomic plane.

Graphene: a crystal one atom thick

Peel off exactly one graphite sheet and you have graphene — a single layer of carbon atoms in a honeycomb pattern. It is the purest possible answer to guide 1's question about the surface-to-volume ratio: graphene is all surface. Every atom sits on the outside; there is no interior to hide in. Its in-plane geometry is inherited straight from graphite — the C–C bond is 1.42 angstrom and the hexagons repeat with a lattice constant a of 2.46 angstrom — but now that net stands alone, a true membrane of matter one atom thick.

Here is a subtlety worth pausing on, because it is the whole lattice-is-not-the-crystal lesson made vivid. The honeycomb is NOT a Bravais lattice. Remember why: in a Bravais lattice every point has an identical surrounding, but the two carbons in a hexagon point in opposite directions — their neighbourhoods differ by a flip. So graphene is properly described as lattice plus motif: a simple triangular (hexagonal) Bravais lattice carrying a two-atom motif, the A and B carbons. The eye-catching honeycomb is the crystal you see; the underlying triangular lattice is where the motif gets stamped. Miss that split and you will mis-index its diffraction; keep it and graphene slots neatly into everything you learned in the lattice rung.

Beyond carbon: hBN and the dichalcogenide sandwich

Carbon is not alone. Hexagonal boron nitride (hBN) is graphene's insulating twin: the very same honeycomb, but now boron and nitrogen alternate around each hexagon. Because B and N are different, an inversion centre that graphene enjoys is broken, and hBN becomes a wide-gap insulator — often nicknamed "white graphene." Its lattice constant, about 2.50 angstrom, is within roughly 1.8 percent of graphene's 2.46 angstrom. That near-match makes hBN the flat, clean, electrically dead substrate on which the best graphene devices are built — and, as we will see, the small mismatch is not a nuisance but a design knob.

The richest family are the transition-metal dichalcogenides (TMDs), formula MX2 — molybdenum disulfide MoS2, tungsten diselenide WSe2, and dozens more. Their monolayer is not a single atomic plane but a three-plane sandwich: a plane of metal atoms filled between two planes of chalcogen (S–Mo–S), about 6.5 angstrom thick overall. How the metal is caged inside sets everything. In the 2H form the metal sits in a trigonal-prismatic cage; in the 1T form it sits in an octahedral one. Both give the metal a coordination number of 6, yet the geometry differs — a case of polytypism, the same chemistry stacked two different ways — and 2H MoS2 is a semiconductor while 1T MoS2 is a metal. Structure, once again, dictates property.

Thinning a TMD to a single layer is a size effect in the spirit of guide 2, and the payoff is spectacular. Bulk MoS2 has an indirect band gap of about 1.29 eV, but a single S–Mo–S sandwich has a direct gap near 1.8 to 1.9 eV — and a direct-gap crystal glows. That flip is quantum confinement in action: the monolayer is a natural quantum well, thin enough that squeezing the electrons in the third direction reshapes the whole electronic structure. Notice too that the band gap keeps changing with the number of layers — one, two, three — which quietly proves that even the "weak" van der Waals coupling is not zero. The sheets do talk to each other; they are just whispering.

Van der Waals stacking: how the sheets register

Now stack the sheets back up. Because the interlayer glue is only weak van der Waals, the sheets are almost free to choose how they sit on one another — the stacking sequence, the very idea you met with close-packed atoms, now scaled up to whole molecular sheets. Graphite's favourite is AB, or Bernal, stacking: every second sheet is shifted sideways so that half of its carbons sit directly over the hexagon centres of the sheet below. A rarer alternative is ABC rhombohedral stacking. This is precisely the ABAB-versus-ABCABC choice from the close-packing rung — polytypism again — except the repeating unit is a covalent sheet rather than a single atom, and the whole arrangement is van der Waals stacking.

Weak coupling has a delicious consequence: different stackings cost almost the same energy, so they can coexist in one crystal, and neighbouring sheets can even slide over one another — the lubricating trick behind graphite. But do not mistake "weak" for "irrelevant." How the sheets register still changes what the stack does: bilayer graphene behaves quite differently in AB stacking than when the two layers are given a twist. That last word — twist — is the door to the most exciting idea in the whole field, and it is where the graphite we started with turns from a passive material into an engineered one.

Atomic Lego: heterostructures and the moiré superstructure

Here is the payoff, and it is a beautiful contrast with guide 3. In ordinary epitaxy, to grow crystal B on crystal A their lattices must nearly match; any mismatch is paid for in strain and, past a critical thickness, in a network of misfit dislocations stitched into the interface. The registry is rigid, because chemical bonds reach across the join. Van der Waals materials tear up that rulebook. The interlayer bond is weak and non-directional, so you can lay any 2D crystal on any other regardless of lattice match — no strain to relieve, no misfit dislocations to nucleate. Stack graphene, then hBN, then MoS2, like sheets of atomic Lego, and you have built a van der Waals heterostructure: an engineered superlattice assembled by hand, layer by chosen layer.

And now the twist — literally. Overlay two nearly-identical lattices with a small rotation between them and you get a beat pattern, exactly like two window screens held at a slight angle: a new, much larger periodicity called a moiré superstructure. Its period follows a simple rule. For a small twist angle theta, the moiré wavelength is L approximately equal to a / theta, with theta in radians. Take graphene, a = 0.246 nm, and twist by 1.1 degrees (theta = 0.0192 rad): L is about 0.246 / 0.0192 = 12.8 nm — roughly 50 times the atomic spacing. A tiny lattice mismatch does the same job even with no twist: graphene on hBN, with delta about 1.8 percent, gives L approximately a / delta = 0.246 / 0.018 = about 14 nm. Either way, a giant new lattice appears out of near-coincidence.

Take two 2D layers whose lattices nearly match — the same material twisted, or two materials with a small size mismatch delta.
Overlay them across the van der Waals gap and rotate the top layer by a small angle theta.
Walk along the stack: at some spots the two layers' atoms line up (in register), a little further they drift out of register, then back in — a slow beat.
That beat repeats with a long period L approximately a / theta (twist) or a / delta (mismatch) — tens of times the atomic spacing.
That long period is a tunable superlattice the atoms never had alone; at special "magic" twist angles it dramatically reshapes the electrons.

Two honest cautions to carry up the ladder. First, a moiré is a real, long-period modulation, but it is generally incommensurate with the atomic lattice — the two periods need not fit in a whole-number ratio, so strictly a twisted stack may have no exact 3D translational periodicity at all, a cousin of the aperiodicity you met with quasicrystals. Second, "van der Waals" never meant "no interaction": near magic angles the layers actually relax toward good registry, and the moiré coupling can flatten the electronic bands enough to produce correlated insulators and superconductivity. Keep both idealizations honest — a 2D material is a few atoms thick, not truly flat, and van der Waals stacking is weak, not free. Next guide: roll a graphene sheet into a nanotube, and shrink a metal until it is almost all grain boundary.