What Is a Crystal?

Learning Objectives

After completing this section you will be able to:

1. Define what makes a material crystalline and distinguish crystals from amorphous solids using both structural and experimental criteria.
2. Decompose any crystal into its two building blocks — a lattice and a basis — and explain why this decomposition is unique and powerful.
3. Write the lattice-vector equation $\mathbf{R} = n_1\mathbf{a}_1 + n_2\mathbf{a}_2 + n_3\mathbf{a}_3$ and explain every symbol's physical meaning.
4. Identify the experimental signatures of crystallinity in X-ray, electron, and neutron diffraction patterns.
5. Explain why translational periodicity enables Bloch's theorem and makes plane-wave DFT codes like VASP computationally feasible.
6. Connect crystallographic descriptions to modern materials informatics: DFT input files, crystal-structure databases, and machine-learning representations.

The Big Picture: From Mineral Cabinets to Quantum Computers

The idea that solids are built from repeating units is surprisingly old. In 1784, the French mineralogist René Just Haüy dropped a calcite crystal, noticed that every fragment broke into perfect rhombohedra, and proposed that all crystals are built by stacking identical "molécules intégrantes" — tiny building blocks repeated in space. He could not see atoms, but his geometric insight was correct.

Over the next century, this idea matured:

• Auguste Bravais (1850) proved mathematically that only 14 fundamentally different lattice types exist in three dimensions, classifying all possible "repetition rules."
• Evgraf Fedorov and Arthur Schoenflies (1891) independently enumerated the 230 space groups — the complete catalog of symmetries a periodic 3D pattern can possess.
• Max von Laue (1912) directed X-rays at a copper sulfate crystal and observed sharp diffraction spots, proving for the first time that crystals really are periodic on the atomic scale and that X-rays are waves.
• William Henry and William Lawrence Bragg (1913) solved the first crystal structures (NaCl, diamond) from diffraction data, launching modern materials science.

The central idea: A crystal is not just "a pretty rock." It is matter with long-range translational order — a precise mathematical symmetry that determines how it diffracts radiation, conducts electrons, vibrates thermally, responds to stress, and interacts with light. Understanding that order is the key to predicting and engineering material properties from first principles.

What Defines a Crystal?

A crystal is a solid whose atoms (or ions, or molecules) are arranged in a pattern that repeats periodically in three dimensions. This periodicity means that if you translate the entire structure by certain specific vectors, it maps exactly onto itself.

The Three Pillars of Crystallinity

1. Translation invariance: there exist three non-coplanar vectors $\mathbf{a}_1, \mathbf{a}_2, \mathbf{a}_3$ such that shifting the structure by any integer combination of these vectors leaves it unchanged.
2. Complete tiling: the repeating unit (the unit cell) fills all of three-dimensional space without gaps or overlaps, like bricks in a wall.
3. Long-range order: knowing the arrangement in one unit cell lets you predict the arrangement arbitrarily far away. This is what distinguishes a crystal from a liquid or a glass.

The Formal Statement

A structure possesses translational periodicity if there exists a set of translation vectors $\{\mathbf{R}\}$ such that the atomic density $\rho(\mathbf{r})$ satisfies:

Here $\rho(\mathbf{r})$ is the electron density (or, equivalently, the probability of finding an atom) at position $\mathbf{r}$. The set of all valid translation vectors $\mathbf{R}$ forms a lattice, which we will define precisely below.

The Wallpaper Analogy

Before jumping to three dimensions, build your intuition with a two-dimensional analogy: wallpaper. A wallpaper pattern has a small motif — a flower, a geometric shape — repeated at regular intervals across a flat surface. If you know the motif and the two repeat vectors, you can reconstruct the entire wall.

A crystal works the same way, extended into three dimensions. There is a small repeating unit (a group of atoms) that tiles all of 3D space without gaps or overlaps. The repeat rule is the lattice, and the motif is the basis.

The 17 wallpaper groups classify every possible 2D periodic pattern. They are the two-dimensional analogues of the 230 space groups that classify all possible 3D crystal symmetries (Section 7 of this chapter). The jump from 2D to 3D adds complexity, but the underlying principle is identical: classify all distinct ways a motif can be repeated with translational symmetry.

Lattice + Basis = Crystal

The single most important concept in this entire textbook is the decomposition:

The symbol $\otimes$ here means "place a copy of the basis at every lattice point." This is a convolution: the crystal is generated by convolving the lattice (an infinite set of mathematical points) with the basis (a finite set of atoms).

The Lattice

A Bravais lattice is the set of all points that can be reached from any one point by integer translations along three linearly independent vectors $\mathbf{a}_1, \mathbf{a}_2, \mathbf{a}_3$. The general lattice point is:

Symbol-by-Symbol Explanation

What does this equation say in plain language? Starting from any lattice point, you can reach every other lattice point by taking $n_1$ steps along $\mathbf{a}_1$, then $n_2$ steps along $\mathbf{a}_2$, then $n_3$ steps along $\mathbf{a}_3$. The lattice is the set of all points you can reach this way.

Geometric picture: Think of the lattice vectors as the edges of a parallelipiped (a "squished box"). Stacking copies of this box in all three directions fills all of space — that is the lattice.

Key Properties

• Discrete: the points are isolated, not continuous. There is a minimum distance between neighbors.
• Infinite: the lattice extends to infinity in all directions (in VASP, periodic boundary conditions simulate this).
• Identical environment: every lattice point has exactly the same surroundings. This is the defining property of a Bravais lattice.
• Non-unique choice: the lattice vectors $\mathbf{a}_1, \mathbf{a}_2, \mathbf{a}_3$ are not unique. There are infinitely many valid choices that generate the same lattice. We typically choose the primitive vectors (smallest cell) or the conventional vectors (highest symmetry cell).

The Unit Cell

The unit cell is the region of space that, when translated by all lattice vectors, fills all of space exactly once. The primitive cell is the smallest such region (it contains exactly one lattice point). The volume of the primitive cell is:

This is the scalar triple product of the three lattice vectors — geometrically, the volume of the parallelepiped they define.

The Basis

The basis specifies the physical content of each unit cell: which atoms are present and where they sit. It is a list of $N$ atoms, each described by:

1. Species: the chemical element (e.g., Cd, Se, Mn, O).
2. Position: expressed in fractional coordinates relative to the lattice vectors.

If the $j$-th basis atom has fractional coordinates $(x_j, y_j, z_j)$, its Cartesian position within the unit cell at lattice site $\mathbf{R}$ is:

Why fractional coordinates? They are dimensionless and lattice-independent. If you change the lattice constant (e.g., under pressure), the fractional coordinates stay the same while the Cartesian positions scale accordingly. This makes them the natural choice for crystal structure databases and VASP input files.

Simple vs Complex Bases

The basis can range from a single atom (elemental metals) to hundreds of thousands of atoms (protein crystals). The lattice + basis decomposition works for all of them.

Interactive: Build a Crystal

Use the interactive visualization below to explore how the lattice + basis decomposition works. Select a crystal type, then step through the three stages: first the abstract lattice, then the basis atoms in one unit cell, and finally the full crystal.

Examples Across Materials and Experiments

The lattice + basis framework is not just a textbook abstraction. It is the language used every day by experimentalists and computational scientists across physics, chemistry, and engineering.

From Simple Metals to Complex Oxides

The Same Framework Everywhere

Whether you are indexing powder diffraction peaks in a geology lab, designing a new solar cell absorber, or running a VASP calculation on a supercomputer, you always start the same way: specify the lattice vectors and the basis. The framework is universal.

Example: Zinc Blende CdSe

Cadmium selenide in the zinc blende structure is the prototype material for this textbook. Let us dissect it completely using the lattice + basis framework.

Structure Description

• Lattice: Face-centered cubic (FCC) with lattice constant $a = 6.052$ Å.
• Basis: Two atoms — Cd at $(0, 0, 0)$ and Se at $(\tfrac{1}{4}, \tfrac{1}{4}, \tfrac{1}{4})$ in fractional coordinates.
• Coordination: Each Cd is tetrahedrally surrounded by 4 Se atoms, and each Se by 4 Cd. The nearest-neighbor Cd–Se distance is $d = \frac{a\sqrt{3}}{4} \approx 2.62$ Å.
• Space group: $F\bar{4}3m$ (number 216).

VASP POSCAR File

The POSCAR file is the first input VASP reads. It encodes the lattice vectors and atomic positions — exactly our lattice + basis decomposition. Click on any line to see a detailed explanation.

1CdSe zinc blende
26.052
3  1.000000  0.000000  0.000000
4  0.000000  1.000000  0.000000
5  0.000000  0.000000  1.000000
6Cd Se
74  4
8Direct
9  0.000  0.000  0.000   ! Cd
10  0.000  0.500  0.500   ! Cd
11  0.500  0.000  0.500   ! Cd
12  0.500  0.500  0.000   ! Cd
13  0.250  0.250  0.250   ! Se
14  0.250  0.750  0.750   ! Se
15  0.750  0.250  0.750   ! Se
16  0.750  0.750  0.250   ! Se

Experimental Evidence: How Do We Know?

How do we know that crystals are really periodic? The answer is diffraction: when waves (X-rays, electrons, or neutrons) scatter from a periodic structure, they produce sharp, discrete peaks at specific angles. This is the experimental fingerprint of crystallinity.

X-Ray Diffraction (XRD)

The workhorse of crystallography. X-ray wavelengths (~1 Å) are comparable to interatomic distances, so crystals act as natural diffraction gratings. The positions of the diffraction peaks give the lattice parameters; the intensities give information about the basis (which atoms are where). For CdSe zinc blende, the strongest peaks appear at $2\theta$ angles corresponding to the (111), (220), and (311) Miller planes.

Electron Diffraction (TEM/SAED)

In a transmission electron microscope, a focused electron beam produces a selected-area electron diffraction (SAED) pattern — a 2D array of spots that is a direct image of the reciprocal lattice (Chapter 3). Each spot corresponds to a set of crystal planes satisfying the Bragg condition. Single-crystal SAED patterns are used to determine crystal orientation and identify unknown phases.

Neutron Diffraction

Neutrons scatter from atomic nuclei (not electrons), making them sensitive to light atoms (hydrogen, lithium) and to magnetic order. Neutron diffraction is essential for determining the positions of light atoms in metal hydrides and the magnetic structure of materials like MnO.

Crystalline vs Amorphous

Not all solids are crystals. Amorphous materials (glasses, many polymers, amorphous silicon) lack long-range translational order. They may have short-range order — nearest-neighbor bond lengths and angles are well-defined — but no periodicity beyond a few atomic distances.

Use the interactive below to see how introducing atomic disorder transforms a perfect crystal into an amorphous solid. Pay attention to how the simulated diffraction pattern changes.

Why Periodicity Enables Computation

The central promise of crystallography for computational physics is this: because a crystal is periodic, we only need to solve the quantum-mechanical equations for one unit cell — not for the ~$10^{23}$ atoms in a macroscopic sample. This is not just a convenience; it is the difference between possible and impossible.

Bloch's Theorem (Preview)

In a periodic potential, the solutions to the Schrödinger equation take a special form known as Bloch states:

What does Bloch's theorem say in plain language? Because the crystal potential is periodic, the wave function at any point in the crystal can be reconstructed from the wave function in just one unit cell, multiplied by a plane-wave phase $e^{i\mathbf{k} \cdot \mathbf{r}}$. Different choices of $\mathbf{k}$ probe different wavelengths of electronic oscillation through the crystal.

From One Cell to Infinity: The VASP Workflow

1. Define the unit cell in the POSCAR file (lattice vectors + basis).
2. VASP imposes periodic boundary conditions: the cell is replicated infinitely in all three directions.
3. Expand wave functions in a plane-wave basis set up to an energy cutoff (set in INCAR via ENCUT).
4. Solve the Kohn–Sham equations at each $\mathbf{k}$-point on a discrete grid (the KPOINTS file).
5. Integrate over $\mathbf{k}$-space to obtain total energy, forces, band structure, density of states, and other properties.

Modern Computing & Materials Informatics

Crystallography is not a relic of the 19th century. It is the natural language of periodic matter, and as such it appears at the foundation of every modern computational and data-driven approach to materials science.

DFT and Periodic Boundary Conditions

All major DFT codes (VASP, Quantum ESPRESSO, ABINIT, CASTEP) assume periodicity. The lattice vectors define the simulation box; the basis defines the atomic content. When you run a DFT calculation, you are exploiting the crystal structure at every step:

• $\mathbf{k}$-space sampling: the Brillouin zone (the unit cell of the reciprocal lattice) is sampled with a discrete grid. More symmetry in the crystal means fewer $\mathbf{k}$-points are needed — symmetry reduces computational cost.
• Plane-wave basis: the periodic part of Bloch states is expanded in plane waves, which are natural eigenfunctions of the translation operator.
• Symmetry reduction: space-group operations reduce the irreducible Brillouin zone, cutting the number of independent $\mathbf{k}$-points by up to 48× for cubic crystals.

Crystal Structure Databases

The crystallographic description (lattice + basis) is how all structure databases store materials:

Machine Learning for Materials

Modern ML approaches to materials discovery rely heavily on crystallographic representations:

• Graph neural networks (GNNs) such as CGCNN, MEGNet, and M3GNet represent crystals as graphs where nodes are atoms and edges encode distances and lattice periodicity. The lattice vectors are essential for correctly handling periodic boundary conditions.
• Symmetry-aware descriptors like SOAP, Smooth Overlap of Atomic Positions, encode local atomic environments in a way that respects the translational and rotational symmetry of the crystal.
• Crystal structure prediction algorithms (USPEX, CALYPSO, AIRSS) explore the space of possible lattice + basis combinations to find new stable phases.
• Generative models (DiffCSP, CDVAE) learn to generate new crystal structures by operating directly on lattice parameters and fractional coordinates — exactly the quantities defined in this section.

Computational Exploration: Building a Crystal in Python

Let us put theory into practice. The Python code below uses the Atomic Simulation Environment (ASE) to build the zinc blende CdSe crystal, inspect its structure, create a supercell, and export it in formats ready for VASP and crystallographic databases. Click on any line to see a detailed explanation.

1from ase.build import bulk
2from ase.visualize import view
3from ase.io import write
4
5# Build zinc blende CdSe from its space group
6cdse = bulk('CdSe', crystalstructure='zincblende', a=6.052)
7
8# Inspect the unit cell
9print("Lattice vectors (angstrom):")
10print(cdse.cell[:])
11
12print("\nAtom positions (fractional):")
13for atom in cdse:
14    frac = cdse.cell.scaled_positions(atom.position)
15    print(f"  {atom.symbol}: ({frac[0]:.3f}, {frac[1]:.3f}, {frac[2]:.3f})")
16
17print(f"\nNumber of atoms: {len(cdse)}")
18print(f"Space group: F-43m (216)")
19
20# Create a 2x2x2 supercell for visualization
21supercell = cdse.repeat((2, 2, 2))
22print(f"Supercell atoms: {len(supercell)}")
23
24# Export to VASP POSCAR format
25write('POSCAR', cdse, format='vasp')
26
27# Export to CIF for database submission
28write('CdSe_zincblende.cif', cdse)

Summary

In this section we established the foundational vocabulary of crystallography:

• A crystal is a solid with long-range translational order. Its atomic density satisfies $\rho(\mathbf{r} + \mathbf{R}) = \rho(\mathbf{r})$ for all lattice translations $\mathbf{R}$.
• Every crystal decomposes uniquely into a lattice (abstract grid) and a basis (atoms per grid point): Crystal = Lattice $\otimes$ Basis.
• The lattice is generated by three primitive vectors: $\mathbf{R} = n_1 \mathbf{a}_1 + n_2 \mathbf{a}_2 + n_3 \mathbf{a}_3$, where $n_1, n_2, n_3 \in \mathbb{Z}$.
• The basis specifies atom types and fractional coordinates within the unit cell.
• Zinc blende CdSe is an FCC lattice ($a = 6.052$ Å) with a two-atom basis: Cd at $(0,0,0)$ and Se at $(\tfrac{1}{4}, \tfrac{1}{4}, \tfrac{1}{4})$.
• Experimentally, crystallinity is detected by sharp diffraction peaks (XRD, electron, neutron). Amorphous materials show broad halos.
• Periodicity enables Bloch's theorem, which reduces the infinite crystal to a finite computation — the foundation of all VASP and plane-wave DFT calculations.
• Modern materials informatics (databases, ML, crystal structure prediction) all operate on the lattice + basis representation defined here.

In the next section, we will make the lattice concrete by defining lattice vectors, unit cells, and the six parameters $(a, b, c, \alpha, \beta, \gamma)$ that fully characterize the geometry of any lattice.