The fundamental property of a crystal is its triple periodicity and a crystal may be generated by repeating a certain unit of pattern through the translations of a certain lattice called the *direct* lattice. The macroscopic geometric properties of a crystal are a direct consequence of the existence of this lattice on a microscopic scale. Let us for instance consider the natural faces of a crystal. These faces are parallel to sets of lattice planes. The lateral extension of these faces depends on the local physico-chemical conditions during growth but not on the geometric properties of the lattice. To describe the morphology of a crystal, the simplest way is to associate, with each set of lattice planes parallel to a natural face, a vector drawn from a given origin and normal to the corresponding lattice planes. To complete the description it suffices to give to each vector a length directly related to the spacing of the lattice planes. As we shall see in the next section this polar diagram is the geometric basis for the *reciprocal lattice* .

On the other hand, the basic tool to study a crystal is the diffraction of a wave with a wavelength of the same order of magnitude as that of the lattice spacings. The nature of the diffraction pattern is governed by the triple periodicity and the positions of the diffraction spots depend directly on the properties of the lattice. This operation transforms the *direct space* into an associated space, the *reciprocal space* , and we shall see that the diffraction spots of a crystal are associated with the nodes of its reciprocal lattice.

The reciprocal lattice is therefore an essential concept for the study of crystal lattices and their diffraction properties. This concept and the relation of the direct and reciprocal lattices through the Fourier transform was first introduced in crystallography by P. P. Ewald (1921).

Let **a**, **b**, **c** be the basic vectors defining the unit cell of the direct lattice. The basic vectors of the reciprocal lattice are defined by: ^{}

(2.1) |

The modulus of **a*** is equal to the ratio of the area of the face *OBCG* opposite to **a** to the volume of the cell built on the three vectors **a, b, c**. Referring to Fig. 1, we may write:

(2.2) |

From the definition of the reciprocal lattice vectors, we may therefore already draw the following conclusions:

(i) Each of the three vectors **a*, b*, c*** is normal to a set of lattice planes of the direct lattice (**b, c; c, b; a, b**) and their moduli are respectively equal to the inverse of the spacings of these three sets of lattice planes. The basic vectors of the reciprocal lattice possess therefore the properties that we were looking for in the introduction. We shall see in the next section that with each family of lattice planes of the direct lattice a reciprocal lattice vector may be thus associated.

(ii) The dimensions of the moduli of the reciprocal lattice vectors are those of the inverse of a length. For practical purposes the definition equations (2.1) may be rewritten after the introduction of a scale factor which has the dimension of an area:

(2.3) |

From relations 2.1 it can readily be shown that the two sets of basic vectors satisfy the following equations:

(2.4) |

The two sets of equations (2.1) and (2.4) are equivalent and equations (2.4) are sometimes used as the definition equations of the reciprocal lattice. These relations are symmetrical and show that the reciprocal lattice of the reciprocal lattice is the direct lattice.

(a) *with each node of the reciprocal lattice whose numerical coordinates have no common divider can be associated a set of direct lattice planes*

Let *M* be a reciprocal lattice point whose coordinates *h*, *k*, *l* have no common divider (*M* is the first node on the reciprocal lattice row **OM**), and *P* a point in direct space. We may write:

(2.5) |

Let us look for the locus of all points *P* of direct space such that the scalar product should be constant. It is a plane normal to *O* and passes through the projection *H* of *P* on **OM** (Fig. 2). Using 2.4, we find easily that the equation of this plane in direct space is given by

(2.6) |

Let us now assume that *P* is a node of the direct lattice:

The locus of *P* is a lattice plane of the direct lattice. Its equation is:

(2.7) |

Since all numbers in the left hand side are integers, we find that *C* is also an integer. With each value of *C* we may associate a lattice plane and thus generate a set of direct lattice planes which are all normal to the reciprocal vector **OM** (Fig. 3). The distance of one of these planes to the origin is given by:

The lattice planes have, as expected, an equal spacing:

(2.8) |

(2.9) |

*This is the fundamental relation of the reciprocal lattice which shows that with any node M of the reciprocal lattice whose numerical coordinates have no common divider we may associate a set of direct lattice planes normal to* **OM***. Their spacing is inversely proportional to the parameter along the reciprocal row * **OM**.

In order that the correspondence between direct and reciprocal lattice should be fully established, the converse of the preceding theorem should also be demonstrated. This will be done in paragraph 2.2(c).

It is interesting at this point to give an interpretation to the reciprocal lattice points whose numerical coordinates have a common divider. Let us consider such a point for which:

where Let *d*_{h1k1l1} be the spacing of the direct lattice planes associated with *M*. The fundamental law of the reciprocal lattice may be written:

We may also write:

(2.10) |

In other words, with the reciprocal lattice node *M* may be associated a set of fictitious planes in direct space whose spacing is *n* times smaller than the real lattice spacing. We shall see that in diffraction by crystal lattices a reciprocal lattice point may be associated with each Bragg diffraction: if the coordinates of this point have no common divider, Bragg's law is satisfied to the first order (2*d* sin = ); if they have a common divider, *n*, Bragg's law is satisfied to the *n*th order (2*d* sin = *n*), one may also say it is satisfied to the first order for the fictitious lattice planes of spacing *d*/*n* (2*d*/*n* sin = ) and this is what is actually always done in practice.

(b) *Miller indices*

Let us consider one particular lattice plane of equation

*hx* + *ky* + *lz* = *C*

We conclude that the lattice plane intercepts, along the three axes, lengths which are inversely proportional to three integers which have no common divider. This is the so-called *Law of Rational Indices* or Hauy Law. The three indices are called the Miller indices.

The planes which are crystallographically the most important ones are the densest ones, that is those with the largest spacing. Equation (2.9) tells us that they are associated with the shortest vectors in reciprocal lattice and that their Miller indices are therefore small. This is the reason why Hauy's law was also called the law of *simple* rational indices.

(c) *The reciprocal law: to each set of direct lattice planes corresponds a reciprocal lattice vector*

Let us consider a set of direct lattice planes of equation:

*hx* + *ky* + *lz* = *C*

Since *x*, *y*, *z* may be integers, *h*, *k*, and *l* are also integers. If *C* = 1, corresponding to the first plane in the family, *h*, *k* and *l* have no common divider. Let us now consider the reciprocal lattice vector

Its scalar products with the vectors **QR** and **RS** (Fig. 4) are respectively equal to:

They are both equal to zero, which shows that the reciprocal lattice vector is normal to the set of direct lattice planes; the scalar product of **ON**_{hkl} by **OP** where *P* is any direct lattice node in a plane of the set can be written in the form of equation (2.6). The reciprocal theorem is thus demonstrated.

Let us now call **e**_{i} the basic vectors of a vectorial space and *x*^{i} the coordinates of a given vector **x**. We may write:

(3.1) |

(3.2) | |

Quantities with a subscript transform in a change of coordinate like the basic vectors and are called *covariant* ; those with a superscript transform like the coordinates and are called *countervariant* . Let us now consider the scalar products:

(3.3) |

(3.4) |

x^{i} = x_{j}g^{ij} | (3.5) |

(3.6) |

Let us now introduce the following set of vectors

(3.7) |

This set of vectors constitutes a set of basic vectors. To show this we may simply transform equation (3.1):

(3.8) |

The vectors **e**^{j} constitute therefore a set of basic vectors and the *x*_{j} are the coordinates of **x** with respect to this base. They are called *countervariant* basic vectors. They are also identical to the basic vectors of the reciprocal space. This can easily be demonstrated by showing that they satisfy the basic relations (2.3) of the reciprocal space vectors. Let us consider the scalar products .Using (3.7), (3.4) and (3.6), we may write:

Let *V* be the volume of the unit cell. In a change of coordinate:

(3.9) |

(3.10) |

Let us now assume that the base **e**_{j} is orthonormal. There comes:

(3.11) |

We have then demonstrated the following general result:

(3.12) |

From (3.6) we know that

(3.13) |

It is easy to show the following relation, equivalent to (3.12):

(3.14) |

From (3.12), (3.13) and (3.14), we obtain finally:

Relation (3.7) is the most convenient one to use to compute the reciprocal lattice parameters or any quantity related to them. Let *a, b, c* and be the direct lattice parameters. The doubly covariant coefficients of the metric tensor are then:

(3.15) |

Its determinant, that is the square of the volume of the direct lattice unit cell is equal to:

(3.16) |

By inversing 3.15 we obtain the doubly contravariant of the metric tensor, *g*^{ij}

(3.17) |

Using (3.17), we can easily obtain the following relations:

(3.18) |

Let us consider a direct lattice vector

and a reciprocal lattice vectorUsing (2.3), their scalar product is equal to:

*hu* + *kv* + *lw*

Let us consider two direct lattice vectors:

Their vector product is equal to: Using the definition (2.1) of the basic reciprocal vectors, we may write:

(4.1) |

This shows that the vector product of two *direct* lattice vectors is easily expressed in terms of the basic *reciprocal* vectors.

The vector product of two vectors **n**_{1} and **n**_{2} respectively parallel to these two rows is normal to the set of lattice planes and is therefore parallel to the reciprocal lattice vector associated with the lattice planes. If *h, k, l* are its indices, we may therefore write:

(4.2) |

Let *h _{1}*,

(4.3) |

Let us consider a face-centered lattice. It is well known that the basic vectors **a**, **b**, **c**, of the elementary cell are given in terms of the vectors **a, b, c** of the face centered cell by (Fig. 5):

(4.4) |

In a similar way, the basic vectors **a**,**b**, **c** of the elementary cell of a body centered lattice are given in terms of the basic vectors of the multiple cell by (Fig. 6):

(4.5) |

Let us now look for the reciprocal lattice of the face-centered lattice. Its unit cell vectors are given by, using (2.1) and (4.4):

(4.6) |

Noting that the face-centered cell is of the fourth order, we find:

We may thus expressThis may also be written:

This relation shows that the reciprocal lattice of a face-centered lattice is a body centered lattice whose multiple cell is defined by 2**a***, 2**b***, 2**c***. If we index the reciprocal lattice defined by **a***, **b***, **c***, that is the reciprocal lattice of the multiple lattice defined by **a, b, c**, we find that only the nodes such that

Let us consider a plane monochromatic wave incident on a crystal and let **k**_{o} = s be its wave vector. Each scatterer will diffuse this wave in every direction with the same wavelength (coherent scattering). The total amplitude scattered in a particular direction **s**_{h} will be obtained by summing the amplitudes scattered in this particular direction by all scatterers, taking into account their phase relations. Let *A* and *B* be two homologous points in the structure, that is **AB** = **r** is a direct lattice vector. The phase differences between the waves scattered by *A* and *B* is equal to:

(5.1) |

There will be diffraction of the incident wave by the crystal if the wavelets diffracted by all homologous points are in phase, that is if is equal to an integer times 2 whatever the direct lattice vector **r**. The phase may also be written:

(5.2) |

The modulus of the diffusion vector has the dimension of the reciprocal of a length. **R** can therefore be expanded in reciprocal space:

The position vector **r** can in the same way be expressed in terms of its coordinates *u*, *v*, *w* in direct space. Applying relations (2.3), we may therefore write the phase difference in the following way:

(5.3) |

We may note that *u*, *v*, *w* being the coordinates of a direct lattice vector are integers. If is to be equal to an integer times 2 whatever *u*, *v*, *w*, we conclude that *h*, *k*, *l* are necessarily also equal to integers; in other words, *the diffusion vector is a reciprocal lattice vector* . This is the diffraction condition in reciprocal space. Bragg's law and the Ewald sphere construction are easily deduced from this result.

Let *O* be the origin of the reciprocal lattice and **IO** and **IH** vectors respectively equal to **s** and **s**.The vector **OH** is therefore equal to **R** (Fig. 7). If the diffraction condition is satisfied, *H* is a reciprocal lattice node. We have therefore the following construction: we draw through *O* a line parallel to the incident direction, let , then draw a sphere centered in *I* with radius 1/. If it passes through another reciprocal lattice node *H*, there is a reflected beam parallel to *IH*.

We may notice in the triangle *IOH* that *OH*/2 = *IH* sin ,calling the angle between *IO* or *IH* with the bissectrix of *OIH*, that is with the trace of the set of direct lattice planes associated with the node *H*.

We know from (2.8) that

where*A reciprocal lattice node may thus be associated with each Bragg reflection* .

This result can also be obtained directly through the properties of Fourier transforms. The basic assumption of the geometrical theory of diffraction is that the amplitude of the incident wave at each scatterer is constant. This assumption is acceptable if the interaction between the incident wave and the scatterers is small enough. The total diffracted amplitude in a given direction is therefore simply equal to the sum of the amplitudes scattered in this direction by every scatterer, taking into account their phase relationships. It is equal to:

(5.4) |

(5.5) |

- ...by:
- In this article the symbol is used to represent a vector product and commas (e.g.
**a, b, c**) to represent the triple product [which in this case is the volume of the unit cell].