Symmetry in Crystals PDF

Symmetry in crystals CARMELO G I A C O V A Z Z O The crystalline state and isometric operations Matter is usually classified into three states: gaseous, liquid, and solid. Gases are composed of almost isolated particles, except for occasional collisions; they tend to occupy all the available volume...

Symmetry in crystals CARMELO G I A C O V A Z Z O The crystalline state and isometric operations Matter is usually classified into three states: gaseous, liquid, and solid. Gases are composed of almost isolated particles, except for occasional collisions; they tend to occupy all the available volume, which is subject to variation following changes in pressure. In liquids the attraction between nearest-neighbour particles is high enough to keep the particles almost in contact. As a consequence liquids can only be slightly compressed. The thermal motion has sufficient energy to move the molecules away from the attractive field of their neighbours; the particles are not linked together permanently, thus allowing liquids to flow. If we reduce the thermal motion of a liquid, the links between molecules will become more stable. The molecules will then cluster together to form what is macroscopically observed as a rigid body. They can assume a random disposition, but an ordered pattern is more likely because it corresponds to a lower energy state. This ordered disposition of molecules is called the crystalline state. As a consequence of our increased understand- ing of the structure of matter, it has become more convenient to classify matter into the three states: gaseous, liquid, and crystalline. Can we then conclude that all solid materials are crystalline? For instance, can common glass and calcite (calcium carbonate present in nature) both be considered as crystalline? Even though both materials have high hardness and are transparent to light, glass, but not calcite, breaks in a completely irregular way. This is due to the fact that glass is formed by long, randomly disposed macromolecules of silicon dioxide. When it is formed from the molten state (glass does not possess a definite melting point, but becomes progressively less fluid) the thermal energy which remains as the material is cooled does not allow the polymers to assume a regular pattern. This disordered disposition, characteristic of the liquid state, is therefore retained when the cooling is completed. Usually glasses are referred to as overcooled liquids, while non-fluid materials with a very high degree of disorder are known as amorphous solids. A distinctive property of the crystalline state is a regular repetition in the three-dimensional space of an object (as postulated as early as the end of the eighteenth century by R. J. Haiiy), made of molecules or groups of molecules, extending over a distance corresponding to thousands of molecular dimensions. However, a crystal necessarily has a number of defects at non-zero temperature and/or may contain impurities without losing its order. Furthermore: 1. Some crystals do not show three-dimensional periodicity because the 2 1 Carmelo Giacovazzo basic crystal periodicity is modulated by periodic distortions incom- mensurated with the basic periods (i.e. in incommensurately modulated structures, IMS). It has, however, been shown (p. 171 and Appendix 3.E) that IMSs are periodic in a suitable (3 + d)-dimensional space. 2. Some polymers only show a bi-dimensional order and most fibrous materials are ordered only along the fiber axis. 3. Some organic crystals, when conveniently heated, assume a state intermediate between solid and liquid, which is called the mesomorphic or liquid crystal state. These examples indicate that periodicity can be observed to a lesser or greater extent in crystals, depending on their nature and on the thermo- dynamic conditions of their formation. It is therefore useful to introduce the concept of a real crystal to stress the differences from an ideal crystal with perfect periodicity. Although non-ideality may sometimes be a problem, more often it is the cause of favourable properties which are widely used in materials science and in solid state physics. In this chapter the symmetry rules determining the formation of an ideal crystalline state are considered (the reader will find a deeper account in some papers devoted to the subject, or some exhaustive or in the theoretical sections of the International Tables for Cryst~llography).[~~ In order to understand the periodic and ordered nature of crystals it is necessary to know the operations by which the repetition of the basic molecular motif is obtained. An important step is achieved by answering the following question: given two identical objects, placed in random positions and orientations, which operations should be performed to superpose one object onto the other? The well known coexistence of enantiomeric molecules demands a second question: given two enantiomorphous (the term enantiomeric will only be used for molecules) objects, which are the operations required to superpose the two objects? An exhaustive answer to the two questions is given by the theory of isometric transformations, the basic concepts of which are described in Appendix 1.A, while here only its most useful results will be considered. Two objects are said to be congruent if to each point of one object corresponds a point of the other and if the distance between two points of one object is equal to the distance between the corresponding points of the other. As a consequence, the corresponding angles will also be equal in absolute value. In mathematics such a correspondence is called isometric. The congruence may either be direct or opposite, according to whether the corresponding angles have the same or opposite signs. If the congruence is direct, one object can be brought to coincide with the other by a convenient movement during which it behaves as a rigid body. The movement may be: (1) a translation, when all points of the object undergo an equal displacement in the same direction; (2) a rotation around an axis; all points on the axis will not change their position;  Symmetry in crystals ( 3 (3) a rototranslation or screw movement, which may be considered as the combination (product) of a rotation around the axis and a transl-ation along the axial direction (the order of the two operations may be exchanged). If the congruence is opposite, then one object will be said to be enantiomorphous with respect to the other. The two objects may be brought to coincidence by the following operations: (1) a symmetry operation with respect to a point, known as inversion; (2) a symmetry operation with respect to a plane, known as reflection; (3) the product of a rotation around an axis by an inversion with respect to a point on the axis; the operation is called rotoinversion; (4) the product of a reflection by a translation parallel to the reflection plane; the plane is then called a glide plane. (5) the product of a rotation by a reflection with respect to a plane perpendicular to the axis; the operation is called rotoreflection. Symmetry elements Suppose that the isometric operations described in the preceding section, not only bring to coincidence a couple of congruent objects, but act on the entire space. If all the properties of the space remain unchanged after a given operation has been carried out, the operation will be a symmetry operation. Symmetry elements are points, axes, or planes with respect to which symmetry operations are performed.. In the following these elements will be considered in more detail, while the description of translation operators will be treated in subsequent sections. Axes of rotational symmetry If all the properties of the space remain unchanged after a rotation of 2nIn around an axis, this will be called a symmetry axis of order n ; its written symbol is n. We will be mainly interested (cf. p. 9) in the axes 1, 2, 3, 4, 6. Axis 1 is trivial, since, after a rotation of 360" around whatever direction the space properties will always remain the same. The graphic symbols for the 2, 3, 4, 6 axes (called two-, three-, four-, sixfold axes) are shown in Table 1.1. In the first column of Fig. 1.1 their effects on the space are illustrated. In keeping with international notation, an object is represented by a circle, with a + or - sign next to it indicating whether it is above or below the page plane. There is no graphic symbol for the 1 axis. Note that a 4 axis is at the same time a 2 axis, and a 6 axis is at the same time a 2 and a 3 axis. 4 1 Carmelo Giacovazzo Table 1.1. Graphical symbols for symmetry elements: (a) axes normal to the pfane of projection; (b) axes 2 and 2, ,parallel to the plane of projection; (c) axes parallel or inclined to the plane of projection; (d) symmetry pfanes normar to the plane of projection; (e) symmetry planes parallel to the plane of projection Fig. 1.1. Arrangements of symmetry-equivalent objects as an effect of rotation, inversion, and screw axes.  Symmetry in crystals 1 5 Axes of rototranslation or screw axes A rototranslational symmetry axis will have an order n and a translational component t, if all the properties of the space remain unchanged after a 2nln rotation around the axis and the translation by t along the axis. On p. 10 we will see that in crystals only screw axes of order 1, 2, 3, 4, 6 can exist with appropriate translational components. Axes of inversion An inversion axis of order n is present when all the properties of the space remain unchanged after performing the product of a 2nln rotation around the axis by an inversion with respect to a point located on the same axis. The written symbol is fi (read 'minus n' or 'bar n'). As we shall see on p. 9 we will be mainly interested in 1, 2, 3, 4, 6 axes, and their graphic symbols are given in Table 1.1, while their effects on the space are represented in the second column of Fig. 1.1. According to international notation, if an object is represented by a circle, its enantiomorph is depicted by a circle with a comma inside. When the two enantiomorphous objects fall one on top of the other in the projection plane of the picture, they are represented by a single circle divided into two halves, one of which contains a comma. To each half the appropriate + or - sign is assigned. We may note that: (1) the direction of the i axis is irrelevant, since the operation coincides with an inversion with respect to a point; (2) the 2 axis is equivalent to a reflection plane perpendicular to it; the properties of the half-space on one side of the plane are identical to those of the other half-space after the reflection operation. The written symbol of this plane is m; (3) the 3 axis is equivalent to the product of a threefold rotation by an inversion: i.e. 3 = 31; (4) the 4 axis is also a 2 axis; (5) the 6 axis is equivalent to the product of a threefold rotation by a reflection with respect to a plane normal to it; this will be indicated by 6 = 3/m. Axes of rotoreflection A rotoreflection axis of order n is present when all the properties of the space do not change after performing the product of a 2nln rotation around an axis by a reflection with respect to a plane normal to it. The written symbol of this axis is fi. The effects on the space of the 1, 2, 3, 4, 6 axes coincide with those caused by an inversion axis (generally of a different order). In particular: i = m, 2 = 1, 3 = 6, 4 = 4, 6 = 3. From now on we will no longer consider the ii axes but their equivalent inversion axes. 6 1 Carmelo Giacovazzo Reflection planes with translational component (glide planes) A glide plane operator is present if the properties of the half-space on one side of the plane are identical to those of the other half-space after the product of a reflection with respect to the plane by a translation parallel to the plane. On p. 11 we shall see which are the glide planes found in crystals. Symmetry operations relating objects referred by direct congruence are called proper (we will also refer to proper symmetry axes) while those relating objects referred by opposite congruence are called improper (we will also refer to improper axes). Lattices Translational periodicity in crystals can be conveniently studied by con- sidering the geometry of the repetion rather than the properties of the motif which is repeated. If the motif is periodically repeated at intervals a, b, and /Cl /Cl /Cl ,,Cl /Cl ,,Cl ,Cl c along three non-coplanar directions, the repetition geometry can be fully o H , H? ?H ? H ? H? H? H described by a periodic sequence of points, separated by intervals a, b, c /C1 /Cl /Cl /Cl &C1 7 1 /Cl along the same three directions. This collection of points will be called a lattice. We will speak of line, plane, and space lattices, depending on H H H H whether the periodicity is observed in one direction, in a plane, or in a three-dimensional space. An example is illustrated in Fig. 1.2(a), where HOCl is a geometrical motif repeated at intervals a and b. If we replace the molecule with a point positioned at its centre of gravity, we obtain the lattice of Fig. 1.2(b). Note that, if instead of placing the lattice point at the centre of gravity, we locate it on the oxygen atom or on any other point of the motif, the lattice does not change. Therefore the position of the lattice with respect to the motif is completely arbitrary. If any lattice point is chosen as the origin of the lattice, the position of any other point in Fig. 1.2(b) is uniquely defined by the vector where u and v are positive or negative integers. The vectors a and b define a parallelogram which is called the unit cell: a and b are the basis vectors of the cell. The choice of the vectors a and b is rather arbitrary. In Fig. 1.2(b) four possible choices are shown; they are all characterized by the property that each lattice point satisfies relation (1.1) with integer u and v. Nevertheless we are allowed to choose different types of unit cells, such as those shown in Fig. 1.2(c), having double or triple area with respect to those selected in Fig. 1.2(b). In this case each lattice point will still satisfy (1.1) but u and v are no longer restricted to integer values. For instance, the point P is related to the origin 0 and to the basis vectors a' and b' through ( 4 v) = (112, 112). The different types of unit cells are better characterized by determining the number of lattice points belonging to them, taking into account that the ~ i ~ (a). R~~~~~~~~~ of a graphical motif as an points on sides and on corners are only partially shared by the given cell. example of a two-dimensional crystal; (b) The cells shown in Fig. 1.2(b) contain only one lattice point, since the corresponding lattice with some examples Of four points at the corners of each cell belong to it for only 114. These cells primitive cells; (c) corresponding lattice with some examples of multiple cells. are called primitive. The cells in Fig. 1.2(c) contain either two or three  Symmetry in crystals 1 7 points and are called multiple or centred cells. Several kinds of multiple cells are possible: i.e. double cells, triple cells, etc., depending on whether they contain two, three, etc. lattice points. The above considerations can be easily extended to linear and space lattices. For the latter in particular, given an origin 0 and three basis II vectors a , b, and c, each node is uniquely defined by the vector = ua + ub + W C. (1.2) The three basis vectors define a parallelepiped, called again a unit cell. a Fig. Notation for a unit cell. The directions specified by the vectors a , b, and c are the X , Y, Z crystallographic axes, respectively, while the angles between them are indicated by a, 0, and y, with a opposing a , opposing b, and y opposing c (cf. Fig. 1.3). The volume of the unit cell is given by where the symbol '.' indicates the scalar product and the symbol ' A ' the vector product. The orientation of the three crystallographic axes is usually chosen in such a way that an observer located along the positive direction of c sees a moving towards b by an anti-clockwise rotation. The faces of the unit cell facing a , b, and c are indicated by A, B, C, respectively. If the chosen cell is primitive, then the values of u, u, w in (1.2) are bound to be integer for all the lattice points. If the cell is multiple then u, u, w will have rational values. To characterize the cell we must recall that a lattice point at vertex belongs to it only for 1/8th, a point on a edge for 114, and one on a face for 112. The rational properties of lattices Since a lattice point can always be characterized by rational numbers, the lattice properties related to them are called rational. Directions defined by two lattice points will be called rational directions, and planes defined by three lattice points rational planes. Directions and planes of this type are also called crystallographic directions and planes. Crystallographic directions Since crystals are anisotropic, it is necessary to specify in a simple way directions (or planes) in which specific physical properties are observed. Two lattice points define a lattice row. In a lattice there are an infinite number of parallel rows (see Fig. 1.4): they are identical under lattice translation and in particular they have the same translation period. A lattice row defines a crystallographic direction. Suppose we have chosen a primitive unit cell. The two lattice vectors Q , and Q,,,,, ,, ,, with u, u, w, and n integer, define two different lattice points, but only one direction. This property may be used to characterize a direction in a unique way. For instance, the direction associated with the vector Q9,,,, can be uniquely defined by the vector Q,,,,,with no common factor among the indices. This direction will be indicated by the symbol [3 1 21, to be read as 'three, one, two' and not 'three hundred and twelve'. Fig. 1.4. Lattice rows and planes. 8 1 Carmelo Giacovazzo When the cell is not primitive u, v, w, and n will be rational numbers. Thus Q112,312,-113 and Q512,1512,-5,3 define the same direction. The indices of the former may therefore be factorized to obtain a common denominator and no common factor among the numerators: Q, ,-,,, = Q3,6,6,-,6 + [3 9 -21 to be read 'three', nine, minus two'. Crystallographic planes Three lattice points define a crystallographic plane. Suppose it intersects the three crystallographic axes X , Y , and Z at the three lattice points ( p , 0, 0 ) ) (0, q, 0 ) and (0, 0 , r ) with integer p, q, r (see Fig. 1.5). Suppose that m is the least common multiple of p, q , r. Then the equation of the plane is x'lpa + y'lqb + z'lrc = 1. If we introduce the fractional coordinates x = x ' l a , y = y ' l b , z = z l / c , the equation of the plane becomes xlp + y l q + z l r = 1. (1.3) Multiplying both sides by m we obtain hx + ky + lz = m (1.4) Fig. 1.5. Some lattice planes of the set (236). where h , k , and 1 are suitable integers, the largest common integer factor of which will be 1. We can therefore construct a family of planes parallel to the plane (1.4), by varying m over all integer numbers from -m to +m. These will also be crystallographic planes since each of them is bound to pass through at least three lattice points. The rational properties of all points being the same, there will be a plane of the family passing through each lattice point. For the same reason each lattice plane is identical to any other within the family through a lattice translation. Let us now show that (1.4) represents a plane at a distance from the origin m times the distance of the plane The intercepts of the plane (1.5) on X, Y , Z will be l l h , l l k and 111 respectively and those of (1.4) m l h , m l k , m l l. It is then clear that the distance of plane (1.4) from the origin is m times that of plane (1.5). The first plane of the family intersecting the axes X , Y , and Z at three lattice points is that characterized by a m value equal to the least common mutiple of h , k , I. We can therefore conclude that eqn (1.4) defines, as m is varied, a family of identical and equally spaced crystallographic planes. The three indices h , k , and 1 define the family and are its Miller indices. To indicate that a family of lattice planes is defined by a sequence of three integers, these are included within braces: ( h k I ). A simple interpretation of the three indices h, k , and I can be deduced from (1.4) and (1.5). In fact they  Symmetry in crystals 1 9 indicate that the planes of the family divide a in h parts, b in k parts, and c 0 k. (110) (010) in 1 parts. Crystallographic planes parallel to one of the three axes X, Y, or Z are defined by indices of type (Okl), (hol), or (hkO) respectively. Planes parallel to faces A, B, and C of the unit cell are of type (hOO), (OkO), and (001) respectively. Some examples of crystallographic planes are illustrated in Fig. 1.6. As a numerical example let us consider the plane (zio) which can be written as Fig. 1.6. Miller indices for some crystallographic planes parallel to Z ( Z i s supposed to be normal to the page). The first plane of the family with integer intersections on the three axes will be the 30th (30 being the least common multiple of 10, 15, and 6) and all the planes of the family can be obtained from the equation lox + 15y 62 = m,+ by varying m over all integers from -m to +m. We observe that if we divide p, q, and r in eqn (1.6) by their common integer factor we obtain + + x/3 y/2 z/5 = 1, from which Planes (1.7) and (1.8) belong to the same family. We conclude that a family of crystallographic planes is always uniquely defined by three indices h, k, and 1 having the largest common integer factor equal to unity. Symmetry restrictions due to the lattice periodicity and vice versa Suppose that the disposition of the molecules in a crystal is compatible with an n axis. As a consequence the disposition of lattice points must also be compatible with the same axis. Without losing generality, we will assume that n passes through the origin 0 of the lattice. Since each lattice point has identical rational properties, there will be an n axis passing through each and every lattice point, parallel to that passing through the origin. In particular each symmetry axis will lie along a row and will be perpendicular to a crystallographic plane. Let T be the period vector of a row passing through 0 and normal to n. We will then have lattice points (see Fig. 1.7(a)) at T, -T, T', and T". The vector T' - T" must also be a lattice vector and, being parallel to T, we will have T' - T" = mT where m is an integer value: in a scalar form 2 cos (2nIn) = m (m integer). (1.9) Equation (1.9) is only verified for n = 1, 2, 3, 4, 6. It is noteworthy that a 5 axis is not allowed, this being the reason why it is impossible to pave a room only with pentagonal tiles (see Fig. 1.7(b). (b) A unit cell, and therefore a lattice, compatible with an n axis will also be Fig. 1.7.(a) Lattice points in a plane normal to the compatible with an ii axis and vice versa. Thus axes I, 3, 3, 4, 6 will also be symmetry axis n passing through 0.(b) Regular allowed. pentagons cannot fill planar space. 10 1 Carmelo Giacovazzo Let us now consider the restrictions imposed by the periodic nature of crystals on the translational components t of a screw axis. Suppose that this lies along a row with period vector T. Its rotational component must correspond to n = 1, 2, 3, 4, 6. If we apply the translational component n times the resulting displacement will be nt. In order to maintain the periodicity of the crystal we must have nt = p T , with integer p , or For instance, for a screw axis of order 4 the allowed translational components will be (0/4)T, (1/4)T, (2/4)T, (3/4)T, (4/4)T, (5/4)T,...; of these only the first four will be distinct. It follows that: (1) in (1.10) p can be restricted within 0 s p < n ; (2) the n-fold axis may be thought as a special screw with t = 0. The nature of a screw axis is completely defined by the symbol n,. The graphic symbols are shown in Table 1.1: the effects of screw axes on the surrounding space are represented in Fig. 1.8. Note that: If we draw a helicoidal trajectory joining the centres of all the objects related by a 3, and by a 32 axis, we will obtain, in the first case a right-handed helix and in the second a left-handed one (the two helices are enantiomorphous). The same applies to the pairs 4, and 4,, 61 and 6,, and 6, and 6,. 4, is also a 2 axis, 6, is also a 2 and a 32, 64 is also a 2 and a 3,, and 63 is Fig. 1.8. Screw axes: arrangement of symmetry- also a 3 and a 2,. equivalent objects.  Symmetry in crystals I 11 We will now consider the restrictions imposed by the periodicity on the translation component t of a glide plane. If we apply this operation twice, the resulting movement must correspond to a translation equal to pT, where p may be any integer and T any lattice vector on the crystallographic plane on which the glide lies. Therefore 2t = p T , i.e. t = ( p / 2 ) T.As p varies over all integer values, the following translations are obtained OT, (1/2)T, (2/2)T, (3/2)T,... of which only the first two are distinct. For p = 0 the glide plane reduces to a mirror m. We will indicate by a, b, c axial glides with translational components equal to a / 2 , b / 2 , c / 2 respectively, by n the + diagonal glides with translational components ( a b ) / 2 or ( a + c ) / 2 or + ( b c ) / 2 or ( a + b + c ) / 2. In a non-primitive cell the condition 2t = p T still holds, but now T is a lattice vector with rational components indicated by the symbol d. The graphic symbols for glide planes are given in Table 1.1. Point groups a n d s y m m e t r y classes In crystals more symmetry axes, both proper and improper, with or without translational components, may coexist. We will consider here only those combinations of operators which do not imply translations, i.e. the combinations of proper and improper axes intersecting in a point. These are called point groups, since the operators form a mathematical group and leave one point fixed. The set of crystals having the same point group is called crystal class and its symbol is that of the point group. Often point group and crystal class are used as synonyms, even if that is not correct in principle. The total number of crystallographic point groups (for three- dimensional crystals) is 32, and they were first listed by Hessel in 1830. The simplest combinations of symmetry operators are those characterized by the presence of only one axis, which can be a proper axis or an inversion one. Also, a proper and an inversion axis may be simultaneously present. The 13 independent combinations of this type are described in Table 1.2. When along the same axis a proper axis and an inversion axis are simultaneously present, the symbol n/ri is used. Classes coinciding with other classes already quoted in the table are enclosed in brackets. The problem of the coexistence of more than one axis all passing by a common point was first solved by Euler and is illustrated, with a different approach, in Appendix 1.B. Here we only give the essential results. Let us suppose that there are two proper axes I , and l2 intersecting in 0 (see Fig. 1.9). The I , axis will repeat in Q an object originally in P, while 1, will Table 1.2. Single-axis crystallographic point groups Proper axis Improper axis Proper and improper axis 1 = i) (i[i - 1 2 = m- 212-= 2jm 3 3 31 (3[3 = 3) 4 4 414 = 4/m 6 6 = 3/m 616 = 6/m 5 + 5 + 3 = 13  t.' 12 ( Carmelo Giacovazzo 11 Table 1.3. For each combination of symmetry axes the minimum angles between axes are given. For each angle the types of symmetry axes are quoted in parentheses Combination of cu (ded B (ded Y (ded.- P '.-- symmetry axes 2 3 2 2 2 2 90 90 (22) (2 3) 90 (2 2) 90 (2 3) 90 60 (22) (2 2) ---0 4 2 2 90 (2 4) 90 (2 4) 45 (2 2) 6 2 2 90 (2 6) 90 (2 6) 30 (2 2) 0 2 3 3 54 44'08" (2 3) 54 44'08" (2 3) 70 31 '44" (3 3) Fig. 1.9. Arrangement of equivalent objects 4 3 2 35 15'52" (2 3) 45 (2 4) 54 44'08" (4 3) around two intersecting symmetry axes. repeat in R the object in Q. P and Q are therefore directly congruent and this implies the existence of another proper operator which repeats the object in P directly in R. The only allowed combinations are n22, 233, 432, 532 which in crystals reduce to 222, 322, 422, 622, 233, 432. For these combinations the smallest angles between the axes are listed in Table 1.3, while their disposition in the space is shown in Fig. 1.10. Note that the combination 233 is also consistent with a tetrahedral symmetry and 432 with a cubic and octahedral symmetry. Suppose now that in Fig. 1.9 1, is a proper axis while 1, is an inversion one. Then the objects in P and in Q will be directly congruent, while the object in R is enantiomorphic with respect to them. Therefore the third operator relating R to P will be an inversion axis. We may conclude that if one of the three symmetry operators is an inversion axis also another must be an inversion one. In Table 1.4 are listed all the point groups characterized by combinations of type PPP, PII, IPI, IIP (P=proper, I = improper), while in Table 1.5 the classes with axes at the same time proper and improper are given. In the two tables the combinations coinciding with previously considered ones are closed within brackets. The Fig. 1.10. Arrangement of proper symmetry axes for six point groups.  Symmetry in crystals 1 13 Table 1.4. Crystallographic point groups with more than one axis PPP PI1 I P I IIP (43 2 ---- 432 rnlm Table 1.5. Crystallographic point groups with more than one axis, each axis being proper and improper simultaneously results so far described can be easily derived by recalling that: If two of the three axes are symmetry equivalent, they can not be one proper and one improper; for example, the threefold axes in 233 are symmetry referred by twofold axes, while binary axes in 422 differing by 45" are not symmetry equivalent. If an even-order axis and a ? axis (or an m plane) coexist, there will also be an m plane (or a ? axis) normal to the axis and passing through the intersection point. Conversely, if m and ? coexist, there will also be a 2 axis passing through ? and normal to m. In Tables 1.2, 1.4, and 1.5 the symbol of each point group does not reveal all the symmetry elements present: for instance, the complete list of symmetry elements in the class 2/m33 is 2/m 2/m 2/m?3333. On the other hand, the symbol 2/m% is too extensive, since only two symmetry operators are independent. In Table 1.6 are listed the conventional symbols used for the 32 symmetry classes. It may be noted that crystals with inversion symmetry operators have an equal number of 'left' and 'right' moieties; these parts, when considered separately, are one the enantiomorph of the other. The conclusions reached so far do not exclude the possibility of crystal- lizing molecules with a molecular symmetry different from that of the 32 point groups (for instance with a 5 axis). In any case the symmetry of the crystal will belong to one of them. To help the reader, some molecules and their point symmetry are shown in Fig. 1.11. It is very important to understand how the symmetry of the physical properties of a crystal relates to its point group (this subject is more extensively described in Chapter 9). Of basic relevance to this is a postulate  E E E E E E E E E Ekkbmm.-2E341mmo E E E E E E E EEE E EkEnmm. - 2 E 9 3 ~ ~ m wE E  Symmetry in crystals 1 15 2. The variation of the refractive index of the crystal with the vibration direction of a plane-polarized light wave is represented by the optical indicatrix (see p. 607). This is in general a three-axis ellipsoid: thus the lowest symmetry of the property 'refraction' is 2/m 2lm 2/m, the point group of the ellipsoid. In crystal classes belonging to tetragonal, trigonal, or hexagonal systems (see Table 1.6) the shape of the indicatrix is a rotational ellipsoid (the axis is parallel to the main symmetry axis), and in symmetry classes belonging to the cubic system the shape of the indicatrix is a sphere. For example, in the case of tourmaline, with point group 3m, the ellipsoid is a revolution around the threefold axis, showing a symmetry higher than that of the point group. We shall now see how it is possible to guess about the point group of a crystal through some of its physical properties: 1. The morphology of a crystal tends to conform to its point group symmetry. From a morphological point of view, a crystal is a solid body bounded by plane natural surfaces, the faces. The set of symmetry- equivalent faces constitutes a form: the form is open if it does not enclose space, otherwise it is closed. A crystal form is named according to the number of its faces and to their nature. Thus a pedion is a single face, a pinacoid is a pair of parallel faces, a sphenoid is a pair of faces related by a diad axis, a prism a set of equivalent faces parallel to a common axis, a pyramid is a set of planes with equal angles of inclination to a common axis, etc. The morphology of different samples of the same compound can show different types of face, with different extensions, and different numbers of edges, the external form depending not only on the structure but also on the chemical and physical properties of the environment. For instance, galena crystals (PbS, point group m3m) tend to assume a cubic, cube-octahedral, or octahedral habit (Fig. 1.12(a)). Sodium chloride grows as cubic crystals from neutral aqueous solution and as octahedral from active solutions (in the latter case cations and anions play a different energetic role). But at the same temperature crystals will all have constant dihedral angles between corresponding faces (J. B. L. Rome' de l'Ile, 1736-1790). This property, the observation of which dates back to N. Steno (1669) and D. Guglielmini (1688), can be explained easily, following R. J. Haiiy (1743-1822), by considering that faces coincide with lattice planes and edges with lattice rows. Accordingly, Miller indices can be used as form symbols, enclosed in braces: {hkl). The indices of well-developed faces on natural crystals tend to have small values of h, k, 1, (integers greater than six are rarely involved). Such faces correspond to lattice planes with a high density of lattice points per unit area, or equivalently, with large intercepts alh, blk, cll on the reference axes (Bravais' law). An important extension of this law is obtained if space group symmetry (see p. 22) is taken into account: screw axes and glide planes normal to a given crystal face reduce its importance (Donnay-Harker principle). The origin within the crystal is usually chosen so that faces (hkl) and (hit) are parallel faces an opposite sides of the crystal. In Fig. 1.13 some idealized crystal forms are shown. (b) The orientation of the faces is more important than their extension. The Fig. 1.12. (a) Crystals showing cubic or cube- orientations can be represented by the set of unit vectors normal to them. octahedral or octahedral habitus, (b) crystal with This set will tend to assume the point-group symmetry of the given crystal a sixfold symmetry axis. 16 1 Carmelo Giacovazzo independently of the morphological aspect of the samples. Thus, each sample of Fig. 1.12(a) shows an m3m symmetry, and the sample in Fig. 1.12(b) shows a sixfold symmetry if the normals to the faces are considered instead of their extensions. The morphological analysis of a crystalline sample may be used to get some, although not conclusive, indication, of its point-group symmetry. 2. Electrical charges of opposite signs-may appear at the two hands of a polar axis of a crystal subject to compression, because of the piezoelectric effect (see p. 619). A polar axis is a rational direction which is not symmetry equivalent to its opposite direction. It then follows that a polar direction can only exist in the 21 non-centrosymmetric point groups (the only exception is the 432 class, where piezoelectricity can not occur). In these groups not all directions are polar: in particular a direction normal to an even-order axis or to a mirror plane will never be polar. For instance, in quartz crystals (SOz, class 32), charges of opposite sign may appear at the opposite hands Fig. 1.13. Some simple crystal forms: (a) cinnabar, HgS, class 32; (b) arsenopyritf, FeAsS, of the twofold axes, but not at those of the threefold axis. class mmm; (c) ilmenite, FeTiO,, class 3; (d) gypsum, CaSO,, class 2/m. 3. A point group is said to be polar if a polar direction, with no other symmetry equivalent directions, is allowed. Along this direction a per- manent electric dipole may be measured, which varies with temperature (pyroelectric effect, see p. 606). The ten polar classes are: 1, 2, m, mm2, 4, 4mm, 6, 6mm, 3, 3m. Piezo- and pyroelectricity tests are often used to exclude the presence of an inversion centre. Nevertheless when these effects are not detectable, no definitive conclusion may be drawn. 4. Ferroelectric crystals show a permanent dipole moment which can be changed by application of an electric field. Thus they can only belong to one of the ten polar classes. 5. The symmetry of a crystal containing only one enantiomer of an optically active molecule must belong to one of the 11 point groups which do not contain inversion axes. 6. Because of non-linear optical susceptibility, light waves passing through non-centrosymmetric crystals induce additional waves of frequency twice the incident frequency. This phenomenon is described by a third-rank tensor, as the piezoelectric tensor (see p. 608): it occurs in all non- centrosymmetric groups except 432, and is very efficientL7]for testing the absence of an inversion centre. 7. Etch figures produced on the crystal faces by chemical attack reveal the face symmetry (one of the following 10 two-dimensional point groups). Point groups in one and two dimensions The derivation of the crystallographic point groups in a two-dimensional space is much easier than in three dimensions. In fact the reflection with respect to a plane is substituted by a reflection with respect to a line (the same letter m will also indicate this operation); and ii axes are not used. The total number of point groups in the plane is 10, and these are indicated by the symbols: 1, 2, 3, 4, 6, m, 2mm, 3m, 4mm, 6mm. The number of crystallographic point groups in one dimension is 2: they are 1 and m = (I).  Symmetry in crystals 1 17 The Laue classes In agreement with Neumann's principle, physical experiments do not normally reveal the true symmetry of the crystal: some of them, for example diffraction, show the symmetry one would obtain by adding an inversion centre to the symmetry elements actually present. In particular this happens when the measured quantities do not depend on the atomic positions, but rather on the interatomic vectors, which indeed form a centrosymmetric set. Point groups differing only by the presence of an inversion centre will not be differentiated by these experiments. When these groups are collected in classes they form the 11 Laue classes listed in Table 1.6. The seven crystal systems If the crystal periodicity is only compatible with rotation or inversion axes of order 1, 2, 3, 4, 6, the presence of one of these axes will impose some restrictions on the geometry of the lattice. It is therefore convenient to group together the symmetry classes with common features in such a way that crystals belonging to these classes can be described by unit cells of the same type. In turn, the cells will be chosen in the most suitable way to show the symmetry actually present. Point groups 1 and i have no symmetry axes and therefore no constraint axes for the unit cell; the ratios a:b:c and the angles a , P, y can assume any value. Classes 1 and are said to belong to the triclinic system. Groups 2, m, and 2/m all present a 2 axis. If we assume that this axis coincides with the b axis of the unit cell, a and c can be chosen on the lattice plane normal to b. We will then have a = y = 90" and P unrestricted and the ratios a:b:c also unrestricted. Crystals with symmetry 2, m, and 2/m belong to the monoclinic system. Classes 222, mm2, mmm are characterized by the presence of three mutually orthqgonal twofold rotation or inversion axes. If we assume these as reference axes, we will obtain a unit cell with angles a = P = y = 90" and with unrestricted a:b:c ratios. These classes belong to the orthorhombic system. For the seven groups with only one fourfold axis [4,4,4/m, 422,4mm, 42m, 4/mmm] the c axis is chosen as the direction of the fourfold axis and the a and b axes will be symmetry equivalent, on the lattice plane normal to c. The cell angles will be a = P = y = 90" and the ratios a:b:c = 1:l:c. These crystals belong to the tetragonal system. For the crystals with only one threefold or sixfold axis [3, 3, 32, 3m, 3m, 6, 6, 6/m, 622, 6mm, 62m, 6/mm] the c axis is assumed along the three- or sixfold axis, while a and b are symmetry equivalent on the plane perpendicular to c. These point groups are collected together in the trigonal and hexagonal systems, respectively, both characterized by a unit cell with angles a = /3 = 90" and y = 120°, and ratios a:b :c = 1:1:c. Crystals with four threefold axes [23, m3, 432, 43m, m3m] distributed as the diagonals of a cube can be referred to orthogonal unit axes coinciding with the cube edges. The presence of the threefold axes ensures that these directions are symmetry equivalent. The chosen unit cell will have a = P = y = 90' and ratios a :b:c = 1:1:1. This is called the cubic system. 18 1 Carmelo Giacovazzo The Bravais lattices In the previous section to each crystal system we have associated a primitive cell compatible with the point groups belonging to the system. Each of these primitive cells defines a lattice type. There are also other types of lattices, based on non-primitive cells, which can not be related to the previous ones. In particular we will consider as different two lattice types which can not be described by the same unit-cell type. In this section we shall describe the five possible plane lattices and fourteen possible space lattices based both on primitive and non-primitive cells. These are called Bravais lattices, after Auguste Bravais who first listed them in 1850. Plane lattices An oblique cell (see Fig. 1.14(a)) is compatible with the presence of axes 1 or 2 normal to the cell. This cell is primitive and has point group 2. If the row indicated by m in Fig. 1.14(b) is a reflection line, the cell must be rectangular. Note that the unit cell is primitive and compatible with the point groups m and 2mm. Also the lattice illustrated in Fig. 1.14(c) with a = b and y # 90" is compatible with m. This plane lattice has an oblique primitive cell. Nevertheless, each of the lattice points has a 2mm symmetry and therefore the lattice must be compatible with a rectangular system. This can be seen by choosing the rectangular centred cell defined by the unit vectors a' and b'. This orthogonal cell is more convenient because a simpler coordinate system is allowed. It is worth noting that the two lattices shown in Figs. 1.14(b) and 1.14(c) are of different type even though they are compatible with the same point groups. In Fig. 1.14(d) a plane lattice is represented compatible with the presence of a fourfold axis. The cell is primitive and compatible with the point groups 4 and 4mm. In Fig. 1.14(e) a plane lattice compatible with the presence of a three- or a sixfold axis is shown. A unit cell with a rhombus shape and angles of 60" and 120" (also called hexagonal) may be chosen. A centred rectangular cell can also be selected, but such a cell is seldom chosen. (b) m ; 2 m m (c) m;2mm I I i i I Fig. 1.14. The five plane lattices and the corresponding two-dimensional point groups. (d) 4;4mm  Symmetry in crystals 1 19 Table 1.7. The five plane lattices - -- Cell Type of cell Point group Lattice parameters of the net Oblique P 2 a, b, y Rectangular P, C 2mm a, b, y = 90" Square P 4mm a = b, y = 90" Hexagonal P 6mm a=b,y=120" The basic features of the five lattices are listed in Table 1.7 Space lattices In Table 1.8 the most useful types of cells are described. Their fairly limited number can be explained by the following (or similar) observations: A cell with two centred faces must be of type F. In fact a cell which is at the same time A and B, must have lattice points at (0,1/2,1/2) and (1/2,0, 112). When these two lattice translations are applied one after the other they will generate a lattice point also at (1/2,1/2,0); A cell which is at the same time body and face centred can always be reduced to a conventional centred cell. For instance an I and A cell will have lattice points at positions (1/2,1/2,1/2) and (0,1/2,1/2): a lattice point at (1/2,0,0) will then also be present. The lattice can then be described by a new A cell with axes a ' = a/2, b' = b, and c' = c (Fig. 1.15). It is worth noting that the positions of the additional lattice points in Table 1.8 define the minimal translational components which will move an object into an equivalent one. For instance, in an A-type cell, an object at + ( x , y, z) is repeated by translation into ( x , y + m/2, z n/2) with m and n integers: the shortest translation will be (0,1/2,1/2). Let us now examine the different types of three-dimensional lattices grouped in the appropriate crystal systems. J Fig. 1.15. Reduction of an I- and A-centred cell Table 1.8. The conventional types of unit cell to an A-centred cell. Symbol Type Positions of Number additional of lattice lattice points points per cell P primitive - 1 I body centred (112,1/2, l r 2 ) 2 A A-face centred (0,1/2,1/2) 2 B B-face centred (1/2,0,1/2) 2 C C-face centred (1/2,1/2,0) 2 F All faces centred (112,112, O), (1/2,0,1/2) 2 ~0,112,112~ 4 R Rhombohedrally (1/3,2/3,2/3), (2/3,1/3,1/3) 3 centred (de scription with 'hexagonal axes') 20 ( Carmelo Giacovazzo Triclinic lattices Even though non-primitive cells can always be chosen, the absence of axes with order greater than one suggests the choice of a conventional primitive cell with unrestricted a , p, y angles and a:b:c ratios. In fact, any triclinic lattice can always be referred to such a cell. Monoclinic lattices The conventional monoclinic cell has the twofold axis parallel to b, angles a = y = 90", unrestricted p and a :b:c ratios. A B-centred monoclinic cell with unit vectors a, b, c is shown in Fig. 1.16(a). If we choose a' = a , + b' = b, c' = ( a c ) / 2 a primitive cell is obtained. Since c' lies on the (a, c) plane, the new cell will still be monoclinic. Therefore a lattice with a B-type monoclinic cell can always be reduced to a lattice with a P monoclinic cell. An I cell with axes a, b, c is illustrated in Fig. 1.16(b). If we choose a' = a, b' = b, c' = a + c, the corresponding cell becomes an A monoclinic cell. Therefore a lattice with an I monoclinic cell may always be described by an A monoclinic cell. Furthermore, since the a and c axes can always be interchanged, an A cell can be always reduced to a C cell. An F cell with axes a, b, c is shown in Fig. 1.16(c). When choosing + a' = a, b' = b, c' = ( a c ) / 2 a type-C monoclinic cell is obtained. There- fore, also, a lattice described by an F monoclinic cell can always be described by a C monoclinic cell. We will now show that there is a lattice with a C monoclinic cell which is not amenable to a lattice having a P monoclinic cell. In Fig. 1.16(d) a C cell with axes a, b, c is illustrated. A primitive cell is obtained by assuming + a' = ( a + b ) / 2 , b' = ( - a b ) / 2 , c' = c, but this no longer shows the features of a monoclinic cell, since y' # 90°, a' = b' # c ' , and the 2 axis lies along the diagonal of a face. It can then be concluded that there are two distinct monoclinic lattices, described by P and C cells, and not amenable one to the other. Orthorhombic lattices In the conventional orthorhombic cell the three proper or inversion axes are parallel to the unit vectors a, b, c, with angles a = /3 = y = 90" and general a:b:c ratios. With arguments similar to those used for monoclinic lattices, the reader can easily verify that there are four types of orthorhombic lattices, P, C, I, and F. Tetragonal lattices In the conventional tetragonal cell the fourfold axis is chosen along c with a = p = y = 90°, a = b, and unrestricted c value. It can be easily verified that because of the fourfold symmetry an A cell will always be at the same time a B cell and therefore an F cell. The latter is then amenable to a tetragonal I cell. A C cell is always amenable to another tetragonal P cell. Thus only two different tetragonal lattices, P and I, are found. Fig. 1.16. Monoclinic lattices: (a) reduction of a Cubic lattices B-centred cell to a P cell; (b) reduction of an In the conventional cubic cell the four threefold axes are chosen to be I-centred to an A-centred cell; (c) reduction of an F-centred to a C-centred cell; (d) reduction of a parallel to the principal diagonals of a cube, while the unit vectors a, b, c C-centred to a P non-monoclinic cell. are parallel to the cube edges. Because of symmetry a type-A (or B or C)  Symmetry in crystals 1 21 cell is also an F cell. There are three cubic lattices, P, I, and F which are not amenable one to the other. Hexagonal lattices In the conventional hexagonal cell the sixfold axis is parallel to c, with a = b, unrestricted c, a = /3 = 90") and y = 120". P is the only type of hexagonal Bravais lattice. Trigonal lattices As for the hexagonal cell, in the conventional trigonal cell the threefold axis is chosen parallel to c, with a = b, unrestricted c, a = /3 = 90°, and y = 120". Centred cells are easily amenable to the conventional P trigonal cell. Because of the presence of a threefold axis some lattices can exist which may be described via a P cell of rhombohedral shape, with unit vectors a R , bR, CR such that aR= bR = cR, aR= PR = YR, and the threefold axis along the UR + bR + CR direction (see Fig. 1.17). Such lattices may also be described by three triple hexagonal cells with basis vectors UH, bH, CH defined according to[61 These hexagonal cells are said to be in obverse setting. Three further triple hexagonal cells, said to be in reverse setting, can be obtained by changing a H and bH to -aH and -bH. The hexagonal cells in obverse setting have centring points (see again Fig. 1.17)) at (O,O, O), I ,I ,I , (113,213,213) while for reverse setting centring points are at It is worth noting that a rhombohedral description of a hexagonal P lattice is always possible. Six triple rhombohedral cells with basis vectors a;, bk, Fig. 1.17. Rhombohedra1 lattice. The basis of the rhombohedral cell is labelled a, b,, c,, the basis of the hexagonal centred cell is labelled a, b,, c, (numerical fractions are calculated in terms of the c, axis). (a) Obverse setting; (b) the same figure as in (a) projected along c,. 22 1 Carmelo Giacovazzo ck can be obtained from a H , bH, CH by choosing: U ~ = U H + C H , bk=b,+ cH, ck=-(aH+bH)+cH a k = -aH + CH, bk = -bH + CH, ck = a H+ bH + cH and cyclic permutations of a;, bk, ck. Each triple rhomobohedral cell will have centring points at (O,0, O), (113,1/3,1/3), (213,213,213). In conclusion, some trigonal lattices may be described by a hexagonal P cell, others by a triple hexagonal cell. In the first case the nodes lying on the different planes normal to the threefold axis will lie exactly one on top of the other, in the second case lattice planes are translated one with respect to the other in such a way that the nth plane will superpose on the (n + 3)th plane (this explains why a rhombohedral lattice is not compatible with a sixfold axis). When, for crystals belonging to the hexagonal or trigonal systems, a hexagonal cell is chosen, then on the plane defined by a and b there will be a third axis equivalent to them. The family of planes (hkl) (see Fig. 1.18) divides the positive side of a in h parts and the positive side of b in k parts. If the third axis (say d) on the (a, b) plane is divided in i parts we can a introduce an extra index in the symbol of the family, i.e. (hkil). From the same figure it can be seen that the negative side of d is divided in h k + Fig. 1.18. Intersections of the set of crystallographic planes ( h k l ) with the three parts, and then i = -(h +k). For instance (1 2 -3 5), (3 -5 2 I), symmetry-equivalent a, b, daxes in trigonal and (-2 0 2 3) represent three plane families in the new notation. The hexagonal systems. four-index symbol is useful to display the symmetry, since (hkil), (kihl), and (ihkl) are symmetry equivalent planes. Also, lattice directions can be indicated by the four-index notation. Following pp. 7-8, a direction in the (a, b) plane is defined by a vector + (P - 0)= ma nb. If we introduce the third axis d in the plane, we can + write (P - 0)= ma + nb Od. Since a decrease (or increase) of the three coordinates by the same amount j does not change the point P, this may be represented by the coordinates: u = m - j, v = n - j, i = -j. + If we choose j = (m n)/3, then u = (2m - n)/3, v = (2n - m)/3, i = -(m + n)/3. In conclusion the direction [mnw] may be represented in the new notation as [uviw], with i = -(u + v). On the contrary, if a direction is already represented in the four-index notation [uviw], to pass to the three-index one, -i should be added to the first three indices in order to bring to zero the third index, i.e. [u - i v - i w]. A last remark concerns the point symmetry of a lattice. There are seven three-dimensional lattice point groups, they are called holohedries and are listed in Table 1.6 (note that 3m is the point symmetry of the rhombohedral lattice). In two dmensions four holohedries exist: 2, 2mm, 4mm, 6mm. The 14 Bravais lattices are illustrated in Fig. 1.19 by means of their conventional unit cells (see Appendix 1.C for a different type of cell). A detailed description of the metric properties of crystal lattices will be given in Chapter 2. The space groups A crystallographic space group is the set of geometrical symmetry opera- tions that take a three-dimensional periodic object (say a crystal) into itself.  Symmetry in crystals 1 23 Triclinic Cubic Trigonal Fig. 1.19. The 14 three-dimensional Bravais Hexagonal lattices. The total number of crystallographic space groups is 230. They were first derived at the end of the last century by the mathematicians Fedorov (1890) and Schoenflies (1891) and are listed in Table 1.9. In Fedorov's mathematical treatment each space group is represented by a set of three equations: such an approach enabled Fedorov to list all the space groups (he rejected, however, five space groups as impossible: Fdd2, Fddd, 143d, P4,32, P4132). The Schoenflies approach was most practical and is described briefly in the following. On pp. 11-16 we saw that 32 combinations of either simple rotation or inversion axes are compatible with the periodic nature of crystals. By combining the 32 point groups with the 14 Bravais lattices (i.e. P, I, F,...) one obtains only 73 (symmorphic) space groups. The others may be obtained by introducing a further variation: the proper or improper symmetry axes are replaced by screw axes of the same order and mirror planes by glide planes. Note, however, that when such combinations have more than one axis, the restriction that all symmetry elements must intersect in a point no longer applies (cf. Appendix l.B). As a consequence of the presence of symmetry elements, several symmetry-equivalent objects will coexist within the unit cell. We will call the smallest part of the unit cell which will generate the whole cell when applying to it the symmetry 24 1 Carmelo Giacovazzo Table 1.9. The 230 three-dimensional space groups arranged by crystal systems and point groups. Space groups (and enantiomorphous pairs) that are uniquely deter- minable from the symmetry of the diffraction pattern and from systematic absences (see p. 159) are shown in bold-type. Point groups without inversion centres or mirror planes are emphasized by boxes Crystal Point Space system group groups Triclinic [i3 p1 i P1 Monoclinic P2, P2,, C2 m Pm, PC, Cm, Cc 2/m P2/m, P2,/m, C2/m, P2/c, P2,/c, C2/c Orthorhombic 12221 P222, P222,, P2,2,2, P2,2,2,, C222,, C222, F222, 1222, 12,2121 mm2 Pmm2, PmcP,, Pcc2, PmaP,, PcaS,, PncZ,, PmnZ,, Pba2, Pna2,, Pnn2, Cmm2, Cmc2,, Ccc2, Amm2, Abm2, Ama2, Aba2, Fmm2, Fdd2,lmm2, lba2, h a 2 mmm Pmmm,Pnnn,Pccm,Pban,Pmma,Pnna,Pmna,Pcca, Pbam, Pccn, Pbcm, Pnnm, Pmmn, Pbcn, Pbca, Pnma, Cmcm, Cmca, Cmmm, Cccm, Cmma, Ccca, Fmmm, Fddd, Immm, Ibam, Ibca, lmma Tetragonal p4, p41, p4,, p4, 14, 14, 4 P4. 14 Cubic (231 P23, F23, 123, P2,3,_12,3 - ~ m 3~, dFm3,, F a , lm3, ~ a 3 I& , PG2, Pa&?, F$32, Y. 2 , 1432, P?,32, P4,32, l4?32 P43-m, F43m, 143m, P43n, F43c, 1Gd m3m Pm_3m, Pn3n, PrnBn, Pn3m, Fm3m, Fm&, F b m , F&c, lm3m, la3d operations an asymmetric unit. The asymmetric unit is not usually uniquely defined and can be chosen with some degree of freedom. It is nevertheless obvious that when rotation or inversion axes are present, they must lie at the borders of the asymmetric unit.  Symmetry in crystals 1 25 According to the international (Hermann-Mauguin) notation, the space- group symbol consists of a letter indicating the centring type of the conventional cell, followed by a set of characters indicating the symmetry elements. Such a set is organized according to the following rules: 1. For triclinic groups: no symmetry directions are needed. Only two space groups exist: PI and PI. 2. For monoclinic groups: only one symbol is needed, giving the nature of the unique dyad axis (proper and/or inversion). Two settings are used: y-axis unique, z-axis unique. 3. For orthorhombic groups: dyads (proper and/or of inversion) are given along x, y, and z axis in the order. Thus Pca2, means: primitive cell, glide plane of type c normal to x-axis, glide plane of type a normal to the y-axis, twofold screw axis along z. 4. For tetragonal groups: first the tetrad (proper and/or of inversion) axis along z is specified, then the dyad (proper and/or of inversion) along x is given, and after that the dyad along [I101 is specified. For example, P4,lnbc denotes a space group with primitive cell, a 4 sub 2 screw axis along z to which a diagonal glide plane is perpendicular, an axial glide plane b normal to the x axis, an axial glide plane c normal to. Because of the tetragonal symmetry, there is no need to specify symmetry along the y-axis. 5. For trigonal and hexagonal groups: the triad or hexad (proper and/or of inversion) along the z-axis is first given, then the dyad (proper and/or of inversion) along x and after that the dyad (proper and/or of inversion) along [1?0] is specified. For example, P6,mc has primitive cell, a sixfold screw axis 6 sub 3 along z, a reflection plane normal to x and an axial glide plane c normal to [ ~ I o ]. 6. For cubic groups: dyads or tetrads (proper and/or of inversion) along x , followed by triads (proper and/or of inversion) along [ I l l ] and dyads (proper and/or of inversion) along. We note that: 1. The combination of the Bravais lattices with symmetry elements with no translational components yields the 73 so-called symmorphic space groups. Examples are: P222, Cmm2, F23, etc. 2. The 230 space groups include 11 enantiomorphous pairs: P3, (P3,), P3,12 (P3212), P3,21 (P3,21), P41 (P43), P4J2 (P4322), P4,&2 (P4&?,2), P6i (P65), P6, (P64), P6,22 (P6522), P6222 (P6422), P4,32 (P4,32). The ( + ) isomer of an optically active molecule crystallizes in one of the two enantiomorphous space groups, the ( - ) isomer will crystallize in the other. 3. Biological molecules are enantiomorphous and will then crystallize in space groups with no inversion centres or mirror planes; there are 65 groups of this type (see Table 1.9). 4. The point group to which the space group belongs is easily obtained from the space-group symbol by omitting the lattice symbol and by replacing 26 1 Carmelo Giacovazzo the screw axes and the glide planes with their corresponding symmorphic symmetry elements. For instance, the space groups P4Jmmc, P4/ncc, 14,lacd, all belong to the point group 4lmmm. 5. The frequency of the different space groups is not uniform. Organic compounds tend to crystallize in the space groups that permit close packing of triaxial ellipsoids.[81According to this view, rotation axes and reflection planes can be considered as rigid scaffolding which make more difficult the comfortable accommodation of molecules, while screw axes and glide planes, when present, make it easier because they shift the molecules away from each other. Mighell and Rodgers examined 21 051 organic compounds of known crystal structure; 95% of them had a symmetry not higher than orthorhom- bic. In particular 35% belonged to the space group P2,/c, 13.3% to PI, 12.4% to P2,2,2,, 7.6% to P2, and 6.9% to C21c. A more recent study by ~ i l s o n , [ ' ~based ] on a survey of the 54599 substances stored in the Cambridge Structural Database (in January 1987), confirmed Mighell and Rodgers' results and suggested a possible model to estimate the number Nsg of structures in each space group of a given crystal class: Nsg = Acc exp { -BccE21sg - Ccclmls,) where A,, is the total number of structures in the crystal class, ,, is the number of twofold axes, [m],, the number of reflexion planes in the cell, B,, and Cc, are parameters characteristic of the crystal class in question. The same results cannot be applied to inorganic compounds, where ionic bonds are usually present. Indeed most of the 11641 inorganic compounds considered by Mighell and Rodgers crystallize in space groups with orthorhombic or higher symmetry. In order of decreasing frequency we have: Fm3m, Fd3m, P6Jmmc, P2,/c, ~ m 3 m ~, 3 m C2/m,, C2/c,.... The standard compilation of the plane and of the three-dimensional space groups is contained in volume A of the International Tables for Crystallog- raphy. For each space groups the Tables include (see Figs 1.20 and 1.21). 1. At the first line: the short international (Hermann-Mauguin) and the Schoenflies symbols for the space groups, the point group symbol, the crystal system. 2. At the second line: the sequential number of the plane or space group, the full international (Hermann-Mauguin) symbol, the Patterson symmetry (see Chapter 5, p. 327). Short and full symbols differ only for the monoclinic space groups and for space groups with point group mmm, 4/mmm, 3m, 6/mmm, m3, m3m. While in the short symbols symmetry planes are suppressed as much as possible, in the full symbols axes and planes are listed for each direction. 3. Two types of space group diagrams (as orthogonal projections along a cell axis) are given: one shows the position of a set of symmetrically equivalent points, the other illustrates the arrangement of the symmetry elements. Close to the graphical symbols of a symmetry plane or axis parallel to the projection plane the 'height' h (as a fraction of the shortest lattice translation normal to the projection plane) is printed. If h = 0 the height is omitted. Symmetry elements at h also occur at height h + 112.  Symmetry in crystals 1 27 4. Information is given about: setting (if necessary), origin, asymmetric unit, symmetry operations, symmetry generators (see Appendix l.E) selected to generate all symmetrical equivalent points described in block 'Positions'. The origin of the cell for centrosymmetric space groups is usually chosen on an inversion centre. A second description is given if points of high site symmetry not coincident with the inversion centre occur. For example, for ~ n 3 ntwo descriptions are available, the first with origin at 432, and the second with origin at 3. For non-centrosymmetric space groups the origin is chosen at a point of highest symmetry (e.g. the origin for ~ 4 2 c is chosen at 4lc) or at a point which is conveniently placed with respect to the symmetry elements. For example, on the screw axis in P2,, on the glide plane in PC, at la2, in P ~ a 2 at ~ ,a point which is surrounded symmetrically by the three 2, axis in P2,2,2,. 5. The block positions (called also Wyckoff positions) contains the general position (a set of symmetrically equivalent points, each point of which is left invariant only by application of an identity operation) and a list of special positions (a set of symmetrically equivalent points is in special position if each point is left invariant by at least two symmetry operations of the space group). The first three block columns give information about multiplicity (number of equivalent points per unit cell), Wyckoff letter (a code scheme starting with a at the bottom position and continuing upwards in alphabetical order), site symmetry (the group of symmetry operations which leaves invariant the site). The symbol adoptedr9]for describing the site symmetry displays the same sequence of symmetry directions as the space group symbol. A dot marks those directions which do not contribute any element to the site symmetry. To each Wyckoff position a reflection condition, limiting possible reflections, may be associated. The condition may be general (it is obeyed irrespective of which Wyckoff positions are occupied by atoms (see Chapter 3, p. 159) or special (it limits the contribution to the structure factor of the atoms located at that Wyckoff position). 6. Symmetry of special projections. Three orthogonal projections for each space group are listed: for each of them the projection direction, the Hermann-Mauguin symbol of the resulting plane group, and the relation between the basis vectors of the plane group and the basis vectors of the space group, are given, together with the location of the plane group with respect to the unit cell of the space group. 7. Information about maximal subgroups and minimal supergroups (see Appendix l.E) is given. In Figs. 1.20 and 1.21 descriptions of the space groups Pbcn and P4222 are respectively given as compiled in the International Tables for Crystallog- raphy. In order to obtain space group diagrams the reader should perform the following operations: 1. Some or all the symmetry elements are traced as indicated in the space-group symbol. This is often a trivial task, but in certain cases special care must be taken. For example, the three twofold screw axes do not intersect each other in P2,2,2,, but two of them do in P2,2,2 (see Appendix 1.B).