SQU Journal for Science, 2016, 21(2), 139-149 © 2016 Sultan Qaboos University 139 Symmetry of the Pyritohedron and Lattices Nazife O. Koca*, Aida Y. Al-Mukhaini, Mehmet Koca, and Amal J. Al-Qanobi Department of Physics, College of Science, Sultan Qaboos University, P.O. Box 36, PC 123 Al-Khoud, Muscat, Sultanate of Oman. *Email: nazife@squ.edu.om. ABSTRACT: The pyritohedron consisting of twelve identical but non regular pentagonal faces and its dual pseudoicosahedron that possess the pyritohedral (Th) symmetry play an essential role in understanding the crystallographic structures with the pyritohedral symmetry. The pyritohedral symmetry takes a simpler form in terms of quaternionic representation. We discuss the 3D crystals with the pyritohedral symmetry which can be derived from the Coxeter-Dynkin diagram of D3. Keywords: Pseudoicosahedron; Pyritohedron; Lattice; Coxeter groups and Quaternions. كات الفراغيةيوالشب البايرايتذو التركيب المشابه لمعدن تماثل متعدد األوجه القنوبيجمعه المخيني، محمد كوجا وأمليوسف عايدة نظيفة كوجا، المحتوي على اثني عشر وجها متماثال كل منها هو مضلع خماسي غير منتظم إن متعدد األوجه ذو التركيب المشابه لمعدن البايرايت :صمستخل ورات ذات ونظيره شبيه ذي العشرين وجها المنتظم والذي يتميز بتناظر األوجه الموجود في معدن البايرايت يلعبان دورا أساسيا في فهم تراكيب البل يرايت صورة بسيطة باستخدام التمثيل المعتمد على العدد التخيلي الرباعي. تناقش هذه الورقة التماثل الموجود في معدن البايرايت. يأخذ تماثل معدن البا .𝐷3دنكن -البلورات ثالثية األبعاد والتي لها تماثل شبيه بمعدن البايرايت والتي يمكن اشتقاقها من مخطط كوكستر لتركيب المشابه لمعدن البايرايت، الشبيكة الفراغية، مجموعات كوكستر، شبيه ذي العشرين وجها المنتظم، متعدد األوجه ذو ا: مفتاحيةكلمات الرباعيات. 1. Introduction ymmetry describes the periodic repetition of structural features. Any system exhibits symmetry if the action of the symmetry operations leaves the system apparently unchanged. Crystals possess a regular, repetitive internal structure, therefore they have symmetry. Coxeter groups are the symmetry groups generated by reflections [1]. They describe the symmetry of regular and semi-regular polytopes in arbitrary dimensions [1]. The Coxeter- Weyl groups acting as discrete groups in 3D Euclidean space generate orbits representing vertices of certain polyhedra [2-3]. Rank-3 Coxeter-Weyl groups 𝑊(𝐷3)and 𝑊(𝐵3 ) ≈ 𝐴𝑢𝑡(𝐷3) define the point tetrahedral and octahedral symmetries of the cubic lattices [4]. In this work we use Coxeter-Dynkin diagram 𝐷3 and construct the pyritohedral group and the related polyhedra in terms of quaternions by finding the vertices of the pseudoicosahedron with pyritohedral symmetry. We organize the paper as follows. In Section 2 we introduce quaternions and their relevance to O (3) and O (4) transformations. The sets of quaternions defining the binary tetrahedral and binary octahedral groups are given. Section 3 explains Coxeter-Dynkin diagram of 𝐷3 in which the simple roots (lattice generating vectors) and the group generators are expressed in terms of imaginary quaternionic units and the reflection planes of the diagram 𝐷3 are identified as certain planes of the unit cube. Section 4 deals with the construction of the icosahedron, dodecahedron, pseudoicosahedron and pyritohedron from the pyritohedral group derived from the 𝐷3 diagram as its automorphism group. It is noted that the truncated octahedron, the Wigner-Seitz cell of the BCC lattice, splits into two pseudoicosahedra which are mirror images of each other. In Section 5 we discuss the pseudoicosahedron and its dual pyritohedron to give a description of those crystals possessing pyritohedral symmetry. S mailto:nazife@squ.edu.om NAZIFE O. KOCA ET AL 140 2. Quaternions and their Relevance to the Isometries of the O (3) and O (4) Transformations Quaternions are vectors in four dimensions provided with a rule for multiplication that is associative but not commutative, distributive through addition, contains an identity, and for each nonzero vector in four dimensions has a unique inverse [5] The quaternionic imaginary units 𝑒𝑖 , (𝑖, 𝑗, 𝑘) = (1, 2, 3) satisfy the relation: 𝑒𝑖 𝑒𝑗 = −𝛿𝑖𝑗 + 𝑖𝑗𝑘 𝑒𝑘 (1) where ijk  is the Levi-Civita symbol that is completely antisymmetric in the indices. A real quaternion q can be written in general as 𝑞 = 𝑞0 + 𝑞1 𝑒1 + 𝑞2𝑒2 + 𝑞3𝑒3 (2) with 𝑞0, 𝑞1, 𝑞2, 𝑞3 ∈ ℝ, ℝ being the set of real numbers; 𝑞0 is called the scalar part and (𝑞1 𝑒1 + 𝑞2𝑒2 + 𝑞3𝑒3) is the vector part of a quaternion. The conjugate of a quaternion is defined as �̅� = 𝑞0 − 𝑞1𝑒1 − 𝑞2𝑒2 − 𝑞3𝑒3. (3) The norm of unit quaternion q is |𝑞| = √𝑞�̅� =1 and 𝑞−1 = �̅�. The scalar product of two arbitrary quaternions 𝑝 and 𝑞 is defined as (𝑝, 𝑞) = 1 2 (�̅�𝑞 + �̅�𝑝) = 1 2 (𝑝�̅� + 𝑞�̅�) . (4) The transformations of an arbitrary quaternion 𝑡 → 𝑝𝑡𝑞 and 𝑡 → 𝑝𝑡̅𝑞 define orthogonal transformation of the group 𝑂(4). It is clear that the above transformations preserve the norm 𝑡𝑡̅ = 𝑡0 2 + 𝑡1 2 + 𝑡2 2 + 𝑡3 2 . We define the above transformations as abstract group operations by the notations 𝑡 → 𝑝𝑡𝑞 ∶= [𝑝, 𝑞]𝑡, 𝑡 → 𝑝𝑡̅𝑞 ∶= [𝑝, 𝑞]∗𝑡. (5) Dropping also t, the pair of quaternions define a set closed under multiplication. The inverse elements take the forms [6]: 1 1 [ , ] [ ,q], ([ , ] ) [ , ] .p q p p q q p       (6) With the choice of q p , the orthogonal transformations define a three parameter subgroup (3)O . The transformation [ , ] and [ , ]p p p p  leaves the 𝑆𝑐(𝑡) = 𝑡0 invariant. Therefore in 3D Euclidean space one can assume that the quaternions consist of only vector components, namely 1 1 2 2 3 3 t t e t e t e   , satisfying 𝑡̅ = −𝑡. With this restriction to the 3D space the element [ , ]p p  takes a simpler form [ , ] [ , ].p p p p    Therefore the transformations of the group (3)O can be written as [ , ]p p where the sandwiching operator [ , ]p p represents the rotations around the vector 𝑉𝑒𝑐(𝑝) = (𝑝1, 𝑝2, 𝑝3) and [ , ]p p is a rotary inversion [1]. Reflection of a vector Λ represented as a quaternion with respect to a plane orthogonal to the unit pure quaternion 𝑞 can be written as Λ → −𝑞Λ̅𝑞. (7) If we apply another reflection to the vector Λ it will lead to a rotation. The product of two reflections is a rotation. We will display some of the finite subgroups of quaternions related to the tetrahedral and octahedral groups. The set 𝑇 is given by the group elements 𝑇 = {±1, ±𝑒1, ±𝑒2, ±𝑒3, 1 2 (±1 ± 𝑒1 ± 𝑒2 ± 𝑒3)} (8) and is called the binary tetrahedral group of order 24. Another set of 24 quaternions is defined by 𝑇 ′ = { 1 √2 (±1 ± 𝑒1), 1 √2 (±1 ± 𝑒2), 1 √2 (±1 ± 𝑒3), 1 √2 (±𝑒1 ± 𝑒2), 1 √2 (±𝑒2 ± 𝑒3), 1 √2 (±𝑒3 ± 𝑒1) } . (9) SYMMETRY OF PYRITOHEDRON AND LATTICES 141 The set 𝑂 = 𝑇 ∪ 𝑇′ forms binary octahedral group of order 48. These sets play an important role in the definition of the Pyritohedral subgroup. 3. Finite Coxeter Groups, Cartan Matrix and Root Systems In 1934 Coxeter classified all finite Euclidean reflection groups [7-8]. A Coxeter group is a group 𝑊(𝐺) which has a presentation with a very special form 𝑊(𝐺) = ⟨𝑟1, 𝑟2, … , 𝑟𝑛 |(𝑟𝑖 𝑟𝑗 ) 𝑚𝑖𝑗 = 1⟩ (10) where 𝑟1, 𝑟2, … , 𝑟𝑛 are the reflection generators of the group. All information of a root system can be encoded by Coxeter-Dynkin diagrams and the Cartan matrix C. The simple roots 𝛼1, 𝛼2, … , 𝛼𝑛 of the Coxeter-Dynkin diagram are the vectors orthogonal to certain hyperplanes with respect to which the simple reflection acts as an arbitrary vector 𝜆 as [1]: 𝑟𝑖 𝜆 = 𝜆 − 2(𝜆,𝛼𝑖) (𝛼𝑖,𝛼𝑖) 𝛼𝑖 . (11) The Cartan matrix is a square integer matrix that links the simple roots of a given group through the following relation: 𝐶𝑖𝑗 = 2(𝛼𝑖,𝛼𝑗) (𝛼𝑗,𝛼𝑗) . (12) The inverse of the Cartan matrix is related to the metric of the dual (reciprocal) space as: 𝐺𝑖𝑗 = (𝐶 −1)𝑖𝑗 (𝛼𝑗,𝛼𝑗) 2 (13) The basis vectors in the dual space are called the weight vectors, denoted by 𝜔𝑖 , satisfying the scalar product: 𝐺𝑖𝑗 = (𝜔𝑖 , 𝜔𝑗 ). The simple roots and weight vectors are related to each other by (𝜔𝑖 , 2𝛼𝑗 (𝛼𝑗,𝛼𝑗) ) = 𝛿𝑖𝑗 and 𝛼𝑖 = 𝐶𝑖𝑗 𝜔𝑗 . 3.1 Coxeter-Dynkin diagram of 𝑾(𝑫𝟑) The Coxeter-Dynkin diagram of 𝐷3 with the quaternionic simple roots is shown in Figure 1. The angle between two connected roots is 120°, otherwise they are orthogonal. Figure 1. The Coxeter-Dynkin diagram 𝑫𝟑 with quaternionic simple roots. An arbitrary quaternion 𝜆 when reflected by the operator 𝑟𝑖 with respect to the hyperplane orthogonal to the quaternion 𝛼𝑖 , is given in terms of quaternion multiplication [6] as 𝑟𝑖 𝜆 = − 𝛼𝑖 √2 𝜆̅ �̅�𝑖 √2 : = [ 𝛼𝑖 √2 , − 𝛼𝑖 √2 ] ∗ 𝜆 :=[ 𝛼𝑖 √2 , 𝛼𝑖 √2 ] , 𝑖 = 1, 2, 3. (14) The generators of the Coxeter group 𝑊(𝐷3) are then given in the notation of (14) by 1 1 2 1 2 1 2 1 2 1 1 1 1 [ ( ), ( )] [ ( ), ( )] 2 2 2 2 r e e e e e e e e         , 2 2 3 2 3 2 3 2 3 1 1 1 1 [ ( ), ( )] [ ( ), ( )] 2 2 2 2 r e e e e e e e e         , (15) 3 2 3 2 3 2 3 2 3 1 1 1 1 [ ( ), ( )] [ ( ), ( )] 2 2 2 2 r e e e e e e e e         . 𝛼2 = 𝑒2 − 𝑒3 𝛼1 = 𝑒1 − 𝑒2 𝛼3 = 𝑒2 + 𝑒3 NAZIFE O. KOCA ET AL 142 They generate the Coxeter-Weyl group 𝑊(𝐷3) of order 24, isomorphic to the tetrahedral group, the elements of which can be written compactly by the notation 𝑊(𝐷3) = {[𝑇, �̅�] ∪ [𝑇 ′, −�̅� ′]}. (16) Here 𝑇 and 𝑇′ are the sets of quaternions given in (8-9). When the simple roots are chosen as in Figure 1, then the weight vectors are determined as 𝜔1 ≡ (100) = 𝑒1, 𝜔2 ≡ (010) = 1 2 (𝑒1 + 𝑒2 − 𝑒3), (17) 𝜔3 ≡ (001) = 1 2 (𝑒1 + 𝑒2 + 𝑒3). Using the orbit definition 𝑊(𝐷3)(𝑎1𝑎2𝑎3) ≔ (𝑎1𝑎2𝑎3)𝐷3 all the orbits can be determined. 3.2 Coxeter-Dynkin diagram 𝑾(𝑫𝟑) with Dynkin diagram symmetry The symmetry of the union of the orbits (010)𝐷3 ∪ (001)𝐷3 requires the Dynkin- diagram symmetry : 𝛼1 → 𝛼1, 𝛼2 ↔ 𝛼3 as shown in Figure 2. It leads to the transformation on the imaginary quaternions 𝛾: 𝑒1 → 𝑒1, 𝑒2 → 𝑒2 , 𝑒3 → −𝑒3. The Dynkin-diagram symmetry operator reads 𝛾 = [𝑒3, −𝑒3] ∗ = [𝑒3, 𝑒3] which extends the Coxeter group 𝑊(𝐷3) to the octahedral group 𝐴𝑢𝑡(𝐷3) ≈ 𝑊(𝐷3) ∶ 𝐶2 of order 48, the automorphism group of the root system of 𝐷3 . 𝑂ℎ ≈ 𝐴𝑢𝑡(𝐷3) = {[𝑇, ±�̅�] ∪ [𝑇 ′, ±�̅�′]}. (18) Figure 2. The Dynkin-diagram symmetry . The maximal subgroups of the octahedral group 𝐴𝑢𝑡(𝐷3) ≈ 𝑂ℎ, each of order 24, are shown in Figure 3. In the next section we work on pyritohedral group. Figure 3. The maximal subgroups of the octahedral group. The maximal subgroups Chiral octahedral group Tetrahedral group Pyritohedral group Octahedral group α 1 α 2 α 3 SYMMETRY OF PYRITOHEDRON AND LATTICES 143 4. Pyritohedral Group and Related Polyhedra Pyritohedral symmetry is well known in crystallography as the symmetry of the pyritohedron. The group of this symmetry is an important discrete point group of crystallography. The polyhedra with this symmetry can be derived from the Coxeter-Dynkin diagram of 𝐷3. The pyritohedral group ]},[],{[ TTTTTh   consists of rotation generators in addition to the Dynkin diagram symmetry  of D3. The group can be represented as 𝑇ℎ = 𝐴4: 𝐶2 =< 𝑟1𝑟2, 𝑟1𝑟3, 𝛾 > where (: ) denotes the semi direct product. This is the rotational symmetry of a cube with stripes on its faces as shown in Figure 4. The rotation generators 𝑟1𝑟2 𝑎𝑛𝑑 𝑟1𝑟3 of D3 generate the subgroup ],[ )( TT C DW A  2 3 4 of order 12, where A4 is an even permutation of 4 vertices of a tetrahedron. The chiral tetrahedral group 4ATT ],[ consists of 8 rotations by 1200 around the 4 diagonals of the cube, 3 rotations by 1800around the x, y and z axes, and the unit element [9]. Pyrite crystals often occur in the forms of cubes with striated faces as in Figure 4, octahedra and pyritohedra (a solid similar to dodecahedron but with non-regular pentagonal faces) or some combinations of these forms. Figure 4. The cube with stripes on its faces possesses the pyritohedral symmetry. 4.1 Construction of the vertices of the pseudoicosahedron We denote a general vector of D3 by 1 1 2 2 3 3a a a      where the weight vectors are defined in (17). We note that the orbit of the pyritohedral group generated by the vector  with different values of 𝑎𝑖 forms different polyhedra. For instance: 𝑎1 = 1, 𝑎2 = 𝑎3 = 0 corresponds to an octahedron. 𝑎1 = 𝑎3 = 0, 𝑎2 = 1 or 𝑎1 = 𝑎2 = 0, 𝑎3 = 1 corresponds to a cube. 𝑎1 = 0, 𝑎2 = 𝑎3 = 1 corresponds to a cuboctahedron. 𝑎1 ≠ 0, 𝑎2 = 𝑎3 ≠ 0 gives ))(( 21 1 2 11 ee a a ea  , and the orbit corresponds to a pseudoicosahedron as discussed below. Applying some of the rotation elements of the pyritohedral symmetry on a general vector of 1 1 2 2 3 3 a a a      , one can generate five triangles sharing the vertex  as shown in Figure 5(a), which consist of two equilateral and three isosceles triangles. The vertices 2 1 2 1 2 , and ( )r r r r   form an equilateral triangle with an edge length squared: 2 2 1 1 2 2 a a a a  . Similarly, the vertices 2 1 3 1 3 , and ( )r r r r   form the second equilateral triangle with the edge length squared: 2 2 1 1 3 3 a a a a  . The line between the vertices  and 2 3 r r  has a length squared: 2 2 2 3 a a . If we impose all the edge lengths to be equal, we would have five equilateral triangles around one vertex and obtain the relation NAZIFE O. KOCA ET AL 144 2 2 2 2 2 2 1 1 2 2 1 1 3 3 2 3 a a a a a a a a a a       . (19) Suppose the Dynkin-diagram symmetry as shown diagrammatically in Figure 2 leaves the general vector  invariant   . This symmetry exchanges 𝜔2 and 𝜔3, so that 332211233211  aaaaaa  . (20) From (19) one gets ,32 aa  and a general vector can be written in terms of quaternions as 1 1 2 3 1 1 2 ( ( )) ((1 ) )a x a x e xe         (21) where 32 1 1 aa x a a   is a parameter which could be computed from (19). If we take a general value for x in the vector given in (21), the action of the pyritohedral group on  will generate the set of 12 vectors (apart from the scale factor a1): 𝑇ℎ  = ±(1 + 𝑥)𝑒1 ± 𝑥𝑒2, ±(1 + 𝑥)𝑒2 ± 𝑥𝑒3, ±(1 + 𝑥)𝑒3 ± 𝑥𝑒1. (22) Factoring (19) by a1 one obtains the equation 2 1 0x x   . The solutions of this equation are 1 5 1 5 and 2 2       . These values of x will lead to five equilateral triangles sharing the vertex  . Substituting and   respectively for x in (22) we obtain two sets of 12 vertices as: 1 2 2 3 3 1 { , , }e e e e e e         , (23a) 1 2 2 3 3 1 { , , }e e e e e e         . (23b) These sets of vertices represent two mirror images of an icosahedron with a scale factor difference. Multiplying the vertices in (23b) by 3  one obtains the following set of quaternions: 1 2 2 3 3 1 { , , }e e e e e e         . (24) The vectors here have the same norm as in (23a). The icosahedron represented by (23a) is shown in Figure 6(a). Table 1 summarizes the action of the pyritohedral group on  for various values of x: (a) (b) Figure 5. (a) Five triangles meeting at one vertex, (b) The normal vectors of the triangles surrounding . SYMMETRY OF PYRITOHEDRON AND LATTICES 145 Table 1. The action of the pyritohedral group for various values of x. 211 xeex  )( hT Polyhedra generated x or x )( 21 ee   or )( 21 ee   1 2 2 3 3 1 { , , }e e e e e e         or 1 2 2 3 3 1 { , , }e e e e e e         Icosahedron 0x or 1x 1e or 2e 321 eee  ,, Octahedron 2 1 x )( 21 2 1 ee  133221 eeeeee  ,, Cuboctahedron Other x 211 xeex  )( 1 2 2 3 3 1(1 ) , (1 ) , (1 )x e xe x e xe x e xe         Pseudoicosahedron The vertices of the dual of the icosahedron, say the set of vectors of (23a), can be determined as [9]: 1 2 2 3 3 1 1 { , , } 2 e e e e e e           , (25a) 1 2 3 1 ( ) 2 e e e   . (25b) The 20 vertices in (25a and 25b) represent a dodecahedron as shown in Figure 6 (b). Its mirror image can be obtained by replacing   . (a) (b) Figure 6. (a) Icosahedron, (b) dodecahedron (dual of the icosahedron). The vectors in (25b) represent the vertices of a cube which is invariant under the pyritohedral symmetry. Similarly, the other 12 vertices of (25a) form another orbit under the pyritohedral symmetry. They are obtained for x in (22) leading to a pseudoicosahedron. NAZIFE O. KOCA ET AL 146 4.2 Construction of the vertices of the pyritohedron from the pseudoicosahedron We can compute the dual of the pseudoicosahedron in equation (22) for an arbitrary value of x. This can be achieved by determining the vectors normal to the faces of the pseudoicosahedron in (22). The normal vectors of the five triangles in Figure 5 (b) can be determined as follows: The normal vector of the triangle with the vertices 2 1 2 1 2 , , ( )r r r r   can be taken as 3 because 1 2r r is a rotation around 3  for 1 2 3 3 r r   . Similarly, the normal vector of the triangle with the vertices  2 3131 ( and ), rrrr can be taken as 2 that is 2231  rr . The vertices generated by the pyritohedral group from either 2 or 3 would lead to the vertices of a cube given in (25b). The vectors normal to the isosceles triangles shown in Figure 5 (b) can be computed as follows: 1 1 2 4 1 3 5 1 3 (1 ) , (1 ) , (1 ) . b e x e b x e e b x e e          The vectors 2 and 3 are in the same orbit under the pyritohedral group. Moreover, )( 23   is orthogonal to 211 xeex  )( which means the scalar product of )( 23   and  is zero: 023  )),((  . On the other hand, one can prove that 541 and bbb , are also in one orbit under the pyritohedral group. They form a plane orthogonal to the vector 211 xeex  )( . As these five vertices should determine the same plane then one can show that 0 21  )),((  b is the necessary condition to determine the scale factor  . From this relation we obtain 𝜌 = 1+2𝑥 2(1+𝑥)2 where the scale factor implies that 𝑥 ≠ 1 and 𝑥 ≠ − 1 2 . Then the vectors 1 4 5 2 3 , , , and b b b     determine a pentagon, non-regular in general (four edges of the same length and one edge different), as shown in Figure 7. Applying the pyritohedral group on these 5 vertices we obtain 20 vectors where ℎ = 𝑥 (𝑥 + 1)⁄ . The following 20 vertices of the pyritohedron split into two orbits: 12 vertices form a pseudoicosahedron and 8 vertices (±𝑒1 ± 𝑒2 ± 𝑒3) form a cube . {±(1 − ℎ2)𝑒1 ± (1 + ℎ)𝑒2, ±(1 − ℎ 2)𝑒2 ± (1 + ℎ)𝑒3, ±(1 − ℎ 2)𝑒3 ± (1 + ℎ)𝑒1}, (±𝑒1 ± 𝑒2 ± 𝑒3) (27) The pyritohedron has a geometric degree of freedom with two limiting cases as shown in Figure 8. (26) 𝜌𝑏4 𝜌𝑏5 𝜌𝑏1 𝜔2 𝜔3 Figure 7. The plane determined by the vectors normal to the faces of the pseudoicosahedron in equation (22). https://en.wikipedia.org/wiki/Limiting_case_(mathematics) SYMMETRY OF PYRITOHEDRON AND LATTICES 147 When h = 0, the 12+8 vertices in (27) represent the centers of edges and the vertices of a cube as in Figure 8 (a). When h =1, the vertices in (27) represent a rhombic dodecahedron as shown in Figure 8 (b). The regular dodecahedron represents a special intermediate case (𝑥 = −𝜎 or 𝑥 = −𝜏) where all edges and angles are equal. 5. Vertices of the Pseudoicosahedron and the Pyritohedron in a Lattice The simple cubic lattice consists of the vector 𝑚1𝑒1 + 𝑚2𝑒2 + 𝑚3𝑒3 where 𝑚𝑖 ∈ 𝒁, 𝑖 = 1,2,3. Now, we can discuss the pseudoicosahedron and its dual pyritohedron relevant to crystallography. Many candidates of pseudoicosahedra and its dual pyritohedra can be obtained as a structure in the simple cubic lattice. The vertices of pseudoicosahedron and its dual pyritohedron for various x (and corresponding h) values are given in Table 2 and are plotted in Figure 9. As mentioned in Section 4.2, when 𝑥 → 0 (ℎ → 0 ) the pseudoicosahedron is converted to an octahedron and the pyritohedron becomes a cube. The set of vertices of the pseudoicosahedron and the pyritohedron belong to the simple cubic lattice. The unit cubic cell can be stacked in the pseudoicosahedron or in the pyritohedron as long as the vertices are chosen from the simple cubic lattice. Table 2. Pseudoicosahedron and its dual (pyritohedron) for arbitrary x values. x h Vertices of pseudoicosahedron Vertices of pyritohedron (up to a scale factor) 5 3 5 8 ±8𝑒1 ± 5𝑒2, ±8𝑒2 ± 5𝑒3, ±8𝑒3 ± 5𝑒1) For 𝑎1 = 3 with faces isosceles triangles with edge ratios 1,1,√50 49⁄ . {±39𝑒1 ± 104𝑒2, ±39𝑒2 ± 104𝑒3, ±39𝑒3 ± 104𝑒1}, 64(±𝑒1 ± 𝑒2 ± 𝑒3) 1 1 2 (±2𝑒1 ± 𝑒2, ±2𝑒2 ± 𝑒3, ±2𝑒3 ± 𝑒1) with faces isosceles triangles of edge ratios 1,1, √ 2 3 . {±3𝑒1 ± 6𝑒2, ±3𝑒2 ± 6𝑒3, 3𝑒3 ± 6𝑒1}, 4(±𝑒1 ± 𝑒2 ± 𝑒3). 1 10 1 11 (±11𝑒1 ± 𝑒2, ±11𝑒2 ± 𝑒3, ±11𝑒3 ± 𝑒1) For 𝑎1 = 4, faces of with isosceles triangles of edges 1,1,√2 111.⁄ {±110𝑒1 ± 132𝑒2, ±110𝑒2 ± 132𝑒3, ±110𝑒3 ± 132𝑒1}, 111(±𝑒1 ± 𝑒2 ± 𝑒3). Figure 8. Special cases of pyritohedron for (a) h =0, (b) h =1. NAZIFE O. KOCA ET AL 148 Pseudoicosahedron Pyritohedron (dual of pseudoicosahedron) 𝑥 = 5 3 ℎ = 5 8 𝑥 = 1 ℎ = 1 2 𝑥 = 1 10 ℎ = 1 11 Figure 9. The pseudoicosahedron and its dual (pyritohedron) for various x values. For a general x, the union of two pseudoicosahedra 1 2 2 3 3 1 (1 ) , (1 ) , (1 )x e xe x e xe x e xe         , 1 2 2 3 3 1 (1 ) , (1 ) , (1 )xe x e xe x e xe x e         (28) represents the vertices of a non-regular truncated octahedron which can be derived as the orbit of the Coxeter group 𝑊(𝐵3) denoted by 𝑎1(1, 𝑥, 0)𝐵3 . This belongs to the simple cubic lattice if 𝑎1 and 𝑎1𝑥 = 𝑎2 are integers, which implies that x should be a rational number. In other words, as long as 𝑎1 and 𝑎2 are integers the pseudoicosahedron with the vertices can be embedded in a simple cubic lattice. ±(𝑎1 + 𝑎2)𝑒1 ± 𝑎2𝑒2, ±(𝑎1 + 𝑎2)𝑒2 ± 𝑎2𝑒3, ±(𝑎1 + 𝑎2)𝑒3 ± 𝑎2𝑒1 (29) 6. Conclusion The polyhedra possessing the pyritohedral symmetry have been constructed in terms of quaternions. Crystals with pyritohedral symmetry exist in the form of a stratified cube, an octahedron and a pyritohedron. It is expected that crystal structures in the form of a pseudoicosahedron as well as a pseudoicosidodecahedron may exist. The relevance of the pseudoicosahedron to pyritohedral crystals could stimulate research in application to material science. Representation of the pyritohedral symmetry and crystal vectors in term of quaternions is more demanding, but may lead to a new understanding of crystal structures and their symmetries. We are anticipating that this finding will form a link between quaternions and crystallography. SYMMETRY OF PYRITOHEDRON AND LATTICES 149 References 1. Coxeter, H.S.M. Regular polytopes (3rd ed.) New York: Dover Publications, 1973. 2. Koca, M., Al-Ajmi, M. and Koc, R. Polyhedra obtained from Coxeter groups and quaternions, J. Math. Phys, 2007, 48, 113514-113527. 3. Koca, M., Koca, N.O. and Koc, R. Catalan solids derived from three dimensional root systems and quaternions, J. Math. Phys, 2010, 51(4), 043501-043513. 4. Conway, J.H. and Smith D.A. On quaternions and octonions: Their geometry, Aritmetic and symmetry, A.K. Peters, Ltd. Natick, M.A., 2003. 5. Hamilton, S.W.R. Lectures on Quaternions, London: Whittaker, 1853. 6. Koca, M., Koc, R. and Al-Barwani, M. Non-crystallographic Coxeter group H4 in E8. J. Phys. A: Math. Gen. A, 2001, 34,11201-11213. 7. Coxeter, H.S.M. Discrete groups generated by reflections. Annals of Mathematics, 1946, 35, 588-621. 8. Steinberg, R. Finite reflection grouups. Trans. Am. Math. Soc., 1959, 493-504. 9. Koca, M., Koca, N.O. and Al-Shu’eili, M. Chiral Polyhedra derived from Coxeter diagrams and quaternions, SQU J. Sci., 2011, 1, 16, 63; arXiv: 1006.3149. Received 7 June 2016 Accepted 4 September 2016