Really, your lattices are all primitive, Mr. ~ravais!

'
Sidney F. A. Kettle
School of Chemical Sciences, University of East Anglia, Norwich NR4 TTJ, U.K.

Lars J. ~ o r r b ~ ~
Department of Chemistry and Chemical Engineering, Royal Military College of Canada, Kingston, Ontario, Canada K7K 5L0,
Scientific and technological developments over the past
few years have increased attention on the solid state. This
is reflected in, for example, a recent issue of this Journal
(October 1991)that contains no fewer than four articles on
solid state chemistry. In a sense chemistry has been well
prepared for this development, for fundamental aspects of
ihesolid state have been part of undergraduate coukes for
generations, namely the basic crystal structures of metals,
Halts, and simple minerals. ~ o n c i ~surh
t s as unit cells. lattices, crystal systems, and the close-packing of spheres
are
almost everywhere taught early in the chemistry program.
However, over time there has been little change in the way
these tooics have been oresented-tvoicallv.
bv describine
.
the seven crystal syste6s and iilustrating'the 14 ~ r a v a ;
lattices. with the relationshios between the axes a. b. c of
be
their u k t cells. ~ e ~ r e s e n t a t i vexamples
e
of this
found in two well-regarded textbooks by Oxtoby and
Nachtrieb (1)and by Alberty and Silbey (2).Given this
solid tradition, one might assume that the basics of structural chemist6 are secure parts of the subject and that no
review is needed. The purpose of this paper is to question
such an assumption. w e d o so because the app;oach of
solid-state physicists and solid-state spectroscopists is
rather different to that traditionally followed by chemists
and crystallographers. Our first concern is with the 14
Bravais lattices.
~

may

Bravais Lattices
Auguste Bravais (3, 4) published his classical work in
1850, some 60 years before X-ray diffraction was discovered. The work was based on his thorough knowledge of
eeometrical crvstalloeraohv and eeneral mathematical
eonsiderations ;bout t%e h1l"ing of space with parallelepipeds. His results became relevant to the later development
of X-ray crystallography, a subject that in turn became
highly relevant to chemistry So, Bravais and his lattices
became part of every chemist's education. Bravais found
that there are onlv five olane (two-dimensional) and 14
space (three-dimensiouaf) lattices. The 3D-lattices have
since become known to chemists as the seven primitive
and the seven centered Bravais lattices. As seen in Table 1,
the latter are not evenly distributed amonmt the seven
crystal ~ y s t e m s The
. ~ unit cells traditionaliy associated
with the Bravais lattices all have the full point = u p svmmetry (the holohedry) of the crystal system to h i i h ihey
'The title. which ~roDerlvshould have been "En effet.vos rgseaux
sont
tous
-~~- o&nitifs.
-, M. ~raiais!".
-~
, was chosen as homaoe to the areat
"
Frencn pnysfclslana crysta ographer A~gLsteBrava~s,1811-1863,.
'Autnor to whom correspondence sno~ldoe aooressw
3Each cryaa system s characterize0 by a cena n point grow (its
hoiohedry)of finite order, the operations ofwhich bring the lattice into
self-coincidence;see Table 1. Every lattice is also brought into selfcoincidence by an infinite number oftranslations. ~ranslationalinvariance is perhaps the most important feature of crystalline materials
and is, in a sense, the motivation for this paper.
4Thecrystal is assumed to be infinite,a valid assumption for most
purposes and certainly if one is interested in the bulk physical prop
erties ofthe crystal; but see footnote5.
~~

~~~

~~~

~~

~0~

~~

~~

Table 1. The Seven Crystal Systems, the 14 Bravais

Lattices and the Seven Holohedries.
System

Lattices

Cubic

Holohedry
(Point Group)

Point Group
Order

P, I, F

Oh

48

Hexagonal

D6h

24

Tetragonal

P, l

0
4
h

16

Trigonal'

P (= R)

D3d

12

Olthorhombic

P, I, C, F

02h

P. c

C2h

8
4

C!
~.

2-

Monoclinic
Triclinic

P = primitive, I = body-centered, F = face-centered.C = end-centered

'The trigonalsystsm is alsocalled the rhombohedra1 because of the shape

primitive unit cell;the symbol R may be encountered instead of P,

01 its

belong, although an exception is generally encountered for

the hexagonal system.
Given this background, it may come a s a surprise to
learn that, for the solid state physicist, all 14 Bravais lattices are primitive and that primitive unit cells exist for
each and every lattice. This is not to say that the chemist's
view of the seven centered Bravais lattices is wrone. but
perhaps that the future development of the subject would
be served bv a chanw in oresentation. What. then. is the
physicist's approach~howdoes it differ from the chemist's,
and what are the conseauences of these differences? To explore this further we turn to the translations that generate
the entire lattice from a unit cell.
Translations
An invariant subgroup of every crystallographic space
group is the group of all primitive translations, a group
here denoted T. These translations, when acting on the eutire crystal,
serve to superimpose the crystal onitselfjlt is
.
this group of translations that is of p&icular concern to
the solid-state physicist. From T the Brillouin zone is derived and from it, in turn, the energy levels associated with
crystals (5).Ifthe energy levels are to be correct, the group
of all translations must be chosen correctly. A difference
between physicists and chemists is that physicists are interested in Brillouin zones and therefore in translations of
unit cells. Chemists are interested in the contents of unit
cells, in the atoms and the molecules (we simplify!). To explore this difference further we turn to the contents of a
orimitive unit cell. There is. as Bravais recoenized. no sinile choice of the shape and symmetry of a primitive unit
cell. However. the volume and the atomic contents of a
primitive unit cell are constant for any given crystal swucture. The operations of the complete set of primitive translations, of T, on this unit cell Eontent
the positions of all the atoms of the crystal. Indeed, the only
condition a primitive unit cell has to fulfil is to generate
the entire crystal when the operations of T are applied.
~

Volume 70 Number 12 December 1993

959

tals are very large compared to

typical translation vectors, all
unit cells "behave the same" and
the spectra can be interpreted in
terms of the vibrations of a single, primitiue unit cell. But if a
centered, n-times larger unit cell
is chosen, the predicted number
of vibrational modes will be ntimes too large and the spectral
predictions correspondingly incorrect.
Perhaps now t h a t chemists,
too, are becoming increasingly
interested in crystal energy levels, it is time to reconsider their
traditional approach to the nonprimitive Bravais lattices. To do
so, t h e meaning of "primitive
translation" must be elaborated.
Primitive Unit Cells
A lattice is the periodic collection of points generated by all
possible linear combinations of
three primitive translation vectors PI, Pz, ~ 3 ,

[--

fJ/
-7@7-

a
%

+-

The translation group, T, consists of all operations t, where nl,

nz, ns are integers running through
all positive and negative values5.
The origin of the lattice and the
identity operation are defined by:
n,=nz=n3=O=to
For each translation t there is
also a translation -t. Therefore
every lattice point is a center of

generators,
vectors
symmetric.
symmetry;
pl, p,,
all
The
must
lattices
p3,
primitive
the
be linearly
are
primitive
centrolattice
in-

dependent and a s short as possible.

More specifically,they are al-p,
2
ways chosen such that the sum of
,,'
,
the squares of their lengths,p? +
Pa
D
pzZ+ pa2, is a minimum. I n many
cases pl, p,, and p3 are nonorthogonal. For every lattice there
r I=,I
are three such primitive generators that relate any given lattice
point to t h r e e of i t s nearestFigure 1. The seven conventionally primitive unit cells showing three primitive translation vectors pi, p2, neighboring lattice points. For
m which are equal to the crystallographer's usual lattice parameters a, b, c. Note that the "atoms'are the seven primitive Bravais latlattice points (see footmote 6).
tices pl, p,, p3 correspond to the
Legend: (A) Cubic, (6) Hexagonal, (C) Tetragonal, (D) Rhombohedra1 (Trigonal),(E) Orthorhombic, crystallographer,s usual unit
(F) Monoclinic, and (G) Triclinic.
edges a, b, c, see Figure 1. InFigure 2 we give diagrams of the
primitive unit cells of the seven centered Bravais lattices
However, when the set of all translations operates on a
nonprimitive unit cell, the crystal is generated two or four
where, again, pl, p,, p, define the cell edges. The seven
times ouer for body-centered or face-centered cells, respeccentered Bravais lattices differ in that a, b, o do not define
tively (and three times over for rhombohedra1 cells on hexthe directions of the primitive generators. However, pl, p,,
p3 are easily expressed in terms of a, b, c , a s is shown in
agonal axes). To avoid such a multiplication, one has to
work with a subset of T: 112, 1/4, and 113, respectively. But
in so doing, one obtains a n incomplete Brillouin zone and
the density of states will be off be a factor of 2,4, or 3. This
'In reality, the total number of translation operations is finite and
is perhaps most easily illustrated, in a n inverse sense, by
identical to the number of primitive unit cells in the crystal. For crystals of the sizes used in solid state experimentaltechniques,this is a
reference to vibrational spectroscopy. Because the wavevery large number, typically in the range 10'~-10~'.
lengths used in studies of the vibrational spectra of crys+

960

Journal of Chemical Education

Figure 2. The seven conventionally centered unit cells showing a convenient choice of corresponding primitive unit cells. The three primitive
translation vectors PI, p2. p3 are notequal to the crystallographer's usual lattice parameters a, b, c, but are chosen in accord with the definitions
detailed in Table 2. Note that the "atoms' are lanice points (see footnote 6).
Legend: (A) Body-centered cubic, (0) Face-centered cubic (C) Body-centered tetragonal, (D) Body-centered orihorhombic, (E) Face-centered
onhorhombic, (F) End-centered orthorhombic, and (G)End-centered monoclinic.
Table 2, w h i c h i s adapted f r o m B u r n s a n d Glazer (6).
Apart
from some flexibility t h a t m a y arise f r o m point group symmetry, t h e p r i m i t i v e translation vector set i s n o w defined
for a l l space lattices. E v e r y p r i m i t i v e unit cell i s t h e n a
parallelepiped a n d contains one lattice point. I t i s importa n t t o note t h a t w e are here dealing with empty lattices,
t h a t is, lattices w i t h o u t a n y chemical content. Therefore
t h e "atoms" in a l l o u r figures are lattice points a n d n o t r e a l
In a l l body-centered a n d face-centered cases t h e p r i m i t i v e unit cells a r e rhombohedrally shaped a n d have lower
symmetries t h a n t h e corresponding Bravais lattices. In
o t h e r words, these unit cells do n o t show t h e full p o i n t

61t is a very common mistake among students-and textbook writers!-to

fail to distinguish properly between lattice points and atoms.
The unit cells associated with a lattice are always empty and have, in
the oresent context of solid state science. one of seven svmmetries.
~ o i e~~,
v e r filled
.
unit cells.
as encountered in real crvstal'structures.
~~~,~~~~~~
,~
have one of 230 symmetries. In most rextooow the Lnr cels of tne
BBC and FCC Ian ces are exempl flea wlh the crysta strmdres of
a-Fe and Cu, respectively. In both cases the basis, the atoms associated with each lattice point, is one single atom. This means that a
model (of the kind normally used in teaching) of an empty unit cell
and one containing atoms lookthe same. However, this only happens
when the basis is a sinole atom. Proaressina to such a simole structdre as CsC ma6es tneo tference betweenis empty pr m i ve cdb c
Jnlt cel ano the I I ea one very OOVOLS
~

~~~

~~~

~~

Volume 70

Number 12 December 1993

~~

961

Table 2. Primitive Generators for the Seven Centered

Bravais Lattices as Functions of the Lattice Parameters
a, b, c of the Conventional Centered Unit Cells.
The lengths d the unit cell edges are a =Jal,b = Ibl, c = Icl, and the
lengths of the generators are pi= Ipil Capital letters in brackets indicate the corresponding diagram in Figure 2.

Body-centeredcubic (A)
pl=(a+b-cY2
p,=(a+b+c)/2
p3=(a-b+cY2
a=b=c;pl=p2=p3.
The primitive unit cell is rhombohedra1(a = 109.47')with Dad
symmetry. Note that in Fig. 2A the 3-fold axis of the rhombohedral unit cell is along the body diagonal through the lattice
point where p, ends.

Face-centeredcubic (6)
pl = (a + bY2
p, = (b + dl2
p3 = (a + c)i2
a=b=c;pl=pz=p3
The primitive unit cell is rhombohedra1 (a = 60 ') with D3d
symme*. Note that in Fig. 2B the 3-fold axis of the rhombo
hedral unit cell is along the body diagonal through the origin.

Body-centeredtetragonal (C)
pl=(a+b-c)/2
pz = ( a+ b + cY2
p3=(a-b+cY2
a=b+c;pl=pz=p3.~
The primitive unit cell is a distorted rhombohedron of C,symmetry.

Figure 3. The compression of an elongated rhombohedral unit cell

(A) along ih threefold axis gives at one point in the defornlation the
primitive unit cell of the FCC lattice (0). At this point the angle sub
tended by adjacent "waist-band edges is 60'. As this angle increases, a trigonal lattice is again obtained until it reaches 90'(C),
when a Drimitive cubic lattice is obtained. Yet more comDression
oives a thoonal lattice aaain until the anale eauals 109.47..'the "tetrahedral &gk. At h
I s po nt tne BCC atlice ' s generated(~,.The
~ l mate
t
compression through a lrlgonal attice gives, at an angle ot
120' the two-amensona hexagonal an ce (not snown,.

~0~

Body-centeredorthorhombic (D)
pl=(a+b-cY2
pa = (a + b + cX2
pB=(a-b+c)/2
a#b#e;pl=pz=p3.
The primitive unit cell is a distorted rhombohedron of Cisymmetry.

Face-centeredorthorhombic (E)
group symmetry of their lattices. In the orthorhombic endcentered case the primitive unit cell is ofD% symmetry, so
the holohedry is retained, but the cell angles are no longer
all 90'. Likewise, in the end-centered monoclinic case the
primitive unit cell is of the holohedral symmetry ( C d , but
the anele between D. and D
. is not 90'. so the basic monoclinic &nditions a r l not ui(e1d. I n all seven cases the axes
defmed bv the ~rimitivegenerators are inconvenient for.
and so avoided Ly, the c ~ t a l l o g r a p h e r .
The discussion above is echoed in the ambiguity of the
primitive unit cell of the hexagonal crystal system. This
unit cell is not hexagonal but of Du, symmetry. For that
reason it is common to choose a three times larger unit cell
(containing three lattice points instead of one) which does
have a sixfold axis. Thus, it is triply-primitive and truly
hexagonal; see Figure 1B.

Relations among Bravais Lattices

The aooroach detailed above enables a rather different
and, ho&ully, illuminating way of lookingat the relationshio among the Bravais lattices. This is nowhere better
the trigonal (rhombohedral) primitive unit
seen than
cell shown i n Figure 3 and the sequence of different prim-

962

Journal of Chemical Education

p, = (a + b)/2
pz = (b+ dl2
p3 = (a + cU2
a+b+c;pl+p2#p3.
The primitive unit cell is a distorted rhombohedron of Cisymmetry.

End-centeredorthorhornbic (F)
pl = (a - bU2
p2=(a+b)/2
p3=0
a#b#c;p,=pz#p3.
The primitive unit is cell is prismatic of symmetly DZh.The
angle between pl and pz is neither 60' nor 90'.

End-centeredmonoclinic (G)
p, = (a - bY2
h=(a+b)/2
Ps=c
a f b f c ; fj#90';p1=p2#p3.
The primitive unit is cell is prismatic of C2*symmetly. The
angle between pl and p2 # 90'. The two-fold ads is along y.

itive unit cells generated a s it is compressed along the

threefold axis. Note that in trigonal unit cells all edges are
of equal length. In Figure 3A there is an angle of 45' between limbs of the zig-zag "waist band" around the middle
of the cell containing six of its comers. As the cell is compressed along its &reefold axis a point will be reached
where this angle becomes 60' (Fig. 3B). Each set of three
adjacent waistband corners now define an equilateral triangle and thus a threefold axis. There are, then, three such
threefold axes. Together with the threefold axis inherent in
the original trigonal unit cell, there is a total of four threefold axes. We are in the cubic m s t a l svstem! Acomoarison
between Flgures 2A and 3B shows that our rhombohedral
cell is in fact the orimitive unit cell corresoondine toa FCC
lattice. compre& the trigonal nnit cell hthe<until the
waist band angles eaual90' (Fie. 3 0 . Now each corner of
the waist banzlies i n a threeford axis. We have found the
usual primitive unit cell of the primitive cubic lattice.
compr&s yet more. When the waiit band angle is 109.47',
the tetrahedral angle. each waist band corner lies on a
fourfold axis. As ~Gmres3D and 2A show, we have then
arrived at the primitive unit cell correspondina to the 13CC
lattice. This development hinges on the f a d thBt the primitive unit cells of the two centered cubic lattices are both
trigonal, of D3d symmetry Similar but perhaps less striking relationships exist between the centered Bravais lattices and the corres~ondineorimitive unit cells for the tetragonal and orthoihombic-&ystal systems.

Conclusion

Although the contents of this article may offer novel

ideas for some chemists. it would not have surorised Anguste Bravais. As reference to his original work'shows, he
well knew that primitive nnit cells exist for every one of
the 14 space lattices. We have not been able to determine

inst when the seven centered lattices came to dominate the

thinking of chemists, but it seems to have been, not sururisinelv. at the advent of X-rav crvstdloera~hv.For the
x-ray '&~stallographerworking wiih Weissenberg or any
of the other ~hotucrra~hic
techniaues.
. . observed diffraction
patterns are often more readily interpreted in terms of
centered unit cells than in terms of their primitive counterparts. Thus, it is almost certainly becan& oftheir practical
value to the crystallographer that the centered unit cells
have become so well known. Unfortunately, this seems to
have divided solid state scientists in two camps-the phvsicists and spectroscopists on the one hand and the ;himists and crystallographers on the other. To enhance the understanding of the basis of solid state chemistry, we believe
that it is necessary for chemists to become as familiar with
the 14 primitive unit cells of spectroscopy and solid state
physics a s they are with the well-established crystallographers' seven primitive and seven centered unit cells.
After all, Anguste Bravais gave us both!

-..

Acknowledgement

One of us (SFAK) is indebted to the Department of National Defence of Canada for financial s u o ~ o r(ARP-erant
.t
FUHGC).
.A

Llterature Cited

N. H. Rincipks ofModom Chemislry, 2nd ed.: Saunders:

Phlladelohia. 1990:o 631.
2. Albelt, R. A,; Silbey, R. J.Physiool Chemistry, W~ley:New Y o 4 1992:p 787.
3. Bravais, M. A,, "Memoire s u r 1es aystemea folmea po'des poinfs disfribub
d ~ ~ l r P n m o sur
n t unpion ou dons i'paop'. J. l'&lePolyt~chnique 1860, Cohkr
1. Oxtoby, D. W ;Nachtrieb,

..

*,?.~lhm
1-128
, XJX.
~

4. Bravaia, M.A. on T k S ~ ~ fP~dbP~iitt*lorlyDist~~fodod"odPlo"eo~

f f
in S-;
Shsler,A. J. Trans.; CrystallogaphieSaciety d h e r i e q MemoireNm.
her One; The Book Concern: Hanemk, MIC1949.
5. Kettle. S. F.A,; Nomby, L. J. J Ckm. Ed=. 1890,67,1022.
6. Buma. G.:G b ~ e rA M.Sooco Gmvos for Solid St=& Seknfiafs. 2nd od.: Academic

Volume 70 Number 12 December 1993

963