American Mineralogist, Volume 63, pages 219-229, 1978

The origins of color in minerals

Kunr Nlsseu
Bell Laboratories
Murray Hill, NewJersey07974

Abstract

Four formalisms are outlined. Crystal field theory explainsthe color as well as the fluores-
cencein transition-metal-containingminerals such as azuriteand ruby. The trap concept,as
part of crystal field theory, explains the varying stability of electron and hole color centers
with respectto light or heat bleaching,as well as phenomena such as thermoluminescence.
The molecular orbital formalism explainsthe color of charge transfer minerals such as blue
sapphire and crocoite involving metals, as well as the nonmetal-involvingcolors in lazurite,
graphite and organically colored minerals.
Band theory explains the colors of metallic minerals; the color range black-red-orange-
yellow-colorlessin minerals such as galena, proustite, greenockite,diamond, as well as the
impurity-causedyellow and blue colors in diamond. Lastly, there are the well-known pseudo-
chromatic colors explained by physical optics involving dispersion,scattering,interference,
and diffraction.

Introduction The approach here used is tutorial in nature and

referencesare given for further reading or, in some
Four distinct physical theories (formalisms) are instances,for specificexamples.Color illustrationsof
required for complete coveragein the processes by some of the,principlesinvolved have been published
which intrinsic constituents, impurities, defects, and in an earlier less technical version (Nassau, 1915a).
specificstructures produce the visual effects we desig- Specific examples are given where the cause of the
nate as color. All four are necessary in that each color is reasonably well established,although reinter-
provides insights which the others do not when ap- pretationscontinue to appear even in materials,such
plied to specificsituations.An outline with examples as smoky quartz (Nassau, 1975b), which have been
is givenin Table l. extensivelyinvestigatedover many years.
This classificationgoes well beyond the traditional Specificcausesfor color in the majority of minerals
idiochromatic ("self-colored," as from a major in- are in fact not known, and detailed investigations,
gredient), allochromatic ("other-colored," as from often requiring precision spectroscopy, impurity
an impurity), and pseudochromatic("false-colored" analysis down to ppm levels, magnetic resonance,
as originating from physical optical effects),although and sometimes radiation and even laboratory syn-
thesethree are, in fact, indirectly included in several thetic procedures,may be necessaryfor unambiguous
placesin Table l. The current approachestranscend conclusions.As an example, the deep blue of dia-
the conventionalwisdom which attempts to attribute monds such as the "Hope" almost certainly origi-
blue-greencolors to copper, deep blue to cobalt, red nates in merely a few boron atoms per million of
o r g r e e nt o c h r o m i u m , a n d s o o n . T o g i v e o n l y o n e carbon atoms (this also provides the electricalcon-
example: deep blue colors may be produced by ductivity). Previouslythe blue color had been attrib-
idiochromatic copper (shattuckite), allochromatic uted to aluminum presentat much higher concentra-
cobalt (blue spinel), by a color center (Maxixe-type tion, but synthesisdemonstratesthat aluminum in
irradiated beryl), by chargetransferbetweeniron and this instancedoes not produce a blue color, whereas
titanium (blue sapphire),by molecular orbital effects b o r o n d o e s ( C h r e n k o , 1 9 7 3 ) .
in S; units (lapis lazuli), by boron acceptors in a Many different systemsof designationare used for
band gap ( b l u e diamond), and by p s e u d o c h r o m a t i c the energy of quantized systems(e.g. energyas wave
light interference (blue regions in opal). numbers in cm-', electron volts eV, kcal/mole, etc.;
/ 78/0304-02r9s0200
0003-004x 2t9
220 NASSAU.' COLOR IN MINERALS

T a b l e l . T w e l v e t v o e so f c o l o r i n m i n e r a l s

Col-or Cause Typlcal mlnerals Formallsm

Transltlon metal compounds Afmandlte, ma]achite, turquolse Crystal fleld theory

Transltion metal impurlties Cltrlne, emeral-d, ruby Crystal fleld theory
Color centers Amethyst, fluorlte, smoky quartz Crystal fleld theory
'1
Chrnoa trrnefan R l rra crnnhl ra annnal to aqttrila Molecular orbital theory
Onoenln mrtaFialq Amber, coral, graphlte Molecu1ar orbltal theory

C o n d u c to r s L'nnnen. Irnn. s-l Ilrcr Rand i:haanrr

S e m lc o n d u c t o r s Galena, proustlte, pyrlte, sulfur Rand fha^rv

Doped semiconductors Blue dlamond, yel1ow diamond Rand fhonrrr

f)i <ncrq'l nn "H1rtr" rn TrocrFn qems Physlcal optlcs

S..ttarino Moonstone, rrstarsrr, rreyesrl Pl.rrrcl nrl ^nil.q

Int erference frldescent chalcopyrlte Phrrci nrl nn1:l nq

Diffract lon Opal Phrrc{ nrl nnl:l n<

o r t h e e q u i v a l e n tl i g h t w a v e l e n g t hi n n m , p m , o r A , There are no unpaired electrons when all electron

or its frequency in sec 1 or Hertz, Hz). The eV scale shells are completely full or empty as in Cr6+, Ce4+,
used in this article and its relationship to the spec- or Cu+, when this type of color cannot result.Of
trally pure colors is illustrated in Figure l thesg transition elements, Fe is the most plentiful
(about 5Voof the earth's crust) and is therefore the
The crystal field formalism dominant color contributorin minerals.
Color best explained by the crystal field formalism Consider now a mineral containing, say, some tri-
i n v o l v e sp r e d o m i n a n t l yi o n i c c r y s t a l sc o n t a i n i n gi o n s valent chromium. The electronic configuration of
with unpaired electrons.These usually originate in Cr8+ is ls"2s22p"3s23p"3f:here only the three un-
elementswith partially filled d shells such as V, Cr, paired electronsin Ihe 3d shell can interact with vis-
M n , F e , C o , N i , a n d C u , o r i n e l e m e n t sw i t h p a r t i a l l y ible light so as to produce absorption and color. Due
filled / shells such as the actinidesand lanthanides. to the quantization of energy, a system such as Cr3+
can exist only in a limited number of states, the
"energy levels." Theseenergylevelscan be shown on
ELECTRON
WAVELENGTH VOLTS a diagram such as Figure 2, by their energy content
coLoR (i) (ev) with respectto the "ground state," i.e. the stateof the
ion when it is not interactingwith any kind of radia-
tion and is in its lowest energy state.

lo-l'-l ,,oa\

/BLUE-GREET

z2
--7-
#.-
,ry,/
0
RED
z FLI.JORESCENCE
.tEg_^^-.,^.

+ABSORPTION -AESORPTPN

(c) (e) (f ) EMERALO

(a) (b) (c) RUBY

Fig. l. Colors, wavelengths (A), and energy (eV) for spectral Fig. 2. Energy levels, transitions, and color absorptions in ruby
pure light (SW is short and LW long wavelength ultraviolet). (a) to (e) and in emerald (f).
 NASSAU: COLOR IN MINERALS
The actual calculationof such energylevelsis quite
complex and, for a given atom, has to include the
following parameters:the valencestate,the symmetry
of the ion's environment (i.e. the coordination poly-
hedron of chargeswhich producesthe "crystal field,"
also sometimes called the "ligand field") including
any distortions from ideal tetrahedral, octahedral,
etc.; and the strength of the crystal field, i.e. the
nature and strength of the bonding. The result of
such a calculation is an energy level schemesuch as
that of ruby, (about lVo Cr'+ substitutedfor A13+in
the distorted octahedral sites of AlrO, with very
s t r o n g b o n d i n g r e s u l t i n gi n a n A l - O d i s t a n c eo f l . 9 l
A) shown in Figure2a.
The crystal field calculation not only gives the ex-
cited energy levels(only the first three are shown in
Figure 2a) with respect to the ground state A, but
also a set of "selection rules," covering the probabil-
ity of parallel electron paths. For example,excitation
(light absorption) can result in this instance in a
change from the ground state A to statesC and D
with high probability, but the changefrom A to B is
relatively forbidden (Fig. 2b). De-excitation cannot
occur from D or C back to A but must first pass
through state B with the emission of heat to the
lattice as in Figure 2c; this can now be followed by
light emission, the red fluorescenceof ruby, from
level B back to the ground state A as in Figure 2d.
Additional complicationsnot so far mentioned in-
clude temperature and other broadening of energy
levels and the fact that, since the crystal field can
differ with orientation in an anisotropic crystal, so
can the energy level scheme,resulting in pleochroism
(dichroism in the case of rub'y). The actual unpola-
rized absorption spectrumof ruby shown in Figure 2e
illustrates how the broadened absorbing D and C
levels produce violet and green-yellow absorption
b a n d s .T h e r e m a i n i n gl i g h t t r a n s m i s s i o np r o d u c e st h e
strong red color of ruby with its blue (purple) cast.
Some of the energy absorbed into the C and D ab-
sorption bands (as well as into other absorption
bands at higher energiesin the ultraviolet not shown)
reappears as red fluorescence from energy level B
(which remains quite narrow), thus adding to the red
glow of this gemstone.It can be seento be a coinci-
dence that both the color as well as the fluorescence
of ruby is red. It should be noted that the presenceof
Fe impurity can completely "quench" the fluores-
cence by siphoning off energy from the fluorescence
level B without significantlyaffectingthe absorption-
causedcolor.
The effect of a changein the strength of the crystal
221
field is well illustratedby comparing ruby to emerald,
in which Cr3+ also substitutesfor Al3+. The distorted
six-coordinatedsite and Al-O distancesin emerald
are similar to those in ruby, but the effect of the
added Be and Si and the changein structurefrom the
hexagonal close-packedoxygen framework of AlrOt
to the more open structure of BeaAlrSiuO'a leads to
metal-oxygen bonds of higher covalency than in
ruby, a lower electron density on the oxygens, and
accordingly produces a crystal field of reduced
strength.The resulting relativelysmall shifts of the C
and D absorption bandsin Figure 2f essentiallyelimi-
nate the red transmissionbut intensify the blue-green
transmission, leading to the characteristicemerald
color. The fluorescencelevel B is almost unchanged,
and here again the strong red fluorescenceadds to the
quality of the gemstonecolor when not quenchedby
iron impurities. A similar change in crystal field re-
sults in a changefrom red to green as the concentra-
tion of CrrO. in AlrOg is increased,resultingin a less
ionic bonding and a reduced crystal field.
Alexandrite (Cr3+ in AlrBeOr) has an absorption
schemeintermediate betweenthose of ruby (Fig. 2e)
and emerald (Fig. 2f). The almost equally balanced
red and blue-greentransmissionbands result in the
eye perceivingeither the one or the other as domi-
nant, depending on the energy distribution of the
incident light (appearinggreenin daylight and red in
incandescentlight), and are responsiblefor the so-
called "alexandrite effect." Note that a different ion,
V3+, in a different lattice, AlrO', producesalmost the
identical effect in natural or synthetic "alexandrite-
colored corundum." The alexandrite effect has, in
fact, been observedin a number of different minerals
(White et al., 1967).
A last complication of crystal-field-caused color is
that the sameion may enter severaldifferent sitesin a
mineral, thus providing several sets of energy level
schemesand resulting in superimposed absorption
bands. If there is transfer of electronsamong differ-
ent sites, then the molecular orbital formalism de-
scribed below can produce additional absorptions.
Detailed treatments of crystal field conceptshave
b e e n g i v e n b y B u r n s ( 1 9 7 0 ) a n d W o o d ( 1 9 6 9 ) ,a n d
may also be found in recent inorganic texts such as
that of Cotton and Wilkinson (1972). Intermediate
treatmentsby Loeffier and Burns (1976),Frye(1974),
and Fyfe (1964) can be recommended.
The crystal field formalism is appropriate for three
types of color: those causedby transition metals ei'
ther as a major ingredientor as an impurity, and that
produced by color centers.
222 NASSAU: COLOR IN MINERALS

T a b l e 2 . I d i o c h r o m a t i cc o l o r a t i o n i n t r a n s i t i o nm e t a l c o m o o u n d s * be present.Laboratory synthesisof mineralswith the

controlled addition of impurities may even be needed
Cerlum: P.rlaita t'rrallnr'\

Chromlum: Cn^n^ira lnranoal. / , n h ^ a h i ^ F ^ ^ ^vi v+v^f

PrrvvrrrLr us / -r - sAu\ .
\ / ,
f o r u n a m b i g u o u se x p l a n a t i o n so f c o l o r .
rrrrar^\ri to I aracn )

C o b a l - t:
As discussedabove, the same impurity can cause
Erythrlte, roselite, spherocobaltite
(plnk ) widely different colors, such as Cr-caused ruby red
Azurlte, chrysocolla, turquolse (Otue);
dJ.ntAse mAlant\ita I\ 'bor n! va rarrl ), . ^v ,u , Pn rFr lvt ea lrad\
\rqu/
and emerald green. However, there are some emer-
Iron: Lazrillie (hlrre):
'ltrdl^ekil:p
.li\rine 1\ o r eva u. )r r. / ,
6r
.ln^.difa
alds which derivetheir color either partially or totally
frpd). .aooycnitF o^arhita
( ye1low ) f r o m v a n a d i u m( W o o d a n d N a s s a u , 1 9 6 8 ) .
M a n g -asnpeesses: a r t iRt eh o d o c h r o s l t e . rhodonite (p1nk) :
(orange) ; ganopnyttite'
(ye-L_row, Color centers
Nlckel: Bunsenlte (green)
Uranium: Autunite, carnotlte (yellow) I curite, The unpaired electron which produces color by
masuylre (orange)
light absorption into excited statesdoes not have to
Charge transfer may also be present in some of be located on a transition-elemention; under certain
these. circumstancesit can be located on a non-transition-
element impurity ion or on a crystal defect such as a
Idiochromatic colors causedbv maior transition-metal missing ion. Both of these can be the causeof color
constiluents centers.If an electronis presentat a vacancy,we have
Severalexamplesof this type of color are given in an "electron" color center; if an electron is missing
Table 2 and many more can be deduced with some from a location where there usually is an electron
certainty from merely knowing the chemicalcompo- pair, we have a "hole" color center.Color centerscan
sition of minerals containing essentialtransition-ele- also be treated in the band formalism, and a good
ment ingredients.Nevertheless,there are various pit- review is that of Schulman and Compton (1962).
falls: some valencestatesneed not causecolor (e.g. N{any color centersare known, but the exactcolor-
monovalent Cu); the color may be disguisedor modi- causing mechanism has been establishedin only a
fied by impurities; molecular orbital effectsmay also very few instances.One of these is the purple "F-
b e o p e r a t i v e ;a n d s o o n . center" or Frenkel defect of fluorite, one of many
types of color center which can form in fluorite. Fig-
Allochromatic colors causedby transition-metal impu- ure 34, is a two-dimensional representationof the
rities CaF, structure. There are severalways by which an
Some well-established examples of impurity- F - i o n c a n b e m i s s i n gf r o m i t s u s u a lp o s i t i o n :t h i s c a n
causedcolor are given in Table 3. The touchstonesof occur during growth or when energeticradiation dis-
this type of color are the nature of the streak, which placesan F- ion from its usual position to another
usually representsthe color of the pure compound, point in the crystal;we can also createsuch centersby
and the occurrenceof such a mineral in various col- growing fluorite in the presenceof excessCa, or by
ors; it is usually safeto assumethat the lightest color removing some F from a crystal by the application of
representsthe pure mineral. The causeof color can- an electric field.
not necessarilybe ascribed to an impurity just be- Since the crystal must remain electricallyneutral,
causethe impurity is known to be presentin a signifi- an electron usually occupies the empty position to
c a n t a m o u n t . O t h e r i m p u r i t i e s a t a l o w e r produce the F-center or "electron color center" as
concentrationmay be involved, or a color centermay shown in Figure 3B. This unpaired electron can now
exist in excited states, the energy of which is con-
Table 3. Allochromatic coloration by transition metal impurities trolled by the same crystal field factors described
previously; color and fluorescencecan now occur by
Chromlun: Emerald, grossularite, hiddenlte, Cr-
Jade, Cr-tourmallne (green); alexandrlte the same types of transitions as describedabove.
(green-red)i ruby, sp1ne1, topaz (red)
Cobalt : (b1ue )
In the caseof smoky quartz, there is a "hole color
Lusaklte
fron: Aquamarine, tourmal-lne (green); chrysoberyl, center." A necessaryprecursor to this color center is
cltrine, idocrase, orthoclase (ye11ow);
rrgreened amethysttt (green); jade (green- t h e p r e s e n c ei n t h e q u a r t z ( F i g . a A ) o f i m p u r i t y A l 3 +
yelIow-b rown ) substituting for Sin+ions with some alkali (e.9.Na+)
Manganese: Morganite, spodumene, tourmallne (plnk);
andaluslte (green, ye11ow) or a hydrogen ion (H+) nearby to maintain electro-
Nlckef: Chrysoprase, Nl-opal, (green)
Vanadlum: Apophyll1te, V-enerald, r r t s a v o r l t e I neutrality (Fig. aB). Interstitial Al does not serveas a
(green); tralexandrlte-
V-grossularite precursorto form smoky quaftz. Most natural quartz
colored sapphlrerr (green-red)
contains substitutionalAl at the severalhundred ppm
 NASSAU: COLOR IN MINERALS 223

These and other color centersare listed in Table 4.

Note that irradiation with energyas low as ultraviolet
light can produce color centers, and that some of
these colors are not stable to light exposure.These
phenomena can be described on energy level dia-
grams as shown in Figure 5, using the trap concept.
The diagram of Figure 5a could be that of smoky
qvarrz. Starting with colorlessquartz (containing the
(a F substitutional Al precursor) in the ground state A,
radiation is absorbedwith the systemmoving into the
co2+ -^21
co2+ X excited state D. Due to a selectionrule the system
F
must move to state B, as shown in Figure 5b, before
f

it can return to the ground state.

FI
State B, the color center, is a "trapped" state with
F i g . 3 . F l u o r i t e s t r u c t u r e( s c h e m a t i c ) (: A ) n o r m a l , ( B ) c o n t a i n i n g an energy barrier from B to C which must be sur-
an "F-center" where a fluorine ion has been replaced by an
mounted beforethe color centercan disappear.While
electron.
in State B light can be absorbed into other excited
the smoky
level. but this in itself does not produce color, since levelssuch as E and F (Fig. 5c), and cause
color. This color will remain until enough energy is
there are no unpaired electrons. If such quartz is
supplied (e.5. by heating to 400"C) to permit the
irradiated with X-rays, gamma rays, neutrons, etc.,
system to overcome the barrier and return to the
or if it was exposedto radioactivematerialsemitting
ground state, as in Figure 5d. The ground state is
low levelsof such radiation over geologic time peri-
ods, then one of a pair of electrons on an oxygen colorless because the transition to the lowest absorb-
adjacentto each Al can be ejectedfrom its position as ing excitedlevel D requiresa larger energythan that
shown in Figure 4B, leaving unpaired electrons. presentin visible light. In caseof a hole color center,
These
'lhole" defects(the hole indicatesthat there is the trap is not physically located at the center (see
now only one electronwhere there had formerly been below); the energydiagram, applying to the hole-trap
a pair) can now have excited energy levels and light- complex,, is the same as that for an electron color
absorbingtransitionswhich produce the smoky color center. In the energylevel schemeof Figures5a to 5d,
o f q u a r t z ( N a s s a u ,1 9 7 5 b ) . the energy of formation of the color center(A to D) is
The electron displaced in the process depicted in large and the energy for its destruction or "bleach-
Figure 48 becomes"trapped" somewhere else in the ing" (B to C) is also relatively large. The light-stable
quaftz crystal without causing any additional light color centers of Table 4 are of this tvpe.
absorption.If the smoky quartz is now heated (say to
400oC-see Nassau and Prescott, 1977) these dis-
placed electronscan be releasedfrom their traps and
return to the hole center; all electrons are again
paired and the quartz returns to colorless. The
bleachingtemperatureis thus not a characteristicof
the color center itself and often varies.On re-irradia-
RADIATION
tion the color can once again be restored.
Amethyst is a similar color center, involving Fe
instead of Al. Depending on the location and envi-
ronment of the Fe the color obtained on heating
amethystis either yellow (citrine as in most amethyst)
or green(the "greenedamethyst" from Montezuman,
Rio Prado. Brazill Rose and Lietz, 1954). Even
though the yellow or green color after heating is an B
allochromatic transition-metal impurity color, the F i g . 4 . Q u a r t z t r u c t u r e( s c h e m a t i c ) (: A ) n o r m a l ' ( B ) c o n t a i n i n g
s
amethystcolor itself derivesfrom a color center.Irra- Al'* substituted for an Si'+ with an H+ for charge neutrality'
diation after heatingcan recreatethe color centerand Radiation ejectsone of a pair of electrons from an 02- and leaves
the amethyst color. the "hole" color center of smoky quartz.
224 NASSAU: COLOR IN MINERALS

Table 4. Color centers Phosphorescence involves an even smaller bleach-

Essenllallv
ing energy B to C than that necessaryfor light bleach-
stable to 11ght: Amebhyst, ftuorlte
(purple); smoky.quartz (brown to black); ing. This occurs when the return to the ground state
lrradlated dlamond (green, ye1low, brown, black,
blue, pink)i some natural and lrradlated topaz
involves the emissionof light of energy correspond-
(b1ue ) ing to the C to A return to the ground stateand when
Rapldly f?dlng on light exposure: Maxixe-type
beryf (intense blue); some irradiated toDaz the"thermal excitationspresentat room temperature
(brown); lrradiated sapphire (ye1low); uitra-
violeL light lrradiared hackmanite ( p u r p l e ) are sufficient to carry the system slowly over the
Other colors probably due to color centers: barrier at C. Phosphorescenceis shown by some
zlrcon (brown); calcite (ye11ow); barite,
i; fluorite, calcite, diamond, elc. Fluorescence and
celestite (b1ue); pleochroic haloes 1n fluorite,
mlca, etc,
phosphorescence"lifetimes" grade into each other;
rapid phosphorescence with a very small barrier may
The schemeof Figure 5e illustratesa systemwith a seem to be merely fluorescence. If the trap BC is a
large formation energy but a small bleachingenergy. little deeper, thermoluminescence can result;the fluo-
Since B to C is here within the range of energies rescence occurs at a fairly well-defined temperature
presentin visiblelight, the systemwill bleachby light- on heating, when the thermal excitation permits rapid
energy-inducedleakageover the barrier C at the same leakage out of the trap. Phosphorescence and
time that color is observedfrom absorption into E thermoluminescence are not confined to color centers
and F. This state of affairs applies to most of the (any more than fluorescenceis), but the energylevel
light-bleachingcolor centers listed in Table 4. For schemescan neverthelessbe representedas in Figure
example, Maxixe-type deep blue irradiated beryl 5 whereversuch phenomenaoccur.
(Nassauet al., 1976)and brown irradiated topaz will The bleaching processin the caseof a hole center
both fade after some days exposure to light or on involves the releaseof electronstrapped elsewherein
heating for a short period to 100 or 200.C. When the crystal. Since these electronsmay be trapped at
kept in the dark, the color will persist over many various sites,there may be a broad bleachingrangein
years(more than 60 yearsin the caseof Maxixe beryl; a crystal and also different bleaching temperatures in
Nassau et al., 1976). different specimensas was, in fact, observed in a
If both formation and bleachingenergiesare low, seriesof smoky quaftz specimens(Nassau and Pres-
the diagram of Figure 5f applies, for example, to cott, 1977). This implies a variation in the barrier
p h o t o c h r o m i ch a c k m a n i t e .I n t h i s c o l o r l e s sm a t e r i a l height C, although the color center itself (8, E, F in
even ultraviolet light with energy lessthan 5eV (note Fig. 5) remains the same.
that gamma rays range up to over I million eV) is The touchstone for a color center is bleachingby
sufficientto be absorbedvia level D and produces a light or heat and reformation of the color on irradia-
red color center (B, E, F) which bleachesrapidly on tion. This is not foolproof, however (compare the
e x p o s u r et o l i g h t . blue to green change in beryl discussed in the final
section), and years of investigationsemploying opti-
cal spectroscopy,paramagneticresonance,and other
techniques have been required to characterizefully
the few color centers which are well understood.

The molecular orbital formalism

This applies to severaldifferent types of situations
where electrons are involved which are not simply
located on single atoms or ions as in the crystal field
formalism, but must be consideredto be present in
multi-centeredorbits. The resultsvary dependingon
whether metal-metal, metal-nonmetal,or nonmetal-
nonmetal centersare involved. In the latter two cases
the type of bonding is usually predominantly co-
LARGE AD. SMALL AD,
SMALL 8C SMALL 8C valent. The result of a molecular orbital treatment is
Fig. 5. Energy-level schemes showing "traps" involved in similar to that of a crystal field treatment; both for-
p h o s p h o r e s c e n c e .t h e r m o l u m i n c e s c e n c e .and bleaching color malisms result in a set of energylevelsand associated
centers. transition probabilities. Some examplesare summa-
 NASSAU: COLOR IN MINERALS 225

rized in Table 5. An introduction to molecular orbital Table5. Colorscausedby molecularorbital transitions

theory has been given by Ballhausen and Gray
A. Me!al-metal charge transfer:
Q96a); charge transfer has been reviewedby Robin -
p.1*-p25-7psj--Fet-: CordJerite, vlvlanite
a n d D a y ( 1 9 7 0 )a n d o r g a n i cc o l o r b y M a s o n ( 1 9 7 0 ) . (bIue ) , magnetlte, etc. (black? )
p"2+-1,14+7pu3*-rt3*r Kyanite, blue sapphire
M etal-metal charge transfer ( Dfue J
y.2+-yh4+7yr3*-M.3*r Manganlbe, blxbyite, etc.
Consider blue sapphire, a variety of corundum, (b1ack? )
Alros, with both Fe and Ti impurities. Each of the E Metal-nonmetal charge transfer:
-
impurity ions can exist in two valence states, and Crot-ot-: crocoite ( orange )

thereforein two combinations:(a) Fe2+and Tin+;and v5*-o2-, vanadlnite (orange)

(b) Fe3+and Ti3+. A single electroncan be causedto Me!aL-Sulfur: Pyrlte' narcaslte, etc. (see
band-gap semlconductors )
transferfrom the Fe to the Ti by light absorption and
C. Eleetrons not on metal ions:
back again. Since state (b) has a higher energy than in lapls 1azul1 (b1ue)
sl: Lazurite
state (a), the transition from (a) to (b) involves the r
J
electrons: Graphite (b1ack)
absorption of energy,producing a broad intenseab- Orqanic piAments: amber, lvory (brown); coral
(pink, green' blue, black);
sorption band at the red end of the spectrum and (rJa, lr-ac[); pearl
(brown to black)
b1!umen, flgnlte, ebc'
resulting in the well-known deep blue color. The Fe-
Ti distances are 2.65A in the c direction, along the
optic axis, but 2.79A in the perpendiculardirection; to the central metal, i.e. O - Cr in this case. The
therefore, the observed dichroism is expected. result is a broad absorption at the blue end of the
Another chargetransfer processinvolving only Fe, spectrum and the transmittance of the other wave-
namely (c) Fe'z+* Fe3+and (d) Fe3+* Fe'+ can also lengths,leading to an orange color.
occur. In the caseof sapphire this processproduces Similar remarks apply to vanadinite, wulfenite,
absorption only in the infrared; it is, however, of and scheelite.The last is not colored when pure, since
color significancein the hydrated iron phosphatevi- the charge transfer absorption is not in the visible,
vianite. When fresh, this compound contains only but the bright fluorescenceof scheelitederives from
Fe'+ and is almost colorless.As air-oxidation con- an excited molecular orbital state of the charge trans-
verts some of the Fe2+ to Fe3+,charge transfer be- fer O - W in the WO?* units.
comespossibleand resultsin the deepblue color. The As in crystal field calculations,the point symmetry
iron in vivianite is presentin two different sites,and and geometricdistortions of the moleculargroup and
one of these oxidizes preferentially; there is tri- the strength of bonding are important parameters.
chroism with the deepblue color seenonly with light Charge transfer transitions usually have high transi-
polarized along the direction connectingthe two dif- tion probabilities, thus giving intense colors, and
ferent Fe sites,i.e. along the Fe'+-Fe'+ direction. tend to dominate crystal fleld transition colors when
The color of cordierite has also been attributed to both are present,although they may alsooccur in the
this type of Fe2+-Fe3+chargetransfer.It is likely that ultraviolet part of the spectrum.
most black mineralscontaining Fe, Mn, Ti, e/c.,such Charge transferis also presentin the chalcogenides
as magnetite, ilmenite, manganite, schorl, etc., owe such as pyrite, with S - metal electron displacement.
their color to some kind of chargetransfer. It should The optical propertiesof thesetypes of minerals are,
be mentioned that the disproportionation 2 Fe3+- however, best understood using the band formalism
Fe2++ Feo+could also be involved. discussedbelow.

M etal-nonmetal charge transfer Electrons not on metal ions

A substancesuch as crocoite, PbCrOr, does not
have any unpaired electrons,since Pb2+, Crt+, and
02 all have closed electron shells.There are, how-
ever, CrO?- molecular units which are covalently
bonded, and the bonding electronswithin such units
must be consideredto belong to the unit as a whole,
i.e. to be in "molecular orbitals." A molecular orbital
treatment shows that there are excited states,corre-
sponding to the transfer of electronsfrom the oxygen
There are several minerals the colors of which
are best described by molecular orbitals not in-
volving any metal ions. A particularly striking
color is the deep blue of lazurite (lapis lazuli),
(Na,Ca)r(AlSi)'2024(Sr,SOo), which contains sulfur
molecular units and no unpaired electrons. Many
different possibilities have been discussed,but the
excited levelsof S . molecular units appearto explain
the color (Cotton et al., 1976).
 zlo NASSAU: COLOR IN MINERALS

In graphite there are sheetsof connectedsix-mem- ENERGY

FROM
bered rings of carbon atoms, and the "zr-electrons" FERMI
can move more or lessfreely along thesesheets.The SURFACE
result is strong light absorption, specular reflection,
and anisotropic electrical conductivity, somewhat
similar to that in metals, as discussedbelow.
Lastly, there are a number of mineral substances
which owe their color to organic pigments, many of
t
ENERGY
biological origin. Some of theseare listed in Table 5
and here color, once again, originates in molecular
orbital type of transitions within covalently-bonded
m o l e c u l a ru n i t s .
A N U M B E RO F _ B
The band theory formalism ELECTRONS
The crystal field and molecular orbital theories F i g . 6 . B a n d d i a g r a m o f a typical metal (A) showinglight
apply to electronslocated on ions, at defects,and on a b s o r p t i o n t r a n s i t i o n s i n ( B ) .
groups of atoms. Band theory treats electronsas be-
longing to the crystal as a whole and appliesto a wide vidual energy levels are broadenedinto "bands," as
range of materials, as indicated in Table 6. General shown in Figure 6,4,.At each energy there is room for
accountsof band theory can be found in solid-state only so many electrons(the densityof states)and the
physics texts such as that by Kittel (19i.6). availablenumber of electronsfills the band structure
up to the "Fermi surface." In most metals there is
The band theory of metals
one continuous band extendingto high energies,but
In a crystal of a metal such as copper or an alloy the density of states is different at different energies.
such as brass,each metal atom contributes its outer The surfaceof such a metal can absorb light of any
electrons,those usually involved in chemical bond- energy,with electronsfrom the filled part of the band
ing, to a joint pool. These electronsare free to move being excitedup to empty parts of the band, as shown
throughout the whole crystal while being constrained in Figure 68. Most of this light is immediately re-
only by the periodic potential associatedwith the emitted at the surface (the metallic luster), but the
atoms in the crystal structure. The relatively free efficiency of this processdepends on the variation of
movement of such a pool of electrons in a metal the density of state with energy, i.e. the shape of the
results in the high electrical and thermal con- band above the Fermi surface. This produces the
ductivities as well as in the metallic luster and specu- color differencesamong metals and alloys such
as
lar reflection.In a typical metal, there are more than those listed in Table 6. Since there are still selection
1023electrons per cm3, all essentiallyequivalent to rules (basedon which energylevel it was that broad-
each other, i.e. "degenerate."The result of quantum ened into the various parts of the band), the actual
considerationsapplied to such a systemis that indi- identification of the shape of the band with the color
of the metal is very complex.
Table 6. Band theory colors
Band theory and semiconductors
Col9.uctors (mqLal1ic color and luster)
There is a large group of minerals where the bond-
f,rements: uopper, gold, iron, s11ver, mercury, ing is predominanfly covalent (electron sharing) and
etc.
Al1oys: Analgam, irldosmlne, meteoritlc nlckel_ where the average number of bonding electrons is
aron
S e m i c o n d u c t ors
exactly four per atom. This is the casefor elementsof
--T6ii6frIEEfrErgap:
Altalte, galena (opaque, - gray the fourth column of the periodic table such as dia-
Lo black ) ; cobalt i te, narcasiLe, pyr:te,
smaltite (opaque, mera-L1ic) mond, IV-IV compounds such as moissanite(SiC),
Mediun-band-gap: proustite,
.Cinnabar,
(orange); greenockiLe,
pyrargyrite II-VI compounds such as greenockite (CdS), and
\reo); reargar orplnent,
sulphur (ye1low) many others. Band theory applied to such materials
Wlde-band-gap: Diamond, sphaleriLe, zincite
( colorl ess ) showsa gap betweentwo separatebands,the lower of
Wige-band-gao. semlcg+ductors with inpurLties which, the "valence band," is exactly filled with elec-
uonor lnpurlty: Nltrogen in diamond (yellow)
Acceptor inpurity: Boron in dlanond (tlue) trons, while the upper, the "conduction band," is
completelyempty. As shown in Figure 7A, the Fermi
 NASSAU, COLOR IN MINERALS 227

surfaceis thus at the top of the valenceband and the CONDUCTION

BANO
space between the top of the valenceband and the llt
bottom of the conduction band is the "band-gap," ENERGY

the energy of which is usually designated,E". t

Light absorption occurs as in Figure 78, with ab-
sorption possible at all energiesabove the band-gap
energyE r, but not below. Becauseof limitations asso-
ciated with the band-gap, the luster is not usually
metallic exceptat very small valuesof Er. If the band-
gap is smaller than the visible-light range (Fig. 7C),
then all light energiescan be absorbed and a dark
gray or black color results, as in galena with E, less
than 0.4eV. Such "narrow band-gapsemiconductor"
materials provide rectifying and amplifying charac-
teristics in suitable geometries (silicon and germa-
nium diodes, transistors,solar cells;note that galena
crystals served as point contact rectifiers in the early
"crystal" radio sets).Minerals such as pyrite can also
be treated as narrow-band-gap materials (Pridmore
and Shuey, 1976).lf a narrow-band-gapsemiconduc-
tor is heated to a high enough temperature so that
thermal excitation can bridge the band-gap,then the
electrons excited into the conduction band will cause
the material to behave as a metal.
At the other extreme, "large-band-gap semicon-
ductors" with E, larger than the range of visible light
will not be able to absorb in the visible and will thus
be colorless,as in pure diamond and sphaleritewith
.8, about 5 l/2 eY and 3 l/2 eV respectively.
With an intermediateband-gapof about 2eV, as in
proustite, only red light is transmitted; all other col-
ors have energies larger than E, and therefore are
absorbed,as in Figures 7C and 7D. With a band-gap
of about 2.5eY, as in greenockite,only blue and violet
in Figure 7C are absorbed and the resulting color in
Figure 7D is yellow. The overall sequenceof band-
gap colors is thus black/red/orangelyellow /color-
less.When mixed crystalsof CdS (yellow, Es :2.5eY)
and CdSe (black, E, : l.6eV) are prepared,the inter-
mediate orange and red colors are indeed obtained
(Fig. l2 in color in Nassau, 1975a).Some band-gap
semiconductor minerals are listed in Table 6. Semi-
conducting ore minerals are the subject of a book by
S h u e y( 1 9 7 5 ) .
I mpurities in semiconductors
Medium- and large-band-gapsemiconductorsdo
not conduct electricity at room temperature when
pure, since the ambient excitation is insufficient to
bridge the band-gap. The presenceof certain impu-
rities can affect both the conductivity as well as the
light absorption.
^!ig.Tl"'?s* B c o
F i g . 7 B a n d d i a g r a m s o f a t y p i c a l s e m i c o n d u c t o r( A ) , s h o w i n g
l i g h t a b s o r p t i o nt r a n s m i s s i o n s( B ) ; t h e d e p e n d e n c eo f c o l o r o n t h e
e n e r g yo f l i g h t i n ( C ) l e a d st o t h e c o l o r r e m a i n i n g f o r v a r i o u s b a n d
g a p e n e r g i e si n ( D )
Consider nitrogen impurity atoms substituting for
carbons in a diamond crystal (8, about 5 l/2 eY).
SinceN has one more electron than C, each nitrogen
donates one extra electron above the Fermi surface
and these "donor" electrons form an impurity level
within the band-gap, as shown in Figure 8A. The
nitrogen impurity level is still 4eV below the con-
duction band (a "deep" donor) and is, in fact, some-
what broad. It can only absorb a little violet at 3eV,
thus giving a yellow color. A typical deep-yellow
diamond may contain one nitrogen atom for every
100,000carbon atoms. The reduction in the band-
gap is insufficient to permit electrical conductivity at
room temperature.
Boron, having one electron less than carbon, acts
in a similar way to produce "acceptor levels" in the
band-gap. These are "hole" levelswhich can accept
electronsfrom the valenceband and are located only
0.4eV above the top of the valenceband, as shown in
Figure 88. Here again the impurity band is not just a
single level, but has a complex structure, resulting in
CONOUCTION
BAND
T-
I
BAND
GAP
i t/?ev
-l-'.*
AB
Fig.8. Banddiagrams diamondand
for (A) yellowN-containing
(B) blue B-contarning
diamond.
228 NASSAU: COLOR IN MINERALS

a blue color. Al, which also has one lesselectronthan
C and is presentin significantamounts in some blue
diamonds, had been previously thought to provide
the acceptor; however, laboratory synthesis showed
that B additions gave blue and that Al did not
(Chrenko, 1973). Since the boron acceptor level is
"shallow," ambient thermal excitation can raiseelec-
trons from the valence band into the acceptor level,
and the resulting holes in the valenceband can then
conduct electricity. Type IIB blue conducting dia-
monds such as the "Hope" contain typically one
boron atom per million carbon atoms.
Blue irradiated diamonds do not conduct electric-
ity, thus providing one distinction betweenthe boron
acceptorand the irradiation-producedblue color cen-
ter in diamond. Gemological testing (Liddicoat,
1977) is required to distinguish between nitrogen-
donor yellow diamond and irradiated yellow dia-
mond.
Color caused by physical optics effects
Electrons are not involved in the last four types of
colors summarizedin Table 7. Theseeffectsare well-
coveredin optical mineralogy and optics texts and an
extendeddiscussionis therefore not appropriate.
With respect to scattering, it may be noted that the
luster in pearl originates in overlapping platelets of
aragonite. The terminology used for adularescence,
aventurescence,schiller, etc. is very confused. The
effect observedin moonstone orthoclaseand also in
some albite (a "floating" white or bluish shimmer) is
usually attributed to a layer structure. Light scattered
from colloidal particles can produce a very similar
effect, as in unhomogenizedmilk or in some moon-
stone imitations made by heat-treating synthetic
Table 7. Pseudochromatic colors caused by physical optics effects
Color Based on Dlsoersion:
rrFirerrln high dispersion gem-stones
h. hd 21F . a^ Fr r i {l a i hd
Color Based on Scatterina:
@s, tigerrs-eye.
Asterism as 1n star corundum, garnet, quar9z.
Lrrstcr as ln naapl f^l l^tad
spinel (presumablyto precipitatea secondphase).It
is therefore possible that the bluish moonstone effect
is caused by light scattering from very fine particles.
The effect is sometimes termed opalescenceand is
also seen in the white to bluish body color of some
opal; it is unrelated to the fire in opal.
One other distinction may need emphasis.The col-
ors deriving from interference in a thin film, as in the
tarnish film on bornite, is the well-known sequence
black/gray / blue-gray/white/yellow / orange/red (end
of t he fi rst order )/vi olet/blue / gr een/ et c. T hesecolors
become washed out along the sequenceand are not
pure spectral colors. The exact color depends only
on the film thickness,the refractiveindex of the film,
and the natureof the incidentlight.
In the case of diffraction there is also interference
of light waves, but from a regular diffraction grating
as in opal. In this casethe colors are spectrallypure
and dependonly on the grating spacingand the angle
of observation. A given region of an opal can diffract
only a limited range of colors. For example,for a flat-
slab of opal with face-centeredcubic sphere packing
with a spacingof 3A,, the color range is green to red;
with 2.5A blue to yellow, and with 2,{ violet to blue-
green. A detailed description of opal has been given
by Sanders (1968) and labradorite is discussedby
Bolton et al. (1966).
Concluding remarks
The effect of irradiation and heat on the color of
minerals well illustrates the danger of rash con-
clusions concerning the origins of thesecolors. Usu-
ally irradiation produces electron or hole color cen-
ters, while heating permits the trapped electronsand
holes to recombine again. The process is reversible
and heat-bleachedamethystcan usually be recolored
by irradiation. An apparently similar change, the
converting of golden or green beryl to blue aquama-
rine on heating can also be reversedby irradiation.
such
Nevertheless,in this case a color center is not in-
as dla-
a+r ^hti ,,h t i +^- ^+^
volved, but merely a valencechangewhich eliminates
a yellow-producing Fe8+absorption at the violet end
of the spectrum on heating; the resulting Fe2+ does
trl a hr,,^l+6

apophyllite, flbrous asbestos, gypum not absorb in the visible and the original color can be
i / < r t i r e n c r \ ^ t ^

Aventurescence as in sunstone, aventurlne alblte,

restored by irradiation (Wood and Nassau, 1968).In
aventurlne quaTtz, schlLler spar, s1lver-sheen the case of pink tourmaline both an electron color
obs1d1an.
Adularescence as in moonstone (bluish), milky opal. center as well as a hole color center have been pro-
Color Based on Interference: posed,but an examination of a large number of speci-
Interference effects in thln f1lms such as the
tarnish film on chalcopyrlte, columbite, and mens by irradiation and heating indicated the occur-
bornite and wlthin the cracks of irls quaruz.
Color Based on Dlffractlon: rence of at least seven differently-behaving pinks
Dlffraction grating produced by perlodlc spacings (Nassau, 1975c). Clearly, much caution must be used
as in opaI, labradorlte, and lr1s agate.
in explaining specificcolors.
 NASSAU: COLOR IN MINERALS
Further complicationsarise from the fact that
morethan one type of color-causing mechanism can
be presentin a mineral.Color may be causedby very
low level impurities in the presenceof much larger
inactive impurities (e.g. the B us. Al in diamond
discussedabove).Adding to suchdifficultiesare the
many early guesses as to color causeswhich, dueto
repeatedcopyingfrom onetext to another,frequently
end up as apparent"facts."
Quantum theory providesthe framework for the
variouscauses ofcolor in crystals,and it is encourag-
ing that this field is presentlyunderintensiveinvesti-
gation,sincecolor is, after all, oneof the moststrik-
ing characteristic of the majority of mineralsand
gems.
Acknowledgments
Helpful discussionswith D. L. Wood and M. B. Robbins of the
Bell Labs and B. M. Loeffier of MassachusettsInstitute of Tech-
nology are gratefully acknowledged.
References
Ballhausen,C. J. and H. B. Gray (1964) Molecular Orbital Theory.
Benjamin,New York.
Bolton, H. C., L. A. Bursill, A. C. Mclaren and R. G. Turner
(1966) On the origin of the color in labradorite. Phys. Status
Solidi, 18, 221-230.
Burns, R. G. (1970) Mineralogical Applications of Crystal Field
Theory. Cambridge University Press,Cambridge, England.
Chrenko, R. M. (1973) Boron, the dominant acceptor in semi-
conducting diamond. Phys Reu., 87, 4560-4567.
Cotton, F. A., J. B. Harman and R. M. Hedges(1976) Calcu-
lations of the ground state electronic structures and electronic
spectra of di- and trisulfide radical anions by thc scattered
wave-SCF-Xa method. J. Am. Chem Soc..98. 1417-1427.
- and C. Wilkinson (1972) Aduanced Inorganic Chemistry,
3rd ed. Interscience,New York.
Frye, K. (1974) Modern Mineralogy. Prentice Hall, New York.
Fyfe, W. S. (1964) Geochemistry of Solids. McGraw-Hill, New
York.
229
Kittel, C. (1976) Introduction to Solid State Physics,5th ed. John
Wiley and Sons,New York.
Liddicoat, R. T., Jr. (1977) Handbook of Gem ldentification, llth
ed. Gemological Institute of America, Los Angeles.
L o e f f i e r ,B . M . a n d R . G . B u r n s ( 1 9 7 6 )S h e d d i n gl i g h t o n t h e c o l o r
of gems and minerals. Am. Scientist, 64, 636-647.
M a s o n , S . F ( 1 9 7 0 ) C o l o r a n d e l e c t r o n i cs t a t e so f o r g a n i c m o l e -
cufes. In K. Venkataraman, Ed., The Chemistry of Synthetic
Dyes, p. 169-221. Academic Press,New York.
Nassau, K. (1975a)The origins of color in minerals and gems.
Lapidary J , 29,920-8, 1060-70, 1250-8, 1521.
- (1975b) A reinterpretation of smoky quartz. Phys. Slatus
Solidi. A29.659-663.
- ( 1 9 7 5 c ) G a m m a r a y i r r a d i a t i o n i n d u c e d c h a n g e si n t h e
c o f o r o f t o u r m a l i n e s .A m . M i n e r a l . , 6 0 , 7 1 0 - 7 1 3 .
- a n d B . E . P r e s c o t t( 1 9 7 7 )S m o k y , b l u e , g r e e n i s h _ y e l l o w
and
other irradiation-related colors in quartz. Mineral. Mag., 41,
301-312.
and D. L. Wood (1976) The deep blue Maxixe-type
color center in beryl. Am Mineral., 6l, 100-107.
Pridmore, D. F. and R. T. Shuey (1976) The electrical resistivity of
galena, pyrite, and chalcopyrite. Am. Mineral, 6l,248-259.
Robin, M. B. and P. Day (1970) Mixed valencechemistry-a
survey and classification. In H. J Emeleus and A. G. Sharpe,
Eds., Aduances in Inorganic Chemisty and Radiochemistry,Vol.
10, p 247-422. Academic Press, New York.
Rose, H and J. Lietz (1954) Ein griin verfiirbbarer Amethyst.
N alurwissenschaften, 4 l, 448.
Sanders,J. V. (1968) Diffraction oflight by opals.Acta Crystallogr ,
A24,427-434.
Schulman, J. H. and W. D. Compton (1962) Color Centers in
Solids. Pergamon Press, New York.
Shuey, R. T. (1975) Semi-ConductingOre Minerals Elsevier, Am-
sterdam.
W h i t e , W 8 . , R . R o y a n d J . M . C r i c h t o n ( 1 9 6 7 )T h e a l e x a n d r i t e
effect: an optical study. Am. Mineral , 52,867-71.
W o o d , D . L . ( 1 9 6 9 ) S p e c t r a o f i o n s i n c r y s t a l s .I n S . N u d e l m a n
and S. S. Mitra, Eds., Optical Properties of Solids, p. 571-593.
P l e n u m P r e s s ,N e w Y o r k .
- znd K. Nassau (1968) The characterization of beryl and
emerald by visible and infrared absorption spectroscopy. lz.
Mineral. 53, 777-800.
Manuscript receiued, June 20, 1977; accepted
for publication, December 2, 1977.