Content uploaded by Adrian A. Finch
Author content
All content in this area was uploaded by Adrian A. Finch on Apr 30, 2015
Content may be subject to copyright.
ORIGINAL PAPER
A. A. Finch ÆD. E. Hole ÆP. D. Townsend
Orientation dependence of luminescence in plagioclase
Received: 31 October 2002 / Accepted: 12 April 2003
Abstract The orientation dependence of the lumines-
cence of a well-characterized plagioclase crystal at
room temperature and 40 K is reported. A beam of
H
+
ions was used to provide the excitation. Ion beam
luminescence provides emissions effectively from the
bulk of the material, and therefore minimizes the
contribution to the luminescence from atypical regions.
The intensity of the luminescence is strongly orienta-
tion-dependent. The intensity and photon energy,
particularly of the red/infrared and yellow emission
bands, vary significantly. We interpreted this as
resulting from Fe
3+
and Mn
2+
activator ions,
respectively, on crystallographic sites with low point
symmetry. An emission at 860 nm was also signifi-
cantly orientation-dependent. The blue luminescence
showed the least variability. At room temperature, a
350 nm near-UV emission was noted, whereas at 40 K,
emissions were at 240, 260, 300 and 340 nm. UV
emissions may result from Na
+
diffusion along inter-
faces within the plagioclase, notably albite-law (010)
twins. This variability has significant consequences for
the use of single-crystal quantitative luminescence
techniques. We have also studied the dependence of
the peak intensities and profiles during prolonged ion
beam bombardment with heavier (He
+
) ions. Broad-
ening of the red-infrared emission is interpreted as
reflecting growing amorphization of the sample.
Keywords Feldspar ÆIonoluminescence ÆIon beam
luminescence
Introduction
Luminescence is light emission when energy is deposited
into a material. The means of delivering energy can be
divided into two broad types: stimulation, where the
magnitude of the incident energy is less than that of the
emitted light, and excitation, when the incident energy is
greater. Forms of stimulation include heat (thermolu-
minescence, TL) and laser light (optically stimulated
luminescence, OSL); excitation includes electron beam
irradiation (cathodoluminescence, CL) or X-irradiation
(radioluminescence, RL). Whatever the nature of the
incident energy, the energy cascades associated with the
subsequent luminescence are often similar, and hence
different forms of excitation/stimulation explore subtly
different aspects of luminescence centres within materi-
als. Many minerals are luminescent, and applying these
signals to solve geological problems is an important and
expanding field. For example, OSL and TL of frame-
work silicates are the basis for luminescence dating of
Quaternary sediments, and CL petrography is a wide-
spread mineral-prospecting tool.
In low-symmetry materials, structural anisotropy
causes the physical properties of a crystal to be orien-
tation-dependent. Such dependence within individual
grains is often ignored when studying fine-grained
rocks or powders, in which many crystals are randomly
oriented, and the response is therefore averaged over
all possible orientations. However, studies of single
grains of anisotropic minerals or rocks with fabrics (in
which minerals are aligned) will demonstrate orienta-
tion dependence. Feldspar, the subject of the present
study, has monoclinic or triclinic symmetry, with the
result that the physical properties of the crystal are
independent in all three dimensions. For that reason,
the luminescence of feldspar single grains will depend
(to some degree) on orientation of the crystal. Changes
in luminescence as a function of orientation can also
provide insights into the nature and origins of lumi-
nescence centres.
Phys Chem Minerals (2003) 30: 373–381 Springer-Verlag 2003
DOI 10.1007/s00269-003-0327-1
A. A. Finch (&)
Centre for Advanced Materials and School of Geography &
Geosciences, University of St Andrews, Irvine Building,
St Andrews, Fife KY16 9AL, UK
e-mail: aaf1@st-and.ac.uk
D. E. Hole ÆP. D. Townsend
School of Engineering & Information Technology,
University of Sussex, Pevensey Building, Brighton,
Sussex, BN1 9QH, UK
The present study examines how the intensity and
photon energy of the major emission bands change
normal to each of the {100} form faces. We measure the
orientation dependence of the luminescence of a well-
characterized plagioclase (AF/96/4), which demon-
strates all four major luminescence bands found in
feldspar perpendicular to (100), (010) and (001) faces.
Analysis of this sample allows us to examine simulta-
neously the orientation dependence of the major emis-
sion bands in feldspar.
Choice of incident energy
Excitation/stimulation for light-emission studies can be
achieved by a variety of routes. For studying all lumi-
nescence centres in solids, excitation with high incident
power density is employed, and, for ease of analysis, the
primary excitation needs to be a focused, parallel beam.
This rules out many forms of incident energy, since TL
and OSL, for example have low power densities, and
conventional RL uses an unfocused source. Cathodo-
luminescence (CL) is a widely used method whereby an
electron beam excites the sample. Because electrons are
charged, the beam can be focused and, by increasing the
acceleration potential and current, large incident power
densities (e.g. 10
2
Wm
)2
) can be achieved. However, the
majority of the energy from electrons is deposited within
the outermost submicron-thick layer of the sample,
providing only luminescence characteristic of the surface
region. Comparisons of RL and CL responses in alkali
feldspars (Brooks et al. 2002) have shown that feldspar
surfaces provide luminescence that is highly atypical of
the bulk. For anisotropic media, such as feldspar, the
nature of the surface is likely to be orientation-depen-
dent.
Recently, experiments have been performed using ion
beams as an excitation source for luminescence (iono-
luminescence, IL, or ion beam luminescence, IBL). Ion
beams have high incident power densities, and can be
focused to analytical spots typically of areas of mm
2
or
less. Ion beams effectively penetrate the sample bulk to,
e.g., tens of lm depths in contrast to CL, in which the
incident beam is stopped within a few nm of the surface
and provides a surface-dominated response. For H
+
or
He
+
ions at MeV energies (the conditions used in the
present study, see below), >90% of the incident energy
is deposited fairly evenly as electronic excitation along
the ion track. At the end of the track (e.g. in a 1-lm
width zone), nuclear collision accounts for the remain-
ing <10% of the energy. Most of that energy is dissi-
pated in the form of atomic displacements, which give
rise to thermal rather than electronic processes, and
therefore contributes relatively little to the overall
luminescence. Therefore relatively little of the IL signal
derives from regions atypical of the bulk, either at the
surface or at the end of the track. IL is therefore
an important form of excitation for luminescence
studies, since it is high-energy, focusable, and provides a
response dominantly from the bulk. Homman et al. (1995)
first performed IL on plagioclase and, more recently,
Brooks et al. (2002) compared RL, IL and CL responses
from microperthitic alkali feldspar, comparing and
contrasting the three excitation methods. IL is particu-
larly appropriate for understanding the orientation
dependence of luminescence, the focus of the present
study. A disadvantage of IL is that some sample deg-
radation results from the ion beam interaction. The
sample undergoes structural damage particularly at the
end of the ion track, and this can modify the sample,
particularly when using heavy ions at high acceleration
potentials and high beam currents. However, light ions
(such as H
+
) at moderate incident power densities (e.g.
10
4
Wm
)2
) cause little sample degradation in feldspar
over the time-scales of the experiments described below
(<10 h). Note also that sample damage and modifica-
tion also occur with other forms of excitation, particu-
larly CL.
Material studied
The sample studied (AF/96/4) is a cleavage fragment from a
remarkable 1 m
3
plagioclase megacryst, part of an exceptionally
coarse gabbro unit in the Klokken centre, Gardar Province, South
Greenland. Parsons (1979, 1981) gives descriptions of the geologi-
cal context of the unit, and Emeleus and Upton (1976), Upton and
Emeleus (1987) and Upton et al. (2003) review Gardar geology as a
whole. The Klokken intrusion has been dated at 1166.6 ±
1.2 Ma (Burgess et al. 1992). The plagioclase was sampled as
cleavage fragments 15 ·10 ·10 in size, from which small frag-
ments (e.g. 0.5 ·0.5 ·0.5 cm) were broken in the lab. The pla-
gioclase is dark bottle green in hand specimen with obvious
repeated (010) twinning. In thin section, the feldspar is perfectly
transparent (pristine) with little evidence for late-stage hydrother-
mal alteration. It has a nominal composition An
51
Ab
45
Or
4
determined by X-ray fluorescence, but the feldspar contains small
inclusions of magnetite, olivine, biotite and apatite (Hobbs 1999),
and these (particularly biotite mica) may have exaggerated the
apparent Or component. By X-ray diffraction the sample conforms
to the pattern of a triclinic, ordered intermediate plagioclase.
However, the lm-scale multiple (010, albite) twinning gives the
sample an effectively monoclinic symmetry when analyzed on the
mm scale (as in the present study), since at this scale the responses
of the left-and right-handed triclinic twin domains are averaged.
Electron diffraction using the transmission electron microscope
shows pairs of ediffraction spots typical of intermediate plagio-
clase. Additional weak fsatellites are observed around adiffraction
spots. The composition is close to the solvus of the Bøggild inter-
growth, but there is no evidence for Bøggild-type intergrowths in
TEM images, and there is no Schiller effect (indicative of inter-
growths of the size of the wavelength of light) in hand specimen.
There are, however, occasional thin disc-like precipitates 10–40 nm
long and 2 nm wide, which may be exsolved K-feldspar (M.A.
Carpenter, personal communication, 1998). AF/96/4 is therefore a
relatively K-rich e
2
-plagioclase.
Analytical methods
IL investigations were carried out under vacuum using the 3 MeV
Van der Graaf accelerator facility at the University of Sussex
generating a spot size of 0.25 cm
2
. For investigations of tem-
perature dependence, a proton beam at 0.95 MeV generated the
luminescence with an ion current of 100 nA. The beam generated
374
an incident power density of 2kWm
)2
.H
+
ions were used to
minimize the collision damage caused to the sample. IL data were
collected every 10 K between 40 and 160 K and then at 20 K
intervals up to 300 K. Cryogenic temperatures were achieved using
an 8 W He compressor, and thermal contact between the com-
pressor head and the sample was maintained using Ag dag. Under
the conditions outlined above, the incident beam deposits <0.1 W
of power, and analysis of standard materials with strong temper-
ature dependence do not show significant (e.g. <5 K) heating at
the sample surface. When ramping the temperature, the sample was
irradiated only whilst data were being acquired, to reduce the
exposure. A typical ion dose during an experimental run was
5·10
15
H
+
ions. Repeat analyses of the same samples show
excellent reproducibility in the luminescence profiles, although
differences in the positioning and size of different samples (i.e. the
way in which the position is optimized with respect to the system
optics) causes some variation in intensity. IL analysis was per-
formed on freshly exposed feldspar cleavage surfaces for (010) and
(001) and a fracture surface for the (100) face — the morphology of
the cleavage fragments allows the orientations to be identified
unambiguously. The results are the averages of typically three runs
on each face. Ideally, we seek to arrange the primary beam and the
secondary optics exactly normal to the crystal surface, but such a
geometry would cause interaction between the primary beam and
the earthed secondary optical system. We therefore oriented the
sample such that the ion beam encounters the surface of the sample
at an angle of 22from the normal to the sample face, and the
light is collected 22from the perpendicular in the opposite direc-
tion. The incident beam and the emitted light vary by 45. Such a
geometry allows both the incident beam and emitted light to be
close to perpendicular with respect to the sample, whilst also
ensuring that the primary beam and light collection system do not
interfere. A goniometer was not used. Using a standard modelling
package, this geometry gives an estimated penetration range of the
H
+
ions into silica as 10 lm. Light emissions were collected by a
quartz fibre optic coupled to a f/4 SpectroPro 300i monochroma-
tor. The detector used was a Roper Scientific image intensified
CCD camera operated using the WinSpec software package. The
system operates in the range 200–1100 nm by performing two
separate spectral analyses between 200 and 600 and 500 and
1100 nm, respectively, and then matching the spectra in the 500–
600 nm region. Raw intensity counts were typically integrated to
>10 000 for the brightest peak, compared with a background of
200 counts. The intensity data were corrected for system response
against a W lamp by assuming a black-body radiation profile.
Wavelength was calibrated against the emissions of an Hg lamp
and the resolution at the analytical conditions applied corre-
sponded to 2 nm. Sample degradation experiments were carried
out on the (001) face at room temperature, using 2 MeV
4
He
+
ions.
Beam currents were ramped between 50, 1000 and 2500 nA to al-
low a range of integral ion doses between 10
13
and 10
17
ions over
time periods of hours.
Feldspars have been known for several decades to emit light
when excited (e.g. Smith and Stenstrom 1965). The prominent
luminescence bands observed in feldspars, their nature and proba-
bly causes are given in Table 1. In addition to the prominent bands,
Brooks et al. (2002) reported a minor orange band at 620 nm that
they identified as a defect-related structure. Preliminary CL analysis
showed AF/96/4 demonstrates four major luminescence bands in
the red/IR, yellow, blue and near UV spectral regions. Finch and
Klein (1999), Go
¨tze et al. (2000) and Brooks et al. (2002) give
further summaries of feldspar luminescence.
Results and discussion
Sample degradation during H
+
analysis
Ion beam implantation can cause some sample degra-
dation. Brooks et al. (2002) showed that degradation
was observed in alkali feldspars, particularly after pro-
longed implantation with He
+
ions, but that degrada-
tion with H
+
was very slow. In order to monitor sample
damage, we performed comparable IL runs at room
temperature before and after the IL experiments. Fig-
ure 1 shows the emission from the (010) cleavage face
before and after implantation by 10
17
H
+
ions. Note
that this dose is greater than that experienced by samples
in the present study (4·10
15
ions). The sample does
show subtle changes in the relative intensities of the
emissions as a result of the implantation. The greatest
change is observed around 750 nm, in which there is a
reduction in intensity during irradiation by <10%. If
the sample is allowed to recover for 18 h, there is no
modification of the spectrum. Similar responses are ob-
served for the beam dependence of other orientations. IL
therefore does show some minor modification of the
sample during implantation, but this is very small, par-
ticularly in comparison with the orientation dependence
at the heart of the present study. It is important to note
that such modification is also observed using other
forms of excitation, such as CL. We conclude that IL
provides conditions with stabilities as a function of time
that are comparable to, or better than, other forms of
excitation.
Light emission at room temperature
The luminescence response at room temperature shows
broad bands in the red/IR (700–820 nm), yellow
(560 nm) and blue (420 nm) spectral regions with
evidence for emission in the UV (350 nm) (Fig. 2a).
Table 2 provides estimates of the peak positions and
intensities as a function of temperature and sample
orientation. The red/IR peak is the most intense, and the
asymmetry of that band provides evidence for an extra
emission on the long-wavelength side of the IR peak,
with a maximum at 860 nm. In addition, there is a
Table 1 Summary of major feldspar luminescence bands and possible origins
Band Wavelength /nm Energy /eV Possible origin Reference
UV 280 4.42 Strain and/or ionic diffusion Garcia–Guinea et al. (1999)
UV 350 3.54 Strain and/or ionic diffusion Garcia–Guinea et al. (1999)
Blue 400–450 3.00–2.75 O Paramagnetic defect
Eu
2+
activation
Finch and Klein (1999)
Go
¨tze et al. (1999)
Yellow 580 2.14 Mn
2+
activation Telfer and Walker 1976)
Red/IR 750 1.65 Fe
3+
activation Brooks et al. (2002)
375
small emission band at 500 nm. Figures 2 and 3 show
the luminescence normal to the three {100} form faces at
room temperature and 40 K, respectively. The first
diagram in each shows the intensity as a function of
wavelength, the second the log(intensity) as a function of
photon energy. Differences between the intensities and
peak positions of many emissions are observed. For
example, in the room-temperature spectra (Fig. 2), a
shift in the peak position of the red/IR emission is
present as the orientation is changed. Whereas normal to
the (001) face the maximum of this broad band occurs at
757 nm, it occurs at 765 nm for the other two. In
addition, differences in the position of the 555 nm
peak are evident. Perpendicular to the (100) plane, the
peak position has shifted to shorter wavelengths
(550 nm). Differences between the emissions in the
blue and UV regions are most obvious in a graph of the
photon energy versus log(corrected intensity) (Figs. 2b,
3b). Note that near-UV luminescence (3.53 eV,
350 nm) is evident.
Fig. 2a The luminescence spectra normal to the three {100} form
faces as corrected intensity versus wavelength for samples at room
temperature. bRepresentation of the data in a(room temperature),
presented as log(corrected intensity) versus photon energy
Fig. 1 Comparison of the IL spectra perpendicular to the (010) face
before and after the ion beam experiment. Note that there are subtle
differences in the spectra caused by the implantation, but these are
small compared with the changes discussed in the text as a function of
orientation and temperature. The sample modification is typical of all
the orientations
Table 2 Summary of the lumi-
nescence emissions observed in
AF/96/4 as a function of or-
ientation and temperature
Temperature Peak position Peak intensity Comments
Room temperature 350 nm 3.53 eV 3·10
7
All faces
415 nm 2.99 eV 1.6 ·10
8
All faces
550 nm 2.25 eV 3.2 ·10
8
(001) face
560 nm 2.21 eV 3.3 ·10
8
(100) face
560 nm 2.21 eV 2.6 ·10
8
(010) face
757 nm 1.64 eV 1.0 ·10
9
(001) face
765 nm 1.62 eV 1.6 ·10
9
(100) face
765 nm 1.62 eV 1.2 ·10
9
(010) face
860 nm 1.44 eV 4·10
8
Small peak — difficult to resolve
exact energies and intensities
40 K 240 nm 5.16 eV 8 ·10
6
Strongest from (001) face
260 nm 4.76 eV 4 ·10
6
Strongest from (001) face
300 nm 4.14 eV 8 ·10
6
Strongest from (001) face
340 nm 3.61 eV 5 ·10
7
All faces
415 nm 2.99 eV 1.2 ·10
8
All faces
500 nm 2.48 eV 1.4 ·10
8
All faces
568 nm 2.18 eV 1.3 ·10
8
(100) face
570 nm 2.17 eV 7 ·10
7
(010) face
782 nm 1.58 eV 3.3 ·10
9
(100) face
790 nm 1.57 eV 3.0 ·10
9
(001) face
790 nm 1.57 eV 2.2 ·10
9
(010) face
910 nm 1.36 eV 6·10
8
All faces
376
Light emission at 40 K
The analysis at 40 K (Fig. 3) bears many similarities to
the data at room temperature. The red/IR emission is
over double the intensity, accompanied by an energy
shift, moving from 1.62 to 1.57 eV as the tempera-
ture falls. Similarly, the IR shoulder to the peak at
1.44 eV (860 nm) moves to 1.36 eV (910 nm).
There are subtle changes to peak positions in the yellow
(2.17 eV) region, whereas the blue emission appears little
changed in energy (3.0 eV). Figure 3b shows the
log(intensity) as a function of photon energy, and is
particularly useful in showing detail in the near-UV.
Emissions at 4.14, 4.76 and 5.16 eV (corresponding to
300, 260 and 240 nm, respectively) are absent at room
temperature (Fig. 2b), but present at 40 K (Fig. 3b) and
particularly strong perpendicular to the (001) face.
The luminescence as a function of orientation is
markedly different at both temperatures. The red/
infrared emission at 40 K varies in intensity by 30%
between the faces analyzed, with the emission weakest
normal to the (010) face; some small shifts in the peak
position may also occur. This is a smaller variation
than observed at room temperature. There are also
changes in the intensity and position of yellow and UV
bands. The orientation dependence appears least
marked in the blue.
Detailed analysis of the intensities of luminescence
from the three planes is more complex. Variations in
sample size and positioning in the sample holder
introduce variability into the intensities of light re-
corded, and so comparisons of absolute emission
intensities are difficult. The profiles of the luminescence
normal to the (100) and (010) faces are similar and can
be broadly matched by magnifying the raw (010)
counts by 30%. The data perpendicular to these fa-
ces, therefore, although they have different intensities,
have near-identical profiles. In contrast, the profile
normal to the (001) face is quite different. No scaling of
the (001) profile allows it to be matched with the
response from either of the other planes.
Ion beam dose dependence
We examined the dependence of luminescence of one
feldspar face, (001), to ion beam degradation at room
temperature. Brooks et al. (2002) first performed this
type of experiment on feldspar, examining microperth-
itic alkali feldspar. We present here a description of the
ion beam degradation of the AF/96/4 plagioclase.
Briefly, the dependence of luminescence as a function of
ion beam interaction allows important insights into the
nature of luminescence centres and their stability to
high-energy electronic and structural modification. Note
that for these experiments, we used heavier ions (He
+
)
at much greater incident power densities than in the
temperature dependence studies above.
We have considered the intensity of the IR, yellow,
blue and 350 nm UV bands as a function of implanta-
tion dose (Fig. 4). Luminescence intensity falls as the
implantation progresses for all the emission bands. In
addition, there are inflexions in the intensity (labelled A
and B, Fig. 4) which correspond to the points during the
experiment at which the ion beam current was increased.
So that the total implantation dose can be achieved
within convenient time scales, we performed the exper-
iments at three beam currents, 50 nA, 1 lA and 2.5 lA.
The points A and B (Fig. 4) correspond to the doses at
which the current was increased. These changes dem-
onstrate that the sample is responding not only to the
total number of ions implanted, but also to the
implantation dose rate. For all but the red/infrared
band, the change from 50 nA to 1 lA is associated with
a jump in the luminescence, whereas at the highest beam
current there is a notable drop for all the emission
bands. This is similar to the behaviour of the red/IR
band observed in microperthite by Brooks et al. (2002).
Brooks et al. considered the possible interpretations of
this dose-dependent behaviour in feldspar. A key idea
presented by these authors was that at higher dose rates,
there was an increased probability of ions arriving close
together in time and space, forming aggregate defect
structures that develop at the expense of the single ion
Fig. 3a The luminescence spectra normal to the three {100} form
faces as corrected intensity versus wavelength for samples at 40 K.
bRepresentation of the data in a(40 K), presented as log(corrected
intensity) versus photon energy. Arrows identify features at 2.2 and
4.2 eV that are different for the (001) data
377
luminescence centres. Such behaviour would be most
likely for luminescence cascades with long decay times.
In the study of AF/96/4, we observe that the dose rate
dependence is strongest in the UV 350 nm and blue
emissions, but less marked for the red and yellow
emissions (Fig. 4). The decay time of the blue emission
in feldspar is relatively short (e.g. 6 ls, Finch and Klein
1999), whereas those of the yellow and red/infrared
bands are orders of magnitude longer (e.g. 2 ms, Finch
and Klein 1999). If that model for the dose-rate depen-
dence were correct, then the clearest dose-dependent
effects would be observed for the red and yellow bands,
with the longest decay times. This is not the case. We
therefore attribute the dose-dependent behaviour re-
ported here and by Brooks et al. (2002) to a combination
of sample charging (deflecting the incoming beam) and
heating. At the highest currents, the incident beam
deposits a power density of 0.1 MW m
)2
. In contrast
to studies at smaller currents, analysis at 2.5 lA caused
significant sample heating. A thermocouple immediately
next to the sample showed heating by 10 K during
experiments at high currents. The dose-rate dependence
of feldspars is the subject of future study (A. A. Finch,
personal communication).
In addition to variability in the intensities of the
emissions, there are changes in the peak profiles. We
observe that the red/IR emission becomes asymmetrical
and moves to shorter wavelengths during the implanta-
tion (Fig. 5). This is similar to behaviour reported by
Brooks et al. (2002), who explained this peak shift in
terms of changes to the coordination sphere of the
activator Fe
3+
ion caused by the ion bombardment.
Comparison of the profiles at 1 ·10
16
and 1 ·10
17
ions
shows no change in the luminescence profile, demon-
strating that the local structural modification is largely
complete by 10
16
ions. The 850 nm band reduces
in intensity with dose, apparently linearly with the
750 nm band, suggesting that the luminescence centres
causing these bands have common features. However, it
is impossible to resolve any developing asymmetry in the
850 nm emission (as might be expected by analogy
with the 750 nm emission), given its proximity to the
dominant IR band.
Discussion
The orientation dependence of particular emissions
provides insights into the nature of the luminescence
centres, particularly when this is coupled with analyses
at room and cryogenic temperatures. Centres with high
point symmetries will give little orientation dependence,
whereas those in low symmetry sites will give emission
energies and intensities that are significantly different as
the crystal is rotated. In addition, the variation in band
energy, intensity and orientation dependence as tem-
perature is changed also provides insights into the way
in which the luminescence cascade couples with the
mineral lattice. We interpret below the observed orien-
tation dependence in each of the four major emission
bands.
UV luminescence
UV TL emissions at 290 and 340 nm were reported in
alkali feldspars by Garcia–Guinea et al. (1999). Na
+
ions in albites (as high, low and mon-albite) have
highly anisotropic thermal displacement ellipsoids at
room and cryogenic temperatures (Previtt et al. 1976;
Harlow and Brown 1980; Smith and Artioli 1986;
Armbruster et al. 1990). Garcia–Guinea et al. (1999)
Fig. 4 Dose dependence of the four main luminescence bands at
room temperature. The blue and red emissions are shown in a,the
yellow and UV bands in b.Thelines marked A and B mark the points
at which the ion current was increased from 50 to 200 nA (A)and
200 nA to 1 lA(B)
Fig. 5 The dose dependence of the profile of the red/infrared emission
as a function of integral dose. The profile shifts towards the red/
orange regions during implantation, but the shift is complete by 10
16
ions
378
suggested that thermally assisted Na
+
ionic diffusion,
predominantly along (010) planes, was responsible for
the 290 nm emission. They also attributed the 340 nm
emission to [AlO
4
]centres on the feldspar framework.
Our data differ somewhat from their observations. We
note a 350 nm (3.53 eV) emission, significantly offset
from 340 nm. The contrasting composition between the
alkali feldspars studied by Garcia–Guinea et al. and the
plagioclase of the present study may explain the de-
creases in the emission energy of this feature. Inter-
atomic distances in An
50
Ab
50
plagioclase are smaller
than in alkali feldspars, and thermally assisted ionic
diffusion will therefore occur at lower energies. How-
ever, we observe little significant orientation depen-
dence in the emission at room temperature, in contrast
to Garcia–Guinea et al. (1999).
The majority of features in the near-UV are
apparent at 40 K. Small but significant peaks occur at
240, 260, 300 and 340 nm (5.16, 4.76, 4.14 and
3.61 eV, respectively); (Fig. 3b). The first three emis-
sions are particularly strong normal to the (001) face,
whereas the 340 nm emission appears less orientation-
dependent. The 340 nm emission at 40 K may be
analogous to the 350 nm band at room temperature,
and the energy shift would indicate a luminescence
centre coupled to lattice vibrations and/or interatomic
distances. Alternatively, the two emissions may derive
from different centres, in which case, the latter may
equate to the 340 nm emission of Garcia–Guinea et al.
(1999). The exact origins and complexity of the UV
emissions remain problematic. Garcia–Guinea et al.
considered ionic diffusion along surfaces or interfacial
boundaries to be responsible for the emission. Loss of
singly charged ions caused the development of [AlO
4
]-
type centres. Highly strained interfacial states are
unrepresentative of the bulk, and they may provide
atypical luminescence, the understanding and determi-
nation of which are extremely difficult. AF/96/4 is
similar to the majority of plagioclases in having re-
peated polysynthetic twinning according to the (010)
albite twin law. The near-UV emissions are weakest in
the (010) orientation. The orientation dependence of
the low-temperature UV emissions suggests coupling
with an anisotropic feature of the structure. Both the
(100) and (001) orientations provide luminescence from
significant numbers of albite twin planes, whereas
(010)-oriented excitation would not encounter signifi-
cant numbers of twin planes at the penetration depth
of the ion beam. This is consistent with the observed
orientation dependence of the luminescence. In addi-
tion, AF/96/4 contains small submicroscopic K-feld-
spar inclusions observed using TEM. These may
provide alternative loci for the development of strain-
related luminescence. UV emissions in feldspars are
highly complex but it is clear they are related to
luminescence associated with interfaces. These are
highly anisotropic features of the feldspar structure
which give rise to luminescence with notable orienta-
tion and temperature dependence.
Blue luminescence
The blue emission in feldspars has been linked both to
Eu
2+
activation (Go
¨tze et al. 1999) and a paramagnetic
oxygen defect on the tetrahedral silicate framework
(e.g. Finch and Klein 1999). Blue luminescence is
common to the majority of framework silicates and
paramagnetic oxygen is the dominant luminescence
centre in the majority of rock-forming feldspars. Unlike
yellow and red/IR emissions, the photon energy of the
blue emission is unchanged as a function of tempera-
ture, occurring at 3.00 eV at both 40 K and room
temperature. The overall intensity changes from room
temperature to 40 K, but the intensity from each face is
broadly similar. This is consistent with a centre unre-
lated to an activator ion on a specific crystallographic
site, but rather a centre with a defect-related origin.
Yellow luminescence
Light emission in feldspars between 540 and 570 nm (the
yellow-green spectral region) is linked to the presence of
Mn
2+
ions substituting for Ca
2+
(Telfer and Walker
1976). Because of its exchange for Ca, yellow lumines-
cence is most commonly observed in plagioclase. Go
¨tze
et al. (2000) reported a linear relationship between Mn
content and luminescence intensity. This luminescence
band has clear orientation dependence (Figs. 2, 3), with
both the peak position and intensity changing. The
yellow emission is brightest normal to the (100) face at
both room temperature and 40 K. The emission was not
detected perpendicular to the (001) face at 40 K.
Luminescence anisotropy in the yellow emission behav-
iour has been inferred for some time, since (010) poly-
synthetic (albite) twinning is readily visible in yellow
luminescent plagioclases using CL petrography. No
correlation between photon energy in the yellow region
and plagioclase composition is reported (Go
¨tze et al.
2000) and this may be, at least in part, a consequence of
the four non-equivalent crystallographic sites in which
Mn
2+
can reside. We observe that the energy and
intensities of the yellow emission are decreased at low
temperatures (Figs. 2, 3). Decreases in photon energy
with cooling follow shrinkage of the unit cell and
therefore indicate a broad correlation with the inter-
atomic bond distances, in particular crystallographic
sites. Such observations suggest that the complexity in
the reported intensity and photon energy of the yellow-
green emission in the literature results from samples of
different compositions showing different degrees of Mn
order, coupled with the orientation dependence of this
band.
Red/IR luminescence
Our data extend to 1100 nm and provide a full profile of
both the near and far sides of the red/IR peak. Red/IR
379
emissions in feldspars are believed to relate to Fe
3+
substituting for Al
3+
on the feldspar framework. Geake
and Walker (1975) and White et al. (1986) identify the
emission as a result of the
4
T
1
to
6
A
1
transition of tet-
rahedral Fe
3+
. Some authors (e.g. Rae and Chambers
1988) have commented that there appears little correla-
tion between the intensity of the red peak and the
amount of Fe determined by electron probe microanal-
ysis (EPMA) or secondary ion mass spectrometry
(SIMS). This may result (at least in part) from the
presence of submicroscopic inclusions of iron oxides
within feldspars, which influence microanalysis but do
not contribute to luminescence. Note also that pairing,
clustering or precipitation of impurities can all reduce
luminescence efficiency. Finch and Klein (1999) studied
the decay times and profiles of red/IR CL of a micro-
perthite. They showed that the decay time of the red
emission in these feldspars could only be modelled
successfully by the superposition of two or more dif-
ferent emissions, and demonstrated shifts in the peak
positions between samples. In detail, the variation in
the peak profiles and energies of the red/IR emission
may relate to the presence of Fe
3+
across crystallo-
graphically inequivalent sites (Finch and Klein 1999;
Brooks et al. 2002). In plagioclases, red/IR emissions
are reported to vary from 745 to 690 nm as a func-
tion of anorthite content (Go
¨tze et al. 2000; Krbetschek
et al. 2002). AF/96/4, with a composition in the
middle of the plagioclase series, might be expected
therefore to have an emission at 720 nm. However,
the red/IR maximum in our data (755 nm) is beyond
even the range of the most An-rich feldspars quoted by
Go
¨tze et al. (2000).
The peak profile in energy (Figs. 2b and 3b), which is
normally Gaussian for a single emission band, is skewed
to the low-energy side for the IR emission in AF/96/4,
reflecting the presence of a second band at lower energy.
Krbetschek et al. (2002) and Go
¨tze et al. (2000) both
reported an emission 850 nm in potassium feldspars
(adularia, orthoclase and microcline) which corresponds
well to our observations, although that emission is
not previously reported from plagioclase. Those authors
were unable to assign a specific defect centre to the
band.
Implications for quantitative luminescence studies
The present study identifies significant orientation
dependence in the intensity and (for some emissions)
photon energy of the luminescence. For example, the
intensity of the IR band varies by 60% depending on
orientation at room temperature, and small shifts in
peak position are also observed. Quantitative methods
that utilize luminescence in single grains of low-
symmetry minerals must accommodate orientation
dependence. For example, Habermann (2002) proposed
that quantitative CL might be used as a chemical
analytical tool with precision and limit of detection
equivalent to particle-induced X-ray emission (PIXE).
Habermann illustrated this suggestion with studies of
the yellow emission of plagioclase as an indicator of
Mn contents, an emission we demonstrate here as
having significant orientation dependence. The preci-
sion of such analysis would be highly dependent on
knowing, and being able to accommodate, the orien-
tation of the crystal. In addition, quantitative analysis
of OSL and TL is used for dating of Quaternary
sediments and archaeological artefacts. Such studies are
normally carried out on aliquots of samples containing
significant numbers of grains. However, in recent years,
the development of single-grain OSL readers has
allowed luminescence studies to be performed on
individual crystals. In principle, anisotropy of the
luminescence response, if significant, might strongly
influence the luminescence measured from grains lying
in a variety of random orientations. Even assuming
that orientation does not affect the regeneration doses
(something that remains to be determined), certain
orientations would provide brighter luminescence and
therefore more precise age estimates. Precision in dat-
ing might be enhanced by rotating the single crystals
into particular orientations before TL or OSL analysis.
Anisotropy in low-symmetry minerals is clearly a
problem that must be addressed fully for highly precise
measurements of the luminescence properties of single
crystals to be achieved.
Acknowledgements The present work was supported by NERC.
A.A.F. acknowledges tenureship of a BP/Royal Society of Edin-
burgh Fellowship. Michael Carpenter (University of Cambridge) is
thanked for assistance with the TEM characterization of AF/96/4.
The Carnegie Trust for the Universities of Scotland and the Percy
Sladen Trust have contributed to the costs of fieldwork in Green-
land. The manuscript has benefited from the reviews of Karl
Malmqvist and two anonymous reviewers.
References
Armbruster TH, Burgi HB, Kunz M (1990) Variation of dis-
placement parameters in structure refinements of low albite.
Am Mineral 75: 135–140
Brooks RJ, Finch AA, Hole DE, Townsend PD, Wu Z-L (2002)
The red to near-infrared luminescence in alkali feldspar. Con-
trib Mineral Petrol 143: 484–494
Burgess R, Kelley SP, Parsons I, Walker FDL, Worden RH (1992)
Ar
40
–Ar
39
analysis of perthite microtextures and fluid inclusions
in alkali feldspars from the Klokken syenite, south Greenland.
Earth Planet Sci Lett 109: 147–167
Emeleus CH, Upton BGJ (1976) The Gardar Province of South
Greenland. In: Escher A, Watt WS (eds) Geology of Greenland.
Grønlands Geologiske Undersøgelse, Copenhagen: pp 152–181
Finch AA, Klein J (1999) The causes and petrological significance
of cathodoluminescence emissions from alkali feldspars. Con-
trib Mineral Petrol 135: 234–243
Garcia-Guinea J, Townsend PD, Sanchez-Muoz L, Rojo JM (1999)
Ultraviolet-blue ionic luminescence of alkali feldspars from
bulk and interfaces. Phys Chem Miner 26: 658–667
Geake JE, Walker G (1975) Luminescence of minerals in the near-
infrared. In: Karr C (ed) Infrared and Raman spectroscopy of
lunar and terrestrial minerals. Academic Press, New York, pp
73–89
380
Go
¨tze J, Habermann D, Neuser RD, Richter DK (1999) High-
resolution spectrometric analysis of REE-activated cathodolu-
minescence (CL) in feldspar minerals. Chem Geol 153: 81–91
Go
¨tze J, Krbetschek MR, Habermann D, Wolf D (2000) High-
resolution cathodoluminescence studies of feldspar minerals. In:
Pagel M, Barbin V, Blanc P, Ohnenstetter D (eds) Cathodolu-
minescence in geosciences. Springer, Berlin Heidelberg New
York, pp 245–270
Habermann D (2002) Quantitative cathodoluminescence (CL)
spectroscopy of minerals: possibilities and limitations. Mineral,
Petrol, 76: 247–259
Harlow GE, Brown GE (1980) Low albite: an X-ray and neutron
diffraction study. Am Mineral 65 986–995
Hobbs AN (1999) Dislocations in plagioclase feldspar. unpublished
MPhil Thesis, University of Hertfordshire, UK
Homman NPO, Yang C, Malmqvist KG, Hanghøj K (1995) Pla-
gioclase studies by ionoluminescence (IL) and particle-induced
X-ray Emission (PIXE) employing a nuclear microprobe.
Scanning Microscopy Supplement, 9: 157–165, Scanning
Microscopy International, Chicago, USA
Krbetschek MR, Go
¨tze J, Irmer G, Rieser U, Trautmann T (2002)
The red luminescence emission of feldspar and its wavelength
dependence on K,Na,Ca composition. Mineral Petrol 76: 167–
177
Parsons I (1979) The Klokken Gabbro-syenite complex, south
Greenland: cryptic variation and origin of inversely graded
layering. J Petrol 20: 653–694
Parsons I (1981) The Klokken Gabbro-syenite complex, south
Greenland: quantitative interpretation of mineral chemistry.
J Petrol 22: 233–260
Previtt CT, Sueno S, Papike JJ (1976) The crystal structures of high
albite at high temperatures. Am Mineral 61: 1213–1225
Rae DA, Chambers AD (1988) Metasomatism in the North Qoˆ roq
centre, South Greenland: cathodoluminescence and mineral
chemistry of alkali feldspars. Trans Roy Soc Edinburgh 79:
1–12
Smith JV, Artioli G (1986) Low albite, NaAlSi
3
O
8
: neutron dif-
fraction study of the crystal structure at 13 K. Am Mineral 71:
727–733
Smith JV, Stenstrom RC (1965) Electron-excited luminescence as a
petrologic tool. J Geol 73: 627–635
Telfer DJ, Walker G (1978) Ligand field bands of Mn
2+
and Fe
3+
luminescence centres and their site occupancy in plagioclase
feldspars. Modern Geol 6: 199–210
Upton BGJ, Emeleus CH (1987) Mid-Proterozoic alkaline mag-
matism in Southern Greenland: the Gardar Province. In: Fitton
JG, Upton BGJ (eds) Alkaline igneous rocks. Special Publica-
tion, Geological Society London 30: 449–471
Upton BGJ, Emeleus CH, Heaman LM, Goodenough KM, Finch
AA (2003) The Gardar province: repeated rifting events in the
mid-Proterozoic, South Greenland. Lithos (In press)
White WB, Masako M, Linneham DG, Furukawa T, Chandrase-
khar BK (1986) Absorption and luminescence of Fe
3+
in single
crystal orthoclase. Am Mineral 71: 1415–1419
381