Comparison of different NaGdF4:Eu3+ synthesis routes and their influence on its structural and luminescent properties

Article abstract of DOI:10.1016/j.jpcs.2005.01.002

Eu³⁺:NaGdF₄ samples were obtained via co-precipitation in aqueous solution (CP), reversed micelle (RM) method, reaction between solid GdF₃ and NaF solution (SR) as well as a solid-state reaction at high temperatures (SS). The synthesised materials were characterised using X-ray powder diffractometry, TEM microscopy, infrared spectroscopy and TGA analysis. For discussion of optical properties excitation and emission spectra were recorded and emission decay times were measured. The CP and RM methods allow to obtain powders with crystallite size of ～10 nm, which may be smoothly increased to about 1 μm during post-fabrication heat treatment. Differences in structural and especially in optical properties of phosphors prepared by different techniques are emphasised and applicability of wet-chemistry routes for synthesis of fluoride powders is argued.

Full text of DOI:10.1016/j.jpcs.2005.01.002

Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
www.elsevier.com/locate/jpcs
Comparison of different NaGdF4:Eu3C synthesis routes and their influence
on its structural and luminescent properties
a,
b
Mirosław Karbowiak , Agnieszka Mech , Artur Bednarkiewicz ,
b
*
Wiesław Stre˛k , Leszek Ke˛pinski
b
b
´
aFaculty of Chemistry, University of Wroclaw, ul. F. Joliot-Curie 14, 50-383 Wroclaw, Poland
bW. Trzebiatowski Institute of Low Temperature and Structure Research, Polish Academy of Sciences, P.O. Box 1410, 50-950 Wroclaw, Poland
Received 20 August 2004; revised 20 January 2005; accepted 30 January 2005
Abstract
Eu3C:NaGdF4 samples were obtained via co-precipitation in aqueous solution (CP), reversed micelle (RM) method, reaction between
solid GdF3 and NaF solution (SR) as well as a solid-state reaction at high temperatures (SS). The synthesised materials were characterised
using X-ray powder diffractometry, TEM microscopy, infrared spectroscopy and TGA analysis. For discussion of optical properties
excitation and emission spectra were recorded and emission decay times were measured. The CP and RM methods allow to obtain powders
with crystallite size of w10 nm, which may be smoothly increased to about 1 mm during post-fabrication heat treatment. Differences in
structural and especially in optical properties of phosphors prepared by different techniques are emphasised and applicability of wet-
chemistry routes for synthesis of fluoride powders is argued.
q 2005 Elsevier Ltd. All rights reserved.
Keywords: A. Nanostructures; C. X-ray diffraction; D. Luminescence; D. Optical properties
1. Introduction
A solid state (SS) reaction conducted at temperatures of
w650 8C is commonly used for synthesis of complex
fluorides in a polycrystalline form. However, this process
requires strictly inert or even reactive gas atmosphere, since,
the elevated temperature promotes strongly the migration of
O2K and OHK ions into the fluoride lattice, which may
result in formation of oxygen–rare-earth ion aggregates,
degrading seriously the quantum efficiency of the material.
An alternative method for synthesis of fluorides is a co-
precipitation from solution. This method allows for better
control of morphology and homogeneity of obtained
materials and limits strongly the oxygen contamination in
obtained compounds [3].
The luminescence spectroscopy of rare-earth (RE) ions
in the vacuum ultraviolet (VUV) spectral range has attracted
a considerable attention in the past few years. This results
from the search for new efficient VUV-excited phosphors
for mercury-free fluorescent lamps and plasma display
panels [1,2]. Fluorides are very well suited for such
applications, since they have very wide transparency in
spectral range from VUV to far IR, weak crystal field and
small refractive index. Due to low phonon frequencies
(Zumax!450 cmK1) the excited states of lanthanides
embedded into fluoride hosts exhibit long lifetimes, which
leads to a small probability of non-radiative transitions and
the number of luminescent levels is usually larger than for
oxide crystals doped with Ln3C ions.
Hitherto, synthesis of NaLnF4 from solution has been
reported in a few papers only. Hexagonal NaGdF4:Eu3C has
been synthesised by hydrothermal method at 240 8C [4].
Single-phase hexagonal fluorides with nominal composition
of NaY1KxKyYbxPryF4 (with 0%x%1 and y%0.05) have
been prepared by co-precipitation technique [3]. O’Connor
et al. [5] have reported synthesis of NaMnF3 and KMnF3
nanoparticles in aqueous solution of restricted environment
of reverse micelles. In our recent paper [6], we have
*
Corresponding author. Tel.: C48 71 3757304; fax: C48 71 3282348.
E-mail address: karb@wchuwr.chem.uni.wroc.pl (M. Karbowiak).
0022-3697/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpcs.2005.01.002

M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
1009
reported synthesis and some preliminary structural and
luminescent results obtained for nanosized NaGdF4:Eu3C
prepared by co-precipitation method (CP). We have
discerned no substantive differences between emission
spectra of the CP sample heated at 650 8C and the sample
synthesised by a SS reaction accomplished at the same
temperature. Therefore, it has been concluded, that the CP
synthesis may be regarded as an attractive method for the
preparation of complex fluorides, since, it is easier, does not
require expensive high purity rare-earth fluorides as starting
materials, results in powder composed of smaller particles
and the distribution of dopant ions is more homogenous.
Being encouraged by these first promising results we
have decided to explore further applicability of wet-
chemistry routes for synthesis of fluoride powders. Besides
the co-precipitation technique, the Eu3C:NaGdF4 was
prepared by microemulsion (reversed micelles) method as
well as by reaction of solid GdF3 with NaF solution at 90 8C.
Influence of method of synthesis on morphology of obtained
powders was investigated by TEM and X-ray powder
diffraction methods. Amount of water and other volatile
products was controlled by IR spectroscopy and TGA
analysis. Excitation and emission spectra were recorded to
get insight into optical properties of the material. These
results are compared with those obtained for Eu:NGF
phosphors synthesised by the solid-state (SS) reaction.
solution was disrupted with the addition of a 1:1
methanol/chloroform mixture. The resulted nanoparticles
were separated by centrifuging, washed several times with
methanol and dried at 50 8C.
2.3. Reaction between solid GdF3 and NaF solution (SR)
In the typical reaction, Gd2O3 (w0.7 g) was dissolved in
hot concentrated solution of hydrochloric acid and the
resulted solution was slowly evaporated almost to dryness.
The obtained GdCl3$xH2O was dissolved in w50 ml of
water and GdF3 was precipitated by adding excess of 40%
HF. To the resulted GdF3 powder, separated by centrifuging
and placed in a thick wall Teflon bottle, the solution of NaF
(w1.6 g in 50 ml of water) was added. The bottle with hot
solution was tightly closed with the cap and the mixture was
kept for different experimental times, ranging from few
hours to several days in 90 8C, with continuous stirring. The
resulted powder was separated by centrifuging, washed
several times with distilled water containing small amount
of NH4HF2, then two times with methanol and dried at
50 8C.
2.4. Solid state reaction (SS)
Eu3C:NaGdF4 (Eu:NGF-SS) sample was also prepared
by a conventional mixing and firing method. For this, a well
ground stoichiometric (1:0.99:0.01 molar ratio) mixture of
NaF, GdF3 and EuF3 was placed in a glassy carbon boat and
heated for 6 h at 680 8C in ArC(10%)SF6 atmosphere.
Eu3C(1 mol%):NaGdF4 powders obtained by co-pre-
cipitation method (CP), microemulsion method (RM),
GdF3(s)CNaF reaction in solution (SR) and solid state
reaction (SS) will be henceforth denoted as Eu:NGF-CP(T),
Eu:NGF-RM(T), Eu:NGF-SR(T) and Eu:NGF-SS, respect-
ively, where number in brackets indicates the temperature of
post-fabrication heat treatment—see Table 1 for details.
2. Experimental
2.1. Co-precipitation method (CP)
The (1%)Eu3C:NaGdF4 (Eu:NGF-CP) was synthesised
by a co-precipitation process applying the procedure
described in Ref. [6].
2.2. Reversed micelle method (RM)
A quaternary water-in-oil microemulsion system con-
tained CTAB (cetyltrimethylammonium bromide) as sur-
factant, n-butanol as co-surfactant, n-octan as the oil phase
and an aqueous solution of GdCl3 or NaF has been used as
reaction media for preparation of Eu3C:NaGdF4 (Eu:NGF-
RM) nanoparticles. In typical synthesis, the microemulsions
were prepared by adding successively, with continuous
stirring, 12.3 ml of n-butanol, 62.6 ml of n-octan and 10 ml
of aqueous solution, which corresponds to wZ
[H2O]/[CTAB] value of 16.8, to 12 g (0.033 mol) of
CTAB. For precipitation of Eu:NGF-RM two microemul-
sions were prepared separately. The first one contained
0.1 M solution of GdCl3 and 0.001 M solution of EuCl3,
while the other contained 0.7 M NaF. The microemulsion
containing NaF was added dropwise to equal volume of that
containing GdCl3 and EuCl3 at room temperature with
continuous stirring. After different reaction times, ranged
from few minutes to a few dozen hours, the micellar
2.5. Measurements
X-ray analysis was performed with DRON-2 powder
diffractometer using Ni filtered Cu Ka radiation. TEM
images were recorded with Philips CM 20 SuperTwin
microscope operating at 200 kV. Room temperature lumi-
nescence and excitation spectra were recorded with a
0.25 nm resolution at a SPF 500 Spectrofluorimeter
equipped with 300 W Xe-lamp. Both excitation and
emission spectra were corrected for the sensitivity of the
detection system. The kinetic behaviour was recorded with
Tektronix TDS 380 oscilloscope after excitation with
308 nm line of XeCl excimer laser by Lambda Physic or
355 nm line of THG Nd:YAG laser. The pulse width was
10 ns with repetition rate 10 Hz in both cases. Infra-red
spectra were obtained with a Bruker 113v FTIR
spectrophotometer.

1010
M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
Table 1
Investigated Eu3C(1 mol%):NaGdF4 samples, synthesis details, cell parameters and crystallite size determined from powder XRD measurements
c
Sample (abbreviation)a
Method of synthesisb
Cell volume
Crystallite sized
˚
Lattice constants (A)
3
˚
(A )
a
c
Eu:NGF-SS (SS)
Eu:NGF-CP (CP)
Eu:NGF-CP(300)
(CP(300))
SS reaction at 680 8C
6.019(1)
6.037(4)
6.033(3)
3.608(1)
3.595(7)
3.599(1)
113.21
113.47
113.44
(w1 mm)
11 nm (20 nm)
13 nm
CP, aged for 10 days in mother liquor
Sample Eu:NGF-CP, heated for 20 h at
300 8C in Ar
Eu:NGF-CP(680)
(CP(680))
Sample Eu:NGF-CP, heated for 5 h at
680 8C in ArC(10%)SF6
6.021(1)
3.607(1)
113.24
(300 mm)
Eu:NGF-RM (RM)
Eu:NGF-RM(300)
(RM(300))
RM reaction, 36 h, as-prepared
Sample Eu:NGF-RM, heated for 20 h at
300 8C Ar
6.046(8)
6.044(5)
3.594(2)
3.597(1)
113.77
113.79
8 nm (12 nm)
13 nm
Eu:NGF-RM(680)
(RM(680))
Sample Eu:NGF-RM, heated for 5 h at
680 8C ArC(10%)SF6
SR reaction, 7 days
6.025(2)
3.599(1)
113.15
(w0.1–1 mm)
Eu:NGF-SR (SR)
Eu:NGF-SR(300)
(SR(300))
6.024(4)
6.025(4)
3.603(1)
3.603(1)
113.23
113.27
(0.5–1 mm)
–
Sample Eu:NGF-SR heated for 20 h at
300 8C in Ar
Eu:NGF-SR(680)
(SR(680))
Sample Eu:NGF-SR, heated for 5 h at
680 8C in ArC(10%)SF6
6.024(4)
3.605(1)
113.29
–
a
In brackets the abbreviated sample names are given, used in the figure descriptions.
b
c
d
SS, solid state reaction; CP, co-precipitation; RM, reversed micelles; SR, GdF3CNaF reaction in solution.
ꢀ
Indexed on the basis of a hexagonal cell, space group P6,
Determined from Scherer’s formula, values given in brackets are obtained from TEM photograph.
3. Results and discussion
The lattice constants determined for Eu:NGF-CP sample
heated at 680 8C are also practically the same as those for SS
sample (Table 1). One may notice that a cell volume slightly
increases along reduction of crystallite size. Although the
changes are rather small, expanding of unit cell such is
expected for smaller crystallites.
3.1. Structure and morphology
Fig. 1(a) shows X-ray powder diffractogram measured
for Eu:NGF-SS sample prepared by the solid state (SS)
reaction. The observed pattern was indexed on the basis of
Fig. 1(d) and (e) present XRD patterns of RM-made
samples obtained after 36 h of reaction in solution (sample
Eu:NGF-RM) and for the same powder heated additionally
for 6 h at 680 8C (sample Eu:NGF-RM(680)). The average
crystallite size of as-prepared RM-made powder is equal to
8 nm, which value is slightly smaller than obtained for
Eu:NGF-CP sample. Similarly, the average sizes of
nanocrystallites are for heat-treated RM-made powders
smaller than those obtained for analogous CP-processed
samples—Table 1.
ꢀ
˚
a hexagonal cell, space group P6, with aZ6.019(1) A and
˚
cZ3.608(1) A, which stays in good agreement with lattice
constants determined for NaEuF4 [7].
Fig. 1(b) and (c) show X-ray powder diffraction patterns
recorded for as-prepared and heated at 680 8C CP-made
Eu:NGF samples, respectively. In presented XRD patterns
lines characteristic for a hexagonal NaGdF4 are observed
only, which proves, that a single-phase crystalline powder
precipitates from solution, and that the hexagonal phase of
NaGdF4 is stable in investigated temperature range (20–
680 8C). The (021) peak was chosen for determination of the
average crystallite size by applying Scherrer’s formula [8]
Patterns (f) and (g) in Fig. 1 present results of XRD
measurements for samples obtained during reaction
between precipitated GdF3 and an excess of NaF solution.
The XRD pattern of the obtained after 7 days of reaction
time Eu:NGF-SR sample is characteristic for the pure
hexagonal form of NaGdF4 and was indexed with the lattice
constants close to that obtained for Eu:NGF-SS powder—
see Table 1. Due to the low solubility of GdF3, the reaction
with NaF is slow and, in contrast to previous methods, the
product in the form of a coarse powder is obtained.
Therefore, the heat treatment of SR-made samples at 300
or 680 8C does not change noticeably the lattice constants.
However, for sample heated at 680 8C small peaks
indicative of the presence of GdF3 appear (Fig. 1(g)).
Since, the NaGdF4 is stable up to 835 8C, that is temperature
at which it melts incongruently forming a liquid phase and a
solid solution with fluorite-type of structure [9], appearance
0:9l
cos q 32 K320
L Z
(1)
À
Á1=2
where L is the average size of the crystallites (in nm), l
denotes the wavelength (in nm), 3 is the full width at half-
maximum (in radian) of a diffraction line located at q, and 30
is the scan aperture of the diffractometer. A narrowing of
diffraction lines observed for sample heated at elevated
temperature reflects the changes in a average size of
nanocrystallites, which increases from 11 nm for as-
prepared powder to w300 nm for sample heated at
680 8C. Comparison of powder diffractogram obtained for
sample heated at 680 8C with that of sample synthesised by
SS method proves the full crystallinity of the former.

M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
1011
rather irregular (Fig. 2(a2)). Moreover, the size distribution
is not uniform and particles in w0.1–1 mm range may be
found. Fig. 2(b1) shows microstructure of Eu:NGF-CP
sample. Similarly as for RM-made sample the spherical and
slightly agglomerated grains are observed. The difference
between grain size determined from TEM (Z20 nm on
average) and crystallite size determined from XRD (11 nm)
is larger than for RM samples. Moreover, a larger fraction of
amorphous phase is observed—see Fig. 2(b2). The grains of
CP-made powder heated at 680 8C (Fig. 2(b3)) have
similarly irregular shape as those observed for RM-made
material, although the particles are smaller, with average
size equal to 300 nm. Picture presented in Fig. 2(b4), taken
with larger magnification, shows that the heat-treated CP-
made powder is composed of well-formed monocrystallites.
The morphology of Eu:NGF-SR powder presented in
Fig. 2(c) differs significantly from both the RM- and CP-
made materials. This powder is composed of small well
formed and elongated in shape monocrystallites with the
size in the 0.5–1 mm range. The grains are well separated
and no traces of amorphous phase are observed.
(g) SR(680)
(f) SR
(e) RM(680)
(d) RM
(c) CP(680)
(b) CP
3.2. IR spectra and TGA analysis
For solution processed samples the presence of water
molecules adsorbed on surface of the powder is expected.
Moreover, for fluoride material prepared by a wet-chemistry
route hydroxyl or oxygen ions are likely to be incorporated
into the lattice, substituting for FK anions. Fig. 3 presents IR
reflectance spectra recorded for as-prepared Eu:NGF
samples (spectra a and b) as well as for the same samples
heated at 300 and 680 8C. The structured bands observed
around 2280 and 2880 cmK1 are due to an oil used to
suspend the powder. As expected, the as-prepared co-
precipitated Eu:NGF-CP sample contains considerable
amounts of water. This is evidenced by the broad band
around 3350 cmK1 as well as narrower band observed at
1640 cmK1, indicative of antisymmetric and symmetric
stretches and HOH bending modes [10]. In addition to water
molecules or hydroxyl groups bounded to the surface of the
powder, one could also expect hydroxyl groups incorpor-
ated into fluoride lattice. It has been suggested in Ref. [6],
that the relatively narrow peak observed at 3682 cmK1, on
the slope of the water absorption band and marked by an
arrow in Fig. 3, should be assigned to vibrations of the OHK
groups incorporated into the host lattice. The lattice
incorporated OH-groups disappear for CP material treated
at 680 8C (Fig. 3(h)). The peak at 3682 cmK1 is already
present for untreated sample, which suggests that OHK
groups enter the lattice during precipitation of powder,
rather than migrate from the surface during the heat
treatment of the wet sample.
(a) SS
10
20
30
2
40
(degree)
50
60
θ
Fig. 1. Room temperature X-ray diffractograms (using Cu Ka radiation, lZ
0.1541 nm) of hexagonal Eu:NGF samples synthesised by solid state
reaction (SS) (a), co-precipitation method (CP) (b, c), reversed micelle
method (RM) (d, e) and by reaction of solid GdF3 with NaF solution (SR)
(f, g). Numbers in parentheses indicate temperature of post-fabrication
heat-treatment in ArC(10%)SF6 atmosphere. See Table 1 for samples
descriptions.
of GdF3 at 680 8C cannot result from partial sample
decomposition. It rather seems, that even after 7 days of
reaction the equilibrium between GdF3 and NaGdF4 is not
completely shifted towards the latter form. This suggests,
that small amounts of unreacted GdF3 should be present in
all the samples, However, while for unheated sample the
GdF3 is rather in amorphous form, during the heat treatment
its crystallinity improves and intensity of observed powder
diffraction peaks increases.
Fig. 2(a1) presents TEM photograph taken for Eu:NGF-
RM specimen. The powder is composed of spherical
particles, which are, however, considerably agglomerated
and some residue of amorphous phase may be also noticed.
The average grain size determined from TEM is equal to
12 nm and is slightly larger than obtained from XRD
(8 nm). Reflections observed in SAED pattern shown in
inset of Fig. 2(a1), are considerably broadened due to small
size of the particles. The heat treatment at 680 8C leads to
the considerable growth of grains, however, their shapes are
For sample heated at 300 8C considerable amount of the
water, probably that bounded to the surface of the powder,
evaporates, however, heating at 680 8C is necessary for
complete removing of the stronger bounded hydroxyl groups.

1012
M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
Fig. 2. TEM photographs showing the microstructure of Eu:NGF fluorides: (a1) as-prepared powder obtained by reversed micelle method (Eu:NGF-RM), (a2)
RM sample heated at 680 8C in ArC(10%)SF6 atmosphere (Eu:NGF-RM(680); (b1,b2) as-prepared co-precipitation-processed sample (Eu:NGF-CP), (b3,b4)
CP sample heated at 680 8C in ArC(10%)SF6 atmosphere (Eu:NGF-CP(680); (c) as-prepared sample resulted from reaction of solid GdF3 with NaF solution
(Eu:NGF-SR). Diffraction rings observed in SAED pattern obtained for sample Eu:NGF-RM are shown in inset of photograph (a1). See Table 1 for samples
descriptions.
It should be emphasized, however, that when fluoride powders
are simply heated, even in inert atmosphere, the OHK groups
may transform into oxide impurities. Therefore, applying of
fluorinating agent (e.g. SF6) during the heat-treatment is
indispensable for obtaining of pure fluoride phosphors. It is
worth to note, that although the wet-chemistry route was
applied for synthesis of Eu:NGF phosphor, the content of
water (or hydroxyl groups) for sample heated at 680 8C in
ArC(10%)SF6 atmosphere is at the same low level as that
observed for sample obtained by solid state (SS) reaction at
680 8C—compare Fig. 3(h) and (j).
as may be concluded from higher intensity of water
absorption bands of the former, however, the peak at
3682 cmK1, assigned to OH groups incorporated into
lattice, is less pronounced for RM powder than for
Eu:NGF-CP. It is worth noting, that for Eu:NGF-CP sample
both water absorption bands, at 3350 and 1640 cmK1
,
decrease on heating, while for Eu:NGF-RM(300) sample
these bands are also disappearing, but in adjacent of the
latter a new band appears at 1560 cmK1. This band should
be most probably also assigned to vibrations of OHK
groups, however, bound in different way than in other
samples. It seems, that they may result from migration of
surface-bound OH groups into the fluoride lattice during
The content of water in not-heated RM sample (Fig. 3(a))
is slightly larger than for as-prepared CP sample (Fig. 3(b)),

M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
1013
contain water adsorbed on the material surface. This leads to
conclusion, that just surface bounded water is removed from
CP and RM powders in 50–300 8C temperature range. It is
worth to notice, that water content observed for Eu:NGF-
SR(680) sample (heated at 680 8C) is the lowest among all
investigated samples, and is even lower than for Eu:NGF-SS
powder.
SS
SR(680)
CP(680)
(j)
(i)
RM(680)
SR(300)
CP(300)
RM(300)
(h)
(g)
The TG and DTA curves were obtained for CP-, RM- and
SR-made samples on heating in 40 – 800 8C temperature
range, with the heating rate of 10 8C minK1 in a static
atmosphere of N2. For Eu:NGF-CP sample the efficient
dehydration process is observed in 50–300 8C temperature
range. At this temperature the weight loss is equal to 0.98%,
which corresponds to 0.14 mol of water/1 mol of NaGdF4.
The weight loss increases to 1.1% at 680 8C, the temperature
at which water is completely removed. For RM sample the
weight loss due to the dehydration process is three times
larger and equal to 2.8%, which corresponds to 0.43 mol of
water/1 mol of NaGdF4. However, for this sample water is
efficiently removed at lower temperature than for CP
sample, namely in 50–160 8C temperature range. The TG
curve obtained for Eu:NGF-SR sample proves the con-
clusions already drawn from the IR spectra, that this sample
does not contain surface adsorbed water. Only insignificant
weight loss of w0.12% is observed in 50–300 8C range. The
DTA curves do not show any well-defined endothermic
peaks, which indicates that for all the three samples water is
being gradually removed along the temperature rise.
(f)
(e)
(d)
SR
CP
RM
(c)
(b)
(a)
1500
2000
2500
3000
3500
4000
-1
Wavenumber (cm
)
Fig. 3. Infrared spectra of Eu:NGF powders samples synthesised by co-
precipitation method (CP), reversed micelle method (RM) and by reaction
of solid GdF3 with NaF solution (SR): (a–c) as-prepared, (d–f) and (g–i)
heated in ArC(10%)SF6 atmosphere at 300 and 680 8C, respectively. The
upper trace (j) was recorded for Eu:NGF-SS sample obtained by solid-state
reaction at 680 8C. The peak indicated by an arrow was assigned to
vibrations of the OHK groups incorporated into the host lattice. See Table 1
for samples descriptions.
3.3. Excitation and emission spectra
Fig. 4(a) and (b) present excitation spectra recorded in
the 420–210 nm range for Eu:NGF-CP and Eu:NGF-
CP(680) samples, respectively. The sharp lines located at
a vicinity of 250, 275 and 311 nm should be assigned as the
8
8
8S7/2/6DJ, S7/2/6IJ and S7/2/6PJ transitions of the
Gd3C ion, respectively. The low-intensity lines in 350–
420 nm range are due to F0,1/(5LJC5GJ) transitions of
the heat treatment of the wet sample. The band at energy of
1560 cmK1 is observed exclusively for Eu:NGF-RM(300)
sample and, as it will be shown in further part of this paper,
only for this one among all investigated samples, intensity
of emission observed for specimen heated at 300 8C is lower
than for unheated counterparts. The Eu:NGF-RM(680)
powder (heated at 680 8C) contains only traces of water
and do not differ in this regard from Eu:NGF-CP(680) and
Eu:NGF-SS samples.
7
Eu3C. In the excitation spectrum obtained for Eu:NGF-
CP(680) sample, the excitation lines are observed at the
same wavelengths as for as-precipitated sample, and the
small differences in lines intensities are observed, only.
Besides, one may notice, that for sample heated at
higher temperature lines are narrower, e.g. FWHM of the
8S7/2/6PJ line at 311.9 nm is equal to 1.1 and 0.9 nm for
sample Eu:NGF-CP and Eu:NGF-CP(680), respectively.
Analogous changes in excitation spectra of as-precipitated
and heat-treated powders could be observed for samples
prepared by RM and SR methods—see insets in Figs. 8
and 9. There are no noticeable differences between the
excitation spectra of Eu:NGF-CP(680) and Eu:NGF-SS
samples (Fig. 4(b) and (c)). Likewise, samples prepared by
different methods (CP, RM or RS) and heat treated at given
temperature, exhibit very similar spectra.
Eu:NGF-SR sample exhibits significantly lower content
of water as compared to non-heated Eu:NGF-CP sample, as
is evidenced by low intensity of water absorption band in
spectrum presented in Fig. 3(c), and only vestigial band at
3682 cmK1 of OHK groups incorporated into lattice may be
noticed. For this sample the amount of water is at similar
level as observed for CP- or RM-made samples dried at
300 8C. In contrast to CP- or RM-made material, only
insignificant water loss is observed for SR sample heated at
300 8C. This is a consequence of considerably higher
crystallinity of Eu:NGF-SR powder, which almost does not
There are three different crystallographic sites for the
Gd3C and NaC cations in the hexagonal form (space group

1014
M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
differences in occupancy of 1a and 1f sites. Such assignment
(c) SS
agrees with our observations and is further supported by
expected differences in strength of crystal-field affecting
Eu3C ions in both sites. For NaEuF4 the Eu(1a)–F distances
(Gd3+ 8S7/2 6IJ
)
˚
are equal to 2.339 (3!) and 2.383 (6!) A, whereas
corresponding distances for Eu(1f)–F are longer and equal
˚
(Gd3+
)
8S7/2 6PJ
(Gd3+
)
8S7/2
6DJ
to 2.368 (3!) and 2.465 (6!) A [7]. The CF splitting of the
7F1 multiplet, calculated as energy difference between lines
(Eu3+
)
7F0,1 (5LJ+5GJ)
A and C or B and D is equal to 97 and 86 cmK1
,
respectively. These prove the above assignment, because
for 1a site a stronger CF is expected due to shorter Eu–F
distances. The determined splitting corresponds well to the
value observed for Eu3C in LiYF4, equal to 96 cmK1 [11].
The presence of two distinct sites is also evidenced by the
5D1/7F0 transition, which is composed of two lines—see
inset in Fig. 5. We do not observe changes in CF splitting of
the 7F1 multiplet for Eu:NGF-SS, Eu:NGF-CP(680) and
Eu:NGF-RM(680), which results from the fact, that lattice
constants for all these samples do not differ considerably.
Similarly, we do not observe reliance between the sites
population and method of synthesis.
(b) CP(680)
(a) CP
Fig. 6 presents enlarged part of excitation spectra
3C 7
recorded in Eu
F
0,1
/(5LJC5GJ) transition range for
300
350
400
as-prepared CP, RM and RS samples as well as for the same
powders heat-treated. For both CP and RM samples the shift
of 0.5 nm towards longer wavelengths is observed between
as-prepared sample and that heated at 680 8C. Such shift
does not occur for SR samples, which indicates, that it is due
to changes in crystallite size and resulting from this
differences in crystal-field strength. The Gd3C(Eu3C)–FK
distances are slightly different for small crystallites of as-
prepared powders than for larger crystallites of heat-treated
CP and RM samples, which affects the CF strength and
results in the shift of energy levels. Such changes in
200
250
300
350
400
450
Wavelength (nm)
Fig. 4. Excitation spectra of: (a) as-prepared Eu:NGF-CP sample, (b)
Eu:NGF-CP(680) sample heated at 680 8C in ArC(10%)SF6 atmosphere,
and (c) Eu:NGF-SS sample. All spectra were recorded at room temperature
while monitoring the 614.0 nm emission line, corresponding to the
5D0/7F2 transition of the Eu3C ions. Observed lines result from the
8S7/2/6DJ, S7/2/6IJ and S7/2/6PJ transitions of the Gd3C ion and
7F0/(5LJ C 5GJ) transitions of the Eu3C ions. See Table 1 for samples
descriptions.
8
8
measured
calclated
3C
ꢀ
P6) of NaGdF4 [7]. The 1a site is occupied fully by Gd
ions, and the 1f site is occupied randomly by Gd3C and NaC
ions. The third site, 2h, is half occupied by NaC ions. Gd3C
ions in sites 1a and 1f are ninefold coordinated by FK ions
and both have C3h point symmetry. It was shown recently
for NaEuF4 and Eu3C:NaYF4 [4], that the third site 2h, for
which irregular octahedral co-ordination resulting in C3 site
symmetry was suggested, may also be weakly occupied by
Eu3C ions.
C
D
A
B
522
524
526
528
Wavelength (nm)
c) Eu:NGF-RM
b) Eu:NGF-CP
a) Eu:NGF-SS
The JZ1 level in crystal-field (CF) of C3h symmetry
should split into two levels (a non-degenerate and twofold
degenerate CF levels). In Fig. 5, which presents enlarged
part of emission spectra in 5D0/7F1 transitions range four
lines are observed, which proves that Eu3C in NaGdF4 are
present in both 1a and 1f sites. Similar results were reported
586
588
590
592
594
596
598
600
Wavelength (nm)
for low-temperature emission spectrum of NaGdF4:Ce3C
,
Fig. 5. Enlarged part of emission spectra in 5D0/7F1 transition range of
Eu3C in (a) Eu:NGF-SS, (b) Eu:NGF-CP and (c) Eu:NGF-RM powders.
The lines assignment is taken from Ref. [13]. Inset shows that 5D1/7F0
line observed in emission spectrum of Eu3C in Eu:NGF-CP (recorded under
274 nm excitation of a Xe-lamp) is composed of two transitions.
Eu3C [13]. The authors have assigned lines ACC and BCD
to Eu3C ions on the 1a and 1f sites, respectively, arguing
that lines A and C are broader and of higher intensity
compared to B and D. The latter effect results from

M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
1015
5D2 7FJ
5D0
7FJ
3
J= 2
4
3
4
5
J=1
5D1 7FJ
2
4
5D3 7FJ
J= 1
J= 0 1
2
3
4
2
3
SR(680)
SR(300)
SR
5H3 7FJ
(d) Eu:NGF-SS
(c)
(c) Eu:NGF-CP(680)
Gd3+
(b) Eu:NGF-CP(300)
8
6P7/2 S7/2
RM(680)
(a) Eu:NGF-CP
RM(300)
RM
(b)
300
400
500
Wavelength (nm)
600
700
Fig. 7. Emission spectra of (a) as-prepared Eu:NGF-CP sample; (b) and (c)
Eu:NGF-CP(300) and Eu:NGF-CP(680) samples heated in in ArC
(10%)SF6 atmosphere at 300 and 680 8C, respectively. The spectrum of
Eu:NGF-SS sample is also included for comparison (d). The spectra were
recorded under 274 nm excitation of a Xe-lamp, which corresponds to the
CP(680)
CP(300)
CP
5
8S7/2/6IJ transition of Gd3C ions. Observed lines are assigned to D0–
(a)
390
3/7FJ transitions of Eu3C, except for the line at 311 nm, which is due to
the 6P7/2/8S7/2 transition of Gd3C. See Table 1 for samples descriptions.
395
400
405
Wavelength (nm)
5
the 325–360 nm region have been ascribed to H3/7FJ
transitions of Eu3C [13].
Fig. 6. Enlarged part of excitation spectra in 7F0/(5LJC5GJ) transition
range of Eu3C recorded for as-prepared as well as heat-treated at 300 and
680 8C samples, prepared by co-precipitation method (a), reversed micelle
method (b) and in reaction between solid GdF3 and NaF solution (c).
5
For the Eu:NGF-CP sample the emission from the D0
and 5D1 levels of the Eu3C ions is observed almost
exclusively (Fig. 7(a)). This results from a considerable
amount of water and hydroxyl groups present in the fluoride
structure, which quenches the emission from the higher
positioned levels. During the heat-treatment the OHK
groups are being gradually removed from the material and
crystallite size and Gd3C(Eu3C)–FK distances are not
observed for SR material.
3.3.1. D1–3/5D0 emission intensity ratio
5
the emission from the D2 and D3 multiplets of the Eu3C
ion appears—Fig. 7(b) and (c). The D1–3/5D0 emission
5
5
5
The dependence of nonradiative WNR transition rate
constant on the energy gap (DE) from a given electronic
state to the next lower state and on the number of effective
phonons (p) is expressed as [12]
intensity ratio of 0.55 was determined for the as-prepared
sample and this ratio increases to 0.75 for sample heated at
300 8C. For the sample Eu:NGF-CP(680) heated in 680 8C a
more than a factor of two larger value, equal to 1, 4, was
obtained and a very similar emission intensity ratio of 1, 5
was determined for the Eu:NGF-SS sample. This indicates,
that independently on the method of synthesis, a co-
precipitation process followed by a heat treatment or the
solid-state reaction, phosphors with only vestigial amounts
of the hydroxyl groups incorporated into the lattice may be
obtained.
WNR Z A eKðaDEÞ Z A eKðbpÞ
(2)
where A, a and b are constants characteristic of the material.
5
Therefore, the emission from D0 is less affected by the
presence of OHK groups than emission from 5D1–3 and this
allows to treat the D1–3/5D0 emission intensity ratio as a
5
measure of amount of the OHK groups incorporated into
lattice [6]. This ratio will be used in the further part of this
paper for comparison of luminescent properties of powders
synthesised by different methods. This is illustrated in
Fig. 7, which presents emission spectra recorded for
Eu:NGF-CP, Eu:NGF-CP(300) and Eu:NGF-CP(680)
samples. The spectra were obtained with 274 nm excitation
of Xe lamp, which corresponds to the 8S7/2/6IJ absorption
of Gd3C, and they are scaled to have the same integrated
intensity of the 6P7/2/8S7/2 transition of Gd3C at 310.6 nm.
Other lines observed in the spectra are assigned as 5DJ/7FJ
transitions of Eu3C. Some weak emission lines observed in
Fig. 8 presents emission spectra of RM-made material
8
recorded upon exciting in S7/2/6IJ level of Gd3C
(274 nm). In contrast to CP sample, for which a weak
emission from 5D2 and 5D3 levels of Eu3C is noticeable (Fig.
7(a)), emission from analogous levels is completely
quenched for unheated Eu:NGF-RM sample and transitions
from 5D0 and 5D1 levels are observed exclusively in
Fig. 7(a). For RM sample heated at 300 8C emission from
5D2 appears, although its relative intensity is lower than for
Eu:NGF-CP(300) sample. Moreover, for Eu:NGF-RM (300)

1016
M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
c) RM(680)
b) RM(300)
(c) SR(680)
(b) SR(300)
(a) SR
a) RM
350
250
300
400
(c) SR(680)
(b) SR(300)
(a) SR
250
300
350
400
(c) RM(680)
Wavelength (nm)
Wavelength (nm)
(b) RM(300)
(a) RM
300 350 400 450 500 550 600 650 700 750
Wavelength (nm)
300
400
500
600
700
Wavelength (nm)
Fig. 9. Emission spectra of (a) as-prepared Eu:NGF-SR sample; (b) and (c)
Eu:NGF-SR(300) and Eu:NGF-SR(680) samples heated in in ArC
(10%)SF6 atmosphere at 300 and 680 8C, respectively. The spectra were
recorded under 274 nm excitation of a Xe-lamp, which corresponds to the
8S7/2/6IJ transition of Gd3C ions. Inset shows excitation spectra recorded
for the same samples while monitoring the 614.0 nm emission line,
corresponding to the 5D0/7F2 transition of the Eu3C ions. See Table 1 for
samples descriptions.
Fig. 8. Emission spectra of (a) as-prepared Eu:NGF-RM sample; (b) and (c)
Eu:NGF-RM(300) and Eu:NGF-RM(680) samples heated in ArC
(10%)SF6 atmosphere at 300 and 680 8C, respectively. The spectra were
recorded under 274 nm excitation of a Xe-lamp, which corresponds to the
8S7/2/6IJ transition of Gd3C ions. Inset shows excitation spectra recorded
for the same samples while monitoring the 614.0 nm emission line,
corresponding to the 5D0/7F2 transition of the Eu3C ions. See Table 1 for
samples descriptions.
observed for Eu:NGF-SR and Eu:NGF-SR(300) samples,
and only slightly larger value of 0.7 for sample heated at
sample the overall emission is significantly lower than for
unheated Eu:NGF-RM sample, which differs this material
from CP-made samples, where increase in emission
intensity was observed for heat-treated powder. The origin
of this effect is not clear and additional investigations would
be required for explanation of this phenomenon. It may be
somehow related to observation that Eu:NGF-RM(300)
sample became considerably grey on heating. This probably
results from formation of FK ion deficiency in the fluoride
lattice. Among the different factors, which may be
responsible for this behaviour, the content of OHK groups,
which are incorporated in sites otherwise occupied by FK
ions, should be considered. IR spectra (Fig. 3) and TG
analysis prove that content of water and hydroxyl groups is
significantly larger for RM than for CP samples. During the
heat-treatment the OHK groups are being gradually
removed from the powder, which results in formation of
lattice defects. These defects form additional paths for non-
radiative de-excitation of excited levels and quench the
emission of Gd3C and Eu3C ions. For RM sample heated at
680 8C the structure defects are repaired and emission
intensity is similar to that observed for Eu:NGF-CP(680)
5
680 8C. The similar D1–3/5D0 emission intensity ratio of
w0.7 was reported for sample prepared by hydrothermal
method [4]. Since, neither the OHK groups content nor
the crystallite size is changed by heat treatment of SR
5
powder, it also does not affect considerably the D1–3/5D0
emission intensity ratio. This results from larger crystallite
size of SR material, as compared with CP or RM samples.
5
However, it is not clear why the D1–3/5D0 emission
intensity ratio of SR sample heated at 680 8C is lower
then those of Eu:NGF-CP(680) and Eu:NGF-RM(680)
samples, even though the last two specimens contain
slightly larger amounts of hydroxyl groups than Eu:NGF-
SR(680) sample. This effect cannot be also accounted for by
differences in grain size, because it is comparable for all
heated samples.
3.3.2. Gd3C–Eu3C energy transfer
An important factor determining potential applications of
phosphors is efficiency of energy migration from Gd3C to
Eu3C ions. In our case, when excitation spectra were
recorded upon monitoring of Eu3Cemission (line 1 in
Fig. 10) the efficiency of energy transfer may be
qualitatively assessed by comparison of relative intensities
of Gd3C and Eu3C excitation bands, which may be
expressed as the ratio of intensity of (Gd3C)8S7/2/6IJ
(line Gd excit. in Fig. 10) and (Eu3C)7F0/(5LJC5GJ) (line
Eu excit. in Fig. 10) transitions. The higher value of this
ratio should point to more efficient energy transfer. The
above ratio has very similar values equal to 7.1 and 6.5 for
Eu:NGF-SS(680) and Eu:NGF-CP(680) samples, respect-
ively. The value almost a factor of two larger, equal to 12.6
5
and Eu:NGF-SS samples, with comparable D1–3/5D0
emission intensity ratio of 1.3.
Fig. 9 presents emission spectra recorded for raw
Eu:NGF-SR sample, as well as for SR-made samples heated
at 300 and 680 8C. In contrast to as-prepared CP and RM
samples relatively strong emission is observed for unheated
SR sample. Moreover, there are no changes in emission
spectra recorded for as-prepared SR sample and that heated
at 300 8C, which result from absence of surface bounded
5
water. The D1–3/5D0 emission intensity ratio of 0.6 was

M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
1017
was determined, which is lower than for Eu:NGF-SS and
Eu-NGF-CP(680) samples and suggests, in connection with
observed highest Gd/Eu absorption intensity ratio observed
in excitation spectra, that energy transfer from Gd3C to
Eu3C ions is for this sample the most effective.
6DJ
40
30
20
10
0
6IJ
6PJ
3.3.3. Emission lifetime
5
5HJ
Table 2 presents the lifetimes of DJ (JZ0–3) levels
measured for as-precipitated and heat-treated CP samples as
well as for SS sample. The lifetimes of 5D0 level for CP, RM
and RS samples heated at 680 8C are very similar
(6.2–6.5 ms) and are close to value determined for SS
sample (6.2 ms). Similarly, very comparable lifetimes of
5D1–3 levels are observed for heat-treated CP and RM
samples as well as for SS sample, however, significantly
shorter values have been determined for SR sample. This
may be directly linked to lower intensity of emission
5D4
5GJ
5LJ
5D3
5D2
5D1
5D0
2
5
transitions from D1–3 levels observed for SR sample, as
compared to SS or heat treated CP and RM samples. The
1
5
lifetimes of D1–3 levels of unheated samples are, as one
could expect, shorter than for appropriate heat-treated
counterparts, which results from the presence of OHK
groups and enhanced non-radiative transition rate constant.
However, unexpectedly the lifetime of 5D0 level is for
unheated CP and RM samples equal to 10.1 ms and is
significantly longer than for heated samples. Such behaviour
is not observed for SR sample, which suggest that observed
effect is connected with crystallite size. This may be due to
the fact that in smaller nanocrystallites, the local symmetry
of lanthanide ion tends to increase [14], which leads to
7F4
7F3
7F2
7
8S7/2
F0,1
Eu3+
Gd3+
Fig. 10. Schematic energy diagram for Gd3C and Eu3C ions in NaGdF4.
Levels important for discussion of results are included into this figure, only.
was determined for Eu:NGF-RM(680) sample, and rela-
tively low value of 4.7 for Eu:NGF-SR(680) sample. Thus,
energy migration processes strongly depends on synthesis
routes, however, the factors responsible for observed
dependence are not clear.
5
enhancement of the D0/7F1 transition and results in the
5
increase of D0 level lifetime. However, this may be also
connected with observed dependence of fluorescence life-
times on the index of refraction of the medium surrounding
the emitting ion [15]. It was reported in Ref. [15] that
Similar assessment of energy transfer efficiency may be
done by comparison of relative intensity of Gd3C and
Eu3C emission observed upon excitation in Gd3C. For this
end the ratio of integral intensities of (Gd3C)6P7/2/8S7/2
(line 2 in Fig. 10) and (Eu3C)5D0/7F1 (line 1 in Fig. 10)
transitions was calculated. In this case, the lower value of
this ratio the higher efficiency of energy transfer, assuming,
that Gd3C ions are deexcitated only by radiative emission
or transfer of excitation energy to Eu3C ions (which is
equivalent to assumption, that other non-radiative pro-
cesses have a similar rate constants for all samples). This
ratio has again similar values for Eu:NGF-SS and Eu:NGF-
CP(680) samples, equal to 0.13 and 0.10, respectively.
Significantly lower value of 0.036 was obtained for
Eu:NGF-SR(680) sample. However, since, for this sample,
as was indicated above, also the lowest value of Gd/Eu
absorption intensity ratio was observed, the low value of
Gd/Eu emission intensity ratio does not result from
efficient energy transfer, but indicates rather the existence
of other paths for non-radiative quenching of Gd3C excited
levels. For Eu:NGF-RM(680) sample the value of 0.085
5
radiative lifetime of the D0 excited state of Eu3C ions in
nanocrystalline Y2O3 samples is about four times longer
than in the micron size powder of the same material. The
magnitude of lifetime enhancement is determined by the
fraction of the sample volume occupied by nanocrystals
(filling factor) which for nanopowders is much lower than 1.
3.4. Applicability of wet-chemistry routes for synthesis
of fluorides luminofores
3.4.1. Co-precipitation (CP) synthesis
CP method is a relatively straightforward route for
synthesis of NaGdF4 fluorides. In contrast to SS synthesis,
which requires pure and dry GdF3 as starting material,
significantly cheaper Gd2O3 may be used in CP method. The
important for this synthesis is applying of excess of NaF,
otherwise a mixture of GdF3 and NaGdF4 is obtained. The
first steps of synthesis are performed in mild conditions in
temperature of 90 8C and may be accomplished in Teflon
vessels. For complete removing of water molecules and OH
groups heating at 680 8C in ArCSF6 atmosphere is

1018
M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
Table 2
The lifetimes of 5DJ (JZ0–3) levels of Eu3C measured for samples obtained by different method
Sample
Lifetime (ms)
5D0
5D1
5D2
5D3
Eu:NGF-CP
10.1
6.5
2.95
3.6
2.44
0.37/1.6a
0.74/2.0a
0.1/0.4a
0.5/1.6a
0.16/0.5a
0.21/0.7a
0.9/2.5a
Eu:NGF-CP(680)
Eu:NGF-RM
3.1
0.14/0.93a
10.1
6.2
1.8/4.8a
Eu:NGF-RM(680)
Eu:NGF-SR
3.5
2.9
0.86/1.7a
1.7
6.2
2.2
Eu:NGF-SR(680)
Eu:NGF-SS
6.3
2.9
6.2
3.3
2.9
5
5
See Table 1 for sample description. The lifetimes were monitored at 592, 552, 510 and 465 nm, which correspond to 5D0/7F1, D1/7F2, D2/7F3 and
5D3/7F4 transitions, respectively.
a
Values obtained from the fit of the decay curve to biexponential function.
sufficient, which may be performed in a platinum or glassy
carbon boat placed in a quartz tube. The optical perform-
ance of CP made material does not differ from material
prepared by SS reaction. The important advantage of CP
method is formation of powder with grain sizes in
nanometer scale (10–15 nm). Moreover, the grains size
may be smoothly changed by adjusting the temperature and
duration of heat treatment. It is also expected that for CP
sample the doping substance should be more homoge-
neously distributed. In Ref. [3] a similar co-precipitation
process was applied for synthesis of fluorides in Na7F-
(Y,Yb,Pr)F3 system. The authors have obtained NYF:Pr and
NYYbF:Pr fluorides with the average grain sizes of 0.9 mm.
Moreover, for these fluorides only very weak signal
corresponding to OHK groups was observed in IR spectra,
and the TGA curves for SS and CP samples were similar.
These observations are in contrast to our findings, and
probably may be related to differences in grain sizes, which
in our case are in nanoscale dimensions, resulting from
lower solubility of NaGdF4 as compared to NaYF4.
to obtain powders with particles remarkably smaller than
that resulted from co-precipitation method. This may
suggest, that reversed micelles are not stable in fluoride
systems. Similar independence of particle size on w ratio
have been observed during the synthesis of NH4MnF3 [16]
and KMnF3 [5] nanopowders. To explain this behaviour the
authors have admitted, that once the particles are pre-
cipitated inside the micelle core, the surfactant is not
efficiently adsorbed on the micelle surface, and does not
protect the particles against the further growth. To some
extent the control of particle growth could be achieved by
diminishing the reaction time, and similar effect has been
also observed in our case. However, the more detailed
discussions of this process will not be possible until some
additional experiments are accomplished, and it will be
presented in one of our subsequent papers.
Although the control of particle size is not sufficiently
possible with RM process, powders obtained by this method
have slightly smaller grain size, are less agglomerated and
crystallite size determined from XRD is closer to grain size
obtained from TEM photograph, as compared to CP-made
material.
In GdF3-MF (MZLi, Na, K and Cs) system only NaGdF4
and KGdF4 could have been obtained by co-precipitation
method. Although LiGdF4 precipitates from solution, it
always contains admixture of GdF3, which results from low
solubility of lithium fluoride, which cannot be applied at
sufficient excess. The solubility of CsF is considerably
larger, but it seems that also CsGdF4 is considerably better
soluble than GdF3, and the latter compound is the only
product of CP reaction.
3.4.3. GdF3(s)CNaF reaction (SR)
The difference in solubility enables transformation of
solid GdF3 into NaGdF4 by heating of the former
compound in NaF aqueous solution at 90 8C. This process
may be seen as related to hydrothermal method, however,
it is accomplished at significantly milder conditions, under
normal pressure and at temperature below the boiling point
of the solvent. Among fluorides of MGdF4 (MZLi, Na, K,
Cs) type only NaGdF4 could have been obtained by this
process. Unexpectedly, the GdF3(s) does not react with KF
solution, although KGdF4 may be synthesised by co-
precipitation or microemulsion method. Due to low
solubility of GdF3 the growth of NaGdF4 particles is
very slow and results in large monocrystalline particles,
with size in 0.5–1 mm range, which in contrast to CP or
RM-made powders, do not contain surface adsorbed water.
The overall water content is for as-prepared SR sample
3.4.2. Microemulsion synthesis (RM)
Microemulsion synthesis has been tried with hope, that it
allows to obtain fluoride nanoparticles with controllable
size. By the confinement of the microemulsion droplets with
manipulating the wZ[H2O]/[CTAB] ratio it should be
possible to control the size of resulted powders. To check
the influences of water content on nanoparticle size,
microemulsion with different w ratio in the 5–20 range,
were prepared. However, the w ratio has not affected the
average size of particles. Moreover, we have not been able

M. Karbowiak et al. / Journal of Physics and Chemistry of Solids 66 (2005) 1008–1019
1019
significantly lower than for CP or RM processed materials
Acknowledgements
and comparable to CP or RM samples dried at 300 8C.
However, the lowest water content and best crystallinity of
SR sample is not accompanied by superior optical proper-
ties, compared to CP or RM samples.
This work was supported by the Polish Committee for
Scientific Research within the Project No. 7 T09A 094 20,
which is gratefully acknowledged. One of the authors (Artur
Bednarkiewicz) is a Polish Foundation for Science scholar-
ship holder.
4. Summary
References
In this paper, we have proved that precipitation from
solution followed by an appropriate thermal treatment is a
successful synthesis method for preparation of fluoride
phosphors. The highest practical potential has the co-
precipitation process (CP), since, it is relatively simple,
cheap (lanthanide oxides may be used as starting materials
instead of more expensive and hardly available at required
purity lanthanide fluorides) and there is no differences in
optical properties between material prepared by CP method
and solid state reaction at high temperatures. The precipi-
tation method yields powders with crystallite size in
nanometric scale, which additionally may be smoothly
changed up to w1 mm, what is an important advantage of
this method of synthesis compared to SS reaction.
[1] C.R. Ronda, J. Alloys Comp. 225 (1995) 534.
[2] G. Blasse, B.C. Grabmaier, Luminescent Materials, Springer, Berlin,
1994.
[3] N. Martin, P. Boutinaud, R. Mahiou, J.-C. Cousseins, M. Bouderbala,
J. Mater. Chem. 9 (1999) 125.
[4] F. You, Y. Wang, J. Lin, Y. Tao, J. Alloys Comp. 343 (2002) 151.
[5] C.J. O’Connor, V. Kolesnichenko, E. Carpenter, C. Sangregorio,
W. Zhou, A. Kumbhar, J. Sims, F. Agnoli, Synthetic Met. 122 (2001)
547.
´
[6] A. Mech, M. Karbowiak, L. Kepinski, A. Bednarkiewicz, W. Strek,
J. Alloys Comp. 380 (2004) 315.
[7] D. Zakaria, R. Mahiou, D. Avignant, M. Zahir, J. Alloys Comp. 257
(1997) 65.
[8] B.E. Warren, Diffraction by Imperfect Crystals, in X-Ray Diffraction,
Dover Publications, Inc., New York, 1990. p. 251.
[9] R.E. Thoma, H. Insley, G.M. Hebert, Inorg. Chem. 5 (1966) 1222.
[10] K. Nakamoto, Infrared and Raman Spectra of Inorganic and
Coordination Compounds, Wiley, New York, 1978. p. 227.
The interesting finding of this study is that emission
5
decay of the D0/7F1 transition of Eu3C is longer for
¨
[11] C. Gorller-Walrand, K. Binnemans, L. Fluyt, J. Phys.: Condens.
unheated NaGdF4 sample than for the same heat-treated
powder, in spite of larger content of OHK groups observed
for the former specimen. We proposed, that this effect may
be attributed to the increase of the local lanthanide ion
symmetry for smaller nanocrystallites [14] or it may be due
to the fact that radiative lifetime strongly depend on
effective index of refraction of the media consisting of
nanoparticles and the substance filling space between them
medium [15].
Matter. 5 (1993) 8359.
[12] B. Henderson, G.F. Imbusch, Optical Spectroscopy of Inorganic
Solids, Clarendon, Oxford, 1989.
[13] H.S. Kiliaan, J.F.A.K. Kotte, G. Blasse, Chem. Phys. Lett. 133 (1987)
425.
[14] Y.V. Malyukin, A.A. Masalov, P.N. Zhmurin, Phys. Lett. A 316
(2003) 147.
[15] R.S. Meltzer, S.P. Feofilov, B. Tissue, H.B. Yuan, Phys. Rev. B 60
(1999) R14012.
[16] M. Roth, R. Hempelmann, Chem. Mater. 10 (1998) 78.