Clay Minerals (1999)34, 127-135

Use of an air brush to spray dry samples

for X-ray powder diffraction
S. H I L L I E R

Macaulay Land Use Research institute, Craigiebuckler, Aberdeen, AB15 8QH, UK

(Received 1 July 1997; revised 5 January 1998)

ABSTRACT: The construction and operation of a spray drier is described where the spray is
produced using an air brush, essentially a miniature spray gun. The spray-dried products consist of
spheres 50-60 gm in diameter and typical product recoveries are 80%, a marked improvement over
simple two-nozzle systems. The spray-dried samples are easy to load into XRD powder holders and
present a smooth surface and relatively constant bulk density to the X-ray beam. Problems of
preferred orientation are effectively eliminated and the resulting X-ray powder pattems are
completely reproducible by different operators.

The importance of sample preparation for X-ray quantitatively by comparing the intensities of
powder diffraction cannot be overemphasized observed (using both spray-dried and non spray-
(Brindley, 1984; Bish & Reynolds, 1989). One of dried samples) with calculated X-ray powder
the characteristics most commonly desired is a diffraction patterns for several minerals. They
completely random arrangement of the individual concluded that problems of preferred orientation
particles in a powder sample. All types of X-ray were minimized, possibly eliminated, by spray-
powder diffraction, ranging from structure to drying.
quantitative analysis, rely on the preparation of Nevertheless, as discussed by Bish & Reynolds
such random powder samples in order to obtain the (1989), spray drying is not yet widely used as a
correct relative intensities of all the peaks in the method of sample preparation for X-ray powder
diffraction pattern. Unfortunately, many materials, diffraction, possibly because of the high cost of
notably clay minerals, are orientation prone, and the commercially available equipment (>s and
preparation of a random powder sample is not a possibly because it does not appear suitable for small
simple matter. This is reflected by the large number samples of <1 g. Although commercial systems are
of methods that have been suggested in attempts to expensive, an instrument similar to that built by
produce random powders (see Bish & Reynolds, Smith et al. (1979a) can be constructed for a
1989, for a recent account). Indeed, it is probably moderate sum of ~s (materials and labour).
fair to say that truly random powder samples are However, according to Smith et al. (1979a), product
achieved less often than not. Among the techniques recovery with their insmament ran typically at 50%
used, one of the most effective is a method known as a result of the larger spray droplets not drying
as spray-drying (Jonas & Kuykendall, 1966; Hughes before they reached the bottom of the chamber. This
& Bohor, 1970; Smith et al., 1979a,b). Essentially, is in spite of the fact that the chamber was quite
this method consists of spraying a sample as an large, 120 cm in height and 90 cm in diameter. For
aqueous suspension into a heated chamber so that it many applications this probably represents an
dries in the form of the spherical spray droplets. unacceptably high loss, especially if the amount of
The resulting dry product consists of thousands of sample available is limited.
tiny spherical agglomerates of the sample compo- Central to the spray-drying process is the
nents. Smith et al. (1979a) evaluated this method production of the spray itself. This may be

(C) 1999 The Macaulay Land Use Research Institute

 128 S. Hillier

achieved by one of four different methods, namely: EQUIPMENT AND METHODS

rotary atomization, pressure atomization, pneu-
matic atomization, or ultrasonic atomization. The C o n s t r u c t i o n o f the s p r a y d r y e r
spray dryer constructed by Smith et al. (1979a)
used pneumatic, also known as two-fluid, atomiza- The spray dryer consists of two parts, the
tion whereby a stream of compressed air from a spraying system and the drying chamber (Fig. 1).
nozzle impacts a water/sample slurry exiting from The construction of the drying chamber is similar
another closely spaced nozzle and breaks it up into to that described by Smith et al. (1979a), but is
a spray. Compared to the other methods of smaller in size. The chamber is made from 2 mm
atomization, pneumatic atomization characteristi- aluminium sheet formed into a cylinder of 45 cm
cally produces the widest distribution of droplet diameter and 90 cm in height. A Watlow Thinband
sizes (Shaw, 1990). In the course of designing a Heater of two-piece construction (drawing 1500 W
laboratory size spray dryer at the Macaulay per half, and connected to form a band 45 cm in
Institute, trials were made with a design similar diameter and 15 cm in height) is clamped around
to that built by Smith et al. (1979a) using various the base of the cylinder. Due to differential
sizes and arrangements of nozzles, and a range of contraction and expansion of the cylinder and the
air pressures. In all cases, product recovery was heater, strips of aluminium are fixed above and
disappointingly low, never better than 50%, and below the heater to hold it in place. The heater is
clearly the result of the wide distribution of droplet controlled via a United Automation solid-state
sizes formed by this simple method of atomization. temperature controller with a range of 0 150~
Basically, it proved impossible to avoid producing connected to a stainless steel temperature sensor.
a large proportion of droplets that were much too The lid of the chamber is made from a sheet of
big to dry before reaching the bottom of the aluminium formed into a 90 ~ cone with a 2.5 cm
chamber, the same problem encountered by Smith diameter hole at its apex through which the
et aL (1979a). samples are sprayed. The lid is permanently
From these tests it became clear that one of the fixed to the cylinder by brackets and the joint
keys to spray drying small samples on a laboratory sealed with thermal cement. The base of the
scale, for the purpose of X-ray powder diffraction, chamber is supported 0.5 cm above the workbench
is the production of a spray with a relatively small by two insulated blocks of aluminium bolted to the
9 and uniform droplet size distribution. This can be sides.
achieved at a relatively moderate cost (<s by The spray is produced by a Crescendo No. 175-7
replacing the two-nozzle system of atomization air brush manufactured by Badger Air-Brush Co.
with an air brush, essentially a miniature spray gun. 9128 Belmont Av., Franklin Park, IL 60131, USA.
The spray from an air brush is also produced The air brush is fitted with its largest spray head
pneumatically but the two fluids are mixed regulator (slightly enlarged from 1.6 to 2 mm
internally resulting in a narrower and more easily diameter) and needle designed for spraying the
controlled droplet size distribution. Although rotary, most viscous materials. Samples are fed into the air
pressure and ultrasonic atomization can also brush via a disposable plastic pipette trimmed to fit
produce narrow droplet size distributions, rotary with the air brush sample inlet via a small piece of
and pressure methods require large drying chambers silicon rubber tubing. Larger samples, up to 60 ml,
due to the high exit velocities of the sprays, whilst can also be fed from the paint bottles supplied with
ultrasonid equipment is expensive. Furthermore, the the brush, but the gravity-fed arrangement using the
ultrasonic arrangement in one commercial apparatus pipette is usually more convenient as it ensures that
frequently suffers from the development of large all of the sample is sprayed. The air brush is
liquid drops on the probe tip, and 30% recovery is arranged to spray vertically downwards into the
considered exceptional, although this may also be chamber and is simply held in position while
due to other aspects of its design (Chipera pers. spraying by hand. Holding the air brush by hand
comm.). The purpose of this paper is to describe the also enables the suspension to be continuously
construction and operation of a laboratory-scale agitated during spraying, or even to stop, shake, and
spray drier where the spray is produced by an air restart, in order to avoid problems of segregation or
brush and to illustrate the results which have been increasing viscosity due to sedimentation in the
obtained by this method. container.
 Spray drying by air brush 129

Sample holder
~ Airbrush

Ternperature
sensor
:: i i ~.~..

Compressed ! ~
air supply

Cylinder
(45cm X 90cm)

Heater

- - ~ , -- Temperature
Sheetof paper / ~ controller

FIG. 1. Diagram to show the construction and operation of the spray drier.
9 1999 The Macaulay Land Use Research Institute.

Sample preparation suspension needs to have as high a content of solids

as possible in order to minimize drying time and to
Essentially, the preparation of samples for spray ensure that the droplets retain a spherical shape as
drying involves nothing more than making a they dry. On the other hand, the suspension must
concentrated aqueous suspension of the material still be sufficiently dilute to be sprayed easily
that can be sprayed through the air brush into the through the air brush at low pressures without
drying chamber. However, the preparation of clogging. Experience has shown that the best
samples for spray drying is one of the most results, in terms of product recovery, are obtained
important aspects of the process and probably the by spraying at as low a pressure as possible
one that is most likely to result in failure ( 1 0 - 1 5 psi). Observation suggests that this is
(Lusasiewicz, 1989). This is because sample because the higher the pressure, the higher the
preparation involves a delicate balance between proportion of very small droplets that are produced,
two conflicting aims. On the one hand, the sample in fact so small (<10 ~tm) that they tend to remain
130 S. Hillier

in suspension in the hot air, and are carried away in Because of the large quantities of PVA solution
the atmosphere rather than sedimenting to the required, smectites are undoubtedly the most
bottom of the chamber. In addition, low pressures difficult samples to spray dry. In addition to the
allow the droplets to pass more slowly through the fact that the high water content of the droplets
chamber giving a longer contact time with the hot makes them more difficult to dry, it also tends to
air and a better chance of drying before they land result in many more smaller particles, which are
on the paper underneath. The concentration much more difficult to recover from the air stream,
(viscosity) of the suspension also affects the size and a more irregular shape than those Which are
of the spray droplets that form, more concentrated recovered, probably due to the drastic change in
suspensions forming larger droplets, but being more volume of the droplet during drying. Nevertheless,
difficult to spray at low pressures. by operating at pressures as low as 5 - 1 0 psi, even
Sample suspensions are prepared by dispersing clay fractions consisting of smectites can be
the solid material in a 0.5% (w/v) aqueous solution successfully spray dried with solid product recov-
of polyvinyl alcohol (PVA) of 30,000-70,000 eries of up to 70%.
molecular weight. The PVA acts as a binder As far as clay minerals are concerned, a simple
giving strength to the dried product. However, for method of determining a reasonable concentration
samples with a high clay content it has been found of solid to PVA solution is to add the solution
that the addition of a binder is not essential. progressively to the clay until such point that the
A d d i t i o n o f a s i n g l e drop o f 1 - o c t a n o l resulting suspension does not adhere to the sides of
(CH3(CH2)7OH) prevents foaming during disper- its container after vigorous shaking. As a general
sion, makes transfer of samples between containers rule, most suspensions are limited to between
easier, and reduces the incorporation of air bubbles 30 40% solids by v o l u m e before viscosity
into the dried granules. As mentioned previously, increases excessively (Lukasiewicz, 1989).
the suspension needs to have as high a content of
solids as possible, but the amount of solid that can
Operation o f the spray dryer
be suspended in the PVA solution depends on the
nature of the solid itself. For samples of muscovite, The first step is to heat the drying chamber. The
wollastonite, pyrite, siderite and corundum, Smith exit air t e m p e r a t u r e from the c h a m b e r is
et al. (1979a) used 3 - 4 g of powder suspended in monitored by the sensor and typically takes
6 - 7 ml of a 1% aqueous solution of polyvinyl - 1 0 - 1 5 min to reach the maximum temperature
alcohol (PVA). This equates to solid to liquid ratios of 150~ At this maximum setting, the air
(w/v) ranging from 1:1.5 to 1:2.3. In the present temperature on the surface beneath the chamber
investigation, all non-clay minerals have been reaches -90~ E x i t air t e m p e r a t u r e s o f
successfully spray dried using a solid to liquid 100-150~ are suitable for most samples, lower
ratio of ~1:2. In contrast, the solid/liquid ratios for temperatures resulting in insufficient drying. Once
clay minerals are much more variable and some the desired temperature is attained, a sheet of A2
clays need to be diluted considerably in order to paper is placed beneath the chamber. The
make a suspension with the desired rheological temperature sensor is then removed from the
properties. In general, the finer grained the sample chamber, the air brush positioned above it, and
and/or the greater the amount of swelling clay the sample immediately sprayed into the chamber
layers present, the more PVA solution necessary to through the hole in the lid. This typically takes
make a suitable suspension. This is because for a one to a few minutes, depending on the air
given volume concentration of solids in suspension, pressure and the viscosity of the sample. Once
the mean distance separating particles decreases all the sample has been sprayed, the paper is
with decreasing particle size resulting in more removed from beneath the chamber and the sample
interaction between particles (Lukasiewicz, 1989). collected by pouring it into a vial. Between
An extreme case is illustrated by that of Na- samples the air brush is easily cleaned by spraying
montmorillonite. For the raw product (SWy-I) and back-flushing water at high pressure (40 psi).
suspended by shaking, a ratio of 1:13 was necessary The sample chamber does not normally require
to make a suspension that can be sprayed, and for a any cleaning between samples, but can be blown
well dispersed <2 lain fraction of this clay, this ratio out with compressed air if desired, and is cleaned
was increased to 1:60. periodically when not in use.
 Spray drying by air brush 131

RESULTS AND DISCUSSION Most of the loss appears to be due to the former
mechanism, but losses by both are reduced by
Product form and recovery spraying at low pressure. The difference in recovery
between the samples is probably due to the effect of
Typical examples of spray-dried products are suspension concentration. At 40% (w/v) the kaolinite
shown in Fig. 2. Based on measurements made on suspension is much more concentrated than the
these and many other similar photomicrographs, the montmorillonite suspension at 7.1% (w/v). Although
products consist of spheres with an average size not all of a sample is recovered, for most materials
~50 gm. Weighing measured volumes of the recovery o f - 8 0 % is typical, and represents a
products and assuming typical densities and a significant improvement on the 50% attained by
porosity -30%, both between and within the Smith et al. (1979a). Without doubt the most difficult
spheres, i.e near ideal close packing, suggests that samples to process are clay fractions of well
the majority (~75%) are solid rather than hollow dispersed (e.g. Na-saturated) smectites for which
spheres. Three of the samples are of spray-dried recovery may be as low as 20%.
<2 gm clay fractions (Fig. 2A,B,C,D), that of the
muscovite and kaolinite produced quite smooth and
Random powder patterns
uniform spheres, whereas the spheres of the
montmorillonite have a more irregular surface. The aim of spray drying materials for X-ray
One of the samples shown illustrates the effect of powder diffraction is to produce a sample which
the grain size of the materials on the form of the can be loaded into a powder holder without
product (Fig. 2E). Although the granules of the inducing a preferred orientation to any of the
spray-dried talc are roughly spherical, they have a phases present. To assess the effect of spray drying
quite irregular surface due to the relatively large as a means to this end, powder patterns were
size of the component particles. Examples of a recorded from samples of chlorite (CCa-1) and
shale and sandstone which have been prepared by kaolinite (KGa-2) prepared as both freeze-dried and
spray drying the slurry obtained directly from a spray-dried powders. The samples were mixed in a
McCrone mill are also shown (Fig. 2G,H). 1:1 proportion by weight with corundum, wet
Product recovery was determined by weighing the ground and homogenized in a McCrone mill for
amount of material recovered by pouring it off the 12 min, and the resulting slurries freeze dried. After
sheet of paper after spraying measured amounts of freeze drying, a portion of each sample was re-
suspensions of known concentrations. Two clays suspended in 0.5% PVA solution and spray dried.
were used for this purpose: Na-montmorillonite Both chlorite and kaolinite are prone to orientation
(SWy-1) and a kaolinite (AIPB) of unknown origin. with respect to 001, whereas corundum is generally
The results are given in Table 1. For both samples considered to be free from problems of preferred
the best recovery was obtained at the lower pressure orientation to the extent that it is frequently chosen
of 10 psi, especially for the montmorillonite. In both as an internal standard for quantitative phase
cases the total loss is due to the loss of two fractions: analysis. Both f r e e z e - d r i e d and spray-dried
the fraction of droplets so small that they remain in samples were loaded into powder holders and
the air stream, and the fraction of droplets which are XRD patterns recorded from 2 - 7 5 ~ on a
not dry and so splatter on the paper as they land. Siemens D5000 with a 0/0 goniometer, using Co

TABLE 1. Results of product recovery tests.

Sample identity Origin Air pressure Concentration Wt of clay Wt clay % clay

(psi) of clay (wt%) sprayed (g) recovered (g) recovered

Kaolinite AIPB Unknown 10 40 6.34 5.39 85

Kaolinite AIPB Unknown 15 40 7.41 5.99 81
Smectite (SWy- 1) Wyoming 10 7.1 1.07 0.82 77
Smectite (SWy-1) Wyoming 15 7.1 0.97 0.56 58
132 S. Hillier

.......... ,,,~i~ ~~ g, ~ ~
...... ,L:?)? ;:

.............. :

FIO. 2. Scanning electron photomicrographs of spray-dried samples. (A) kaolinite (KGa-lb) <2 btm fraction; (B)
muscovite (No 5 Macaulay collection) <2 Ixm fraction; (C) montmorillonite (SWy-1) <2 lam fraction; (D) close-
up of C; (E) talc (No 8 Macaulay collection) as supplied by Hopkins & Williams Ltd; (F) corundum (No 3
Macaulay collection) <10 ~tm as supplied by Aldrich/Sigma Chemicals Ltd.; (G) shale, Reading beds, McCrone
milled 12 min; (H) glauconitic sandstone, Denmark, McCrone milled 12 min. All scale bars are 50 I~m long.
 Spray drying by air brush 133

radiation, diffracted beam monochromator, 1~ slits to 0.82 and 0.75 indicating significant preferred
and counting for 1 s per 0.02~ step. orientation, equivalent to compaction of the chlorite
Because the spray-dried powders flow very plates from a uniform distribution by 18 and 25%,
easily, loading of the powder holder is accom- respectively. For the kaolinite:corundum mixtures, a
plished simply by pouring in an excess of powder, similar trend was observed with the March function
gently tapping to distribute it evenly across the refining to 1.01 and 1.03 for the spray-dried
cavity and swiping away the surplus powder by samples, and to 0.94 for both of the freeze-dried
passing the edge of a glass slide across it to leave a samples. Quantitative phase analysis results from
smooth flat surface. Indeed, the spray-dried Siroquant were accurate to within 2% absolute for
powders are so fluid that problems of sample all spray-dried samples regardless of whether or not
stability might be anticipated with a 0/20 goni- preferred orientation was refined. In contrast,
ometer if the sample is tilted far from horizontal. In analyses of the freeze-dried chlorite:corundum
contrast, the freeze-dried powders were both lumpy mixtures were not accurate (66% chlorite, 34%
and fluffy, and much more cohesive than the spray- corundum, in the worst case) unless preferred
dried samples. As such, gentle tamping of the orientation was refined. Although neither of the
sample surface was unavoidable in loading them Rietveld refinements of the clay minerals is exact
into the holders. due to the presence of stacking disorder, as far as it
The low-angle portions of the resulting XRD is possible to judge by this method, it appears that
patterns are shown in Fig. 3. All four samples were the samples which have been spray dried are
run twice, the second run after a different operator randomly oriented, whereas the freeze-dried
had emptied and reloaded the sample holders. In all samples show significant preferred orientation.
patterns the intensities of the reflections from
corundum are virtually identical, except for being
slightly less intense in the spray-dried samples, CONCLUSIONS
presumably due to a somewhat lower bulk density
of these powders. Similarly, the patterns for each Spray drying on a laboratory scale is easily
spray-dried mixture are identical. Patterns recorded accomplished with a small drying chamber by
from the freeze-dried samples, however, show using an air brush to produce the spray. The
marked and variable increases in the intensity product consists of spheres with an average
ratio of the basal to the non-basal reflections of diameter of 50 gm, and 70-80% recovery is
the clay minerals compared to the spray-dried typical for most samples. The spray-dried powders
patterns. Clearly, the increases in intensity of the are easily loaded into XRD powder holders in a
001 reflections in the freeze-dried samples must be form with a smooth surface and relatively constant
due to preferred orientation acquired during the bulk density. The resulting XRD powder patterns
loading of these samples into the holder. In are reproducible and appear to be random,
contrast, for the spray-dried samples, the degree exhibiting no preferred orientation. Consequently,
of any preferred orientation is clearly much less, this method of sample preparation is ideally suited
and is unaffected by the loading and packing to problems such as quantitative phase analysis.
process. Furthermore, because wet grinding is preferable to
in an attempt to assess the degree of preferred dry grinding for the reduction of particle size for
orientation in the samples, observed XRD patterns X-ray powder diffraction, a scheme such as
were compared to patterns calculated using the McCrone milling followed by spray drying of the
Rietveld quantitative phase analysis program, resulting slurry, can be used as a routine, rapid, and
Siroquant. In all cases without refinement of reproducible method of sample preparation.
preferred orientation, the fit between observed and
calculated patterns was best for the spray-dried
ACKNOWLEDGMENTS
samples. When refinement of preferred orientation
was included in the analysis, the March function Thanks are due to Allan Wilson and Gordon Ewen for
(Dollase, 1989) for the spray-dried chlorite:cor- construction of the drying chamber, and to Martin Roe
undum mixtures refined to 0.99, indicative of for SEM photography. The manuscript was improved
virtually no preferred orientation. In contrast, the by the constructive comments of Steve Chipera and an
March function for the freeze-dried samples refined anonymous referee.
134 S. Hillier

K
E

~D

N ~g

~.~
o

m o ~2
q)

g
o

$+64=
,o ~ZE

-e=
m~

-g
_d
~ . ~ "7,

N ~g~

~~

' .

~7
 Spray drying by air brush 135

REFERENCES 1780 1786.

Jonas E.C. & Kuykendall J. R. (1966) Preparation of
Bish D i . & Reynolds R.C. Jr. (1989) Sample prepara- montmorillonites for random powder diffraction.
tion for X-ray diffraction. Pp. 73 99 in: Modern Clay Miner. 6, 232-235.
Powder Dij]?action (D.L. Bish & J.E. Post, editors) Lukasiewicz S.J. (1989) Spray-drying ceramic powders.
Reviews in Mineralogy, Volume 20, Mineralogical
J. Am. Ceram. Soc. 72, 617-24.
Society of America, USA. Shaw F.V. (1990) Spray-drying: A traditional process
Brindley G.W. (1984) Quantitative X-ray mineral
for advanced applications. Ceram. Bull. 69,
analysis of clays. Pp. 411-438 in: Crystal
Structures of Clay Minerals and their X-ray
1484-1488.
Smith S.T., Snyder R i . & Brownell W. E. (1979a)
Identification (G.W. Brindley & G. Brown, editors).
Mineralogical Society, London. Minimisation of preferred orientation in powders by
Dollase W.A. (1989) Correction of intensities for spray-drying. Advances in X-ray Analysis, 22,
preferred orientation in powder diffractometry: 77-87.
Application of the March model. J. Appl. Cryst. Smith S.T., Snyder R.L. & Brownell W.E. (1979b)
19, 267-272. Quantitative phase analysis of Devonian shales by
Hughes R. & Bohor B. (1970) Random clay powders computer controlled X-ray diffraction of spray-dried
prepared by spray-drying. Am. Miner. 55, samples. Advances in X-ray Analysis, 22, 181 191.