HIGH-SPEED IMAGING OF LIQUID METAL ATOMIZATION BY

TWO DIFFERENT CLOSE-COUPLED NOZZLES

Steven P. Mates* and Gary S. Settles**

ABSTRACT

High-speed flow visualization is used to examine the physics of metal powder production via
close-coupled gas atomization. The aim of this study is to evaluate current-generation close-coupled
nozzles on the basis of their ability to atomize liquid metal effectively. Results of this evaluation will be
used to guide efforts to improve nozzle design, using gas dynamic principles, in order to enhance metal
powder production capability. An annular HPGA (converging) nozzle and an annular Ünal (converging-
diverging) nozzle are investigated as examples of current-generation technology. Both nozzles are used
to atomize molten tin at similar operating conditions, with gas pressures and metal flow rates
commensurate with current standards for nozzles of this scale. Strobe-illuminated schlieren visualization
is used to image features of the supersonic atomizing gas flow and liquid metal breakup simultaneously,
both close to the nozzle tip and up to 12 nozzle diameters downstream. Water is also atomized to study
the influence of fluid surface tension on atomization physics. Results show that the low-speed
recirculating wake flows at the blunt tips of both nozzles shield liquid metal from high-velocity gas flow,
thereby delaying effective atomization. Observations also indicate that primary breakup by both nozzles
is highly unsteady, and that secondary atomization of large molten ligaments persists well beyond the
nozzle tip. Potential improvements to nozzle design include elimination of the blunt-base recirculation
region and the establishment of a supersonic jet of maximum length.

INTRODUCTION

Metal powder production via close-coupled gas atomization is a complex process involving a
wide array of physical phenomena. Powder is produced by the violent disruption of a stream of molten
metal with a coaxial, co-directional supersonic gas flow. To date this process is not well understood, due
in part to its inherent complexity. A better understanding of the physics involved is desirable in order to
meet current demands for more cost-efficient production, improved control of powder properties, and
finer mean particle sizes.
In particular, there appears to be strong a potential for enhancing powder production capability
through improved nozzle design. First, current nozzle technology, as published in the literature, seems to
be under-developed. For example, since molten metals typically have high surface tensions, very high
gas velocities, often supersonic, are required to atomize them effectively. It is surprising, therefore, that
most nozzles achieve these high gas velocities inefficiently by operating a sonic gas orifice at very high
pressure ratios, producing strong, energy-dissipating shock waves. Converging-diverging nozzles, well
known in aerospace disciplines, can be designed to produce shock-free supersonic flow at any practical
Mach number. However, this technology has not been fully exploited by the powder metallurgy industry.
Second, little advantage has been taken of the large body of fundamental research dealing with twin-fluid
atomization and its application to fuel spray technology [1]. This work represents the most extensive
knowledge base on the subject and remains largely untapped.
For these reasons a research program was initiated at the Penn State Gas Dynamics Lab to: 1)
examine the physics of liquid metal atomization using current-generation close-coupled nozzles, and
characterize the gas dynamics involved, 2) evaluate the ability of current nozzles to produce fine metal
powder, and 3) explore ways to improve nozzle design, using gas-dynamics principles and related fuel
atomization technology, in order to enhance powder production capability.
In this study we examine an HPGA nozzle and a Ünal nozzle. Our laboratory nozzles, described
in previous work [2], were constructed based on nozzle dimensions published by NIST [3], Anderson et.
al. [4] and Ünal [5]. They are included in this study for two reasons. First, both are reported to be
particularly effective at producing fine powder [4,6]. Second, they are the only nozzles that have been
studied openly in the literature, and are thus the only ones available for evaluation. It is possible that
other nozzles are in use in the industry, but details of their construction are proprietary.
The HPGA nozzle is similar to a widely-used family of nozzles that can be characterized as sonic,
using either an annular gas orifice or discrete jets, with a tapered liquid metal delivery tube. Because of
its perceived ubiquity, the HPGA nozzle is regarded as most representative of current technology. The
Ünal nozzle employs an annular converging-diverging supersonic gas nozzle, and has a straight-walled
delivery tube. In order to compare these nozzles on an equal basis, they are operated in identical fashion
in this study. Therefore the Ünal nozzle is operated in a downwards orientation in this study rather than
an upwards orientation as was the case in Ünal’s original work.
The first step in the evaluation of current nozzle technology involved a comparative study of the
gas dynamic behavior of these nozzles over a representative range of operating pressures, and was
reported in [2]. The nozzle flowfields were visualized using the schlieren technique, and aspiration
pressures were measured. The HPGA nozzle was found to produce a highly-under-expanded flowfield
and strong shock waves at elevated operating pressures. Pitot pressure measurements indicated that the
flow reached Mach 5 just outside the nozzle exit. The Ünal nozzle was found to better expand the gas
flow because of its converging-diverging design. Both nozzles produced large, low-speed, recirculating
wakes at the blunt tips of their respective liquid delivery tubes. Qualitative agreement of both the flow
visualization results and the aspiration pressure measurements with earlier work [3,4,6] showed that the
HPGA and Ünal nozzles examined in [2] and the present study are, from a gas dynamics standpoint,
accurate representations of the original nozzles they were intended to emulate.
The aim of the present study is to complete the next step in this evaluation, namely to examine
the physics of liquid atomization using the HPGA and Ünal nozzles. To do this, the laboratory atomizer
described in [2] was used to atomize small quantities of water and molten tin. Tin was selected because
its low melting temperature and low toxicity made it easy to handle. Although tin is regarded as “easy” to
atomize because of its low surface tension compared to other metals of interest, such as stainless steel [7],
it is representative of metals in general, which have much higher surface tensions than all other common
fluids. Water, having a very low surface tension compared to tin, was atomized as well to illustrate the
dominant role this fluid property plays in the atomization physics. Gas-to-metal mass flow ratios used in
these preliminary experiments were selected based on levels regarded as being economically desirable
[8].

High-speed flow visualization was used to visualize features of the supersonic atomizing gas flow
and liquid breakup simultaneously. The present laboratory atomizer is particularly useful for this purpose
because it offers an unobstructed view of the nozzle and its flowfield, which facilitates high-quality
imaging of the atomization process. Previous attempts to visualize liquid metal atomization [3,9] have
suffered from limited optical access to the nozzles, which are often confined within enclosed vessels on
commercial equipment. In addition, visualization of the gas flow during metal atomization has never
before been reported. The results of this study are used to make qualitative assessments of the
atomization performance of these nozzles and to suggest design alternatives.

EXPERIMENTAL EQUIPMENT AND METHODS

Dimensions of the present HPGA and Ünal nozzles are shown in Figure 1. The nozzles have
throat areas of 9mm2 and 13 mm2, respectively. The nozzles’ liquid metal delivery tubes have identical
tip diameters (8 mm) and liquid orifice diameters (2 mm). These dimensions were made similar in order
to facilitate performance comparisons. The nozzles are operated at pressure ratios between 14-55
(plenum pressures between 2-5.5 MPa), where the pressure ratio is defined as the ratio of the plenum
stagnation pressure to ambient (atmospheric) pressure (Po/Pa). Details of the high pressure air delivery
system and atomizer construction are given in [2].

HPGA Nozzle Ünal Nozzle

Figure 1. HPGA and Ünal nozzles. Dimensions are in millimeters.

Figure 2 is a schematic of the laboratory atomizer. A key feature of the design is the combination
liquid reservoir/delivery tube, which is made entirely from quartz. The liquid reservoir has a practical
capacity of about 1 kg of tin. The delivery tube, made from 6 mm o.d. by 2 mm i.d. quartz capillary
tubing, is fused to the bottom of the reservoir. The delivery tube is inserted into the aluminum centerbody
to insulate the molten tin as it flows down through the centerbody and out the nozzle tip during
atomization. Secondary heating of the centerbody by two 35W strip heaters was required to keep the tin
from freezing in the delivery tube. The inner diameter of the delivery tube was selected to provide a
realistic gas-to-metal mass flow ratio during atomization [8]. Tin is melted and superheated by a 410W
electric tube heater mounted around the reservoir. Metal flow is initiated by pulling the quartz stopper
rod that blocks the entrance to the delivery tube at its junction with the reservoir.
The gas-to-metal mass flow ratios used in the study are estimated to range between 1 and 2 for
both nozzles, with metal flow rates estimated to be between 0.03-0.05 kg/s. Both the metal flow rate and
the gas-to-metal mass flow ratio increase with increasing operating pressure. Metal flow rates were
estimated by measuring water flow rates over a range of operating pressures and extrapolating the results
to molten tin, similar to the method used by Strauss [10]. Gas flow rates were calculated based on the
measured plenum pressure and the nominal throat areas of the nozzles. The uncertainty in these estimates
is high, possibly as much as 40 percent for the gas-to-metal ratios, but is considered acceptable for the
purposes of this study.

Figure 2. Schematic of the laboratory atomizer. Not to scale.

Features of the supersonic flowfields produced by the nozzles are visualized using the schlieren
technique [11]. A General Radio Strobotac was used as the light source of the schlieren system, shown in
Figure 3, in order to “freeze” fast-moving droplets during atomization. At the magnifications used in this
study, the Strobotac flash duration of 1.2 µs could stop the motion of droplets moving up to
approximately 100 m/s in the flowfield.
 Images of atomization were captured using two different methods. For high-magnification
imaging of liquid breakup within 3-4 diameters of the nozzle tip, a CCD video camera was used. The
Strobotac was synchronized with the video camera so that one flash occurred for each non-interlaced field
acquired by the CCD array. Frontlighting via a second Strobotac was often used with this arrangement to
help visualize details on the surface of the spray. RS-170 video output from the camera was recorded on
S-VHS videotape and later digitized using a PC-based frame grabber. The CCD array could resolve
objects in the flowfield down to approximately 100 µm.
Low-magnification images of atomization up to 12 diameters downstream of the nozzle were
captured with a 35mm film camera. The 35mm film provided the high spatial resolution required to
capture fine details in the atomization spray at low magnification. The Strobotac was triggered by the
camera shutter so that one flash occurred each time the shutter opened. A motor drive was used to
automatically wind the film after each picture so that up to three frames could be exposed per second. At
the highest framing rate approximately 10 photos could be taken during a typical atomization run. Black
and white negatives were printed on photographic paper and then scanned into digital format.

Figure 3. Schematic of the schlieren visualization system.

RESULTS

In the following discussion of results, all distances mentioned while describing atomization are
given relative to the diameter of the liquid delivery tube tip (8 mm, see Figure 1).

HPGA Nozzle
Gas-Only Flowfield. A time-averaged schlieren image of the “open-wake” HPGA flowfield is shown in
Figure 4, along with a sketch indicating important flow features. A more complete description of the
flowfield over a range of operating pressures is given in [2]. Of special interest to this study is the low-
speed wake flow at the nozzle tip. The nature of the wake flow is very similar to the separated “blunt-
base” flows produced by military projectiles, which have been studied since the 1940’s. These complex
flows are known to contain low-speed recirculating zones very close to the blunt base. Recently, Herrin
and Dutton [12] made detailed wake velocity measurements behind a cylindrical afterbody in supersonic
flow. Their measurements showed that the recirculating zone extends roughly one diameter downstream
of the blunt base, after which the flow on centerline begins to re-accelerate in the freestream direction.
Their measurements also indicated that radial gas velocities along the base are very low and that the
pressure distribution along the base is nearly uniform. These results are useful in understanding the gas
dynamics of the wake flow at the tip of the present HPGA nozzle.

(a) (b)

Figure 4. “Open-wake” HPGA flowfield. (a) Schlieren image (time averaged). (b) Flowfield sketch.

Atomization of Water. Near-field images of the HPGA nozzle atomizing water at three different
operating pressures are shown in Figures 5a-c. The outer surface of the water spray, illuminated by
frontlighting, is medium gray in contrast to the lighter supersonic gas flow surrounding it. Its smooth
appearance belies the fact that it is actually made up of very small, fast-moving water droplets that are too
small to be resolved by the video camera and are moving too fast to be stopped by the strobe light.
It is evident from Figures 5a-c that the shock and expansion waves in the supersonic flowfield
strongly influence the shape of the liquid spray boundary in the near-field (2-3 diameters). These waves
give the spray a peculiar exterior contour, marked by a sharp cusp caused by the reflection of the internal
shock off the constant-pressure wake boundary. As the operating pressure rises, the internal shock grows
stronger and the cusp becomes more exaggerated. This distinctive contour also appears in images of
supersonic atomization of both liquid metal and water published elsewhere [3,9,13]. In addition, it is
noteworthy that the presence of water in the flowfield does not significantly alter the wave structure from
gas-only operation (Figure 4a). It is also important to note that the shock waves in the flowfield do not
appear to enhance liquid breakup directly.
 (a) (b) (c)

(d)

Figure 5. Atomization of water using the HPGA nozzle at Po/Pa= (a) 14, (b) 31, and (c) 51. Restricted
liquid flow in (d) shows large ligaments in the low-speed wake.

Less evident from the first three images in Figure 5 is that the water spray has a complex three-
dimensional structure close to the nozzle tip. Inspection of the videotape from which these images were
taken revealed that the water spray near the nozzle tip consists of a hollow “curtain” of fine droplets
around the outer edge of the wake. Hidden underneath this curtain are much larger liquid structures
inside the wake itself. To see these structures clearly, the water flow rate was reduced until the outer
curtain of droplets nearly disappeared, as shown in Figure 5d.
From the evidence shown in Figure 5, it appears there are two different primary breakup
mechanisms at work. Some water is drawn out into the shear layer at the nozzle lip, where gas velocity is
high, and is atomized into very fine droplets immediately. The remaining liquid flows into the wake and
is atomized less effectively by low-velocity gas. The first mechanism has been described by investigators
of both the HPGA and Ünal nozzles [3,4,14], who described that the recirculating gas flow in the wake
forces liquid to flow radially out along the blunt base of the delivery tube to the nozzle lip, where it is
stretched into a thin film and vigorously atomized. Figure 5d shows, however, that this “filming”
mechanism does not fully describe the primary atomization of water observed here, which has an obvious
and distinct dual-mode character.

Tin Atomization. Figure 6 shows molten tin atomization close to the HPGA nozzle tip. The first two
images in the figure show how the introduction of liquid metal into the flowfield affects the gas flow
pattern. Figure 6a, just prior to the start of atomization, shows the HPGA flowfield at a pressure ratio of
51, just after wake closure [2], with the characteristic Mach disk present two diameters from the nozzle
tip. In Figure 6b, taken 1/30 of a second later, molten tin is flowing and the wake has “opened up.” The
recompression shock has shifted away from centerline, the Mach disk has disappeared, and the wake has
grown in size. The shift to open-wake operation is clearly caused by the sudden flow of molten tin into
the wake, raising a question as to whether closed-wake operation can be sustained during atomization.
This is potentially important for atomization using HPGA-type nozzles, since closed-wake atomization
may be preferable [2]. However, this issue is set aside for future study.

(a) (b)

(c) (d) (e)

Figure 6. Tin atomization using the HPGA nozzle at Po/Pa=51. Images a-b are consecutive frames 1/30
sec apart showing how molten tin flow causes the wake to “open.” Images c-e show unsteady variation in
both amount and distribution of liquid in near-field during atomization. Arrows indicate fine droplets.

Discrete random images of tin atomization are shown in Figure 6c-e. These images are a small
sample of the hundreds of still frames recorded on videotape during a single atomization run. They show
that primary breakup of molten tin is highly unsteady. From one frame to the next, both the amount and
distribution of liquid in the flowfield varies significantly. While Figure 6c shows a large amount of liquid
spread throughout the wake, much less is present in Figure 6d. The comparatively modest amount of
liquid in Figure 6e is symmetrically distributed, while in Figure 6d nearly all of it is in the right half of the
flow. The cause of the unsteadiness is not well understood, but it is likely to be a consequence of
complex interactions between the molten tin and the wake flow. Unsteadiness of primary atomization
may be important because of its potential negative impact on resulting powder size distributions. Broader
distributions are possible if the metal flow rate is changing with time, as it appears to be in Figure 6,
because this may cause unpredictable variations in the relative amount of gas energy being applied to
atomize the liquid metal. Although further study is needed, there is some evidence that the flow of liquid
metal during atomization may be pulsatile. A precedent for this is found in the rocket fuel atomization
literature [15], which indicates pulsatile flow may be caused by certain combinations of gas and liquid
flow rates and nozzle dimensions. However, for present purposes, pulsatile flow is viewed as a
disadvantage.
Also evident from Figure 6 is the striking difference in the degree of liquid tin breakup in the
near-field compared to the earlier case of the atomization of water. Within the low-speed wake, which
stretches two diameters from the nozzle, the molten tin is disrupted into mostly large, irregular structures,
in contrast to water, much of which was immediately atomized into fine droplets. It appears that the
“filming” mechanism responsible for creating fine water droplets does not occur with molten tin. Only
near the end of the wake do the large structures break down into smaller ones, and fine droplets (which
appear as dark, shadowy patches in the flowfields shown in Figure 6) begin to appear. Thus Figure 6
shows that primary tin breakup occurs mostly under the influence of low-velocity gas within the wake,
and that secondary atomization by high-velocity gas commences beyond the wake, when the flow re-
accelerates to supersonic speed on centerline.
Figure 7 is a typical example of a far-field image showing tin atomization up to 12 diameters
from the nozzle, again illustrating how fine droplet production begins in earnest only near the end of the
wake. (The pressure ratio in this case is 31.) The wake contains large molten structures which appear to
be untouched by the high-velocity flow surrounding the wake.

Figure 7. Far-field image of tin atomization at Po/Pa=31. NOTE: Image has been rotated 90 degrees.

Figure 7 also shows that some large-scale molten metal ligaments persist far downstream of the
nozzle tip. Such a ligament is visible almost 5 diameters downstream of the nozzle tip surrounded by a
“cloud” of fine droplets. Similar observations were reported by Miller et. al. [9], suggesting that this may
be common, though contrary to general perception.
The fact that secondary breakup of large liquid metal ligaments continues far downstream has
several implications regarding the resulting powder characteristics. First, the existence of large molten
ligaments together with many small, presumably-solid particles in a turbulent jet suggests there is a high
probability of collisions between large molten particles and small solid ones. Such collisions may
promote the formation satellites in the resulting powder, which is a familiar but undesirable characteristic
of gas-atomized powders. Second, since gas velocity decreases with increasing distance from the nozzle
due to viscous dissipation, and at the same time the molten ligaments are accelerating, atomization will
not be as effective far downstream as it is close to the nozzle, where relative gas velocities are higher.
Larger solid particles may therefore result, thus widening the final powder size distribution.
 These observations lead to two conclusions about the qualitative performance of the HPGA
nozzle. First, the low-speed recirculating wake flow at the blunt tip of the delivery tube works against
fine powder production by sheltering molten metal from high-velocity gas. It may therefore be
advantageous to minimize the size of the wake by reducing the width of the blunt delivery tube tip. Such
a design would enhance the interaction between molten metal and high velocity gas in the nozzle near-
field. (This is standard practice in the design of fuel atomizers [1].) Second, since secondary breakup
continues to occur far downstream of the nozzle tip during normal operation, the nozzle should be
designed to maintain high gas velocity over as long a distance as possible in order to completely atomize
the liquid metal. It is known that sonic nozzles operated at high pressure ratios, such as the HPGA
nozzle, are not well suited for this purpose [16].

Ünal Nozzle

Gas-Only Flowfield. Figure 8 shows a time-averaged schlieren image and a flowfield sketch of the Ünal
nozzle operating near its design pressure ratio. Further details of the flowfield produced by this nozzle
over a range of operating pressures are given in [2]. The converging-diverging design of this nozzle
produces a better pressure match between the nozzle exit flow and the ambient atmosphere than does the
sonic HPGA design. This better match eliminates the strong internal shock present in the flowfield of the
latter nozzle. The other important difference between the Ünal and HPGA nozzle flowfields is that the
straight-walled delivery tube of the Ünal nozzle produces a larger wake than does the tapered delivery
tube of the HPGA nozzle.

Tin Atomization. Figure 9 shows the Ünal nozzle atomizing tin at a pressure ratio of 29, close to its
design pressure. The first image (Figure 9a) illustrates that the wake flow, by itself, contributes very little
to the breakup of molten tin. The liquid stream issuing from the nozzle is only slightly deformed by the
low-velocity gas flow in the wake. Clearly, the recirculating flow in the wake does not force the molten
metal to flow radially along the base of the delivery tube tip as described by Ünal [14]. However, the
metal flow rate used here may be close to an order-of-magnitude larger than what Ünal found to be
optimum for the “filming” mechanism to work properly [17]. In addition, the nozzle is used to atomize
tin in a downwards rather than upwards orientation, thus changing the direction with which gravitational
forces act on the liquid. Both may contribute to the difference in primary breakup physics from what
Ünal observed.

(a) (b)
Figure 8. Ünal nozzle operating at close to the design pressure. (a) Schlieren image (time averaged). (b)
Flowfield sketch.

(a) (b) (c)

Figure 9. Near-field images of tin atomization using the Ünal nozzle at Po/Pa=29 (design pressure).
Image (a) shows that the molten tin stream flowing into the wake on start-up is only slightly deformed by
low-velocity wake flow. Images (b) and (c) show atomization is similar to HPGA but with a noticeably
larger wake. Arrows indicate fine droplets.

The next two images (Figure 9b-c) show that primary breakup of molten tin is much the same as
it was using the HPGA nozzle. Large molten structures are present in the wake close to the nozzle tip,
and fine droplets are rare in this region. Also similar to the HPGA case is the fact that primary breakup is
highly unsteady. What is obviously different is that the larger wake of the Ünal nozzle delays fine
powder production for three diameters or more downstream of the nozzle tip, a significant increase
compared to the HPGA nozzle case. Thus it appears that the tapered delivery tube tip of the HPGA
nozzle promotes contact between high-velocity gas and molten metal closer to the nozzle by virtue of the
smaller wake it produces.
A far-field image of tin atomization using the Ünal nozzle is shown in Figure 10. It clearly shows
the elongated region over which large-scale structures are being inefficiently atomized by low-velocity
gas flow. By four diameters downstream, these large structures have mostly broken up. Similar to HPGA
atomization, however, most photographs taken during this run show some evidence of large-scale
ligaments being atomized farther downstream. In theory, the Ünal nozzle’s converging-diverging design
is better able to maintain high gas energy far downstream [16], which should enhance breakup of these
large ligaments. However, it cannot be determined from the images alone whether or not this design
feature improves atomization performance. Conclusive evidence must be obtained through comparisons
of powder size distributions, which are planned. From a qualitative standpoint, therefore, both the HPGA
and Ünal nozzles perform similarly when atomizing at similar conditions.
Figure 10. Far-field image of tin atomization using the Ünal nozzle at Po/Pa= 29. NOTE: Image has been
rotated 90 degrees.

SUMMARY

The physics of liquid metal atomization by two current-generation close-coupled nozzles was
examined using high-speed flow visualization. Imaging liquid breakup phenomena simultaneously with
the supersonic atomizing gas flow was used to better characterize both the gas dynamics of the process
and the mechanisms by which liquid is atomized by these nozzles. The objectives of the study were to
explore the physics involved and to qualitatively address the ability of these nozzles to atomize liquid
metal effectively. The results are used to identify the potential for improved nozzle design, using gas
dynamic principles and technology transfer from related fields, in order to enhance powder production
capability. Both an HPGA nozzle and a Ünal nozzle were used to atomize molten tin at operating
pressures and gas-to-metal mass flow ratios commensurate with current standards for nozzles of this
scale. Their performance was compared directly for the first time.
High-speed images of tin atomization using the HPGA nozzle showed that liquid breakup at the
nozzle tip is unsteady. Shock waves in the flowfield produce a peculiar contour in liquid spray boundary
close to the nozzle tip. These waves do not appear to enhance liquid breakup, however. The low-speed
wake at the blunt delivery tube tip effectively shields the molten metal from high-velocity gas close to the
nozzle, delaying effective atomization for up to two diameters from the nozzle tip. Although primary
breakup by the ‘filming’ mechanism was observed when atomizing water, it does not seem to occur with
molten tin. Far-field visualization of atomization up to ten diameters downstream shows that large molten
ligaments continue to undergo secondary breakup far from the nozzle tip, contrary to common perception.
The existence of these large structures far downstream may promote satellite formation and broad particle
size distributions, neither of which are desirable.
High-speed images of tin atomization using the Ünal nozzle show that, from a qualitative
standpoint, the physics of liquid breakup is similar to the HPGA nozzle. The principal difference is that
the larger wake flow produced by the Ünal nozzle delays effective atomization to even farther
downstream from the nozzle tip. Far-field images of atomization show that secondary breakup of large
ligaments is just as common as in the HPGA nozzle case. Although the converging-diverging design of
this nozzle is potentially better able to atomize these far-field ligaments, it remains to be determined
whether finer particles are produced as a result. Quantitative comparisons of powder size distributions are
planned to address this issue.
These observations of tin atomization have suggested ways to improved nozzle design that may
enhance atomization performance. An improved nozzle would produce a small wake at the nozzle tip,
thus giving the molten metal less opportunity to avoid immediate contact by high-velocity gas in the near-
field. Further, an improved nozzle would be designed to maintain supersonic gas flow over the longest
distance possible in order to effectively atomize liquid far downstream of the nozzle tip. Efforts to design
such a nozzle are underway. The new nozzle will be evaluated through a set of comparative atomization
experiments with the other nozzles, including comparisons of tin powder size distributions. High-frame-
rate imaging is also planned to further study the unsteady nature of liquid metal atomization by these
nozzles.

ACKNOWLEDGMENTS

This work is supported by the National Science Foundation under grant No. CTS-9221863. The authors
wish to thank J. D. Miller and Lori J. Dodson for their assistance.

REFERENCES

1 A. H. Lefebvre, Atomization and Sprays, Hemisphere Publishing Corporation, New York, 1989.
2 S. P. Mates and G. S. Settles, “A Flow Visualization Study of the Gas Dynamics of Liquid Metal
Atomization Nozzles”, Advances in Powder Metallurgy and Particulate Materials - 1995,
Part 1, Metal Powder Industries Federation, Princeton, NJ, 1995, pp.15-29.
3 S. D. Ridder, P. I. Espina and F. S. Biancaniello, “Optimization of Inert Gas Atomization”,
Physical Chemistry of Powder Metals, Production and Processing, The Minerals, Metals, and
Materials Society, 1989, pp.163-173.
4 I. E. Anderson, H. Morton and R. S. Figliola, “Fluid Flow Effects in Gas Atomization
Processing”, Physical Chemistry of Powder Metals, Production and Processing, The Minerals,
Metals, and Materials Society, 1989, pp.229-249.
5 A. Ünal, “Effect of processing variables on particle size in gas atomization of rapidly
solidified aluminum powders”, Materials Science and Technology, Vol. 3, Dec. 1987, pp.1029-
1039.
6 A. Ünal, “Influence of Gas Flow Performance of 'Confined' Atomization Nozzles”,
Metallurgical Transactions B, Vol. 20B, 1989, pp.833-843.
7 A. J. Yule and J. J. Dunkley, Atomization of Melts for Powder Production and Spray Deposition,
Oxford Series on Advanced Manufacturing, Clarendon Press, Oxford, England, 1994.
8 S. A. Miller, S. Savkar and R. S. Miller, “A Systematic Approach to Designing Close-Coupled
Gas Atomization Nozzles”, Advances in Powder Metallurgy and Particulate Materials - 1994,
Metal Powder Industries Federation, Princeton, NJ, 1994, pp.95-104.
9 S.A. Miller, S. D. Savkar and R. S. Miller, “Global and Midfield Imaging of Liquid Metal
Atomization”, Advances in Powder Metallurgy and Particulate Materials - 1993, Vol. 1, Metal
Powder Industries Federation, Princeton, NJ, 1993, pp.25-34.
10 J. T. Strauss, “The Effect of Metallostatic Head on Close-Coupled Atomization of Fine
Powders”, Advances in Powder Metallurgy and Particulate Materials - 1994, Metal Powder
Industries Federation, Princeton, NJ, 1994, pp.85-94.
11 W. Merzkirch, Flow Visualization, Academic Press, Orlando, 1987.
12 J. L. Herrin and J. C. Dutton, “Supersonic Base Flow Experiments in the Near Wake of a
Cylindrical Afterbody”, AIAA Journal, Vol. 32, No. 1, January 1994, pp.77-83.
13 K. Issac, A. Missoum, J. Drallmeier and A. Johnston, “Atomization Experiments in a Coaxial
Coflowing Mach 1.5 Flow”, AIAA Journal, Vol. 32, No. 8, August 1994, pp.1640-1646.
14 A. Ünal, “Flow Separation and Liquid Rundown in a Gas-Atomization Process”, Metallurgical
Transactions B, Vol. 20B, Oct. 1989, pp.613-622.
15 M. Glogowski, M. Bar-Gill, C. Puissant, T. Kaltz, M. Milicic and M. Micci, “Shear Coaxial
Injector Instability Mechanisms”, Proceedings of the 30th AIAA/ASME/SAE/ASEE Joint
Propulsion Conference, June 27-29, 1994, Indianapolis, IN.
16 S. D. Scroggs and G. S. Settles, “An Experimental Study of Supersonic Microjets”, accepted for
publication in Experiments in Fluids, submitted Aug. 1996.
17 A. Ünal, “Liquid Break-Up in Gas Atomization of Fine Aluminum Powders”, Metallurgical
Transactions B, Vol. 20B, Feb. 1989, pp.61-69.