From: Optimizing Particle Removal in Ultrasonic=20
Baths Introduction Part cleanliness becomes increasingly =
important as=20
mechanical, optical, and electronic parts shrink in size and formerly=20
inconsequential particulates grow in comparison. To remove even sub-micron particles, =
semiconductor,=20
micro-electromechanical systems (MEMS), optical and hard disk drive =
(HDD)=20
components, micro inertial sensors, laser gyroscopes, compact discs =
(CD=92s) and=20
optical coating applications all employ liquid ultrasonic cleaning baths =
for=20
both intermediate and final cleaning steps. =
In contrast, traditional industrial =
cleaning=20
applications often use liquid sprays to remove machining debris, oils, =
and large=20
particulates from engine blocks, transmission parts, and other cast and =
machined=20
parts of many types. However, intermediate industrial applications; e.g. =
natural=20
gas solenoid valves, injector nozzles, compressors, air bag sensors, and =
many=20
others are now using ultrasonic cleaning baths as well. In these items,=20
particles may not be yield limiting as they are in the high technology =
arena,=20
but they can be just as important to long-term function, and can even be =
potentially catastrophic. Understanding, and optimizing, ultrasonic =
cleaning=20
bath parameters can result in major improvements in part cleanliness, =
cleaning=20
efficiency, and ultimately part reliability and lifetime.=20
Ultrasonic=20
Cleaning Parameters Ultrasonic baths formerly employed =
chlorofluorocarbon=20
(CFC) cleaning fluids and other volatile organic compounds (VOC=92s). =
However, the=20
Montreal Protocol has restricted the use of these fluids to avoid =
potential=20
ozone depletion, consequently today=92s ultrasonic baths use =
predominantly=20
water-based solutions. Thus, the hot DI water ultrasonic bath has become =
one of=20
the most popular cleaning processes in use today, for both high =
technology and industrial =
applications. An excellent CleanTech 1999 paper, entitled =
=93Advanced Ultrasonic Surface =
Cleaning=94,=20
by Ahmed Busnaina[1], presented =
the=20
variations in surface cleaning efficiency of sub-micron sized particles =
on=20
semiconductor wafers using the aqueous ultrasonic bath. There are many parameters affecting the =
cleaning=20
ability of an aqueous ultrasonic bath, including; contamination type, =
quantity,=20
and application manner; water temperature, auxiliary cleaning agent and=20
concentration; ultrasonic power, frequency, and sweep rate; filter =
particle=20
size, recirculation rate, and bath flow dynamics.=20
Mr. Busnaina=92s paper examines particle =
removal=20
efficiency by an aqueous ultrasonic cleaning bath, versus temperature,=20
ultrasonic power, ultrasonic frequency and cleaning duration. The =
contamination=20
particles used in those tests were typically single sized, sub-micron, =
plastic=20
calibration particles; and the particle counts were performed on the =
cleaned=20
part, after completion of =
the aqueous=20
cleaning and subsequent drying. In contrast, this paper is focussed on =
the=20
industrial, rather than the high technology, realm of industry. Natural =
silicate=20
particles, spanning a large size range, were used as test contaminants =
in place=20
of single sized plastic spheres; and real time particle counts were =
measured in the wash bath, not on the =
dried=20
part! Traditionally, particle counters do not =
fare well in=20
the wash bath, due to the =
cleaning=20
agents and their concomitant bubbles, which appear as false particles. =
However,=20
by taking special precautions, the graphs in this paper show the actual =
particle=20
removal process, taking place directly in the wash bath! One graph =
includes the=20
characteristic signature of bubble interference in comparison. Part=20
Cleanliness Tools Many techniques are traditionally used to =
determine=20
part cleanliness; water break test, contact angle, UV photoelectric =
emission,=20
optically stimulated electron emission (OSEE), scanning electron =
microscopy=20
(SEM), secondary ion mass spectroscopy (SIMS), Auger electron =
spectroscopy=20
(AES), electron spectroscopy for chemical analysis (ESCA), evaporative =
rate=20
analysis (MESERAN), nonvolatile residue analysis (NVR), surface scatter =
particle=20
counting, thermogravimetric analysis, total organic carbon analysis =
(TOC), phase=20
imaging, gas chromatrography (GC), mass spectrophotometry (MS), and =
Fourier=20
transform infra-red spectroscopy (FTIR). Most of these tests measure chemical =
contamination,=20
while the few that do measure particulates cannot be used in a liquid=20
medium. Particle Counting Advantages and =
Concerns The application of optical single =
particle counters=20
to operating ultrasonic baths provides a convenient means of monitoring =
the=20
cleaning process by particle size and in real time. However, several =
questions=20
have been raised regarding particle counting=92s efficacy and the =
application of=20
its possibly unique advantages. Does the=20
particle counter function properly while sampling from an operating =
ultrasonic=20
bath? Does the ultrasound energy disrupt its particle counting =
ability? Can an optical=20
particle counter yield stable data when used in a detergent bath? Can particle=20
removal efficiency be determined using a particle counter? Can an=20
=93optimal=94 process be established to clean all parts, even if =
confined to a=20
single part type and contaminant? Finally, if a=20
single =93optimal=94 cleaning pass is achievable, how can it be =
maintained? These questions were investigated in a =
series of=20
tests using silicate contaminants, typical of many environmental and =
industrial=20
soils, and utilizing an on-line particle counter sampling from within an =
operating ultrasonic bath. Test=20
Configuration A simple recirculating bath system was constructed using a 150 =
ml glass=20
beaker, 0.3 micron absolute filter, a small heated ultrasonic bath, and =
a=20
battery operated portable particle counter with its own internal =
pump. The beaker contained 100 ml of aqueous cleaning fluid, =
typically 93-100%=20
DI water, which was filtered and recirculated at a high 100% per minute =
rate,=20
i.e. one bath refresh each minute. This recirculation speed provided a =
rapid=20
particle capture rate for better enumeration of total particle load. The =
system=20
is shown in figure 1, below. Figure 1 =96 Particle Counter, Filter, and Ultrasonic Bath Test =
Setup Test =
Procedure The general test procedure began by =
verifying the=20
initial beaker and system clean up to an acceptable background particle =
level=20
for particles larger than 2 microns. Particle counts less than 10 =
particles/ml=20
were considered acceptable. Counts of 5-6 particles per ml were readily =
achieved=20
and, with sufficient patience, even levels of 0-1 p/ml were observed. =
Refer to=20
Graphs 1 & 2 for =
examples of=20
beaker clean up. Please note, =
all graphs=20
show cumulative particle counts, for the sizes noted, on a per =
milliliter=20
basis. A cleaned test item was then suspended in the bath and the =
minute by=20
minute particle count of the wash water was recorded. Once the =
=93cleaned=94 part=20
achieved a low particle value, often arbitrarily set to 10-20 p/ml, then =
a=20
second contaminated test item replaced the original and the =93dirty=94 =
part=20
particle response was recorded. =20
The cleaning agent type and/or its concentration was varied and =
the test=20
repeated on identically contaminated items. Test=20
Contaminant The parts were contaminated using a water solution of known =
concentration=20
of AC Fine Test Dust (ACFTD), which was then baked, or air-dried, onto =
the test=20
item. ACFTD is an ASTM (American Society for Testing and Materials) =
recognized=20
material, composed of a known distribution of mostly silicate =
particulates.=20
This now obsolete material is currently being replaced with a =
series of=20
ISO certified, and markedly more expensive, test dusts of similar =
composition,=20
but with better documented particle distributions. Test =
Parameters Tests were performed for variations in item material, shape,=20
contamination level, cleaning agent type and concentration. All tests =
were=20
performed at a nominal bath temperature of 50 =B0C, =
filtered recirculation rate of 100 ml/min (1 bath change per minute), =
and=20
continuous ultrasonic agitation (45 W average/135 W peak power at 38.5+2 =
kHz=20
swept), unless otherwise stated. Test =
Items A number of common test items were also employed to present a =
variety of=20
materials, surface finishes, and shapes. These items proved to be =
amazingly=20
varied in their ability to harbor and dislodge particulates. Results Aluminum Sheet Metal Originally, simple strips of =
2024-aluminum alloy,=20
1.5=94wide x 2.4=94 long x 1/16=94 were tested in a heated DI water =
ultrasonic bath.=20
These test coupons were cleaned and tested to confirm their ability to =
achieve=20
low background particle count levels. The particle response of a typical =
cleaned=20
coupon is shown in Graph 3. This graph shows =
the pulsed=20
ultrasonic particle response initially of the beaker itself, and then =
the=20
Aluminum test coupon. Ultrasonic energy was experimentally applied every =
10=20
minutes in this test, but was applied continuously for all other tests =
shown in=20
the remaining graphs. The beaker quickly achieved very low =
background=20
counts, but despite leaving the coupon in the bath for over an hour, the =
particle counts never returned to the low baseline values of the initial =
bath.=20
Examination of the cleaned and dried coupons revealed a copper-oxide =
colored=20
stain. Since this alloy of aluminum contains about 5% copper, we suspect =
the=20
copper was being exposed by the effects of the heated DI water. We =
observed the=20
same effect on commercial aluminum foil, which was exposed only to the =
vapor=20
from the DI bath. After this failure to achieve a low =
particle count=20
baseline using an aluminum coupon, stainless steel disks were tested to =
avoid=20
generation of particles by possible chemical corrosion.=20
Stainless Steel Stainless steel rod, grade #303, =
measuring 1 =BD=94 in=20
diameter was cut into approximate =BC=94 thick pieces, then machined =
flat, and=20
washed prior to particle count testing. No staining of any kind was =
observed=20
after the hot DI ultrasonic cleaning. However, high particle counts were =
again=20
observed for the duration of the bath cleaning, with no indication that =
the=20
parts would ever clean up. Observe Graph 4 for the particle =
cleanup=20
rate of an uncontaminated Stainless Steel coupon. It too appears to =
reach a=20
constant particle level until the test is terminated.=20
From the failures of both these metals to =
clean up=20
satisfactorily, we postulated the cause might be due to the surface =
texture of=20
the metal. Both metal coupons displayed the typical porous surface =
finish of=20
unpolished metal. To test =
this=20
theory, we tested glass microscope slides, which had mostly smooth =
glassy=20
portions, but also a rough etched section. =
Glass Slides Glass microscope slides, measuring 1=94 x =
3=94, with a =BE=94=20
frosted end on one side, were tested next and demonstrated excellent =
clean up,=20
provided the frosted end was not immersed. See Graph 5 for these test =
results. Based on their superior particle cleanup =
ability,=20
smooth glass slides were then used for subsequent contamination testing. =
However, the particle removal between slides of different contamination =
levels=20
proved inconsistent. Refer to Graph 6, which shows less particle =
removal from=20
contaminated slides than for =93clean=94 slides in a detergent =
(Citranox)=20
solution. Visual inspection of the inconsistent =
slides revealed=20
residual stains where the particle solution had been baked onto the =
surface.=20
Apparently, variations in how the contaminant dried onto the surface had =
produced differing adhesion strengths and thus removal rates. The bath particle concentration was tested by directly =
injecting=20
100 to 500 ml=20
of ACFTD particle solution, i.e. 50-250 mg=20
of actual dust directly into the bath. Within the range of experimental =
error,=20
this produced the expected linear results, see Graph 7.=20
Two sets of 5 slides each were similarly=20
contaminated. One set was rapidly dried at 200F and then run in the =
bath, while=20
the other set was dried overnight at room temperature before testing. As =
expected, the quickly dried contamination dislodged fewer particles than =
the=20
slowly dried slides. Graph 8 shows markedly =
fewer=20
particles are liberated from the baked on slides as compared to the free =
bath=20
injection (Graph 7). Also, generally more =
particles were=20
released from the air-dried slides of equal contamination (not shown), =
but their=20
release appeared more erratic and drawn-out. Cleaning Agents Remove Residual =
Baked On=20
Contamination Uniformly contaminated glass test slides =
were used as=20
test coupons to evaluate the effects of several commercial cleaning aids =
when=20
added to the DI water of the ultrasonic bath. =
The cleaning agent manufacturer =
recommends to first=20
try a low concentration solution, followed by an anticipated =
over-concentration,=20
then an intermediate concentration that might prove nominal. However, we =
chose=20
to steadily increase, or decrease, the concentration of the cleaning =
agent in=20
our experimental application for the sake of test convenience.=20
Cleaning agents are formulated for =
different=20
contamination conditions, e.g. acid or base solution and plastic, metal, =
or=20
glass parts. Selecting the correct cleaning agent can be part science =
and part=20
art. Manufacturers often assist in selecting the correct product for =
each=20
particular application. Citranox[2] Citranox cleaning=20
agent was tested for its particle removing ability at different =
concentrations=20
using ACFTD contaminated glass microscope slides. Graph 9=20
shows the particle removal =
results=20
of 3%, 5%, and 7% solutions of Citranox in DI water upon three =
identically=20
contaminated glass slide test coupons. Other results for concentrations =
of 0.5 %=20
and 1.75% (not shown) reveal a nominal concentration of about 3%; see =
table=20
below. %=20
Citranox 0.5% 1.75% 3% 5% 7% Particle Count=20
Peak 53 97 165 120 95 Detergent 8 After the fashion of the Citranox testing =
above,=20
Detergent 8 was similarly tested at 1% and 3% concentrations and those =
results=20
appear in Graph 10. Both =
concentrations yield=20
essentially identical particle removal peaks of about 100 p/ml. However, =
so much=20
difficulty with false mixing counts was encountered that no additional=20
concentrations were tested. The low initial particle removal peaks, =
trouble in=20
achieving low background counts, rapid recovery of slide counts to base =
line=20
values, and highly visible residue on the slides after being washed and =
dried,=20
portend against using Detergent 8 for this application.=20
Triton=20
X-100 The Triton X-100 results generally follow the above Citranox =
example, as=20
shown in Graph 11. The following =
table lists=20
the peak particle count removal values versus Triton X-100 =
concentration. %=20
Citranox 1% 3% 5% 7% Particle Count=20
Peak 99 125 73 101 Again, the 3% concentration appears to be =
optimal for=20
removing particles. The chart appears to show a possible inflection =
point for 7%=20
concentration, but Graph 11 reveals three =
distinct=20
peaks for the 3% concentration, whereas there is only one such peak for =
the 7%=20
solution. Thus, many more particles were removed at 3% concentration, =
though=20
they did not occur all in the first peak. Note =96 Triton X-100 left an oily =
residue on the=20
slides and required great amounts of rinsing with hot water to clean it=20
thoroughly from the particle counter and tubing/beaker system.=20
Summary Surface=20
Texture Simple smooth glass =
microscope=20
slides provided a convenient vehicle for testing basic contamination =
removal=20
techniques and verifying theory. However, real-world parts do not =
typically=20
resemble smooth glass and are frequently complex in structure and =
material=20
finish. These factors can greatly complicate optimal cleaning =
determination, as=20
was observed with the simple aluminum and stainless coupon results shown =
earlier. Particle Counts During Sonication The particle count shedding behavior =
(particulate=20
cleaning) of both test coupons and common machined and molded pieces =
(not shown=20
here) were successfully observed in=20
situ, within the bath, during ultrasonic cleaning.=20
Measuring particle counts directly in =
ultrasonic=20
baths is possible, even a bath containing a sudsy cleaner, as evidenced =
by the=20
near zero particle counts in a clean sonicating bath and the proper =
measurement=20
of varying amounts of known contaminants Clean =
Up Time Particle=20
clean up rate varied according to the manner in which the part was =
contaminated.=20
Low temperature evaporation of the contamination conveying fluid from =
the part=20
resulted in lower particle surface adhesion, while high temperature =
evaporation=20
produced stubborn bonding of particulates to the test part. Different materials, of different shapes =
and surface=20
finishes, may require a surprisingly varied range of cleaning conditions =
to=20
achieve acceptable cleanliness. =20
Under some conditions, an ultrasonic bath cleaning process can =
=93stall=20
out=94, providing no further removal of the remaining particulates =
despite=20
indefinite extensions of cleaning time. Universal vs. Optimal Cleaning =
Process It may be possible, but probably not =
economical, to=20
establish a single universal =
process=20
to clean each and every item completely, every time. However, it may =
require=20
such a large investment in equipment, time, and cleaning agents, that it =
would=20
be prohibitively expensive and would result in inefficient and wasteful=20
over-cleaning of most items. However, an =93optimal=94 cleaning =
=93recipe=94 may be able=20
to be formulated for each part =
type=20
by running several parts through an exhaustive matrix of conditions and =
measure=20
their resultant cleanliness by traditional methods. Again, this challenging task =
would need=20
to be repeated for each part type. Alternative Optimal Cleaning An alternate technique may be to monitor =
particulate=20
removal in real time, for each part type, thus =93optimizing=94 particle =
removal for=20
each batch of components being cleaned, regardless of part size, shape, =
and=20
contamination level. Each =93recipe=94 would accommodate a =
specific part type,=20
material, shape, and finish, while the degree of contamination would be=20
monitored on-line to determine the nominal cleaning duration.=20
By whatever means an =93optimized=94 =
process is=20
determined, it may need subsequent and frequent verification and/or =
alteration,=20
especially if the part contamination loading varies, or even if the =
surface=20
texture or finish of the part is altered. =
Conclusions The in-situ particle counting of =
an=20
ultrasonic bath, during the =
cleaning=20
process =96 even with cleaning agents - can provide a direct means of =
measuring=20
when parts have been cleaned of particulates to the desired level. Rote =
reliance=20
on previously =93optimized=94 process conditions need not be blindly =
followed, but=20
can be periodically (or even continuously) verified.=20
However, on-line particle counting =
results should be=20
correlated with another standardized cleanliness test on the finished =
part, to=20
ensure the cleaning bath did not =93stall out=94 by reaching its maximum =
cleaning=20
ability before all the particles were removed. =
If varying cleaning results are =
encountered, despite=20
following an =93optimized=94 process, on-line particle monitoring can =
provide a tool=20
to reveal the cause, and then be used to establish and maintain a truly=20
=93optimized=94 cleaning process. Continuous in-process particle monitoring =
may=20
alleviate the rigorous cleaning conditions and shorten the duration =
developed=20
for =93worst case=94 contamination levels, which are then imposed on =
each and every=20
part. Thus on-line particle monitoring can be employed to conserve =
cleaning=20
resources, while increasing cleaning throughput, without jeopardizing =
part=20
cleanliness.=20
Graph 4: =
Stainless Steel=20
-=20
Cleanup Plateaus =96 DI Water =20
(particles/ml)
(particles/ml)
Graph 1: Beaker=20
Cleanup =96 in DI Water
Graph 2: Beaker Cleanup =96 =
Without &=20
With Ultrasonics -=20
in DI Water
Graph 3: Aluminum -=20
Cycled Sonication =96 in DI Water=20
Graph 5: Clean=20
Glass Slides =96 Smooth vs. Frosted -=20
DI Water
[1]=20
Advanced Ultrasonic Surface =
Cleaning,=20
Ahmed Busnaina, Ph.D. , Clean =
Tech 99=20
Conference, Microcontamination Research Laboratory, Clarkson =
University
[2] Citranox and Detergent 8 are registered =
trademarks of=20
Alcon Company, Newark, New=20
Jersey
Triton X-100=20
was=20
formerly a trademark of the DuPont Company, but is now widely=20
produced.