AQUEOUS CLEANING AND VALIDATION FOR SPACE SHUTTLE PROPULSION

HARDWARE AT THE WHITE SANDS TEST FACILITY

Steven D. Homung and Paul Biesinger -31

AlliedSignal Technical Services Corp. Team
NASA Johnson Space Center White Sands Test Facility
Las Cruces, New Mexico

Mike Kirsch_Harold Beeson, and Kathy Leuders

NASA Johnson Space Center White Sands Test Facility
Las Cruces, New Mexico

INTRODUCTION

The NASA White Sands Test Facility (WSTF) has developed an entirely aqueous final cleaning and
verification process to replace the current chlorofluorocarbon (CFC) 113 based process. This process has been
accepted for final cleaning and cleanliness verification of WSTF ground support equipment. The aqueous process
relies on ultrapure" water at 50 °C (323 K) and ultrasonic agitation for removal of organic compounds and
particulate. The cleanliness is verified by determining the total organic carbon (TOC) content and filtration with
particulate counting. The effectiveness of the aqueous methods for detecting hydrocarbon contamination and
particulate was compared to the accepted CFC 113 sampling procedures. Testing with known contaminants, such as
hydraulic fluid and cutting and lubricating oils, to establish a correlation between aqueous TOC and CFC 113
nonvolatile residue (NVR) was performed. Particulate sampling on cleaned batches of hardware that were randomly
separated and sampled by the two methods was performed.

This paper presents the approach and results, and discusses the issues in establishing the equivalence of
aqueous sampling to CFC 113 sampling, while describing the approach for implementing aqueous techniques on
Space Shuttle Propulsion hardware.

BACKGROUND

The application of aqueous cleanliness verification is based on a Kennedy Space Center developed
procedure, which has been adopted as ASTM method 131-96) and the method for sampling for residual organics is
based on ASTM method G-144-96. 2 In the cleaning verification procedure, parts are sampled in 50 °C (323 K)
ultrapure water using a 25-kHz, 1000-W ultrasonic bath. The organic content of this water is then determined using
a high temperature, catalytic combustion TOC analyzer. The particulate is determined by filtration and counting the
particulate on the filter as a function of size.

Cleaning operations at WSTF encompass support for materials, components, propulsion test systems, and a
large Shuttle depot activity. As part of these operations, components and piece parts are cleaned for a variety of
services including oxygen, hydrogen, hydrazines, and nitrogen tetroxide. The majority of the parts is constructed of
stainless steel and is resistant to corrosion in aqueous media. Cleanliness levels to 50A are required for high
pressure oxygen service. For precision cleaned parts, the standard cleaning process consists of two major elements:
precleaning and final cleaning, and cleanliness verification. The precleaning process is shown in Fig. I. Parts are
first inspected, and any gross contamination is removed. For stainless steel parts, the cleaning consists of an alkaline
cleaner with ultrasonic agitation, rinse, a phosphoric acid cleaner with ultrasonic agitation, rinse, and then drying in a
HEPA-filtered oven. Following drying, the parts are visually inspected, and if passed, are transferred to the Class
100 clean room. The accepted CFC-based final cleaning, shown in Fig. 2, is a CFC 113 rinse, followed by
a final rinse of 100 mL per 0. I m 2(1 if-") of surface. The rinse is then collected for filtering and NVR. The results

" Ultrapure v,ater defined having resistivit) of 12.5 to 18 MOhm.

169
 Cleana"
Alkaline I

WaIst
[Link]

hosphoric
Aci_ Tap W_er
Rinse
C[©an_ J

Niu'ogen
Blow Dt_ Dry
HEPAOven

Fig. 1. Standard Precleaning Process

r CFC-113
Clean
Final
Rinse
i of the NVR and particulate sampling are
expressed in passing levels def'med in the
' Filter for
cleaning specifications, such as JHB 5322. 3
CFC-II3 Rinse I Particle Count
For NVR, the levels are <1, <2, <3, and <4
i
I mg/f_ of sampled surface area and are
V ¥

i Nitrogen
t Dry Blow J
I
!
Evaporate
NVR
for
I designated A, B, C, and D, respectively.
Particulate levels are usually defined by the
largest particle acceptable, 50, 100, 200,
T 300, and 400p. per 0.1 m 2 (1 _) of sampled
i Package and surface area. The proposed aqueous final
I Label
l cleaning and verification process is shown in
Fig. 3. The final cleaning procedure consists
Fig. 2. CFC Final Cleaning and Verification Process of a rinse with filtered hot ultrapure water,
agitation with ultrapure water in a 1000-W
ultrasonic bath, followed by final rinse.
IUltrapure W_ef Cleanliness verification consists of a 1.27-
cm (0.5-in.) tip ultrasonic probe delivering
[ _---
U t,-upu re Wal--er-'ler approximately 600 W of power into the
U lx's_,_ c Bzlh j sampling container for a total of 30 seconds.
Ultml_re waier
A small aliquot of the solution is then
Rnse removed for sampling by TOC, and the
remainder of the water is filtered for the
particle count. The use of the ultrasonic
Y

!tqtrapurew.._[ probe allows a higher ultrasonic energy

density in the sampling container. This
__._F
aqueous process is directed at piece parts
"u,_,-w=_,'L_ _,_ Fil,,rf_ [
Spcay Sl_rlphng] i p_'licle Count i which are not corrosion or ultrasonic
sensitive, and do not have small passages or
r'--}
:
!
Nitrogen
Blow Dry
....
i Oven Dry
--- _ Label
Package and entrapment spaces.

Fig. 3. Aqueous Final Cleaning and Verification Process

170
 APPROACH

Thetestingusedtocertifytheaqueousalternative
toCFC113sampling wasdividedintothreephases:
development,batchpartssampling,
andprocessvalidation.
Inthedevelopment
phase, equipmentandprocedures
weretestedtodeterminethemostpromisingmethod ofaqueouscleanliness
verification.
Developmenttesting
was
alsoperformedtodeterminethattherewasarelationship
between thecontaminant
levelsandthesamplingresponse.
The batch parts sampling was aimed at assessing the efficiency of contaminant removal, both organic and particulate,
compared to CFC 1 ! 3. More extensive batch parts sampling tests were then conducted to compare the efficiency of
removal of contaminants. The batch parts sampling was performed only to establish the efficiency of the cleanliness
verification and did not include the f'mal cleaning step. CFC 113 was used as the standard against which the results
were compared. Process validation was performed to determine the reliability of the process.

ORGANIC CARBON METHODS

In the development testing, numbered stainless-steel and -aluminum plates, 5 X 7.6 cm (2 X 3 in.) were
used. These plates were fabricated according to ASTM-G 121. 4 A representative plate is shown in Fig. 4. A tare
weight on each plate was obtained and either single contaminants or contaminant mixtures in a CFC 113 carrier
contaminated the plate. These contaminants were representative of the types of contaminants encountered in WSTF
cleaning operations. The plates were re-weighed; to determine the contaminant loading, then sampled using the
aqueous process. The results of the TOC analysis were compared with the original contaminant loading to determine
percent recovery for each of the contaminants. These tests were performed over a range of contaminant loading, and
a slope of the fitted data determined the efficiency of recovery.

In batch sampling, representative batches of fittings numbering bet_veen 10 and 30 parts were contaminated
by a mixture of compounds in CFC 113, allowed to dry, then divided into two groups at random. An example of
these fittings is shown in Fig. 5. One group was sampled by the aqueous method and the other using CFC 113. In
both cases, a final cleaning or rinse did not precede the sampling, as is usually the case with CFC 113 cleanliness
verification. The TOC response was compared to the CFC 113 NVR. Results were tabulated in the form of JHB-
5322 passing levels. Because Krytox '_ is often used as a thread lubricant in the assembly of systems, it was treated
separately from the other contaminants.

PARTICULATE COUNTING METHODS

Because of the difficulty of uniformly contaminating plates or a batch of parts at a low enough particulate
level, cleaned batches of parts were used for the particulate studies. The particulate contamination on these parts
was due to residual particles that were not removed during the cleaning process.

The development testing and the batch sampling for the particulate consisted of cleaning batches of fittings
numbering between 10-30 parts and separating them into two groups at random. One group was sampled by the
aqueous method and was followed by filtering to collect the particles. The other group used CFC 113 with the
standard filtering technique. In both cases, a final cleaning or rinse did not precede the sampling, as is usually the
case with CFC 113 cleanliness verification. The total number of particles on the filter was counted according to size.
The passing levels for the aqueous and CFC 113 sample pairs were tabulated using JHB-5322 passing levels.

ORGANIC AND PARTICULATE PROCESS TESTING

Process testing was performed on actual hardware. The aqueous method, which included an aqueous
ultrasonic final cleaning step, was followed by the accepted CFC 113 method. Particulate and NVR passing levels
were recorded for each batch. This sequential sampling was performed to assess the possibility of obtaining a
passing level with the aqueous method followed by a failure with the CFC 113 method. The viability of this
technique in a production environment was also evaluated in terms of throughput and recleaning required due to
excessive failures. A wider variety of parts were sampled in this testing and included disassembled valves and filters
as well as fittings and short lengths of tubing. An example of a disassembled filter assembly is shown in Fig. 6.

171
 Fig.4. Representative
Plates

: ?.+i__ -,

Fig. 5. Example of Fittings

Fig. 6+ Disassembled Filter Element

172
 RESULTS

ORGANIC CARBON METHODS - DEVELOPMENT TESTING

Tare weights were obtained on numbered stainless-steel plates 5 by 7.6 cm (2 by 3 in.). A contaminant
mixture consisting of 0.5 grams of each contaminant, including hydraulic, vacuum pump, lubricating, and cutting
oil were dispersed in 200 mL ofCFC 113. Approximately 0.25 mL of this contaminant mixture was placed on each
plate and allowed to air dry. The mass ofthe plate plus the contaminant was determined. For higher contaminant
levels, this procedure was repeated several times./Before each sampling, the TOC was determined on the water to
serve as a blank or reference.

The coupon was placed in a 600-mL beaker and covered with 400 mL of ultrapure water at 50 °C (323 K).
The coupon was agitated ultrasonically for 1 minute with 0.3-second pulses for a total ultrasonic exposure time of
30 seconds. A gastight syringe was used to remove 250 _L of the water, which was injected into the TOC analyzer.
From the measured concentration, the mass removed by this technique was calculated. The results of this testing is
represented by a graph in Fig. 7. This initial testing showed a good correlation with a slope of approximately 0.65.
This slope indicates there is a linear increase in the TOC response with the contaminant mass.

To assess the recovery of Krytox _ by the aqueous method, approximately 2 mg was deposited on the
coupon directly from the tube and sampled as previously described. No TOC response could be found in this case.
However, Krytox ® was detected in the form of particles when the water was filtered. Further work on the sampling
for Krytox ® was done from this particle count data. An example of Krytox ® particles detected by aqueous ultrasonic
sampling is shown in Fig. 8.

24 • Water Blank
/ ORGANIC BATCH TESTING

20
The batches of parts, primarily fittings,
numbering at least 10-30 pieces were contaminated
22 I --" TOC Response
using the same solution as in the development
testing. These were then dried, and the batch was
split into two groups at random. One group was

,°I
added to a 600-mL beaker, covered with 400 mL of
50 °C (323 K) ultrapure water and ultrasonicated for
60 seconds. 250 I-tLwas then removed for TOC
4 analysis. The second group was sampled using the
standard CFC I 13 rinse with 100 mL from which the
2
nonvolatile residue was obtained in mg. The
o 411_-ili_,lial* o-J-,- )',- .... -I
l

) ) i [ i t ) ) . i ( i L equivalent TOC was calculated from the

O 2 4 6 8 10 12 14 16 18 20 22 24
concentration obtained from the TOC by the
Contaminant Loading. mg following equation:
Fig. 7. Recovery of Organic Contaminants

Equivalent TOC (mg) = C(ppm) * V(L)/S(mg CFC/mg TOC) (1)

where
C the concentration obtained from the TOC
V = the volume of ultrapure water used in liters, and
S = the experimentally determined sensitivity factor

The sensitivity factor was obtained in this case by plotting the milligram response (ppm by volume) of the
TOC versus the results obtained from the NVR to determine the slope of the line.

173
 Fig.8. Example
ofKrytox
° Particles

From the batch testing, a sensitivity factor of 0.4 was determined and applied in all subsequent tests. To
compare the NVR and TOC sampling methods, the results were distributed into the discreet, integer passing levels
used in the cleaning specifications. For example, a result of 1.4 mg would pass a 2 mg or B level; a result of 0.8 mg
would pass a 1 mg or A level. Afigure of merit, AW, which expresses the relative efficiency ofthe two sampling
techniques was also calculated by subtracting the CFC 113 passing level from the aqueous passing level. When AW
is zero, the results of the two sampling techniques are equal. When AW is negative, the CFC 113 showed more
contamination on the parts. When AW is positive, the aqueous method showed more contamination on the parts.

The number of observations of each result was then tabulated to give a distribution of the number of
observations for each value of&W. In Fig. 9, AW is plotted versus number of observations. The mean or average
passing level is 1.2 AW, and is on the side of the distribution where the aqueous technique is more effective. The
mode, or most likely occurring value of AW is zero, where the two techniques yield equal results. The median AW,
the value with equal numbers of observations above and below, is 1, is also on the side of the distribution where the
aqueous technique is more effective. From this set of data, the aqueous ultrasonic sampling is equal to or more
effective than CFC 113 and samples 91 percent of the time in detecting residual contamination.
2s
PARTICULATE BATCH TESTING
C FC- 113 Aqueous Ultra sonic

20 Et_ective Effective The batches of parts (primarily of at least

,_ More ] More
10-30 pieces or fittings) were precleaned and
is _ transferred to the Class 100 clean room where the
batch was split into two groups at random. One
oo group was then added to a 600-mL beaker, covered
lo with 400 mL of 50 °C (323 K) ultrapure water, and
_E ultrasonicated for 60 seconds. All of the 400 mL
Z
Er--! _--q

s _ L [ r'-_ the filter was counted. The second group was

I _, _ was filtered, and the total number of particles on
o I--1 __ _I , _l _ sampled
v,hich wasusing
then a filtered.
rinse of 100
The mL
total ofnumber
CFC 113,
of
-6 -4 -2 0 2 4 s particles on the filter was counted as a function of
AW size. Using the passing levels from the cleanliness
specification, the aqueous and CFC 113 particle
Fig. 9. Distribution of AW for Organic Passin_ _ counts were then assigned the appropriate passing
Levels on Batch Parts level.

174
 Afigureof merit, AP, which expresses the relative efficiency of the two sampling techniques was also
calculated by subtracting the CFC 113 passing level from the aqueous passing level. When AP is 0, the results
of the two sampling techniques are equal. When AP is negative, the CFC 113 detected a higher particulate level
on the pans. When AP is positive, the aqueous method detected a higher particulate level on the parts. The
number of observations of each result was then tabulated to give a distribution of the number of observations for
each value of AP. In Fig. 10, AP is plotted versus number of observations. The average AP is 53 and is on the
side of the distribution where the aqueous technique is more effective. The mode, or most likely occurring value
of AP is zero, where the two techniques yield equal results, and the median AP with equal numbers of
observations above and below is also 0. From the tabulated data, the aqueous sampling for particulate is equal
to or more effective than the CFC 113 and samples 73 percent of the time.

KRYTOX ® BATCH TESTING

12
The studies and experience gained at
WSTF indicate through the CFC 113 cleanliness
10
verification and the aqueous ultrasonic sampling
that Krytox ® is detected in the particulate count
rather than the NVR. To assess the effectiveness of
the aqueous technique in detecting Krytox ® by
i B
particle count, six batches of four threaded fittings 6
were contaminated by a heavy application and
distribution of Krytox ® grease over the threads.
These batches were cleaned using the precleaning Z
process. Following precleaning and inspection,
three batches were sampled aqueously without the
final cleaning step; and CFC 113, without the final
cleaning step, sampled three batches. Out of the
o
three batches, aqueous sampling detected Krytox ®
-B00 -600 -400 -20o 0 200 400 600 S00
in one batch of fittings, and CFC I 13 detected
Krytox ® in two batches of fittings. The CFC 113
sampled batches all passed a level A, or 1.0 mg AP
NVR. To relate this particulate detection of
Krytox _ to a comparable number for NVR, a Fig. I0. Distribution of AP for Particulate Passing
calculation was performed to determine the mass, Levels on Batch Parts
which would result from a particulate passing
level if all the allowed particulate were Krytox r.
As shown in Table 1, a sample passing the 200
level with the maximum allowable number of Table I. Worst Case NVR Ec[uivalent
Krytox ® particles would have less than 0.90 mg of
Krytox ® present. This represents the worst case Specification Passing Mass Equivalent as
Level Krytox ®
NVR equivalent of a Krytox ® particle count.
(mg)
Process Testing
300 1.62
200 0.90
In this phase of testing, batches of 100 0.10
hardware were final cleaned and sampled using
50 0.01
aqueous methods, and then final cleaned and
sampled using CFC 113 methods. To accomplish
this, the glass beakers for the ultrasonic sampling
were replaced with a metal container along with an
internal basket. The metal container and basket
generated particulate during the ultrasonic
exposure.

175
 Theprocess
waschanged tousea 50
rinseof50°C(323K)waterfor the 45
particulate sampling. The sampling for CFC,I 13 Aqueous
more more
organics was unchanged from that used for 40 Effective Effective
the batch testing. The passing levels were
35
compared for both particulate and organic
recovery. Out of a total of 300 trials, the 30

aqueous ultrasonic with TOC detection

25
registered five failures and CFC 113 NVR
gave no failures at the A passing level. For 20

particulate counts, the figure of merit AP,

15 -
was applied to 92 sampling runs. The
results are tabulated in Fig. i I where AP is 10 -

plotted versus number of observations of

5 -
each value of AP. The mode, or most likely
occurring value of AP is 0, where the two 0 ....... 1 "1 ! I 4

-600 -400 -200 0 200 400 600

techniques yield equal results. The average
AP is 32, and the median value of AP with
equal numbers of observations above and AP
below is zero. Based on this testing, this Fig. 11. Distribution of AP for Particulate Process Testing
process has been accepted for final cleaning
and cleanliness verification of WSTF ground
support equipment.

AQUEOUS CLEANING FOR PROPULSION HARDWARE

CFC replacement is a critical issue for Space Shuttle Propulsion hardware. Adaptation of a new process
required a clear technical approach to the selection of alternative verification methods. This technical approach
consisted of a part-by-part analysis with consideration given to configuration, materials, and types of contaminants.

Part configurations were considered critical, especially evaluating any potential entrapment areas where
drying processes may not be effective. Since most of the flight hardware repair work involves total unit disassembly,
down to its piece-part level, the part configurations during cleaning are generally very simple. Therefore, piece-part
configurations were perfect candidates for using the water-verification process. In addition, where there were
potential ultrasonic damage concerns, either testing was performed to verify that there were no harmful effects or the
configuration in question was deemed to be suitable for a solvent verification. Examples of this include filter
assemblies and anodized parts.

The part material was also a consideration. All soft goods were cleanliness verified using solvents.
Although many soft goods could have been cleaned and verified aqueously, many have questionable aqueous
verification applications. The aqueous verification method was also discarded where there were corrosion concerns,
specifically when drying methods were in question. Materials such as 400-series stainless or titanium are specific
examples of materials that will not be verified aqueously.

The final evaluation criteria dealt with the types of contaminants likely to be found on the Shuttle. Most of
the flight hardware requiring cleaning at WSTF is exposed to nitrogen tetroxide or monomethylhydrazine propellants
or their combustion products. All of these contaminants are soluble in water, and therefore the water cleaning and
verification method is effective. The other main types of contaminants are assembly lubricants, such as fluorocarbon
grease. The WSTF data indicate that aqueous verification methods are effective at verifying the presence of this
class of contaminants.

176
 CONCLUSIONS

These resultsapplyto piece parts which were primarily stainless steel and resistant to corrosion in aqueous
media. Contaminants used in this study were representative of those encountered at WSTF. The following
conclusions are made:

Aqueous ultrasonic cleanliness verification does not increase the risk of passing a contaminated part over
the current CFC 113 cleanliness verification technique.

The use of the aqueous ultrasonic sampling with TOC detection as a substitute for the CFC 113 NVR is
effective and was found to be more conservative than CFC 113.

The use of aqueous ultrasonic sampling followed by rinse sampling for particulate is effective and was
shown to be equivalent to CFC 113 particulate sampling.

Detection of Krytox ® is by particulate count and is conservative for particulate levels below level 200.

Aqueous methods are applicable to propulsion type hardware, although, rigorous engineering evaluations
including configuration, material, and contaminant reviews, must be thoroughly documented especially
when changing certified processes.

REFERENCES

1. ASTM G-131-96. Practice for Cleaning Materials and Components by Ultrasonic Techniques. ASTM Vol.
14.02, pp. 1431-1435.

. ASTM G-I44-96. Standard Method for Determination of Residual Contamination of Materials and
Components by Total Carbon Analysis Using a High Temp. Combustion Analyzer. ASTM Vol. 14.02, pp.
1454-1457.

3. NASA. Contamination Control Requirements Manual. JHB 5322, Revision C, NASA Johnson Space Center,
Houston, TX.

4. ASTM G-121-93. Standard Practice for Preparation of Contaminated Test Coupon for the Evaluation of
Cleaning Agents. ASTM Vol. 14.02, pp. 1381-1384.

177