Urea and Ammonia Removal from De-Ionized Water
via Steam Purification
By Jeffrey J. Spiegelman, President, RASIRC and Russell J. Holmes, Test Lab Manager,
RASIRC
11760 Sorrento Valley Road, San Diego, CA 92121 • 858-259-1220 • [email protected]
Abstract
De-Ionized (DI) water is ubiquitous within the Fab. Purity and consistency of DI is critical to
manufacturing process. Urea is a known contaminant that is difficult to control and has large
seasonal fluctuations. Normally stable at room temperature, Urea has a high conversion rate to
Ammonia when boiled. DI water is often converted to steam and then injected into cleanroom
air to control humidity. We will present data that shows that Urea converts to Ammonia during
boiling. Ammonia is a known airborne molecular contaminant and directly leads to photo resist
T-topping and wafer image defects. We will also present data that shows the ability to remove
both Urea and Ammonia from the steam. Urea was reduced from 2200 ppb to 2.6 ppb and
Ammonia was removed from 1468 ppt to 116 ppt.
Introduction
Urea and Ammonia are serious contaminants in the fabrication of semiconductors. As device size
and geometry shrink, the susceptibility to damage associated with airborne contaminants grows.
Airborne Molecular Contaminants (AMC) are introduced through the humidification of the
cleanroom air and through the wafer cleaning and manufacturing processes. Reducing AMC to
ppt levels is critical to achieving high yield and quality.
Urea is a very small, stable amine. The individual atoms that make up a urea molecule come
from carbon dioxide, water, aspartate and ammonia. Urea is produced commercially from
synthetic ammonia and carbon dioxide. More than 90% of world production is destined for use
as a fertilizer. It has the following characteristics:
Molecular Weight: 60.06
Specific Gravity: 1.32 @ 20C/4C
pH: 7.2 (10% in water)
% Volatiles by volume @ 21°C: 0
Boiling Point: Decomposes
Melting Point: 132-135°C
Source: http://en.wikipedia.org/wiki/Urea
Contamination in Humidification
The concentration of Urea in DI water, used in the humidification process, varies seasonally and
with sources of the city water supply. Fertilizers, auto emissions, and human and animal sources
contribute to the presence of Urea and Ammonia. When DI water containing Urea is boiled, the
Urea converts to Ammonia, contaminating the cleanroom air and manufacturing processes.
Urea is not easily rejected by reverse osmosis membranes. It is non-polar, so not removed by deionized water processes, and chemically stable so not easily destroyed by UV sterilization
processes. Urea is also not typically continuously monitored in UHP water.
Humidity in cleanroom air is tightly controlled to ensure consistent processing and control static
electricity. Winter air and dry climates frequently require adding water vapor to raise water
vapor content. Commonly UHP DI water is converted into steam via commercial steam
generators, then injected into cleanroom air as part of the makeup air system.
When Urea present in DI water is injected in the steam generator it becomes a two-fold
contaminant. Urea has limited volatility converting from a solid to decomposition at 132.7°C. Its
solubility increases with water temperature increasing seven-fold from 20°C to 100°C.
Urea does not readily precipitate in the steam generator. As the DI water continues to boil off,
the Urea continues to increase. Since the decomposition temperature (132°C) is slightly above
typical operating temperature for a steam generator (125°C), a large concentration of Urea can
build up the steam source water.
As the Urea concentration increases Urea conversion into Ammonia increases. In addition, the
amount of Urea transported in micro droplets increases and is transported by the steam.
So steam generator creates two-fold problem: entrained Urea and combusted Urea which is seen
as Ammonia. The steam is injected into the makeup air and then carried into the cleanroom
system.
Chemical filter systems as part of the cleanroom air handling system will remove the Ammonia.
These filters are capacity limited and the higher the Ammonia challenge, the shorter the useful
life. Without endpoint detection, their efficiency gradually declines leading to reduced yield or
increased variability in process. While chemical air filters can remove Ammonia they are not
particularly efficient for scrubbing Urea. Urea being stable and small is free to travel and land on
any surface including silicon. The Urea will lay dormant until exposed to temperatures exceeding
133°C when it will decompose into its carbon and nitrogen constituents.
Deleterious Effects during Manufacturing
Various processes are negatively affected by the presence of Urea and Ammonia. The
International Technology Roadmap for Semiconductors (ITRS), defines a maximum Ammonia
concentration of <0.75 ppb to guarantee default-free production using deep UV (DUV)
technologies. In a sub100µM environment, AMC now pose the same or greater threat as particle
contaminants.
In lithography, controlling Ammonia levels makes the difference between conformity and
uncontrolled variation. ‘T-Topping’ is a real danger, resulting from chemically amplified resists.
Other structural defects include incorrectly imprinted line width and short circuits. Ammonia can
also deposit on optical surfaces, causing equipment downtime.
Atomic Layer Deposition (ALD) depends on water vapor for High-K film formation. The
technique requires the proper molecule be available and not replaced by competitive species that
will disrupt the lattice structure. ALD is increasingly popular for creating thin films for gate
dielectrics, capacitor dielectrics and diffusion barriers. Contamination is a high risk in this
process because ALD is slower and performed at lower temperatures.
Wafer cleaning eliminates contaminants and is an important step in photolithography and other
depositional techniques. AMC deposits can generate localized micro environments where
standard cleaning processes have reduced effectivity. The use of ultra high purity steam with low
ppb levels of Urea and Ammonia keeps the process from introducing new contaminants.
While the above processes are very different, they all suffer in the presence of Urea and
Ammonia. With a reliable source of Urea/Ammonia free steam, AMC from the humidification
process can be eliminated and chemical filter challenges reduced, leading to reduced cost and
better environmental wafer control.
New Purification Material
Up to now no steam purification technology has been commercially available. RASIRC has
developed a new hydrophilic membrane. The membrane is nonporous and selective for water
vapor. Figure 1 shows selectivity of up to a million to one water molecules over nitrogen. In
addition, the glass transition temperature is above 180˚C, well above the boiling point of water.
The membrane has a high flow rate capable of supplying water vapor needed in a fab-wide
humidification system, which exceeds 2600 lbs/hr for state-of-the-art fabrication.
PERMEABILITY (BARRERS)
10,000,000
1,000,000
100,000
10,000
1,000
100
10
1
0.1
H 2O
H 2S
CO 2
(cm 3 -cm )/(cm 2 -cm H g)-10
C 2H 4
CO
H2
O2
Ref Ion Power
Figure 1. Permeability of Membranes
N2
Testing the New Material
Previous test results showed a high
efficiency for removing metals but the
membrane had not been tested for the
efficiency of removing Urea or Ammonia.
Refer to the 2006 SPWCC paper entitled
“Alternative Method and Device to Purify
and Deliver Water Vapor” for more detail
on this testing.
To determine the viability of this new
purification material, an experiment was
performed to determine if Urea will
transfer across the hydrophilic membrane.
Figure 2. The RASIRC Intaeger UHP
A steam generator system was developed to provide the controlled delivery of pure steam to the
purifier. The source water, pre- and post-purified steam were tested for Urea concentrations.
Furthermore, Ammonia concentrations were measured to determine if Urea converts into
Ammonium and the amount of Ammonium that would transfer from the boiler and through the
membrane. The source water was doped with Urea to generate a significant challenge that could
be easily monitored. This challenge far exceeded what would normally be seen within a Fab.
Manifold Setup
Initially, one and a half liters of water was mixed with 3 milligrams (mg) of Urea to produce a
2.2 parts per million Urea solution. 50 milliliters (mL) of this solution was poured into a sample
bottle. The bottle was capped and rotated to coat the sides of the bottle with the solution. The
bottle was emptied, refilled with 175 mL, capped, labeled, and bagged. The remainder of the
solution was poured into a quartz boiler. Figure 3 is a schematic of the manifold used for this
experiment.
Figure 3. Water Sample Collection Manifold
The steam was generated in a quartz boiler using infrared heat lamps. Pressure of the steam was
controlled through a Koyo’s Direct Logic 06 Program Logic Controller (PLC), which monitored
head pressure with a thin film pressure transducer with a full scale range of 30 psia. The steam
from the boiler was fed into the Intaeger Steam Purifier Assembly (SPA). The SPA consisted of
a series of vertical, thin wall, tubular membranes within a 0.5 inch outer diameter Teflon®
housing. The SPA works as a cross flow filtration device. The membrane is nonporous, but has a
high specific transfer rate for water molecules. Of the steam that enters the SPA, approximately
20% exits the outlet of the SPA and is referred to as condensate or pre-purified steam. The
remaining 80% permeates across the membrane. This purified steam is referred to as permeate.
The permeate and condensate from the SPA were sent to separate quartz condensers. The water
from the two condensers was allowed to collect into sample bottles downstream. The bypass
ports would be left open for the duration of the test to prevent pressure from building within the
test manifold.
Sample Collection
Before collecting the condensate and permeate samples, the test manifold had to be conditioned.
Both sample ports were opened and both bypass ports were closed. UHP steam was generated
and run through the test SPA. 50 mL of liquid water was allowed to pour through both sample
ports. After reopening both bypass ports, sample bottles were attached to both sample ports with
sampling tubes that were cut and installed with flare fittings. The bottles were conditioned by
filling them with 50 mL of water, removing them from the manifold, capping them, rotating
them to coat the sides of the bottles with collected water, and emptying the bottles. The bottles
were then reattached to the manifold. After collecting over 175 mL of water, both sample ports
were closed and the sample bottles were removed, capped, labeled, and bagged.
After collecting the condensate and permeate sample, a sample of the solution in the boiler was
collected. Initially, 50 mL was poured into the sample bottle. The bottle was capped and rotated
to coat the sides of the bottle with the solution. The bottle was emptied, refilled with 175 mL of
the solution from the boiler, capped, labeled, and bagged. The four samples were sent for Urea
and Ammonium analysis.
Results and Discussion
Table 1 is a summary of the results. As shown, the overall reduction from initial solution to
permeate was 99.88% for Urea and 92.1% for Ammonium. The reduction from condensate to
permeate was 94.58% for Urea and 89.62% for Ammonium. The relatively high contamination
levels shown in the final solution are because the contaminants were concentrated in the solution
due to the evaporation of the water. After subtracting the amount of solution removed for the
initial sample, 1.275 L of solution was placed in the boiler. With the given concentration, the
Urea and Ammonium amounts initially in the boiler were 2.81 milligrams (mg) and 1.87 µg
respectively. After collecting the condensate and permeate samples, the amount of solution left
in the boiler was approximately 225 mL. Using the final solution sample results, the Urea and
Ammonium amounts left in the boiler were 2.25 mg and 0.489 mg respectively. The increase in
Ammonium indicates that the Urea will convert to Ammonium inside the boiler.
Lower Detection Limit
Initial Solution Sample
Condensate Sample
Permeate Sample
Final Solution Sample
Urea(ppb) Ammonium(ppt)
1
20
2200
1468
48
1117
2.6
116
10000
217247
Table 1. Results from Urea and Ammonium Analysis
The low volatility of Urea can be seen by the five-fold increase in Urea in the water that
remained in the boiler. The ability of Urea to convert to Ammonia with heat is seen by the
increase in Ammonia from 1.5 ppb to 217.2 ppb—an increase of almost 150 times. The longer
the water boils the higher the concentration in the boiler for both Urea and Ammonia. As these
numbers increase, the contaminant load in the steam will continue to increase.
While Urea is not highly volatile, the condensate samples collected illustrate the ability for Urea
to be entrained with the steam and deposit downstream of the steam generator. The use of the
steam purifier reduced the transfer rate by 95%. In addition, the conversion rate of Urea to
Ammonia is high—over 10%. Once converted it is highly volatile and is easily transported with
the steam. The use of a steam purifier reduced the amount of Ammonia by 90%.
Urea in ultrapure DI water is a variable contaminant that is not easily measured and typically not
continuously monitored. The low volatility of Urea leads to a continuous build up in the source
water in the steam generator. As this concentration increases, the net contaminant loading of
Ammonia and Urea in the steam used to humidify clean room air increases. This can add
significantly to the AMC loading in the fab cleanroom. By using steam purifiers before injection
into the cleanroom make up air, 95% of the Urea can be reduced as well as 90% of the
Ammonia. The result would be reduced AMC challenge of silicon manufacturing process and
significantly longer chemical filter life.
Summary
The test results indicate that a steam purifier can be used to reduce high levels of Urea and
Ammonia contaminants that may concentrate in de-ionized water. Further, the test shows that
Urea does covert to Ammonia when boiled. As a result, Ammonia will be introduced as an AMC
unless it is captured in the steam phase. The manifold in the test effectively captured a high
percentage of this contaminant.
Bios
Jeffrey Spiegelman has a BS in bioengineering and MS in Applied Mechanics from University
of California at San Diego. He has over 50 international patents and publications. Previously,
he was founder and president of Aeronex until it was purchased by Entegris in 2003. In 2005, he
founded RASIRC to address process purity and delivery issues around next generation
chemistries, with an initial focus on water vapor.
Russell J. Holmes has a BS in chemical engineering from the University of California at San
Diego. Previously, he was employed as an applications engineer at Aeronex/Mykrolis/Entegris
for more than 5 years. He is the author of several patents and publications concerning
purification for the semiconductor industry.
References
“The International Technology Roadmap for Semiconductors”, 2003 Edition.
Hansen, Jeff (Texas Instruments). “Organic Anions in Multiple UPW Polish Loop Systems”.
Presented at SPWCC 2006.
Spiegelman, Jeff (RASIRC). “Alternative Method and Device to Purify and Deliver Water
Vapor”. Presented at SPWCC 2006.