EPA-650/2-73-032
November 1973
Environmental Protection Technology Series
A fllTIR
I
55
\
UJ
O
:|:|?j|^i$^iji^iii:i:i:;li:j:i:i:::
-------
EPA-650/2-73-032
DEVELOPMENT OF A HIGH-PURITY FILTER
FOR HIGH TEMPERATURE
PARTICULATE SAMPLING
AND ANALYSIS
by
A. L. Benson, P. L. Levins,
A. A. Massucco, and J. R. Valentine
Arthur D. Little, Inc.
Acorn Park
Cambridge, Massachusetts 02140
Contract No. 68-02-0585
Program Element No. 1AA010
Project Officer: Dr. Kenneth T. Knapp
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
November 1973
-------
FOREWORD
The work described In this report was monitored for the
Environmental Protection Agency by Mr. John Davis and Dr. Kenneth Knapp.
The authors are indebted to all program participants for their
cooperation and effort, but especially to Dr. D. W. Lee, Dr. E. T.
Peters, and Mr. W. J. Smith (consultant) for their advice and
inspiration.
Emission spectrographic analyses were performed by Kent
Laboratories of Waltham, Massachusetts. X-ray fluorescence analyses
were performed by Columbia Scientific Industries, Inc., of Austin,
Texas, and the sponsor. Neutron activation analyses were performed
by the sponsor.
This report has been reviewed by the Environmental Protection
Agency and approved for publication. Approval does not signify that
the contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
ii
Arthur D Little; Inc
-------
SUMMARY
The purpose of this program has been to develop and characterize a high-
purity filter medium for measuring participates in stack gases at 230 to
540°C. Sufficient purity to allow optimum use of the detection limits
of atomic absorption (AAS) and flame emission spectrometry (FES), and
reproducible collection efficiency of 99% for dense particles as small as
0.05 pm were primary objectives.
The work has included analysis and purification of filter materials, fil-
ter handsheet preparation, and filter characterization. The latter
includes the determination of filter suitability for use in X-ray fluor-
escence and neutron activation analysis of collected particulates.
Johns-Manville Co. 99.2% silica fibers have been made into handsheet
filters with satisfactory efficiency, temperature resistance, cost (about
$l/ft2), strength (about 1 lb/in.), and flexibility. The strength and
flexibility result from a novel combination alkaline and 800°C heat
treatment. The heat treatment also decreases extractable impurities and
moisture sensitivity. The application of this treatment to machine pro-
duction of filter media should be evaluated.
Almost all purity requirements for optimum AAS and FES analysis have
been achieved. Major extractable impurities in excess of AAS and FES
detection limits are lead, iron, aluminum, chromium, titanium, and zinc.
The filters appear highly promising for X-ray fluorescence analysis but
should be further evaluated. Suitability of the filters for neutron
activation analysis is uncertain but should also be further evaluated.
Higher purity silica fibers and other similar fibers have been developed
by several companies recently (but not in time to be evaluated in this
program). These fibers should be evaluated.
Collection efficiency has been measured with liquid dioctyl phthalate
(DOP) particles of 0.3 ym diameter. Efficiencies of 99 to 99.99% have
been achieved by using fibers of various diameters. Efficiency of
collection of solid, high-density particles has not been measured, but
we expect the 99.9% DOF efficiency filters to be at least 99% efficient
on such particles.
The strengthened filters are Insensitive to humidity (similar to Gelman
A glass filters), insoluble in most acids and organic solvents, and
slightly alkaline (pH 9 to 10).
ill
Arthur D Little Inc
-------
TABLE OF CONTENTS
*• '
Page
FOREWORD ii
SUMMARY
LIST OF TABLES
I. INTRODUCTION 1
II. REVIEW AND PLANNING 2
A. STACK GAS SAMPLING FOR PARTICULATES 2
B. SILICA FIBER APPROACH 3
C. ATOMIC ABSORPTION AND FLAME EMISSION SPECTROMETRY 5
D. X-RAY FLUORESCENCE ANALYSIS 6
E. NEUTRON ACTIVATION ANALYSIS 6
F. EXTRACTION OF FILTERS FOR ANALYSIS 7
G. MINIMUM DETECTABLE CONCENTRATIONS 9
III. ANALYSIS AND PURIFICATION OF SILICA FIBERS 12
A. CHEMICAL ANALYSIS APPROACH 12
B. EXTRACTABLE IMPURITY SURVEY BY EMISSION
SPECTROGRAPHY . 12
C. EFFECT OF HEATING ON EXTRACTABLE IMPURITIES 14
D. EFFECT OF MACHINE MANUFACTURING PROCESS ON
EXTRACTABLE FILTER IMPURITIES 20
IV. FILTER HANDSHEET DEVELOPMENT 25
A. FILTER STRENGTH VS. FIBER BEATING CONDITIONS,
HEATING TIME, AND FIBER LENGTH 25
B. EFFECT OF ALKALINE TREATMENT AND HEAT
TREATMENT ON FILTER STRENGTH 28
Arthur D Little Inc
-------
TABLE OF CONTENTS (continued)
Page
C.. ALKALINE TREATMENT AND HEAT TREATMENT
WITHOUT WATER WASH 32
D. EFFECT OF FIBER SIZE AND PURITY ON TENSILE
STRENGTH 38
E. SPRAY APPLICATION OF SODIUM HYDROXIDE AND
SODIUM CARBONATE TO PALLFLEX "TISSUQUARTZ" 38
F. TENSILE STRENGTH VS. FIBER SIZE AND PROCESS
CONDITIONS WITH SPRAYED NaOH 41
G. SPRAY APPLICATION OF 70% CODE 106/30% CODE 108
"MICROQUARTZ" FILTERS WITH NaOH AND Na2C03 46
V. HANDSHEET FILTER CHARACTERISTICS 49
A. PRODUCTION OF HIGH-PURITY HANDSHEETS 49
B. FILTRATION EFFICIENCY, FLOW RESISTANCE,
AND TENSILE STRENGTH 50
C. EFFECT OF HUMIDITY OF FIBER FILTER WEIGHT 50
D. FILTER pH 54
E. SOLUBILITY IN ACID, ALKALINE, AND ORGANIC MEDIA 54
F. EXTRACTABLE FILTER IMPURITIES 54
G. SUITABILITY OF FILTERS FOR X-RAY FLUORESCENCE
ANALYSIS AND NEUTRON ACTIVATION ANALYSIS OF
COLLECTED PARTICULATES 58
VI. REFERENCES 64
APPENDIX 65
vi
Arthur D Little, Inc
-------
LIST OF TABLES
Table Page
1 COMPARATIVE MAXIMUM SENSITIVITIES FOR ATOMIC
MEASUREMENT TECHNIQUES 10
2 MINIMUM DETECTABLE CONCENTRATIONS OF IMPURITIES
IN GLASS FIBER FILTERS 11
3 EMISSION SPECTROGRAPHIC ANALYSES OF ACID EXTRACTS
FROM FIBERS 15
4 EMISSION SPECTROGRAPHIC ANALYSES OF QUARTZ FIBERS
BEFORE AND AFTER ACID EXTRACTION 16
5 EFFECT OF HEATING FOR 1 HOUR ON EXTRACTABLE
IMPURITIES IN QUARTZ FIBERS 18
6 EFFECT OF HEATING ON EXTRACTABLE IMPURITY LEVELS
OF CODE 108 MICROQUARTZ 19
7 QUALITATIVE COMPOSITION OF ACID EXTRACTS FROM
FIBERS HEATED AT 800°C 21
8 QUALITATIVE COMPOSITION OF ACID EXTRACTS FROM
AS IS AND IGNITED FIBERS 22
9 QUALITATIVE COMPOSITION OF ACID EXTRACTS FROM
98.5% Si02 FIBERS AND FILTERS 24
10 EFFECT OF FIBER DISPERSION CONDITIONS ON FILTER STRENGTH 26
11 EFFECT OF HEAT TREATMENT TIME AT 800°C ON FILTER STRENGTH 27
12 EFFECT OF LONG FIBERS (1/4" ASTROQUARTZ) ON
FILTER STRENGTH 29
13 EFFECT OF ALKALINE TREATMENT AND HEAT TREATMENT
ON FILTER STRENGTH 30
14 EFFECT OF HEATING TEMPERATURES ON FILTER STRENGTH 31
15 EFFECT OF ALKALINE TREATMENT AND HEAT TREATMENT
ON GELMAN QUARTZ FIBER FILTERS 33
16 EFFECT OF ALKALINE TREATMENT AND HEAT TREATMENT
ON PALLFLEX "TISSUQUARTZ" 34
1
17 EFFECT OF NaOH APPLICATION TO PALLFLEX "TISSUQUARTZ"
ON 12" x 12" MOLD 35
vii
Arthur D Little Inc
-------
LIST OF TABLES (continued)
Table Page
18 EFFECT OF OMITTING WATER WASH ON TENSILE STRENGTH
OF PALLFLEX "TISSUQUARTZ" 36
19 EFFECT OF SHEET FORMATION CONDITIONS ON TENSILE STRENGTH 37
20 EFFECT OF NaOH CONCENTRATION AND APPLICATION TO SHEET
UNDER SUCTION 39
21 EFFECT OF FIBER SIZE AND PURITY ON TENSILE STRENGTH 40
22 EFFECT OF 0.002N NaOH SPRAY TIME ON TENSILE STRENGTH
OF PALLFLEX "TISSUQUARTZ" 42
23 EFFECT OF Na2C03 CONCENTRATION ON TENSILE STRENGTH
PALLFLEX "TISSUQUARTZ" 43
24 EFFECT OF BEATING TIME AND FIBER SLURRY CONCENTRATION
ON TENSILE STRENGTH OF 50% CODE 106/50% CODE 108
"MICROQUARTZ" HANDSHEETS 44
25 EFFECT OF BEATING TIME AND SHEET FORMATION CONDITIONS
ON TENSILE STRENGTH OF 100% CODE 106 "MICROQUARTZ"
HANDSHEETS 45
26 TENSILE STRENGTH OF 70% CODE 106 AND 30% CODE 108
"MICROQUARTZ" HANDSHEETS SPRAYED WITH 0.001N NaOH 47
27 TENSILE STRENGTH OF 70% CODE 106/30% CODE 108
"MICROQUARTZ" HANDSHEETS SPRAYED WITH Na2C03 48
28 PHYSICAL PROPERTIES OF HIGH-PURITY HANDSHEET SAMPLES
PREPARED ON .STAINLESS STEEL EQUIPMENT . 51
29 FILTER SLURRY pH 57
30 MAJOR EXTRACTABLE IMPURITIES BY AAS IN 99.2% SILICA
FIBERS AND FILTERS 59
31 "MICROQUARTZ" FILTER IMPURITIES MEASURED BY
X-RAY FLUORESCENCE 61
32 "MICROQUARTZ" FILTER IMPURITIES MEASURED BY NEUTRON
ACTIVATION ANALYSIS 62
viii
Arthur D Little; Inc
-------
DEVELOPMENT OF A HIGH-PURITY FILTER
FOR HIGH TEMPERATURE
PARTICULATE SAMPLING AND ANALYSIS
I. INTRODUCTION
As a result of recent improvements in trace metal analytical techniques,
sources of many potentially hazardous chemicals in the atmosphere can be
located by sampling and analysis of stack gas particulates. However,
strong, high-purity filter media are not available for high-temperature
use (230 to 5AO°C). Furthermore, the efficiency of filter media in
current use under some conditions is suspected to be inadequate.
The primary purpose of this program was to develop a suitable filter, as
follows:
1. Sufficient purity to allow optimum use of the detection
limits of atomic absorption and flame emission spectrophotometric analy-
sis for Ag, Al, As, B, Ba, Be, Cd, Ca, Co, Cr, Cu, Fe, Hg, Hf, In, Mn,
Mo, Ni, P, Pb, Sb, Sn, Ta, Te, Ti, V, W, Zn, Rh, Mg, K, and Zr.
2. Reproducible collection efficiency of 99% for spherical
particles of 5 g/cc minimum density and 0.05 urn minimum diameter at a
face velocity of 100 cm/sec.
3. Sufficient strength to resist fiber "blowoff," tearing,
cracking, etc.
4. Reasonable cost.
A secondary purpose was to determine the following filter characteristics:
1. Applicability of filter for in-situ X-ray fluorescence and
neutron activation analyses.
2. Effect of humidity on filter weight.
3. Solubility in acid, alkaline, and organic media.
The work has consisted of review and planning studies, analysis and pur-
ification of candidate filter materials, filter handsheet development,
and filter characterization. These activities are discussed in the
following sections.
Arthur D Little Inc
-------
II. REVIEW AND PLANNING
A. STACK GAS SAMPLING FOR PARTICULATES
Stack sampling can be an extremely difficult task with severe constraints
on materials and equipment. One of the most obvious constraints Is high
temperature. Some additional factors of relevance to; this filter develop-
ment program are as follows:
1. Particulate loadings can be highly variable. At very low
loadings, high-velocity sampling with high-purity filters is necessary
for trace-level analytical work so that excessive sampling times can be
avoided. At high loadings, high filter capacity for particulates
(resistance to plugging) is Important.
2. Stack gases often contain gases that corrode or adsorb on
filters, resulting in particulate measurement errors.
3. Stack access is commonly limited to 3" diameter sampling
ports. When the sample must be obtained inside the stack, the choice of
filter type and size is very limited.
Sampling stacks for particulate emissions is similar in many ways to many
other particulate sampling tasks. Filters of high purity, high efficiency,
high weight stability, low flow resistance, adequate capacity, and con-
sistent uniformity are commonly needed but not available. In addition,
however, stack sampling filters require resistance to temperatures of
230 to 540°C.
Most particulate sampling filters for ambient temperatures have been dis-
cussed in references 1 td 3. Relatively recent additions to these avail-
able filters are GE Nuclepore membranes and Selas silver membranes.
Filters currently used for high-temperature stack sampling and their
limitations are as follows:
• Teflon Membranes - Temperature resistance is limited to
about 260°C. Flow resistance is high.
• Porous Ceramics1* - (e.g., Norton alundum thimbles). Flow
resistance is high, and the necessary purity is expected to be difficult
to achieve.
• Selas Silver Membranes - Flow resistance and reactivity are
high. Membrane weight change is expected to result from reaction with
H2S and acid gases.
• High-Efficiency Glass Fiber Filters - Purity is inadequate
but this filter is useful to about 540°C and is specified for use in the
EPA stack sampling train. The commercial suppliers and performance
characteristics of this filter are discussed in the Appendix.
Arthur D Little Inc
-------
B. SILICA FIBER APPROACH
High-efficiency silica fiber filter material can be obtained commercially
from Pallflex Products Corporation ("Tissuquartz") for $1.00/ft2 or less.
It is made on a stainless steel Fourdrinier type of papermaking machine
with Johns-Manville 98.5% "Microquartz" fibers. The material is weak and
too moisture-sensitive for gravimetric measurements, but suitable heat
treatment and/or chemical treatment should eliminate these problems.
Because "Microquartz" is prepared by leaching blown glass fibers, residual
extractable impurities are very low. Heat treatment to evaporate vola-
tile metals and close the porous leached structure is expected to further
decrease extractable impurities. Purity should also be increased by using
the following higher purity fibers:
1. General Electric Co. 10 yim Drawn, Fused Quartz Fibers
These fibers are reported to be the strongest and highest purity (50 to
100 ppm total impurities) commercially available quartz fibers, but con-
tinued availability is somewhat uncertain. Price is about $100/lb. GE
states that 10 to 15 impurities are present, but that the major ones are
Al and Na. This fiber is available from GE in continuous lengths only
and will require cutting to half-inch lengths or less for filter prepa-
ration.
XT.
2. U, P. Stevens & Co. 7 to 9 pm Diameter Drawn, Fused Quartz Fibers
(Astroquartz)
The standard purity quoted for this material is 99.95% silica, but up to
99.98% purity lots can be specified. Fibers are chopped to various
lengths, including 1/4" and 1/2". Price is $35 to $75/lb.
3. J. P. Stevens & Co. Blown. Fused Quartz Wool (Astroquartz)
These fibers are similar in purity to the chopped, drawn Astroquartz.
They are crimped and vary in size from 1 to 12 pm diameter and 1 to
2-1/2" long. It will be necessary to chop these fibers to a maximum
length of 1/2" and possibly as short as 1/8" to permit proper dispersion
and sheet formation. Price is $40 to $75/lb.
A. Johns-Manville 99.2% to 99.5% "Microquartz"
(0.5 Co 0.7 ym diameter and 0.7 to 1.6 ym diameter in unspecified
lengths.)
Because "Microquartz" is the best currently available fiber for making
the desired filters and because "Microquartz" development work is still
in progress, we will identify all currently available grades in terms of
process conditions, as follows:
Arthur D Little Inc
-------
a. 98.5% "Microquartz" - Made with one leach in "old process"*
equipment at 190°F with technical grade h^SO^ and tap wash water. Price"
$4 to $8/lb.
b. 99.2% "Microquartz" - Made with one leach in "old process"*
equipment at 190°F with technical grade I^SOi, and deionized wash water.
Price $15 to $35/lb.
c. 99.5% (best effort) "Microquartz" - Made with two leaches
in "new process"* equipment at 190°F with reagent grade l^SOi, and
deionized wash water. Johns-Manville will aim for, but not guarantee,
99.5%. Actual analyses can be obtained. Availability and price of
production quantities is uncertain.
High-efficiency glass fiber filters and silica fiber filters are made
commercially by conventional papermaking processes. The same processes
should be capable of making high-purity filters, provided that:
1. Raw materials (including water and drying air) are high
purity.
2. Equipment is clean (preferably stainless steel).
3. Housekeeping is immaculate. Work area must be kept
clean and free of contaminants.
Because silica fiber filters have potential for meeting the program
objectives, they were selected for further study. We also considered
other high-temperature fibers5*6 but find them less attractive than
silica. For example, boron nitride (Carborundum Co.) is expensive ($350
to $600/lb), only 99% pure, and would not permit boron analyses. Graphite
could gain or lose weight by gas adsorption or oxidation. Organic fibers
such as Kynol II (phenolic) have inadequate temperature resistance.
*Johns-Manville is developing a new process on a pilot scale for "Micro-
quartz" production. Process changes are expected to produce stronger,
purer fibers, as follows:
1. Lead leaching tanks have been replaced with stainless steel
tanks. (Sulfuric acid is still used for leaching because of low
volatility.)
2. Fibers will probably be dewatered by centrifugation rather than
roll-pressing. This change should result in less fiber breakage and
could result in increased filter strength.
3. Gas-fired conveyor dryer has been replaced with an electrically
heated tumbling dryer. This change is intended to increase fiber
purity.
Arthur D Little Inc
-------
Toward the end of this program, we learned that ICI has developed high- ,
purity alumina and zirconia fibers similar in size and price to "Micro-
quartz," but we have not evaluated them. Another candidate fiber is being
developed by 3M Co. Production quantities of the ICI fibers are scheduled
to be available in the latter half of 1974.
C. ATOMIC ABSORPTION AND FLAME EMISSION SPECTROMETRY
Atomic absorption and flame emission are but two members of a broad class
of techniques in which the sample is first vaporized to produce disso-
ciated atomic species. The atoms are then electronically excited, and
either the absorption or emission of radiation (in the visible or ultra-
violet region) associated with the electronic transition is measured and
related to concentration.
In flame emission, the functions of vaporization, dissociation, and
electronic excitation are all performed by the flame itself. As the
electronically excited atoms relax to their ground state, they emit radi-
ation of a wavelength characteristic of the particular atom and transition.
In atomic absorption, in its most commonly employed form, the function of
the flame is only to vaporize the sample and produce ground state atomic
species. The electronic excitation is produced by a source of mono-
chromatic light, usually a hollow cathode discharge tube; the amount of
light absorbed is measured and related to the atomic concentration.
For the alkali and alkaline earth metals, flame emission is the more
sensitive of the two techniques; for other species, the reverse is usually
true. The relative sensitivities of the two techniques depend largely on
the ionization potential of the element in question. Alkali metals have
a low ionization potential and are easily excited, even in relatively low.
temperature air/acetylene flames. As one moves to the right across the
periodic table, ionization potentials increase, the flame becomes less
effective in producing ionization, and sensitivity falls. For elements
with higher ionization potentials, some flame emission sensitivity can be
regained by using a hotter flame such as nitrous oxide/acetylene. In
atomic absorption, electronic excitation by the flame.is disadvantageous,
since a large proportion of the atoms must be initially in their ground
states if a significant amount of the exciting hollow-cathode radiation
is to be absorbed and consequently measured.
Both techniques, as commonly practiced, involve the aspiration into the
flame of a solution of the sample to be measured. However, there are a
variety of approaches by which atomic absorption measurements may be made
without utilizing a flame. Probably the best known is the "cold vapor"
technique in which mercuric ion in solution is first reduced to metallic
mercury by stannous ion, and the metallic mercury is then sparged out of
the solution with air or nitrogen. The gas stream containing the mercury
vapors is passed into a gas cell, and the ultraviolet absorption is
measured. Elimination of the flame and its associated noise permits one
to achieve a mercury detection limit of a few nanograms. Other alterna-
tives to flame vaporization include electrically heated filaments and
graphite furnaces.
Arthur D Little Inc
-------
D. X-RAY FLUORESCENCE ANALYSIS
When a sample is irradiated by sufficiently energetic photons, usually
X rays from a tube with a chromium or tungsten target inner-shell elec-
trons of the sample atoms are completely ejected from the atom. The
resulting high-energy vacancy is then immediately filled by an electron
from an outer shell of the atom, and a secondary X-ray photon is emitted.
The energy of this "fluorescent" X ray is proportional to the atomic
number of the element and also depends upon the particular valance shell
which is filled.
The sensitivity of analyses by X-ray fluorescence depends both on the
efficiency with which the exciting X rays are absorbed and also on the
efficiency with which the fluorescent X-radiation can escape to the
detector. Elements with an atomic number less than 11 (sodium) are quite
transparent to the exciting X rays and, hence, cannot be measured.
X-ray fluorescence, like neutron activation discussed in the following
section, is usually employed as a nondestructive analytical tool in which
the entire sample, filter plus collected particulate, is subjected to
analysis. Thus, in contrast to atomic absorption and flame emission,
the total level of impurities in the bulk filter must be considered.
X-ray fluorescence measurements are extremely sensitive to the presence
of other elements in the sample matrix. The presence of small amounts
of a heavy metal is of particular concern when one is attempting to
measure a lighter species. Matrix effects also become very important
when one is trying to measure a sample thicker than 10 to 100 microns,
as is the case in the direct measurement of particulate trapped through-
out deep filters (particularly if materials such as glass, which itself
absorbs X-radiation, is used in constructing the filter).
E. NEUTRON ACTIVATION ANALYSIS
Radioactivation consists of the absorption of an elementary particle,
such as a neutron, a proton, or an alpha-particle, into the nucleus of
an atom; the latter becomes unstable and spontaneously emits electro-
magnetic radiation and/or energetic particles. By observing the energies
and intensities of emission, both qualitative and quantitative analytical
information can be obtained. Neutron activation is the most common and
most generally useful. This reaction converts atoms of a particular
Isotope to the next higher isotope of the same element. Thus, for
example, sodium-23 absorbs thermal neutrons to become radioactive
sodium-24. The subsequent radioactive decay, involving emission of both
beta and gamma rays, yields stable 21*Mg.
For analysis of particulates collected on porous filters, the neutron-
activation method can be applied to samples "as received," provided that
the filter substrate is sufficiently free of the elements sought, and
that it is suitable for exposure to an intense neutron flux. Filters
made only of hydrocarbons, such as polyethylene or polypropylene, are
Arthur D Little Inc
-------
ideal because of the complete insensitivity of carbon and hydrogen to
neutron bombardment. Filters containing nitrogen and oxygen are satis- :i
factory because these elements are insensitive to thermal neutrons.
Silica fiber filters are susceptible to neutron activation and require
more careful consideration. One isotope, 30Si, will produce 31Si, with
a half-life of 2.6 hours; its radiation could be troublesome in the case
of long irradiations of several days or weeks. The decay of 31Si is
almost entirely by beta radiation, which is easily distinguished from
the gamma radiation of analytical value.
When long irradiations are needed to activate long-lived elements in the
particulate samples (Fe, Cr, Zn, etc.), any interference produced by the
31Si radioactivity that cannot be suppressed by the detector itself can
be eliminated by suitable radiochemical separation. Radiochemical sepa-
ration for other long-lived nuclides may be equally advantageous.
Facilities for conducting neutron-activation analysis should be capable
of producing a thermal neutron flux of at least 1012n/cm2 sec in order
to achieve adequate sensitivity for all important trace-elements. Avail-
ability of comparable fast-neutron fluxes of known intensity (separate
from the thermal-neutrons) is also desirable for some special situations.
Ideally, the analysis would be done without chemical separation and
would, therefore, require the availability of a high-resolution (Ge:Li)
semiconductor gamma spectrometer facility, equipped with special compu-
tation capabilities for direct calculation of trace-element concentrations
and correction of matrix-interference effects wherever they may occur.
F. EXTRACTION OF FILTERS FOR ANALYSIS
For atomic absorption and flame emission analyses, particulates on filters
must first be put into solution by simple dissolution in water, acid
digestion, fusion, or leaching with solutions containing complexants.
The fact that a dissolution step is required can ease the requirements
for filter purity in two ways. First, only a small proportion of certain
of the impurities within the filter may actually leach out during dissolu-
tion of the sample. Secondly, impurity levels in the reagents required
for dissolution may become limiting factors before impurities leaching
out of the filter itself become important.
Filter extraction should be either preceded by or combined with an ashing
step to remove organic matter. Three types of ashing procedures currently
used for analysis of particulates on filters are as follows:
1. Ashing in a Muffle Furnace at 500°C
Past studies have indicated serious losses of certain metals, namely,
lead, zinc, cadmium, and copper. The cause of these losses has been
ascribed to either vaporization or formation of acid-insoluble silicates.
A recent study7 provides good evidence for the latter and indicates that
muffle furnace ignition in the absence of silicates (i.e., paper filters
Arthur D Little Inc
-------
in platinum ware) gives good recovery of the above metals.
2. Low-Temperature Plasma Ashing at 50 to 250°C in a Low-Pressure
Oxygen Atmosphere
Thompson et al8 show data which indicate that plasma ashing minimizes
losses of most metals, but there is still some question regarding losses
of highly volatile species such as arsenic and mercury.
3. Wet Ashing with Nitric-Perchloric Acid Mixtures
Wet ashing combines the ashing and extraction steps and gives similar
results to plasma ashing.7 However, the acids are somewhat hazardous and
must be of very high purity in order to avoid blank problems.
As a filter purification step, any of these or other ashing procedures
can be used. For example, ashing may be combined with heat treatment to
increase purity and strength.
Ideally, a universal solvent for the particulates of interest, which does
not dissolve silica, should be used both for filter purification and par-
ticulate analysis. In practice, however, a variety of solvents will
probably be used, depending on the particulate elements to be analyzed.
We have therefore reviewed the analytical methods for particulates con-
taining metallic compounds to guide the selection of a filter purification
and analytical extraction procedure.
Procedures used to dissolve particulates have generally involved treatment
of the previously ashed sample for 1 to 3 hours with hot (redistilled)
nitric and/or hydrochloric acids. The most frequently used reagents are
nitric acid (30% to 60% concentrations) and a nitric hydrochloric acid
mixture (32 ml of 40% HN03 + 8 ml of 19% HC1). In one study, dustfall
was treated with nitric, hydrochloric, and perchloric acids and mixtures
of these acids, and hydrochloric was the most successful for extracting
Fe, Cu, Pb, and Zn.9 However, the complete solubility of oxides of
elements such as aluminum, boron, and zirconium in any of these acids is
questionable.
A possible alternative extractant of analytical interest is hydrofluoric
acid, which would dissolve virtually all of the elements of interest.
However, use of HF presents other problems, including the need for plat-
inum containers, the complete dissolution of the glass filter (which will
release all impurities contained in the glass fiber), and the possible
loss of some metals by volatilization.
Perchloric acid offers no obvious advantages over nitric acid in extrac-
tion mixtures but would likely be included in any wet ashing procedure.
The presence of at least some hydrochloric acid may be useful since this
will function as a complexant for many of the metal species, and may aid
in stabilizing the very low concentrations expected.
Arthur D Little Inc
-------
Based on the above review, we decided to use a 1- to 3-hour boiling, high-
purity, nitric-hydrochloric acid treatment for analytical extraction.
This treatment will dissolve the majority of possible compounds of the
metals of interest.
G. MINIMUM DETECTABLE CONCENTRATIONS
Detection sensitivities for most elements of concern when measured by
atomic absorption, flame emission, and X-ray fluorescence spectrometry
as well as neutron activation analysis are presented in Table 1.
A comparison of the detectabilities shown in Table 1 indicates that atomic
absorption is probably a more important technique for trace analysis of
air pollutant particulates than flame emission. In only a few instances
(the alkali metals and alkaline earths) is flame emission more sensitive
than atomic absorption.
It must be stressed that the levels presented in Table 1 are generally
the lowest attainable detection limits. The presence of one or more of a
wide variety of chemical interferences and/or imperfect optimization of
instrument operating parameters can, and usually do, markedly decrease the
sensitivity which is actually obtained. For practical purposes, the
detection limits normally obtained in most laboratories are probably up
to 10 times higher than the levels presented in the table.
In order to establish filter purity goals, we computed the maximum con-
centrations of elements which should be allowed in the filter based on the
best atomic absorption and flame emission detection limits, normal extrac-
tion procedures, and the normal filter weight. This list, presented in
Table 2, gives the minimum concentration which can be detected by atomic
absorption and flame emission in 25 ml of extract from a single 4-inch
diameter (0.85 g) filter.
Arthur D Little Inc
-------
TABLE 1
COMPARATIVE MAXIMUM SENSITIVITIES
FOR ATOMIC
MEASUREMENT
TECHNIQUES
/ «\
Detection Llmitvt*'
Element
Mg
Al
K
Ca
Ti
V
Cr
Mn
Fe
Co
Ni
Cu
Zn
As
Zr
Mo
Ag
Cd
In
Sn
Sb
Te
Ba
Hf
Ta
W
Hg
Pb
AAF
US/ml (b)
0.003
0.04
0.005
0.0005
0.1
0.02
0.005
0.002
0.005
0.005
0.005
0.005
0.002
0.1
5
0.03
0.005
0.005
0.05
0.06
0.1
0.3
0.05
NA
5
3
0.2
0.03
FES
pg/ml(c)
0.005
0.005
0.0005
0.0001
0.2
0.01
0.005
0.005
0.05
0.05
0.03
0.01
0.002
50
3
0.1
0.02
2
0.002
0.5
20
200
0.001
NA
5
0.5
40
0.2
XRF
yg/cm2(d)
0.1
0.1
0.1
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
NAA
(e)
Ug
0.5
0.004
0.2
4
0.1
0.002
0.3
0.0001
2
0.01
0.7
0.002
0.1
0.005
0.8
0.1
0.004
0.005
0.00006
0.03
0.007
0.03
0.02
0.0006
0.1
0.004
0.08
0.5
NA - Sensitivity not available.
(a) Under optimum conditions, presence of interferences or use of other
than optimum instrumental operating parameters can produce signifi-
cant individual or overall degradations (see text).
(b) Atomic absorption with flame (Source, Refs. 10, 11).
(c) Flame emission (Source, Refs. 10, 11).
(d) X-ray fluorescence (Source, Ref. 12).
(e) Neutron activation, weight of total sample can be in range of 0.1-lOg
(Source, Ref. 13).
10
Arthur D Little, Inc
-------
TABLE 2
MINIMUM DETECTABLE CONCENTRATIONS OF IMPURITIES IN GLASS FIBER FILTERS*
Atomic Absorption
Element
Ag
Al
As
B
Ba
Be
Ca
Cd
Co
Cr
Cu
Fe
Hf
Hg
In
K
Mg
Mn
Mo
Ni
P
Pb
Rh
Sb
Sn
Ta
Te
Ti
V
W
Zn
Zr
Na
Si
Ug in
Extract
0.13
1.0
2.5
1000**
1.3
0.75**
0.013
0.13
0.13
0.13
0.13
0.13
380**
0.03**
. 1.3
0.13
0.075
0.05
0.75
0.13
6300**
0.75
0.75
2.5
1.5
125
7.5
2.5
0.5
75
0.05
125
0.05
2.5
yg per g
of Fiber
0.16
1.2
3
1200
1.6
0.9
0.016
0.16
0.16
0.16
0.16
0.16
450
0.04
1.6
0.16
0.09
0.06
0.90
0.16
7500
0.9
0.9
3.0
1.8
150
9.0
3.0
0.6
90
0.06
150
0.06
3
Ug per
sq. ft.
1.6
12
30
12000
16
9
0.2
1.6
1.6
1.6
1.6
1.6
4500
0.4
16
1.6
0.9
0.6
9
1.6
75000
9
9
30
18
1500
90
30
6
900
0.6
1500
0.6
30
Flame Emission
Ug in
Extract
0.5
0.13
1250
NA
0.025
NA
0.0025
50
1.3
0.13
0.25
1.3
NA
1000
0.05
0.013
0.13
0.13
2.5
0.75
NA
5.0
0.5
500
13
125
5000
5.0
0.25 .
13
0.05
75
0.01
125
Mg per g
of Fiber
0.6
0.16
1500
0.03
0.003
60
1.6
0.16
0.3
1.6
1200
0.06
0.016
0.16
0.16
3.0
0.9
6.0
0.6
600
16
150
6000
6.0
0.3
16
0.06
90
0.012
150
yg per
sq. ft.
6
1.6
15000
0.3
0.03
600
16
1.6
3
16
12000
0.6
0.2
1.6
1.6
30
9
60
6
6000
160
1500
60000
60
3
160
0.6
900
0.1
1500
NA = Non-applicable.
*Detection limits based on extraction of a 4-in. diam filter weighing 0.85 g
(density = 10 g per sq ft), a final extract volume of 25 ml, and detection
limit data given in Table 1.
**Detection limits as reported in Atomic Absorption Methods Manual, 1971,
Perkin-Elmer Corporation.
11
Arthur D Little Inc
-------
III. ANALYSIS AND PURIFICATION OF SILICA FIBERS
A. CHEMICAL ANALYSIS APPROACH
Since the detection limits of each of the analytical techniques of inter-
est is known, it was not necessary to analyze the filters by each of
these techniques all of the time. The need was simply to show that the
elements of interest are present below the detection limits. Initially,
our objectives were:
1. To survey candidate fibers for acid-extractable
impurities for comparison with AAS and FES
detection limits
2. To observe the effects of possible processing steps
(e.g., leaching and ignition) on the amounts of
acid-extractables.
Because of the large number of elements to be considered (32) and the
high cost of complete AAS analyses, we used emission spectrography to
estimate the impurities in portions of the dried acid extracts, and used
these data to select a few key elements for further analysis by AAS.
When final candidate filters were prepared, they were extracted and
analyzed for key elements via atomic absorption spectrometry. They
were also analyzed by X-ray fluorescence and neutron activation
in a preliminary way.
B. EXTRACTABLE IMPURITY SURVEY BY EMISSION SPECTROGRAPHY
In order to estimate purity of some candidate fibers and to test the
effect of acid leaching, we carried out a series of three extractions
and emission spectrographic analyses on each of the following types of
high-purity quartz fibers:
1. Code 108 Johns-Manville 99.2% silica "Microquartz"
(0.7 to 1.6 vim diameter fibers of unspecified length)
2. General Electric quartz (10 )jm diameter continuous
filament)
3. J. P. Stevens Standard Astroquartz Mat (7 to 9 ym
diameter crimped fibers bonded with 4 to 5% poly-
vinyl alcohol). Purity was greater than 99.95%.
The fibers were extracted without prior ashing except that a duplicate
portion of the GE fiber was heated at 500°C for one hour and then carried
along in the extraction scheme. The leachings were performed in both
Vycor and Pyrex containers (50 and 100 ml crucibles and beakers).
12
Arthur D Little Inc
-------
Separate blanks were carried through for both Vycor and Pyrex vessels,
Prior to use, all glassware and equipment were soaked in 611 HC1, and
then rinsed well with distilled water.
The extraction sequence was:
leachate
The acid solvent was a dilute aqua regia prepared by mixing 45 ml of 70%
HN03, 10 ml of 37% HC1, and 45 ml of deionized water.. The acids were ACS
Reagent Grade (Fisher Scientific Co.).
The samples were transferred into the extraction vessels with a minimum
of handling using cleaned glass and polyethylene tools. The continuous
filament GE fiber was cut with a pair of stainless steel surgical scissors
after winding an appropriate amount onto a polyethylene rod. Weighed
amounts of fiber were used which could be fully covered by 30 ml of
extraction solvent.
The extraction vessels were covered with watchglasses and heated on a hot
plate at just below the boil for one hour. The hot plate was set in a
large hood and was protected on top and in front by sheets of polyethylene
mounted on a frame to minimize "fallout" into the samples.
The extract was transferred with the aid of a clean glass rod into a
13
Arthur D Little Inc
-------
clean container for evaporation and the extraction was repeated with a
fresh 30 ml portion of solvent. This second extract was transferred into
another clean container and a third extraction carried out. Following the
third extraction and transfer, the fiber mass was washed well with deion-
ized water and dried in a clean container on the hot plate.
All extracts were evaporated to 5 ml volume, using a hot plate at a rate
to avoid spattering. Then approximately 25 mg of lithium nitrate matrix
agent (Johnson Matthey, spectrographically analyzed) were added to each
and evaporation to dryness was completed. The lithium salt was added to
facilitate emission spectrographic estimation of alkali and alkaline earth
elements and to aid the mechanical collection of the residue.
The residues were scraped up and removed with cleaned polyethylene tools
to avoid any abrasion of the glass, and submitted to Kent Laboratories
for emission spectrographic analysis. In addition, portions of the neat
and extracted fibers and a full set of reagent and container blanks were
submitted. Results are listed in Tables 3 and 4. The circled values in
Table 3 are near or below the AAS detection limits.
With few exceptions, the amounts of impurities in Table 3 vary little
from the first to the third extract, indicating that the impurities are
likely distributed within the fiber, rather than present as an easily
removed surface contamination.
The fiber analyses (Table 4) indicate no great decrease in impurity
levels resulting from acid extractions. This also suggests that the
impurities are distributed uniformly in the fiber and difficult to
extract. The apparent large increase in sodium after acid extraction is
an anomaly, but sodium is not one of the trace elements under consider-
ation.
C. EFFECT OF HEATING ON EXTRACTABLE IMPURITIES
Table 3 also suggests that heating the fibers to 500°C reduces the extract-
able impurity content. We therefore used atomic absorption spectrometry
(AAS) to verify the preliminary emission spectrography results. Fibers
used in these experiments were:
1. 99.2% silica Code 106 Johns-Manville "Microquartz." This
fiber was not previously tested. It is the smallest diameter silica
fiber of this purity available from Johns-Manville.
2. 99.2% silica Code 108 Johns-Manville "Microquartz" (same lot
used in previous tests).
3. 99.98% silica J. P. Stevens Chopped Astroquartz. This fiber
is 1/4" long and was not previously tested. The previously tested J. P.
Stevens Astroquartz was a different lot of less certain purity (over
99.95%), and was in mat form with 4-5% polyvinyl alcohol binder.
14
Arthur DLittklnc
-------
TABLE 3
EMISSION SPECTROGRAPHIC ANALYSES
Impurities
Element
Aluminum
Barium
Boron
Calcium
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Nickel
Sliver
Tin
Titanium
Zinc
Zirconium
AA Detection
Limit*
U8 per g
of fiber
1.2
1.6
1200
0.02
0.2
0.2
0.2
0.9
0.09
0.06
0.2
0.2
2
3
0.06
ISO
Emission
Spec.
Detection
Llmlt-pg
0.2
0.05
0.5
0.1
0.05
0.05
0.1
0.1
0.1
0.01
0.1
0.01
0.1
0.1
2
0.5
J.P.S. Astroquartz
(Sample wt. » 0.2g)
Extraction
First
45
NO
0.8
17
4.5
7
29
0.7
17
ND
0.9
2
7
5
ND
ND
Second
70
ND
0.5
B.5
7.5
3
18
ND
4.5
ND
0.5
1
1.2
5
ND
ND
Third
15
ND
ND
34
2.5
65
17
ND
3
ND
0.1
0.7
2.5
10
ND
ND
Total
130
-
©
60
15
75
64
-------
TABLE 4
EMISSION SPECTROGRAPHIC ANALYSES OF .QUARTZ FIBERS
BEFORE AND AFTER ACID EXTRACTION
(pg per g of
Element
Sodium
Aluminum
Lead
Iron
Titanium
Calcium
Magnesium
Nickel
Zirconium
Barium
Strontium
Chromium
Copper
Silver
Zinc
Manganese
J.P.S. Astroquartz
Before After
<1
10
<0.5
10
<0.5
10
3
3
<0.5
1
<0. 1
<0. 1
3
0.3
20
<0.1
5000
10
<0.5
100
<0.5
30
1
20
<0.5
<0.1
<0. 1
<0. 1
3
1
<1
<0.1
fiber)
J-M Microquartz
Before After
300
1000
300
200
100
70
30
20
20
10
10
5
3
1
<1
1
1000
1000
1000
200
100
70
30
1
20
10
10
5
3
<0.1
<1
1
GE lOp Fibers Heated GE Fibers*
Before After After Extraction
-------
4. General Electric 10 ym Quartz Fiber (same lot used in
previous test).
Samples of 2.0 g ± 0.01 g each of the fibers were ignited for one hour
at either 5AO°C or 800°C in previously ignited Vycor crucibles. The
fibers were then extracted, as previously described, with 60 ml of dilute
aqua regia for three hours at just below the boil. The extract was then
decanted off, evaporated nearly to dryness to eliminate much of the acid
and diluted to 50 ml with deionized water. Some samples were extracted
again for one hour in the same manner.
\
The extracts were quantitatively analyzed by AAS for lead, zinc, copper,
aluminum, and iron. The results are shown in Tables 5 and 6. These
results were corrected for blank analyses (no fiber). Pyrex and Vycor
beaker blanks were almost identical; therefore, we used acid-washed Pyrex
beakers for most of these analyses.
During these analyses, filter handsheet development work progressed to
the point where the Johns-Manville "Microquartz" was identified as the
best available fiber for making filters. Therefore, whereas analytical
results on all fibers are reported, we will emphasize the "Microquartz"
results in interpreting the data.
The Code 106'"Microquartz" appears to be slightly purer than the Code 108.
Heating the fibers to 540°C did not reduce the extractable metal contents
but heating to 800°C looks promising, as follows:
1. Extractable zinc and copper levels in J-M "Microquartz" are
reduced to our AAS detection limit by this treatment.
2. Extractable lead levels in J-M "Microquartz" are reduced by
1/2 or 2/3 by one hour at 800°C, but are still above the minimum detec-
tion limit.
Extractable iron levels in J-M "Microquartz" are not reduced by preheating.
Extractable aluminum levels in J-M "Microquartz" were at the detection
limit before heating but doubled after preheating. The increase may be
due to sample contamination with alumina from the furnace lining.
It should be noted that the practical detection limits (signal/noise
ratio =2) of our AAS equipment (Perkin-Elmer Model 303 with Recorder
Readout Accessory and Hewlett-Packard Model 7127A Recorder), as shown in
Tables 5 and 6, are about three to four times higher than those given in
Table 2. However, these detection limits are probably characteristic of
the real day-to-day variation in limits which can be attained in a working
laboratory with common types of currently used equipment.
To determine the effect of heating the fibers to 800°C on the other
extractable elements, a complete qualitative metals survey was done by
emission spectrographic analysis. These samples were prepared by
17
Arthur D Little Inc
-------
TABLE 5
>
c
c:
EFFECT OF HEATING FOR 1 HOUR ON
EXTRACTABLE
IMPURITIES IN QUARTZ FIBERS
Impurities Measured by AAS -
J-M Microquartz
Code 106
J-M Microquartz
Code 108
J.P.S. Astroquartz
GE 10 micron fiber
Not
Ignited
19
<2.5
<2.5
Lead
550°C
18
33
<12*
<12*
800° C
<12*
5.5
18
<2.5
<2.5
Not
Ignited
0.45
0.35
1.1
Zinc
550°C
0.25
0.5
0.5
1.8
US per g
of Fiber
Copper
800°C
<0.25
<0.25
<0.25
0.25
0.37
Not
Ignited
0.8
0.8
1.6
550°C
<0.8
<0.8
<0.8
2.5
800°C
<0.8
<0.8
<0.8
<0.8
<0.8
*Detection limit for older lead hollow-cathode lamp
replaced part way through tests.
o
-------
Element
TABLE 6
EFFECT OF HEATING ON EXTRACTABLE IMPURITY LEVELS OF CODE 108 MICROQUARTZ*
AAS
Detection Limit
Impurities Measured by AAS - ug per g of Fiber
Unheated Fibers
1st Extraction
After 1 hr @ 800°C
2nd Extraction
1st Extraction
2nd Extraction
Lead 4b
Iron 0.4
Aluminum 11
39
38
7.1
11
<11
13
15 13
12 11
6.0 15
6 8
<11 22
19 23
<4
7
9
6
26
13
a. Corrected for reagent blanks.
b. The difference between this value and the 2.5 value in Table 5
reflects daily variation in detection limit.
C
1
D
cr
o
-------
evaporating 25 ml aliquots of the same diluted extracts that were used
in Table 5, adding 20 mg of high-purity lithium nitrate as a carrier,
and evaporating to dryness. Results are shown in Table 7. They have been
approximately corrected for the "blank" by eliminating those elements
which were found to be present either at the same or at a higher level as
a reagent blank (extraction acids plus lithium nitrate carrier) carried
through the same procedure. The qualitative results on the acid extracts
have been related approximately to the level of extractables in the
original fiber sample by multiplying the extract solids level by 0.02
(extract from 1 gram of fiber was mixed with 0.02 grams of lithium
nitrate carrier prior to analysis).
Table 7 indicates that the only "Microquartz" impurities exceeding the AA
detection limits in Table 2 are aluminum, iron, titanium, and chromium.
However, AA data in Tables 5 and 6 indicate that extractable lead levels
also exceed AA detection limits.
Table 8 shows an overall comparison of acid-extractable materials in the
three candidate fibers before and after heating. The data for the non-
ignited fibers is taken from Table 3. These data indicate that the
levels of acid-extractable species are generally one order of magnitude
lower after ignition to 800°C for one hour. The decrease agrees with the
decrease shown by quantitative atomic absorption measurements of lead,
zinc, and copper in the same extracts reported in Tables 5 and 6.
Table 8 agrees with previous AA results on iron which indicate no reduc-
tion due to heating, and the tenfold reduction in "Microquartz" aluminum
level shown in Table 8 suggests that the doubled aluminum concentration
indicated by AA in Table 6 was due to contamination.
This section indicates that 99.2% Johns-Manville "Microquartz" meets
almost all extractable impurity requirements after heat treatment for one
hour at 800°C.
D. EFFECT OF MACHINE MANUFACTURING PROCESS ON EXTRACTABLE
FILTER IMPURITIES
Up to this point, chemical analyses on this program have been limited to
candidate fibers for use in the filter. As a guide to planning the
analysis of several filter handsheets by atomic absorption, we have
attempted to estimate the kinds and amounts of acid-extractable impurities
which could be added to a filter during machine manufacturing by analyzing
both the fibers and the finished filters.
We know that Pallflex "Tissuquartz" is made by machine. We understand
that the machine is almost entire stainless steel and that "Tissuquartz"
consists primarily of 98.5% Si02 Johns-Manville Code 106 "Microquartz."
Ideally, we should have analyzed the same lot of raw material (fiber)
and finished product (filter) to determine the effect of manufacturing on
extractable impurities. However, the best raw material samples we could
obtain were from Johns-Manville Co. and in larger sizes (98.5% Code 108
20
Arthur D Little Inc
-------
TABLE 7
QUALITATIVE COMPOSITION OF ACID EXTRACTS FROM FIBERS HEATED AT 800°C
Approximate Approximate
Level in ... Level in
Extract Solids'1 ' Fiber (c)
21
Arthur D
>10%
0.1-1%
100-1000 ppm
30-300 ppm
10-100 ppm
3-30 ppm
1-10 ppm
0.3-3 ppm
2-20 ppm
1-10 ppm
0.2-2 ppm
0.1-1 ppm
0.02-0.2 ppm
0.01-0.1 ppm
(a)
Corrected for reagent blank (acids
'This is estimate of level in extra<
(c)
Calculated by multiplying level in
Code 108
J-M , . J.P.S. GE Quartz
Microquartzv ' Astroquartzv ' Fiber
actract -olid- x 1 (w£ Qf f±b^ .
Elements in
Reagent Blank
Li
Fe, Na
Ca, Cu, Mg
.Al, Cr, Ni
Si, Ti
Pb
-------
TABLE 8
K>
to
QUALITATIVE COMPOSITION OF ACID EXTRACTS FROM AS IS AND IGNITED **' FIBERS
tTlJ^!^ TS J'M "i«°<">a"z J'P-S- Astroquartz
oi Acid Extractables ...... ,. •
in Fiber As is Ignited^ ' As is Ignited v
>100 pg/g Al
10-100 Al, Pb, B Cr, Cu, Fe
Ca, Mg
1-10 Cu, Fe, Sn Al, Fe, Ti Ni, Ag, Sn Al, Fe, Ti
Ti, Ca, Mg Ti
0.1-1 Cr, Mn, Ni Cr, Pb, Cu Pb, B Cr, Cu, B
Ag, Ba B, Ca, Mg Ca, Mg
.01-0.1 Mn, Ni, Ba Mn, Ba Pb , Mn, Ni
Ba
GE Quartz
(a)
As is Ignited v '
Al, Cu, Fe
Ca
Cr, Pb, Ni Fe
Sn, Ti, Mg
Mn, Ag, B Al, Cr,
Ca, Mg
Ba Pb, Mn,
Ti, B
Cu
Ni
zr
-t
D
(a)
Ignited for 1 hour at 800°C.
Sn not detected (limit ca. 1 yg per g).
Ag not detected (limit ca. 0.01 ug per g).
o
-------
and Code 112 "Microquartz").
We analyzed the "Microquartz" fibers and "Tissuquartz" filters by the
emission spectrography procedure reported in Section III.B. Samples of
1.0 g each were extracted for three hours with 60 ml of hot dilute aqua
regia. The extract was decanted off and evaporated to about 2 ml. Then
0.020 g of high-purity lithium nitrate was added and the remaining liquid
was evaporated. The solids were then analyzed by Kent Laboratories.
Results are presented in Table 9. The data have been corrected for ele-
ments found in a reagent blank carried along with the samples.
The data in Table 9 indicate that the Pallflex manufacturing process has
very little effect on acid-extractable impurities. The major differences
in impurity levels are a slight increase in tin in "Tissuquartz" Roll
No. 2 compared to both "Microquartz" fibers, and a slight decrease in
lead in both "Tissuquartz" rolls compared to Code 108 fiber. However,
because the fibers and filters were not actual raw material and finished
product samples, these differences are not considered significant. We
conclude only that the Pallflex filter manufacturing process probably
does not greatly affect the purity of the raw material fibers.
23
Arthur D Little, Inc
-------
TABLE 9
QUALITATIVE COMPOSITION OF ACID EXTRACTS FROM 98.5% Si02 FIBERS AND FILTERS
Approximate
Level in
Extract Solids2
>10%
1-10%
0.3-3%
0.1-1%
300-3000 ppm
100-1000 ppm
30-300 ppm
10-100 ppm
3-30 ppm
1-10 ppm
Approximate
Level in
Fiber3
0.01-0.1%
20-200 ppm
10-100 ppm
2-20 ppm
1-10 ppm
0.2-2 ppm
0 . 1-1 ppm
0.02-0.2 ppm
Code 108
J-M
"Microquartz"
Si, Na
K
Ba, Pb
Ca, Cu
Sn, Ti
Ni
Code 112
J-M
"Microquartz"1
Si
Na
Sb, Ca, Mg
Pb
Ba, Sn, Ti, Ni
Al, B
Roll //2
Pallflex1
Si, Na
K, Sn
Ca, Mg
Cu, Pb
Ni, Ti
Roll #3
Pallflex1
Si
Na
K
Ca, Mg
Sb, Cu, Pb
Ba, Cr, Sn,
Ni
Ti
B
Elements in
Reagent Blank
Li
Na
K, Fe
Ca
Cu, Mg, Si, Pb
Ba, Cr, Sn, Ni
Mn
KJ
D
ET
Corrected for reagent blank (acids plus lithium nitrate)
Estimate of concentration level in extract solids which are primarily lithium nitrate
Calculated by multiplying level in extract solids x 0-02 frt-off lithium nitrate solids)
' & 1 (wt of fiber taken)
o
-------
IV. FILTER HANDSHEET DEVELOPMENT
As indicated in Section II.B, high-efficiency quartz fiber filters are
commercially .available but are too weak and moisture-sensitive for
gravimetric measurements. We view the low. strength (about 0.4 Ib/in.)
as the principal problem and attribute it to the use of fragile, porous
(leached glass) fibers, the absence of a binder, and weak fiber-to-fiber
attractive forces. The filter can be strengthened by heat treatment at
about 1000°C but it becomes brittle and non-foldable. We therefore
focused our efforts on increasing strength without loss of purity or
flexibility.
Filter handsheets were made by dispersing fibers in water, pouring the
slurry into a papermaking mold, and draining the water (through a woven
wire screen) to form a fibrous web. Strength was measured with an
Instron Tester at a jaw gap of 2" and a crosshead speed of 2"/min. We
measured filter efficiency and flow resistance with the liquid dioctyl
phthalate (OOP) smoke test (ASTM D 2986-71).
A. FILTER STRENGTH VS. FIBER BEATING CONDITIONS, HEATING TIME
AND FIBER LENGTH
A standard laboratory method of preparing aqueous fiber slurries in fil-
ter development work is to use a Waring blender. We expected this method
to produce severe fiber damage, so we investigated several beating con-
ditions, as shown in Table 10.
In these experiments, 0.8 g of Code 106 and 1.2 g of Code 108 Johns
Manville 99.2% silica "Microquartz" fibers were added to 2000 ml of dis-
tilled water in a gallon-size stainless steel Waring blender and adjusted
to pH 2-3 with 1:1 HC1. After beating, the slurries were cast into a 6"
diameter handsheet mold, gravity-drained, vacuum damp-dried, and dried
between blotters at 116°C on a drum drier. The results suggested that
3-5 minutes beating at low speed produced stronger filters than either
less severe or more.severe beating.
As previously discussed, chemical analysis and purification studies indi-
cated that heat treatment at 800°C for 1 hour greatly reduced extractable
impurity levels of many metals. We therefore prepared samples as above
and determined the effect of 800°C heat treatment on filter strength.
The results, as shown in Table 11, indicate that 30 minutes was optimum
but inadequate (0.24 Ibs/in. tensile strength).
The silica fibers used above are air blown and therefore of variable
unknown lengths. The average length, however, is believed to be well
under 1/4". Because fibers of 1/4" length or greater generally increase
filter strength, we prepared additional samples as above, except that we
replaced half of the Code 108 fibers with l/4"-long, 7-9 pm diameter
Astroquartz fibers (40% Code 106, 30% Code 108, and 30% Astroquartz).
Both as-received Astroquartz (silane-coated) and heat-cleaned Astroquartz
25
Arthur D Little, Inc
-------
TABLE 10
EFFECT OF FIBER DISPERSION .CONDITIONS
ON FILTER STRENGTH
Filter
Sample
17-A
15-B
17-B
21-A-l
21-A-2
17-C
17-D
Speed
High
Low
Low
Low
Low
Low
Low
Voltage
110
110
110
110
110
60
40
Time
3
1
3
5
10
1
1
Tensile Strength
Ib/in.
0.17
0.13
0.17
0.19
0.10
0.13
«0.10**
Fiber Loss*
wt %
13.5
7.5
13.5
14.0
23.5
7.5
— —
*Wt % of fibers lost through the screen during handsheet formation.
**Very poor dispersion—many lumps.
26
Arthur D Little; he
-------
TABLE 11
EFFECT OF HEAT TREATMENT TIME
AT 8000C ON FILTER STRENGTH*
Filter
Sample
21-6A
21-6B
21-6C
21-6D
Tensile Strength, Ib/in.
Min. at 800°C
5
30
60
120
Before 800°C
0.12
0.12
0.12
0.12
After 800°C
0.11
0.24
0.21
0.21
*Beating conditions: 1 min. at low speed.
27
Arthur D Little, Inc
-------
(24 hrs at 425°C) were used. The fibers appeared well dispersed and of
reasonable length, but did not increase sheet strength, as shown in Table 12.
B. EFFECT OF ALKALINE TREATMENT AND HEAT TREATMENT ON FILTER STRENGTH
To prevent silica fibers from sticking together in aqueous dispersions,
it has been found necessary to use low pH (2.0 to 4.5). This effect is
not well understood, but it suggested the possibility that high pH might
somehow strengthen the filter. We therefore prepared several handsheets,
as before, except that fibers were beaten at low speed from 1 to 3
minutes, and damp-dry filters were treated with various alkaline solu-
tions and washed.*
Results, as shown in Table 13, indicate that treatment with 0.05 to 0.1N
sodium hydroxide, followed by rinsing with distilled water and heating
30 minutes at 800°C produced the best filter. Strength (0.8 lb/in.),
flexibility, and filtration performance (99% OOP efficiency at 320 cm/min)
are all satisfactory.
Because all raw materials can be high purity, this treatment is considered
highly promising and a major breakthrough in the program. The cost of
this filter is expected to be satisfactory. As indicated in Section II.B,
the cost of Johns-Manville 99.2% Microfibers is about $25/lb. Fiber cost
would therefore be about $0.50/ft2. Total filter cost should be about
$1.00/ft2.
The Table 13 data also indicate that potassium hydroxide, ammonium hydrox-
ide, and tetramethylammonium hydroxide increase filter strength signifi-
cantly, but not as much as sodium hydroxide. Sodium chloride does not
increase strength at all. The lithium hydroxide results appear anomalous.
Sodium carbonate should be further evaluated at lower concentrations.
To determine the effect of higher temperatures on filter strength, the
0.05N NaOH and rinse treatment was repeated except that three tempera-
tures were used. Results, as shown in Table 14, Indicate that 30 minutes
at 800°C is better than 30 minutes at 870°C or 930°C. However, the
*Alkaline treatment and washing procedure is as follows:
After sheet formation by gravity draining, suck dry for 5 seconds and
shut off vacuum pump. Pour 100 cc of alkaline solution carefully around
the edge of the sheet to wet the entire sheet. Suck dry and shut off
pump. Add another 100 cc and suck dry. Repeat this with 200 cc more
(400 cc total). After sucking dry, wash the sheet with 1000 cc of dis-
tilled H20. The first 300 cc are added without suction (poured around
the sheet so as not to upset the fiber mass). Suck dry, leave suction
on, and pour the rest of the distilled H20 directly on the sheet (in
approximately 10-15 sec). Shut off pump; lift off sheet; place on
blotters and dry 15 minutes on drum drier at 116°C. Heat treat at 800°C
for 30 minutes.
28
Arthur D Little; Inc
-------
TABLE 12
EFFECT OF LONG FIBERS (1/4" ASTROQUARTZ)
ON FILTER STRENGTH*
Sample No. Astroquartz Tensile Strength
Ibs/in.
12-A None (control) 0.12
12-E Silane coated** 0.11
12-F Heat cleaned*** 0.13
*Filter contained 40% Code 106 and 30% Code 108 Microfibers and
30% Astroquartz. Code 106 and Code 108 fibers were beaten
1 minute at low speed.
**Silane-coated Astroquartz (2 g) was agitated in a 1-qt Waring blender
for 5 minutes at lower than normal speed with 400 g water at pH 2-2.5
(with HC1) and 0.01 g Daxad 11 (Dewey and Almy Chemical Co. surfactant).
The dispersion was diluted to 3 SL, 10 cc of 0.5% aq. solution of Percol
156 (Allied Colloids, Inc., anionic polyacrylamide dispersing agent)
added, and bucketed 25 times.
***Heat cleaned 1/4-in. Astroquartz (0.6 g) was agitated in 1-qt Waring
blender 4 min. at low speed with 400 g water at pH 2-2.5 (with HC1)
and 0.01 g Daxad 27.
29
Arthur D Little, Inc
-------
TABLE 13
rr
-t
D
Filter
No.
18A-1
18B-1
18C-1
18D-1
18E-1
21-2
21-3
21-4
21-5**
23A
23B
26-2
26-3
26-4
26-5
I
Beating
Time at
JFFECT OF ALKALINE TREATMENT*
AND HEAT TREATMENT ON FILTER STRENGTH
Before 800'C
Heat Treatment
Alkaline
Low Speed Solution
1
1
1
1
1
1
1
1
1
3
3
3
3
3
3
None
1 N
1 N
25% N
25% N
1 N
1 N
0.1
0.1
0.05
0.01
0.05
0.05
NaOH
NaOH
NaOH
NH.OH
N NaOH
N NaOH
N NaOH
N NaOH
N KOH
N
).NOH
0.05N LiOH
0.05N
NaCL
Rinse Tensile
Water Strength
Ib/in.
None
None
Dist H-0
None
Dist HO
pH 4 H20
Dist H_0
None
Dist HO
Dist HO
Dist H20
Dist HO
Dist HO
Dist HO
Dist H20
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
12
40
35
55
20
23
28
35
33
34
20
26
20
31
34
Comments
Control
Brittle, cannot
be folded
Flexible
Fairly flexible
Fairly flexible
Flexible
Flexible
Slightly flexible
Very flexible
Very flexible
Very flexible
Very flexible
Very flexible
Very flexible
Very flexible
After 30
min @ 800°C
Tensile
Strength
Ib/in.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.33
.40
.75
.50
.49
.39
.72
.82
.83
.48
.51
.52
.08
.33
Comments
Flexible
Brittle, cracks
when folded
Stiff, can be folded
Brittle, cracks
folded
Brittle, cracks
folded
Flexible
Flexible
Brittle
Flexible
Flexible
Flexible
Flexible
Flexible
Flexible
Flexible
when
when
*Alkaline solution volume was 400 ml. Rinse water volume was about 1000 ml.
**Four additional Type 21-5 filters averaged 0.74 Ib/in. tensile strength and were quite flexible.
efficiencies at 320 cm/min of 99.0 to 99.4% at 22 to 26 mm HO AP, respectively, were obtained.
OOP
-------
TABLE 14
EFFECT OF HEATING TEMPERATURES
ON FILTER STRENGTH
(30 min heating)
Filter No.
Temperature
°C
Tensile Strength
Ibs/in.
Remarks
2 3-A
800
0.83
Flexible, foldable
25-4A
870
0.50
Brittle, non-
foldable
25-4B
930
0.68
Brittle, non-
foldable
31
Arthur D Little, Inc
-------
effect of time on strength should also be determined.
The Gelman Instrument Co. provided us with two samples of quartz fiber
filter paper made by W. & R. Balston, Ltd. We do not know whether these
sheets were made by hand or machine, but understand that an inorganic
binder has been added to one sample to improve strength. We treated the
dry filters with 0.05N NaOH and rinsed and heated them, as described
above, to determine the effect on strength. Results are shown in
Table 15. All samples were flexible and foldable.
The low tensile strength of the untreated Gelman Type E Quartz suggests
that the filter may have been damaged in the mail. Gelman's tensile
data was much higher (about 0.26 vs. 0.08 lb/in.).
The Gelman Type F Quartz sample was difficult to wet during the alkaline
treatment (suggesting a hydrophobia organic binder) but was strengthened
substantially by heating (from 0.62 to 0.90 lbs/in.).
We also treated dry Pallflex "Tissuquartz" with 0.05N NaOH and rinsed and
heated it, as described in Section B, to determine the effect on strength
and filtration performance. Results are shown in Table 16. All samples
were flexible and foldable. These results show that "Tissuquartz" can be
substantially strengthened by this treatment (from 0.28 to 0.79 lbs/in.)
without degrading filter performance.
To gain some insight into scale-up problems, we treated some Pallflex
material in the same manner on a larger handsheet mold (12" x 12"). This
mold is not equipped with a vacuum pump (as is the 6" mold) and drains
by gravity only. It was difficult to apply the sodium hydroxide and wash
water without damaging the sheet. The sheet had improved strength but
much lower filtration performance, as shown in Table 17. The results
confirm that the wet sheet was damaged by the application of the alkaline
and wash liquids and suggest that the liquids be applied carefully with
the sheet under suction.
C. ALKALINE TREATMENT AND HEAT TREATMENT WITHOUT WATER WASH
We investigated the effect of omitting the water wash by treating
"Tissuquartz" filters with 0.001N and 0.002N sodium hydroxide. We
applied 50 ml to a dry 6" diameter filter, pumped it damp-dry, and
repeated the step. Each unwashed filter was then dried 15 minutes on a
drum drier at 116°C and heated at 800°C. Results, as shown in Table 18,
indicate that 0.002N NaOH and 30 minutes at 800°C produced high-strength
filters (1.14 to 1.30 lbs/in.). All filters could be folded without
cracking, indicating that higher NaOH concentrations without washing
should be evaluated.
To confirm these results as well as to investigate the effect of slurry
dilution and draining velocity, we prepared several slurry batches and
formed sheets in the 6" diameter mold, as shown in Table 19.
32
Arthur D Little Inc
-------
TABLE 15
Filter No.
24-I-EA
24-I-EB
24-I-EC
24-I-FA
24-I-FB
24-I-FC
Type
Gelman
Quartz
Gelman
Quartz
Gelman
Quartz
Gelman
Quartz
Gelman
Quartz
Gelman
Quartz
EFFECT OF ALKALINE TREATMENT AND
HEAT TREATMENT ON GELMAN
QUARTZ FIBER FILTERS
Wt 6 in.
Treatment Diam Disc
Type E None 1.80
Type E 0.05 N NaOH
Type E 0.05" NaOH
+ 30 min at
800°C
Type F None 1.65
Type F 0.05 N NaOH
Type F 0.05 N NaOH
+ 30 min at
800°C
Tensile
Strength
Ibs/in.
0.08
0.09
0.42
0.62
0.80
0.90
33
Arthur D Little, Inc
-------
TABLE 16
EFFECT OF ALKALINE TREATMENT AND
HEAT TREATMENT ON
PALLFLEX "TISSUQUARTZ"
Tensile Filtration Properties @ 320 cm/min
Filter Strength % OOP Efficiency Flow Resistance
Ibs/in. mm HJ)
As received
0.28
99.99
36
After treatment
and washing
0.41
After 30 minutes
at 800°C
0.79
99.99
38.5
34
Arthur D Little, Inc
-------
TABLE 17
EFFECT OF NaOH APPLICATION
TO PALLFLEX "TISSUQUARTZ" ON 12" x 12" MOLD
• Tensile Filtration Properties at 320 cm/min
Filter Strength % DOP Efficiency Flow Resistance
Ibs/in. mm HO
As received
0.28
99.99
36
After treatment
and washing
0.41
After 30 minutes
at 800°C
0.62
97.8
21
35
Arthur D Little, Inc
-------
TABLE 18
EFFECT OF OMITTING WATER WASH ON TENSILE STRENGTH
OF PALLFLEX "TISSUQUARTZ"
NaOH
Concentration Time at 800°C Tensile Strength
(mins.) (Ibs/in.)
0-001 N 15 0.60 to 0.80
" 30 0.90 to 0.91
" 60 0.83 to 0.92
°-002 N 15 0.72 to 0.77
30 1.14 to 1.30
60 0.75 to 0.89
36
Arthur D Little Inc
-------
TABLE 19
EFFECT OF SHEET FORMATION CONDITIONS ON TENSILE STRENGTH
Sample Sheet Formation Conditions Tensile Strength
(Ibs/in.)
1. Gravity drain 2 1 of slurry 0.12 to 0.14
2. Mix 2 1 of slurry with 2 1 0.18 to 0.18
of pH2 tap water in mold
and gravity drain
3. Pump out 2 1 of slurry 0.43 to 0.46
4. Mix 2 1 of slurry with 21 0.20 to 0.23*
of pH2 tap water in mold
and pump out slurry
*Sheet was nonuniform due to mold swirl.
37
Arthur D Little Inc
-------
In each case, we beat 1 g of 99.2% Code 106 Johns Manville old process
"Microquartz" for 1 minute in 2 £ of tap water adjusted to pH 2 with HC1.
Beating was done at low speed in a 6 £ stainless steel Waring blender.
We then added 1 g of 99.5% Code 108 new process "Microquartz" and beat
another minute. After formation, sheets were treated with 0.002N NaOH
without washing as above, dried on a drum drier for 15 minutes at 116°C,
and heated 30 minutes at 800°C. The results are shown in Table 19. All
sheets could be folded without cracking. The results suggest that pumped
drainage of dilute fiber slurries increases strength, and that more con-
centrated NaOH should be evaluated.
To evaluate the effect of NaOH concentration, additional slurries were
prepared as above and drained by pumping. While still under suction,
the sheets were treated with 500 ml of NaOH, pumped to damp dryness, and
treated again. Unwashed samples were then dried on a drum drier for
15 minutes at 116°C and heated 30 minutes at 800°C. Results, as listed
in Table 20, indicate that NaOH concentration should not exceed 0.005N
without washing.
D. EFFECT OF FIBER SIZE AND PURITY ON TENSILE STRENGTH
Previous handsheet samples with adequate strength have all been made with
99.2% Johns-Manville old process Code 106 and Code 108 "Microquartz.."
To investigate the effect of fiber size and purity on tensile strength,
we prepared the fiber slurries shown in Table 21. In each case, we beat
2 g of fibers for 2 minutes in 2 £ of tap water adjusted to pH 2 with
HC1. Beating was done at low speed in a 6 £ stainless steel Waring
blender. Sheets were drained by pumping and treated with 500 ml of
0.002N NaOH with suction as above (twice). Unwashed sheets were dried
on a drum drier for 15 minutes at 116°C and heated 30 minutes at 800°C.
Table 21 indicates that 99.5% Code 108 new process "Microquartz" produces
weaker sheets than 99.2% Code 108 old process "Microquartz." This may
result from excessive leaching and weakening of the 99.5% fibers. How-
ever, these fibers should be further evaluated. Table 21 also indicates
that the larger Code 112 fibers reduce sheet strength.
The high strength of some of these sheets is attributed to the beating
and forming conditions. The stronger sheets in Table 21 cracked very
slightly when folded, indicating that NaOH concentrations should not
exceed 0.002N when sheets are prepared in this way.
E. SPRAY APPLICATION OF SODIUM HYDROXIDE AND SODIUM CARBONATE
TO PALLFLEX "TISSUQUARTZ"
In work thus far, NaOH has been applied to filter handsheets by pouring
from a beaker. To evaluate spray application of alkali, we placed
Pallflex "Tissuquartz" on a dry 6-inch diameter handsheet mold under
suction and sprayed it with 0.002 N NaOH in tap water for various times.
We used an H. D. Hudson* 2-gallon "Clipper" 6215 galvanized sprayer with
*A03 E. Illinois Street, Chicago, Illinois 60611
38
Arthur DLittleilnc
-------
TABLE 20
EFFECT
NaOH
Concentration
OF NaOH CONCENTRATION AND APPLICATION
TO SHEET UNDER SUCTION
Tensile Strength Remarks
fibs/in.)
0.002 N
0.005 N
0.01 N
0.38 to 0.49 Foldable without cracking*
1.50 to 1.73 Brittle
0.69 to 1.41 Very brittle
*DOP penetration 0.67 to 0.95% at 850 cm/min.
Flow resistance 72 to 76 mm HO at 850 cm/min.
39
Arthur D Little Inc
-------
TABLE 21
EFFECT OF FIBER SIZE AND PURITY ON TENSILE STRENGTH
Sample
1.
2.
3.
4.
99.2%
Code 106
Composition
99.2% 99.5% 98.5% Tensile
Code 108 Code 108 Code 112 Strength
(0.5 to 0.7 (0.7 to 1.6 (0.7 to 1.6 (2.6 to 3.8 (Ibs/in.)
yrc dia) pm dia) urn dia) ym dia)
1 g
1 g
1 g
2 g
1 g
1 g
0.65 to 0.97
1.12 to 1.18
1 g 0.33 to 0.52
0.84 to 1.16
40
Arthur D Little Inc
-------
Cone Nozzle Cap, No. 113-476. Spray rate was about 400 ml/min. Filter,s
were kept under suction until damp dry and were not washed. They were
then dried on a drum drier for 15 min at 116°C and heated 30 min at 800°C.
Results are shown in Table 22.
These results suggest that 15 seconds of spraying is adequate, that even
shorter spray times may be satisfactory, and that each sheet was uni-
formly covered with NaOH.
The different strengths of duplicate sheets may be due to different
degrees of damp dryness of the sheets, caused by poor design of our hand-
sheet mold screen. The screen is supported by a rather thick perforated
plate which retains liquid by capillary action. When the suction is shut
off after damp-drying the handsheet, the sheet is rewetted by the residual
water in the support plate. The amount of rewetting appears to be some-
what variable. We therefore are making a new handsheet mold to eliminate
this problem.
We also sprayed dry "Tissuquartz" filters for 30 sec with sodium carbon-
ate solutions of varying concentration. Results, as listed in Table 23,
suggest that 0.002N to 0.005N Na2C03 is preferable to 0.01 to 0.02N.
F. TENSILE STRENGTH VS. FIBER SIZE AND PROCESS CONDITIONS
WITH SPRAYED NaOH
Previous results suggested that beating of fibers for 2 minutes followed
by pumped drainage of dilute slurries and the NaOH and heat treatment
would produce filters of high tensile strength (1 lb/in.). We therefore
prepared several handsheets to determine the optimum beating time and
slurry concentration.
In each case, we beat 1 g of 99.2% old process Code 106 fiber and 1 g of
99.2% old process Code 108 fiber in 2 I of tap water adjusted to pH 2
with HC1. Beating was done at low speed in a 6 «, stainless steel Waring
blender. Slurries were used either as is or diluted with 2 £ of tap
water adjusted to pH 2 with HC1. Sheets were formed by pumped drainage,
sucked to damp dryness, sprayed for 30 sec with 0.002N NaOH, and sucked
again to damp dryness. Unwashed sheets were then dried on a drum drier
for 15 min at 116°C and heated 30 min at 800°C. Results are listed in
Table 24.
These results indicate an adverse effect on dioctyl phthalate (DOP) smoke
penetration of beating for 5 minutes and dilution of slurry (samples 3
and 9), and an adverse effect on flexibility of beating for 3 minutes
(sample 7). They suggest that the beating time and slurry volume under
these conditions should be 1 to 2 minutes and 2 liters, respectively,
and that 0.001N NaOH may improve flexibility.
Similar experiments were also performed with all Code 106 fiber, except
that two slurries were gravity drained rather than pumped out of the
handsheet mold. Results are listed in Table 25.
41
Arthur D Little Inc
-------
TABLE 22
EFFECT OF 0.002N
OF
Sample Spray Time
(sec)
1 15
2 15
3 15
2>
4 30
5 30
6 45
7 45
8 60
9 60
NaOH SPRAY TIME ON TENSILE
PALLFLEX "TISSUQUARTZ"
Tensile Strength
(Ibs/in.)
0.74 to 0.76
1.17 to 1.27
1.34 to 1.37
0.99 to 1.01
0.90 to 0.93
1.02 to 1.12
1.05 to 1.09
0.71 to 0.72
1.32 to 1.38
STRENGTH
Cracks
When Folded?
No
No
Yes
No
No
No
No
No
Slightly
42
Arthur D Little, Inc.
-------
TABLE 23
EFFECT OF Na^CO^ CONCENTRATION
Sample
1
2
3
4
5
6
7
8
ON TENSILE STRENGTH
Na2C03
Concentration
0.002N
0.002N
0.005N
0.005N
0.01N
0.01N
0.02N
0.02N
"OF PALLFLEX "TISSUQUARTZ"
Tensile Strength
(Ibs/in.)
0.96 to 1.01
0.93 to 1.02
0.82 to 1.00
0.97 to 1.00
0.85 to 0.97
0.80 to 0.87
0.87 to 0.92
1.04 to 1.06
Cracks
When Folded?
No
No
No
No
No
Slightly
Slightly
Slightly
43
Arthur D Little. Inc.
-------
TABLE 24
EFFECT OF
BEATING TIME AND FIBER SLURRY
OF 50% CODE 106/50% CODE 108
Sample
1
3
5
7
9
Beating
Time
(min)
2
2
1
3
5
Slurry
Volume
(liters)
2
4
2
2
2
OOP Penetration
at 320 cm/min
0.34
0.67
0.27
0.26
1.15
CONCENTRATION ON TENSILE STRENGTH
"MICROQUARTZ11 HANDSHEETS
Flow Resistance
at 320 cm/min
(mm H20)
29
26
28.8
30
25.5
Tensile Strength
(Ibs/in.)
1.27 to 1.27
1.66 to 1.72
1.17 to 1.36
1.72 to 1.90
1.62 to 1.76
Cracks
When
Folded?
Slightly
Yes
Slightly
Yes
Yes
c
-t
0
-------
c
TABLE 25
EFFECT OF BEATING TIME AND SHEET FORMATION CONDITIONS
ON TENSILE STRENGTH OF 100% CODE 106 "MICROQUARTZ" HANDSHEET?
Beating Sheet Formation
Sample Time Conditions
(min)
Cracks when
Tensile Strength Folded?
(Ibs/in.)
Remarks
Ui
10
Pump out 2 1 of
slurry
Pump out 4 1 of
slurry
Pump out 2 1 of
slurry
Gravity drain
2 1 of slurry
Gravity drain
2 1 of slurry
0.70 to 0.94
0.63 to 0.83
0.72 to 1.07
Yes
Yes
Yes
Sheet stuck to screen.
Sheet had pinholes and
stuck tightly to screen.
Sheet stuck to screen.
Sheet stuck tightly to
screen.'
Sheet stuck to screen.
o
-------
These results also suggest that beating time should be 1 to 2 minutes,
that the slurry should not be diluted to 4 liters, and that 0.00IN NaOH
may improve flexibility. The pinhole formation in the diluted slurry
sample (#4) may explain the increased DOP penetration of the diluted
slurry sample (#3) in Table 24.
A comparison of Tables 24 and 25 suggest that Code 106 fibers be blended
with at least 10% Code 108 fibers to prevent sticking to the handsheet
mold screen, and increase strength and flexibility.
G. SPRAY APPLICATION OF 70% CODE 106/30% CODE 108 "MICROQUARTZ"
FILTERS WITH NaOH AND Na2C03
To confirm the above indications, we prepared several handsheets as above
except that:
1. Fibers were 70% of 99.2% old process Code 106 "Microquartz"
and 30% of 99.2% old process Code 108 "Microquartz."
2. Beating time was 2 minutes.
3. Slurries were used as is (2 liters volume).
4. Heating time was varied from 30 to 120 minutes. Results
are listed in Table 26.
These results suggest that 0.001N NaOH is too dilute. None of the sheets
cracked when folded, but tensile strength was low. The results confirmed
previous indications that pumped drainage gives stronger filters than
gravity drainage, and suggest that heating for 60 minutes at this low
concentration maximizes strength.
We also prepared several handsheets as above except that 0.002N and
0.005N Na2C03 was used instead of NaOH. Results, as given in Table 27,
agree with the indications in Table 26 that heating for 60 minutes at
this low concentration maximizes strength.
The gravity-drained 0.002N Na2C03 sprayed filters in Table 27 (samples 5
through 8) appear somewhat anomalous in that they are generally stronger
than the pump-drained filters (samples 1 through 4) and one sample (//8)
is much stronger and more brittle than would be expected. These results
may be due to the handsheet screen problem discussed in Section IV.E.
In general, these results indicate that handsheet filters with satisfac-
tory efficiency, temperature resistance, cost, strength, and flexibility
can be made by dilute aqueous alkali spraying and heating. The effect
of the alkaline treatment on purity, moisture sensitivity, pH, etc.
should now be evaluated. The scale-up of the strengthening treatment to
machine production of filter media should also be evaluated.
46
Arthur D Little Inc
-------
TABLE 26
-1
^t
c
—I
D
c:
TENSILE STRENGTH
OF 70% CODE 106
"MICROQUARTZ" HANDSHEETS SPRAYED
Sample
1
2
3
4
5
6
7
8
Type of Handsheet
Mold Drainage
Pumped
Pumped
Pumped
Pumped
Gravity
Gravity
Gravity
Gravity
Time at 800°C
(min)
30
60
90
120
30
60
90
120
AND 30% CODE 108
WITH 0.001N NaOH
Tensile Strength
(Ibs/in.)
0.26 to 0.32
0.51 to 0.57
0.32 to 0.35
0.36 to 0.38
0.22 to 0.23
0.42 to 0.46
0.23 to 0.27
0.43 to 0.44
Cracks
When Folded?
No
No
No
No
No
No
No
No
-------
oo
>
r^
3-
-i
b
{L
R
TABLE 27
i
i
TENSILE STRENGTH OF 70% CODE 106/30% CODE 108
"MICROQUARTZ" HANDSHEETS SPRAYED WITH I
Sample
1
2
3
4
5
6
7
8
9
10
11
12
Concentration
0.002N
0.002N
0.002N
0.002N
0.002N
0.002N
0.002N
0.002N
0.005N
0.005N
0.005N
0.005N
Type of Handsheet
Mold Drainage
Pumped
Pumped
Pumped
Pumped
Gravity
Gravity
Gravity
Gravity
Pumped
Pumped
Gravity
Gravity
Time at 800° C
(min)
30
60
90
120
30
60
90
120
30
60
30
60
Ia0C00
Tensile Strength
(Ibs/in.)
0.34 to 0.38
0.47 to 0.53
0.48 to 0.56
0.56 to 0.56
0.58 to 0.62
0.57 to 0.61
0.33 to 0.35
1.13 to 1.17
1.34 to 1.44
1.21 to 1.24
1.10 to 1.15
0.66 to 0.77
Cracks
When Folded?
No
No
No
No
No
No
No
Yes
Slightly
Slightly
Slightly
No
-------
V. HANDSHEET FILTER CHARACTERISTICS
A. PRODUCTION OF HIGH PURITY HANDSHEETS
One hundred high-purity 6-inch diameter handsheets were prepared with
99.2% old process "Microquartz" fibers for filter characterization pur-
poses, as follows:
% Code 106 % Code 108
No. of Sheets "Microquartz" "Microquartz"
40 90 10
20 70 30
40 50 50
Special care was taken to achieve high purity, including the fabrication
and use of an improved 6-1/2-inch diameter handsheet mold. This mold
was all stainless steel except for a very small carbon steel area which
was painted with zinc chromate primer and coated with Pliobond rubber
adhesive. In each case, 2 grams of fiber and 2 1 of distilled water,
adjusted to pH 2 with Fisher reagent grade HC1, were beaten for 2 minutes
at low speed in a 6 £ stainless steel Waring blender. The slurry was
then poured into the stainless steel handsheet mold and drained by
pumping.
The formation screen was a Millipore Corporation.293 mm diameter stain-
less steel support screen with 0.005" diameter holes spaced 0.005" apart.
To minimize filter contamination, the mold was not filled with water to
screen level before the slurry was added, as had been done previously.
Half of the above filters were sprayed under suction with 0.005 N NaOH
and half with 0.005 N Na2C03 (both Fisher Certified reagent solutions)
in distilled water. The filters were sprayed for 15 seconds at a flow
rate of 250 ml/min with a Millipore stainless steel fan spray gun at a
pressure of 15 psi. The solutions were pressurized in a Sears Roebuck
stainless steel garden sprayer in which the brass air pump was replaced
with a connection to compressed nitrogen. The formation screen was blown
free of loose fibers with compressed nitrogen before each sheet was made.
The filters were damp-dried without rinsing, placed between blotters
lined with Whatman No. 1 analytical filter paper and dried on a drum
drier at 115°C for 15 minutes.
The filters were then placed in stacks of 10 on a thin sheet of Hastelloy
and heated 1 hour at 800°C. The top and bottom filters were considered
contaminated from the Hastelloy and possible dusting from the furnace
lining, and were used for DOP and tensile tests. The remaining filters
were trimmed to 5-3/4" diameter, placed between Whatman No. 1 filter
papers, and sealed in polyethylene envelopes. These filters averaged
49
Arthur D Little Inc
-------
about 1.38 g, which is 7.6 g/ft2 or 4.2 g/80 in.2. Fiber loss in proc-
essing was about 20%.
B. FILTRATION EFFICIENCY, FLOW RESISTANCE, AND TENSILE STRENGTH
OOP and tensile test data were obtained on samples of the 100 handsheets.
Results were highly variable, as shown in Table 28. The variability is
attributed to use of the new stainless steel handsheet mold and alkali
spraying system without thoroughly evaluating their performance. Although
less variability would have been preferred, the filters are satisfactory
for the measurements for which they are intended (chemical analysis, pH
measurements, effects of humidity on filter weight, and filter efficiency
tests).
Despite the variability, the data in Table 28 indicate that filter
efficiency can be increased by using less Code 108 "Microquartz" and more
Code 106 "Microquartz." Table 28 also suggests that the higher efficiency
filters are slightly stronger than those of lower efficiency and that the
sodium carbonate-treated filters are slightly stronger than those treated
with sodium hydroxide.
The effect of filtration velocity on DOF penetration and flow resistance
of 10% Code 108 and 30% Code 108 "Microquartz" filters is shown in
Figures 1 and 2, respectively. Note that penetration reaches a maximum
at about 40 cm/sec and that flow resistance increases linearly with
velocity.
We generated 4.5 pm and 12 ym monodisperse sodium chloride particles with
the Berglund-Liu aerosol generator and attempted to perform filter pene-
tration tests with them by light scattering. However, the particle con-
centrations were too low to be measured by this technique.
Efficiency of collection of solid, high-density particles has not been
measured, but we expect the 99.9% OOP efficiency filters to be at least
99% efficient on such particles.
C. EFFECT OF HUMIDITY ON FIBER FILTER WEIGHT
When used for gravimetric analysis, sampling filters should not change
weight at different humidities. To evaluate this effect, we conditioned
and weighed both the NaOH and Na2C03-treated 10% Code 108 filters in
Table 28 and Gelman A glass fiber filters after conditioning overnight,
using the following sequence:
1. At 105°C
2. Over saturated calcium chloride (32% RH)
3. Over Drierite
4. Over saturated calcium chloride (32% RH)
50
Arthur D Little Inc
-------
TABLE 28
PHYSICAL PROPERTIES OF HIGH-PURITY
% Code 108
"Microquartz"
10
10
30
30
50
50
SAMPLES
No. of
Samples
4
3
2
2
4
4
PREPARED ON
0.005N
Alkali
NaOH
Na2C03
NaOH
Na2C°3
NaOH
Na»CO,
STAINLESS STEEL
DOP Penetration
at 320 cm/min.
0.01 to 0.04
0.005 to 0.04
0.20 to 0.34
0.04 to 0.43
0.26 to 1.10
0.15 to 0.73
HANDSHEET
EQUIPMENT
Flow Resistance
at 320 cm/min.
(mm H20)
37 to 43
29 to 48
25 to 28
24.5 to 34.5
19.5 to 24
22 to 27
Tensile
Strength
Ibs/in.
0.18 to 0.63
0.35 to 0.88
0.17 to 0.42
0.26 to 0.67
0.14 to 0.37
0.13 to 0.49
c:
^*
(1
R
-------
50
o
o
10 20 30 40 50 60 70 80 90 100
Velocity (cm/sec)
FIGURE 1 DOP PENETRATION AND FLOW RESISTANCE VERSUS VELOCITY OF
30% CODE 108/70% CODE 106 "MICROQUARTZ" FILTER STRENGTHENED
WITH 0.006 N NaOH
52
-------
0.05
0.04
E
8 0.03
L.
s
D_
O
OJ
a.
O
O
0.02
0.01
Flow Resistance
O
50
40
30 I
10 20 30 40 50 60 70
Velocity (cm/sec)
80
90
100
FIGURE 2 OOP PENETRATION AND FLOW RESISTANCE VERSUS VELOCITY OF
10% CODE 108/90% CODE 106 "MICROQUARTZ" FILTER STRENGTHENED
WITH 0.005 N NaOH
53
-------
Tests were performed in duplicate, and average weight changes relative to
oven-dry weights are shown in Figure 3. None of the filters was signif-
icantly water-sensitive; all weight changes were less than 0.2%.
In a separate but similar experiment, we compared the effects of humidity
on 2 glass fiber filters and 2 Pallflex "Tissuquartz" filters. One
"Tissuquartz" filter had been strengthened by the sodium hydroxide and
heat treatment described in Section IV.C. Results, as shown in Figure 4,
indicate that "Tissuquartz" is much more water-sensitive than the other
filters (up to 7% weight change). The reduced sensitivity of the
strengthened "Tissuquartz" (similar to glass fiber filters) is attributed
to pore closing by heat treatment.
The experimental technique used in these experiments is not good enough
to permit direct comparison of the previous humidity effects with these
results (filters were exposed to room air during weighings). However,
the relative performance in each set of experiments is believed to be
significant. We conclude that either NaOH or Na2C03-strengthened filters
are highly insensitive to water vapor (similar to Gelman A glass filters).
D. FILTER pH
The alkalinity of sampling filters is believed to be important for gravi-
metric analysis due to possible filter reaction with acid gases. There-
fore, although not a contract requirement, we measured the pH of several
filter slurries. Slurries were prepared by stirring 2 g of filter in
100 ml of distilled water for 10 minutes. Slurry pH was measured when
fresh and after standing 40 hours in watchglass-covered beakers. Results
are given in Table 29.
Pallflex "Tissuquartz" has the lowest pH (7.2). Strengthened "Micro-
quartz" filters are similar to glass (pH 8.3 to 10.1). These results
suggest that the alkaline strengthening treatment may need modification
to reduce filter pH where this is an important consideration.
E. SOLUBILITY IN ACID, ALKALINE, AND ORGANIC MEDIA
The "Microquartz" filters described in Section A above are highly insol-
uble in most acids and organic solvents, but are expected to be soluble
in hydrofluoric acid, hot phosphoric acid, and aqueous alkali. They
will be insoluble in organic solvents due to the use of all inorganic
materials and a final 1 hour heat treatment at 800°C. Solubility in
hot, dilute aqua regia is discussed in the following section.
F. EXTRACTABLE FILTER IMPURITIES
In Section III, we used emission spectrography to estimate the impurities
in extracts of 98.5% and 99.2% "Microquartz" fibers and machine-made
98.5% "Microquartz" filters. These results indicated that the major
54
Arthur DLittklnc
-------
0.16
0.12
O 10% Code 108/90% Code 106
Microquartz Strengthened with NaOH
A Gelman A Glass
• 10% Code 108/90% Code 106
Microquartz Strengthened with
Na2C03
0.08
i
6
*->
| 0.04
-0.04
-0.08
105°C
Sat.
CaCI2
Drierite
Sat.
CaCI2
Humidity Condition
FIGURE 3 EFFECT OF HUMIDITY ON FILTER WEIGHT
55
-------
I
s.
o
D> ..
c 4
I
105°C
O Paltflex Tissuquartz
D MSA 1106 BH Glass
NaOH/Heat-Strengthened
Pallflex "Tissuquartz"
Gelrnan A Glass
Sat.
CaCI2
Drierite
Sat.
CaCI2
Humidity Condition
FIGURE 4 EFFECT OF HUMIDITY ON FILTER WEIGHT
56
-------
TABLE 29
FILTER SLURRY pH
Filter
Gelman A glass
MSA 1106 BH glass
Pallflex "Tissuquartz"
pH of Fresh
Filter Slurry
8.3
10.0
7.2
pH of Filter
Slurry after 40
7.0
8.7
5.8
hrs
10% Code 108/90% Code 106
Filter treated with 0.05N
NaOH and rinsed with
distilled water* 9.3 to 9.8
10% Code 108/90% Code 106
Filter treated with 0.005N
NaOH without rinse* 10.1
10% Code 108/90% Code 106
Filter treated with 0.005N
Na2C03 without rinse* 8.6 to 10.0
7.3 to 7.4
8.0
7.0 to 7.7
*"Microquartz"
57
Arthur D Little Inc
-------
extractable impurities of interest in 99.2% "Microquartz" filters will
be iron and lead, and that aluminum, chromium, and titanium may also be •••<
above minimum detection levels. We therefore extracted and analyzed
several of the 100 "Microquartz" 99.2% filters for these and several
other metals by atomic absorption spectrometry.
A stack of eight trimmed handsheet filters (5.75" diameter) of the 10%
Code 108 and 50% Code 108 "Microquartz" types listed in Table 28 was
bisected to give duplicate samples weighing about 5.5 g each. These were
extracted with a total of 240 ml of hot, dilute aqua regia (90 ml of 70%
HN03, 20 ml of 37% HC1, 90 ml deionized water) for three hours. The
extract was decanted off and the fibers washed with a few ml of 1:1 HC1.
Extracts plus washings were evaporated nearly to dryness and then diluted
to 100 ml with deionized water. Two reagent blanks were carried along
with the samples. The results are presented in Table 30.
These results indicate that almost all purity requirements for optimum
AAS and FES analysis have been achieved. Major extractable elements
above the AAS and FES detection limits are lead, iron, aluminum, chromium,
titanium, and zinc.
Aluminum, copper, iron, and lead levels are about the same in both the
raw materials and the finished filters, but zinc levels are much higher
in the finished filters. The zinc may have been extracted from the
small Pliobond rubber-coated zinc chromate-painted area of the handsheet
mold.
Table 30 also indicates that the extractable impurity levels are about
the same in both NaOH and Na2C03-strengthened filters and in both 10%
Code 108 and 50% Code 108 "Microquartz" filters.
Because the "Microquartz" filters are expected to be used in applications
where the purity of glass fiber filters is inadequate, it is of interest
to compare the extractable filter impurities in these two materials. This
is difficult to do because glass filters are made from different types of
glass and filter purity varies widely from lot to lot. In general, how-
ever, emission spectrographic analyses of several MSA 1106BH and Gelman A
glass fiber filters indicate that they have about ten times as much of at
least one of the following extractable impurities as the "Microquartz"
filters in Table 30:
Cr, Cu, Fe, Mn, Ni, Ti, Zn.
Lead levels in the several glass filters were about 50% higher than in
the "Microquartz" filters.
G. SUITABILITY OF FILTERS FOR X-RAY FLUORESCENCE ANALYSIS AND
NEUTRON ACTIVATION ANALYSIS OF COLLECTED PARTICULATES
Detailed analyses were not carried out using these techniques, but EPA was
able to arrange for the following preliminary evaluations:
58
Arthur D Little Inc
-------
TABLE 30
Ln
VO
>
r-»
c
br
?5"
MAJOR EXTRACTABLE IMPURITIES BY AAS IN 99.2% SILICA FIBERS AND FILTERS
Minimum
Detection Level
(yg/g) "Microquartz" Fiber
Element
Aluminum
Chromium
Copper
Iron
Lead
Magnesium
Manganese
Nickel
Titanium
Zinc
Sodium
AAS
1.2-12
0.2-2
0.2-2
0.2-2
0.9-9
0.09-0.9
0.06-0.5
0.2-2
3.0-30
0.06-0.6
0.06-0.6
FES Code 106 Code 108
0.2-2 22
23
0.2-2
0.3-3 <0.8 <0.8
1.6-1.6 8
15
6.0-60 5.5 11
<12 13
18
0.2-2
0.2-2
0.9-9
6.0-60
0.06-0.6 <0.25 <0.25-
0.01-0.1
-
10% Code 108/90% Code 106
"Microquartz" Filter
NaOH
Strengthened
26
31
3.0
3.4
0.28
0.36
9.8
13.5
A.I
6.2
1.3
1.4
<0.4
<0.4
<2
<2
<20
<20
8.7
10.1
430
550
Na2C03
Strengthened
26
52
6.3
6.6
0.55
0.56
11.1
13.2
4.6
5.2
1.5
2.2
<0.4
<0.4
<2
<2
<20
<20
2.8
3.7
640
650
50% Code 108/50% Code 106
"Microquartz" Filter
NaOH
Strengthened
13
19
2.7
3.1
0.55
0.65
6.7
12.4
8.8
11.2
1.1
1.2
<0.4
<0.4
<2
<2
<20
<20
1.4
1.8
260
280
Na2C03
Strengthened
25
26
4.3
4.8
<0.08
<0.08
0.5
17.3
9.9
15.9
1.0
1.3
<0.4
<0.4
<2
<2
<20
<20
5.2
7.0
370
390
-------
1. X-ray Fluorescence
Several of the 100 handsheets were analyzed by the sponsor on an energy
dispersive X-ray fluorescence spectrometer made by the Lawrence Berkeley
Laboratory, using evaporated foil standards. Samples were counted for
300 seconds, utilizing three secondary X-ray fluorescers. Results are
listed in Table 31.
These results indicate that lead and iron are major impurities (1000 to
3000 ng/cm2) and that these impurity levels increase with coarse fiber
(Code 108) content. This finding is consistent with extractable fiber
impurities (Table 5) and the fiber purification method (leaching), and
suggests that the fine fiber content should be maximized. Impurities in
the 100 to 1000 ng/cm2 range are Ti, Ca, Sr, and Cr.
Identical samples were also analyzed in a preliminary way by Columbia
Scientific Industries on their energy dispersive Model 110 Automated
X-ray Fluorescence Analyzer. The general background for all filters was
about 10% less than that from Whatman 41 paper, and the impurities were
much lower than in glass fiber filters. Again, lead and iron were major
impurities and these levels increased with coarse fiber content (up to
5000 ng/cm2).
The EPA and Columbia results both indicate that the "Microquartz" filters
are highly promising for X-ray fluorescence analysis and should be further
evaluated. This evaluation should include the addition of standard par-
ticulate impurities at various depths in the filter.
2. Neutron Activation
Two of the 100 handsheets were analyzed by neutron activation by the
sponsor. Samples were irradiated in the North Carolina State Pulstar
Reactor at a flux of 1013 neutrons/cm2/sec. One portion of each was
analyzed directly by gamma ray spectrometry and the other was first sub-
jected to radiochemical separation. Results are listed in Table 32.
These results are roughly consistent with the X-ray fluorescence analyses
which indicate that Pb and Fe are major impurities. Pb is not detected
by neutron activation, but Fe was detected at about 10X the concentration
of any other impurity reported by this method.
The total Fe level (205-27A ppm) and the total Mn level (10-11 ppm) are
at least 20X the extractable levels detected by AAS (Table 30). Total
Cr and Zn levels measured by neutron activation are two or more times
the extractable levels. These results confirm that dissolution of par-
ticulates from a filter for atomic absorption or flame emission analysis
eases the requirements for filter purity.
We expected that sodium and aluminum might interfere with the neutron
activation analyses but they were not mentioned in the EPA report.
Extractable levels of these elements were measured by AAS to be about
60
Arthur D Little; Inc
-------
TABLE 31
•^
D
"MICROQUARTZ" FILTER IMPURITIES MEASURED BY X-RAY FLUORESCENCE (ng/cm2)*
Element
S
Cl
K
Ca
Ti
V
Cr
Mn
Fe
Nl
Zn
Cu
Br
Pb
Se
Ba
As
Rb
Sr
Cd
10% Code 108/90% Code 106
"Microquartz" Filter
Strengthened with Na?Co^
<300
<120
<100
170 ± 40
490
<30
70 ± 20
<30
1160
36 ± 9
30 ± 15
25 ± 20
<18
890
<20
<250
<20
25 ± 15
180
<120
30% Code 108/70% Code 106
"Microquartz" Filter
Strengthened with NaOH
<300
<150
<100
235 ± 60
430
<30
64 ± 30
<50
955
28 ± 8
25 ± 10
<30
<20
1930
<20
<300
<20
<25
160
<150
50% Code 108/50% Code 106
"Microquartz" Filter
Strengthened with Na?CO^
<200
<120
<100
150 ± 50
500
<30
150 ± 80
40 ± 30
1360
39 ± 6
36 ± 10
<30
<20
3150
<20
<250
<40
<25
162
<100
*Not corrected for attenuation. Errors ±15% in addition to those listed.
-------
TABLE 32
"MICROQUARTZ" FILTER IMPURITIES
Element
Hg
Ag
Mn
Sb
Cr
Zn
Fe
Sc
Co
MEASURED BY NEUTRON ACTIVATION
10% Code 108/90% Code 106
"Microquartz" Filter
Strengthened with NaOH
0.02
0.05
9.7
15
16
19
205
0.10
2.4
ANALYSIS (ppm)
10% Code 108/90% Code 106
"Microquartz" Filter
Strengthened with Na2C03
—
1.0
11
11
26
10
274
0.10
2.8
62
Arthur D Little Inc
-------
500 ppm Na and 25-50 ppm Al (Table 30).
The suitability of these filters for neutron activation analysis is
uncertain and should be further evaluated, especially for purely instru-
mental analysis (no radiochemical separation). As recommended above
for X-ray fluorescence analysis, additional evaluation should include
the addition of standard particulate impurities. Various irradiation
times and "cooling" times should also be included.
63
Arthur D Little Inc
-------
VI. REFERENCES
1. W. J. Smith and N. Surprenant, "Properties of Various Filter
Media for Atmospheric Dust Sampling," ASTM Proc.t Vol. 53. p. 1122 (1953).
2. L. A. Chambers, "Filter Media for Air Sampling," Industrial
Hygiene Quarterly, 290-296 (December 1954).
3. L. B. Lockhart, et al., Characteristics of Air Filter Media
uaed for Monitoring Airborne Radioactivity, NRL Report 6054, March 1964
(AD 600292).
4. L. B. Bentley et al., "Porous Ceramics in Filtration,"
Filtration and Separation. 9^, No. 2, 207-210 (March-April 1972).
5. L. R. McGreight, et al., Ceramic and Graphite Fibers and
Whiskers, Academic Press, 1965, New York.
6. J. I. Jones, "High Temperature Resistant Fibers from Organic,
Polymeric Precursors," Filtration and Separation, ]_, No. 2, 160-167
(March/April 1970) and No. 3, 303-309 (May/June 1970).
7. T. Y. Kometani, et al., "Dry Ashing of Airborne Particulate
Matter on Paper and Glass Fiber Filters for Trace Metal Analysis by Atomic
Absorption Spectrometry," Env. Sci. & Tech. , £, No. 7, 617-620 (July 1972).
8. R. J. Thompson, et al., "Analysis of Selected Elements in
Atmospheric Particulate Matter by Atomic Absorption," Atomic Absorption
Newsletter, 9_, No. 3, May-June 1970.
9. V. Sugawara and V. Yamazaki, "Determination of Heavy Metals
in Dustfall by Atomic Absorption Spectrophotometry," J. Japan Soc. Air
Pollution. 4. (2), 182-187, Nov. 1970.
10. J. D. Winefordner and R. C. Elser, "Atomic Fluorescence
Spectrometry," Anal. Chem. , 43_ (4) , 24A (1971).
11. E. E. Pickett and S. R. Koirtyohann, "Emission Flame
Photometry—A New Look at an Old Method," Anal. Chem., 41^ (14), 29A (1969).
12. L. S. Birks, "X-Ray Absorption and Emission," Anal. Chem.,
44 (5), 557R (1972).
13. Anon., "Activation Analysis," Gulf General Atomic, San
Diego, Calif.
64
Arthur D Little, Inc
-------
APPENDIX
HIGH-EFFICIENCY GLASS FIBER FILTER PERFORMANCE CHARACTERISTICS
INTRODUCTION
High-efficiency glass fiber filters are commonly used to collect par-
ticulates from both ambient air and stack gases. The filter material
was developed for the Atomic Energy Commission1'2 for ventilation appli-
cations. It is a felt-like material composed of a combination of glass
fibers in mixed sizes (about 0.5 to 3pm).
There are specifications for this filter for processing use (MIL-F-
51079A) but not sampling applications. At the present time, two binder-
less filters (MSA 1106BH and Gelman Type A) have been found the most
acceptable for atmospheric sampling and analysis3 »** and are specified
for use in the EPA particulate stack sampling train.
Because significant fractions of particulates collected in the EPA sam-
pling train have been found in the impingers downstream of the filter,
the efficiency of the filter has been questioned as follows:
• Is the filter efficiency adequate for aerosol particles
smaller than 0.3pm?
• Is the filter efficiency for solid aerosol particles of
various densities as good as the tested efficiency for
liquid DOP aerosol?
• What are the effects of high temperature on filter
performance?
• Do electrostatic charges on the particles and/or filter
reduce filter efficiency?
Several of the more important properties of filters are described
below:
• Efficiency is a measure of the ability of a filter to remove
particles from an air stream. Initial efficiencies of new filters are
most commonly quoted, but efficiency as a function of filter loading
is equally important. Performance is frequently expressed in terms of
penetration, which is a measure of the ability of particles to pass
through a filter. Percent efficiency equals 100% minus percent
penetration.
• Flow resistance is measured by the static pressure drop across
a filter at a given flow rate.
65
Arthur DLittklnc
-------
• Particle collection capacity or filter loading is the amount
of particulate that a filter can hold before flow resistance or penetra-
tion becomes excessive.
The filtration performance of these filters depends on filter composi-
tion, particle size, particle loading, and many other factors. However,
the following supplier's data on MSA 1106BH with 0.3pm diameter dioctyl
phthalate (DOP) liquid aerosol is typical:
Initial Flow
Velocity Efficiency Resistance
(ft/min) (%) (in. H70)
10 99.995 1.58 to 1.58
28 99.98 3.94 to 4.33
Linear filtration velocities in the EPA sampling train at 1 CFM are
approximately as follows:
Filter Diameter Velocity
(in.) (ft/min)
2-1/4 36
3 24
4 12
EFFECT OF VELOCITY AND PARTICLE SIZE ON FILTRATION PERFORMANCE
Figure 1 shows that flow resistance increases linearly with velocity.
Figure 1 also shows that efficiency decreases with velocity up to about
50 ft/min and then increases. The minimum efficiency point results
from the combined effects of particle diffusion and impaction, and
varies with filter construction (usually between 30 to 70 ft/min).
The efficiency on particles other than 0.3pm has been the subject of
much misunderstanding and controversy. It is commonly thought that the
filter collects particles only down to 0.3wm diameter. This misunder-
standing results from the common practice of stating 0.3pm DOP effi-
ciencies without interpretation. The DOF test was designed during
World War II to test high-efficiency gas mask filters against what was
thought at that time to be the most penetrating particle size (0.3pm).
The most penetrating particle size of these filters is now believed to
be between 0.1 and 0.3pm diameter; the peak penetration is about 2 to
3 times the DOP penetration.5»b»7»17 Particles smaller and larger than
the most penetrating are collected with greater efficiency, largely by
diffusion and impaction, respectively.
66
Arthur D Little, Inc
-------
FIGURE 1
100.0
EFFICIENCY OF ABSOLUTE FILTER VS. FLOW RATE
(0.3-Micron Dioctyl Phthalate Aerosol)
• PRESSURE DROP
O EFFICIENCY
99.99
99.98
o
z
UJ
ui 9997
99.96
c
CT
o
99.95
25
50
75 100
VELOCITY (FPM)
125
150
Source: Doyle, et al (Ref. 11)
25
175
-------
EFFECT OF SOLID VS. LIQUID AEROSOLS ON FILTER EFFICIENCY
The efficiency of filters for solid vs. liquid aerosols is also not well
understood. However, there is reasonable agreement that HEPA filters
have equal initial efficiency for both solid and liquid aerosols.8~1(^
FILTER LOADING EFFECTS
As the filter becomes loaded with solid particles, efficiency increases,
whereas the reverse is true of liquid particles. Loading with liquid
particles is not a common problem with stack sampling. However, where
it does occur, HEFA filters should be limited to particulate loadings
of about 20 mg/sq cm of surface area.11 The use of two or more filters
in series will increase loading capacity but will also increase flow
resistance.
EFFECT OF PARTICLE DENSITY ON FILTER EFFICIENCY
The effect of increased particle density on filter efficiency is to
increase efficiency at high velocities, due to impaction. Reduced
efficiencies with heavier particles have been measured at low veloci-
ties, but not understood.
EFFECT OF ELEVATED TEMPERATURE ON FILTER PERFORMANCE
This subject is discussed by Dyment13 as follows:
"At elevated temperature, gas viscosity will increase and there-
fore increased pressure drops will be obtained for the laminar flow
conditions which prevail in high-efficiency paper media under normal
ambient conditions.
"The effect of temperature on fibre filtration mechanisms has been
summarised by Thring and Strauss11* in terms of inertial, interception
and diffuslonal collection. Because of the way in which the three
mechanisms vary with conditions (i.e., fibre and particle sizes, fluid
velocity) a general rule cannot be formulated for the effects of temper-
ature on filtration performance. When the predominant mechanism is
known, however, the effects can be predicted qualitatively. Inertia
and interception efficiency are reduced by increased temperature. For
the larger particles for which inertial impaction is more important,
changes in particle/fibre adhesion at high temperature may exaggerate
or modify this effect. The diffusion mechanism is of increasing
importance as particle size diminishes below one micron. For diffusion
the effect of temperature rise is to increase the fibre collection
efficiency particularly for submicron particles."
Dyment also reported the results of some glass fiber filter tests with
sodium chloride aerosol at temperatures up to 500°C. He found that
68
Arthur D Little, Inc
-------
glass paper shrinks at these temperatures so he preheated the samples
to avoid cracking failure. His data, while extremely limited, suggest
that increasing temperatures to 400°C slightly decreased filter effi-
ciency. His conclusions were as follows:
"In practice the effect of temperature and pressure on filtration
mechanism and performance has not been found a determining factor in
the application of high efficiency filters. The main problems are the
physical and chemical effects of a high temperature environment on the
materials of construction of the filter, which are manifested by re-
duced mechanical strength and resilience or loss of adhesion, leading
to mechanical leakage and loss in efficiency."
M°re rffent hi8h-temperature filter efficiency tests were reported by
First. 5 Heat-shrunk quartz fiber filters were tested at temperatures
up to 950°F with sodium chloride aerosols of 0.14ym mass median diameter
at velocities of 15 to 30 cm/sec. No effect of temperature on effi-
ciency was found.
ELECTROSTATIC EFFECTS ON FILTER EFFICIENCY
The effects of electrostatic charges on filtration efficiency are poorly
understood. The most commonly reported effects are relatively small
increases in efficiency when the particles are charged, and greater
increases when the filter is charged.16*17
The high-efficiency glass fiber filter is reported to collect at least
99% of charged particles of all sizes in the atmosphere.18*19 Adverse
electrostatic effects on filter efficiency may be possible, but appear
unlikely.
69
Arthur D Little, Inc
-------
REFERENCES
1. Arthur D. Little, Inc., Report No. NYO 4527 to USAEC, Development of
a High-Temperature. High-Efficiency Air Filter (August 18, 1953).
2. W. J. Smith, "Concerning the Absolute Filter," J. de Recherches
Atmospheriquea, 205-209 (1966).
3. J. P. Pate and E. C. Tabor, "Analytical Aspects of the Use of Glass
Fiber Filters for the Collection and Analysis of Atmospheric Partic-
ulate Matter," Am. Ind. Hyg. Assoc. J.. 145-150 (March-April 1962).
4. D. N. Kramer and P. W. Mitchel, "Evaluation of Filters for High-
Volume Sampling of Atmospheric Particulates," Am. Ind. Hyg. Assoc. J.,
224-228 (May-June 1967).
5. J. Dyment, Penetration of Glass Fibre Media by Aerosols as a Function
of Particle Size and Gas Velocity. AWRE Report No. 05/69.
6. R. G. Dorman, "The Sodium Flame Apparatus in Routine and Research
Tests of Air Filters," Filtration and Separation. 499-503 (Sept-Oct
1969).
7. J. Dyment, "Use of a Goetz Aerosol Spectrometer for Measuring the
Penetration of Aerosols through Filters as a Function of Particle
Size," Aerosol Science. Vol. 1. 53-67 (1970).
8. R. N. Mitchell, et al., "Comparison of Respirator Filter Penetration
by Dioctyl Phthalate and Sodium Chloride," Am. Ind. Hyg. Assn. J.t
357-364 (June 1971).
9. W. J. Smith and N. Surprenant, "Properties of Various Filter Media
for Atmospheric Dust Sampling," ASTM Proc.. Vol. 53. p. 1122 (1953).
10. B. I. Ferber, et al., Respirator Filter Penetration Using Sodium
Chloride Aerosol. U.S. Dept. Interior, Bureau of Mines, RI 7403
(June 1970).
11. A. Doyle, W. J. Smith, and N. M. Wiederhorn, Collection, Monitoring,
and Identification of Particles in Gas Distribution Systems, Am. Gas
Assn. (Cat. No. 46/OR), 1959.
12. E. A. Ramskill and W. L. Anderson, "The Inertial Mechanism in the
Mechanical Filtration of Aerosols," J. Colloid Science. 6_, No. 5,
pp. 416-28 (October 1951).
13. J. Dyment, "Assessment of Air Filters at Elevated Temperatures and
Pressures," Filtration and Separation, ]_, No. 4, 441-445 (July-Aug
1970).
70
Arthur D Little, Inc
-------
14. M. W. Thring and W. Strauss, Trans. Inst. Chem. Engrs., 41, 248
(1963). ~~
15. M. W. First, "Performance of Absolute Filters at Temperatures from
Ambient to 1000°F," 12th AEC Air Cleaning Conference (Aug. 1972).
16. H. L. Green and W. R. Lane, Particulate Clouds. 2nd ed. Ch. 6,
Van Nostrand (1964).
17. R. G. Dorman, "Filtration," Ch. 8 of Aerosol Science, edited by
C. N. Davies, Academic Press, London and New York (1966).
18. C. F. Moore, et al., "Airborne Filters for the Measurement of
Atmospheric Space Charge," J. Geophysical Research. Vol. 66, No. 10,
p. 3219 (October 1961).
19. R. B. Bent, "Testing of Apparatus for Ground Fair-Weather Space-
Charge Measurements," J. Atm. & Terrestrial Phys.. Vol. 26. p. 313
(1964).
71
Arthur D Little, Inc
-------
|