PRODUCT AND PROCESS DEVELOPMENT
Formation of Filter from Glass Fibers VAN A. WENTE
AND
ROBERT T. LUCAS
Naval Research laboratory, Washington 2 5 , D. C .
T
HE use of glass fibers in the preparation of filter paper is a
comparatively recent development. In 1950 a paper made entirely from glass fibers with filament diameters averaging less than 2 microns was produced successfully for the first time on a Fourdrinier paper machine. This early work was sponsored by the Naval Research Laboratory and took place on the semicommercial facilities of the Xational Bureau of Standards ( 1 , %6,?).
The success of the endeavor, as opposed to previous attempts to produce paper from glass fibers, lay primarily in the use of fibers of very fine diameters. If the average fiber diameter was on the order of 1 micron or less, the paper possessed ample strength t o be handled on ordinary paper mill equipment. Special techniques for the manufacture of these submicron glass fibers were developed by the Owens Corning Fiberglas Corp. and Glass Fibers, Inc. A further advantage in the use of very fine fibers consisted of the important fact that the paper then possessed a notably high efficiency in the filtration of aerosols. In the intervening period many paper products have been produced both experimentally and commercially with giass fibers, which have had their principal applications in the fields of filters and electrical insulation. Some of the work performed a t this laboratory in the development of glass fiber aerosol filter materials is here described. Glass fiber tllter papers were prepared with standard equipment
The glass fiber papers were prepared for the most part using standard paper-making techniques-Le., the fibers were broken apart and dispersed in water in a standard holland beater and then the slurry was diluted further with water and pumped directly to the wire of a Fourdrinier paper machine (7). One important exception in handling glass fibers is t h a t high contact pressures which might crush the fibers cannot be used. Thus, the beater roll and press rolls must do little more than make contact. Calendering is eliminated. Some of the papers described are “handsheets,” for which the fibers were dispersed in a Waring Blendor and then the sheets made b y T A P P I standard methods, again with the omission of all high contact pressures. Data on filtration and air resistance were taken under commonly accepted conditions ( 4 ) . The test smoke consisted of dioctyl phthalate droplets, 0.3 micron in diameter, and all measurements were made at a standard face velocity of 14.2 cm. per second. Resistance to air flow is reported as pressure drop in millimeters of water a t this velocity. I n the process of aerosol filtration the following general equation is found t o be valid:
C = e
X - -F100
where C = concentration of smoke at any point in the filter expressed as a fraction of the initial concentration F = per cent of smoke filtered (removed) per small unit of thickness X = number of unit thicknesses through which smoke passed February 1956
The value of F may be taken as an approximate measure of efficiency when the unit of thickness on which it is based is sufficiently small. For some of the efficiencies reported here, this unit is arbitrarily made a 0.1-mil (10-4-inch, 2.5-micron) layer of paper. A more practical efficiency of filtration, however, depends upon the amount of smoke removed per unit of energy expended. Therefore, in some cases F is expressed as smoke removed per unit of pressure drop across the filter, The unit of energy is taken as 1 mm. of water pressure drop a t the standard flow condition. As there is a definite linear relationship between the thickness and air resistance of a given homogeneous filter paper, conversion may be made from one efficiency rating t o the other:
F/0.1 mil
=
(F/mm. HzO) (mm. H20/0.1 mil)
Filtering efficiency of glass paper i s proportional to surface area of fibers
Fiber diameter exercises the greatest influence upon the properties of a glass fiber paper. As the average fiber diameter decreases, both the air resistance and tensile strength increase (Figure 1). However, as the diameter decreases t o less than 0.6 micron the strength reaches a maximum and thereafter starts to decline. This reversal in strength characteristic of the paper is believed t o be caused by a decrease in the average fiber length as well as by deterioration of the inherent fiber strength which might accompany the greater surface exposure. T h e tensile strength values shown in Figure 1 are little more than relative figures, for the actual strength of a 0.6-micron fiber paper might vary from 20 t o 400 pounds per square inch, depending upon fiber treatment and other conditions. The air resistance as shown in Figure 1 follows the expression:
4.8 Resistance per mil = d2
Since theoretical treatments of the flow resistance of fiber mats, a t constant porosity, state t h a t pressure drop is proportional t o the square of surface area (6), the diameter used is the mean volume-surface diameter, due. This average was calculated using methods suggested by Dallavalle (3): dw
a
End2
where n equals number of fibers in each diameter fraction, d. The average diameter values used in this work, therefore, are based upon the distribution-frequency curves obtained from a set of electron micrographs and upon the foregoing equations. Confusion has frequently arisen among glass fiber manufacturers and users over the true average diameter of a particular sample. It is believed t h a t use of the proper average-i.e., the volume-surface average-should be established. The various averages computed for a typical fiber distribution vary widely, as shown in Table I. T h e volume-surface average may be described as the diameter which a theoretical fiber must have in order t o possess
INDUSTRIAL AND ENGINEERING CHEMISTRY
219
PRODUCT AND PROCESS DEVELOPMENT
I 2 3 FIBER SURFACE A R E A (SOUARE METER PEk GRAH)
Figure 2. Effect of fiber surface area upon filtering abiiity Corrected paper thickness is determined under no load and is approximately 1.7 times thickness determined under 8 Ib. pee sq. inch compression. Letter designations A, 2A, etc., are sometimes used commercially to describe fiber size
0.5 I .o 1.5 2.0 AVERAGE FIBER DIAMETER, d,, (MICRONS)
“0
Figure 1. Effect of fiber diameter upon paper strength and resistance to air flaw
the same amount of surface area per unit weight, as an actual fiber sample. Then the average diameter, d,,, which is numerically the largest, corresponds to the true worth of a particular fiber grade when that worth is expressed as surface area per unit JTeight of fiber. This value of surface area per gram has been found t o correspond almost directly t o the filtering ability of a unit thickness of the fiber, as shown by Figure 2. As all of the fibers yield papers with approximately equal densities (0.22 to 0.24 grams per cc., a,pparent),the units given on the abscissa, may be considered almost a direct measure of the fil1,eringability of any fiber sample. In considering a unit thickness of paper, however, a distinctioll must always be made between the apparent and the true thickness. When measured without any compression, the thickness is about 1.7 times greater than that measured by a standard TAPPI caliper gage, which determines thiclmess under a pressure of 8 pounds per square inch, The thickness used in this .ivor,lris the apparent (TAPPI) thickness unless noted otherwise, as in Figures 2 and 3. Although no technique was found completely satisfactory for deternliliing the “no load” thickness, a simple visual method was used n-ith reasonable accuracy. Readings viere taken by setting a machinist’s micrometer just to make contact with the outer fiber layers of the sheets. Contact Tvas determined by observation through a 10power binocular microscope.
Table I.
Effect of Averaging Method upon Diameter of 3A Glass Fiber
Name of Average Median Nuinher Volume-number Volume-surface
220
Figure 3 compares the trio methods of Prating filtration efficiency when plotted against the average f i h diameter, Although thc relative positions of the two curve+may not be compared, their Bhapes reveal that as the fiber diameter decreaws, there is a sharp increase in the filtration (smoke i-emoved) pcr unit thickness, but a more gradual increase when the efficiencv plotted is the smoke rcmoved per unit of air resistuiice. Paper Formation. There are two distinct13 different types of paper that may conveniently be made from glass fiber. One is madc from stock in which the fihcrs are flocked and the other from
Coniputation
d , Microns
2nd -
0 31 0 42
@z
0 52
Zn
X
Z
Z nd
0 64
AVERAGE FIBER DIAMETER, d,,
(MICRONS)
Figure 3. Influence of fiber diameter on filtration per unit paper thickness and an filtration per unit flow resistance
INDUSTRIAL AND ENGINEERING CHEMISTRY
voi. 48, N ~ 2.
PRODUCT AND PROCESS DEVELOPMENT dispersed fibers. When the fibers are separated in ordinary water, inch in they rapidly flocculate, giving fiber clumps around diameter. However, when the p H value of the dispersion is reduced t o 3.5 or below, the fibers remain dispersed. Figure 4 illustrates these two basic conditions of formation and shows that the web being formed from flocked fiber consists of more randomly oriented fiber than the web formed from dispersed stock. The “neutral” or flocked-fiber paper is considered t o be relatively three-dimensional, compared t o the “acid” dispersedfiber paper where the fibers are primarily oriented in two dimensions-Le., in the plane of the paper. In addition t o having greatly different surface smoothness, the two types of paper have different strength and filtration properties. These properties, in particular the paper’s strength, are also sensitive to the concentration of fiber in the stock as shown for a typical 0.6-micron (3A) glass fiber paper in Figures 5 and 6. In Figure 5 it may be seen that a sharp strength increase occurs for acid-formed sheets as the fiber concentration is reduced, It is believed that fiber interweaving and other factors which arise from greater individual freedom for the dispersed fibers begin to come into play below the concentration of 0.1%. Standard conditions for the formation of a 3A glass fiber handsheet a t NRL call for a concentration of 0.015%. Hence, the difference in properties between “acid” and “neutral” sheets is always large. The concentrations normally used on a Fourdrinier paper machine, however, are between 0.1 and 0.2%. The dynamic conditions which exist on such machines apparently shift both curves toward the right, for exactly the same differences result between acid-formed and neutral-formed papers when they are made on a machine as result when they are made in the more static handsheet mold. This compensation afforded by the high degree of movement on a paper machine has not been found to alter significantly the properties of neutral-formed paper. Although the filtration performances of the two types of paper (Figure 6) do not differ as greatly as their strengths, the difference is ncvertheless significant under conditions of use. The dissimilarity of fiber orientation in the paper is believed t o be the primary cause of the variance.
300r
,
I
1
OO
0.1 0.2 0.3 0.4 FIBER CONCENTRATION IN H A N D S H E E T MOLD ( P E R C E N T )
0.5
Figure 5. Influence of acidity and fiber concentration on paper strength Paper made in TAPPI standard mold from 0.6-micron (3A) glass fiber. Acid pH 3.2; neutral pH 7.8
The more common anionic surface agents have very little effect upon the properties of glass fiber paper. An interesting result was obtained when glass fibers (7001, 2-micron fiber and 3oq;’, 0.6-micron fiber) were dispersed in water containing 0.01% ’ each of cationic and nonionic surface agents (Triton K-60 and Triton X-100, respectively). The resulting paper possessed almost no strength, was very fragile, but performed very well as a filter. When the paper was treated further, first by burning off the organic surface agent a t 600” F. and then by rewetting the sheet with plain water and redrying, significant changes occurred, as shown in Table 11. By oxidiz-
-
NEUTRAL
I
I
01
0.2
F I B E R CONCENTRATION
IN H A N D S H E E T MOLD ( P E R C E N T ) Figure 6. Influence of acidity and fiber concentration on filtration per unit flow resistance ( a ) FLOCKED FIBERS (NEUTRAL)
Figure
4.
( b ) DISPERSED FIBERS (ACID)
Formation of paper with flocked and dispersed fibers
Dispersing Agents. Glass fibers may be dispersed in water by means other than simply reducing the p H value to below 3.5. Addition to the slurry of Calgon (sodium hexametaphosphate) or certain cationic surface active agents will also separate fiber clumps. The use of Calgon is particularly effective a t a p H of 3.5, where it serves to disperse the fibers further and thereby improve the strength. Surface active agents of the quaternary ammonium type will effectively disperse glass fibers, especially when they are used in conjunction with a nonionic surface agent. February 1956
Paper prepared as described in Figure 5
ing out the organic film which had been deposited upon the fibers, strength was greatly increased, for this film behaved as a lubricant, permitting slippage between fibers. Smoke penetration was somewhat lower and, hence, efficiency higher than it had been initially. When the burned sheet was rewetted, its density, air resistance, and strength increased. However, the penetration, instead of decreasing as expected with the higher air resistance, increased by a factor of 2.8. The increase in density was an understandable result of action by the water surface tension which pulled the fibers together as the water evaporated. The sheet thickness thereby decreased, giving a density more comparable (0.17 gram per cc.) t o that normally obtained. The
INDUSTRIAL AND ENGINEERING CHEMISTRY
22 1
PRQDUCT AND PROCESS DEVELOPMENT dried. As the binder content increased, the strength rose rapidly, approaching values normally associated with fiber reinforced plastics. Simultaneously the air resietance and the relative smoke penetration also increased, with the resistance increasing more slomly and even going through a slight minimum a t first. Efficiency of filtration naturally decreased as a result of the higher penetration and resistance values. Results similar to these are obtainable when the binder is added as an aqueous dispersion t o the paper stock. In this case, however, surfaw active agents that’ are present in resin dispersions can seriously impair strength, especially if the agents are cationic in nature. i
I
I
1
I
I
i
I
1
Summary 2
v,
500
The surface area of glass fibers per unit weight has been found to be almost directly proportional t,o the ability of a glass paper to filter fine aerosol particles. The volume-surface average fiber diameter should therefore be used for comparing various fiber samples and for predicting filtration performance. A convenient method for determining this average diameter consists of measuring the standard air-flow resistance per mil of paper thickness and applying the formula:
CL
2 400 z u
300 v, W
d
200
G? W z +-
too
I
0 Figure 7.
I
I
I
I
2
3
i
I
I
I
4 5 6 7 RESIN BINDER (PERCENT)
1
I
8
9
IO
Effect of adding binder to glass fiber paper
0.6-micron (3A) acid-formed Fourdrinier glass paper saturated with toluene solution of poly(viny1 chloride)-type resin (Geon 200 X 20)
most significant result of this experiment consisted of the simultaneous increase in air resistance and in smoke penetration that occurred after the paper was rewetted. This phenomenon would seem t o indicate that an increase in the apparent average fiber diameter was brought about possibly by the action of surface tension as the water evaporated.
Table II. Effect of Burning and Rewetting upon Glass Paper Formed from Fibers Dispersed b y Cationic Agent Rel. Filtration ThickTensile .Air A. B.
C.
Paper A s made Surface agent burned out Rewetted and redried
ness, AIils 45
Strength, Resistance, Smoke Efficiency, Lb./Sq. Inch 3 I m . HpO Pen. F / l I m . H20 2 90 1 0 9 0
45
20
89
O R
9 4
32
47
94
2 8
7 7
Binders. In general, binders may be used to increase the strength of glass fiber paper, but they concurrently decrease its filtration efficiency. A possible mechanism for this adverse effect would be one similar t o the action of water during the second drying described in the preceding section. Typical effects of the addition of a binder was illustrated in Figure 7. The binder used t o obtain these results was a poly(viny1 chloride) “sohtion”-type resin (Geon 200 X 20). Previously dried papers were saturated with solutions of various concentrations and then re-
d,,
=
d z ~
When pnper is formed at sufficiently low fiber concentrations from fibers that have been specially dispersed, its tensile strength increases over fivefold but filtration efficiency is significantly reduced. Fibers may be dispersed by acidifying t,he paper stock (pH 3.5 or lower) or by adding certain dispersing agents mch as Calgon or cationic surface agenrs. The average fiber diameter of a paper may be altered after the paper is formed. The presence of binders or surface active agents can greatly influence the appaxent fiber diameter and consequently affect the filtration efficiency of a given fiber sample, Therefore, in the manufacture of filter materials, all agents which might possess surface activity must be eit,her absolutely eliminated or used mith extreme care. Literature cited (1) Callinan, T. D., and LucaF, R. T., “Electrical Properties of Glass
Fiber Paper,” Saval Research Laboratory, NRL Rept. 4042 (Oct. 30. 1952). (2) Callinan, T. D., Lucas, R. T., and Bowers, R. C., Elec. M f g . 48, 94-7 (August 1951). (3) Dallavalle, J. AI., “iGcromeretics,” 2nd ed., p. 47, Pitman, Sew York, 1948. (4) Knudson, H. W., and White, Locke, Sam1 Research Laboratory, hTRLRept. P-2642 (Sept. 14, 1945). (5) Langmuir, I., “Filtration of Aerosols and Development of Filter Materials,” Office of Scientific Research and Development, OSRD Rept. 865 (Sept. 4, 1942). (6) O’Leary, M. J., Hobbs, R. B., Missimer, J. K., and Erving, J. J., TAPPI 37 KO. 10’1.446-50 (October 1954). (7) O’Leary, M. J., Scribner, B. W., Missimer, J. K., and Erving, J. J., Ibzd., 35 (NO.7) 289-93 (July 1952). RECEIVED for review April 3, 1956. A C C E P T E D October 18, 1956. Division of Industrial and Engineering Chemistry, Symposium on Respirntory Protective Devices and Civil Defense, 127th Meeting, .4CS, Cincinnati, Ohio, March-April 1955.
E N D OF PRODUCT AND PROCESS D E V E L O P M E N T SECTION
222
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
Vol. 48, No. 2