Fire Barrier Coating and Fire Barrier Plywood

This invention relates to a fire barrier and more particularly to a fire resistant laminate for retarding flame spread and flame penetration. This invention also relates to a fire barrier composition, a substrate coating with the composition, a laminate based on the coated substrate and a method of making the composition.

Numerous methods have been developed for controlling the ravages of fire. The rationale for each of these methods is found in the physics of combustion, the chemistry of flame, and the engineering of fire control systems. The disclosure herein presented treats with the use of coatings as a fire control methods; a barrier system intended to prevent the ignition of and spread of flame along a combustible substrate.

The prior art discloses four major types of fire-protective coatings: (1) ablative; (2) intumescent; (3) sublimation and gas- or vapor-producing; (4) "ceramic" or inorganic. Ablative coatings are generally thick and heavy and costly, thus limiting their application to special uses. Intumescent coatings are generally soft and easily abraded, and have limited duration of protection--10 to 15 minutes when exposed to fire. Sublimative or gas-producing coatings have even shorter duration of protection--after their snuffing gases have been produced and wafted away from the surface, they are left without a protecting mechanism. They have found their major use in the treatment of fibers and clothing. Ceramic coatings, as the name implies, require high-temperature curing in order to form the ceramic bond--which many structural or building components cannot withstand. Inorganic coatings, such as Portland cement, gypsum, calcium aluminate cement, phosphate-bonded cement, metal-pigment loaded silicate coatings (sodium, potassium, ethyl, etc.), high-temperature silicone, and magnesium "oxychloride" and magnesium "oxysulphate" cements have all been proposed. This disclosure treats specifically with magnesium "oxychloride" in combination with calcium aluminate cements and other high-temperature resistant binders.

It is remarkable that in the long history of oxysalt cements, so few attempts have been made to convert these materials into a fire barrier in the form of thin paints. No doubt this is, in part, due to the prior major shortcomings of the products and the inability of the developers to overcome them. Amongst these shortcomings are: brittleness and rigidity, along with the proneness to spalling or decrepitating under heat. Unless mixed in stoichiometric ratios, the result is unsightly efflorescence ("white bloom") of the gauging salts (magnesium chloride, magnesium sulphate, magnesium carbonate) which continually condenses onto the surface of the coating. In other cases, especially in the oxysulphates, the excess liquids tend to synerese (squeeze out) and leave a greasy wet film at the surface. The coatings exhibit poor aging and weatherability characteristics, as well as solubility in water and rain, thereby making them unusable in outdoor situations. (The rain and high humidity leach the chloride and sulphate from the formation, leaving void spaces as the result of the leaching process, which in turn leads to a disintegration of the coating.) They have a strong tendency to be hygroscopic, picking up moisture from the air, and thereby presenting a continuously wetted surface; they show poor resistance to common solvents; and there is a loss of tensile strength and cohesiveness after exposure to flame, causing cracking which exposes the substrate to ignition and subsequent combustion. In addition, some coatings are relatively soft and easily scratched and abraded, and unable to resist scrubbing (oxysulphate coatings are markedly softer than oxychloride coatings). Thin coatings, although they adhere well to substrates, frequently shrink in drying, and show "mud-cracking" (crazing), thus exposing the undrlying substrate to direct flame impingement.

Thus, there exists a need in the art to develop a durable, quick-setting, non-combustible thin coating capable of protecting a variety of susceptible flammable substrates from ignition ("fire-barrier"), one which at the same time has all the attributes of a good paint and, additionally, aids in overcoming the above-listed shortcomings in the prior art. Another need is to develop a "fire-barrier" coating which does not spall or decripitate or crack under the heat of flame and thus expose the underlying substrate. A further need is to develop an inexpensive Class "A" fire-retardant coating as defined by the requirements of the National Fire Protection Association (NFPA), utilizing inorganic non-toxic components, and based on magnesium oxide-magnesium chloride "oxysalt" chemistry. Other needs are to develop a coating which would bond to wood, plywood, and glypsum (plaster wallboard, to fiberglass board and batting, and to other substrates, imparting to them fire-barrier properties and increased flexural strength when incorporated as an element of the laminate structures. In addition, a barrier coating must possess a suite of other physical characteristics: hardness, compressive and tensile strengths, cohesiveness, adhesion, and a variety of other properties, including cosmetic attributes, which keeps the coating's integrity under the stresses of fire exposure, so that it does not crack, spall, flake off, or otherwise lose its ability to act as a fire barrier. The coating itself must be noncombustible, so that it does not contribute to the fuel, nor aid the progress of a flame front.

Although natural wood as a material of construction has many desirable properties, some of its shortcomings, such as its lack of strength in thin layers, and its inherent combustibility, have restricted the scope of its application. The economic necessity of utilizing "waste" subproducts, and the notable product improvement in strength and appearance has led to the widespread use of laminated wood products in the form of plywood, wood panelling, and veneers, such as doorskin, as well as veneer-covered compressed particle and waferboard. Construction and industrial plywoods are widely used for floor, wall and roof sheathing; decorative hardwood plywood is used for interior panelling as well as for furniture and many specialty applications. The 1984 edition of "U.S. Industrial Outlook" reported that the market for wood panel products in 1983 reached $6.6 billion, of which the industrial softwood panels represented 65% of the shipments; hardwood panels represented 25%, and particleboard held 10% of the market. 2.60 Billion square feet (surface measure) of hardwood plywood was produced in the U.S.A. in 1983, versus 3.9 billion square feet in 1977 (a reduction of 33%), while an additional 1.48 billion square feet were imported. (Softwood plywood volumes far exceed those of hardwood plywood; viz. over 20 billion square feet per year.)

If one of the major drawbacks to the use of plywood--its flammability--could be economically counteracted without necessitating the use of claddings, fire resistive glypsum wallboard shielding, space isolation, etc. now required by construction codes, the market and applications for plywood would be further extended.

Plywood and panelling is manufactured by bonding together layers (plies) of thin sheets of wood (veneer). The layers are glued together with the grain direction of adjacent layers at right angles. The veneer is usually rotary-peeled from logs, but may also be sliced or sawn. After trimming, drying and grading, the veneers go to blue spreaders, where adhesive is applied and the plywood panel is laid up. The plywood is generally hot-pressed in large multiopening heated hydraulic presses. The application of both heat and pressure cures the glue.

Construction and industrial softwood plywood is made in two types: Interior and exterior, on the basis of resistance of the panels to moisture, and in thicknesses ranging from 1/4" to 11/8". Decorative hardwood plywood is available in thicknesses from 1/8" to over 2". The standard size of a plywood sheet is 4'.times.8', but other sizes and thicknesses can be manufactured on special order. Typical plywoods are manufactured in 3 to 5 layers, but with 3 to 6 plies. Although weights will vary with the species of wood used, 5/16" thick unsanded panels weigh approximately 1.0 pound per square foot; 3/4" panels weigh 2.2 pounds per square foot, and 1" thick panels weigh 3.0 pounds per square foot. Hardwood plywood is generally a decorative wall penaling product, but is also used structurally, for example, in the structural integrity of mobile and manufactured homes.

Becaue of plywood's cross-laminate structure, it can support highly concentrated loads, even at unsupported edges. Shear strength varies with the kind of glue used, even though the glue line is very thin--a few mils thick (dry).

Several types of water-dispersable glues are used in plywood bonding, which vary in their resistance to moisture and are one of the bases used for grading as exterior or interior grace. Most glues used are thermosetting, and these vary in the temperature, temperature-time, and temperature-time-pressure required to "cure", as well as "pot-life", "open time", viscosity, "spreadability" and other parameters essential for a suitable bonding agent. Softwood plywoods generally use phenolic adhesives (phenol-formaldehyde); decorative hardwood plywoods generally use urea formaldehyde resins. The phenolic glues, which are highly water-resistant, will cure in 4 to 5 minutes at temperatures as low as 300.degree. F., but they have a short shelf life. The urea formaldehyde adhesives can be catalyzed with ammonium sulphate at 0.5 to 1.0% to bring about a lower temperature cure, and typically require from 150 to 300 psi pressure (depending on the density of the wood species being bonded) at 230.degree. to 260.degree. F. for 3 to 5 minutes. Resorcinol and phenol-resorcinol resins are highly waterproof, and if properly catalyzed do not require heating for curing--that is, they bond at room temperatures. Paraformaldehyde is a commonly used catalyst in bringing about the cross-linking reaction. However, in comparison to the phenolics and ureas, they are very expensive, and hence are not used in plywood or panelling lamination, but are used in laminating heavy timbers and beams ("Glulam"). Other widely used laminating adhesives are based on melamines (melamine formaldehyde, melamine urea, etc.), but these are used mainly in furniture decorative laminates, etc.

Most of the glues used are flammable, although some are "self-extinguishing". They impart little fire resistance to the plywood.

The fire hazard associated with the flammability of wood products is the principal limiting factor in the use of laminated woods as a material of construction. Almost all local, state and regional governing bodies have enacted building Codes which regulate the use of wood products in construction, and which dictate other norms for the prevention, early detection, and extinguishment of fires. (Most "advanced" countries also have national building codes). Some fire-preventive methods are "structural", i.e., isolation of the wood members behind fire barriers; others are "spatial"--separating wood members by "fire breaks"; and still other require the use of sprinkler or water deluge system installations. "Chemical" methods are also available, although not as widely used. These fall into two major categories: (a) Impregnation with fire retardants; or (b) Coating with paints which retard the spread of flame. Both are effective for short periods of time in reducing flame spread, but neither is significantly effective in increasing the resistance of wood to degradation under sustained fire exposure, or in preventing reduction of its load-bearing capacity.

The National Fire Protection Association (NFPA) has established a Standard (#703-1979) for both "Fire Retardant Impregnated Wood" and "Fire Retardant Coatings for Building Materials." The American Wood Preservers Association has a Standard (AWPA C-27) with a requirement for a flamespread of 25 or less, with no evidence of significant progressive combustion when tested for 30 minutes duration under the ASTM E-84 Test Method. "Fire Retardant Treatment of Lumber and Plywood" is also covered by Federal Specification MIL-L-19140 D (Navy). U.S. Federal Specification SS-A-118(b) covers the use of flame retardant coatings on unfinished interior plywood and other construction products.

Impregnation with fire retardant chemicals is an expensive procedure and a complicated process, since, in order to be effective, the chemicals must be applied under pressure in a closed pressure cylinder which requires that a vacuum be pulled first on the load to extract the air from the wood cells prior to pressurization. Chemicals commonly used in this impregnation are water-soluble sulphate, borax, boric acid, zinc chloride, sodium dichromate and combinations thereof. Some of the salts are corrosive to metal fasteners and hardware (nails, screws, bolts, hinges, etc.); some prevent surface painting; some species of wood must be scored to ensure penetration of the treating solution; and being water-soluble, the salts tend to leach out if exposed to weather. American Plywood Assocation (APA) mandates that allowable stresses be reduced 1/6th, and the modules of elasticity be reduced 1/10th for fire-retardant chemical pressure-impregnated plywood below normal design standards for untreated plywood. In addition, the fire retardant treatment of wood does not prevent the wood from decomposing and charring under fire exposure. The rate of fire penetration through treated wood is approximately the same as for untreated wood.

Fire retardant coatings are also used to reduce the surface flammability of wood and laminated wood products. Prior art disclosed four major types of fire-protective paints: (1) Ablative; (2) intumescent; (3) sublimating and gas- or vapor-producing; and (4) "ceramic" or inorganic. Ablative coatings are generally thick and heavy and costly, thus limiting their application to special uses. Intumescent coatings are generally soft and easily abraded, and have limited duration of protection--10 to 15 minutes--when exposed to fire before the delicate char cracks and is eroded away, leaving the substrate unprotected. They cannot be used internally ("submerged") between the wood plies since their mode of action is inhibited by the glue line bond strength. The adhesive prevents their normal 10 times (or greater) expansion. Sublimative or gas-producing coatings have even shorter duration of protection--after their snuffing gases have been produced and wafted away, they are left without a protective mechanism. They have found their major use in the treatment of fibers and clothing where only short-term exposure to flame is a major requirement. Ceramic coatings, as the name implies, require high temperature curing in order to form the ceramic bond, which many structural and building wood components cannot withstand. Inorganic coatings, such as Portland Cement, gypsum, calcium aluminate cement, phosphate-bonded cements, silicate coatings, high temperature silicones, and magnesium "oxychloride" and magnesium "oxysulphate" cement coatings have all been proposed.

H. Ellis, in a companion U.S. Letters Patent application, now pending, entitled "FIRE BARRIER: I. COATINGS", of which this disclosure is an extension, in part, discloses seven type formulations for inorganic coatings, consisting of synergistic combinations of two and three non-combustible inorganic bonding systems which efficiently prevent flame spread on the surface of susceptible substrates, and at the same time prevent flame from penetrating through the substrate.

The American Society for Testing Materials (ASTM) has issued a series of Standard Test Methods for evaluating the protective abilities of fire retardant coatings and constructions (E-119, "Fire Tests of Building Construction and Materials"; D-3806, "Small Scale Evaluation of Fire Retardant Paints (2-Foot Tunnel Method)"; E-286, "Surface Flammability of Building Materials Using an 8-Foot Tunnel Furnace"). E-84, a 25-foot "Tunnel" test (the most widely accepted test), has its counterpart in Underwriters' Laboratories (UL) #723; and the National Fire Protection Assocation (NFPA) #25 test methods. The tests determine the protection a coating affords its substrate; and the comparative burning characteristics of coatings by evaluating flame spread over the surface when ignited under controlled laboratory conditions. The different tests are not comparable, although the results are correlatable. None of them describe or apprise the fire hazard of assemblies under actual fire conditions, but the results are usable as an element in fire-risk assessment.

The tests also give the basis for classifying the performance of the coatings based on their Flame Spread Index Values into Class A (or I), Class B (or II) and Class C (or III), used by most Building Codes. Index value classification depends on flame spread of the candidate fire retardant coated products or materials of construction, in comparison to known standards. Asbestos-cement board is arbitrarily rated at 0 flame spread; Red Oak lumber is rated at 100. Materials with the lowest rate of flame spread (0-25) are classified as Class I, and are permitted in areas where fire hazard is most severe (such as exitways or unsprinklered buildings); materials with rates from 26 to 75 are Class II, and are permitted in areas of intermediate severity; materials rated 76 to 200 (such as Ps-1 uncoated construction plywood) are Class III, and are permitted in most rooms when suitably protected by other design factors. Hospitals, institutions and buildings of public gatherings require Class I products or fire-protective barriers. There is also a Class D (or IV), with Index Values of from 201 to 500, and a Class E (or V), with Index Values of over 500.

Most untreated softwood plywoods and decorative hardwood plywoods fall into Class III, cut this varies, depending on wood species, thickness, and glue type. These types of plywood constitute by far that great majority of laminated wood boards. Therefore, they are accepted for interior finish only, and not for exitways. If they could be upgraded in Class I economically, the range of their potential application would be extended considerably. The market for these products would also be considerably enlarged by the resulting savings in associated construction costs.

In evaluating fire safety of laminated wood structures, two types of fire propagation must be considered: Flame spread and fire resistance.

Flame spread--a measurement of the maximum extension of the flame on a sample--is a property of the surface material once fire has started, not of the structure. (Under test conditions, it is a ratio, not inches, versus the arbitrary standards). Flame spread is an indication of how fast fire can extend over a room's surface, but says little about the third dimension of fire spread, i.e. the depth of fire penetration through the substrate. Most of the fire retardant coatings influence only the flame spread rate, that is, they function to lower the lateral extension of the flaming surface per unit of time, but have little influence on the "Burn-Through Rate", since the coatings are a surface application. One exception to this statement is the disclosure in the Ellis patent application referred to above, in which the surface coating prevents the ignition of the substrate so that a carbonized insulative char forms under the coating--pyrolysis without flame--and in an area restricted to the immediate zone of the initial fire. Fire Endurance (or Resistance) is "The resistance against fire penetrating a wall, floor or roof, either directly or through a high rate of heat transfer that might cause combustible materials to be ignited on the `cold` side of the wall away from the actual fire." In short, it indicates the ability to resist "burn-through" and measures the containment of the fire within a room or building. The standard test for measuring fire resistance is ASTM E-119; and resistance rating is expressed in the hours or the minutes that the construction withstands the test.

The disclosure presented herein defines a novel method for increasing significantly the "fire endurance" of laminated wooden structures, and for decreasing significantly (if not completely preventing) flame spread in these laminated structures by interposing a non-flammable high temperature resistant coating, in a variety of arrangements, between the plies of the laminate, and, optionally, on the surface of the "plywood."

The prime objective of this invention is to impart to the various types of laminated wood products generally termed "plywoods", which term includes panelling, doorskin, and other structures composed of multi-layered veneers, sufficient fire retardency so that by themselves, without the need of other structural barriers, they form a fire barrier to both surface flame spread and burn-through (fire endurance), in short, develop a fire-barrier plywood which meets the requirements for a Class I (or A) fire retardancy classification, and thus permit the use of these materials in those situations now barred to their use by fire hazards or construction code restraints. By obviating the necessity of using supplemental fire barriers, such as claddings, gypsum wallboard, and other structural alternatives, the use of fire barrier plywoods could lower construction costs, simplify design, decrease weight, add usable space to a structure, and lower insurance rates.

Another objective is to overcome the deficiencies of the currently-used alternative systems for imparting fire retardancy to plywoods viz. pressure-impregnating the entire mass of the wood with fire retardant salts and resins, or surface coating alone with fire retardant paints and mastics. The use of pressure-impregnated salts is costly, more than doubling the price of untreated plywood; reduces the allowable stress limits by 16% and modulus of elasticity by 10%; may be corrosive to metals, and is water-leachable, thus restricting their outdoor use. Surface coatings are effective for only limited periods of time when exposed to fire; are usually soft and easily abradable; are usually applied on site after the installation, hence only cover the exposed, easily reached areas; and are effective mainly against surface flame spread for limited periods of time. Their efficiency in preventing burn-through is highly variable, and of relatively short duration.

Still another objective is to develop a fire barrier coating which is relatively inexpensive, simple to apply, and which requires little modification of the procedures or equipment presently used in the manufacture of plywood.

Another objective is to develop a fire barrier coating which in itself has sufficient adhesion to the various woods used in plywood construction to substitute for the currently used bonding glues, and/or be compatible with those glues, so that they may be applied together utilizing existing equipment.

Still other objectives will become evident as the teachings of this invention are disclosed.

This invention aids in fulfilling these needs in the art. This disclosure crosses over and unites four major technologies: as a composition of material, it falls into the category of air-setting "hydraulic: binding cements, belonging to the group of caustic magnesia cements, generally called Sorel Cements, and more specifically as magnesium oxychloride cements; as a product it falls into the category of a coating or paint; as a use or application, it falls into the category of fire-control devices; and its associated characteristics permit it to be used for--and thus categorized as--a material of construction.

Specifically, presented herein is a description of a novel magnesium "oxychloride" cement coating, co-bonded with high alumina calcium aluminate cement and a colloidal silica, which overcomes the numerous deficiencies in the prior art coatings of this type, and the application of this coating, either alone, or in combination with a non-woven spunbond polyester fabric or a non-woven or woven glass fabric in the form of a thin laminate ("composite coating") to fire susceptible substrates to form a fire barrier capable of preventing ignition and flame spread of the structure. The coating is non-combustible in itself, and has adequate tensile and cohesive strength to maintain its integrity for the duration of the fire. It has excellent adhesion to a wide variety of substrates, and can be included as a structural component in the construction of gypsum wallboard, plywood, and other types of laminates.

Prior art magnesia cement compositions consisted of: (a) "binder ingredients"; (b) the "activator" ingredients; (c) "fillers"; and (d) "adjuvants".

(a) The basic binding ingredient, a "cement", consists of a lightly calcined (heat-activated) magnesium oxide (MgO) which, when combined with the aqueous MgCl.sub.2 or MgSO.sub.4 salt solution, crystallizes into a network of interlocKing crystals (and other ionic bonding systems) that gives strength to the system by bonding the fillers contained therein. In the present disclosure, high-alumina calcium aluminate cement and colloidal silica are added as compatible co-binders, imparting to the magnesium "oxychloride" cement important beneficial characteristics, as detailed below.

(b) An aqueous "activator" fluid (called a gauging liquid) which is a salt solution of magnesium chloride (MgCl.sub.2) or magnesium sulphate (MgSO.sub.4), or a combination thereof. Both salts are hydrated salts (6H.sub.2 O and 7H.sub.2 O, respectively) easily soluble in water. Practice shows that to be effective in bringing about the setting of the MgO, the solutions must be concentrated (40% to 60%), generally expressed as a specific gravity, in either specific gravity or in degrees Baume units (sp. gr. 1.22 to 1.26, or, 26.degree. to 32.degree. Baume). These salts are best added in a stoichiometric ratio for the reaction, which does not leave an excess of salt that can effloresce to the surface or leach out in water or humid conditions; or react with the CO.sub.2 in the air to form compounds which cause a disruption of the structure by occuping more space than originally occupied in the initial setting of the cement. In the present disclosure, the salt solutions are used as the common hydrating agent for both the MgO cement and the calcium aluminate cement. As demonstrated below, the use of this solution to hydrate the monocalcium aluminate obviates the necessity of maintaining an optional stoichiometric ratio of salt to MgO.

(c) Fillers--any of a wide variety of types of inert materials, such as sand, gravels, crushed rocks, silica flour, pumice, vermiculite, volcanic ash, perlite, wood shavings, sugarcane bagasse, asbestos, mineral fibers, etc. are added to impart special properties or lower the overall cost of the concrete.

(d) Adjuvants--a wide variety of accessory chemicals have been added to the slurry mix to bring about specific changes in the product, e.g. to make the concrete harder, to speed up the setting, to slow down the setting, to prevent the magnesium salt efflorescence, to waterproof the composition, to prevent shrinking (and hence shrinkage cracks), to prevent the absorption of atmospheric moisture, to decrease "harshness" in the mix and impart spreadability (smoothness), to decrease water or fluid demand, and etc.

As in most hydraulic cementitious systems, the binding ingredient or cement is intimately blended with the dry filler materials and any dry adjuvants and stored until needed. The dry ingredients are then mixed with the liquid ingredients, and the slurry poured or placed. The chemical reaction takes place, which "sets" the cement and forms the concrete. The mass, which is fluid at first, gradually thickens, becomes less "workable", forms a gel (initial set) which gradually and continuously becomes harder over a four-hour period. The hardening continues and strength increases for several days so that by the third to fifth day maximum strength has been reached.

The coating, subject of this disclosure, consists of a two-part mixture: Part "A", the dry powdered binder and fIller components; Part "B", the gauging or activator liquid components. The binder contains magnesium oxide, with specific characteristics, and high alumina calcium aluminate cement, along with fillers (silica flour, titanium dioxide, and, optionally, expanded vermiculite and alumina trihydrate). The liquid component consists of magnesium chloride solution of 1.26 specific gravity, colloidal silica dispersed in dimethyl formamide (DMF), and an anionic surfactant, sodium tetradecyl sulphate. As is common practice in the art, part "A" is added to part "B", with adequate mixing to form a slurry of the proper consistency for painting or coating, and the slurry permitted to set into a hard monolithic refractory layer. When used as a composite or laminated coating, either a non-woven spunbond polyester fabric, or a non-woven or woven fiberglass fabric is interposed between two thin layers of the coating. The coating or composite coating is applied to the desired substrate. When exposed to flames of up to 2000.degree. F., the coatings retain their integrity and prevent the flame front from advancing. They themselves do not burn, and they prevent the direct ignition and flaming of the substrate for as long as 30 minutes--sufficient time, in most cases, for the original fire source to die out or bypass the protected area. Although they are efficient thermal insulators, as long as any water of hydration remains in the coating they are not designed to be long-term heat barriers, hence, after varying periods of time when wood substrate temperatures reach 700.degree. to 800.degree. F., wood substrates will char or carbonize, but do not ignite or burn. Similarly, plastic substrates, foam or sheet or block depending on their particular time-temperature response, will char but not ignite, and thus do not contribute to the advancing flame front.

While the excellent high temperature and fire-resistive properties of coatings made from magnesium oxysaklts have long been recognized, relatively little advantage has been taken of these in attempting to adapt thin coatings to commercial use as fire retardants or fire barriers, due, in large measure, to major deficiencies in physical properties. There is widespread use of these materials in thick layers, 1/2" to 4" or even 6" in thickness, as insulative and fire-barrier mastics, grouts and cements. Although these thick insulations are frequently termed "coatings", this disclosure treats only with the use of magnesium "oxychloride" concrete as a thin paint or paint laminate.

The coatings and coating laminates described herein have the function of preventing the spread of fire by defensively protecting combustible substrates from igniting, preventing the flame from spreading and extending its area of destruction along the substrate, and thus prevent the accumulation of the combustible (and highly toxic) gases in the stucture. "Preventing" is an absolute term which would signify "fire-proofing." However, "fire-proof", "non-combustible" and even "fire-resistant" are all ambiguous terms unless qualified by a description of the fire conditions under a recognized test method. The American Society for Testing Materials (ASTM) does not recognize the term "fire-proofing" since nothing, under the most severe conditions, cannot be fired. Commercially, the term is used to indicate the slowing down or delaying--and occasionally preventing--of the ignition and spread of flame of the substrate underlying the protective coating or cladding.

Currently available commercial "fire-retardant" coatings, such as intumescent-type fire-retardant paints which when exposed to high temperatures form an insulating blanket, are intended to function as their designation implies, to delay ignition for about 10 to 15 minutes, and reduce the surface burning rate of interior combustible finish materials. They do not render the underlying materials non-combustible, but serve only to retard ignition for a brief period of time. That delay furnishes an extra measure of escape time, affords time for implementation of other fire-fighting methods, and reduces structural damage. In contrast, the fire-barrier coatings, subject of this disclosure, are designed to prevent ignition and fire-spread of the underlying substrate by interposing an incombustible layer between the flame and the substrate. Fire-retardanct coatings are rated as to "flame spread" when tested under standard widely accepted test conditions (e.g. ASTM E-84, Underwriters' [UL] #723, National Fire Protection Association [NFPA] #25, or ASTM D-3806).

In development work, for comparative evaluation of candidate coating formulations and assessment of the numerous variables which enter into such formulations and compositions, ASTM Standard Test Method D-3806, "Small-Scale Evaluation of Fire Retardant Paints (2-foot Tunnel Method" is preferred to the Much more elaborate and time-consuming 8-foot tunnel test (ASTM E-286) and 25-foot tunnel test (ASTM E-84). D-3806 "determines the protection a coating affords its substrate and the comparative burning characteristics of coatings by evaluating the flame spread over the surface when ignited under controlled conditions in a small tunnel. This establishes a basis for comparing surface burning characteristics of different coatings without specific consideration of the end use parameters that might affect surface burning characteristics under actual fire conditions." In addition to the experimental flame spread rate, the test determines the weight of the panel consumed, fuel contributed to the fire, time of afterflaming, smoke emission and other characteristics. The Standard measures and describes the properties of the coatings in response to heat and flame under controlled laboratory conditions, but it warns that it should not be used to describe or appraise the fire hazard of assemblies under acutal fire conditions. However, the test may be used as an element in fire risk assessment.

The test consists of impinging directly onto the surface coating a 2000.degree. F. gas flame supplying heat at 5085 BTU's per hour for 4 minutes, while the 2-foot long by 4-inch wide test panel is suspended above the burner, coated surface facing downward, at an upward angle in a chamber. The angle of the test panel and the draft of the chamber encourages any flame in the coating and substrate initiated from the burner to proceed upward along the substrate. The extent of this flame front is measured and compared with the known substrate standards (0 and 100).

"Flame spread" is the ratio of the measured extent of the burning of the substrate protected by the candidate paint, contrasted with a known incombustible rated standard substrate (cement--asbestos board) as 0 and an unprotected highly combustible wood substrate, such as Red Oak, with a rating of 100.

More specifically, "Flame spread" is a ratio, not inches. The maximum extension of the flame on the sample is measured. The maximum extent of the flame (flame spread) of the asbestos board is considered as 6.5", with an adjustment factor of 4.45. The length of the asbestos board flame is subtracted from the flame length of the test specimen, and the difference is multiplied by the adjustment factor to yield the flame spread of the test sample. For example, in Table II, Test j.2 of the Type II coating alone on isocyanurate foam board, the flame spread on the speciment was 7" long, minus 6.5" length for asbestos board, which yields an 0.5" difference, and when multiplied by the 4.45 factor, equals the reported flame spread of 2.2.

"Percent Loss" is the total weight of the specimen prior to the test minus the total weight of the specimen after the test, .times.100. This includes loss of moisture due to steam vaporization, as well as losses due to combustion. In the case of the coatings, subject of this disclosure, most of the loss in weight when applied to susceptible substrates is due to this water-of-hydration loss from the coating under the area of flame impingement, without significant deleterious effects on the coating.

Twenty readings are taken in determining "smoke density," and summed. This is divided by 2,000. The value of Red Oak, considered a standard with a value of 1924, is subtracted, and the resultant is multiplied by 100 to convert the specimen's value into a percentage which relates to the value of the Red Oak as 100% and asbestos board as 0%.

Flame spread is considered the main factor associated with testing fire retardant coatings. The "fire-retardant" coatings currently available are rated as low as 10, or as high as 60 to 70. The lower the rating, the longer the coating will retard flames. In the higher ranges, e.g. 60 to 70, the coating affords little or no fire protection.

The coatings and laminates described herein, which are rated 0 flame spread (as well as 0 fuel contributed to the fire and 0 smoke developed) are a significant advance in the art of protective coatings in that they are "fire barriers," not merely "fire-retardants." They are designed to interpose a noncombustible, coherent, durable layer between a flame and a combustible substrate, and thus prevent the ignition of the substrate and and the subsequent flame-spread along the substrate. Under operational fire conditions, following a period of time, the initial fire source will have exhausted its local fuel supply and burned itself out.

TABLE I lists the several formulas for the fire barrier coatings developed in this disclosure. The numbers in parentheses listed below correspond to the numbers in the "Materials"0 column of TABLE I.

The first five items of part "A" are all finely comminuted solid powders, while the sixth is a granulated solid of -16 mesh. Each material is first dried in a hot air oven at 200.degree. F., mixed together, and, except for item (6), intimately blended by mulling. The surfactant (9) is added drop-by-drop to the MgCl.sub.2 solution (7) and the colloidal silica (10) is added slowly to (7)+(9). This causes an opalescence in the clear solution, but no precipitate. The reaction is highly exothermic. There is a marked increase in the viscosity of the solution, which decreases again upon standing. The mixed powders, (1)+(2)+(3)+(4)+(5)+(6), are then added to the liquid phase (part "B")--(7) or (8)+(9)+(10), slowly, with constant mechanical stirring, so that each particle is completely wetted out by the liquid. A smooth suspension of paint-like consistency results. This is applied, while still fluid, by any common method of coating application (brushing, rolling, spraying, etc.). Pot life is short--20 to 30 minutes, with the paint gradually thickening until it becomes a soft gel. The gel gradually hardens with initial set at 1 to 11/2 hours, and final set in 2 to 4 hours. Pot life and set time can be controlled by varying the solids to liquid ratio--most easily by increasing (or decreasing) the amount of the fluid phase. Final set time of the coating is significantly decreased by elevating the cure temperature to 100.degree.-120.degree. F. In these formulations, the final set of the coating, that is, the hardness and other surface characteristics, is relatively insensitive to the solids:liquid ratio. Upon drying, the coating is a brilliant white; porcelain like; refractory hard (Mohs 5.5-6.0); and crackfree, with essentially zero shrinkage. After 48 to 72 hours of cure time, the coating is highly resistant to direct flame impingement, and withstands 1900.degree. to 2000.degree. F. without cracking for 30 to 45 minutes and longer, when suitably supported on a fibrous matrix (non-woven spunbond polyester, woven or non-woven fiberglass fabric, fiberglass fibers, etc.) which absorbs the thermal stresses of differential expansion and crystalloraphic alterations accompanying the firing.

(1) Three types of MgO (magnesia) have been used in the principal binding system which differ in reactivity as a result of originating from different sources and receiving differing heat treatments. In the formulations listed in Table I, "OXYMAG" grade from Basic Chemicals, Inc. was used--a calcined magnesia ore from Nevada, with 1.5% lime (CaO), 95% MgO, 3% loss on ignition, and with 99.6% passing 200 mesh. It is treated to eliminate any tendency to expand due to the free lime. Its bulk density is twice that of seawater-derived magnesias, but has great binding power for fillers and high reactivity with MgCl.sub.2 and MgSO.sub.4. Alternatively, both "MAGOX" 98 HR and "MAGOX" 98 LR magnesias (also from Basic Chemicals, Inc.) have been used with equal success in the fire barrier coating, but for different applications. These "MAGOX" materials are calcined magnesia from sea water (brine) sources. HR refers to high reactivity; LR refers to low reactivity. Both contain 97% to 98% MgO, 1% lime (CaO) and only 0.15% Fe.sub.2 O.sub.3, so that they are white. Due to the method of calcination, the specific surface area of the HR grade is 68 M.sup.2 /gm, while the LR grade shows 36 M.sup.2 /gm. Both have a median particle size of less than 2 microns. Both have been used successfully--the only modification required is the quantity of MgCl.sub.2 solution necessary to make a paint of the required consistency. Since their densities are lower than the "OXYMAG", they bulk more (30 lbs/ft.sup.3 vs. 48 lbs/ft.sup.3) and a greater volume of material, with a larger surface area, is required to yield the weights shown in the type formulation. This requires approximately 20% to 25% greater fluid volume to wet out and produce a fluid paint. Where open pot time is more important than rapid initial set, the LR grade can be used. On a volume-of-paint basis, the MAGOX grades are less expensive than the OXYMAG grade. Other producers, such as Martin-Marietta Corporation, produce a variety of sea water (brine) magnesia grades (which correspond to the HR grades of Basic), "MAG-CHEM 40", a very high reactive grade; "MAG-CHEM 30", a high reactive grade; and "MAG-CHEM 10", a low reactive grade.

The order of blending and mixing is of some importance. The solids are added to the fluid mixture with constant stirring, so that each powder particle is wetted out and no lumps remain. After mixing is complete, the paint, using the HR grade of MgO, has a pot life of 15 minutes before thickening to a point which makes brushing or rolling difficult. Using the LR grade, the pot life is extended to 20 to 25 minutes, with the final setting extended to 3 to 4 hours.

(2) High Alumina Calcium Aluminate Cement: Both ALCOA (Aluminum Company of America) CA-25 (High Alumina Refractory Cement) and LONE STAR LaFARGE, INC. SECAR 71 or 80 have been used as a co-binding system with the magnesium "oxy" cements, with equal success. The materials are equivalent. The principal constituent of these HAC (high alumina cements) is mono-calcium aluminate, (CaO+2.5 Al.sub.2 O.sub.3). They are hydraulic cements containing a high percentage of alumina (Al.sub.2 O.sub.3), approximately 80%, and calcium oxide (CaO), approximately 17%-18%. (In contrast, Portland cements are calcium silicates; Plaster-of-Paris cements are calcium sulphates). Water is the generally-accepted "activating" agent, which causes the essentially anhydrous mixture (only 1.8% loss on ignition) to form a strong interlocking crystalline bond in setting. Setting is rapid, initial set taking place in 1/2 to 1 hour, and final set in 4 to 5 hours, and maximum strengths are achieved in one day. Useful service temperature of the set concrete is 3200.degree. F. It is widely used in castable refractories. The HAC lose strength at intermediate temperature ranges of 900.degree.-1200.degree. F., but achieve much higher strengths through ceramitization at the 1500.degree. F. and higher temperature ranges. They resist the corrosive action of calcium and magnesium sulphates and chlorides. The cements are white; they are finely ground, with a specific surface of about 10,000 cm.sup.2 /gm. (Blaine). Compressive strength of a 1:3 mortar at 7 days is about 9400 psi, while "neat" cements, depending on the cement:water ratio, can achieve 8,000 to 18,000 psi.

The strengths developed with these high alumina cements are significantly greater than when lower grade calcium aluminate cements are used, e.g. U.S. Steel Corp., Atlas Cement Divisions's "LUMNITE" cement, with 47.0% Al.sub.2 O.sub.3, 34.3% CaO and 7.4% Fe.sub.2 O.sub.3, or "REFCON" cement, with 58.5% Al.sub.2 O.sub.3, 33.5% CaO and 7.4% Fe.sub.2 O.sub.3, both of which are brown in color; or Lone Star LeFarge "FONDU" cement, with 38%-40% Al.sub.2 O.sub.3, 39% CaO and 15% Fe.sub.2 O.sub.3.

The calcium aluminate cement is not to be confused with alumina (Al.sub.2 O.sub.3), frequently called out as either a high strength "filler" in cemented products, or as a "binder" in refractories and ceramics. In the latter case, it functions as a high-strength and high-temperature binder only after high-temperature fusion in a furnace.

In the listed formulations the "water" of hydration for the monocalcium aluminate is supplied by the magnesium chloride (or magnesium sulphate) solution. Magnesium hydroxide, the principal ingredient of the "mag-oxy" cements, according to one current theory, retards the setting of the alumina cement; but the magnesium sulpate high concentration accelerates the set. Similarly, magnesium chloride in low concentration retards the set, but in high concentration accelerates the set. In the formulations, listed in TABLE I, the entire body of cementitious materials develop and initial set in 11/2 to 2 hours, and a final set in 3 to 4 hours.

The set cement is highly resistive to the leaching action of water, and is impervious to the corrosive action of sulphates and chlorides (both of which have severe effects on Portland Cement Concrete).

The addition of the high alumina calcium aluminate cement as a co-binder is completely compatible with the MgO binder system, and adds considerably to the hardness and abrasion resistance of the coatings, as well as improving resistance to leaching of the magnesium salts in the set cement by rain, thus improving the waterproofness of the product. In addition, it effectively prevents the efflorescence of the dried magnesium salts and the syneresis of the soluble salts.

(3) Silica flour. A standard commercial commodity available from many producers, consists of finely ground (smaller than 200 mesh) silica sand (silicon dioxide, SiO.sub.2) low in iron, glistening white. It is used as an incombustible low-cost filler. It affords rheological (flow) control to the fluid paint, while supplying hardness and abrasion resistance and innumerable sites for crystal stress relaxation in the cured films, especially at elevated temperatures when the silica goes through its own phase changes.

(4) Titanium Dioxide (TiO.sub.2). A standard commercial commodity available from many producers. This ingredient, used at a relatively low concentration, functions as a whitening agent--not only decoratively, but more importantly as an infrared radiant heat reflector. TiO.sub.2 has great "hiding power," and imparts a brilliant white to the product, obviously of very low emissivity (.alpha.), and high spectral reflectance (.xi.) in visible light, but also at infrared wavelengths. Since burning gases and flames emit a high percentage of their heat as radiant heat, the low emissivity and high reflectance of these coatings contribute to lowering the surface temperature on the coating and its underlying substrate.

(5) Alumina Trihydrate (Al.sub.2 O.sub.3.3H.sub.2 O) with approximately 35% bound water, is widely used as a fire retardant filler for plastics. Its mode of action is similar to that of the hydrated "magoxy" cements. The hydrate retards flame by serving as a "heat-sink," absorbing heat in the process of decomposition, and liberating water vapor (steam) which serves as a gaseous blanket protecting the substrate. The grade used in the type formulas was SB 331 from Solem Industries, Inc., Norcross, Ga., with a mean particle size of 10 microns, and TECHFILL A-112, with a mean particle size of 12 microns, produced by Great Lakes Mineral Co. of Troy, Mich. Both products are brilliant white with high specular reflectivity. The alumina trihydrate is used to substitute for the silica flour as a filler. While the coatings it produces are not as hard as the silica flour coatings, this decrease is not significant, and is compensated for by the additional water of hydration it adds.

(6) Expanded Vermiculite. Because of its low "K" factor and non-combustibility, expanded vermiculite has long been used for pour-in insulation, insulating concrete, and a wide variety of high-temperature insulations. Its fusion point is rated at 2200.degree. to 2400.degree. F., and it has a specific heat of 0.20. When added as a filler to the part "A" binder components, while its conribution to the heat-barrier properties of the coating is not insignificant (since it occupies, by volume, almost 20% of the fluid paint, and 25% of the dry paint) its main function is to serve as a stress absorber. The grade of expanded vermiculite used in Type Coating IV is size No. 3, with a density of 5 to 6 pounds per cubic foot. This commercial grade is screened through a -16 mesh sieve, and only the material passing through the mesh is used. Domestic U.S.A. vermiculite tends to be dark brown in color, with South African grades tend to be a light cream color and require lesser TiO.sub.2 to hide the off-color it imparts to the coating. In plant manufacturing, the particles should be slightly wetted with the MgCl.sub.2 (or MgSO.sub.4) solution prior to adding to the total mix, but for field application, it is added as a dry ingredient to the rest of the powders in part "A", and ennobled by these powders in the mixing. The particles disperse quite readily in the mix, and there is little tendency to segregate. In the dry paint, they impart to the coating a "pebbly" surface, which contributes to the reflectivity of the coating. Thus, in painting the particles are evenly dispersed.

During exposure to flame at 1800.degree. to 1950.degree. F., the crystallite components of the "oxychloride" cement undergo phase changes which result in compression stresses on the integrity of the film. The vermiculite particles are not subject to these phase changes at those temperatures, and retain their integrity. They are thus capable of serving as stress-relaxation points. Due to their lamellar structure with air voids between the layers, the particles are compressible, and thus take up the stresses of the thermal expansion which accompanies the phase changes of the cement components. This prevents the "popping" (spalling or decrepitating) and subsequent cracking of the coatings common to many rigid and brittle "ceramic" films. As noted in the examples presented below, not all formulations require the inclusion of vermiculite particles, and they are not normally added to those coatings included in the laminates made with fabric interlays, which take up the stress points in a similar manner.

(7) MgCl.sub.2 Solution in Water--1.26 sp. gr. (30.degree. Baume). Dow Chemical Co., Tech, Flake Grade, supplied in a dry form as 99.0% MgCl.sub.2.6H.sub.2 O. The material is a thin white opaque flake, easily soluble in water, but high deliquescent. The hexahydrate material contains 46% to 52% MgCl.sub.2. The hexahydrate is a commodity article of commerce supplied by many chemical manufacturers. Because of the hygroscopicity of the hexahydrate, it cannot be packed dry with MgO and other solid materials into a one-package magnesium oxychloride cement product. Formerly covered by ASTM Specification #C 276 (now discontinued), the Dow Chemical Co. Product Data and Sales Specification is generally accepted as the standard. The MgCl.sub.2.6H.sub.2 O is soluble to the extent of 155 mg/100 cc distilled water at 20.degree. C. A 28% solution based on MgCl.sub.2 yields 30.degree. Baume at 60.degree. F. In practice, 1500 gms of the hexahydrate salt dissolved in 1000 cc of city tap water yields 1800 cc of a 1.26 sp. gr. (30.degree. Baume) solution, with a pH of 6.0. It is recommended that the solution be allowed to "mature" for 24 hours prior to use; and it is stable indefinitely, after preparation, in closed containers. Practice shows that a minimum concentration of a sp. gr. of 1.18-1.20 is required to successfully bring about the SOREL cementitious reaction. The more concentrated the solution, the harder the final product. Too great a quantity of MgCl.sub.2 leads to efflorescence under humid conditions. Both the salt and solution are corrosive to ferrous metals, so precautions must be taken in the preparation of the salt solution. Also, because of the hygroscopicity, the flake salt must be kept from exposure to the air.

This solution, in the proportion disclosed in the formulas listed in TABLE I, is the "activator" or gauging solution, used to wet out and hydrate the MgO and the calcium aluminate cement, and is the principal liquid vehicle for the aqueous-based fire-barrier paint.

(8) Alternatively, the gauging solution may be MgSO.sub.4 at a specific gravity of 1.25, or 29.degree. Baume. This is the common Epsom salts, (MgSO.sub.4.7H.sub.2 O, a flaked, colorless crystalline material), a commodity product, produced by many chemical manufacturers. Formerly covered by ASTM Specification #C-277, this specification is now discontinued. The hydrated salt is readily soluble in water and highly endothermic--chilling the water in the process of dissolving. 500 gms of the salt, dissolved in 1 liter (or 4.169 lbs in 1 U.S. gallon) of water yields a sp. gr. of 1.25 or 26.degree. Baume. This makes a net 41% solution. In tap water, the resulting pH is 5.3. The reaction between the MgO and MgSO.sub.4 is well known in the art of making fire-resistive mastics (and stucco), but the resulting "concrete" is noticeably softer than when made with MgCl.sub.2. Also, the product shrinks considerably more, foring surface and deep cracks. This tendency to shrink is counteracted by the calcium aluminate cement and/or by the addition to the sulphate solution of phosphate salts or MgCl.sub.2 salts.

Because of the lower hygroscopicity of the MgSO.sub.4.7H.sub.2 O, it is frequently packaged with the MgO into a one-component oxysulphate cement base requiring only the addition of water. However, because the preferred formulas listed in TABLE 1 contain other fluid ingredients besides the salt solution, they require a 2-component packaging system.

A third alternative fomulation calls for the mixing of the above-formulated magnesium chloride solution with the magnesium sulphate solution in proportions of 1:1, 2:1, or 3:1. This formulation hardens the product, and reduces the tendency to shrink and crack.

(9) Niaproof Anionic Surfactant 4, supplied by Niacet Corp., Niagara Falls, N.Y. 14302, is a 27% aqueous solution of sodium tetradecyl sulphate, [C.sub.4 H.sub.9 CH(C.sub.2 H.sub.5)C.sub.2 H.sub.4 CH(SO.sub.4 Na)CH.sub.2 CH(CH.sub.3).sub.2 ]. Whereas many surfactants (dispersants) were precipitated by the high magnesium salt solution, or cause a coagulation or gelling of the colloidal silica, or cause objectionable foaming during the mixing of the paint, NAS 4 met all compatibility requirements. Its function is to reduce the surface tension of the liquid phase ("B") of the formulation, and thus enable it to wet out the solid particles more efficiently. It acts as a dispersant for the powders in the formula. As such, it reduces the quantity of fluid required to make a satisfactory paint viscosity (the "water of convenience"), and thus yields a stronger paint film, with less tendency to shrink-crack (craze) while excess fluid evaporates.

(10) Colloidal Silica, or slica sols. NALCOAG 2325, a 35% colloidal dispersion of silica in the non-aqueous polar solvent dimethyl formamide (DMF), from Nalco Chemical Co., Chicago, Ill.

The particles of silica are discrete, non-crystalline, spherical, submicron (average particle diameter of 20 millimicrons; average surface area 150 M.sup.2 /gms). The dispersion contains 35% silica as SiO.sub.2. When diluted with water, the DMF dispersion yields a pH of 5.

The dried particles of the silica develop strong adhesive and cohesive bonds. Under high temperatures of up to 200.degree. F., they show little crystallographic transformation. In the Type I and Type VI formulas in which used, the colloidal silica performs the function of a densifying agent; a waterproofing agent; a hardening agent; a surface gloss-producing agent ("porcelainizing"); a co-bonding agent along with the "magoxy" and calcium aluminate cements; and an adhesive aiding bonding of the coating to non-porous substrates. Colloidal silicas have been widely used as binders for fillers in the production of high-temperature mineral and inorganic fiber shapes (principally thermal insulation, e.g. APOLLO spacecraft heat shields); as frictionizing (anti-skid) agens in coatings; as catalysts; as anti-tacking agents in polyolefinic films; in protecting coatings to increase abrasion resistance of decorative laminates; as polishing agents (e.g. silicon wafers) etc.

By far the most commonly used type of dispersion is the aqueous, negatively charged sodium ion-stabilized form, usually supplied as 15%, 30%, 40% or 50% SiO.sub.2 concentration, depending on the particle size of the silica. All are highly alkaline. Other types are stabilized with ammonium ion or chloride ion (positive particle charges); and instead of a suspensoid in water, are dispersed in acidic solutions or oils, or coated with aluminum acetate, etc. Colloidal silicas are produced by DuPont Industrial Chemicals Division of E. I. DuPont de Nemours; by the PQ Corp. (formerly Philadelphia Quartz Co.); by Nalco Chemical Co., and by several others in the U.S.A., as well as in Japan, Germany, England, etc.

While all types of colloical silicas were not tested in the development of the listed formulations, five types were tested, with sufficiently differing results (mainly, compatabilities amongst the various other fluid and powder ingredients) to lead to the selection of the DMF as the preferred type of dispersion. When 30% and 40% alkaline aqueous dispersions of the commonly used sodium ion stabilized form are added to the high salt content MgCl.sub.2 or MgSO.sub.4 solutions, the silica precipitates out and leaves a soft semi-solid gel, containing numerous silica clumps.

Alumina acetate coated silica sol (NALCOAG ISJ-613) is aqueous based and completely miscible with the liquid phase of the listed formulations. Containing 3% colloidal alumina, plus 19% colloidal silica, ISJ-613 is acidic, and has a pH of 4, the Al.sub.2 O.sub.3 and SiO.sub.2 particles are positively charged, the ISJ-613 yielded excellent porcelain-like hard waterproof coatings, but the paint is "harsh", i.e. while the fluid paint is smooth, it shows poor flow and brush-mark fill in. It also reduces the pot life of the paint. However, it markedly accelerates the initial and final setting of the coating to as low as 1 to 11/2 hours.

In the type formulation I, calling for the use of the DMF dispersed colloidal silica, it may be noted that the addition of the NALCOAG 2325 to the magnesium chloride solution generates considerable heat--a highly exothermic reaction--but while it causes an opalescence in the salt solution, there is no precipitation. There is an increase in solution viscosity, which is compensated for by the addition of a surfactant, NAS 4.

Colloidal silicas should not be confused with other types of silicas or silicates. They are significantly different from soluble silicates (e.g. sodium, potassium, lithium or other alkali metal silicates); organic silicates (e.g. ethyl silicate); fumed silica ("Cab-o-Sil", "Aerosil"); microfine silicas; amorphous silicas; silica gels; silica flour or silica powder.

To accomplish the objectives listed above, the basic SOREL cement reaction was modified at several critical points. In order to increase the fluidity of the paint so that it could be brushed, sprayed, or rolled onto the substrate, with good flow and "fill-in" to eliminate streaks, the ratio of activator or gauging fluid was increased beyond that customary in the art. At such ratios of fluid MgO to solids (approximately 1 cc of a 1.26 sp. gr. MgCl.sub.2 solution fluid per gram of MgO), the customary result would be excessive syneresis of the gauging fluid resulting in an "oily" surface, followed by crystallization of the salt, depositing a heavy efflorescence of white powder, followed in turn by hygroscopic adsorption of atmospheric moisture, leaving a continuously wet surface. The novel formulation removes the criticality of maintaining the correct stoichiometric balance between the magnesium salts and the magnesium oxide in order to prevent the efflorescence and syneresis, since the high alumina calcium aluminate cement, utilized as a co-binder, in its own hydration/crystallization reaction binds all the excess salt solution. The formulation facilitates the adding of sufficient fluid to permit ease of working, and smooth coating flow without the deleterious effects of excess activator solution. Excess fluid normally results in high film shrinkage and unacceptable crazing of the coating. The formulations herein presented, by utilizing the salt solutions as the "water of hydration" for the calcium aluminate cement have completely obviated this problem, and the coatings are crack-free. They have remained crack-free during almost two years of aging under exposure to the fumes and fluorescent lighting of an industrial chemical laboratory.

The surfactant (9), Niaproof 4, by "wetting out" the powered binders and fillers, and performing this function in the high salt content milieu, decreases the fluid requirements for a workable formulation, thus contributing to the strengthening of the coating.

In the setting process, the coatings show essentially zero shrinkage, and in addition to the absence of shrinkage cracking, they do not lift off from the substrate. They exhibit excellent adhesion to metal, wood, fiberboard, concrete, wall paper, fiberglass fabric and board, clay and ceramic tiles, asbestos cement, and many plastics--including plastic foams. The coatings do not adhere to fluoroplastics nor to polyester film (however, they adhere tenaciously to polyester fabrics). The separation from polyester film is complete, imparting a smooth, glossy surface to the coating. This "release paper" characteristic is made use of in lining the surfaces of molds into which the formulations are cast; or by covering the coatings with the polyester film so that when the film is stripped off, it leaves a smooth glossy surface. They can be made to bond to the films by using an intermediary track coat of an acrylic latex or other adhesive.

The coatings set equally well whether covered by the polyester film or exposed to air.

The addition of the calcium aluminate cement and/or colloidal silica, both acting as co-binders, imparts a surface hardness (Mohs 5.5-6.0) and abrasion resistance exceeding that of prior art "oxychloride" coatings, and brings to the normally much softer "oxysulphate" coatings a hardness (Mohs 4.0-5.0) almost equal to that of the "oxychloride" coatings. (It should be noted that one of the major drawbacks to the use of "oxysulphate" floors and walls, stucco, fireproof mastic coatings, despite the many advantages of this material, resides in its "softness." The use of calcium aluminate cement overcomes this failing).

A high alumina calcium aluminate cement (70% to 80% Al.sub.2 O.sub.3) is preferred to the cements which contain lower concentrations of alumna 47% to 58%). These latter are brown in color and detract from the high reflectivity white which is desired in the product, but, more importantly, contain much higher concentrations of calcium oxide (CaO), 34% to 39%, versus 17%, which has a disruptive effect on the crystallites of the cement at high temperatures, and leads to catastrophic cracking of the coating when exposed to flame.

Novelty also rests in the use of colloidal silica, dispersed in dimethyl formamide (10) as a compatible co-bonding system with the other two co-bonding systems (1+7) and (2+7). Thus the system contains three bonding systems, all of which have differing characteristics, and retain their cohesiveness at different temperatures. The "magoxychloride" loses its water of hydration at 570.degree. F., and starts to calcine at 1000.degree. F. The colloidal silica is stable to 2000.degree. F., and the calcium aluminate cement to 3200.degree. F., with transitory weakness at 800.degree. to 1200.degree. F. while it undergoes phase transformation.

Coatings I-V are self-levelling and with final set, usually in 2 to 21/2 hours, they are brilliant white, smooth, hard (Mohs 5.5-6.0), dense, difficult to scratch except with pointed steel objects, and capable of faithfully reproducing the finest detail of a mold. They develop a high gloss if cast against a smooth plastic film (e.g. polyester or polypropylene) or a plastic mold without the need of a release agent. If cured with the surface exposed, type I's surface stays glabrous (shiny), but the others develop a matte finish. The coatings may be tinted or colored by the addition of water- or oil-based mineral pigments, such as the Hoechst Corp. "Colanyl" pastes (yellow, brown, green, etc.); and the coatings may be overprinted by silk-screening or other printing methods to give a decorative pattern. They may be covered with wallpaper, with the coating serving as a high temperature resistant adhesive, if the paper is applied while the print is still in the gel stage.

The coatings are resistant to water leaching and to the action of many polar and non-polar and aromatic solvents. They are abrasion resistant. They are crack-free, age without noticeable changes, and exhibit little shrinkage in curing.

As discussed above, the TiO.sub.2 imparts a brilliant whiteness to the coatings, which reflects most of the incident infrared radiation of a flame, and thus contributes to the lowering of the surface temperature of the substrate. Most of the heat transferred by a flame is radiative (radiation). While flat black paint will absorb 92-97% of this incident radiation, a white glossy paint will absorb only 21% and reflect the other 79%. This difference between a white reflective surface and a dull absorptive surface can mean a lowering of several hundred degrees surface temperature in the former. NASA has determined that increasing the reflection of radiant heat from 30% to 80%F resulted in a decrease of surface temperature from 2300.degree. F. down to 1900.degree. F. Thus, the high radiation reflective surface imparted by the glossy white coatings described in this disclosure should operate at considerably cooler temperatures and require a longer period of time before their protected surfaces reach ignition temperature.

Another novel feature of these types of formulations, compared to the prior art, is the use of colloidal silica dispersion in DMF (10), which further increases the hardness, density, abrasion resistance, and hydrophobicity, and waterproofs the coating, further decreasing any tendency for rain to leach the magnesium salts. It also prevents the coating from picking up moisture from the air. The colloidal silica imparts a high sheen and porcelain-like surface to the coating.

The coatings can be applied by brushing, spraying, dipping, roller coating, silk-screening, or by any other convenient method. Several layers of coating can be applied, on top of one another. The freshly applied layers show excellent bonding to the older base layer.

The coatings can be texturized by adding sand, gravels, vermiculite, expanded perlite, etc.

The formulations, currently, because of the short shelf life after mixing, must be packaged as a ready-to-use two-component system: Part "A", the blended powered solids; and Part "B", the liquid components, to be mixed at volumetric ratios which correspond to the metric weights shown in TABLE I.

When coated onto both surfaces of a free-standing woven or non-woven natural fiber or synthetic resin fabric or fiberglass fabric, the cured coatings impart complete fire-retardancy to the fabric. Under the 1950.degree. F. temperature and high velocity of a propane torch for 30 to 60 minutes, the fabric does not combust, and there is zero flame-spread from the area of the cone of impingement of the flame. The organic fabric pyrolyzes into a carbonized skeleton within the ennobling coating, and the fiberglass melts, but the structural strength of the coating holds the fabric together; nor does the flame penetrate to the opposite side of the barrier. The fabric serves both as a stroma or supporting framework, and as a network of stress-relief foci. The numerous minute fibrils interlock with the crystallites as they form, leaving compressible points at which the volumetric changes associated with the heat-engendered phase changes in the cements have an opportunity to release their thermodynamic energies. They also serve to increase the tensile strength of the coatings.

The thermal dissipation of polyester or glass fabric is not as high as a metallic screen would be, it is better than that of the magnesium "oxychloride" cement itself, and thus reduces localized temperature build-up. No optimum or preferred fabric porosity has been established, but should be in the range of 50% to 95%. The HOESCH Trevira spunbond grade 1110 polyester fabric used in the examples given below weighs 3.3 oz. per square yard, has a porosity of 85%, and has a melting point of 485.degree. to540.degree. F.

A variety of laminating fabrics has been used; non-woven spunbond polyester (geotextile); woven polyester; nylon; jute (burlap); cotton, wool; steel wool; aluminum and carbon steel, -16 mesh window screening, woven cloth (graphite), cloth's woven aramid ("Kevlar") fabric; etc. Materials of choice are the woven fiberglass, where it is desired to keep to a minimum smoke and gases generated by the pyrolysis of organic materials, and the spunbound polyester, which imparts desired physical properties (high tensile strengths) until called upon to serve as a barrier in an actual fire.

This laminate system of a coating on both surfaces of a fabric can be laid on a susceptible substrate (wood, fiberboard, composition board, plastic laminates, foamed plastics, etc.) with similar results. The coating on the undersurface acts as a high-temperature bonding adhesive for the fabric-to-substrate interface. The paint layer on the top surface forms the refractory-hard coating. If the coating laminate is laid between two mating substrates, the top layer also acts as another high-temperature bonding adhesive. Whether on the surface of one fire-susceptible substrate, or bonded between two substrates, the coating laminate serves as a fire barrier. The coating laminate adds considerable tensile strength to the structure. Since the coating material fills the voids and spaces within and between the fibers of the mesh of the fabric, the laminate carries more weight.

The efficiency of the supporting "textile" in the laminate is due to a combination of: (1) fabric porosity (i.e. the volume of voids divided by the total fabric volume, measured as a percentage); (2) the heat resistance or melting point of the fibers; and (3) the thermal conductivity of the individual fibers. This latter allows for a thermal gradient to be established between the hot spot immediately beneath the flame, and cooler areas further away from the flame. Since the duration of effectiveness of a fire barrier is in part a function of the quantity of hydrate material available for release of water vapor, a laminate is a better insulator than the coating alone. Although two or three coats of paint alone can be applied to a surface to build up a dry thickness of 15 to 18 mils, a single laminate layer can build up to 25 mils thickness, while two or three layers of laminate can build up to 1/8" or more--all possessing the characteristics of the paint: refractory hard and abrasion resistant, and brilliant white; capable of serving as a fire barrier for 20 to 30 minutes when exposed to a flame of 1950.degree. to 2000.degree. F.

Both the coatings alone, and the coating laminates--single or multiple--can be sawn, nailed, stapled, or fastened with screws, without fracturing.

Instead of using woven or non-woven fabrics for increasing the tensile strength and serving as the stress-relief face for the coatings, a variety of types of chopped fibers could be used to serve the same functions, e.g. glass, steel, polyester, aramid, nylon and other synthetic and natural materials.

Type I formula (TABLE I) intended for outdoor use, was made by the method indicated in the teachings (vide supra). In this formula, the colloidal silica (10) not only becomes a densifying agent and a waterproofing agent, but also becomes a co-bonding agent along with (1) and (2). Since the viscosity increases fairly rapidly, and the "pot-life" of a flowable coating is 15 to 20 minutes, batch size is limited to the quantity which can be applied to that period of time. It may be noted that when the colloidal silica (10) is added to the magnesium chloride, the reaction is highly exothermic. An opalescence develops, along with an increase in viscosity, which is reduced by the addition of the surfactant (9). The coatings were brushed onto a variety of substrates: 6".times.6" fired ceramic tiles; 3".times.6" concrete cylinders; 6".times.15" of woven burlap (jute) fabric, and HOECHST Trevira non-woven needle-punched spunbound polyester fabric, grade 1110, 3.3 oz. per square yard; 6".times.15".times.14" white pine board; 6".times.15".times.3/8" plywood. The fabrics were coated on both surfaces. However, the coatings do not adhere to polyester film ("Mylar") and peeled off cleanly, leaving a smooth glossy surface.

The coatings started to gel in 1 hour, and initial set took place in 90 minutes. Final set varied from 21/2 to 3 hours. The dried, fully-cured paint has a density of 1.56, equal to a weight of 97.4 lbs/ft.sup.3.

All coatings dried crack-free and refractory hard, resisting scratching by a steel knife blade (Mohs hardness 6.0). The coatings were rigid, and, in thin sections, brittle, brilliant, white, smooth, highly glossy, and have remained unchanged in appearance for 20 months up to the time of the preparation of this disclosure. Unlike the other coatings--II, III, IV and VII, the types I and VI, which contain colloidal silica, do not pick up moisture from the air, and the weight of the dry coating remains constant. Since the coatings were brush-applied, they varied in thickness from 4 to 6 mils thick, when dry. The samples were submitted to a battery of tests. The coatings were permitted to "cure" for 1 to 2 days prior to being exposed to a "standard" laboratory flame test.

Flame Tests: Using a needle-tip propane torch of 1950.degree. F. flame temperature, so that the tip of the blue cone of the flame played directly onto the painted surface from a distance of 3/4" to 1", the coated burlap fabric and polyester fabric, mounted vertically, were separately exposed to the flame for 15 and 30 minute periods. Tests were replicated four times with similar results. When tested on the uncoated base fabric, the burlap ignited and burned, creating a flaming torch within 1 second. The polyester fabric ignited within 1 second and melted, and became self-extinguishing when the flame was removed, but in the continued presence of the flame, burning continued until the fabric was reduced to a molten mess. The coated surfaces were rendered "fire-proofed" under conditions of this test. There was zero flame spread. The 1" diameter area exposed to the flame first became "red hog", then "white hot", and stayed in this condition for the full 15 and 30 minutes of the test. The fabric beneath the coating pyrolyzed and outgassed only in the area directly under the flame and in a penumbral circle of an additional 1/2" to 1" diameter. The coatings themselves did not crack, and while they did calcine, they retained their integrity and the source flame did not penetrate through the opposite side. Paper and other flammable materials held against the back of the fabric, directly under the flame impingement, carbonized but did not ignite. The "flame front" did not extend beyond the area of direct flame exposure.

Two layers of paint were applied on consecutive days to the wood panels (white pine board and plywood) with the result that the dried coatings were approximately 8 mils thick. The adhesion between the two coats was excellent, and no line of demarcation could be noted. When exposed to the propane torch test, if the full energy of the 1950.degree. F. flame was applied immediately to the coating surface, the superficial layer would "pop" or decreptitate, due to the rapid volatilization and expansion of the contained residual moisture (not water of hydration). Bringing the surface temperature up gradually over a period of 1 minute, by first playing the flame over a larger surface area, obviated this decrepitation. For burn periods of 15 minutes, the results on the two boards were similar. The coatings held their integrity and there was zero flame spread of the substrate from the area immediately beneath the flame. The coating calcined only in the 1" diameter area directly under the flame and in an additional 1" pneumbra (total diameter of approximately 3"). The substrated, after an initial 5-minute period, showed signs of carbonizing (pyrolysis), giving off thin wisps of smoke, but did not ignite or combust. The depth of the carbonization gradually continued for the full 15 minutes before the first indication of slight browning reached the back of the 1/4" thick white pine board. Smoke density increased up to the 10 minute mark, after which it reached almost zero. At no time did the substrate flame, nor was there a "flame-front". The area of pyrolysis of the substrate was limited to that of the area of calcination of the coating (total 3" diameter). The degree of pyrolysis varied with the depth. At the end of 15 minutes while carbonization was complete immediately beneath the coating, there was a gradual diminution until at the rear of the panel there was only the initial indications of carbonization. The plywood panel smoked more densely due to the use of organic resin and adhesives between the layers, but also showed a lesser degree of pyrolysis. In contrast, the uncoated boards both flamed within 1 minute, and the flame front extended up the board unchecked while being fired, and the flaming continued after the removal of the activating flame.

The coatings continued to adhere well to the substrate, without indication of debonding. The calcined area of the coating could be scraped away, but did not peel away or flake off of its own accord.

This series of tests indicated that the coatings themselves were non-combustible, and were capable of acting as a barrier to the ignition of susceptible wood and fabric substrates, and thus preventing the extension of a flame front.

In preparation for a certified ASTM D-3804 two-foot tunnel test for flame spread, smoke density, and fuel contribution to be run by an independent test laboratory (Applied Research Laboratories of Miami, Fla.) a 233/4".times.37/8".times.1/4" white pine panel was coated with Type I paint, and a layer of HOECHST Trevira non-woven spunbond geotextile fabric #1110 to form a single laminate. As indicated previously, such a construction consists of a base layer of paint applied to the bare wood, a strip of the polyester fabric of the same size as the board, rubbed or rolled in with a rubber roller or a squeegee, in order to saturate the fabric and a top layer of paint applied immediately. The fabric weighed 6.8 gms; the pine board, 150.9 gms; while the laminate picked up a total of 205 gms of paint (dry weight) equal to 2.23 gms of paint per square inch. The total coating laminate had a thickness of approximately 25 mils. The dried paint was brilliantly white, refractory hard, and crack-free.

Results of the ASTM D-3804 test, with a full 5 minute burn (instead of the customary 4 minute burn) at 2000.degree. F., showed: Flame spread 17.8; Fuel contributed=0; Smoke density 56.6; Percentage loss in weight 11.8%. The control white pine panel showed: Flame spread=73.0; Fuel contributed=0; Smoke density 578.9%; and Weight loss=74.64%. The coated panel was barely scorched on the surface; there was no significant penetration of the coating or laminate, while the bottom half of the uncoated control was completely consumed, and the upper half reduced to a badly cracked, completely charred stick of charcoal. The smoke density and percentage loss in weight in the coated panel is principally due to the moisture vapor (steam) driven off during the heating of the coating--as it is supposed to be--in protecting the underlying susceptible substrates.

The coating adhered tenaciously to the 3".times.6" concrete cylinder, leaving a smooth, glossy refractory-hard highly abrasion resistanc impervious coating. Untinted, the coatings were a highly reflective bright white. The addition of minute quantities of HOECHST "Colanyl" pigments to the basic Type I formula permitted the coloring of the concrete coatings to be bright green, yellow, brown and black.

The coated ceramic tiles were weighed and immersed in tap water for periods of 7 and 14 days, after which they were dried and reweighed. There was no noticeable loss of weight, nor changes in the physical properties of hardness, nor gloss, which would indicate increase in porosity due to leaching of soluble salts.

A series of aggressive solvents was flooded over the surface of the coatings on the ceramic tiles, and each permitted to stand for 15 minutes, the excess drained, and the remainder dried off with paper toweling. The coatings did not soften, and showed only a minor loss of surface gloss to some of the treatment. Resistance to the following solvents was thus demonstrated: SOLVESSO 150 (a highly aromatic hydrocarbon); ACCUSOL S/T (20% ethylene glycol in monoethyl ether); Copier machine mix (a high-flash mineral spirit); MOBAY SOLVENT BLEND (4 parts cellulose acetate+1 part xylene+1 part methyl ethyl ketone (MEK)+1 part butyl acetate); kerosene, acetone and xylene.

The fluid paint was poured into a patterned plastic cup with raised printing, and rotated. The excess was poured off, leaving a thin shell mold. Upon drying and removal of the cup, the 6-mil thick layer of paint was not only smooth, brilliant white and crack free, but showed a remarkably faithful reproduction of the finest detail of the mold.

Thus the Type I coating not only performs a role as a fire barrier, but in addition has the necessary attributees of a good paint.

Where outdoor use and exposure to rain is not contemplated, the colloidal silica (10) may well be omitted from the formulation, and Types II and III of Table I utilized for interior use. These differ only in the use of the surfactant (9) dispersing and wetting the binder powders of part "A" so well that this permits a reduction of almost 15% in the amount of gauging fluid required for an equivalent paint viscosity. Type II paint has a pH of 8.5 and weighs 14.26 lbs. per gallon (1.71 gms/cc). When the respective coatings are cured, there is little difference between them. Neither shows any cracking nor appreciable shrinkage. Both exhibit refractory hardness (Mohs 5.5), brilliant whiteness, high gloss and excellent adhesion to various substrates. Hence the results of the testing--the fire barrier properties under different configurations--will be treated together. Omitting the colloidal silica exposes the coating to moisture pick-up from humid atmospheres, and the cured coatings show an increase in weight until psychrometric equilibrium is established.

During a two-year period, over 100 "batches" of Types II and III formulations were made up to coat several hundred types, sizes, and configurations of test specimens. The only difference in the formulations was the grade of MgO (1) utilized. The prior discussion covers the characteristics (vide supra). There was no noticeable difference in the final results of the cured coatings. However, due to its longer "open-pot" time, Basic Chemical Co.'s Magox 98 LR is the preferred grade.

The various configurations of test specimens are categorized below; specimens made up for in-house fire-exposure screening tests and formal specimens made to the standards of ASTM D-3806 and submitted to an independent testing laboratory (Applied Research Laboratories of Miami, Fla. for certified flame testing in their "Two-foot Tunnel.:

[a] Coating applied to bare fabric

1. Non-woven spunbond polyester

2. Glass fiber cloth (woven fiberglass)

3. Woven jute (burlap)

x. Fabrics 1 and 2 were also used as the fabrics forming part of the coating laminate.

xx. Generally, only a single coating laminate was utilized; occasionally, a double coating laminate was applied. Multiple laminates, consisting of 3 to 5 or more layers may also be constructed to form a rigid panel construction of 1/4" to 1/2" thickness, or even up to a 1" thickness.

[b] Coating and/or coating laminate applied to white pine board.

[c] Coating and/or coating laminate applied to laminated plywood.

[d] Coating and/or coating laminate applied to 5/32" wood panelling.

[e] Coating and/or coating laminate applied to thin one-ply doorskin.

[f] Coating and/or coating laminate applied to cedar roof shingles

[g] Coating and/or coating laminate applied to compressed fiberglass "board."

[h] Coating and/or coating laminate applied to gypsum wallboard.

[i] Coating and/or coating laminate applied to polyurethane foam board.

[j] Coating and/or coating laminate applied to isocyanurate foam board.

[k] Coating and/or coating laminate applied to polystyrene foam board.

[l] Coating and/or coating laminate applied to corrugated paperboard.

[m] Coating and/or coating laminate applied to wood fiber (cellulose) board ceiling panel.

[n] Coating and/or coating laminate applied to asbestos-cement board.

[o] Coating and/or coating laminate applied to plywood covered with vinyl wallpaper.

[p] Coating and/or coating laminate applied between two pieces of 3-ply laminated plywood.

[q] Coating and/or coating laminate applied between two pieces of 1-ply doorskin, and between 4 pieces of doorskin (4-ply).

[a] Type II and Type III Coatings on Bare Fabric

[a] 1. Pieces of HOECHST Trevira Spunbond Polyester geotextile fabric, grade 1110, 3.3 oz. per sq. yd., were cut into 51/2" wide.times.14" long pieces. Coated with the paint on both sides. One group permitted to air dry, uncovered; another group was pressed between two sheets of polyester ("Mylar") film and permitted to "set" covered. Coating covered only 10" of the panel, permitting the uncoated 4" length to be used as a control. All coatings cured into a "refractory-hard" brilliant white surface. crack-free. The covered panels were smooth and glossy; and due to the compression, the dried coating on each side was 21/2 to 3 mils thick and weighed a total of 104.7 gms. The coatings in the uncovered fabric was 4 mils thick on each surface. Each coated fabric was exposed several times to the standard 1950.degree. F. propane torch test for 15 and 30 minute periods, all with similar results. The control sections of the panel, flamed and then melted within 1 second. The spot under the direct flame in the coated areas became "white hot" on both surfaces. The area on both sides calcined, but the flame did not break through to the opposite side, no hole was created in the fabric, and the coating did not crack and decrepitate. Occasionally, upon cooldown, the coating in the heated area did crack. The fabric in the heated areas volatilized and left a space betwen the top and bottom coats of paint. In the 1/2" to 3/4" diameter penumbra surrounding the heated areas, the stroma of the fabric was still intact, but the fabric had carbonized. Beyond the penumbra (3" diameter), neither the fabric not the coating showed any signs of change.

[a] 2. An open-mesh woven fiberglass cloth, called SCANDATEX Wall Covering, 3 oz/sq. yd., with a 90 lbs/inch tensile strength, cut into strips 51/2".times.14", was used as the supporting matrix for a similar series of tests--with comparable results. Even after 15 and 30 minutes of exposure, there is zero flame spread from the point of the flame application, with only minor lateral penumbral involvement. While the glass fibers do not volatilize, they do melt, but the strength of the coating maintains structural integrity of the sheet. Since the organic material content of the glass fabric is less than that of the polyester fabric or jute, it might be expected that the production of volatile gasses or smoke would be less, but this was not measured.

[a] 3. An open-mesh woven burlap or jute, 8-mesh, 6.5 oz/sq. yd., cut into strips 51/2".times.14", was similarly coated on both surfaces with the paint, and tested by the propane torch method. Similar results were obtained. After 15 and 30 minute test exposure, there was zero flame spread, and the coatings held their integrity, although calcined.

In the above series of tests, which demonstrate the non-combustibility of the coatings and the ability of the coatings to prevent the penetration of open flame to the opposite surface, the fabrics acted as both a substrate and as a stress-relief "filler" preventing cracking and decrepitation, and indicated the value of this combination in forming a coating laminate for application to fire-susceptible substrates.

TABLE II lists the various combinations of coatings and/or coating laminates applied to a variety of substrates and the results obtained when these were subjected to flame testing to determine their fire-barrier characteristics.

Because of the occasional tendency of Types II and III coatings to decrepitate when heated rapidly, and to crack upon cooldown after exposure to the high temperatures test flames, expanded vermiculite was added to the formulation to serve as stress-relief foci (vide spura, in the teachings). The formulations which include this component follow closely Types II and III, with suitable adjustments in the quantity of the gauging fluid to allow for the higer fluid absorption of this material, in order to maintain the same print viscosity. The substitution of the expanded vermiculite for half of the silica flour did not affect the characteristics of the coating--which dried refractory hard and crack-free.

(a) Coated a 12".times.12".times.1/2" plywood board with a single coat of paint (4-5 mils thick).

(b) 12".times.4".times.1/2" plywood board with a single coat laminate, using Swedish "SCANDATEX" woven fibergalss wall covering.

(c) 237/8".times.31/2".times.5/32" wood wall panelling; double coat of paint alone.

(d) 237/8".times.31/2".times.5/32" wood wall panelling; single coat of paint laminate, using SCANDATEX woven fiberglass.

(a) and (b) were submitted to propane torch tests; (c) and (d) were tested by ASTM-D-3806.

(a) In a series of 6-minute tests, there was noticeable steam emitted at 3 minutes, wisps of smoke at 5 minutes, and, in one case, cracking of the coating. Dissection of the burn area showed no sign of discoloration on the rear of the panel, and carbonized wood to a depth of only 1/8". There was zero flame spread beyond the flame impact area. There was no sign of decrepitation in any test.

(b) In a 15-minute test, steaming started at 51/2 minutes; at 7 minutes, smoke started; at 121/2 minutes, rear of panel showed discoloration. At the end of the test, wood was carbonized to a depth of 3/8" and showed cracks--only under the flame impingement area. There was no flame spread. The coating remained intact and crack-free; and there was no decrepitation, even thought the full flame had been applied immediately to the surface without a preliminary warm-up.

Both (c) and (d) coatings came through the ASTM-D-3806 test in excellent condition. No decrepitation; no cracks; hardly a sign of flame exposure, except for a slight calcination. The highly flammable wood panelling substrate showed minimum carbonization, and the coatings maintained excellent adhesion to the substrate, under the area of direct flame impingement.

Type V formula was developed to determine the effects of adding a well-known fire-retardant material, alumina trihydrate, to the "mag-oxychloride" formula. The resulting coatings were particularly smooth and brilliant white, as well as hard (mohs 4.5), and crack-free; exhibited little shrinkage, and had excellent adhesion to a variety of substrates.

(a) Single coating applied to a 12".times.12" piece of HOECHST non-woven spunbond polyester fabric, grade 1110.

(b) Single coating laminate, using same fabric, applied to a 12".times.12".times.1/2" wood waferboard. Utilized 1.17 gms. of paint per square inch of surface.

(c) Single coat, 4 mils thickness, dry, applied to a piece of 12".times.4".times.1/4" white pine board.

(a) and (b) were tested by the propane torch method; (c) was tested by the "chimney" method.

(a) At 1 minute, whiffs of steam were noted; at 2 minutes, slight smoke arose from the carbonization of the fabric stroma in the area under the impinging flame; at 3 minutes, coating started to calcine and the area under the flame became "white-hot", both on the front and back surface, and developed cracks which radiated out from the center into the penumbral area. In a 15-minute test, although there was no flame spread, the cracks widened and small chunks of coating fell off, leaving a hole in the center flamed area. The central 11/2" diameter area under the direct flame was completely calcined, and there was 1/2" diameter penumbral area surrounding the central area, of lesser degree of calcination. This Type V formulation does not possess the physical resistance of Types I, II or III, in its inability to hold the unsupported coating intact without cracking during the period of calcination.

(b) In 4 replicated tests, at 1 minute steam emission was noted; at 2 minutes paint developed cracks; at 3 minutes paint cracks propagated in all directions and opened wider; at 31/2 minutes, although there was no decrepitation, wood underneath the laminate started to combust (burn)--not merely carbonize with flaming, as would have taken place if the coating had remained intact. By 7 minutes, wood under the coating was in self-sustained combustion.

(c) In the chimney test, two pieces of wood (the coated piece and an uncoated control) are set up in parallel, facing each other, and separated by a 1" spacer at the top. The bottoms of the boards are protected by sheet metal, and the unit is set on top of an open can of STERNO (solidified wood alcohol), which is ignited, so that the 1400.degree. F. flame rises in the "chimney" formed by the two opposing test specimens. At 1 minute, the uncoated control was a blazing inferno, the flames licking up the complete 12" length of the board, with severe charring at the base. At 5 minutes, the basal 6" length of the control board was reduced to char, at which time the tests were terminated. In the coated panel, the coating developed longitudinal and horizontal cracks, and some pieces had exfoliated. The pine board showed a 1/16" thick layer of char under the coating, but there was no indication of combustion or flaming.

Types VI and VII formulations were developed to take advantage of the reputedly higher thermal decomposition temperature and higher water-of-hydration content of magnesium "oxysulphate" cements over magnesium "oxychloride" cements. C. W. REDEKER (1967) found that the "oxysulphate" cements withstand a 400.degree.-500.degree. F. higher temperature. Normally, the "oxysulphate" cements are markedly softer and more abrasive-prone.

Types VI and VII overcame the surface softness problem, and brought the "oxysulphate" coatings up to an acceptable 5.0 and 5.5 Mohs hardness . . . the former by the use of a colloidal silica (as is used in Type I formulation) and the latter by using high alumina calcium aluminate cement, as in Types II, III and IV formulations. Both coatings are brilliant white.

The major drawback of both Type VI and VII formulations is the appearance of surface and deep crazing ("mud-cracking") in the dried coatings when applied as coatings alone on any of the numerous types of substrates listed in Example II. This crazing not only detracts from the cosmetic acceptability of the coating, but also, under flame-testing, the cracks open up and expose the substrate to the direct flame, which results in substrate ignition.

The problem of crazing in the Type VI and VII "oxysulphate" cement coatings has been eliminated by applying the paint onto a non-woven spunbond polyester geotextile fabric. Here the interlaced fibrils of the fabric form the necessary stress-relief foci, so that the coatings set crack-free.

When applied to fabric as a free-standing coating, without a substrate, and subjected to the propane torch test, the coating behaved the same as discussed under Type II and III coatings, tests a.1,--namely that the coatings withstand direct flame exposure for 15 and 30 minutes, without a flame spread beyond the direct impingement area; and without burning holes through the fabric.

Type VI and VII coating laminates, applied to wood substrates. The propane torch test is not sufficiently sensitive to distinguish differences in results between "oxychloride" and "oxysulphate" coatings. Both VI and VII type coatings, when applied to the surface of white pine and laminated plywood, were capable of protecting the substrates from ignition and flame-spread for periods of 15 minutes, and with thicker coatings, for 20 to 25 minutes.

While Type VI and VII coatings can be categorized as "satisfactory," Types I, II and III are the preferred formulations.

From the teachings and test results presented in this disclosure, it is obvious that the paint, subject of this invention, has numerous applications of great utilitarian value.

1. Non-combustible coatings

Since the coating does not burn, it may be used in those confined areas or enclosed spaces which cannot risk the dangers inherent in a paint that burns.

2. High-temperature coatings

Since the coatings withstand 1900.degree. F. for extended periods of time, they may be applied in those situations which require high temperature protective paints (for example, coatings for furnaces, wood stoves, mufflers, or as high-temperature radiation reflectors).

3. Abrasion resistant coatings

Since the coatings are hard and refractory, with great compressive strength, they may be applied over soft surfaces or plastic foams to lend rigidity, puncture-proofness, and abrasion resistance to susceptible substrates.

4. Adhesive bonding agent for wood, wallpaper, etc

The coatings exhibit exceptional bonding strengths for wood and paper, hence may be used as an adhesive for the bonding of wood lamina in the construction of plywood; in the assembly of corrugated paper board or honeycomb panels; or the adhesion of wallpaper or decorative foils to walls or substrates, and at the same time, impart fire-barrier or flame-stop properties.

5. Structural laminates

When multilayers (4 to 12) are built up from paint impregnated non-woven spunbond polyester geotextile fabric, and the layers pressed together while curing, non-magnetic, fire-retardant, structural laminates of tremendous compressive and tensile strengths are formed. Resistant to biodegrading agents, these structural laminates can be made in the form of sheets, tubes, posts, curved panels, etc.

6. Fire-barrier coatings

Potential applications are for roofing; walls; fire doors; lining of school and hospital fire-egress exits; aircraft, boat or ship interiors; coating of horse stables or animal barns; coatings for foam plastics; coatings for plastic pipe, wire and cable conduits; acoustic or decorative ceiling tiles; corrugated paper board containers; fireplace hearths and backing; waste containers; etc.; as well as improving the surface properties of gypsum plaster wallboard.

Disclosed herein are new and useful coatings and coating laminates, in several alternative formulations, which have proven capable of forming "fire-barriers" which prevent flame penetration, ignition, and flame spread to a wide variety of fire-susceptible substrates.

These formulations have overcome the numerous shortcomings of the prior art; viz. salt efflorescence and syneresis of excess fluids; water solubility; high shrinkage; crazing or "mud-cracking"; spalling (decrepitation) under fire; proneness to abrasion due to softness; need to maintain delicate stoichiometric balance of the principal components; deleterious effects of the excess fluid required for smooth paint application; etc.

There has been developed a durable, rapid-setting, non-combustible thin coating capable of protecting susceptible flammable substrates from ignition and flame-spread, which, at the same time, has the necessary characteristics of a good paint, and which bonds firmly to a wide variety of substrates.

Ellis, in his companion U.S. application entitled "FIRE BARRIER: I. COATINGS" describes a series of tests (Table II, p. 1-p. 3 and q. 1) in which a non-combustible inorganic coating was applied between two pieces of 3-ply laminated plywood, and between two and four plies of doorskin, using the coating as the bonding agent (adhesive) for the layups. Results were unexpected, and highly successful. Surface flame spread, even on uncoated surfaces, was limited to little more than the contact area of the flame source, when exposed to a direct 1950.degree. F. propane torch flame, or subjected to the Standard ASTM-D-3806, 2-foot Tunnel test; "burn-through" was prevented, and the flame did not extend to a depth greater than the level of the nearest layer of internal "submerged" coating. In addition to markedly increasing the "fire resistance" of the panels, the "submerged" coating layer was found to influence and limit the surface flame spread. The closer the "submerged" layer was to the surface, the smaller the extent of flame spread. The coating served as an excellent adhesive between the wood plies, and the bond did not delaminate under the heat from either the propane torch or the gas flame (approximately 1900.degree. to 2050.degree. F.). Additionally, the coating can be tinted to match the color of the wood plies, so that the "glue line" is not apparent.

This invention extends the referenced Ellis disclosure and broadens its applicability. The results from a specific type coating formulation have been expanded into a general model for a fire barrier system for plywood and other laminated wood products, applicable to most types of non-combustible coatings.

The general description of the invention is applicable to all types of plywoods and panelling.

The Barrier Coating--Specific embodiments of a non-flammable, inorgaic, high temperature resistant coating, which can be applied to the wood plies by any method known to the art, viz. by brushing, spraying, rolling, roller-coating, are described by Ellis in his copending application. This disclosure contemplates that any formulation which meets these criteria could be used to form the fire barrier layer.

The coating layer can be applied between one or several, or all of the plies, while the laminate is being laid up, prior to pressing and heat curing. It can also be applied to, or can be used to bond, two traditionally bonded multiple-ply layups. In addition, the coating can be applied to the front or top surface, and/or to the rear or lower surface of the laminate. At any of these levels--top, intermediate, or bottom--the coating influences and/or prevents flame spread and fire penetration.

The coating can be used alone as the adhesive or bonding agent, since it bonds tenaciously to the wood fibers. The length of time required for cure is a function of the specific coating formulation and the temperature. In the exemplary formulation discussed herein (a synergistic combination of two or three bonding systems, viz. magnesium oxide/magnesium chloride, high alumina calcium aluminate cement, and a colloidal silica--optional), the coating normally requires two hours to set at ambient temperature, but can be accelerated to 45 minutes at 120.degree. F. The coating, consisting of both hydraulically-setting and chemical-setting systems, can cure anaerobically--without access to air, and hence can cure when layered between the plies of the laminate. The coatings are easily sawed, nailed, or screwed.

The thickness of the coating can be varied at will from a few mils to 1/16" thick. The thicker the coating, the better the fire-barrier performance. The greater the number of plies to be coated, the thinner each glue line needs to be.

The fluid coating can also be applied in the form of an impregnated fabric. Bonding peel strengths are equivalent. When using impregnated fabric, the glue line is thicker (on the order of 1/32" to 1/16") since the fabric holds more coating material. The fabrics add significantly to the flexural strength of the laminate. Fabrics which have been successfully used to date are: Non-woven, spunbonded and needled polyester fabrics (geotextiles) in several weights (2.0 to 4.0 ounces per square yard); non-woven fiberglass veil; woven fiberglass; woven carbon cloth; and woven aramid fabric ("Kevlar"). Fabrics of quartz, nylon, or other natural or synthetic or inorganic fibers could also be utilized. The coating-saturated fabric cures at the same rate as the coating alone.

Where a specific non-combustible coating does not have the required bonding strength to a wood species veneer (ply), and if it is desirable to use conventional bonding adhesives, the coating may still be used as the fire barrier layer. The impregnated fabric referred to above, is air cured into a flat rigid sheet and trimmed to proper size. It may have a smooth surface by curing between release paper, or left with a matte finish by being cured uncovered. All of the commonly used plywood adhesives (phenol formaldehyde, urea formaldehyde, melamines, resorcinol, and phenol resorcinol) show excellent adhesion to the type coatings discussed herein, and would probably show similar adhesion to other cementitious coatings. The cured coating and carrier fabric is laid up between the wood plies, using an adhesive layer on both surfaces, and conventionally bonded by pressure and the heating cycle required of the adhesive.

Finally, the bond or glue line between the wood plies may be a mixture of conventional plywood adhesives and the non-combustible coating. All of the wood adhesives referred to above are compatible with the type aqueous-based non-combustible coating which is described herein. To 80-85% by weight of the freshly-prepared coating is added, with adequate mixing, 15 to 20% by weight of the selected adhesive, and the mix spread or applied to the wood plies in the conventional manner. The hybrid system requires a different time/temperature cure cycle for each of its major components. If the phenolic or urea adhesives are cured under their typical press cycles prior to the cure of the coating, the fluid of the coating boils off and all cementitious properties (as well as fire-barrier properties) are lost. Hence, the coating is permitted to cure first, although the cure may be accelerated by slight heating to 110.degree.-120.degree. F. (or by adding known accelerators for both the MgO-MgCl.sub.2 and calcium aluminate systems). After the coating cures (1 to 2 hours) as stated above, curing can take place "anaerobically" (without exposure to air) under the laid-up plies of the laminate, the remainder of the hybrid system is subjected to the time-temperature-pressure curve of the adhesive component. As noted above, the resorcinol adhesives do not require a heating cycle, so if they are used, the normal pressure lamination process serves for both the adhesive and the coating components.

This hybrid bonding system adds water repellency and enhanced adhesion to the coating without detracting from the latter's fire barrier properties.

The function of the coating, whether applied to the surface of the plywood panel, or applied as a "submerged" layer between one or several or all of the plies, is to present an unbroken non-combustible barrier to the flame front, and thus prevent the ignition of the substrate to which it is applied. The coating is not primarily an insulation, since, with time, conductive heat transfer will bring the substrate temperature near to that of the heat source, if the heat source remains constant and does not die out due to lack of fresh fuel in the immediate area of the initial flame. Those coatings which contain molecularly-bound "water of hydration" serve as an insulation maintaining the substrate at a temperature no higher than 570.degree. F. until all of the water has been driven off--a period of time which depends upon the thickness of the coating. The energy required for this release is approximately 1000 BTU's per pound of water released. The coatings may contain from 40 to 54% by weight of this bound water. Until spontaneous auto-ignition temperature of the wood is reached (approximately 650.degree. F. [340.degree. C.] for 1 hour), in the absence of an igniting flame touching the surface, the substrate will pyrolyze and char, but will not ignite and will not support surface flame nor burn-through. The char itself acts as an insulation, and resists temperature changes to the underlying wood--which is important for strength retention. In Douglas Fir, the char base equilibrates at 550.degree. F. (288.degree. C.), well below the auto-ignition temperature of the wood. One quarter inch (1/4") inward from the char, the maximum temperature reached is 360.degree. F. (182.degree. C.) and at one half inch (1/2") inward, the maximum temperature is 200.degree. F. (93.degree. C.). This illustrates the importance of maintaining a cohesive coating which prevents flame from igniting the substrate.

Many types of non-combustible mostly inorganic coatings based on either hydraulic or chemical setting are capable of performing this function, and meeting the other performance parameters required for a satisfactory coating. These are well known in the art, and many variations and combinations are possible. Among the types of coatings may be mentioned those based on Portland Cement, high alumina calcium aluminate cement, magnesium "oxychloride" and magnesium "oxysulphate"; calcium sulphate (Plaster of Paris), magnesium phosphates, alumina, silicas and silicates, such as colloidal silicas and sodium, potassium, ammonium, and "organic" silicates, refractory compositions, and a wide variety of phosphate-bonded cementitious compounds.

Ellis (1984) in the copending U.S. patent application Ser. No. 603,799, filed Apr. 25, 1984, entitled "FIRE BARRIER: I. COATINGS", referred to above, discloses the following composition, (Type II), as being one type of coating satisfactory for both surface application and "submerged" application, i.e., used either on the top or bottom of the plywood, and between the plies of wood laminates, when applied either alone or as an impregnated fabric. He discusses in detail the materials specifications and behavior for each ingredient. he also discloses six other formulations, all of which would also be satisfactory in this laminated plywood application. These coatings are the result of a synergistic reaction between two disparate binding systems, in which the stoichiometric imbalance between the magnesium oxide and magnesium chloride, caused by the need for sufficient fluidity in the mix to enable this subsystem to be applied as a paint or coating is compensated for by the high alumina calcium aluminate cement utilizing this excess liquid as its "water of hydration." The excess magnesium chloride solution, if not otherwise utilized, would normally lead to a film disruption (cracking and crazing), salt efflorescence, and a weak, soft coating. In these formulations, the MgCl.sub.2 solution is "sopped-up" by the calcium aluminate binding system. The two binding systems complement each other, and retain their bond strength and coating integrity over different temperature ranges which span the normal fire cycle of 1800.degree. to 1950.degree. F.

The yield from this batch size is 700 cc. In the FORMULATION NO. II, presented above, Part "A" is added to Part "B" with adequate mixing to the consistency of a house paint, and applied by brushing, spraying, roller coating, or flooding to yield the desired dried thickness. The fluid paint has a specific gravity of 1.71, and weighs 14.26 pounds per gallon.

The coating develops an initial set in 1 hour, and a final set in 2 to 21/2 hours, and maximum strength in 24 hours. It dries crack-free. It is brilliant white, with high infrared radiation reflectivity. Since the dominant mode of heat transfer of a fire is by radiation, the difference between a white reflective surface (emittance of approximately 0.8) and a dull absorptive surface (emittance 0.3) results in a lowering of the surface temperature of several hundred degrees (at the 1900.degree.-2000.degree. F. range) in the former. Where the surface of the plywood is coated with this white paint, this surface directly faces the fire. Where the coatings are "submerged" and placed between the plies, as the top ply is burned off, it exposes the bright white coating underneath.

Usually two coats are applied to the surface, although one is adequate if an internal or "submerged" coating is also applied. The coating has a relatively low heat conductivity, and thus acts as a thermal insulator for a short period of time. There is little lateral spread of heat when exposed to the direct impingement of a 1950.degree. F. propane flame source of 1" diameter. The zone of influence, or halo (or penumbra), of 500.degree. to 550.degree. F. extends, after 15-minute exposure, only an additional 1" to 11/2" in diameter, beyond which the temperature does not exceed 180.degree. to 200.degree. F.

The coating used alone or impregnated into one of the fabrics listed previously, when cured, forms a hard highly abrasion-resistant ceramic-like layer (Mohs hardness of 5.5). It conforms with exquisite detail to the surface of the wood substrate, and bonds tenaciously. The surface coating can also be tinted any desired color, or overprinted with a pattern; or top-covered with wall paper on the surface. The coating may be cured smooth and glistening (glabrous) or with a matte finish. Coating-impregnated woven fabrics (such as fiberglass fabric) have a pleasant decorative appearance. Surface coatings may be "field-applied", i.e. after panel installation, and handled as a special paint application, or factory-applied coatings can be "touched-up" after installation to repair any damage.

When exposed to flame, the coating retains its integrity during the heat-up and exposure periods. Occasionally, depending on the speed of cool-down, it will show one or two stress-relieving cracks. After 15 to 20 minutes' exposure to the direct flame--at a temperature of approximately 2000.degree. to 2200.degree. F.--the coating will show signs of calcination, but it still retains its integrity. The softer calcined areas can be scraped away, but they do not peel nor flake off on their own accord.

When coated onto or impregnated into a natural fiber or synthetic resin woven or non-woven fabric, such as spunbond polyester, under the 1950.degree. F. propane torch for 30 to 60 minutes, the fabric does not combust, and there is zero plame spread from the area of the cone of impingement of the flame. The organic fabrics pyrolyze into a carbonized skeleton within the ennobling coating (and fiberglass melts), but the structural strength of the coating holds the fabric together, and the flame does not penetrate to the opposite side of the barrier. The fabric serves both as a stroma, or supporting framework, and as a network of stress-relief foci. The numerous minute fibrils interlock with the crystallites as they form, leaving compressible points at which the volumetric changes associated with the heat-engendered phase changes in the cements have an opportunity to release their thermodynamic energies. They also serve to increase significantly the tensile strength and flexured modulus of the laminated plywood.

A variety of types of laminated wood construction were fabricated using the inorganic high-temperature resistant non-combustible coating referred to above: (a) the coating alone, or the coating-impregnated woven or non-woven fabric applied to the top or exposed surface of 1/8", 1/4", 1/2" and 5/8"-thick standard interior/exterior plywood; and 1/8" doorskin; (b) the coating alone, or coating-impregnated woven or non-woven fabric applied to the bottom or unexposed surface of these same products; (c) two, three, and four pieces of 1/4", 3/8" and 1/2" plywood, and 1/8" doorskin and 1/4"-thick wall panelling were utilized as "plies" and bonded together with the coating, or the coating-impregnated fabric as the adhesive, and cured "anaerobically" at ambient temperatures in a press under slight pressure (20 to 30 psi). [Each of the "plies" in these laminations actually consisted of a standard 3-, 4- or 5-ply plywood, or doorskin bonded together with a standard thermosetting organic resin adhesive]; (d) two and three pieces of 1/4" 3-ply plywood were bonded together using GEORGIA-PACIFIC GP 1948, urea-formaldehyde resin adhesive with ammonium sulphate catalyst at 1% concentration, under 100 psi pressure, at 300.degree. F. for 4 minutes, sandwiching between each two pieces of plywood a cured coating-impregnated non-woven 3.0 oz per square yard spunbond polyester geotextile as the "submerged" coating layer; (e) top-surface coating applied to "submerged" coating bonded laminates; and (f) bottom-surface coating applied to "submerged" coating-bonded laminates. (This latter was intended to function as the "final" barrier to prevent ignition and burn-through to any other substrate, e.g. roof sheathing subflooring, wall partitions, etc.).

Two types of fire exposure tests were performed: (a) The propane torch test: The hottest part of the high-velocity 1950.degree. F. flame was directed against the candidate coated sample, usually 9".times.9", or 12".times.12" in size, or its uncoated control. The samples were mounted free standing at a 15% inclination off the vertical, and held steady in a test bracket for the test duration, which extended for as long as 60 minutes. Readings were taken every minute of the temperatures on the back, signs of discoloration, amount of smoke generated, evidence of coating cracking, surface burning, burn-through and temperature peripherally from the center of the flame contact; and (b) ASTM D-3806 "Small Scale Evalution of Fire Retardant Paints (2-foot Tunnel method)." These were certified tests run by Applied Research Laboratories (ARL) of Miami, Fla., on 2-foot-long by 4"-wide test panels. This test consists of impinging directly onto the surface coating a 2000.degree. F. gas flame supplying heat, at 5085 BTU's per hour, for 5 minutes, while the test panel is suspended above the burner, coated surface facing downward, at an upward angle, in a chamber. The angle of the test panel, and the draft of the chamber, encourages any flame in the coating and substrate initiated from the burner to proceed upward along the substrate. The extent of this flame front is measured and compared with the known substrate standards (0 and 100). In addition to flame spread, the smoke density, fuel contributed, and weight loss are also recorded and calculated. Although this test is mainly concerned with flame spread along the surface, weight loss and visual inspection and depth measurement shows fire endurance (resistance, or burn-through protection).

Flame spread is the main factor associated with testing fire-retardant coatings. The "fire-retardant" coatings currently available are rated as low as 10, or as high as 60 or 70. The lower the rating, the longer the coating will retard flame.

The following examples show that under the test conditions:

(a) When high-temperature inorganic non-flammable coatings or coating-impregnated fabrics are applied to the front or "hot" surface of a laminated wood product (plywoods), they are capable of achieving zer0 (0) flame spread, as well as zero (0) smoke density, and zero (0) fuel contributed; and of stopping, or considerably slowing down, fire penetration. This latter is a function of flame exposure time. In a 5-minute test there is little penetration. In a 15 or 20-minute test, there is substrate charring but no flaming;

(b) coatings or coating-impregnated fabrics applied to the rear, or "cold" surface of a laminated wood product, at a depth of 3/8" to 1/2" below the front surface, prevents flame from reaching the substrate on which it may be mounted, limits the burn-through area to a circle of only a couple of inches in diameter; and influences surface flame spread by restricting it to an area only slightly larger than the burn-through circle;

(c) one or several internal or submerged coating layers--coatings alone, or coating-impregnated fabrics between the wood plies, are highly effective in preventing fire penetration and burn-through. The flame is usually stopped by the first coating layer; continued flame exposure leaves a char behind that first coating layer. If not stopped by the first layer, flame penetration is stopped by the second coating layer. Even in exposures for as long as 60 minutes in the normally highly combustible doorskin laminates, made up of four layers of doorskin 1/8" thick, with 3 layers of coating alone between the plies, the flame did not penetrate to the rear surface. Moreover, the flame spread on the surface, as well as the hole which did penetrate to the second layer of coating, was restricted to 2" to 21/2" in diameter. As a general rule, when the surface is uncoated, the closer the first layer of internal or submerged coating is to the surface of the panel, the greater its effectiveness in decreasing surface flame spread;

(d) a surface coating backed up by one or two internal coats at 1/8" to 1/4" depth below the surface coat yielded substantially zero (0) surface flame spread, and a minimum of flame penetration.

The coatings and the coating-impregnated fabrics, applied to the rear of the panels, almost always retained their integrity, remaining whole, and prevented flame from penetrating through to the space on the other side. In practice, this would prevent a chimney effect involving spaces between studs; and if applied to roofing shingles, would prevent ignition of the roofing sheathing. In addition, this rear coating may obviate the necessity for utilizing gypsum wallboard as a rear fire-barrier.

An additional advantage of the coating systems is that by greatly reducing flame spread and penetration, they decidedly decrease smoke emission.

A batch of Type Formulation No. II paint, as given above, was prepared and used to brush coat the surface of a 6".times.15".times.3/8"-thick interior/exterior plywood board. Two coatings were applied, on successive days. Total dry thickness of the coats was 16 mils. The adhesion between the two coats was excellent, and no line of demarcation could be noted. The coating was permitted to cure for two days prior to testing. The sample and its uncoated control were submitted to Propane Torch tests, as described above.

Uncoated Control Panel:

Within 20 seconds after application of the flame, the uncoated board caught fire. The surface flamed in all directions, quickly extending to a diameter of 5" from the epicenter, but principally upward from the point of the torch impact, and continued burning on the surface as well as progressively deeper. The panel completely burned through in three minutes. With the torch removed after 3 minutes, the panel continued to burn both on the surface and through all layers of the complete panel depth. The result was a mass of charred, cracked pieces and the some ash--almost complete combustion in the upper 7" of the panel prior to extinguishing the flaming brand.

Surface Coated Panel:

For the first five minutes of torch application, results were minimal and hardly noticeable. The coating held its integrity, and there was zero flame spread on the surfafce, and no noticeable changes in the substrate. Temperature on the rear surface was about 15.degree. F. above ambient of 76.degree. F. The coating showed signs of calcination only in the 1" diameter area directly under the flame, and in an additional 1" penumbra around the central core--i.e., for a total diameter of 3". At 6 minutes, the substrate wood showed signs of carbonizing (pyrolysis), and thin wisps of smoke emerged, but there was no ignition, nor combustion. This initial "smoking" is principally due to moisture vapor (steam) being drawn off during the heating from both the coating (water of hydration) and from the substrate (adsorbed moisture). Smoke became slightly more dense as the adhesive pyrolyzed, and the smoke density increased up to the 10-minute mark, after which it decreased. The depth of carbonization gradually progressed. At 12 minutes there was the first indication of a slight browning on the back of the board, which increased only slightly until the end of the 15-minute test period. At no time was there any ignition or combustion of the substrate. There was zero surface flame spread. The surface area affected was limited to the 3" diameter of calcined coating, beyond which the coating was entirely unaffected. The coating continued to adhere well to the substrate without indication of debonding. The pyrolyzed area under the point of flame impingement was likewise limited to the approximately 3" diameter directly under the area of the surface flame application. There was no further spread.

This test was replicated several times with essentially similar results. Some tests were run for 20 and 30 minutes. The only noticeable changes were a progressive increase in the depth of carbonization, until full charring on the rear. Continued exposure to the constant 1950.degree. F. propane torch flame finally results, at approximately 28 to 30 minutes, in "burn-through", in which the char itself fractures and finally ignites in the limited area under the flame. However, there is still no further extension of surface flame spread.

This test demonstrates that surface coating alone, using a high temperature non-combustible "paint", with adequate cohesive strength to withstand the thermal stresses without cracking, and the adhesive strength not to debond, could be used as a fire-barrier in preventing both surface flame spread and flame penetration. This test is more severe than that encountered in actual fire situations in that under fire conditions it is unlikely that the flame source would be either this intense or concentrated in such a small area, or for such a prolonged period of time.

Two 24".times.4".times.1/4" pieces of interior/exterior 3-ply laminated plywood were subjected to the ASTM D-3806 test for a 5-minute burn. Tests were performed by an independent test laboratory (Applied Research Laboratories of Miami, Fla.) (ARL).

One coat of Type II Formulation Paint was brush-applied to the surface of one panel, and a layer of 3.0 oz. HOECHST spunbond non-woven polyester fabric laid into the coating, which impregnated and ennobled the fabric. The other panel was left as the uncoated control.

Uncoated Control:

Flame spread was 73.0; percent weight loss, 58.6%; smoke density, 543.4. After burn, remnants were a few broken pieces of charcoal, carbonized throughout.

Coated Panel:

Flame spread was zero (0); percent weight loss 3.0%; smoke density, 9.2%.

The weight loss was due to a loss of moisture in the panel, and the smoke due mainly to the development of steam, since there was no carbonization of the substrate, which showed practically no change in this severe (2000.degree. F. direct-flame impingement for 5 minutes) but limited time test.

Coated Panel:

12".times.12".times.1/2" 5-ply interior/exterior plywood. HOECHST spunbond polyester geotextile 3.0 oz. per square yard, impregnated with Type II formulation coating, applied to the surface of the panel. Tested with Propane Torch. Steam started to emerge at 31/2 minutes; no decrepidation of coating; at 5 minutes, the coating was "white hot", at 7 minutes the coating had calcined, but did not crack under the area of the flame. Slight smoke emission. At 10 minutes, heavy smoke emission. Carbonization under the flame was 1/4" deep. No substrate flaming. Zero surface flame spread, and no surface flaming. Subsurface carbonization was restricted to the area directly under the flame (11/2" diameter) and a penumbral area of an additional 1" diameter--a total of 31/2" in diameter. Test was stopped at 10 minutes.

The impregnated fabric had served as an effective fire-barrier for both the surface and the substrate.

Two coats of Type II formulation coating, applied to the rear surface of 24".times.4".times.5/32" wood wall panelling (Georgia-Pacific 3-ply untreated "Cedar Hollow" simulated wood grain (printed finish) totalling 1 gram of dried paint per square inch of surface. This panel and its uncoated control were subjected to a 5-minute ASTM-D-3806 flame test by Applied Research Laboratories (ARL) of Miami, Fla. Flame was applied to the front (printed finish) surface.

Results were as follows:

Uncoated Control:

Flame Spread 69.0; percent weight loss, 77.79%; smoke density, 584.2%; fuel contributed, 58.6.

Coated Specimen:

Flame spread, 28.9; percent weight loss, 13.22%; smoke density 71.1%; fuel contributed, 0.

This test indicated that applying the coating to the rear of the panel, at a depth 5/32" below the surface, was capable of affecting and decreasing surface flame spread, as well as the depth of flame penetration.

This test was a continuation of Example II. Another 24".times.4".times.5/32" piece of the same type of wall panelling was coated on the front surface with a 6-oz. per square yard open-weave fiberglass fabric, impregnated with the same coating formulation, and subjected to a 5-minute coating formulation, and subjected to a 5-minute ASTM D-3806 flame test. The flame was applied to the rear (uncoated) surface.

Results were: Flame spread, 33.4; percent weight loss, 15.05%; smoke density, 19.7%; fuel contributed, 0.

Except for the significantly lower smoke density attributed to the fact that the printed grain finish on the flat surface was not being burned, the results were similar to Example II. This test gends to confirm the observation that a fire-barrier coating on the opposite side of a panel aids in slowing down flame spread on a surface.

Two pieces of 1/4"-thick, 3-ply, interior/exterior plywood, 12".times.12", were laminated together using Type II paint formulation as the bonding adhesive, in a layer approximately 1/8" thick. The coating was tinted with American Hoechst 17-3010 Colonyl Caramel FK pigment, to yield a pine yellow color, which so matched the color of the plywood that the combination fire--barrier and bonding layer was indistinguishingable from that of the polywood laminate. The coating made a superb adhesive, forming a strong bond between the two plywood panels.

The panel was exposed to the Propane Torch test. The surface of the wood ignited immediately and burned, but the burning was restricted to a zone 2" to 3" in diameter around the flame tip. The burning continued carbonizing the three top lamina of plywood, and reached the coating layer in four minutes. The coating layer was not penetrated by the flame, and for the next 10 minutes, there was no combustion of the panel behind the barrier coating. There was a gradual carbonization (but no flaming, and no flame spread) of this second plywood sheet. Testing continued for 15 minutes, with no further changes.

The coating bond did not delaminate under the influence of the heat of the propane torch, and did not crack. The buried (or submerged) coating had prevented the burn-through of the panel.

A companion test to Example III. Two 12".times.12".times.1/4" 3-ply interior/exterior plywood pieces were boned together using a 12".times.12" piece of Hoechst spunbond non-woven 3.0 oz. per square yard polyester, impregnated with Type II Formulation paint as the bonding agent. This served as an excellent adhesive, forming a bond which showed no line of separation even after the joined panel was subjected to heavy hammer blows.

The panel was tested by the Propane Torch method. Results were similar to Example III. Combustion was restricted to the top panels only, and the flame spread restricted to a 31/2" to 4" diameter circle. Flame penetration did not extend below the coating layer.

This test again demonstrated that the coating serves as an efficient fire barrier to prevent burn-through in laminated plywood.

Two pieces of interior/exterior plywood, 24".times.4" were bonded together. One piece was 1/4" thick (3-ply); the other piece was 1/2" thick (4-ply) using Type II Formulation coating as the adhesive.

Two similar sets were constructed, with the same quantity of coating in each set.

Both sets were subjected to the ASTM-D-3806 5-minute burn test by Applied Research Laboratories of Miami, Fla.

The objective was to study the influence of the "submerged" barrier coating on flame spread and the influence of the nearness to the surface of this coating on that flame spread. Other tests had shown that barrier coatings on the rear of a panel limited the flame spread on the uncoated surface. Hence, in one set, the flame was applied to the 1/4"-thick panel surface; in the other set, the flame was applied to the 1/2"-thick panel surface.

The results were as follows:

In both cases, the flame did not penetrate through the submerged coating layer, and the coating did not debond under the heat of the test. The 1/4" panel surface showed flame scorch on only one-half of the panel length; the 1/2" panel surface showed flame scorch over the entire length of the panel.

The test indicates that the closer the coating is the surface, the greater the flame spread protection.

Two and four pieces of 3-ply veneer doorskin, 8.increment..times.8", were bonded together using Type II formulation between each of the layers as the adhesive. The two-piece unit made a panel 5/16" thick; the 4-piece unit made a panel 1/2" thick. Uncoated control panels were made up using doorskin stapled together with a staple gun. Both were subjected several times to the Propane Torch test.

Control Panels

The doorskin was highly flammable. Within 20 to 30 seconds of applying the torch, they ignited and were completely consumed by flame within 3 to 5 minutes, leaving a few scraps of carbonized charcoal.

Two-Layer Doorskin Sample:

The flame quickly punched a 1" diameter hole through the top layer, until it reached the barrier coating. No flame spread beyond the 1" diameter additional penumbra. Flame was effectively stopped by the coating. On a 25-minute test the rear of the panel gradually carbonized, but did not flame; the coating while "white-hot" remained intact and did not crack; paper placed on the opposite side of the coating scorched, but did not flame.

Four-Layer Doorskin Sample:

The flame punched a 1"-diameter hole through the top layer, which with time enlarged to 3" diameter; flame was stopped by the first barrier coating layer; the second layer of veneer carbonized under the first barrier coating, but did not flame or combust. A total of 46 minutes was required for the rear surface to be carbonized, but at no time during the test did the second, third or fourth layers combust. The three layers of barrier coating, separated by wood char, had formed not only a flame barrier, but a carbonaceous insulation which prevented flame spread. The pyrolyzed area on the rear of the panel was only 1" in diameter.

Other tests with four layer doorskin samples showed that the panels resist the flame for more than 60 minutes without burning through.

The multilayer panels with coating applied between each 1/8" thick veneer, were effective in preventing both surface flame spread and depth penetration, and indicate the possibility of constructing multilayer plywood panels with fire-barrier properties.

It will be apparent from the foregoing discussion, that this invention provides a refractory-hard, high temperature, non-combustible coating which is composed of a mutually compatible and synergistic series of hydraulic-setting and chemical setting inorganic cements, which are caused to set by one and the same activator. For example, in the formulations designated Types I, II, III, IV and V, a magnesium chloride activator solution is employed both as the "gauging" solution for the MgO to cause a chemical reaction which results in a solid solution that is "magnesium oxychloride complex", and is the hydrating agent which supplies the water for the setting of the high alumina calcium aluminate cement.

Within this context, then, this invention provides a fire resistant laminate for retarding flame spread and flame penetration therethrough. The laminate comprises (A) a solid, combustible lamina; and (B) bonded to said lamina (A), a hard, thin, continuous layer comprising an inorganic cementitious composition. Layer (B) is capable of maintaining its structural strength and dimensional stability with zero flame spread when exposed to a flame temperature of 2000.degree. F. for a time period of one hour. The inorganic cementitious composition in a fluid state is capable of adhesively bonding to wood and maintaining an adhesive bond to the wood after the fluid composition is dried to form the solid, thin layer. The inorganic cementitious composition consists essentially of (1) a mixture of two or more hydraulic setting or chemical setting inorganic cements, and (2) an activator. Each of the cements is caused to set at a temperature to about 120.degree. F. within about 45 minutes by the activator. The resulting laminate has a fire retardant classification of Class I (or A).

The above-cited teachings and examples in this disclosure demonstrate that most of the objectives of the invention have been accomplished--at least on a laboratory scale--viz. imparting to plywood, wood panelling, doorskin, and multilayering wood veneers, sufficient fire retardancy so the laminated structures by themselves, without the need of other structural barriers, could form a fire-barrier to both surface flame spread and burn-through, and meet the requirements of a Class I (or A) Flame Spread Index Value.

This has been accomplished by utilizing a high-temperature, non-combustible inorganic coating with good wood adhesive characteristics, either alone, or as a coating-impregnated fabric, as the bonding agent between various combinations of wood veneer plies, and, optionally, as a top surface or bottom surface coating.

This system overcomes many of the deficiencies of currently-used systems, such as pressure impregnating with fire retardant salts, or surface coating with intumescent paints; and will introduce a minimum of operational changes in the manufacturing procedures commonly used for these laminated articles.

Having described my invention in particular embodiments, it is to be understood that various alternatives and substitutions will become apparent to those skilled in the art from a reading of the foregoing descriptions. Accordingly, it is not intended that the scope of the claims appended hereto be limited to the description set forth herein, but rather that the claims be construed as encompassing all the features of patentable novelty which reside in the present invention, including all features which would be treated as equivalent thereof by those skilled in the art to which this invention pertains.

Having described my invention of a novel fire barrier system for laminated plywood and similar constructions, I make the following claims: