- Email: [email protected]
Characterization of the performance of an intumescent fire protective coating
Surface & Coatings Technology 201 (2006) 979 – 987 www.elsevier.com/locate/surfcoat
Characterization of the performance of an intumescent fire protective coating M. Jimenez, S. Duquesne ⁎, S. Bourbigot Laboratoire des Procédés d'Elaboration de Revêtements Fonctionnels (UPRES EA 1040), Ecole Nationale Supérieure de Chimie de Lille, BP90108, F-59652 Villeneuve d'Ascq, France Received 24 August 2005; accepted in revised form 5 January 2006 Available online 28 February 2006
Abstract The aim of this work is to study the efficiency of different intumescent formulations designed to protect steel in the case of hydrocarbon fire. The coating is based on a thermoset epoxy–amine resin system into which fire retardant agents, boric acid and ammonium polyphosphate derivative have been incorporated. The first part of the study evaluates, using large scale industrial furnace tests, the behavior of the thermoset resin containing alone or in combination with additives. It is revealed that in this epoxy resin, the combination between ammonium polyphosphate and boric acid leads to the best protective results. The second part of the study attempts to investigate more precisely the effect and the mode of action of the additives in terms of thermal stability, mechanical resistance and rheological properties using small scale lab tests, to explain why this combination works better than using the two fire retardants used separately. The experiments show that this combination leads to the smallest decrease of viscosity when the resin degrades, the highest mechanical resistance and the highest expansion. © 2006 Elsevier B.V. All rights reserved. Keywords: Intumescence; Fire protection; Structural steel; Ammonium polyphosphate; Boric acid; Epoxy resin
1. Introduction Structural steel loses an appreciable part of its load carrying ability when its temperature exceeds 500 °C. The protection of metallic materials against fire has become an important issue in the construction industry. Indeed, prevention of the structural collapse of the building is paramount to ensure the safe evacuation of people from the building, and is a prime requirement of building regulations in many countries. Intumescent coatings are designed to perform under severe conditions and to maintain the steel integrity between 1 and 3 h when the temperature of the surroundings is in excess of 1100 °C [1–4]. Intumescent coatings represent an increasingly used way to provide passive fire protection to the structural steel that is more and more used in modern architectural designs, whilst at the same time maintaining aesthetic appearance. Intumescence is defined as the swelling of certain substances when they are heated. Intumescent coatings form on heating an expanded multicellular layer, which acts as a thermal barrier that effectively protects the substrate against rapid increase of tem⁎ Corresponding author. Tel.: +33 3 2033 7236; fax: +33 3 2043 6584. E-mail address: [email protected] (S. Duquesne). 0257-8972/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2006.01.026
perature, thereby maintaining the structural integrity of the building. Intumescent coatings contain “active” ingredients bound together by a binder. Generally, three “active” ingredients are used: an acid source (normally ammonium polyphosphate or a mineral acid), a carbon source (such as char forming polymers or polyols) and a blowing agent (e.g. melamine). The formulation of the coating has to be optimized in terms of physical and chemical properties in order to form an effective protective char [5]. Chemical interactions between the “active” ingredients in the formulation lead to the formation of the intumescent char. It is generally accepted [6–9] that first, the acid source breaks down to yield a mineral acid, then it takes part in the dehydration of the carbonization source to yield the carbon char and finally, the blowing agent decomposes to yield gaseous products. The latter cause the char to swell and produce the insulating multi-cellular protective layer. This protective char limits both the heat transfer from the heat source to the substrate and the mass transfer from the substrate to the heat source, resulting in conservation of the underlying material. In this study, the efficiency of intumescent coatings on steel is investigated using a fire test (measurement of the temperature profile on the backside of a coated steel plate when exposed to a hydrocarbon fire) as well as laboratory test to measure the heat
980
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
resistance and its viscoelastic behavior. The combination of different intumescent ingredients in a thermoset epoxy resin is examined. 2. Experimental
& & &
2.1. Materials The binder used is a solvent free thermoset epoxy resin. It is a mixture of the Diglycidylether of Bisphenol A (DGEBA) and an amine curing agent (Polyaminoamide). The resin is cured at ambient temperature. Two main fire retarding agents have been chosen in this study:
&
Madd(T): values of weight given by the TG curve of the additives (APP or boric acid), Mexp(T): values of weight given by the TG curve of the formulation (polymer + additives) that is to say APP or boric acid or both, Mth(T): theoretical TG curve computed by linear combination between the values of weight given by the TG curve of the polymer and of the additives: Mth(T) = (1 − x)Mpoly(T) + xMAPP(T); Mth(T) = (1 − y)Mpoly(T) + yMBoricAcid(T); Mth(T) = (1 − x′ − y′)Mpoly(T) + x′MAPP(T) + y′MBoricAcid(T) where x and x′ are the APP content of the formulations including APP and y and y′ are the boric acid content of the formulations including APP, Δ(T): weight difference curve: Δ(T) = Mexp(T) − Mthe(T).
– A mineral acid: Boric acid (H3BO3) (Aldrich, p.99%). – A commercial Ammonium Polyphosphate (APP) derivative (Fig. 1) supplied by Clariant (Germany). It is an APP coated with a component containing nitrogenous and carbonaceous species.
The Δ(T) curves allow us to show a potential increase or decrease in the thermal stability of the polymeric matrix related to the presence of one or every additive [10].
Four formulations have been studied in this work: the thermoset resin alone, the thermoset resin containing boric acid, the thermoset resin containing the commercial APP derivative and the thermoset resin containing both the APP derivative and boric acid. The formulations were prepared by mixing the components using an Ultra Turrax mixer (600 rpm). For the fire tests the coating was applied at 3.5 mm thickness onto grit blasted steel plates of dimensions 30 × 30 cm2 or 15 × 15 cm2. Free films of the coating and pellets of dimensions 25 × 25 × 3 mm3 (for the rheological measurements) were produced by casting the coating between PTFE plates. For the thermogravimetric analysis the cured coating was ground in liquid nitrogen in an ultra centrifuge mill to produce a fine powder.
Rheological measurements were carried out using a Rheometric Scientific ARES-20A Thermal Scanning Rheometer (TSR) in a parallel plate configuration. Both the viscosity of the coating and its mechanical resistance were measured.
2.3. Rheological measurements
2.3.1. Complex viscosity measurements The TSR was designed for monitoring changes in the rheological properties as a function of temperature and/or time. Samples (25 × 25 × 3 mm3) were positioned between the two plates. A constant normal force (F = 200 Pa) was systematically applied in order to obtain good adhesion between samples and plates, and also to ensure the validity of the results. The program used is a “Dynamic Temperature Ramp Test”: heating program with a heating rate of 10 °C/min in the range 25–500 °C, a strain of 1% and a constant normal force of 10 g (200 Pa).
2.2. Thermogravimetric analysis Thermogravimetric analyses were carried out at 10 °C/min under synthetic air (flow rate: 50 mL/min, Air Liquide grade), using a Setaram TG 92 microbalance. The samples (approx. 10 mg) in powder form were placed in open vitreous silica pans. The precision of the temperature measurements was 1.5 °C over the whole range of temperature (20–800 °C). The curves of weight difference between the experimental and theoretical curves were computed as follows:
&
Mpoly(T): values of weight given by the TG curve of the polymer,
2.3.2. Mechanical resistance Mechanical resistance is evaluated using the protocol described as follows: at t = 0 s, the sample (height, h = 1 mm) is put into the furnace and heated to 500 °C without applying any strain (the upper plate is not in contact with the sample as shown in Fig. 2. This allows the sample to intumesce without any constraint. The upper plate is then brought into contact with the intumesced material and the separation between the plates is reduced linearly (0.02 mm/s). The force is followed as a function of the separation between the two plates (Fig. 2). The upper plate used in this experiment has a diameter of 5 mm in order to increase the pressure on the whole sample and to ensure complete destruction of the char.
O P
2.4. Fire resistance
O n
O N H 4+ Fig. 1. Ammonium polyphosphate.
Industrial furnace tests have been carried by our Industrial sponsor. It is a demonstration test (OTI 95 634) and the specification is to burn a given volume of propane (0.3 kg/s) at a given heat flux (200–250 kW/m2) and at a given distance (1 m)
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
F=0 N
F=0 N
F sample sampl
sample
Z=0 Z
T°C
t = 0sec
T°C
T°C
t = 240sec
t > 240sec
Fig. 2. Test of mechanical resistance.
from the test piece. The burning conditions fit as close as possible the ramp of temperature of a hydrocarbon fire heating curve (about 200 °C/min). The coatings are applied on a steel plate (thickness 3.5 mm) and cured for 1 week under ambient conditions. Thermocouples are attached to the backside of the coated plates. Five thermocouples are used on each plate, so that an average temperature can be obtained. As four plates are tested at the same time the plates are isolated using glass wool. The plates are mounted vertically in the furnace and burnt (Fig. 3) until the thermocouples attached to the backside of the plates reach a temperature higher than 400 °C (the failure temperature). This test produces time/temperature curves and characterizes the heat protective effect of the different coatings in a hydrocarbon fire. 2.5. Bunsen burner test This test has been used to characterize some properties of the char. A high temperature (about 1000 °C) is applied with a Bunsen burner on to a coated plate mounted vertically as shown in Fig. 4. 3. Results and discussion 3.1. Fire protection Fig. 5 shows the evolution of temperature as a function of time on the backside of the steel plates coated with the different
981
formulations. Steel usually loses its main structural properties at around 500 °C. In this case, for safety reasons and because the thermocouple is on the backside of the steel plate, 400 °C has been chosen as failure temperature. Four formulations were chosen and compared to the virgin steel plate (A): the thermoset resin alone (B), the thermoset resin mixed with the APP derivative (C), the thermoset resin mixed with boric acid (D) and the thermoset resin mixed with both (E). The time of failure (when temperature reaches 400 °C) of the steel plate covered with the thermoset resin (curve B) is close to the time of failure of the steel plate alone (curve A). As expected, it implies that the thermoset resin does not provide any protective effect. This organic resin can in fact easily initiate or propagate fire, because it can decompose to yield volatile combustibles when exposed to heat. For this reason flame retardants have to be added into the polymer. APP is a high molecular weight chain phosphate. It is an interesting component because it can be used both as acid source and blowing agent: it is a source of phosphoric acid which speeds up the formation of carbonaceous char and of NH3 which improves the swelling [11–13]. When APP derivative is added to the thermoset resin (curve C), an improvement in performance is observed (time of failure of 11.3 min compared with 5 min for the uncoated steel). Intumescence and charring take place, but the char falls off the plate before the end of the experiment (change of slope at 610 °C). Borax and boric acid are well established as flame retardants and zinc borates have emerged as a replacement for antimony oxides in halogenated fire retardant polymers [8,14]. Addition of boric acid (curve D) to the resin also leads to improved performance, the time of failure is increased to 18.2 min. Development of intumescence is also observed, however the char falls off the plate (rapid change of slope at 400 °C). These falls could be explained by a loss of adherence of the coating on the plate, or by a loss of cohesion of the char or also by the effect of gravity, since the tests are carried out vertically. The best result is obtained when both the APP derivative and boric acid are added to the resin (curve E). The time of failure increases up to 29.5 min and the char adheres to the plate. The results show that the use of only one fire retardant additive (APP derivative or boric acid) leads to a significant increase in the time to failure. The problem is that the tests are carried out on vertically mounted plates and the char must adhere to the plate to provide the required protection. APP derivative or
Fig. 3. Steel plates coated before the test and after the test.
982
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
Fig. 4. Photos of the small scale test using a Bunsen burner, before and during the test.
boric acid, when incorporated in the resin, does not produce chars with the required adhesion. However, the combination of the two fire retardants lead to a high time of failure (29.5 min regard to 18.2 min) and the resulting intumescent char adheres strongly to the plate exhibiting a regular hemisphere shape. In order to further investigate the behavior of the intumescent char and to try to understand why certain chars did not remain stuck to the plate, bunsen burner tests were carried out. In all cases, no deformation of the substrate (steel plate) has been observed. Three plates were evaluated: thermoset resin + boric acid, thermoset resin + APP derivative, thermoset resin + boric acid + APP derivative. The results for the formulations thermoset resin + boric acid, thermoset resin + APP derivative obtained are shown in Fig. 6: – The char of the thermoset resin + APP derivative (Fig. 6(a)) does not adhere to the plate because it is light and crumbly. However, as some coating remains on the plate cohesive rather than adhesive failure appears to be occurring. It suggests that phosphates provide few adhesion of the coating on the plate. – The char developed by the combination of thermoset resin + boric acid (Fig. 6(b)) did not adhere.
1000
Temperature (°C)
800
The information obtained from these small scale tests is in agreement with the results of the industrial furnace tests. The main conclusion is that the combination “phosphate and boric acid” is necessary to allow the formation of species which promote both intumescence and adhesion of the intumesced coating to the steel plate. The furnace tests are interesting in terms of observation of what happens during hydrocarbon fires, but they do not give further information on the effect of the additives on the mechanical properties, thermal degradation and stability, expansion, viscosity, nor on the possible interactions between the components. That is why the same formulations will be investigated in the next part using small scale tests (TGA, TSR, etc.), in order to explain better their behavior in a fire. 3.2. Thermal degradation
D
TG curves of the thermoset resin (A), the thermoset resin + APP derivative (B), the thermoset resin + boric acid (C) and the thermoset resin + boric acid + APP derivative (D) are presented
C
A
– The char developed by the ternary system thermoset resin + boric acid + APP derivative (not presented) did not fall off the plate. It seems that this char combines the positive effect of phosphates promoting adhesion to steel and of borates which produces a very hard char. It may also suggest that interactions and/or reactions might occur between phosphates and borates leading to the formation of species promoting the adhesion of the intumescent coating.
B
600
E
400
200
0 0
10
20
30
40
Time (min) Fig. 5. Evolution of temperature as a function of time on the back side of a steel plate of different intumescent coatings.
Fig. 6. Residues obtained after the Bunsen burner test: (a): Thermoset resin + APP derivative, (b): Thermoset resin + boric acid.
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
in Fig. 7 and the thermal stability characteristics are presented in Table 1. The degradation of the resin (curve A) occurs in two main steps: a first step between 300 and 460 °C, corresponding to about 20 wt.% residual weight and a second step between 460 and 600 °C leading to the total degradation of the resin. The mechanism of thermal degradation of the cured epoxy resin has been studied by different authors. According to Grassie et al. [15] in terms of bond energies, the weakest points in the crosslinked resin network should be the N\C and the O\CH2 bonds. However, the high concentration of hydroxyl groups in the cured resin favors localised intramolecular hydrogen bonding which would promote dehydration. To allow a cured epoxy resin to degrade there must be an initial rupture of bonds. Keenan and Smith [16] suggested that this rupture is primarily homolytic, to produce free radicals which undergo further reactions that are expected to be of a complex nature. They postulated that primary degradation occurs at the N\C6H5 bond and most likely at the C6H5\C(CH3)2\C6H5 bonds. This probably occurs by homolytic fission. Complex secondary reactions will then occur whose nature depends on temperature. One of the initial reactions may be dehydration, as it has been found in other studies. Subsequent complex reactions would yield to the formation of small molecules such as phenols or cresols found during degradation of cured epoxy. Similar results were reported by Peterson-Jones et al. [17] who found evidence that the degradation of amine-cured epoxies involved the homolytic scission of aliphatic bonds and that dehydration and a specific reaction to form methylanilino and benzofuril (Fig. 8) structures possibly preceded degradation to a large extent. The degradation of the mixture thermoset resin + APP derivative (curve B) occurs in four main steps: one step between 200 and 350 °C corresponding to a residual weight of 55 wt.%. This step may partly correspond to the release of ammonia from the ammonium polyphosphate: in fact, APP begins to lose ammonia (Fig. 9) at a temperature slightly above 200 °C (Fig. 10), resulting in a highly condensed polyphosphoric acid [18]. This step also corresponds to the degradation of the agent of coating
Residual weight (wt.-%)
100
80
D 60
C B
40
A
20
983
Table 1 Thermal stability of the mixtures
First step
Tonset ° (°C) Tend (°C) Residual wt. (%) at Tend Second Tonset ° step (°C) Tend (°C) Residual wt. (%) at Tend Third Tonset ° step (°C) Tend (°C) Residual wt. (%) at Tend Fourth Tonset ° step (°C) Tend (°C) Residual wt. (%) at Tend
Thermoset Thermoset resin (A) resin + APP derivative (B)
Thermoset resin + Boric acid (C)
Thermoset resin + boric acid + APP derivative (D)
300
200
100
100
460
350
150
150
20
55
93
93
460
350
150
150
600
450
180
180
0
35
88
85
450
220
220
720
600
500
25
50
60
720
600
500
800
800
800
15
40
35
of the APP derivative, containing nitrogenous and carbonaceous species. A second step occurs between 350 and 450 °C corresponding to a residual weight of about 35 wt.%. Then, a thermally stable material is formed that degrades from 450 °C up to 800 °C in two steps and leads to a residue of about 15 wt.%. Details on interactions between the thermoset resin and APP will be given in a further paper. The degradation of the mixture thermoset resin + boric acid (curve C) occurs in four main steps: two steps overlap between 100 and 180 °C corresponding to a residual weight of 12 wt.%. Then, a thermally stable material is formed that degrades slowly in two steps from 220 °C up to 800 °C. The first two steps of degradation can be assigned to the degradation of boric acid (Fig. 11). The first step of degradation may be attributed to the dehydration of boric acid (H3BO3, M = 62 g/Mol) into metaboric acid (HBO2, M = 44 g/Mol). It is consistent with the mass loss
O
O O
0 0
200
400
600
800
Temperature (°C) Fig. 7. Thermogravimetric analyses.
Fig. 8. Exemple of benzofuril structure.
984
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
O
O
O
O
P
P
P
P
NH 4+ O
O NH 4+
NH 3OH
OHNH 3
-NH 3
O
O
P
P
O
O
H
H NH 3
Fig. 9. Mechanism of degradation of APP.
measured (about 30 wt.%). The second step corresponds to the dehydration of HBO2 into a very hard mineral glass: boron oxide, or B2O3 (59 g/Mol). These crystals begin to break down at 300 °C, and a series of suboxides are produced with partial melting until full fusion is reached at 700 °C. The degradation of the thermoset resin mixed with both additives (curve D) occurs in four steps: two steps overlap between 100 and 180 °C corresponding to a weight loss of 15 wt. %, due to the dehydration of boric acid. A third step occurs between 180 and 500 °C, corresponding to a weight loss of about 25 wt.%. Then, a fourth step, occurring between 500 and 800 °C, leads to a final residue of about 35 wt.%. Additives can interact with the epoxy resin during the degradation process. The difference of weight loss between the experimental and the theoretical TG curves (see part 2.2) has been presented for the three mixtures in Fig. 12. The difference between the experimental and the theoretical TG curves gives information on the reactivity of the resin with APP, boric acid or with both. When the experimental curve is higher than the theoretical one (or when the difference weight loss curve is positive), the loss of weight is lower than expected showing that the reactivity of the resin with the additive leads to a thermal stabilization of the materials. If the experimental curve is lower than the theoretical one (or when the difference weight loss curve is negative), the reactivity of the resin with the additive leads to a thermal destabilization of the materials. The thermoset resin is totally degraded at 600 °C. The comparison of the experimental curve with the calculated curve for the mixture A (thermoset resin containing APP derivative) shows an important decrease of about 20 wt.% in the thermal stability between 300 and 400 °C. This destabilisation suggests a reaction between the additive and the polymeric matrix
3.3. Complex viscosity In order to build on the previous observations, visco-elastic measurements have been carried out and the results are described in the next section. The formation of the effective char occurs via a semi-liquid phase, which coincides with gas 100 90
80
Residual weight (wt.-%)
Residual weight (wt.-%)
100
leading to the formation of volatiles. An increase in thermal stability is then observed from 500 to 800 °C with an increase of 10 wt.% compared to the result expected if no reaction happened. The chemical interactions between the two components lead to the formation of a high thermal stability material. The comparison between the experimental TG curves with the calculated TG curve for mixture B (epoxy resin containing boric acid) shows a slight decrease in the thermal stability between 80 and 130 °C and between 230 and 330 °C but a large increase in thermal stability is observed from 330 to 800 °C. The residue at 800 °C is about 17 wt.% higher than what it would be expected. The two steps corresponding to the dehydration of boric acid occur at lower temperature when incorporated into the epoxy resin and chemical interactions between the two components lead to the formation of a high thermal stability material. The comparison between the experimental TG curves with the calculated TG curve for the mixture C (epoxy resin containing boric acid and APP derivative) shows a very low decrease in thermal stability between 100 and 350 °C while thermal stabilization is observed from 350 °C to 800 °C. The residue at 800 °C is the same as expected by the calculation: it appears that the chemical interactions between the three components lead to the formation of a thermally stable material between 350 and 800 °C but the residue at 800 °C of the final products is as predicted.
60
40
20
80 70 60 50 40 30 20 10
0
0 0
200
400
600
Temperature (°C) Fig. 10. TGA of the APP derivative.
800
0
100
200
300
400
500
600
Temperature (°C) Fig. 11. TGA curve of boric acid.
700
800
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987 40 30
B Δ(T) (wt.-%)
20
A
10
C
0
-10
0
200
400
600
800
Temperature (°C)
-20 -30
Fig. 12. Curves of weight differences (difference between experimental curves and calculated curves).
formation and expansion of the surface [9]. Gases released from the degradation of the intumescent material, and in particular of the blowing agent, have to be trapped and to diffuse slowly in the highly viscous melt degraded material in order to create a layer with appropriate morphological properties as shown in Fig. 13. If the degraded matrix has a too low viscosity, easy diffusion of gases takes place and the gases will not be trapped but rather escape to feed the flame. The viscosity of the degraded matrix in the blowing phase is, as a consequence, a critical factor [19,20]. Fig. 14(a) and (b) compare the expansion and complex viscosity of the four formulations, the thermoset resin (A), the thermoset resin mixed with boric acid (B), the thermoset resin mixed with the APP derivative (C) and the thermoset resin mixed with both Boric acid and the APP derivative (D). For the four coatings, the viscosity remains reasonably steady until 300 °C. From 330 to 350 °C it decreases very rapidly to then increase again quite rapidly until 400 °C at higher temperatures a further slow increase in viscosity is observed. This increase of viscosity after 400 °C can be attributed to a carbonization process, the liquid phase disappears and a highly condensed char structure is produced. For the thermoset epoxy resin, the peak of decrease of viscosity corresponds to the main step of degradation of the resin, which loses 90% of its mass between 300 and 460 °C (as shown in Fig. 7). This phenomenon agrees with one of the main degradation mechanisms of the resin proposed in literature [9]: the degradation begins with cleavage of the N\C bonds, and this leads to a decrease in the crosslink density and therefore a decrease in the viscosity. Expansion is very small in the case of the unmodified resin. The material is quite totally degraded at 500 °C, there is only a small amount of residue observed at the end of the test. When boric acid is added to the resin (mixture B), the viscosity shows a decreasing peak but over a larger range of temperature [300–400 °C] compared with [330–380 °C] in the case of the resin alone. Moreover, the minimum viscosity is higher than the epoxy resin and there is an important expansion of the coating (about 120%). The boric acid acts as an intumescent agent. It is proposed that the formation of B2O3 due
985
to the dehydration of the boric acid leads to the formation of a “glass-like” material which increases the viscosity of the melt (compared with the unmodified resin) and prevents the gaseous decomposition products escaping to feed the flame. When the APP derivative is incorporated into the resin (mixture C), the viscosity decreases in a similar way to the resin alone, but over a larger temperature range [300–500 °C]. The TG curve (see Fig. 7) also shows evidence for degradation at a lower temperature when APP is added to the thermoset resin. During its degradation, as shown in Fig. 9, APP forms phosphoric acid [18] which leads to a decrease in melt viscosity due to the formation of a liquid phase. In addition, the released acid catalyses the degradation of the epoxy resin, leading to degradation at lower temperature. Expansion of the thermoset resin is significantly improved when the APP derivative is added. The APP additive degrades to yield ammonia [7,11] above 200 °C. There is a good correlation between the temperature of the onset of degradation observed by TG and temperature at which the expansion starts (T = 300 °C in both cases). The expansion is attributed to the evolution of volatile degradation products, which are trapped in the structure. APP incorporated into the resin leads to intumescence, even if the residue obtained at 500 °C is unfortunately extremely light and crispy. This, together with the non-controlled release of ammonia, leads to variable expansion results. The presence of both boric acid and the APP derivative in the thermoset resin (mixture D) leads to a higher expansion (140%) and the main peak of viscosity is similar to the peak corresponding to mixture B. The presence of the APP derivative does not seem to influence significantly the rheological behavior of mixture B. The residue obtained is solid and dense. The conditions are favorable to allow slow diffusion of the gaseous degradation products and effective intumescence. This phenomenon, together with a thermally stable char structure (Fig. 7), provides effective protection. To conclude this part, it is noteworthy that swelling starts at about 300 °C, suggesting that the expansion takes place because of the relatively low viscosity of the char combined with the release of volatile degradation products. It is also observed that all the mixtures that contain boric acid exhibit a lower decrease in viscosity. As a consequence, boric acid plays a key role in the intumescent behavior of the coating, acting both as an intumescent agent and char reinforcer. Finally, as the peak of viscosity of the mixture thermoset resin + boric acid + APP derivative is the smallest, the swelling is the highest because the
Fig. 13. Development of the intumescence (α = conversion degree).
986
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
(a)
(b)
D
1e+6
D
120 100
B
80 60
C
40 20
A
0 100
200
300
400
500
Complex viscosity (Pa.s)
relative % of swelling
140
1e+5 1e+4
C
1e+3
A
B 1e+2 1e+1 0
100
Temperature (°C)
200
300
400
500
Temperature (°C)
Fig. 14. (a) Expansion of the formulation. (b) Complex viscosity.
char can accommodate the stresses due to the internal pressure created by the evolving gases. However, an important parameter that also has to be taken into account is the mechanical properties of the char. If the char is destroyed when subjected to external perturbations such as explosion (e.g. Jet fire) or wind turbulence, the coating will not provide effective protection of the steel structure. 3.4. Mechanical resistance of the char In the stabilization phase of the intumescent structure, the change in the viscosity of the charred material under stress may explain the loss of the protective character of the intumescent shield. Indeed, if the shield becomes too hard, the creation and propagation of cracks leading to a rapid degradation of the material occurs. If a char has a good structural, morphological and heat insulative properties but is easily destroyed under a mechanical action, its efficiency is totally lost in the turbulent regime of combustion. Fig. 15 presents the destruction force of the carbonaceous char plotted against the distance between the plates for the following mixtures at 500 °C: the thermoset resin (A), the thermoset resin + APP derivative (B), the thermoset resin + boric
force (g) 1800 1600 1400
D
1200 1000 800 600 400
C
200 0 16
B 14
12
10
8
6
4
2
A 0
gap (mm) Fig. 15. Mechanical resistance at 500 °C of different formulations.
acid (C) and the thermoset resin mixed with both boric acid and APP (D). The curves corresponding to mixtures A and B are not very different: the char is not resistant and a very weak force is sufficient to destroy it. It is only when 1 mm of residue remains that the force increases, and this is due to the fact that the residue has been strongly compressed. The curves corresponding to mixtures C and D give the same results at the beginning: a very weak force is sufficient to destroy the char. But the force begins to increase earlier than in cases A and B; there is a difference of about 2 mm, which means that the char is harder. It appears that the presence of boric acid, and particularly the formation of B2O3, provides better mechanical resistance. 4. Conclusion The aim of this work was to study the efficiency of different epoxy resin based experimental intumescent formulations, designed for the protection of steel in the case of hydrocarbon fires. Tests in big industrial furnaces lead to time/temperature profiles of the different formulations in case of hydrocarbon fires. Ammonium polyphosphate and boric acid provide good performance when they are incorporated separately into the resin, but the intumesced char falls off the plate. Using a Bunsen burner test onto the plate, it is revealed that the intumesced char formed from the epoxy resin containing APP is light and blows away during the experiment. It was also shown that the adhesion of the coating to the steel plate is very weak when phosphorous species are not added into the formulation. The char developed from resin mixed with boric acid slides along and does not remain stuck on the plate. The best result was obtained when APP and boric acid were combined into the resin: the backside of the steel plate reaches 400 °C in 29.5 min regard to 4 min for the epoxy resin alone and the char remains well stuck on the plate. Nevertheless the furnace test does not provide explanations on how the formulation works. Thermogravimetric analyses show that additives interact with the thermoset resin to increase its thermal stability. It is suggested that a reaction occurs between boric acid and ammonium polyphosphate
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
derivative during heating, permitting the adhesion between the coating and the steel plate. Novel experiments developed, using a rheometer, show that the degradation of the resin at about 350 °C creates an important decrease of viscosity. Boric acid seems to be the key additive to reduce this fall of viscosity: in fact it turns by dehydration into boron oxide, a hard glass, which traps the gases, and allows a high expansion and a char with a good mechanical resistance. APP derivative is used both as blowing agent and acid source, as it releases ammonia which is trapped into the char structure to yield phosphoric acid. This mixture behaves like an intumescent system, but the char created is too light and exhibit poor mechanical resistance. The best result is obtained when both additives are combined into the resin: there is a small decrease of viscosity at 350 °C, a high expansion and the char has an appropriate mechanical resistance. References [1] [2] [3] [4]
J.A. Seiner, T.A. Ward, Polym. Paint Colour J. 178 (4207) (1988) 75. G.K. Castle, Loss Prev. 8 (1974) 57. S.C. Upadhya, Paintindia 50 (2000) 45. S. Duquesne, S. Magnet, C. Jama, R. Delobel, Surf. Coat. Technol. 180–181 (2004) 302.
987
[5] S. Bourbigot, M. Le Bras, S. Duquesne, M. Rochery, Macromol. Mater. Eng. 289 (2004) 499. [6] R. Delobel, M. Le Bras, N. Ouassou, F. Alistiqsa, J. Fire Sci. 8 (1990) 85. [7] G. Camino, G. Martinasso, L. Costa, Polym. Degrad. Stab. 27 (1990) 285. [8] H.L. Vandersall, J. Fire Flammabl. 2 (1971) 97. [9] A.R. Horrocks, Polym. Degrad. Stab. 54 (2–3) (1996) 143. [10] R. Delobel, M. Le Bras, N. Ouassou, F. Alistiqsa, J. Fire Sci. 8 (2) (1991) 85. [11] C.F. Cullis, M.M. Hirschler, The Combustion of Organic Polymers, Clarendon Press, Oxford University Press, Oxford, 1981. [12] T. Arthur, K. Quill, Proceedings of Flame Retardant'92, Elsevier, London, 1992, p. 223. [13] W. Lyons, The Chemistry and Uses of fire Retardants, Wiley-Interscience, John Wiley and Sons. Inc. New York, 1970. [14] K.K. Shen, E. Olson, Proceedings of the Conference on Recent Advances in Flame Retardancy of Polymeric Materials 16 (2005) 159. [15] N. Grassie, M.I. Guy, N.H. Tennent, Polym. Degrad. Stab. 14 (1986) 125. [16] M.A. Keenan, D.A. Smith, J. Appl. Polym. Sci. 11 (7) (1967) 1009. [17] J.C. Peterson-Jones, J.A. Percy, R.G.F. Giles, A.M. Stephen, J. Appl. Polym. Sci. 17 (1973) 1877. [18] G. Camino, M.P. Luda, in: G. Camino, M. le Bras, S. Bourbigot, R. Delobel (Eds.), Fire Retardancy of Polymers—The Use of Intumescence, Special Publication, vol. 224, Royal Society of Chemistry, Cambridge, 1998, p. 48. [19] M. Bugajny, M. Le Bras, S. Bourbigot, Fire Mater. 23 (1) (1999) 49. [20] P. Anna, G. Marosi, I. Csontos, S. Bourbigot, M. Le Bras, R. Delobel, Polym. Degrad. Stab. 74 (3) (2001) 423.
Characterization of the performance of an intumescent fire protective coating M. Jimenez, S. Duquesne ⁎, S. Bourbigot Laboratoire des Procédés d'Elaboration de Revêtements Fonctionnels (UPRES EA 1040), Ecole Nationale Supérieure de Chimie de Lille, BP90108, F-59652 Villeneuve d'Ascq, France Received 24 August 2005; accepted in revised form 5 January 2006 Available online 28 February 2006
Abstract The aim of this work is to study the efficiency of different intumescent formulations designed to protect steel in the case of hydrocarbon fire. The coating is based on a thermoset epoxy–amine resin system into which fire retardant agents, boric acid and ammonium polyphosphate derivative have been incorporated. The first part of the study evaluates, using large scale industrial furnace tests, the behavior of the thermoset resin containing alone or in combination with additives. It is revealed that in this epoxy resin, the combination between ammonium polyphosphate and boric acid leads to the best protective results. The second part of the study attempts to investigate more precisely the effect and the mode of action of the additives in terms of thermal stability, mechanical resistance and rheological properties using small scale lab tests, to explain why this combination works better than using the two fire retardants used separately. The experiments show that this combination leads to the smallest decrease of viscosity when the resin degrades, the highest mechanical resistance and the highest expansion. © 2006 Elsevier B.V. All rights reserved. Keywords: Intumescence; Fire protection; Structural steel; Ammonium polyphosphate; Boric acid; Epoxy resin
1. Introduction Structural steel loses an appreciable part of its load carrying ability when its temperature exceeds 500 °C. The protection of metallic materials against fire has become an important issue in the construction industry. Indeed, prevention of the structural collapse of the building is paramount to ensure the safe evacuation of people from the building, and is a prime requirement of building regulations in many countries. Intumescent coatings are designed to perform under severe conditions and to maintain the steel integrity between 1 and 3 h when the temperature of the surroundings is in excess of 1100 °C [1–4]. Intumescent coatings represent an increasingly used way to provide passive fire protection to the structural steel that is more and more used in modern architectural designs, whilst at the same time maintaining aesthetic appearance. Intumescence is defined as the swelling of certain substances when they are heated. Intumescent coatings form on heating an expanded multicellular layer, which acts as a thermal barrier that effectively protects the substrate against rapid increase of tem⁎ Corresponding author. Tel.: +33 3 2033 7236; fax: +33 3 2043 6584. E-mail address: [email protected] (S. Duquesne). 0257-8972/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.surfcoat.2006.01.026
perature, thereby maintaining the structural integrity of the building. Intumescent coatings contain “active” ingredients bound together by a binder. Generally, three “active” ingredients are used: an acid source (normally ammonium polyphosphate or a mineral acid), a carbon source (such as char forming polymers or polyols) and a blowing agent (e.g. melamine). The formulation of the coating has to be optimized in terms of physical and chemical properties in order to form an effective protective char [5]. Chemical interactions between the “active” ingredients in the formulation lead to the formation of the intumescent char. It is generally accepted [6–9] that first, the acid source breaks down to yield a mineral acid, then it takes part in the dehydration of the carbonization source to yield the carbon char and finally, the blowing agent decomposes to yield gaseous products. The latter cause the char to swell and produce the insulating multi-cellular protective layer. This protective char limits both the heat transfer from the heat source to the substrate and the mass transfer from the substrate to the heat source, resulting in conservation of the underlying material. In this study, the efficiency of intumescent coatings on steel is investigated using a fire test (measurement of the temperature profile on the backside of a coated steel plate when exposed to a hydrocarbon fire) as well as laboratory test to measure the heat
980
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
resistance and its viscoelastic behavior. The combination of different intumescent ingredients in a thermoset epoxy resin is examined. 2. Experimental
& & &
2.1. Materials The binder used is a solvent free thermoset epoxy resin. It is a mixture of the Diglycidylether of Bisphenol A (DGEBA) and an amine curing agent (Polyaminoamide). The resin is cured at ambient temperature. Two main fire retarding agents have been chosen in this study:
&
Madd(T): values of weight given by the TG curve of the additives (APP or boric acid), Mexp(T): values of weight given by the TG curve of the formulation (polymer + additives) that is to say APP or boric acid or both, Mth(T): theoretical TG curve computed by linear combination between the values of weight given by the TG curve of the polymer and of the additives: Mth(T) = (1 − x)Mpoly(T) + xMAPP(T); Mth(T) = (1 − y)Mpoly(T) + yMBoricAcid(T); Mth(T) = (1 − x′ − y′)Mpoly(T) + x′MAPP(T) + y′MBoricAcid(T) where x and x′ are the APP content of the formulations including APP and y and y′ are the boric acid content of the formulations including APP, Δ(T): weight difference curve: Δ(T) = Mexp(T) − Mthe(T).
– A mineral acid: Boric acid (H3BO3) (Aldrich, p.99%). – A commercial Ammonium Polyphosphate (APP) derivative (Fig. 1) supplied by Clariant (Germany). It is an APP coated with a component containing nitrogenous and carbonaceous species.
The Δ(T) curves allow us to show a potential increase or decrease in the thermal stability of the polymeric matrix related to the presence of one or every additive [10].
Four formulations have been studied in this work: the thermoset resin alone, the thermoset resin containing boric acid, the thermoset resin containing the commercial APP derivative and the thermoset resin containing both the APP derivative and boric acid. The formulations were prepared by mixing the components using an Ultra Turrax mixer (600 rpm). For the fire tests the coating was applied at 3.5 mm thickness onto grit blasted steel plates of dimensions 30 × 30 cm2 or 15 × 15 cm2. Free films of the coating and pellets of dimensions 25 × 25 × 3 mm3 (for the rheological measurements) were produced by casting the coating between PTFE plates. For the thermogravimetric analysis the cured coating was ground in liquid nitrogen in an ultra centrifuge mill to produce a fine powder.
Rheological measurements were carried out using a Rheometric Scientific ARES-20A Thermal Scanning Rheometer (TSR) in a parallel plate configuration. Both the viscosity of the coating and its mechanical resistance were measured.
2.3. Rheological measurements
2.3.1. Complex viscosity measurements The TSR was designed for monitoring changes in the rheological properties as a function of temperature and/or time. Samples (25 × 25 × 3 mm3) were positioned between the two plates. A constant normal force (F = 200 Pa) was systematically applied in order to obtain good adhesion between samples and plates, and also to ensure the validity of the results. The program used is a “Dynamic Temperature Ramp Test”: heating program with a heating rate of 10 °C/min in the range 25–500 °C, a strain of 1% and a constant normal force of 10 g (200 Pa).
2.2. Thermogravimetric analysis Thermogravimetric analyses were carried out at 10 °C/min under synthetic air (flow rate: 50 mL/min, Air Liquide grade), using a Setaram TG 92 microbalance. The samples (approx. 10 mg) in powder form were placed in open vitreous silica pans. The precision of the temperature measurements was 1.5 °C over the whole range of temperature (20–800 °C). The curves of weight difference between the experimental and theoretical curves were computed as follows:
&
Mpoly(T): values of weight given by the TG curve of the polymer,
2.3.2. Mechanical resistance Mechanical resistance is evaluated using the protocol described as follows: at t = 0 s, the sample (height, h = 1 mm) is put into the furnace and heated to 500 °C without applying any strain (the upper plate is not in contact with the sample as shown in Fig. 2. This allows the sample to intumesce without any constraint. The upper plate is then brought into contact with the intumesced material and the separation between the plates is reduced linearly (0.02 mm/s). The force is followed as a function of the separation between the two plates (Fig. 2). The upper plate used in this experiment has a diameter of 5 mm in order to increase the pressure on the whole sample and to ensure complete destruction of the char.
O P
2.4. Fire resistance
O n
O N H 4+ Fig. 1. Ammonium polyphosphate.
Industrial furnace tests have been carried by our Industrial sponsor. It is a demonstration test (OTI 95 634) and the specification is to burn a given volume of propane (0.3 kg/s) at a given heat flux (200–250 kW/m2) and at a given distance (1 m)
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
F=0 N
F=0 N
F sample sampl
sample
Z=0 Z
T°C
t = 0sec
T°C
T°C
t = 240sec
t > 240sec
Fig. 2. Test of mechanical resistance.
from the test piece. The burning conditions fit as close as possible the ramp of temperature of a hydrocarbon fire heating curve (about 200 °C/min). The coatings are applied on a steel plate (thickness 3.5 mm) and cured for 1 week under ambient conditions. Thermocouples are attached to the backside of the coated plates. Five thermocouples are used on each plate, so that an average temperature can be obtained. As four plates are tested at the same time the plates are isolated using glass wool. The plates are mounted vertically in the furnace and burnt (Fig. 3) until the thermocouples attached to the backside of the plates reach a temperature higher than 400 °C (the failure temperature). This test produces time/temperature curves and characterizes the heat protective effect of the different coatings in a hydrocarbon fire. 2.5. Bunsen burner test This test has been used to characterize some properties of the char. A high temperature (about 1000 °C) is applied with a Bunsen burner on to a coated plate mounted vertically as shown in Fig. 4. 3. Results and discussion 3.1. Fire protection Fig. 5 shows the evolution of temperature as a function of time on the backside of the steel plates coated with the different
981
formulations. Steel usually loses its main structural properties at around 500 °C. In this case, for safety reasons and because the thermocouple is on the backside of the steel plate, 400 °C has been chosen as failure temperature. Four formulations were chosen and compared to the virgin steel plate (A): the thermoset resin alone (B), the thermoset resin mixed with the APP derivative (C), the thermoset resin mixed with boric acid (D) and the thermoset resin mixed with both (E). The time of failure (when temperature reaches 400 °C) of the steel plate covered with the thermoset resin (curve B) is close to the time of failure of the steel plate alone (curve A). As expected, it implies that the thermoset resin does not provide any protective effect. This organic resin can in fact easily initiate or propagate fire, because it can decompose to yield volatile combustibles when exposed to heat. For this reason flame retardants have to be added into the polymer. APP is a high molecular weight chain phosphate. It is an interesting component because it can be used both as acid source and blowing agent: it is a source of phosphoric acid which speeds up the formation of carbonaceous char and of NH3 which improves the swelling [11–13]. When APP derivative is added to the thermoset resin (curve C), an improvement in performance is observed (time of failure of 11.3 min compared with 5 min for the uncoated steel). Intumescence and charring take place, but the char falls off the plate before the end of the experiment (change of slope at 610 °C). Borax and boric acid are well established as flame retardants and zinc borates have emerged as a replacement for antimony oxides in halogenated fire retardant polymers [8,14]. Addition of boric acid (curve D) to the resin also leads to improved performance, the time of failure is increased to 18.2 min. Development of intumescence is also observed, however the char falls off the plate (rapid change of slope at 400 °C). These falls could be explained by a loss of adherence of the coating on the plate, or by a loss of cohesion of the char or also by the effect of gravity, since the tests are carried out vertically. The best result is obtained when both the APP derivative and boric acid are added to the resin (curve E). The time of failure increases up to 29.5 min and the char adheres to the plate. The results show that the use of only one fire retardant additive (APP derivative or boric acid) leads to a significant increase in the time to failure. The problem is that the tests are carried out on vertically mounted plates and the char must adhere to the plate to provide the required protection. APP derivative or
Fig. 3. Steel plates coated before the test and after the test.
982
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
Fig. 4. Photos of the small scale test using a Bunsen burner, before and during the test.
boric acid, when incorporated in the resin, does not produce chars with the required adhesion. However, the combination of the two fire retardants lead to a high time of failure (29.5 min regard to 18.2 min) and the resulting intumescent char adheres strongly to the plate exhibiting a regular hemisphere shape. In order to further investigate the behavior of the intumescent char and to try to understand why certain chars did not remain stuck to the plate, bunsen burner tests were carried out. In all cases, no deformation of the substrate (steel plate) has been observed. Three plates were evaluated: thermoset resin + boric acid, thermoset resin + APP derivative, thermoset resin + boric acid + APP derivative. The results for the formulations thermoset resin + boric acid, thermoset resin + APP derivative obtained are shown in Fig. 6: – The char of the thermoset resin + APP derivative (Fig. 6(a)) does not adhere to the plate because it is light and crumbly. However, as some coating remains on the plate cohesive rather than adhesive failure appears to be occurring. It suggests that phosphates provide few adhesion of the coating on the plate. – The char developed by the combination of thermoset resin + boric acid (Fig. 6(b)) did not adhere.
1000
Temperature (°C)
800
The information obtained from these small scale tests is in agreement with the results of the industrial furnace tests. The main conclusion is that the combination “phosphate and boric acid” is necessary to allow the formation of species which promote both intumescence and adhesion of the intumesced coating to the steel plate. The furnace tests are interesting in terms of observation of what happens during hydrocarbon fires, but they do not give further information on the effect of the additives on the mechanical properties, thermal degradation and stability, expansion, viscosity, nor on the possible interactions between the components. That is why the same formulations will be investigated in the next part using small scale tests (TGA, TSR, etc.), in order to explain better their behavior in a fire. 3.2. Thermal degradation
D
TG curves of the thermoset resin (A), the thermoset resin + APP derivative (B), the thermoset resin + boric acid (C) and the thermoset resin + boric acid + APP derivative (D) are presented
C
A
– The char developed by the ternary system thermoset resin + boric acid + APP derivative (not presented) did not fall off the plate. It seems that this char combines the positive effect of phosphates promoting adhesion to steel and of borates which produces a very hard char. It may also suggest that interactions and/or reactions might occur between phosphates and borates leading to the formation of species promoting the adhesion of the intumescent coating.
B
600
E
400
200
0 0
10
20
30
40
Time (min) Fig. 5. Evolution of temperature as a function of time on the back side of a steel plate of different intumescent coatings.
Fig. 6. Residues obtained after the Bunsen burner test: (a): Thermoset resin + APP derivative, (b): Thermoset resin + boric acid.
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
in Fig. 7 and the thermal stability characteristics are presented in Table 1. The degradation of the resin (curve A) occurs in two main steps: a first step between 300 and 460 °C, corresponding to about 20 wt.% residual weight and a second step between 460 and 600 °C leading to the total degradation of the resin. The mechanism of thermal degradation of the cured epoxy resin has been studied by different authors. According to Grassie et al. [15] in terms of bond energies, the weakest points in the crosslinked resin network should be the N\C and the O\CH2 bonds. However, the high concentration of hydroxyl groups in the cured resin favors localised intramolecular hydrogen bonding which would promote dehydration. To allow a cured epoxy resin to degrade there must be an initial rupture of bonds. Keenan and Smith [16] suggested that this rupture is primarily homolytic, to produce free radicals which undergo further reactions that are expected to be of a complex nature. They postulated that primary degradation occurs at the N\C6H5 bond and most likely at the C6H5\C(CH3)2\C6H5 bonds. This probably occurs by homolytic fission. Complex secondary reactions will then occur whose nature depends on temperature. One of the initial reactions may be dehydration, as it has been found in other studies. Subsequent complex reactions would yield to the formation of small molecules such as phenols or cresols found during degradation of cured epoxy. Similar results were reported by Peterson-Jones et al. [17] who found evidence that the degradation of amine-cured epoxies involved the homolytic scission of aliphatic bonds and that dehydration and a specific reaction to form methylanilino and benzofuril (Fig. 8) structures possibly preceded degradation to a large extent. The degradation of the mixture thermoset resin + APP derivative (curve B) occurs in four main steps: one step between 200 and 350 °C corresponding to a residual weight of 55 wt.%. This step may partly correspond to the release of ammonia from the ammonium polyphosphate: in fact, APP begins to lose ammonia (Fig. 9) at a temperature slightly above 200 °C (Fig. 10), resulting in a highly condensed polyphosphoric acid [18]. This step also corresponds to the degradation of the agent of coating
Residual weight (wt.-%)
100
80
D 60
C B
40
A
20
983
Table 1 Thermal stability of the mixtures
First step
Tonset ° (°C) Tend (°C) Residual wt. (%) at Tend Second Tonset ° step (°C) Tend (°C) Residual wt. (%) at Tend Third Tonset ° step (°C) Tend (°C) Residual wt. (%) at Tend Fourth Tonset ° step (°C) Tend (°C) Residual wt. (%) at Tend
Thermoset Thermoset resin (A) resin + APP derivative (B)
Thermoset resin + Boric acid (C)
Thermoset resin + boric acid + APP derivative (D)
300
200
100
100
460
350
150
150
20
55
93
93
460
350
150
150
600
450
180
180
0
35
88
85
450
220
220
720
600
500
25
50
60
720
600
500
800
800
800
15
40
35
of the APP derivative, containing nitrogenous and carbonaceous species. A second step occurs between 350 and 450 °C corresponding to a residual weight of about 35 wt.%. Then, a thermally stable material is formed that degrades from 450 °C up to 800 °C in two steps and leads to a residue of about 15 wt.%. Details on interactions between the thermoset resin and APP will be given in a further paper. The degradation of the mixture thermoset resin + boric acid (curve C) occurs in four main steps: two steps overlap between 100 and 180 °C corresponding to a residual weight of 12 wt.%. Then, a thermally stable material is formed that degrades slowly in two steps from 220 °C up to 800 °C. The first two steps of degradation can be assigned to the degradation of boric acid (Fig. 11). The first step of degradation may be attributed to the dehydration of boric acid (H3BO3, M = 62 g/Mol) into metaboric acid (HBO2, M = 44 g/Mol). It is consistent with the mass loss
O
O O
0 0
200
400
600
800
Temperature (°C) Fig. 7. Thermogravimetric analyses.
Fig. 8. Exemple of benzofuril structure.
984
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
O
O
O
O
P
P
P
P
NH 4+ O
O NH 4+
NH 3OH
OHNH 3
-NH 3
O
O
P
P
O
O
H
H NH 3
Fig. 9. Mechanism of degradation of APP.
measured (about 30 wt.%). The second step corresponds to the dehydration of HBO2 into a very hard mineral glass: boron oxide, or B2O3 (59 g/Mol). These crystals begin to break down at 300 °C, and a series of suboxides are produced with partial melting until full fusion is reached at 700 °C. The degradation of the thermoset resin mixed with both additives (curve D) occurs in four steps: two steps overlap between 100 and 180 °C corresponding to a weight loss of 15 wt. %, due to the dehydration of boric acid. A third step occurs between 180 and 500 °C, corresponding to a weight loss of about 25 wt.%. Then, a fourth step, occurring between 500 and 800 °C, leads to a final residue of about 35 wt.%. Additives can interact with the epoxy resin during the degradation process. The difference of weight loss between the experimental and the theoretical TG curves (see part 2.2) has been presented for the three mixtures in Fig. 12. The difference between the experimental and the theoretical TG curves gives information on the reactivity of the resin with APP, boric acid or with both. When the experimental curve is higher than the theoretical one (or when the difference weight loss curve is positive), the loss of weight is lower than expected showing that the reactivity of the resin with the additive leads to a thermal stabilization of the materials. If the experimental curve is lower than the theoretical one (or when the difference weight loss curve is negative), the reactivity of the resin with the additive leads to a thermal destabilization of the materials. The thermoset resin is totally degraded at 600 °C. The comparison of the experimental curve with the calculated curve for the mixture A (thermoset resin containing APP derivative) shows an important decrease of about 20 wt.% in the thermal stability between 300 and 400 °C. This destabilisation suggests a reaction between the additive and the polymeric matrix
3.3. Complex viscosity In order to build on the previous observations, visco-elastic measurements have been carried out and the results are described in the next section. The formation of the effective char occurs via a semi-liquid phase, which coincides with gas 100 90
80
Residual weight (wt.-%)
Residual weight (wt.-%)
100
leading to the formation of volatiles. An increase in thermal stability is then observed from 500 to 800 °C with an increase of 10 wt.% compared to the result expected if no reaction happened. The chemical interactions between the two components lead to the formation of a high thermal stability material. The comparison between the experimental TG curves with the calculated TG curve for mixture B (epoxy resin containing boric acid) shows a slight decrease in the thermal stability between 80 and 130 °C and between 230 and 330 °C but a large increase in thermal stability is observed from 330 to 800 °C. The residue at 800 °C is about 17 wt.% higher than what it would be expected. The two steps corresponding to the dehydration of boric acid occur at lower temperature when incorporated into the epoxy resin and chemical interactions between the two components lead to the formation of a high thermal stability material. The comparison between the experimental TG curves with the calculated TG curve for the mixture C (epoxy resin containing boric acid and APP derivative) shows a very low decrease in thermal stability between 100 and 350 °C while thermal stabilization is observed from 350 °C to 800 °C. The residue at 800 °C is the same as expected by the calculation: it appears that the chemical interactions between the three components lead to the formation of a thermally stable material between 350 and 800 °C but the residue at 800 °C of the final products is as predicted.
60
40
20
80 70 60 50 40 30 20 10
0
0 0
200
400
600
Temperature (°C) Fig. 10. TGA of the APP derivative.
800
0
100
200
300
400
500
600
Temperature (°C) Fig. 11. TGA curve of boric acid.
700
800
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987 40 30
B Δ(T) (wt.-%)
20
A
10
C
0
-10
0
200
400
600
800
Temperature (°C)
-20 -30
Fig. 12. Curves of weight differences (difference between experimental curves and calculated curves).
formation and expansion of the surface [9]. Gases released from the degradation of the intumescent material, and in particular of the blowing agent, have to be trapped and to diffuse slowly in the highly viscous melt degraded material in order to create a layer with appropriate morphological properties as shown in Fig. 13. If the degraded matrix has a too low viscosity, easy diffusion of gases takes place and the gases will not be trapped but rather escape to feed the flame. The viscosity of the degraded matrix in the blowing phase is, as a consequence, a critical factor [19,20]. Fig. 14(a) and (b) compare the expansion and complex viscosity of the four formulations, the thermoset resin (A), the thermoset resin mixed with boric acid (B), the thermoset resin mixed with the APP derivative (C) and the thermoset resin mixed with both Boric acid and the APP derivative (D). For the four coatings, the viscosity remains reasonably steady until 300 °C. From 330 to 350 °C it decreases very rapidly to then increase again quite rapidly until 400 °C at higher temperatures a further slow increase in viscosity is observed. This increase of viscosity after 400 °C can be attributed to a carbonization process, the liquid phase disappears and a highly condensed char structure is produced. For the thermoset epoxy resin, the peak of decrease of viscosity corresponds to the main step of degradation of the resin, which loses 90% of its mass between 300 and 460 °C (as shown in Fig. 7). This phenomenon agrees with one of the main degradation mechanisms of the resin proposed in literature [9]: the degradation begins with cleavage of the N\C bonds, and this leads to a decrease in the crosslink density and therefore a decrease in the viscosity. Expansion is very small in the case of the unmodified resin. The material is quite totally degraded at 500 °C, there is only a small amount of residue observed at the end of the test. When boric acid is added to the resin (mixture B), the viscosity shows a decreasing peak but over a larger range of temperature [300–400 °C] compared with [330–380 °C] in the case of the resin alone. Moreover, the minimum viscosity is higher than the epoxy resin and there is an important expansion of the coating (about 120%). The boric acid acts as an intumescent agent. It is proposed that the formation of B2O3 due
985
to the dehydration of the boric acid leads to the formation of a “glass-like” material which increases the viscosity of the melt (compared with the unmodified resin) and prevents the gaseous decomposition products escaping to feed the flame. When the APP derivative is incorporated into the resin (mixture C), the viscosity decreases in a similar way to the resin alone, but over a larger temperature range [300–500 °C]. The TG curve (see Fig. 7) also shows evidence for degradation at a lower temperature when APP is added to the thermoset resin. During its degradation, as shown in Fig. 9, APP forms phosphoric acid [18] which leads to a decrease in melt viscosity due to the formation of a liquid phase. In addition, the released acid catalyses the degradation of the epoxy resin, leading to degradation at lower temperature. Expansion of the thermoset resin is significantly improved when the APP derivative is added. The APP additive degrades to yield ammonia [7,11] above 200 °C. There is a good correlation between the temperature of the onset of degradation observed by TG and temperature at which the expansion starts (T = 300 °C in both cases). The expansion is attributed to the evolution of volatile degradation products, which are trapped in the structure. APP incorporated into the resin leads to intumescence, even if the residue obtained at 500 °C is unfortunately extremely light and crispy. This, together with the non-controlled release of ammonia, leads to variable expansion results. The presence of both boric acid and the APP derivative in the thermoset resin (mixture D) leads to a higher expansion (140%) and the main peak of viscosity is similar to the peak corresponding to mixture B. The presence of the APP derivative does not seem to influence significantly the rheological behavior of mixture B. The residue obtained is solid and dense. The conditions are favorable to allow slow diffusion of the gaseous degradation products and effective intumescence. This phenomenon, together with a thermally stable char structure (Fig. 7), provides effective protection. To conclude this part, it is noteworthy that swelling starts at about 300 °C, suggesting that the expansion takes place because of the relatively low viscosity of the char combined with the release of volatile degradation products. It is also observed that all the mixtures that contain boric acid exhibit a lower decrease in viscosity. As a consequence, boric acid plays a key role in the intumescent behavior of the coating, acting both as an intumescent agent and char reinforcer. Finally, as the peak of viscosity of the mixture thermoset resin + boric acid + APP derivative is the smallest, the swelling is the highest because the
Fig. 13. Development of the intumescence (α = conversion degree).
986
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
(a)
(b)
D
1e+6
D
120 100
B
80 60
C
40 20
A
0 100
200
300
400
500
Complex viscosity (Pa.s)
relative % of swelling
140
1e+5 1e+4
C
1e+3
A
B 1e+2 1e+1 0
100
Temperature (°C)
200
300
400
500
Temperature (°C)
Fig. 14. (a) Expansion of the formulation. (b) Complex viscosity.
char can accommodate the stresses due to the internal pressure created by the evolving gases. However, an important parameter that also has to be taken into account is the mechanical properties of the char. If the char is destroyed when subjected to external perturbations such as explosion (e.g. Jet fire) or wind turbulence, the coating will not provide effective protection of the steel structure. 3.4. Mechanical resistance of the char In the stabilization phase of the intumescent structure, the change in the viscosity of the charred material under stress may explain the loss of the protective character of the intumescent shield. Indeed, if the shield becomes too hard, the creation and propagation of cracks leading to a rapid degradation of the material occurs. If a char has a good structural, morphological and heat insulative properties but is easily destroyed under a mechanical action, its efficiency is totally lost in the turbulent regime of combustion. Fig. 15 presents the destruction force of the carbonaceous char plotted against the distance between the plates for the following mixtures at 500 °C: the thermoset resin (A), the thermoset resin + APP derivative (B), the thermoset resin + boric
force (g) 1800 1600 1400
D
1200 1000 800 600 400
C
200 0 16
B 14
12
10
8
6
4
2
A 0
gap (mm) Fig. 15. Mechanical resistance at 500 °C of different formulations.
acid (C) and the thermoset resin mixed with both boric acid and APP (D). The curves corresponding to mixtures A and B are not very different: the char is not resistant and a very weak force is sufficient to destroy it. It is only when 1 mm of residue remains that the force increases, and this is due to the fact that the residue has been strongly compressed. The curves corresponding to mixtures C and D give the same results at the beginning: a very weak force is sufficient to destroy the char. But the force begins to increase earlier than in cases A and B; there is a difference of about 2 mm, which means that the char is harder. It appears that the presence of boric acid, and particularly the formation of B2O3, provides better mechanical resistance. 4. Conclusion The aim of this work was to study the efficiency of different epoxy resin based experimental intumescent formulations, designed for the protection of steel in the case of hydrocarbon fires. Tests in big industrial furnaces lead to time/temperature profiles of the different formulations in case of hydrocarbon fires. Ammonium polyphosphate and boric acid provide good performance when they are incorporated separately into the resin, but the intumesced char falls off the plate. Using a Bunsen burner test onto the plate, it is revealed that the intumesced char formed from the epoxy resin containing APP is light and blows away during the experiment. It was also shown that the adhesion of the coating to the steel plate is very weak when phosphorous species are not added into the formulation. The char developed from resin mixed with boric acid slides along and does not remain stuck on the plate. The best result was obtained when APP and boric acid were combined into the resin: the backside of the steel plate reaches 400 °C in 29.5 min regard to 4 min for the epoxy resin alone and the char remains well stuck on the plate. Nevertheless the furnace test does not provide explanations on how the formulation works. Thermogravimetric analyses show that additives interact with the thermoset resin to increase its thermal stability. It is suggested that a reaction occurs between boric acid and ammonium polyphosphate
M. Jimenez et al. / Surface & Coatings Technology 201 (2006) 979–987
derivative during heating, permitting the adhesion between the coating and the steel plate. Novel experiments developed, using a rheometer, show that the degradation of the resin at about 350 °C creates an important decrease of viscosity. Boric acid seems to be the key additive to reduce this fall of viscosity: in fact it turns by dehydration into boron oxide, a hard glass, which traps the gases, and allows a high expansion and a char with a good mechanical resistance. APP derivative is used both as blowing agent and acid source, as it releases ammonia which is trapped into the char structure to yield phosphoric acid. This mixture behaves like an intumescent system, but the char created is too light and exhibit poor mechanical resistance. The best result is obtained when both additives are combined into the resin: there is a small decrease of viscosity at 350 °C, a high expansion and the char has an appropriate mechanical resistance. References [1] [2] [3] [4]
J.A. Seiner, T.A. Ward, Polym. Paint Colour J. 178 (4207) (1988) 75. G.K. Castle, Loss Prev. 8 (1974) 57. S.C. Upadhya, Paintindia 50 (2000) 45. S. Duquesne, S. Magnet, C. Jama, R. Delobel, Surf. Coat. Technol. 180–181 (2004) 302.
987
[5] S. Bourbigot, M. Le Bras, S. Duquesne, M. Rochery, Macromol. Mater. Eng. 289 (2004) 499. [6] R. Delobel, M. Le Bras, N. Ouassou, F. Alistiqsa, J. Fire Sci. 8 (1990) 85. [7] G. Camino, G. Martinasso, L. Costa, Polym. Degrad. Stab. 27 (1990) 285. [8] H.L. Vandersall, J. Fire Flammabl. 2 (1971) 97. [9] A.R. Horrocks, Polym. Degrad. Stab. 54 (2–3) (1996) 143. [10] R. Delobel, M. Le Bras, N. Ouassou, F. Alistiqsa, J. Fire Sci. 8 (2) (1991) 85. [11] C.F. Cullis, M.M. Hirschler, The Combustion of Organic Polymers, Clarendon Press, Oxford University Press, Oxford, 1981. [12] T. Arthur, K. Quill, Proceedings of Flame Retardant'92, Elsevier, London, 1992, p. 223. [13] W. Lyons, The Chemistry and Uses of fire Retardants, Wiley-Interscience, John Wiley and Sons. Inc. New York, 1970. [14] K.K. Shen, E. Olson, Proceedings of the Conference on Recent Advances in Flame Retardancy of Polymeric Materials 16 (2005) 159. [15] N. Grassie, M.I. Guy, N.H. Tennent, Polym. Degrad. Stab. 14 (1986) 125. [16] M.A. Keenan, D.A. Smith, J. Appl. Polym. Sci. 11 (7) (1967) 1009. [17] J.C. Peterson-Jones, J.A. Percy, R.G.F. Giles, A.M. Stephen, J. Appl. Polym. Sci. 17 (1973) 1877. [18] G. Camino, M.P. Luda, in: G. Camino, M. le Bras, S. Bourbigot, R. Delobel (Eds.), Fire Retardancy of Polymers—The Use of Intumescence, Special Publication, vol. 224, Royal Society of Chemistry, Cambridge, 1998, p. 48. [19] M. Bugajny, M. Le Bras, S. Bourbigot, Fire Mater. 23 (1) (1999) 49. [20] P. Anna, G. Marosi, I. Csontos, S. Bourbigot, M. Le Bras, R. Delobel, Polym. Degrad. Stab. 74 (3) (2001) 423.