Thermal decomposition of flexible polyurethane foams in air

Fire Safety Journal 111 (2020) 102925

Contents lists available at ScienceDirect

Fire Safety Journal journal homepage: http://www.elsevier.com/locate/firesaf

Thermal decomposition of flexible polyurethane foams in air D.S.W. Pau a, *, C.M. Fleischmann a, M.A. Delichatsios b a b

University of Canterbury, Private Bag 4800, Christchurch, 8140, Canterbury, New Zealand Northeastern University, 360 Huntington Avenue, Boston, MA, 02115, USA

A R T I C L E I N F O

A B S T R A C T

Keywords: Thermogravimetric analysis Differential scanning calorimetry Oxidative thermal decomposition Polyurethane foam Inflection point methods Kinetic properties Heat of reaction

The oxidative thermal decomposition of a non-fire retardant and a fire retardant polyurethane foam is investi­ gated using thermogravimetric analysis and differential scanning calorimetry over 1–60 � C/min of heating rate, in both air and nitrogen environments. From the thermogravimetry results, the oxidative environment of air results in additional oxidative reactions which are heating rate dependent. These reactions compete with the pyrolysis reactions for the same reactants over similar temperature range. For both foams, increasing heating rate gradually reduces the influence from the oxidation of foam while favouring the pyrolysis of polyol. At low heating rate, in oxidative air environment, the fire retardant foam shows significant decomposition across low temperature range. This environment mimics the incipient phase of a fire, and ignition inhibition is possible when this low temperature decomposition is coupled with gas phase fire retardant mechanisms such as neu­ tralisation of reactive radicals. Fire retardant additives also form char residue which notably lowers the decomposition rate while extending the decomposition temperature range, improving the overall thermal sta­ bility. Lastly, the kinetic properties governing the oxidative thermal decomposition of foams are estimated graphically using Inflection Point Methods, and the corresponding exothermic heat of reaction is determined from the heat flow results.

1. Introduction During smouldering combustion, condensed solid undergoes decomposition which is chemically more complex when compared to flaming combustion. This is caused by the presence of oxygen which results in additional oxidative reactions that compete with the pyrolysis reactions in smouldering combustion. A number of researchers listed chronologically, have presented the science and physical phenomena relating to smouldering combustion in great details. Ohlemiller et al. [1] developed a model for downward smouldering combustion of flexible polyurethane (PU) foams against an upward forced flow of air. The model accounted for the heat transfer mechanisms, mainly conduction and radiation which govern the smoulder propagation rate, and also the oxygen dependent heat generation where an accurate quantification of the equivalence ratio is important. Investigating the effects of opposed forced flow, Torero et al. [2] conducted downward and upward smouldering experiments of PU foam, and highlighted the competition between oxygen supply and heat losses with respect to the oxidiser flow rate which governs smouldering combustion. Walther et al. [3] extended

the research, investigating diffusion driven and opposed forced flow smouldering behaviour of PU foam under normal gravity and micro­ gravity, and also normal and oxygen-rich environments. The research demonstrated that without gravity, the transport of oxidiser to the smoulder reaction zone is limited, and as a result, smouldering becomes unsustainable even under oxygen-rich environment. Using a different sample orientation, Chao et al. [4] investigated the transition of smouldering to flaming combustion of horizontal PU foams under nat­ ural convection. The research noted that while the transition is influ­ enced by a number of factors, it is essentially governed by the exothermic oxidation of residual char and the availability of oxygen. Rein [5] presented a summary of the state-of-the-art science and phe­ nomena associated with smouldering combustion. In flaming combustion, the oxygen is mostly consumed in the gas phase, producing flame which is an oxidative reaction zone [6]. While this assumption is valid for most condensed solid, the porous nature of flexible PU foam allows oxygen to permeate the material by means of convection and diffusion hence the influence of oxidative reactions in flaming combustion could be significant [7]. A number of researchers

* Corresponding author. E-mail addresses: [email protected] (D.S.W. Pau), [email protected] (C.M. Fleischmann), [email protected] (M.A. Delichatsios). https://doi.org/10.1016/j.firesaf.2019.102925 Received 15 July 2019; Received in revised form 4 November 2019; Accepted 22 November 2019 Available online 27 November 2019 0379-7112/© 2019 Elsevier Ltd. All rights reserved.

D.S.W. Pau et al.

Fire Safety Journal 111 (2020) 102925

(a)

oxidative thermal decomposition of PU foam by presenting the experi­ mental study on a non-fire retardant (NFR) and a fire retardant (FR) foam using simultaneous differential scanning calorimetry and ther­ mogravimetric analysis (SDT). The series of experiments included four different heating rates, 1, 5, 20 and 60 � C/min, under inert nitrogen environment and oxidative air environment. The kinetic properties relating to the pyrolysis and oxidative reactions are attained using the Inflection Point Methods [14] from the TGA results comprising sample mass history. These properties are the reaction order (n), pre-exponential factor (A) and activation energy (E). The heat of reac­ tion (Δhr) relating to the oxidative thermal decomposition is established from the heat flow measurements attained from DSC. This paper is an expansion of the research described in an earlier published conference proceeding [15].

(b)

Fig. 1. Simultaneous differential scanning calorimetry and thermogravimetric analysis. (a) SDT 600; (b) Sample and reference cups entering purged furnace.

listed chronologically, have demonstrated that the thermal decomposi­ tion behaviour of flexible PU foam in air is more complex than under an inert environment. Bilbao et al. [8] conducted thermogravimetric analysis (TGA) experiments on flexible PU foam at heating rate of 5 � C/min. The results revealed different decomposition behaviour under nitrogen and air environments. In nitrogen, two distinct pyrolysis re­ actions represented by two separate peaks are noted at 200–300 � C and 325–400 � C, respectively. In air, a single decomposition region repre­ sented by a plateau is noted at 200–325 � C. Investigations at multiple heating rates, 5–20 � C/min for flexible PU foams have also been carried out by Chao et al. [9] and Valencia [10] which continues to show notable differences between decomposition in nitrogen and air envi­ ronments. At these higher heating rates, three peaks are noted for decomposition in air, suggesting the overlap of multiple pyrolysis and oxidative reactions. Comparing TGA results at 5 and 176 � C/min, Kr€ amer et al. [7] confirms that oxidative thermal decomposition in air is heating rate dependent as three peaks at 5 � C/min reduce to two at 176 � C/min. Based on TGA results from Chao et al. [9], Rein et al. [11] derived a 2-step pyrolysis mechanism to describe the thermal decomposition of flexible foam under inert nitrogen environment, and included 3 addi­ tional oxidative reactions for the decomposition under oxidative air environment. The proposed 5-step oxidative decomposition mechanism listed as follows was further validated by Valencia et al. [10,12] and Rogaume et al. [13] via detailed experimental studies involving TGA, differential scanning calorimetry (DSC), tubular furnace and Fourier transform infra-red (FTIR) spectroscopy. This improved the under­ standing on the competition between pyrolysis and oxidative reactions and the characterisation of the gaseous products released during each macro-scale decomposition step. Near heating rate experienced in €mer et al. [7] found that 2-step mechanism flaming combustion, Kra dominates in both nitrogen and air environments. Pyrolysis:

2. Simultaneous differential scanning calorimetry and thermogravimetric analysis Fig. 1(a) shows the instrument used, SDT 600. It is a product of TA Instruments which simultaneously acquires both DSC and TGA mea­ surements from a single experiment. The experiments were conducted in dynamic mode where the foam sample was subjected to a constant heating rate within a purged furnace from room temperature (~20 � C) up to a maximum of 600 � C for inert nitrogen environment and 900 � C for oxidative air environment. The sample mass ranged between 3 and 4 mg, comprising flexible foam fragments within a half-filled 90 μL alumina cup as seen in Fig. 1(b). The foam fragments were prepared by shredding a larger foam piece using laboratory tweezers. The sample was tested in open configuration without the presence of a lid to allow the release of volatiles during thermal decomposition. The relatively small sample size ensures negligible thermal gradient spatially through the specimen during the transient heating process thus satisfying the assumption of a lumped capacitance system. SDT 600 has four calibrations which were performed in sequence following a change in heating rate or purge gas. Listed in order, the calibrations were specific to mass, temperature, heat flow and cell constant. Mass calibration determined the weight correction factors specific to the heating rate and temperature range applied. This cali­ bration was performed using manufacturer issued standard weight pieces. Temperature calibration established accurate temperature mea­ surements by relating the measured temperature with the actual known temperature of a physical transition [16], such as the melting of a high purity metal. A single point calibration using 3–10 mg of zinc with a melting temperature of 419 � C was conducted. Zinc sample was placed inside the sample cup, and at the selected heating rate, the calibration was performed from 200 � C below the melting point to 200 � C above. Analysing over the endothermic melting region, the difference between measured and actual temperature were compensated. Heat flow cali­ bration developed the proportionality factor for converting measured voltage signal into heat flow [16]. Specific to the heating rate and temperature range applied, this calibration used the manufacturer is­ sued sapphire disk to develop a heat flow calibration curve according to the sapphire heat capacity. Lastly, the cell constant calibration refined the heat flow calibration curve by applying a new cell constant. The calibration involved setting the existing cell constant to 1, and using the same zinc sample from the previous temperature calibration. The sample was first equilibrated at 550 � C, and then at 250 � C before being heated to 500 � C at 20 � C/min. Analysing the heat flow of the endothermic melting region, the measured heat of fusion was determined. The new cell constant is the ratio of known value to measured value of the heat of fusion. The TGA measurements are the changes in sample mass during decomposition, registered based on the current signal required to correct a taut-band meter movement [17]. The DSC measurements are the changes in enthalpy, registered based on the heat flux concept where the heat flow, dq/dt is determined from the changes in sample temperature

(1) PU Foam → Regenerated Polyol þ Gaseous Isocyanates (2) Regenerated Polyol → Char þ OH Species, H2CO, CH4 and H2O Oxidation: (1) PU Foam þ O2 → Regenerated Polyol þ OH Species, CO2 and H2O (2) Regenerated Polyol þ O2 → Char þ OH Species, H2CO, CH4, CO, CO2 and H2O (3) Char þ O2 → OH Species, H2CO, CH4, CO, CO2 and H2O In the past, the kinetic properties governing the decomposition are determined using graphical methods [8,9], and more recently, numeri­ cal tools such as genetic algorithm are also applied [10,11]. Neverthe­ less, a complete depiction of PU foam thermal decomposition remains scarce as the studies mainly involved flexible foam without fire retar­ dant additives, and at relatively low heating rates, below 20 � C/min in comparison with heating rate achieved in flaming combustion, 200–400 � C/min. This paper aims to supplement the current knowledge on 2

D.S.W. Pau et al.

1.8E-02 1.6E-02

2.0E-02

Prominent influence from oxidation of foam at 1 °C/min.

1.8E-02

1.4E-02

d /dT (1/°C)

1.6E-02

Oxidation of polyol.

1.4E-02

d /dT (1/°C)

2.0E-02

Fire Safety Journal 111 (2020) 102925

1.2E-02 Minimal influence from pyrolysis of polyol at 1 °C/min.

1.0E-02 8.0E-03 6.0E-03

1.0E-02 8.0E-03 6.0E-03

4.0E-03

4.0E-03 Pyrolysis of foam.

2.0E-03 0.0E+00

1.2E-02

100

2.0E-03

Oxidation of char.

200

300 N2

400 500 Temperature (°C)

Air

600

0.0E+00

700

100

200

Oxidation only

300 N2

Air

( a) 2.0E-02

1.8E-02

1.8E-02

1.6E-02

1.6E-02

Reducing influence from oxidation of foam at 20 °C/min.

1.0E-02 8.0E-03

1.0E-02 8.0E-03 6.0E-03

4.0E-03

4.0E-03

2.0E-03

2.0E-03

0.0E+00

0.0E+00

200

300 N2

400 500 Temperature (°C)

Air

Oxidation only

600

700

Pyrolysis of foam.

1.2E-02

6.0E-03

100

700

Prominent influence from pyrolysis of polyol at 60 °C/min.

1.4E-02

Increasing influence from pyrolysis of polyol at 20 °C/min.

d /dT (1/°C)

d /dT (1/°C)

1.2E-02

600

(b)

2.0E-02

1.4E-02

400 500 Temperature (°C)

Minimal influence from oxidation of foam at 60 °C/min. Oxidation of polyol. Oxidation of char. 100

200

300 N2

Oxidation only

( c)

400 500 Temperature (°C)

Air

600

700

Oxidation only

(d)

Fig. 2. Decomposition rate versus temperature of NFR-SB-31 in air and nitrogen, and the derived oxidative reactions. (a) 1 � C/min; (b) 5 � C/min; (c) 20 � C/min; (d) 60 � C/min.

according to the thermal equivalence of Ohm’s Law, as seen in Equation (1) [18]. dq ΔTsam ¼ dt Rt

ref

⋅kf ⋅ku

acrylonitrile polymer. In smaller quantity, other ingredients include inorganic fillers, plasticisers, extenders, antimicrobial agents and pig­ ments. The fire retardant additives within FR-Y-36 comprise melamine and halophosphate, indicated by the higher nitrogen content and the presence of chlorine and phosphorus. These fire retardant additives have solid and gas phase fire retardant capability. In the gas phase, chlorine from the breakdown of hal­ ophosphate neutralises the reactive radicals produced by foam decom­ position [19] while the breakdown of melamine also releases ammonia and nitrogen gas which dilute the concentration of combustible gas to inhibit combustion [20]. Melamine and phosphate structure are also reactive towards the gaseous isocyanates released, forming residue hence reducing the quantity of isocyanates available for gas phase combustion [21,22]. In the solid phase, melamine reacts with phos­ phoric acid from halophosphate breakdown to form thermally stable cross-linked residue or char at varying temperatures, melam at 350 � C, melem at 450 � C and melon at 600 � C [23]. Char on the fuel surface has shielding capability, limiting the heat transfer through solid, the release of combustible volatile and the access of oxygen to mitigate combustion. NFR-SB-31 is not tested to any fire test standards while FR-Y-36 is tested to BS 5852:2006 [24] with wood crib ignition and Technical Bulletin 117 [25]. These are tests involving mock-up chair setup and this

(1)

ΔTsam-ref is the difference between the sample and reference tem­ peratures measured by the thermocouple underneath the platforms. Rt, kf and ku are factory values and calibration constants. The heating rates investigated were 1, 5, 20 and 60 � C/min, and for each heating rate, three replicates were tested. The selected heating rates were limited by the allowable maximum of the instrument at 100 � C/min. 3. Polyurethane foams selection The flexible PU foams investigated are referred to as NFR-SB-31 for the non-fire retardant foam with a density of 31 kg/m3 and FR-Y-36 for the fire retardant foam with a density of 36 kg/m3. Through elemental analysis and based on the chemical elements detected, the chemical composition of NFR-SB-31 is C1H1.77O0.31N0.06 and for FR-Y-36, it is C1H1.69O0.28N0.17Cl0.003P0.001. The polyurethane foams are manufac­ tured from the polycondensation of toluene diisocyanate, water and polyalkoxy polyether polyol. The polyol also contains styrene and 3

D.S.W. Pau et al.

Fire Safety Journal 111 (2020) 102925

4.0E-02

4.0E-02

Single peak and prominent influence from oxidation of foam and polyol at 1 °C/min.

3.5E-02

3.5E-02

1.5E-02

Pyrolysis of foam.

d /dT (1/°C)

d /dT (1/°C)

Negligible influence from pyrolysis of polyol at 1 °C/min.

2.5E-02 2.0E-02

2.5E-02 2.0E-02 1.5E-02 1.0E-02

1.0E-02 5.0E-03 0.0E+00

Single peak and reducing influence from oxidation of foam and polyol at 5 °C/min.

3.0E-02

3.0E-02

5.0E-03

Oxidation of char. 100

200

300 N2

400 500 Temperature (°C) Air

600

700

0.0E+00

800

100

200

300 N2

Oxidation only

(a) 2.0E-02 1.6E-02 1.2E-02 1.0E-02

Reducing influence from oxidation of foam at 20 °C/min.

4.0E-03

100

200

800

Oxidation only

1.2E-02

Minimal influence from oxidation of foam at 60 °C/min.

1.0E-02 8.0E-03 6.0E-03

Oxidation of polyol.

Pyrolysis of foam.

4.0E-03

Pyrolysis of foam.

2.0E-03

1.6E-02

Increasing influence from pyrolysis of polyol at 20 °C/min.

6.0E-03

700

Prominent influence from pyrolysis of polyol at 60 °C/min.

1.8E-02 1.4E-02

8.0E-03

0.0E+00

2.0E-02

d /dT (1/°C)

d /dT (1/°C)

1.4E-02

Air

600

(b)

Separated peak, oxidation of polyol.

1.8E-02

400 500 Temperature (°C)

2.0E-03

Oxidation of char. 300 N2

400 500 Temperature (°C) Air

600

700

0.0E+00

800

Oxidation of char. 100

200

300 N2

Oxidation only

(c)

400 500 Temperature (°C) Air

600

700

800

Oxidation only

(d)

Fig. 3. Decomposition rate versus temperature of FR-Y-36 in air and nitrogen, and the derived oxidative reactions. (a) 1 � C/min; (b) 5 � C/min; (c) 20 � C/min; (d) 60 � C/min.

FR foam has passed the tests, demonstrating no flaming or progressive smouldering at this scale. On reduced scale, FR-Y-36 passes the vertical test of Appendix F, Part I of x25.853 of Federal Aviation Regulations [26] where a sample is exposed to burner’s flame. In AS/NZS 1530.3:1999 [27] where a vertical sample is subjected to gas-fired ceramic panel, FR-Y-36 exhibits ignition (index 18 of 20) and smoke generation (index 5 of 10) but demonstrates negligible flame spread and heat evolution (index 0 of 10). Further details relating to the application of flexible PU foams used in this research have been documented in a dissertation [28].

represents nitrogen and dotted red line represents the derived oxidative reactions. Decomposition under nitrogen environment shows a consis­ tent 2-step pyrolysis mechanism at all heating rates, and as described earlier, this involves the pyrolysis of PU foam followed by the pyrolysis of the secondary product formed, polyol. The decomposition under air environment shows the outcome of a series of reactions overlapping closely across similar temperature range. The decomposition behaviour is also heating rate dependent where the peak of the curve shifts towards higher temperature as the heating rate increases. Assuming that the 2 pyrolysis reactions proceed and remain unaffected by the additional oxidative reactions under air, subtracting the decomposition under ni­ trogen (dashed line) from that of air (solid line) and setting all negative values to zero, yields the derived oxidative reactions (dotted red line). These derived results illustrate the presence of 3 additional oxidative reactions, and as described earlier, these are the oxidation of PU foam, polyol and char. Oxidative reactions generate secondary products similar to the pyrolysis reactions, and also common combustion prod­ ucts such as carbon monoxide and carbon dioxide. From Fig. 2, the pyrolysis and oxidation of foam overlap over a similar temperature range, competing for the same reactant. The oxidative thermal decomposition initiates with the pyrolysis of foam but the competing oxidative reaction is able to accelerate and reach its peak

4. Pyrolysis and oxidative reactions of polyurethane foam thermal decomposition Thermal decomposition of PU foam is described by the plot of decomposition rate, dα/dT versus reaction temperature, T where a peak over a defined temperature range signifies the occurrence of a solid phase reaction. α is defined as the fraction decomposed and the decomposition rate is normalised by dividing dα/dt by the heating rate β. The decomposition behaviour of NFR-SB-31 is shown in Fig. 2(a) – (d) for the different heating rates while the different line types on each plot denote the different environments, solid line represents air, dashed line 4

D.S.W. Pau et al.

-2.50E-03

-2.30E-03

3rd Reaction y = 13288x + 24.977 R² = 0.7293

-2.10E-03

12.0

-2.0 -1.90E-03 -1.70E-03 -1.50E-03 2nd Reaction -4.0 y = 43163x + 67.582 R² = 0.8444 -6.0

1st Reaction y = 38288x + 65.55 R² = 0.954

8.0 4.0

-8.0

ln(k)

ln(d /dT)

-2.70E-03

Fire Safety Journal 111 (2020) 102925

-10.0

(ln(1- )/ )-1/T

3rd Reaction y = -13160x + 24.814 R² = 0.9017

0.0 1.00E-03

1.20E-03

-12.0

-8.0

-14.0

-12.0

-16.0

-16.0

1.40E-03

1.60E-03

1.80E-03

2.00E-03

2.20E-03

2nd Reaction y = -42388x + 66.302 R² = 0.9532

-4.0

1st Reaction y = -37882x + 64.78 R² = 0.9736 1/T (1/K)

(a)

( b)

Fig. 4. Application of Inflection Point Methods on derived oxidative reactions of NFR-SB-31 at 60 � C/min (a) ln(dα/dT) versus [ln(1 – α)/Φ] – 1/T; (b) ln(k) versus 1/T. 1.0E-02

1.0E-02

d /dT (1/°C)

6.0E-03 4.0E-03

c4 = 0.175

c2 = 0.279

c3 = 0.348

8.0E-03

c3 = 0.206

d /dT (1/°C)

c1 = 0.206

8.0E-03

2.0E-03

6.0E-03

c4 = 0.214

c2 = 0.281

4.0E-03 c1 = 0.101

2.0E-03 c5 = 0.134

0.0E+00

100

200

Ox Foam (1)

Py Foam (2)

300

400 500 Temperature (°C)

Ox Polyol (3)

Py Polyol (4)

600

0.0E+00

700

Ox Char (5)

c5 = 0.056 100

Ox Foam (1)

(a)

200 Py Foam (2)

300

400 500 Temperature (°C)

Ox Polyol (3)

Py Polyol (4)

600

700

Ox Char (5)

(b)

Fig. 5. Decomposition rate versus temperature of NFR-SB-31 for estimation of mass fraction ci. (a) 1 � C/min; (b) 60 � C/min.

first. With increasing heating rate, there is a reduction to the influence from foam oxidation (first peak of dotted red line) where the peak value reduces from ~9.1 � 10 3 to 3.1 � 10 3 1/� C. The influence from foam pyrolysis (first peak of dashed line) remains consistent, between ~4.9 � 10 3 and 6.4 � 10 3 1/� C. The competing reactions for foam are fol­ lowed by the oxidation of foam decomposition product, polyol (second peak of dotted red line) which shows a fairly consistent influence at all heating rates, ~7.7 � 10 3 1/� C. Similar to foam, there is also a competing pyrolysis reaction for polyol but this is over a higher tem­ perature range, sufficiently separated from polyol oxidation. As heating rate increases, the influence from polyol pyrolysis (refer to solid line around 300–450 � C region) becomes more prominent where the peak value increases from ~2.3 � 10 3 to 8.0 � 10 3 1/� C. Lastly, the char from pyrolysis and oxidation of polyol undergoes oxidation, and its magnitude is relatively small compared to the other main reactions. Nonetheless, a reduction in the influence of char oxidation is noted as the heating rate increases. The sequence of pyrolysis and oxidative re­ actions for the oxidative thermal decomposition of NFR-SB-31 is found to be consistent with the literature [12,13]. The decomposition behaviour of FR-Y-36 for the different heating

rates is shown in Fig. 3(a) – (d), using the same legends as NFR-SB-31 for the different environments. Note that the vertical axis of (a) and (b) have greater magnitude than (c) and (d). Similar to NFR-SB-31, the decom­ position under nitrogen of FR-Y-36 shows the consistent 2-step pyrolysis mechanism at all heating rates. The decomposition under air also shows overlapping and heating rate dependent reactions but different from NFR-SB-31, the 3 oxidative reactions are not always apparent for FR-Y36. At 1 and 5 � C/min, following the initial pyrolysis reaction of foam, a single peak is noted from the derived oxidative reactions (dotted red line) suggesting that the oxidation of foam and polyol closely overlap each other. This is possibly due to the fire retardant additives within the foam, where the acidic condition generated by the breakdown of hal­ ophosphate and the reactivity of melamine towards the products of foam decomposition might have catalysed the oxidation of foam and polyol, allowing these to proceed at lower temperature under such heating rates [21,22]. The release of gaseous products over low temperature range diminishes the ignitability, and when coupled with gas phase fire retardant mechanisms such as neutralisation and dilution, combustion inhibition becomes more effective. At 20 and 60 � C/min, separated peaks are noted for the oxidation of foam and polyol, and the influence 5

D.S.W. Pau et al.

Fire Safety Journal 111 (2020) 102925

magnitude due to char formation by melamine. According to Chao et al. [4], oxidative reaction is believed to be favored by slow heating and the results in Figs. 2 and 3 support this, showing the oxidative reactions having greater influence than the pyrolysis reactions at low heating rate. The temperature range where the main reactions occurs is ~200 � C for NFR-SB-31, consistently between 200 and 400 � C. However, for FR-Y-36, this range can vary from ~100 � C at low heating rate such as 1 � C/min to ~250 � C at high heating rate such as 60 � C/min. The main reactions occur between 200 and 300 � C for heating rate below 5 � C/min, between 200 and 400 � C at 20 � C/min, and increasing to be­ tween 200 and 450 � C at 60 � C/min. The extended temperature range of FR-Y-36 towards high heating rate is due to the char formation from the breakdown of melamine which improves the thermal stability of the decomposing sample.

Table 1 Refined kinetic properties for NFR-SB-31 at all heating rates. Reaction

Parameter

1 � C/ min

5 � C/min

20 � C/min

60 � C/min

Foam oxidation

n1 A1 (1/s)

18.50 3.96 � 1062 639

20.80 3.04 � 1056 597

13.00 1.45 � 1031 350

22.00 1.36 � 1028 315

0.206

0.186

0.143

0.107

3.50 8.27 � 109 134

4.50 5.68 � 108

6.58 7.73 � 108

9.50 5.41 � 109

121

121

127

0.279

0.314

0.298

0.298

9.50 1.08 � 1051 539

10.30 3.78 � 1052 579

13.30 4.59 � 1061 700

6.70 6.23 � 1028 352

0.206

0.267

0.297

0.298

1.04 1.07 � 1013 189

0.96 2.22 � 1012 181

1.20 1.95 � 1018 251

1.36 2.47 � 1019 265

0.175

0.233

0.262

0.297

4.70 2.34 � 1044 511

excludeda excludeda

excludeda excludeda

excludeda excludeda

excludeda

excludeda

excludeda

0.134

excludeda

excludeda

excludeda

E1 (kJ/ mol) c1 Foam pyrolysis

n2 A2 (1/s) E2 (kJ/ mol) c2

Polyol oxidation

n3 A3 (1/s) E3 (kJ/ mol) c3

Polyol pyrolysis

n4 A4 (1/s) E4 (kJ/ mol) c4

Char oxidation

n5 A5 (1/s) E5 (kJ/ mol) c5

a

5. Application of Inflection Point Methods Based on the decomposition behaviour noted in Figs. 2 and 3, the kinetic properties, n, A and E for the 2 pyrolysis and 3 oxidative re­ actions of the oxidative thermal decomposition of NFR-SB-31 and FR-Y36 are determined using the Inflection Point Methods [14]. This graphical approach has previously been applied to determine the kinetic properties for the thermal decomposition of the same foams in nitrogen environment [29]. Equation (2) shows the temperature dependent rate constant, k is represented as the decomposition rate, dα/dT divided by the nth order kinetic model, (1 – α)n where α is the fraction of sample decomposed. The remainder of the equation on the right hand side is the temperature based Arrhenius expression whereby A is divided by the heating rate, β and R is the universal gas constant, 8.314 J/mol∙K. � � � � dα=dT E A lnðkÞ ¼ ln þ ln (2) ¼ ð1 αÞn RT β

Reaction has been excluded from the model.

At the peak of a reaction or the maximum inflection point, dα/dT is a maximum while d2α/dT2 equals 0. Solving these conditions yield n ¼ (E/ RT2)(1 – α)/(dα/dT) and rearranging Equation (2) yields Equation (3) with Φ ¼ T2(dα/dT)/(1 – α). T, α and dα/dT as input for the determi­ nation of n and Φ are parameters at inflection point. The analysis region for the Inflection Point Methods initiates from the start of a reaction to the peak dα/dT. In this research, the starting point is nominally defined as the temperature where dα/dT reaches 10% of the difference between the starting minima and the peak dα/dT of the reaction.

of these oxidative reactions has diminished with increasing heating rate, from ~2.7 � 10 2 at 1 � C/min to 4.6 � 10 3 1/� C at 60 � C/min. While the influence from pyrolysis of polyol is negligible at 1 and 5 � C/min (refer to solid line around 300–400 � C region), this influence has increased at higher heating rates of 20 and 60 � C/min (refer to solid line around 350–450 � C region) but compared to NFR-SB-31, the peak of polyol pyrolysis is not as apparent. The char oxidation shows a similar trend to NFR-SB-31 in relation to heating rate but with greater Table 2 Refined kinetic properties for FR-Y-36 at all heating rates. Reaction

Parameter

1 � C/min

5 � C/min

20 � C/min

60 � C/min

Foam oxidation

n1 A1 (1/s) E1 (kJ/mol) c1

6.10 1.02 � 1073 753 0.582

22.00 2.57 � 10121 1266 0.454

51.00 7.03 � 10136 1446 0.053

excludeda excludeda excludeda excludeda

Foam pyrolysis

n2 A2 (1/s) E2 (kJ/mol) c2

2.00 5.46 � 109 130 0.418

3.10 4.27 � 1010 137 0.333

4.80 1.20 � 1011 141 0.340

6.70 1.12 � 109 118 0.385

Polyol oxidation

n3 A3 (1/s) E3 (kJ/mol) c3

excludeda excludeda excludeda excludeda

excludeda excludeda excludeda excludeda

3.10 3.80 � 1014 185 0.394

2.90 7.37 � 1010 149 0.372

Polyol pyrolysis

n4 A4 (1/s) E4 (kJ/mol) c4

excludeda excludeda excludeda excludeda

excludeda excludeda excludeda excludeda

excludeda excludeda excludeda excludeda

1.91 3.07 � 1024 331 0.243

Char oxidation

n5 A5 (1/s) E5 (kJ/mol) c5

excludeda excludeda excludeda excludeda

1.10 2.46 � 109 156 0.213

1.70 2.56 � 1010 155 0.213

excludeda excludeda excludeda excludeda

a

Reaction has been excluded from the model. 6

D.S.W. Pau et al.

Fire Safety Journal 111 (2020) 102925

1.0E-02

3.5E-02

9.0E-03

3.0E-02

8.0E-03 7.0E-03

d /dT (1/°C)

d /dT (1/°C)

2.5E-02 2.0E-02 1.5E-02

5.0E-03 4.0E-03 3.0E-03

1.0E-02

2.0E-03

5.0E-03 0.0E+00

6.0E-03

1.0E-03 100

200

300

400 500 Temperature (°C)

Model

Ox Foam

600

700

0.0E+00

800

Py Foam

100 Model

200

300

Ox Foam

1.0E-02

1.0E-02

9.0E-03

9.0E-03

8.0E-03

8.0E-03

7.0E-03

7.0E-03

6.0E-03 5.0E-03 4.0E-03

2.0E-03

1.0E-03

1.0E-03

0.0E+00

0.0E+00

Model

400 500 Temperature (°C)

Py Foam

Ox Polyol

Ox Char

4.0E-03 3.0E-03

300

800

5.0E-03

2.0E-03

200

Ox Polyol

700

6.0E-03

3.0E-03

100

Py Foam

600

(b)

d /dT (1/°C)

d /dT (1/°C)

(a)

400 500 Temperature (°C)

600

700

800

Py Polyol

100 Model

(c)

200

300

Ox Foam

400 500 Temperature (°C) Py Foam

600

Ox Polyol

700

800

Py Polyol

(d)

Fig. 6. Modelled decomposition rate versus temperature for oxidative thermal decomposition of polyurethane foams. (a) FR-Y-36 at 1 � C/min; (b) FR-Y-36 at 20 � C/ min; (c) FR-Y-36 at 60 � C/min; (d) NFR-SB-31 at 60 � C/min.

� � � ��� dα E lnð1 αÞ ¼ R Φ dT

� ln

1 T

�

� � A þ ln β

scheme is adopted as a simplified approach to model the competing pyrolysis and oxidative reactions of the heating rate dependent oxida­ tive thermal decomposition of PU foam. Treating each reaction as a separate component, the model also contains a number of simplified assumptions, including (1) the 2 pyrolysis reactions can proceed, unaf­ fected by the additional oxidative reactions, (2) the ci for each compo­ nent can be reasonably estimated from the area under the curve of dα/dT for each reaction, and (3) the residue generated from each reaction is considered negligible. The last assumption is the least pertinent as py­ rolysis and oxidation of foam is known to generate polyol; however, given there are negligible amount of residue formed following the final oxidation of char, the overall impact of this assumption is perhaps lessened at this scale. � � � � X dα X dα Ei ð1 αÞni ¼ ci Ai exp (4) ¼ dt dt i RT i i

(3)

The determination of the kinetic properties for the derived oxidative reactions of NFR-SB-31 at 60 � C/min is presented in Fig. 4(a) and (b) as example. From the plot of ln(dα/dT) versus [ln(1 – α)/Φ] – 1/T in Fig. 4 (a), as per Equation (3), the value of E/R is determined from the slope of the 1st, 2nd and 3rd reactions, corresponding to the oxidation of foam, polyol and char, respectively. Applying E/R into n yields 37.71, 10.38 and 31.27, respectively for these reactions. Substituting n into Equation (2), from the plot of ln(k) versus 1/T in Fig. 4(b), E is determined from the slope and A is determined from intercept. The pair of values are 315 kJ/mol and 1.36 � 1028 1/s for the 1st reaction, 352 kJ/mol and 6.23 � 1028 1/s for the 2nd reaction, and 109 kJ/mol and 5.98 � 1010 1/s for the 3rd reaction. 6. Determination of kinetic properties

In relation to the second assumption, Fig. 5 illustrates the estimation of ci for the separate pyrolysis and oxidative reactions from the nitrogen, air and derived oxidative dα/dT curves for NFR-SB-31 at 1 and 60 � C/ min. The legend convention uses different line types to represent oxidative (solid) and pyrolysis (dashed) reactions, and different colours to denote the different reactants involved which are foam (black), polyol

Wang et al. [30] described a multiple-component decomposition scheme where the overall decomposition rate, dα/dt is the summation of the decomposition rate of each component, (dα/dt)i based on the respective mass fraction, ci as per Equation (4). This decomposition 7

Fire Safety Journal 111 (2020) 102925

1.6E-02

1.6E-02

1.4E-02

1.4E-02

1.2E-02

1.2E-02

1.0E-02

1.0E-02

d /dT (1/°C)

d /dT (1/°C)

D.S.W. Pau et al.

8.0E-03 6.0E-03

8.0E-03 6.0E-03

4.0E-03

4.0E-03

2.0E-03

2.0E-03

0.0E+00

100

200

300

400 500 Temperature (°C)

Model

Experiment

600

0.0E+00

700

100

200

300

400 500 Temperature (°C)

Model

Experiment

1.6E-02

1.6E-02

1.4E-02

1.4E-02

1.2E-02

1.2E-02

1.0E-02

1.0E-02

8.0E-03 6.0E-03

2.0E-03

2.0E-03 200

300

400 500 Temperature (°C)

Model

Experiment

700

6.0E-03 4.0E-03

100

600

8.0E-03

4.0E-03

0.0E+00

700

(b)

d /dT (1/°C)

d /dT (1/°C)

(a)

600

600

0.0E+00

700

(c)

100

200

300

400 500 Temperature (°C)

Model

Experiment

(d)

Fig. 7. Decomposition rate versus temperature between model and experiment for NFR-SB-31 in air. (a) 1 � C/min; (b) 5 � C/min; (c) 20 � C/min; (d) 60 � C/min.

(green) and char (blue). As an estimate, ci associated with the oxidation of foam, polyol and char is determined from the derived oxidative dα/dT curves while ci for pyrolysis of foam and polyol is attained from the nitrogen and air dα/dT curves, respectively. The trend in Fig. 5 reflects the changes to the influence of the separate reactions with respect to increasing heating rate as explained in Section 4. This is mainly the reduction to the influence of foam oxidation compensated by the in­ crease of the influence from polyol pyrolysis. For modelling, the esti­ mated ci are subjected to further adjustment based on additional qualitative judgements. For instance, the air dα/dT curve (solid line) in Fig. 2(d) suggests the oxidation and pyrolysis of polyol at 60 � C/min are overlapping closely with more balanced distribution in terms of reactant consumed thus in modelling, c3 ¼ 0.348 and c4 ¼ 0.214 seen in Fig. 5(b) are readjusted to be of equal value. The refinement of kinetic properties introduced by Wang et al. [30] is applied to enhance n and ci in order to improve the comparison be­ tween model and experimental dα/dT. The refinement is qualitatively based on matching the peak values of each reaction, between the model and the experiment. Table 1 and Table 2 present the refined kinetic properties for each reaction implemented in the model where the subscript represents the different reactions, (1) oxidation of foam, (2) pyrolysis of foam, (3) oxidation of polyol, (4) pyrolysis of polyol and (5) oxidation of char. A few reactions with low magnitude over an extended

temperature range have been excluded due to the difficulty in calcu­ lating the kinetic properties and modelling the reaction with reasonable accuracy. These includes the final char oxidation, the negligible polyol pyrolysis of FR-Y-36 from 1 to 20 � C/min, and the negligible foam oxidation of FR-Y-36 at 60 � C/min. Also, at 1 and 5 � C/min, the oxidation of foam and polyol for FR-Y-36 are not differentiable as depicted by the significant single first peak of the derived oxidative dα/dT curve (dotted red line) in Fig. 3(a) and (b). Therefore, these merged oxidative reactions are simply denoted by subscript (1). From Tables 1 and 2, the heating rate dependency of the oxidative reactions have resulted in greater variation of the kinetic properties across the heating rates investigated when compared with the pyrolysis reactions. From the trend of ci, there is a notable reduction to the in­ fluence of foam oxidation as the heating rate increases while the influ­ ence of foam pyrolysis remains relatively constant. The reduction is compensated by the increase to the influence from oxidation and py­ rolysis of polyol, particularly the latter showing notable increase at 60 � C/min. Due to the compensating effect amongst the kinetic properties, the trend of n, A and E is less noticeable. According to Wang [31], A and E shift the reaction along the temperature axis whereby lower E or higher A enables the reaction to initiate at lower temperature. n affects the peak and spread of the reaction whereby higher n reduces the re­ action rate, spreading the reaction across wider temperature range. 8

Fire Safety Journal 111 (2020) 102925

4.0E-02

4.0E-02

3.5E-02

3.5E-02

3.0E-02

3.0E-02

2.5E-02

2.5E-02

d /dT (1/°C)

d /dT (1/°C)

D.S.W. Pau et al.

2.0E-02 1.5E-02

2.0E-02 1.5E-02

1.0E-02

1.0E-02

5.0E-03

5.0E-03

0.0E+00

100

200

300 Model

400 500 Temperature (°C)

600

700

0.0E+00

800

100

200

Experiment

300 Model

1.6E-02

1.6E-02

1.4E-02

1.4E-02

1.2E-02

1.2E-02

1.0E-02

1.0E-02

8.0E-03 6.0E-03

2.0E-03

2.0E-03 300 Model

400 500 Temperature (°C)

800

600

700

800

Experiment

6.0E-03 4.0E-03

200

700

8.0E-03

4.0E-03

100

600

(b)

d /dT (1/°C)

d /dT (1/°C)

(a)

0.0E+00

400 500 Temperature (°C)

600

700

0.0E+00

800

Experiment

100

200

300

400 500 Temperature (°C)

Model

(c)

Experiment

(d)

Fig. 8. Decomposition rate versus temperature between model and experiment for FR-Y-36 in air. (a) 1 � C/min; (b) 5 � C/min; (c) 20 � C/min; (d) 60 � C/min.

Tables 1 and 2 consistently shows the compensating effect where high A results in high E, and vice versa. The impact of n is illustrated by FR-Y-36 whereby the significant reduction in the influence of foam oxidation is reflected in the significant increase of n1, from ~6 to 50. Fig. 6 shows examples of the modelled decomposition rate for NFR-SB-31 and FR-Y-36 at a few selected heating rates, including the individual re­ actions that constitute the overall oxidative thermal decomposition. The sequence of the individual reactions shows the modelled oxidative thermal decomposition initiates with the pyrolysis of foam. This is followed closely by the oxidation of foam which forms the first peak of the decomposition illustrated in Fig. 6 with exception to Fig. 6(c) for FR-Y-36 at 60 � C/min where the oxidation of foam is excluded from the model. This is due to the reaction’s minimal influence on decom­ position as noted in Fig. 3(d). The modelled decomposition continues with the oxidation of polyol which forms the second peak of the decomposition for NFR-SB-31 in Fig. 6(d) and the plateau for FR-Y-36 in Fig. 6(b) and (c). FR-Y-36 at 1 and 5 � C/min are exception where the oxidation of polyol is not modelled because this reaction overlaps with the oxidation of foam to form a single peak as seen in Fig. 6(a). Unlike the pyrolysis and oxidative reactions of foam which overlap closely, the pyrolysis of polyol occurs over higher temperature and is relatively distanced from the oxidation of polyol as seen in Fig. 6(d) for NFR-SB31. FR-Y-36 from 1 to 20 � C/min are exception where pyrolysis of

polyol is not modelled as per Fig. 6(a) and (b) due to the reaction’s negligible influence on decomposition as seen in Fig. 3(a)–(c). The py­ rolysis of polyol captures the third and final peak of the main reactions. Lastly, the oxidation of char captures the extended decomposition at high temperature but this reaction is mostly excluded due to its minor contribution to the overall decomposition and the difficulty in accu­ rately estimating its kinetic properties. The comparison between model and experimental decomposition rate at all the heating rates are illus­ trated in Fig. 7 and Fig. 8 for NFR-SB-31 and FR-Y-36, respectively. The solid line represents the experiment while the dotted red line represents the model. The simplified model is able to capture the general trend observed experimentally for oxidative thermal decomposition of foams where the maximum dα/dT shifts from an initial lower temperature region towards a final higher temperature region with increasing heating rate. This is due to the reduction in the influence of foam oxidation and the increase in the influence of polyol pyrolysis. Three peaks for the oxidative ther­ mal decomposition of NFR-SB-31 are evidenced from 5 to 60 � C/min, and these are captured by the overlapping individual reactions modelled. FR-Y-36 exhibits more variation in decomposition behaviour with a single peak at 1–5 � C/min which transitions into a plateau at 20 � C/min, and these features are captured by the model with reasonable accuracy. Note that the vertical scale utilised in Fig. 8(a) and (b) is 9

0.12

15.0

0.12

4.0

0.10

12.0

0.10

3.0

0.08

9.0

0.08

2.0

0.06

6.0

0.06

1.0

0.04

3.0

0.04

100

200

300

-1.0

400

500

600

700

800

Temperature (°C)

0.02

0.0

0.00

-3.0

100

200

300

500

600

700

800

0.02 0.00

Temperature (°C)

(a)

(b) 0.12

150.0

0.12

40.0

0.10

120.0

0.10

30.0

0.08

90.0

0.08

20.0

0.06

60.0

0.06

10.0

0.04

30.0

0.04

0.0 -10.0

100

200

300

400

500

Temperature (°C)

600

700

800

Heat Flow (mW)

50.0

d /dT (1/°C)

Heat Flow (mW)

400

0.02

0.0

0.00

-30.0

100

200

300

400

500

Temperature (°C)

(c)

600

700

800

d /dT (1/°C)

0.0

Heat Flow (mW)

5.0

d /dT (1/°C)

Fire Safety Journal 111 (2020) 102925

d /dT (1/°C)

Heat Flow (mW)

D.S.W. Pau et al.

0.02 0.00

(d)

Fig. 9. Reaction heat flow and decomposition rate versus temperature of NFR-SB-31 in air. (a) 1 C/min; (b) 5 C/min; (c) 20 � C/min; (d) 60 � C/min. �

different to the others. Due to the simplified assumptions and the diffi­ culty in accurately determining the kinetic properties of extended, low magnitude reaction, the model has not been able to capture the final stage of decomposition involving char oxidation. The model also consistently shows the oxidative thermal decomposition initiating and occurring over slightly higher temperature range compared to the experiment, and such inherent mismatch is not uncommon [32]. The likely causes of mismatch include the selection of data points analysed and the adopted graphical technique to determine the kinetic properties, the assumptions made to establish the derived oxidative reactions, and the exponential nature of Arrhenius equation where minor change of inputs translate into relatively greater change of outputs.

�

raise the material temperature from ambient to its decomposition tem­ perature. Similar to the kinetic properties, Δhr is commonly used as input for numerical fire models which describe thermal decomposition and fire spread mechanisms. Δhr is determined through Equation (5) where the numerator is the integral of reaction heat flow, (dq/dt)R be­ tween the initial and final temperatures, Ti and Tf of a reaction, divided by the heating rate, and the denominator is the loss of sample mass over the identical temperature range. � � R Tf dq dt dT⋅dT Ti dt R Δhr ¼ (5) mi mf (dq/dt)R is derived from the experimental heat flow curve by applying a user defined baseline curve to remove a few artefacts, including offset and curvature. Details of the analysis have been illus­ trated for the determination of Δhr for the 2 pyrolysis reactions of foam decomposition under nitrogen [33]. The experimental uncertainties and sensitivity of Δhr are also described in these references [28,33]. Fig. 9 and Fig. 10 show the reaction heat flow curve of NFR-SB-31 and FR-Y-36, respectively for the different heating rates investigated, rep­ resented by the solid line. Also included is the decomposition rate on the secondary y-axis for comparison, represented by the dashed line. The oxidative thermal decomposition consists of 2 pyrolysis and 3

7. Determination of heat of reaction The heat of reaction, Δhr is defined as the amount of energy either absorbed or released in order for a reaction such as oxidation or pyrol­ ysis to proceed [33]. A reaction absorbing energy is endothermic in nature whereas one that releases energy is exothermic. Within the literature, Δhr is also known as the enthalpy of reaction, the heat of pyrolysis, the heat of volatilisation, the heat of vaporisation and the heat of decomposition. There is also the heat of gasification which is different from Δhr, as this term includes the sensible heat or energy required to 10

0.12

25.0

0.12

20.0

0.10

20.0

0.10

15.0

0.08

15.0

0.08

10.0

0.06

10.0

0.06

5.0

0.04

5.0

0.04

100

200

300

-5.0

400

500

600

700

800

Temperature (°C)

0.02

0.0

0.00

-5.0

100

200

300

500

600

700

800

0.02 0.00

Temperature (°C)

(a)

(b) 0.12

100.0

0.12

40.0

0.10

80.0

0.10

30.0

0.08

60.0

0.08

20.0

0.06

40.0

0.06

10.0

0.04

20.0

0.04

0.0

100

200

-10.0

300

400

500

600

700

800

Temperature (°C)

Heat Flow (mW)

50.0

d /dT (1/°C)

Heat Flow (mW)

400

0.02

0.0

0.00

-20.0

(c)

100

200

300

400

500

Temperature (°C)

600

700

800

d /dT (1/°C)

0.0

Heat Flow (mW)

25.0

d /dT (1/°C)

Fire Safety Journal 111 (2020) 102925

d /dT (1/°C)

Heat Flow (mW)

D.S.W. Pau et al.

0.02 0.00

(d)

Fig. 10. Reaction heat flow and decomposition rate versus temperature of FR-Y-36 in air. (a) 1 � C/min; (b) 5 � C/min; (c) 20 � C/min; (d) 60 � C/min.

reflecting the oxidation of the additional char generated by the fire retardant additives. The higher char content couples with heat gener­ ated by the oxidative reactions are driving factors for the process of smoulder. These trends are in line with the literature findings that FR foams might have a greater tendency to smoulder [4,5]. Given the close overlap of the pyrolysis and oxidative reactions, it is not possible to differentiate the respective heat flow of each reaction. Applying Equa­ tion (5) over the 5-step oxidative thermal decomposition region, Δhr and the temperature range considered for NFR-SB-31 and FR-Y-36 at the different heating rates investigated are listed in Table 3. For NFR-SB-31, Δhr are fairly consistent, ranging from exothermic 4895–5301 J/g while for FR-Y-36, comparatively higher magnitude is noted, ranging from exothermic 5649–8437 J/g. This is possibly due to the oxidation of additional char from the fire retardant additives [4]. With respect to heating rate, 1 � C/min achieves significantly higher Δhr for FR-Y-36 compared to the other heating rates as noted in Table 3 suggesting this could be the result of enhanced char oxidation from the higher char yield at low heating rate. However, this is not conclusive as a gradual increase in Δhr is noted where the char yield decreases from heating rate of 5–60 � C/min. As such, char oxidation might not be the only process affecting Δhr, and a better understanding of the chemistry of the other oxidative and pyrolysis reactions is necessary. Assuming combustion efficiency of 90% with effective heat of

Table 3 Heat of reaction and reaction temperature range for NFR-SB-31 and FR-Y-36 at all heating rates. Foam

Parameter

1 � C/min

5 � C/min

20 � C/min

60 � C/min

NFR-SB-31

Ti (� C) Tf (� C) Δhr (J/g) Ti (� C) Tf (� C) Δhr (J/g)

212 512 5288 224 594 8437

228 617 4895 235 641 5649

236 662 5264 236 688 6043

244 664 5301 234 724 6767

FR-Y-36

oxidative reactions. Previously, negative heat flow is noted for the py­ rolysis reactions under nitrogen environment indicating that energy is being absorbed by these endothermic reactions [33]. The positive value of heat flow noted in Figs. 9 and 10 indicates that the oxidative thermal decomposition of foam in air is exothermic overall. This overall reaction releases energy to the surrounding as the positive heat flow from oxidative reactions are greater in magnitude than the negative heat flow from pyrolysis reactions. Comparing between NFR and FR foams, FR-Y-36 achieves higher peak heat flow than NFR-SB-31 with the exception at 60 � C/min. Furthermore, over the char oxidation region, FR-Y-36 consistently achieves higher heat flow than NFR-SB-31 11

D.S.W. Pau et al.

Fire Safety Journal 111 (2020) 102925

combustion, Δhc,eff for NFR-SB-31 and FR-Y-36 of 25400 and 23000 J/g [28], respectively, Δhr in Table 3 amounts to average 19 and 26% of the total energy released for the complete combustion of these foams. The remaining 81 and 74% are released in the gas phase upon ignition, similar to the findings in Ref. [13]. Therefore, it is evident that the oxidative thermal decomposition of FR-Y-36 with higher Δhr contributes a greater portion of the total energy released than NFR-SB-31, high­ lighting a better potential to smoulder where oxidation drives the thermal decomposition. The greater contribution is also aided by the lower Δhc,eff of FR-Y-36, a result of the fire retardant mechanisms such as solid phase char shielding effects and gas phase neutralisation of reac­ tive radicals. For the 2 pyrolysis reactions of foam and polyol, Δhr ranges from endothermic 610–1023 J/g and 164–295 J/g, respectively based on the heat flow in nitrogen [33]. Comparing with air, it is evident that significantly more energy would be released than absorbed over the oxidative thermal decomposition of foam.

References [1] T.J. Ohlemiller, J. Bellan, F. Rogers, A model of smoldering combustion applied to flexible polyurethane foams, Combust. Flame 36 (1979) 197–215, https://doi.org/ 10.1016/0010-2180(79)90060-9. [2] J.L. Torero, A.C. Fernandez-Pello, M. Kitano, Opposed forced flow smoldering of polyurethane foam, Combust. Sci. Technol. 91 (1993) 95–117, https://doi.org/ 10.1080/00102209308907635. [3] D.C. Walther, A.C. Fernandez-Pello, D.L. Urban, Space shuttle based microgravity smoldering combustion experiments, Combust. Flame 116 (1999) 398–414, https://doi.org/10.1016/S0010-2180(98)00095-9. [4] C.Y.H. Chao, J.H. Wang, Transition from smoldering to flaming combustion of horizontally oriented flexible polyurethane foam with natural convection, Combust. Flame 127 (2001) 2252–2264, https://doi.org/10.1016/S0010-2180(01) 00326-1. [5] G. Rein, Smouldering combustion phenomena in science and technology, Int. Rev. Chem. Eng. 1 (2009) 3–18. [6] K. Prasad, R. Kr€ amer, N. Marsh, M. Nyden, T. Ohlemiller, M. Zammarano, Numerical simulation of fire spread on polyurethane foam slabs, Proc. Fire Mater. 11 (2009) 697–708. [7] R.H. Kr€ amer, M. Zammarano, G.T. Linteris, U.W. Gedde, J.W. Gilman, Heat release and structural collapse of flexible polyurethane foam, Polym. Degrad. Stab. 95 (2010) 1115–1122, https://doi.org/10.1016/j.polymdegradstab.2010.02.019. [8] R. Bilbao, J.F. Mastral, J. Ceamanos, M.E. Aldea, Kinetics of the thermal decomposition of polyurethane foams in nitrogen and air atmospheres, J. Anal. Appl. Pyrolysis 37 (1996) 69–82, https://doi.org/10.1016/0165-2370(96)009369. [9] C.Y.H. Chao, J.H. Wang, Comparison of the thermal decomposition behavior of a non-fire retarded and a fire retarded flexible polyurethane foam with phosphorus and brominated additives, J. Fire Sci. 19 (2001) 137–156, https://doi.org/ 10.1106/Q56W-KUDB-0VRT-6HLF. [10] L.B. Valencia, Experimental and numerical investigation of the thermal decomposition of materials at three scales: Application to polyether polyurethane � foam used in upholstered furniture, PhD thesis, Ecole Nationale Sup�erieure de M�ecanique et d’A� erotechnique, 2009. Poitiers, France. [11] G. Rein, C. Lautenberger, A.C. Fernandez-Pello, J.L. Torero, D.L. Urban, Application of genetic algorithms and thermogravimetry to determine the kinetics of polyurethane foam in smoldering combustion, Combust. Flame 146 (2006) 95–108, https://doi.org/10.1016/j.combustflame.2006.04.013. [12] L.B. Valencia, T. Rogaume, E. Guillaume, New method for simulating the kinetic of toxic gases production of upholstered furniture fire, Proc. Fire Mater. 11 (2009) 685–695. [13] T. Rogaume, L.B. Valencia, E. Guillaume, F. Richard, J. Luche, G. Rein, J.L. Torero, Development of the thermal decomposition mechanism of polyether polyurethane foam using both condensed and gas-phase release data, Combust. Sci. Technol. 183 (2011) 627–644, https://doi.org/10.1080/00102202.2010.535574. [14] S.G. Viswanath, M.C. Gupta, Estimation of nonisothermal kinetic parameters from a TG curve by the methods of overdetermined system and inflection point, Thermochim. Acta 285 (1996) 259–267, https://doi.org/10.1016/0040-6031(96) 02917-6. [15] D.S.W. Pau, C.M. Fleischmann, M.A. Delichatsios, Thermal decomposition of flexible polyurethane foams in air, Proc. Int. Semin. Fire Explos. Hazards 9 (2019) 1097–1108, https://doi.org/10.18720/spbpu/2/k19-31. [16] M. Schubnell, Temperature and heat flow calibration of a DSC-instrument in the temperature range between -100 and 160 � C, J. Therm. Anal. Calorim. 61 (2000) 91–98, https://doi.org/10.1023/A:1010156407261. [17] H.S. Kalsi, Electronic Instrumentation, third ed., Tata McGraw-Hill, New Delhi, 2010. [18] G.W.H. H€ ohne, W.F. Hemminger, H.-J. Flammersheim, Differential Scanning Calorimetry, second ed., Springer, Berlin, 2003. [19] H.C. Ashton, Fire retardants, in: M. Xanthos (Ed.), Functional Fillers for Plastics, Wiley-VCH, Weinheim, 2010, pp. 309–350. [20] S.V. Levchik, Introduction to flame retardancy and polymer flammability, in: A. B. Morgan, C.A. Wilkie (Eds.), Flame Retardant Polymer Nanocomposites, John Wiley & Sons Inc., New Jersey, 2007, pp. 1–30. [21] N. Grassie, G.A.P. Mendoza, Thermal degradation of polyether-urethanes: Part 4–Effect of ammonium polyphosphate on the thermal degradation of polyetherurethanes prepared from methylene bis(4-phenylisocyanate) and low molecular weight poly(ethylene glycols), Polym. Degrad. Stab. 11 (1985) 145–166, https:// doi.org/10.1016/0141-3910(85)90107-7. [22] D. Price, Y. Liu, G.J. Milnes, R. Hull, B.K. Kandola, A.R. Horrocks, An investigation into the mechanism of flame retardancy and smoke suppression by melamine in flexible polyurethane foam, Fire Mater. 26 (2002) 201–206, https://doi.org/ 10.1002/fam.810. [23] C. Denecker, J.J. Liggat, C.E. Snape, Relationship between the thermal degradation chemistry and flammability of commercial flexible polyurethane foams, J. Appl. Polym. Sci. 100 (2006) 3024–3033, https://doi.org/10.1002/app.23701. [24] BS 5852, Methods of Test for Assessment of the Ignitability of Upholstered Seating by Smouldering and Flaming Ignition Sources, British Standards Institution, London, 2006, 2006. [25] Technical Bulletin 117-2013, Requirements, Test Procedure and Apparatus for Testing the Smolder Resistance of Materials Used in Upholstered Furniture, Bureau Household Goods and Services, Sacramento, 2019. [26] Federal Aviation Regulations, Appendix F to Part 25, Part I–Test Criteria and Procedures for Showing Compliance with x25.853 or x25.855, Federal Aviation Administration, Washington, 1972.

8. Conclusions The oxidative thermal decomposition of PU foams is represented by a total of 5 reactions, 2 pyrolysis and 3 oxidative. From TG experiments, the 2 pyrolysis reactions demonstrated consistent trend with increasing heating rate while the additional 3 oxidative reactions are noted to be heating rate dependent. A simplified model is developed and applied to simulate the experimental behaviour. The model captures the essential heating rate dependent trend where the influence of foam oxidation reduces and the influence of polyol pyrolysis increases towards higher heating rate. From the kinetic properties, a compensating effect is noted for A and E where the increase of one parameter leads to the increase of the other. For n, the significant reduction to the influence of foam oxidation towards high heating rate leads to the increase of n which is noted for FR-Y-36. The fire retardant additives of FR-Y-36, hal­ ophosphate and melamine appear to show characteristics which potentially improve the foam’s fire performance. These include the notable decomposition over lower temperature region at low heating rate where the gaseous fuel released could be neutralised or diluted by the gas phase fire retardant mechanisms, and at high heating rate, the decomposition occurs over greater temperature range with lower magnitude due to char formation. However, the literature has also showed that FR foams can be prone to smouldering combustion due to the complex char formation and oxidation. This is evidenced by the higher exothermic heat flow and heat of reaction due to the higher char yield of FR-Y-36. Δhr of FR-Y-36 amounts to approximately 26% of the total heat released of complete combustion, greater than 19% achieved by NFR-SB-31. The toxicity of smouldering combustion and its likelihood of transitioning into flaming combustion can diminish the fire performance of FR foams. Going for­ ward, the research can benefit from investigating the impact of varying local oxygen concentration has on the oxidative thermal decomposition, and also the gaseous species production and reaction chemistry. These findings might explain the trends observed in the heat flow curve and the heat of reaction at different heating rates. Declaration of competing interest None. Acknowledgements The authors would like to acknowledge the Chemistry Department for offering the experimental equipment; the lab technicians, Mr. Mike Van der Colk and Mr. Grant Dunlop for their technical support; Dr. KaiYuan Li for the intellectual discussions on the topic of material decom­ position. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. 12

D.S.W. Pau et al.

Fire Safety Journal 111 (2020) 102925

[27] AS/NZS 1530.3:1999, Methods for Fire Tests on Building Materials, Components and Structures - Simultaneous Determination of Ignitability, Flame Propagation, Heat Release and Smoke Release, Standards Australia, Sydney, 1999. [28] D.S.W. Pau, A Comparative Study on Combustion Behaviours of Polyurethane Foams with Numerical Simulations Using Pyrolysis Models, PhD Thesis, University of Canterbury, Christchurch, New Zealand, 2013. [29] D.S.W. Pau, C.M. Fleischmann, M.J. Spearpoint, K.Y. Li, Determination of kinetic properties of polyurethane foam decomposition for pyrolysis modelling, J. Fire Sci. 31 (2013) 356–384, https://doi.org/10.1177/0734904113475858. [30] X. Wang, C.M. Fleischmann, M.J. Spearpoint, Parameterising study of tunnel experiment materials for application to the Fire Dynamics Simulator pyrolysis

model, J. Fire Sci. 34 (2016) 490–514, https://doi.org/10.1177/ 0734904116667738. [31] X. Wang, Fire Dynamics Simulator (FDS) Pyrolysis Model Analysis of Heavy Goods Vehicle Fires in Road Tunnels, PhD Thesis, University of Canterbury, Christchurch, New Zealand, 2017. [32] F. Hache, M. Delichatsios, T. Fateh, J. Zhang, Comparison of methods for thermal analysis: application to PEEK and a composite PEEKþCF, J. Fire Sci. 33 (2015) 232–246, https://doi.org/10.1177/0734904115584154. [33] D. Pau, C. Fleischmann, M. Spearpoint, K.Y. Li, Sensitivity of heat of reaction for polyurethane foams, in: D. Nilsson, P. van Hees, R. Jansson (Eds.), Fire Safety Science–Proceedings of the Eleventh International Symposium, 2014, pp. 179–192, https://doi.org/10.3801/IAFSS.FSS.11-179.

13