Competition in Aluminium Phosphinate-Based Halogen-Free Flame Retardancy of Poly(Butylene Terephthalate) and Its Glass-Fibre Composites

DOI 10.1515/epoly-2014-0029 e-Polymers 2014; 14(3): 193–208

Sven Brehme, Thomas Köppl, Bernhard Schartel* and Volker Altstädt Competition in aluminium phosphinate-based halogen-free flame retardancy of poly(butylene terephthalate) and its glass-fibre composites

Abstract: Aluminium diethylphosphinate (AlPi-Et) and electrical and electronic products as well as in automotive inorganic aluminium phosphinate with resorcinol-bis(di- and transportation applications because of its combina- 2,6-xylyl phosphate) (AlPi-H+RXP) were compared with tion of advantageous properties (1). In the past, the good each other as commercially available halogen-free flame flame-retardancy performance required for most applica- retardants in poly(butylene terephthalate) (PBT) as well tions was achieved using halogenated flame retardants as in glass-fibre-reinforced PBT (PBT/GF). Pyrolysis (2). Nowadays, flame-retardant polymers are expected by behaviour and flame retardancy performance are reported the market and environmental regulations to be halogen in detail. AlPi-H+RXP released phosphine at very low free (3). Phosphorus-based compounds are the most prom- temperatures, which can become a problem during pro- ising alternative among the halogen-free flame retardants cessing. AlPi-Et provided better limiting oxygen index currently in use. (LOI) values and UL 94 ratings for bulk and PBT/GF than Today, metal phosphinates play a major role in the AlPi-H+RXP. Both flame retardants acted via three differ- flame retardancy of PBT. They can be divided into organic ent flame-retardancy mechanisms in bulk as well as in phosphinates and inorganic phosphinates. Organic PBT/GF, namely, flame inhibition, increased amount of phosphinates are salts of dialkyl, diaryl, or alkyl-aryl char, and a protection effect of the char. AlPi-Et was more phosphinic acid, respectively. Today’s products are based efficient in decreasing the total heat evolved of PBT in on formulations containing aluminium or zinc organic the cone calorimeter test. AlPi-H+RXP reduced the peak phosphinates alone (4–6) or in combination with other heat release rate of PBT more efficiently than AlPi-Et. An substances such as melamine cyanurate (7). Inorganic optimum loading of AlPi-Et in PBT/GF was found, which phosphinates, also called hypophosphites, are salts of was below the supplier’s recommendation. This loading non-substituted phosphinic acid; that is, mainly alumin- provides a maximum increase in LOI and a maximum ium hypophosphite has been proposed with and without decrease in total heat evolved. synergists as flame retardants for polyesters (8, 9). The mode of action of organic metal phosphinates Keywords: aluminium phosphinate; flame retardancy; in polyesters is known to a certain extent. Aluminium glass-fibre composite; poly(butylene terephthalate). diethylphosphinate (AlPi-Et) and zinc diethylphosphinate (ZnPi-Et) show a strong gas-phase activity in PBT and a slightly increased residue (10). The formation of *Corresponding author: Bernhard Schartel, BAM Federal Institute of Materials Research and Testing, Unter den Eichen 87, 12205 Berlin, intumescent residue was reported for AlPi-Et in PBT Germany, Tel.: +49 30 8104 1021, Fax: +49 30 8104 1747, (11) and for ZnPi-Et in poly(ethylene terephthalate) (12). e-mail: [email protected] Synergistic improvements in flame retardancy occurred Sven Brehme: BAM Federal Institute of Materials Research and for the combinations of AlPi-Et with nanometric metal Testing, Unter den Eichen 87, 12205 Berlin, Germany oxides in PBT (13, 14). ZnPi-Et combined with nanopar- Thomas Köppl and Volker Altstädt: Department Polymer Engineering, University of Bayreuth, Universitätsstr. 30, 95447 ticles is proposed in polyesthers (15, 16). Most studies Bayreuth, Germany on inorganic metal phosphinates deal with aluminium hypophosphite (AlPi-H); lanthanum and cerium hypophosphite, however, have also been investigated (17). The flame retardancy mechanisms of AlPi-H have been discussed in very few works only. The condensed- 1 Introduction phase activity of AlPi-H in glass fibre-reinforced PBT (PBT/GF) was proposed, whereas gas-phase activity Poly(butylene terephthalate) (PBT) is one of the most was not discussed (18). The degree of understanding important engineering thermoplastics. It is widely used in concerning inorganic metal phosphinates is somewhat 194 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants lower than that for organic metal phosphinates. Fur- An injection mould machine (Allrounder 320 S 500-150, thermore, works in which organic and inorganic metal ARBURG, Germany) with a melt temperature of 250°C phosphinates were compared directly with each other and a mould temperature of 70°C was prepared to mould are very rare, although such comparison is very interest- the compact and the reinforced specimen. Tables 1 and 2 ing, especially with regard to application. provide an overview of the composition of the compact Thus the aim of this work was to provide a better under- and glass fibre-reinforced materials, respectively. standing of AlPi-H’s mode of action in PBT, as well as in PBT/GF, and to show a direct comparison with AlPi-Et. The pyrolysis of the two flame retardants in PBT and the possible 2.2 Morphology interactions during decomposition were assessed in detail using thermogravimetry coupled with FTIR spectroscopy. The morphology of the flame retardants in PBT was The flammability and burning behaviour were assessed by characterised by transmission electron microscopy. The LOI, UL 94, and the cone calorimeter test, respectively. The samples were cut by a microtome UltraCut E instrument active flame-retardancy mechanisms were identified and from Leica and, afterwards, investigated on a Zeiss 902 quantified. The efficiency of the flame retardants, i.e., their transmission electron microscope with an acceleration performance per unit of phosphorus, was assessed. AlPi- voltage of 80 kV. The fracture surfaces of the glass fibre- H+resorcinol-bis(di-2,6-xylyl phosphate) (RXP) released reinforced samples were analysed by a scanning elec- phosphine at relatively low temperatures, which could tron microscope (JSM-IC 848, Jeol) with an acceleration cause problems during compounding. AlPi-Et provided voltage of 15 kV. better LOI values and UL 94 ratings than AlPi-H+RXP. An optimum loading was found for the use of AlPi-Et in PBT/ GF, which was even below the supplier’s recommendation. 2.3 Thermal analysis Thus, this study provides some very important results for the efficient application of the two flame retardants. Thermal analysis experiments were performed on a thermo-microbalance instrument with an automatic sample charger (TG 209 F1 Iris, Netzsch, Selb, Germany). 2 Experimental

Table 1 Composition of the investigated bulk materials. 2.1 Materials and sample preparation Material P-content PBT AlPi-Et AlPi-H+RXP The poly(butylene terephthalate) used in this study was (wt.%) (wt.%) (wt.%) (wt.%) of commercial injection moulding grade (Ultradur B4520, PBT 0 100 – – PBT/AlPi-Et6.3 1.5 93.7 6.3 BASF SE, Germany; TM = 223°C). Exolit OP 1240 consisting of aluminium diethylphosphinate (AlPi-Et) was obtained PBT/AlPi-Et20 4.8 80 20 – PBT/AlPi-H+RXP4.2 1.5 95.8 – 4.2 from Clariant Produkte Deutschland GmbH, Germany. PBT/AlPi-H+RXP20 7.1 80 – 20 Phoslite B85AX consisting of 85 wt.% aluminium phosphinate and 15 wt.% resorcinol-bis(di-2,6-xylyl phosphate) (AlPi-H+RXP) was obtained from Italmatch Chemicals, Table 2 Composition of the investigated glass fibre-reinforced Italy. The materials were melt compounded using a co- materials. rotating twin-screw extruder (Megacompounder ZSK 26 Material P-content PBT/30% AlPi-Et AlPi- MCC, Werner & Pfleiderer, Germany) with a length-to- (wt.%)a glass fibre (wt.%) H+RXP diameter ratio of 44 at a maximum temperature of 250°C. (wt.%) (wt.%) The dried polymer and the flame-retardant additives were PBT/GF 0 100 0 – added through the main feeder. PBT/GF/AlPi-Et4.4 1.0 95.6 4.4 – PBT/GF was produced by adding 30 wt.% glass fibres PBT/GF/AlPi-Et6.3 1.5 93.7 6.3 – from PPG (Fibre Glass ChopVantage HP 3786) with a diam- PBT/GF/AlPi-Et14 3.3 86 14 – eter of 10 μm and an initial length of 4.5 mm. Glass fibres PBT/GF/AlPi-Et20 4.8 80 20 – were added via a co-rotating side feed near the die. PBT/GF/AlPi-H+RXP4.2 1.5 95.8 – 4.2 Test specimens were produced by injection moulding PBT/GF/AlPi-H+RXP20 7.1 80 – 20 after the compounded material was dried for 4 h at 100°C. aWith respect to the whole material. S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 195

Cryo-milled samples of 10 mg were heated in alumina reproducibility was satisfactory (below 10% deviation for pans from 30°C to 900°C at a heating rate of 10°C/min and all characteristics). a constant nitrogen flow rate of 30 ml/min. Blank measurements were used to correct apparatus-specific devia- tions including buoyant forces. Fourier transform infrared (FTIR) spectrometry 3 Results and discussion (TENSOR 27, Bruker Optics, Ettlingen, Germany) coupled with thermogravimetric analysis was used to investigate 3.1 Morphology of the materials volatile decomposition products. The coupled measurements were run with the same parameters as the thermal Both systems, PBT/AlPi-Et and PBT/AlPi-H+RXP, showed analysis, except that a sample weight of 5 mg was used. The a composite structure with particle sizes between a few entire purge-gas flow was transferred to the spectrometer’s micrometers and a nanometre scale (Figure 1A and B). The gas cell. The transfer line and the gas cell were maintained composites differed in terms of the shape and size of the at 270°C during the measurements. Such a high tempera- particles. The AlPi-Et particles showed a rod-like shape ture was necessary to successfully detect diethylphosphinic with a median length of 240 nm. In contrast, the AlPi-H acid and vaporised AlPi-Et. These compounds would not be and RXP particles, respectively, had a significantly lower able to reach the gas cell if a temperature of only 230°C was median length of 110 nm and a spherical shape. Neverthe- used. FTIR spectra were recorded with an optical resolu- less, the particle dispersion in the two composites was tion of 4 cm-1 and in the spectral range of 600–4000 cm-1. quite similar. The spectra were evaluated with regard to characteristic In Figure 1C and D, the morphology of two glass fibre- absorption bands. Reference spectra from databases were reinforced composites is presented as scanning electron used to identify the decomposition products. microscopy images of the fracture surface. The AlPi-Et A Vertex 70 FTIR spectrometer (Bruker Optics) equipped composites show a strong fibre-matrix adhesion, which with an FTIR600 Linkam stage (Linkam Scientific Instru- can be seen in wetted fibres. In contrast, the glass fibres ments Ltd., Chilworth, UK) was used to analyse changes in in the composites with AlPi-H+RXP show relatively poor the condensed phase. Sample powder was melted on a KBr adhesion to the matrix. Therefore an influence on the disc to prepare thin films. The disc was then fixed by a clamp mechanical behaviour is expected, perhaps in combina- in the Linkam stage. A nitrogen flow rate of 150 ml/min was tion with an influence on melt flow and dripping. purged through the Linkam stage during heating from 35°C to 600°C at a heating rate of 10°C/min. The spectra were recorded in the spectral range of 400–4000 cm-1 using the 3.2 Pyrolysis: mass loss and residue transmission mode with a resolution of 4 cm-1. Attenuated total reflection (ATR) IR spectra were PBT decomposed via one decomposition step (Figure 2A recorded on a Tensor 27 FTIR spectrometer (Bruker Optics) and B) at a temperature of maximum mass loss rate (Tmax) in the spectral range of 400–4000 cm-1 with a resolution of around 400°C, leaving a 7.5 wt.% residue (Table 3). The of 4 cm-1. addition of AlPi-Et slightly decreased the starting tempera-

ture of decomposition, whereas Tmax remained unchanged. The materials containing AlPi-Et showed a small second 2.4 Fire behaviour decomposition step (3.7 and 7.5 wt.%, respectively) sub-

sequently to the main decomposition step (Tmax ∼478°C). The flammability (reaction to a small flame) of the Adding 20 wt.% AlPi-Et to PBT increased the residue to materials was determined by the limiting oxygen index 12.6 wt.%. The addition of AlPi-H+RXP decreased the start- (LOI) according to ISO 4589 (specimen dimensions ing temperature of decomposition more strongly than that 150 × 10 × 4 mm) and by UL 94 according to IEC 60695-11- of AlPi-Et. The temperature of the maximum mass loss 10 (specimen dimensions 150 × 12.9 × 3.2 mm). Cone calo- rate was not changed for PBT/AlPi-H+RXP4.2, and a small rimeter testing (Fire Testing Technology, East Grinstead, “early” decomposition step prior to the main step was UK) according to ISO 5660 was used to assess the fire indicated. PBT/AlPi-H+RXP20 clearly showed this small behaviour under forced-flaming conditions. Plate-shaped decomposition step prior to the main step. The main decom- specimens (100 × 100 × 3 mm) were placed in aluminium position step of PBT/AlPi-H+RXP20, i.e., the temperature of trays and exposed to the irradiation (50 kW/m2) of the the maximum mass loss rate, shifted to temperatures lower cone heater. Each material was tested only twice if the than for PBT and the residue increased to 18.8 wt.%. Thus 196 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants

1 µm 1 µm

10 µm 10 µm

Figure 1 Morphology of the investigated materials: Transmission electron microscopy images of (A) PBT/AlPi-Et6.3 and (B) PBT/AlPi- H+RXP4.2. Scanning electron microscopy images of the fracture surface of (C) PBT/GF/AlPi-Et20 and (D) PBT/GF/AlPi-H+RXP20.

PBT/AlPi-H+RXP20 produced more residue than PBT/AlPi- 3.3 Pyrolysis: volatile decomposition Et20. Nevertheless, the amount of residue corresponds to products the calculated residue, assuming a superposition of the contributions of the individual components. Thus there PBT released butadiene, tetrahydrofuran (THF), benzoic is no evidence for any interaction between PBT and AlPi- acid, butyl benzoate, CO, and CO2 as volatile decompo- H+RXP that increases charring. The starting temperature sition products during the maximum of decomposition. of decomposition was 8°C lower than expected from super- These results are in agreement with the literature results position of the individual components of PBT/AlPi-H+RXP. (19). The main decomposition step of PBT/AlPi-Et was The temperatures of the maximum mass loss rates of both characterised by the release of the main decomposi- decomposition steps were 12°C lower than expected from a tion products of PBT. Additionally, the release of diethyl superposition. Thus AlPi-H+RXP slightly destabilised PBT. phosphinic acid (3649, 854, and 770 cm-1) was detected.

100 6 ABPBT C 2320 90 PBT/AIPi-Et6.3 5 80 PBT/AIPi-Et20

PBT/AIPi-H+RXP4.2 1065 2410 991 1121 70 PBT/AIPi-H+RXP20 4 2249 60 50 3 a 40 1742 1265

1101 909 Mass (wt.%) Absorbance

2 1018 2996

30 3084 3107 2980 1605 1584 1809 1832 3583 2350 Mass loss rate (wt.%/s)

20 2308 2872 1 669 2183 10 2100 b 0 0 100 200 300 400 500 600 700 800 900 300 350 400 450 500 4000 3500 3000 2500 2000 1500 1000 Temperature (°C) Temperature (°C) Wavenumber (cm-1)

Figure 2 (A) Mass loss and (B) mass loss rate of the bulk materials and (C) gas-phase FTIR spectra of PBT/AlPi-H+RXP20 taken at curve a (30 min, 329°C) and at curve b (36.5 min, 394°C). S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 197

Table 3 Thermal decomposition characteristics of the investigated bulk materials.

Material T(2%) Tmax early Mass loss early Tmax main Mass loss main Tmax second Mass loss second Residue at (°C) step (°C) step (wt.%) step (°C) step (wt.%) step (°C) step (wt.%) 800°C (wt.%)

PBT 361 – – 399 92.4 – – 7.5 PBT/AlPi-Et6.3 357 – – 401 88.7 473 3.7 7.0 PBT/AlPi-Et20 353 – – 402 79.6 483 7.5 12.6 PBT/AlPi-H+RXP4.2 353 a a 401 91.8 – – 7.7 PBT/AlPi-H+RXP20 323 322 4.5 392 76.3 – – 18.8

Error ± 2°C and ± 1.0 wt.%. aA small decomposition step prior to the main step was indicated, but it was not significant and thus was not considered.

A comparison with the ATR-IR spectra of pure AlPi-Et band at 1690 cm-1. PBT/AlPi-Et formed polyaromatic char revealed that a part of AlPi-Et itself (1457, 1410, 1136, structures, as well as terephthalic acid structures. Addi- 1062, and 770 cm-1) vaporised. The second decomposition tionally, phosphorus-related bands at 1371 and 1277 cm-1 step was characterised by the release of butyl benzoate, (P = Osym, asym) remained after decomposition. -1 - benzoic acid, and diethyl phosphinic acid, as well as by Two bands of AlPi-H (1165 and 1071 cm , PO2 ) over- -1 the vaporisation of AlPi-Et. These results are in agreement lapped with the O–CH2-vibration (1103 cm ) of PBT in with the work of Braun et al. (10). PBT/AlPi-H+RXP. The condensed-phase analysis of PBT/ PBT/AlPi-H+RXP released phosphine (2410, 2320, AlPi-H+RXP (Figure 3A and B) showed a decrease in the 2249, 1121, 1065, and 991 cm-1) at about 322°C during a absorption bands associated with P-H bonds (stretching small decomposition step prior to the main step (Figure vibration: 2408 and 2385 cm-1; deformation vibration: 2C, curve a). Phosphinate or phosphinic acid, respec- 827 cm-1). This corresponds to the decomposition of the tively, disproportionates to phosphine or phosphonic phosphinate. The bands associated with PBT, e.g., aliphaic acid at elevated temperatures. The phosphonic acid dis- C-H bonds (2963 cm-1) and ester bonds (C = O 1714 cm-1, C-O proportionates further to phosphine and phosphoric acid 1269 cm-1), decreased when PBT decomposed. The for- (20, 21). The release of butadiene (3107, 3084, 2996, 1832, mation of terephthalic acid was indicated by a band at 1809, 1605, 1584, 1013, and 909 cm-1), CO (2183 and 2100 1690 cm-1. A broad absorption band at 1094 cm-1, which was -1 -1 cm ), CO2 (2350, 2308, and 669 cm ), benzoic acid (3583, attributed to phosphorus, did not decrease as much as the 1761, 1182, and 1083 cm-1), butyl benzoate (2968, 1742, 1265, bands of PBT did. This broad absorption band dominated and 1101 cm-1), and THF (2980, 2872, and 1083 cm-1) was the condensed-phase FTIR spectrum of PBT/AlPi-H+RXP detected during the main decomposition step of PBT/AlPi- at 600°C (Figure 3C, curve a). The small bands at 1692, H+RXP (Figure 2C, curve b). Thus the main decomposition 1575, 1508, 1423, 1271, 1020, 885, and 731 cm-1 originated step was characterised by the decomposition of the PBT from terephthalic acid, which condensed on the Linkam fraction in PBT/AlPi-H+RXP. cell’s window. The broad absorptions centred at 1094 and The release of volatile phosphorus compounds by 400 cm-1 were attributed to aluminium phosphate. The AlPi-Et and AlPi-H+RXP indicates the potential of both ATR-IR spectrum of the residue, which was obtained from flame retardants for flame inhibition in the gas phase. a thermogravimetry experiment with pure AlPi-H+RXP (Figure 3C, curve b), was very similar to the FTIR spectrum of the low cristobalite form of aluminium phosphate reported in the literature (24). This proves the formation of 3.4 Pyrolysis: changes in the condensed aluminium phosphate from AlPi-H+RXP. phase

The typical absorption bands of PBT’s condensed phase 3.5 Pyrolysis: decomposition pathway -1 -1 -1 (22), i.e., 2963 cm (aliphatic CH2), 1713 cm (C = O), 1267 cm -1 -1 (C-O ester), 1104 cm (O-CH2), and 728 cm (CH aromatic The decomposition mechanism of PBT is well known (19). ring), decreased as decomposition proceeded. The appear- Chain scission starts via a six-membered transition state, ance and increase of a band at 1557 cm-1 was attributed to leading to the formation of butadiene and terephthalic the formation of polyaromatic char structures (23). The acid, which further decompose to benzoic acid and CO2. formation of terephthalic acid, which was found to con- Alternatively, PBT’s decomposition starts with the hydroly- dense on the Linkam cells’ window, was indicated by a sis of the ester bond leading to the formation of terephthalic 198 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants

2963 24082385 1269 1190 827 1262 AB2958 C 1714 1723 1020 1271 1094 1692 1508 1103 885 a 731 e 1423 1575 445 b f 1109 a

c g 558 Absorbance Absorbance Absorbance 1690 72 3 71 2 1233

d h 62 4

e i b 4000 3500 3000 2500 2000 1500 1000 500 4000 3500 3000 2500 2000 1500 1000 500 4000 3500 3000 2500 2000 1500 1000 500 Wavenumber (cm-1) Wavenumber (cm-1) Wavenumber

Figure 3 (A) Condensed-phase FTIR spectra of PBT/AlPi-H+RXP20 taken at curve a (0 min, 30°C), curve b (30 min, 329°C), curve c (33 min, 357°C), and curve d (34 min, 366°C); (B) condensed-phase FTIR spectra of PBT/AlPi-H+RXP20 taken at curve e (35 min, 379°C), curve f (40 min, 428°C), curve g (43 min, 456°C), curve h (45 min, 478°C), and curve i (50 min, 528°C); and (C) condensed-phase FTIR spectrum of PBT/ AlPi-H+RXP20 taken at curve a (60 min, i.e., 2.5 min after the Linkam cell reached 600°C) and at curve b: ATR-IR spectrum of the residue of pure AlPi-H+RXP obtained from a thermogravimetry experiment.

acid and 1,4-butanediol. The 1,4-butanediol eliminates acid, CO, and CO2, the mixed salts decompose to aluminium either one or two molecules of water to form THF or buta- phosphates during the second small decomposition step. diene, respectively. PBT/AlPi-Et released the same decom- The decomposition of PBT/AlPi-H+RXP (Scheme 1) position products as PBT. Thus the addition of AlPi-Et may started with the decomposition of aluminium phosphi- have changed the decomposition mechanism of PBT to only nate to phosphine and aluminium phosphate. AlPi-H a limited extent. Some of the AlPi-Et vaporised and some decomposed prior to the main step. It slightly destabi- decomposed, releasing diethyl phosphinic acid. The forma- lised PBT due to phosphorus acids that formed upon the tion of mixed aluminium phosphinate carboxylate as inter- decomposition of AlPi-H, catalysing the hydrolytic scis- mediate products in the condensed phase was proposed sion of PBT’s ester bond. The change in the decomposi- for PBT/AlPi-Et (10, 25). With the release of ethene, benzoic tion of PBT by the addition of AlPi-H+RXP was limited

Scheme 1 Decomposition pathway of PBT with aluminium phosphinate (AlPi-H). Products in frames were proved in the gas phase. S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 199 to this slight destabilisation, as the same decomposition Table 4 Flammability (reaction to a small flame) and ignitability products as for pure PBT were detected during the main of the bulk materials determined by LOI (error = ± 1.0%), UL 94, and time to ignition (t ) in the cone calorimeter with an irradiation of decomposition step of PBT/AlPi-H+RXP. ign 50 kW/m2.

Material P content (wt.%) LOI (%) UL 94 t (s) 3.6 Fire behaviour of the bulk materials: ign flammability PBT 0 23.3 HB 54 ± 1 PBT/AlPi-Et6.3 1.5 27.5 V-2 43 ± 8 PBT is a flammable polymer with an LOI of 23.3% and an PBT/AlPi-Et20 4.8 52.0 V-0 44 ± 9 PBT/AlPi-H+RXP4.2 1.5 23.2 V-2 48 ± 1 HB classification in the UL 94 test (Table 4). The LOI of PBT/AlPi-H+RXP20 7.1 26.8 V-2 32 ± 2 PBT was increased strongly by adding AlPi-Et, whereas the addition of AlPi-H+RXP yielded only a slight increase The error of tign was based on the maximum deviation of measured in LOI. The UL 94 classification of PBT was improved to values from the average value. Some specimens deformed strongly upon exposure to the cone heater, thus causing a high deviation V-2 with 6.3 wt.% AlPi-Et and to V-0 with 20 wt.% AlPi- in t . Et, respectively. Adding 4.2 wt.% AlPi-H+RXP to PBT ign improved the UL 94 classification to V-2. Increasing the amount of AlPi-H+RXP to 20 wt.% yielded no further improvement in the UL 94 rating because flaming drip- 3.7 Fire behaviour of the bulk materials: ping still occurred. The formation of a flow limit was forced-flaming combustion reported for PBT/AlPi-Et20 but not for PBT/AlPi-H+RXP20 (26), which explains the different UL 94 ratings of these The heat release rate curves (Figure 4A and B) show a two materials. The time to ignition was decreased by both similar decrease in peak heat release rate (pHRR) and flame retardants. The decrease was stronger with 20 wt.% total heat evolved (THE; i.e., total heat released taken AlPi-H+RXP than with 20 wt.% AlPi-Et. This is consistent at flameout) of PBT/AlPi-Et6.3 and PBT/AlPi-H+RXP4.2. with the earlier decomposition of PBT/AlPi-H+RXP in the These two materials have the same phosphorus content thermal analysis. of 1.5 wt.%. PBT/AlPi-Et20 showed a strongly decreased

2000 90 PBT 1800 A PBT/AIPi-Et6.3 80 PBT/AIPi-Et20 1600 PBT/AIPi-H+RXP4.2 70 1400 PBT/AIPi-H+RXP20 60 ) ) 2 1200 2 50 1000 40 800 THR (MJ/m HRR (kW/m 30 600 400 20 200 B 10 0 0 050 100 150 200 250 050 100 150 200 250 Time (s) Time (s)

C DE

Upper side Lower side

Figure 4 (A) Heat release rate and (B) total heat release of the bulk materials during the cone calorimeter test with an irradiation of 50 kW/m2. Fire residues from (C) PBT/AlPi-Et20 and (D) PBT/AlPi-H+RXP20 obtained at 50 kW/m2. (E) Detailed photo shows the upper and the lower sides of the PBT/AlPi-H+RXP20 residue. 200 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants pHRR and an HRR curve shape that is typical for non- 3.8 Fire behaviour of the bulk materials: charring materials (27). This is because charring and quantitative assessment of flame-retar- intumescence occurred by the time of pHRR (122 s) and dancy mechanisms thus could hardly influence the HRR. Nevertheless, a protective effect of the intumescent char (Figure 4C) was Gas-phase activity and charring were quantified by the expected. PBT/AlPi-H+RXP20 showed a strong decrease decrease in the effective heat of combustion and the in pHRR and an HRR curve without a pronounced peak increase in the amount of residue, respectively. The at the end. This curve shape corresponds to a charring residue was increased from 4 wt.% (PBT) to 11 wt.% (PBT/ material (27). The HRR rose until an efficient compact AlPi-Et20). Hence, the amount of fuel released by PBT/ char layer was formed (Figure 4D and E). Then the HRR AlPi-Et20 was reduced to 93% compared to PBT. The THE/ remained constant. One side of the char layer curled TML of PBT/AlPi-Et20 was reduced to 62% due to flame upward during the cone calorimeter test (Figure 4D). inhibition. Combining reduced fuel release and flame Thus the efficiency of the char’s barrier effect was inhibition yielded a reduction to 58% (0.93 × 0.62 = 0.58), reduced. The HRR rose slightly and formed a broad peak which corresponds well with the reduction in THE (59%). at 100 s. Thus the reduced THE of PBT/AlPi-Et20 in comparison Both flame retardants decreased pHRR and THE to PBT is explained well by flame inhibition and reduced as increasing amounts of flame retardant were added fuel release. The pHRR was decreased to 36%. In the first (Table 5). The decrease in pHRR was stronger with AlPi- approximation, a reduction in pHRR to 58% was due to H+RXP. In the comparison of materials with the same gas-phase activity and increased charring as discussed for phosphorus content (1.5 wt.% P), the decrease in THE THE. The protective effect of the intumescent char caused was stronger with AlPi-Et than with AlPi-H+RXP. With a further reduction in pHRR from 58% to 36%, i.e., a rela- the materials with 20 wt.% of flame retardant, the reduc- tive reduction of 38%. tion in THE was stronger with AlPi-H+RXP than with Analogous considerations were made concerning AlPi-Et. The amount of residue was increased by both PBT/AlPi-H+RXP20. The increase in residue to 20% cor- flame retardants as increasing amounts were added. responded to a reduction in fuel release to 83%. THE/ The increase in residue was stronger with AlPi-H+RXP. TML was decreased to 62% due to flame inhibition. Hence, the condensed-phase activity of AlPi-H+RXP was The combination of flame inhibition and reduced fuel stronger. Both flame retardants released compounds con- (0.62 × 0.83 = 0.51) offers a good explanation for the reduc- taining phosphorus into the gas phase, where they acted tion in THE (54%). The pHRR of PBT/AlPi-H+RXP was as flame inhibitors, decreasing the effective heat of com- reduced to 22%. In the first approximation, a reduction bustion (THE/TML). The decrease in THE/TML was simi- in pHRR to 51% was explained by flame inhibition and larly strong for AlPi-Et and AlPi-H+RXP. Hence, the two reduced fuel. The additional decrease in pHRR from 51% flame retardants showed similarly strong gas-phase activ- to 22% corresponded to a relative reduction of 57% and ity. Flame inhibition led to incomplete combustion. The was due to the protective effect of the compact char layer. increase in total smoke release (TSR) and CO yield was Three different flame retardancy mechanisms are due to an increased amount of incomplete combustion active in PBT/AlPi-Et20 and PBT/AlPi-H+RXP20. Con- products and thus supports flame inhibition as one of the densed-phase activity (charring) is strong in PBT/AlPi- main flame-retardancy mechanisms active in PBT/AlPi-Et H+RXP20 (20 wt.% residue), whereas it plays a minor role and PBT/AlPi-H+RXP. in PBT/AlPi-Et20. Gas-phase activity (flame inhibition)

Table 5 Cone calorimeter results of the bulk materials obtained using an irradiation of 50 kW/m2 including peak heat release rate (pHRR), total heat evolved (THE), total mass loss (TML), and total smoke release (TSR).

Material P content pHRR THE Residue THE/TML TSR CO yield (wt.%) (kW/m2) (MJ/m2) (wt.%) (MJ/m2 g) (m2/m2) (kg/kg)

PBT 0 1812 ± 100 76 ± 1 4 ± 1 2.1 ± 0.1 1525 ± 100 0.06 ± 0.01 PBT/AlPi-Et6.3 1.5 1412 ± 100 60 ± 1 5 ± 1 1.7 ± 0.1 2488 ± 100 0.10 ± 0.02 PBT/AlPi-Et20 4.8 657 ± 70 45 ± 1 11 ± 1 1.3 ± 0.1 3486 ± 100 0.16 ± 0.02 PBT/AlPi-H+RXP4.2 1.5 1188 ± 100 66 ± 1 7 ± 1 1.9 ± 0.1 2360 ± 100 0.09 ± 0.01 PBT/AlPi-H+RXP20 7.1 400 ± 40 41 ± 1 20 ± 1 1.3 ± 0.1 3333 ± 100 0.16 ± 0.01

The errors were based on the maximum deviation of measured values from the average value. S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 201 was equally strong (38%) in both materials. The protec- The decrease in PBT’s two main fire risks by AlPi-Et tive effect of the char was strong (38%) for the intumes- and AlPi-H+RXP, respectively, is assessed in Figure 5E cent residue of PBT/AlPi-Et20 and even stronger (57%) for and F. If flame spread was assessed by pHRR/tign, PBT/ the compact char layer of PBT/AlPi-H+RXP20. The three AlPi-H+RXP4.2 showed a decrease in fire load and flame different flame-retardancy mechanisms influenced the spread. PBT/AlPi-Et6.3 exhibited a stronger reduction in different fire properties. The protective effect of the char fire load than did PBT/AlPi-H+RXP4.2, but no reduction decreased pHRR, whereas flame inhibition and charring in flame spread was observed. Increasing the additives’ decreased pHRR and THE. content to 20 wt.% led to a strong decrease in fire load and flame spread for both flame retardants. The reduction of both fire risks was slightly stronger for PBT/AlPi- 3.9 Fire behaviour of the bulk materials: H+RXP20 than for PBT/AlPi-Et20 due to the stronger assessment of efficiency condensed-phase activity (charring and protective effect of the char) of PBT/AlPi-H+RXP20. With the THE The dependence of burning parameters on the phospho- vs. FIGRA plot, AlPi-Et and AlPi-H+RXP were assessed rus content of the materials was analysed to compare as ideal flame retardants because they decreased both the efficiency of AlPi-Et and AlPi-H+RXP (Figure 5A–D). fire risks with rising amounts of flame retardant added. PBT/AlPi-Et showed a decrease in pHRR that was linearly At a phosphorus content of 1.5 wt.%, AlPi-Et decreased dependent on phosphorus content (Figure 5A). The pHRR fire load more strongly than AlPi-H+RXP, whereas AlPi- of PBT/AlPi-H+RXP decreased non-linearly with rising H+RXP reduced fire growth rate slightly more than phosphorus content and a levelling-off was indicated for AlPi-Et did. At 20 wt.% loading of the flame retardants, high phosphorus concentrations. For concentrations of fire load and fire growth rate were reduced further. up to 5 wt.% phosphorus, AlPi-H+RXP was slightly more AlPi-Et decreased the fire growth rate more than AlPi- efficient at decreasing pHRR than was AlPi-Et. The reduc- H+RXP did. AlPi-H+RXP performed better at reducing tion in THE showed a linear dependence on phosphorus fire load than AlPi-Et because of the stronger condensed- content for the materials containing AlPi-Et, whereas the phase activity (charring and protective effect of the char) dependence was non-linear for the materials with AlPi- of PBT/AlPi-H+RXP20. H+RXP. AlPi-Et was more efficient than AlPi-H+RXP in reducing THE for phosphorus concentrations up to about 6 wt.%. THE/TML decreased (i.e., gas-phase activity 3.10 Fire behaviour of the composites: increased) non-linearly with rising phosphorus content flammability for PBT/AlPi-Et. A linear dependence of the reduction in THE/TML on phosphorus content was found for the mate- The addition of glass fibres increased the flammability of rials containing AlPi-H+RXP. AlPi-Et showed a higher PBT, as demonstrated by the lowered LOI (19.8% instead efficiency in decreasing THE/TML than AlPi-H+RXP, i.e., of 23.3% for pure PBT). Glass fibres suppress melt flow the gas-phase activity of AlPi-Et per unit of phosphorus and dripping (29). This effect was also reported for carbon was stronger than that of AlPi-H+RXP. The increase in the fibres and multi-walled carbon nanotubes (30, 31). Hence, amount of residue was linearly dependent on phosphorus heat is no longer taken away from the pyrolysis zone by content for both flame retardants. AlPi-H+RXP was more melt flow. In addition, a “wick effect” of the glass fibres efficient in promoting charring than AlPi-Et, i.e., the con- and improved heat transmission are considered as reasons densed-phase activity of AlPi-H+RXP per unit of phospho- for the increased flammability of PBT/GF. rus was stronger than that of AlPi-Et. Adding AlPi-Et to PBT/GF increased LOI, but only Fire load and flame spread are two of the main fire until a phosphorus concentration of 3.3 wt.% was reached risks. Fire load was directly measured as THE, but there is (Table 6). The use of phosphorus concentrations higher no way to measure flame spread directly in the cone calo- than 3.3 wt.% led to a decrease in LOI. The dependence rimeter test. That is why indices like pHRR/tign and FIGRA of LOI on phosphorus content showed an optimum at [fire growth rate = maximum ratio HRR(t)/t] were proposed about 3 wt.% phosphorus (Figure 6). A levelling-off or to assess flame spread or fire growth, respectively (27). It even a decrease in flame-retardancy performance for high was additionally proposed to assess both the main fire phosphorus concentrations has been reported before and risks at the same time by plotting THE over pHRR/tign ascribed to the inherent flammability of phosphorus (32). (28). Then, ideal flame retardancy corresponds to a shift The addition of AlPi-H+RXP to PBT/GF increased LOI, towards the origin of this plot. but this increase was less strong than with AlPi-Et. The 202 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants

AB2200 80 2000 75 PBT 1800 PBT 70 PBT/AIPi-H+RXP4.2 1600 PBT/AIPi-Et6.3

) 65 2 )

1400 2 1200 60 1000 55 PBT/AIPi-Et6.3 PBT/AIPi-H+RXP4.2

800 PBT/AIPi-Et20 THE (MJ/m pHRR (kW/m 50 600 45 400 PBT/AIPi-Et20 PBT/AIPi-H+RXP20 40 200 PBT/AIPi-H+RXP20 0 35 012345678 012345678 P-content (wt.%) P-content (wt.%) C 2.2 D 22 PBT 20 PBT/AIPi-H+RXP4.2 2.0 18 PBT/AIPi-H+RXP20 16 g)

2 1.8 14 12 1.6 PBT/AIPi-Et6.3 10 PBT/AIPi-H+RXP4.2 PBT/AIPi-Et20 PBT/AIPi-H+RXP20 8 1.4 Residue (wt.%) THE/TML (MJ/m 6 PBT 1.2 4 PBT/AIPi-Et20 PBT/AIPi-Et6.3 2 1.0 0 0 12345678 012345678 P-content (wt.%) P-content (wt.%) E 80 F 80 75 PBT 75 PBT 70 PBT/AIPi-H+RXP4.2 (1.5% P) 70 PBT/AIPi-H+RXP4.2 (1.5% P) 65 65 60 60 PBT/AIPi-Et6.3 (1.5% P) 55 55 PBT/AIPi-Et6.3 (1.5% P) ) ) 2 50 2 PBT/AIPi-Et20 (4.8% P) 50 PBT/AIPi-Et20 (4.8% P) 45 45 40 40 PBT/AIPi-H+RXP20 (7.1% P) PBT/AIPi-H+RXP20 (7.1% P) 35 35

THE (MJ/m 30

THE (MJ/m 30 25 25 20 20 15 15 10 10 5 5 0 0 0 246810 12 14 16 18 20 22 24 26 28 30 32 34 36 0246810 12 14 16 18 pHRR/t(ign) (kW/m2 s) FIGRA (kW/m2 s)

Figure 5 Dependence of (A) pHRR, (B) THE, (C) THE/TML, (D) residue on the phosphorus content of the bulk materials. Fire risk in terms of

(E) fire load (THE) vs. pHRR/tign and (F) THE vs. FIGRA of the bulk materials.

dependence of LOI on phosphorus content was linear for 3.11 Fire behaviour of the composites: AlPi-H+RXP. No levelling-off or decrease for high concen- forced-flaming combustion trations was indicated. The UL 94 rating of PBT/GF was improved to V-0 by adding 14 wt.% AlPi-Et, 20 wt.% AlPi- The HRR of PBT/GF showed an initial increase in HRR, Et, or 20 wt.% AlPi-H+RXP, respectively. The addition of forming a pHRR quite at the beginning of burning

AlPi-Et to PBT/GF did not affect the tign in the cone calo- (Figure 7). Then, as the level of the burning melt decreased rimeter test, whereas tign decreased with rising amounts of below the upper layer of glass fibres, this upper layer with AlPi-H+RXP added. some char on it (as it appeared black) formed a protective S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 203

Table 6 Flammability (reaction to a small flame) and ignitability of decreased and then remained constant until the end, the composite materials determined by LOI (error = ± 1.0%), UL 94, when it exhibited a small second peak. The HRR of PBT/ and time to ignition (t ) in the cone calorimeter with an irradiation ign GF/AlPi-H+RXP20 showed only a small peak at the begin- of 50 kW/m2. ning. Then, the HRR remained constant due to the protective effect of the very compact, thick char layer (Figure 7). Material P content (wt.%) LOI (%) UL 94 tign (s) Similar heat release characteristics were reported for glass PBT/GF 0 19.8 HB 40 ± 4 fibre-reinforced polyamide flame retarded with red phos- PBT/GF/AlPi-Et4.4 1.0 33.9 HB 43 ± 6 PBT/GF/AlPi-Et6.3 1.5 39.0 HB 43 ± 1 phorus (33). PBT/GF/AlPi-Et14 3.3 44.3 V-0 40 ± 5 The first pHRR of PBT/GF was decreased by both PBT/GF/AlPi-Et20 4.8 36.7 V-0 44 ± 1 flame retardants (Table 7). At a phosphorus content of PBT/GF/AlPi-H+RXP4.2 1.5 21.7 HB 33 ± 1 1.5 wt.%, the reduction in the first pHRR was similar for PBT/GF/AlPi-H+RXP20 7.1 31.9 V-0 22 ± 3 AlPi-Et and AlPi-H+RXP in PBT/GF. With a flame retard-

The error of tign was based on the maximum deviation of measured ant content of 20 wt.% in PBT/GF, AlPi-H+RXP reduced values from the average value. Some specimens deformed upon the first pHRR more strongly than AlPi-Et did. The second exposure to the cone heater, thus causing a high deviation in t . ign pHRR was decreased only slightly by AlPi-Et, whereas a clear reduction was observed for AlPi-H+RXP. PBT/GF/ AlPi-Et showed a decrease in THE with rising AlPi-Et layer at the surface. When more and more material was content of up to 14 wt.%. Then THE increased for PBT/GF/ consumed, the layer thickened and HRR decreased. The AlPi-Et20. AlPi-H+RXP reduced the THE of PBT/GF, with shape of HRR changed upon addition of AlPi-Et. The increasing amounts of AlPi-H+RXP added. The residue of pHRR at the beginning decreased. The formed glass fibre/ PBT/GF (30 wt.%) was due mainly to glass fibres. AlPi-Et char layer was more compact, with rising amounts of increased the residue slightly, with rising amounts of AlPi-Et added, because there was more char “gluing” the flame retardant added. AlPi-H+RXP increased the residue glass fibres together (Figure 7). PBT/GF/AlPi-Et materials strongly, with increasing amounts of flame retardant showed a second small pHRR at the end due to thermal added. Hence, AlPi-H+RXP had stronger condensed- feedback from the insulation on the sample holder. As no phase activity in PBT/GF than did AlPi-Et. The effective cracking of the char was observed, this was ruled out as a heat of combustion (THE/TML) decreased with rising reason for the second pHRR (27). The addition of 4.2 wt.% AlPi-Et content until a content of 14 wt.% was reached. AlPi-H+RXP decreased the HRR compared to PBT/GF The decrease up to this point corresponded to an increase (Figure 7). The HRR curve of PBT/GF/AlPi-H+RXP4.2 was in gas-phase activity or flame inhibition, respectively. characterised by a peak of HRR at the beginning. As soon Flame inhibition decreased for AlPi-Et contents higher as an efficient glass fibre/char layer was formed, HRR than 14 wt.%, i.e., for PBT/GF/AlPi-Et20. The addition of 4.2 wt.% AlPi-H+RXP to PBT/GF decreased THE/TML compared to PBT/GF. Thus AlPi-H+RXP showed flame

46 PBT/GF/AIPi-Et14 inhibition as well. Increasing the content of AlPi-H+RXP 44 to 20 wt.% decreased flame inhibition. This suggests that 42 20 wt.% AlPi-H+RXP was already beyond the optimum 40 PBT/GF/AIPi-Et6.3 concentration for PBT/GF. 38 36 PBT/GF/AIPi-Et20 34 PBT/GF/AIPi-Et4.4 32 3.12 Fire behaviour of the composites: 30 LOI (%) 28 PBT/GF/AIPi-H+RXP20 quantitative assessment of flame- 26 retardancy mechanisms 24 22 PBT/GF/AIPi-H+RXP4.2 The increased residue of PBT/GF/AlPi-Et20 (40 wt.% 20 instead of 30 wt.%) corresponds to a reduction in fuel 18 PBT/GF 16 release to 86%. THE/TML was reduced to 79% due to 012345678flame inhibition. The combination of reduced fuel and P-content (%) flame inhibition led to a reduction to 68%, which explains Figure 6 Dependence of LOI on the phosphorus content of the the reduction in THE (68%) well. The pHRR of PBT/GF/ composite materials. AlPi-Et20 was reduced to 46%. In the first approximation, 204 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants

700 700 PBT/GF PBT/GF PBT/GF/AIPi-Et4.4 PBT/GF/AIPi-H+RXP4.2 600 600 PBT/GF/AIPi-Et6.3 PBT/GF/AIPi-H+RXP20 PBT/GF/AIPi-Et14 500 500 PBT/GF/AIPi-Et20 ) ) 2 2 400 400

300 300 HRR (kW/m HRR (kW/m 200 200

100 100

0 0 050 100 150 200 250 300 350 050 100 150 200 250 300 350 Time (s) Time (s)

Figure 7 Heat release rate of the composite materials during the cone calorimeter test with an irradiation of 50 kW/m2, and fire residues from PBT/GF (bottom left), PBT/GF/AlPi-Et20 (bottom centre), and PBT/AlPi-H+RXP20 (bottom right) obtained at 50 kW/m2.

Table 7 Cone calorimeter results of the composite materials obtained using an irradiation of 50 kW/m2 including pHRR, THE, TML, and TSR.

Material P content First pHRR (kW/m2) Second pHRR (kW/ THE Residue THE/TML 2 2 2 (wt.%) (tpHRR in s) m ) (tpHRR in s) (MJ/m ) (wt.%) (MJ/m g) PBT/GF 0 677 ± 40 (70 ± 2) 457 ± 40 (133 ± 5) 59 ± 1 30 ± 1 1.9 ± 0.1 PBT/GF/AlPi-Et4.4 1.0 438 ± 40 (72 ± 3) 442 ± 40 (106 ± 11) 41 ± 1 33 ± 1 1.4 ± 0.1 PBT/GF/AlPi-Et6.3 1.5 367 ± 30 (71 ± 2) 362 ± 30 (114 ± 12) 39 ± 1 31 ± 1 1.3 ± 0.1 PBT/GF/AlPi-Et14 3.3 351 ± 30 (69 ± 3) 402 ± 30 (121 ± 4) 37 ± 1 37 ± 1 1.3 ± 0.1 PBT/GF/AlPi-Et20 4.8 310 ± 30 (73 ± 7) 356 ± 40 (113 ± 10) 40 ± 1 40 ± 1 1.5 ± 0.1 PBT/GF/AlPi-H+RXP4.2 1.5 378 ± 30 (45 ± 1) 253 ± 20 (141 ± 9) 44 ± 1 34 ± 1 1.5 ± 0.1 PBT/GF/AlPi-H+RXP20 7.1 165 ± 15 (30 ± 3) 169 ± 15 (95 ± 10) 35 ± 1 55 ± 1 1.7 ± 0.1

The errors were based on the maximum deviation of measured values from the average value.

a reduction to 68% was due to reduced fuel release and in pHRR to 57% was due to reduced fuel and flame inhi- flame inhibition. The further reduction to 46% corre- bition. The additional reduction to 24%, i.e., a relative sponded to a relative reduction of 32% and was due to the reduction of 58%, was due to the protective effect of the protective effect of the char. char. Analogous considerations were made for PBT/GF/ The three different flame-retardancy mechanisms AlPi-H+RXP20. The increased residue (55 wt.%) cor- that were active in the compact materials were also responded to a reduction in fuel release to 64%. THE/ active in the glass-fibre composites PBT/GF/AlPi-Et20 TML was decreased to 89% due to flame inhibition. and PBT/GF/AlPi-H+RXP20. Condensed-phase activity Combining reduced fuel and flame inhibition yielded a (charring) was strong in PBT/GF/AlPi-H+RXP20 (36% reduction to 57%, which explains the reduction in THE reduced fuel), whereas it played a minor role in PBT/GF/ (59%) as well. The pHRR of PBT/GF/AlPi-H+RXP20 was AlPi-Et20 (14% reduced fuel). Gas-phase activity (flame reduced to 24%. In the first approximation, a reduction inhibition) was stronger (21%) in PBT/GF/AlPi-Et20 than S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 205 in PBT/GF/AlPi-H+RXP20 (11%). The protective effect 3.13 Fire behaviour of the composites: of the char was strong (32%) for the residue of PBT/GF/ assessment of efficiency AlPi-Et20 and even stronger (58%) for the char of PBT/ GF/AlPi-H+RXP20. The three different flame-retardancy The decrease in pHRR was non-linearly dependent on mechanisms influenced the different fire properties phosphorus content for both flame retardants (Figure 8A). of the glass-fibre composites, as was also observed for The reduction in pHRR of PBT/GF/AlPi-Et levelled off at a the compact materials. The protective effect of the char lower phosphorus content than did the decrease in pHRR of decreased pHRR, whereas flame inhibition and charring PBT/GF/AlPi-H+RXP. AlPi-H+RXP was more efficient than decreased pHRR and THE. AlPi-Et in decreasing the pHRR of PBT/GF for phosphorus

800 65 AB750 700 PBT/GF 60 650 PBT/GF 600 55 550 ) PBT/GF/AIPi-Et4.4 ) 2 500 2 50 450 PBT/GF/AIPi-H+RXP4.2 400 PBT/GF/AIPi-Et14 PBT/GF/AIPi-H+RXP4.2 350 45 PBT/GF/AIPi-Et6.3 PBT/GF/AIPi-Et20 300 THE (MJ/m

pHRR (kW/m PBT/GF/AIPi-Et20 250 40 200 150 PBT/GF/AIPi-H+RXP20 PBT/GF/AIPi-Et6.3 100 35 PBT/GF/AIPi-Et4.4 PBT/GF/AIPi-Et14 50 PBT/GF/AIPi-H+RXP20 0 30 0 12345678 012345678 P-content (wt.%) P-content (wt.%) CD2.0 60 58 1.9 PBT/GF 56 PBT/GF/AIPi-H+RXP20 54 1.8 52

g) PBT/GF/AIPi-Et4.4 50 2 1.7 PBT/GF/AIPi-H+RXP20 48 1.6 46 44 1.5 PBT/GF/AIPi-H+RXP4.2 PBT/GF/AIPi-Et20 42 40 PBT/GF/AIPi-Et20 1.4 Residue (wt.%) 38 PBT/GF/AIPi-Et4.4 THE/TML (MJ/m 36 1.3 PBT/GF/AIPi-Et14 34 1.2 32 PBT/GF/AIPi-H+RXP4.2 PBT/GF/AIPi-Et6.3 PBT/GF/AIPi-Et14 30 1.1 28 PBT/GF PBT/GF/AIPi-Et6.3 0 12345678 012345678 P-content (wt.%) P-content (wt.%) E 65 F 65 60 60 PBT/GF PBT/GF 55 55 50 PBT/GF/AIPi-Et6.3 (1.5% P) 50 PBT/GF/AIPi-Et6.3 (1.5% P) 45 PBT/GF/AIPi-H+RXP4.2 (1.5% P) 45 PBT/GF/AIPi-H+RXP4.2 (1.5% P) ) PBT/GF/AIPi-Et20 (4.8% P) ) 2 40 2 40 PBT/GF/AIPi-Et20 (4.8% P) PBT/GF/AIPi-Et4.4 (1.0% P) PBT/GF/AIPi-Et4.4 (1.0% P) 35 35 PBT/GF/AIPi-H+RXP20 (7.1% P) 30 30 PBT/GF/AIPi-Et14 (3.3% P) PBT/GF/AIPi-Et14 (3.3% P) 25 25 THE (MJ/m PBT/GF/AIPi-H+RXP20 (7.1% P) THE (MJ/m 20 20 15 15 10 10 5 5 0 0 0 246810 12 14 16 18 0 12345678910 11 pHRR/t(ign) (kW/m2 s) FIGRA (kW/m2 s)

Figure 8 Dependence of the (A) first pHRR, (B) THE, (C) THE/TML, and (D) residue on the phosphorus content of the investigated composite materials. Fire risk in terms of (E) fire load (THE) vs. first pHRR/tign and (F) THE vs. FIGRA of the composite materials. 206 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants concentrations higher than 2 wt.%, and the maximum 4 Conclusion possible decrease in pHRR was higher for AlPi-H+RXP than for AlPi-Et. THE decreased non-linearly with rising The pyrolysis and fire behaviour of organic aluminium phosphorus content for both flame retardants (Figure 8B). diethylphosphinate and inorganic aluminium phosphi- It showed an optimum at about 3 wt.% phosphorus for nate was investigated in PBT and PBT/GF. A detailed com- PBT/GF/AlPi-Et, whereas levelling-off was observed for parison of the two phosphinates was reported. The results high phosphorus concentration in PBT/GF/AlPi-H+RXP. of the pyrolysis investigation support the decomposition Up to a phosphorus content of 3 wt.%, AlPi-Et was more mechanisms proposed for PBT and PBT/AlPi-Et. The efficient than AlPi-H+RXP in reducing the THE of PBT/ AlPi-H in PBT/AlPi-H+RXP decomposed to aluminium GF. The THE/TML of PBT/GF/AlPi-Et showed a non-linear phosphate, releasing phosphine at temperatures as low as dependence on phosphorus content with an optimum at 320°C. This is quite close to the processing temperatures of 3 wt.% phosphorus (Figure 8C), as was already observed PBT. Therefore, the release of toxic phosphine and a cor- for THE and LOI. The dependence of the THE/TML of PBT/ responding loss of phosphorus are problems that may be GF/AlPi-H+RXP on phosphorus content was non-linear. encountered during the processing of AlPi-H+RXP. AlPi-Et An optimum was indicated, but two different phosphorus does not suffer from such problems during processing. concentrations were not enough to determine its occur- A good performance in the flammability tests LOI rence with any certainty. The amount of residue increased and UL 94 is crucial for the application of flame-retarded linearly with rising phosphorus content for both flame PBT. At 20 wt.% loading, AlPi-H+RXP achieved only a V-2 retardants. AlPi-H+RXP was more efficient than AlPi-Et in rating in bulk PBT, which is not sufficient for most appli- increasing the residue of PBT/GF (Figure 8D). cations. The necessary V-0 rating was achieved with 20 The first pHRR was used to assess flame spread and wt.% AlPi-H+RXP in glass fibre-reinforced PBT. In con- fire growth rate, respectively. As the first pHRR was higher trast to AlPi-H+RXP, AlPi-Et provided a V-0 rating in bulk than the second pHRR and occurred earlier, the first pHRR PBT at 20 wt.% loading. In glass fibre-reinforced PBT, only determined flame spread (pHRR/tign) and fire growth rate 14 wt.% of AlPi-Et was necessary to achieve a V-0 rating. [maximum ratio HRR(t)/t]. AlPi-H+RXP led to only a moderate increase in the LOI of

In the assessment of flame spread by pHRR/tign, bulk PBT (up to 26.8%) and reinforced PBT (up to 31.9%). AlPi-Et decreased fire load and flame spread (Figure 8F). AlPi-Et provided very high LOI values for bulk PBT (up to PBT/GF with 6.3 wt.% and 14 wt.% AlPi-Et, respectively, 52.0%) as well as for glass fibre-reinforced PBT (44.3%). showed a similar performance. Increasing the amount of Thus, AlPi-Et performed much better than AlPi-H+RXP in AlPi-Et to 20 wt.% increased fire load slightly but further flammability tests. reduced flame spread. AlPi-H+RXP decreased the fire load The cone calorimeter results revealed that both flame and flame spread of PBT/GF as rising amounts of additive retardants acted via three different flame-retardancy were added. At a phosphorus content of 1.5 wt.%, AlPi-Et mechanisms in bulk as well as glass fibre-reinforced-PBT. performed much better than AlPi-H+RXP in reducing fire The three mechanisms are flame inhibition, increased load and flame spread. At a flame-retardant content of 20 char, and a protective effect of the char. With the bulk wt.%, AlPi-H+RXP performed slightly better than AlPi- materials, both flame retardants provided equally strong Et. As shown in the THE over FIGRA plot (Figure 8E), the flame inhibition (38%). The increase in charring (17% addition of 4.4 wt.% AlPi-Et decreased the fire growth reduction in fuel release for PBT/AlPi-H+RXP20 instead rate and fire load of PBT/GF. A further increase in the of 7% reduction in fuel release for PBT/AlPi-Et20) and the AlPi-Et content mainly reduced fire growth rate without protective effect of the char (57% instead of 38%) were having much influence on fire load. At an AlPi-Et content stronger for AlPi-H+RXP than for AlPi-Et. With the glass- of 20 wt.%, fire load was even slightly increased. AlPi- fibre composites, AlPi-Et showed stronger flame inhibi- H+RXP decreased fire growth rate and fire load with rising tion than AlPi-H+RXP (21% instead of 11%). AlPi-H+RXP amounts of additive added. At a phosphorus content of provided a stronger increase in charring (36% reduced 1.5 wt.%, AlPi-Et performed much better than AlPi-H+RXP fuel instead of 14%) and a stronger protective effect of the in reducing fire load and fire growth rate. At a flame- char (58% instead of 32%) than AlPi-Et. retardant content of 20 wt.%, the overall performance Based on the cone calorimeter results, the efficiency of the two flame retardants was similar. PBT/GF/AlPi- of the flame retardants was assessed, i.e., the decrease Et20 showed a lower fire growth rate than PBT/GF/AlPi- in pHRR, THE, and THE/TML, and the increase in the H+RXP20, whereas the latter material showed a lower fire amount of char per unit of phosphorus. AlPi-Et decreased load than PBT/GF/AlPi-Et20. the THE and THE/TML of the bulk and the composite S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants 207 materials more efficiently than AlPi-H+RXP. Concerning A synergist can be added to further increase flame-retar- the reduction in pHRR and the increase in charring of the dancy performance if necessary. bulk and composite materials, AlPi-H+RXP was more efficient than AlPi-Et. Most interestingly, an optimum loading Acknowledgements: The authors thank the German of AlPi-Et was found for glass fibre-reinforced PBT. The Research Foundation (DFG, SCHA 730/11-1, AL 474/17-1) highest LOI value and the strongest decrease in THE and for its financial support. H. Bahr and P. Klack (BAM) are THE/TML was achieved with a loading of 12 wt.% AlPi-Et. acknowledged for supporting the fire tests and the evolved The performance was decreased if more or less AlPi-Et was gas analysis, respectively. used. Thus it is recommended to use 12 wt.% of AlPi-Et for efficient flame retardancy in glass fibre-reinforced PBT. Received February 17, 2014; accepted March 26, 2014

References

1. Gallucci RR, Patel BR. Poly(butylene terephthalate). In: Scheirs J, organo-modified sepiolite and Zn phosphinate. Fibers Polym. Long T, editors. Modern olyesters. Chichester, UK: Wiley; 2003. 2011;12(2):166–73; doi:10.1007/s12221-011-0166-5. pp 293–321. 17. Yang W, Tang G, Song L, Hu Y, Yuen RKK. Effect of rare earth 2. Levchik SV, Weil ED. Flame retardancy of thermoplastic polyes- hypophosphite and melamine cyanurate on fire performance of ters – a review of the recent literature. Polym Int. 2005;54(1): glass-fiber reinforced poly(1,4-butylene terephthalate) compos- 11–35. ites. Thermochim Acta 2011;526:185–91. 3. RoHS-Directive 2002/95/EC. EU Parliament and Council; 2003. 18. Yang W, Song L, Hu Y, Lu H, Yuen RKK. Investigations of thermal 4. Racky W, Kleiner H-J, Herwig W. US Patent 3900444; 1975. degradation behavior and fire performance of halogen-free 5. Kleiner H-J, Budzinsky W, Kirsch G. EP 0699708; 1996. flame retardant poly(1,4-butylene terephthalate) composites. J 6. Kleiner H-J, Budzinsky W, Kirsch G. EP 0794220; 1997. Appl Polym Sci. 2011;122(3):1480–8; doi:10.1002/app.34119. 7. Jenewein E, Kleiner H-J, Wanzke W, Budzinsky W. WO 97/39053; 19. Levchik SV, Weil ED. A review on thermal decomposition and 1997. combustion of thermoplastic polyesters. Polym Adv Technol. 8. Klatt M, Leutner B, Nam M, Fisch H. WO 99/57187; 1999. 2004;15(12):691–700; doi:10.1002/Pat.526. 9. Constanzi S, Leonardi M. WO 2005/121232; 2005. 20. Corbridge DEC. Phosphorus 2000: chemistry, biochemistry & 10. Braun U, Bahr H, Sturm H, Schartel B. Flame retardancy technology. Amsterdam: Elsevier; 2000. mechanisms of metal phosphinates and metal phosphinates in 21. Holleman AF, Wiberg E, Wiberg N. Lehrbuch der Anorganischen combination with melamine cyanurate in glass-fiber reinforced Chemie. Berlin: Walter de Gruyter; 1985. poly(1,4-butylene terephthalate): the influence of metal cation. 22. Balabanovich AI, Zevaco TA, Schnabel W. Macromol Mater Eng. Polym Adv Technol. 2008;19(6):680–92; doi:10.1002/pat.1147. 2004;289(2):181–90. 11. Brehme S, Schartel B, Goebbels J, Fischer O, Pospiech D, 23. Holland BJ, Hay JN. The thermal degradation of PET and Bykov Y, Döring M. Phosphorus polyester versus aluminium analogous polyesters measured by thermal analysis-Fourier phosphinate in poly(butylene terephthalate) (PBT): flame transform infrared spectroscopy. Polymer 2002;43(6):1835–47. retardancy performance and mechanisms. Polym Degrad Stab. 24. Rokita M, Handke M, Mozgawa W. Spectroscopic studies of

2011;96(5):875–84. polymorphs of AlPO4 and SiO2. J Mol Struct. 1998;450:213–7. 12. Vannier A, Duquesne S, Bourbigot S, Castrovinci A, Camino G, 25. Braun U, Schartel B. Flame retardancy mechanisms of aluminium Delobel R. The use of POSS as synergist in intumescent phosphinate in combination with melamine cyanurate in glass- recycled poly(ethylene terephthalate). Polym Degrad Stab. fibre-reinforced poly(1,4-butylene terephthalate). Macromol 2008;93(4):818–26. Mater Eng. 2008;293(3):206–17; doi:10.1002/mame.200700330. 13. Gallo E, Braun U, Schartel B, Russo P, Acierno D. Halogen-free 26. Köppl T, Brehme S, Wolff-Fabris F, Altstädt V, Schartel B, Döring M. flame retarded poly(butylene terephthalate) (PBT) using metal Structure-property relationships of halogen-free flame-retarded oxides/PBT nanocomposites in combination with aluminium poly(butylene terephthalate) and glass fiber reinforced PBT. J phosphinate. Polym Degrad Stab. 2009;94(8):1245–53; Appl Polym Sci. 2012;124(1):9–18; doi:10.1002/app.34910. doi:10.1016/j.polymdegradstab.2009.04.014. 27. Schartel B, Hull TR. Development of fire-retarded materials 14. Gallo E, Schartel B, Braun U, Russo P, Acierno D. Fire retardant – Interpretation of cone calorimeter data. Fire Mater.

synergisms between nanometric Fe2O3 and aluminum phosphi- 2007;31(5):327–54; doi:10.1002/Fam.949. nate in poly(butylene terephthalate). Polym Adv Technol. 28. Petrella RV. The assessment of full-scale fire hazards from cone 2011;22(12):2382–91; doi:10.1002/pat.1774. calorimeter data. J Fire Sci. 1994;12(1):14–43. 15. Bourbigot S, Samyn F, Turf T, Duquesne S. Nanomorphology 29. Casu A, Camino G, Giorgi MD, Flath D, Laudi A, Morone V. Effect and reaction to fire of polyurethane and polyamide nano- of glass fibres and fire retardant on the combustion behaviour composites containing flame retardants. Polym Degrad Stab. of composites, glass fibres-poly(butylene terephthalate). Fire 2010;95(3):320–6. Mater. 1998;22(1):7–14. 16. Alongi J. Investigation on flame retardancy of poly(ethylene 30. Perret B, Schartel B, Stöß K, Ciesielski M, Diederichs J, terephthalate) for plastics and textiles by combination of an Döring M, Krämer J, Altstädt V. A new halogen-free flame retard- 208 S. Brehme et al.: Competition between aluminium phosphinate-based halogen-free flame retardants

ant based on 9,10-dihydro-9-oxa-10-phosphaphenanthrene- 32. Levchik GF, Levchik SV, Camino G, Weil ED. Fire retardant action 10-oxide for epoxy resins and their carbon fiber composites for of red phosphorus in nylon 6. In: Le Bras M, Camino G, Bour- the automotive and aviation industries. Macromol Mater Eng. bigot S, Delobel R, editors. Fire retardancy of polymers. The use 2011;296(1):14–30; doi:10.1002/mame.201000242. of intumescence. London, UK: The Royal Society of Chemistry; 31. Schartel B, Pötschke P, Knoll U, Abdel-Goad M. Fire behaviour 1998. pp. 304–15. of polyamide 6/multiwall carbon nanotube nanocompos- 33. Schartel B, Kunze R, Neubert D. Red phosphorus–controlled ites. Eur Polym J. 2005;41(5):1061–70; doi:10.1016/j.eurpol- decomposition for fire retardant PA 66. J Appl Polym Sci. ymj.2004.11.023. 2002;83(10):2060–71; doi:10.1002/app.10144.