Introduction

Among the fabrics employed in the market, cotton has always been an object of great interest. Its economic significance in the global market is shown by its majority share (over 50 %) among fibres (micro-denier such as polyesters and polyamides, elastomeric and lyocell fibres) for apparel and textile stuffs (Gordon and Hsie 2007; Wakelyn et al. 2006). One of the most important issues to overcome for using it in numerous applications is its high flammability (Horrocks 1983). Many studies have been carried out by academia and industries in this century, as recently reviewed (Alongi and Malucelli 2015; Lowden and Hull 2013), in particular for replacing the main market targets represented by Proban® and Pyrovatex® (Horrocks 2011; Alongi et al. 2013b). Among the most innovative solutions, the use of green flame retardants from bio-sources has been also explored (Basak et al. 2014, 2015a, b, c; Malucelli et al. 2014), opening a “new era” for seeking environmentally friendly flame retardants.

Overall, the surface deposition of flame retardants or of a specific coating is the most common methods exploited for reaching the above goal at both the scientific and industrial scale (Alongi et al. 2014a). However, according to a close examination of the scientific literature, generally, no one seems to be interested in the role of cotton as a substrate with its own physical properties, concerning grammage and texture, as examples. Indeed, usually, flame-retardant properties are measured as a relative comparison of untreated and treated fabrics, without taking into account that the flame retardants employed in the study may be efficient for thin and not thicker fabrics. However, this aspect can be of fundamental importance, in particular in the case of fabrics or films. On the other hand, also the fabric texture can play a key role, in particular when a flame retardant is designed for blocking cotton combustion when exposed to a flame, regardless of the gas type, flame length or application time. Thus, the spontaneous question that comes to mind is: how much does the fabric grammage affect cotton combustion? In addition, how can cotton combustion be investigated? What instrumentation can be used and what do the collected results mean? The present article is meant to answer all these questions, highlighting the role of fabric grammage on the resulting cotton combustion behaviour by employing different types of instrumentations, thus providing basic and solid knowledge useful for future studies. More specifically, the flammability of 100, 200 and 400 g/m2 fabrics has been explored by exposing cotton to: (1) an irradiative heat flux (25 or 35 kW/m2) during cone calorimetry tests; (2) a methane flame during horizontal or vertical flame spread tests; (3) a propane flame in specific conditions of oxygen concentration during LOI tests; and (4) pyrolysis and further oxidisation during PCFC tests. Collected results demonstrated a precise and direct relationship between fabric grammage and cotton combustion behaviour.

Materials and methods

Materials

Three scoured cotton fabrics having 100, 200 and 400 g/m2 grammage (hereafter coded as COT100, COT200 and COT400) were purchased from Fratelli Ballesio s.r.l. (Torino, Italy), washed in Marsille soap, ethanol and diethyl ether, and subsequently dried before testing them. The weave design characteristics of the employed fabrics are described in Table 1, which reports the grammage of each fabric, basic weaves and number of threads per unit length assessed by the ISO 3572 (1976) and UNI EN 1049 (1996) standards, respectively. In addition, some pictures of the three fabrics are reported in Fig. 1 (a, b and c for COT100, COT200 and COT400, respectively).

Table 1 Weave design characteristics of the fabrics employed in the present study
Full size table
Fig. 1
figure 1

Pictures of COT100, COT200 and COT400 (a, b and c, respectively)

Full size image

Characterisation techniques

The surface morphology of the treated samples was studied using a LEO-1450VP scanning electron microscopy (SEM; beam voltage: 5 kV). Pieces of fabric surfaces (5 × 5 mm2) and cross sections were cut and pinned up with conductive adhesive tape and gold metallised.

Thermogravimetric analyses (from 50 to 800 °C with a heating rate of 10 °C/min) in both nitrogen and air (60 ml/min for both atmospheres) were carried out for investigating the thermal stability of the different fabrics. To this aim, a TAQ500 thermogravimetric balance was used, placing the samples in open alumina pans (12 mg). The experimental error was 0.5 % for weight and 1 °C for temperature.

LOI tests were performed with a FIRE oxygen index apparatus according to the ASTM D2863 standard (2006). The experimental error was 0.5 %.

PCFC (Fire Testing Technology) was used according to the ASTM D7309 standard (2013). More specifically, the cotton specimen (12.0 ± 0.1 mg) was pyrolysed (60 °C/min) under nitrogen (80 ml/min). Subsequently, these latter products were mixed with a 20 ml/min stream of oxygen prior to entering the combustion furnace, where they were burnt at 750 °C for 10 s. The T initial (T at which combustion starts), peak heat release rate (PHRR) and corresponding temperature (T PHRR) and time (t PHRR) and total heat release (THR) were evaluated. These tests were repeated three times, and the experimental error on PHRR and THR was ±2 %. The instrumental error on T and t was 1 °C and 1 s, respectively.

The resistance to flame exposure was examined by performing horizontal and vertical flame spread tests on rectangular specimens (50 × 150 mm2): more specifically, a blue methane flame (20 mm length) was applied on the short side of the specimen for 3 s. These tests were repeated three times for each configuration. In the horizontal flame spread tests, the afterflame time (length of time the flame persists after the ignition source has been removed on the basis of ISO 13943 2007, expressed in s), burning rate (mm/s) and afterglow time (length of time that there is persistence of glowing combustion after both removal of the ignition source and cessation of any flaming combustion, expressed in s) were assessed. More specifically, the burning rate was calculated evaluating the time required to burn the specimen portion between two lines traced at 2.5 and 12.5 cm. The experimental error for these parameters was ±2 %.

The resistance to a heat flux of square fabric samples (100 × 100 mm2) was investigated using cone calorimetry (Fire Testing Technology). The measurements were carried out under 25- or 35-kW/m2 irradiative heat flux in horizontal configuration, following the procedure described elsewhere (Tata et al. 2011) and derived from ISO 5660 (2002), also suitable for cotton fibres (Ceylan et al. 2013). Parameters such as the time to ignition (TTI, s), flame-out time (FO, s), total heat release (THR, kW/m2) and peak of heat release rate (PHRR, kW/m2) were measured. The latter two parameters were also normalised on specimen initial mass and expressed in kg (namely, nTHR and nPHRR). Average [CO] and [CO2] yields (both expressed in kg/kg) were assessed as well. The experiments were repeated three times for each material investigated to ensure reproducible and significant data; the experimental error was assessed as standard deviation (σ). Prior to combustion tests, all the specimens were conditioned at 23 ± 1 °C for 48 h at 50 % relative humidity in a climatic chamber.

Results and discussion

Fabric texture

Fabric texture of COT100, COT200 and COT400 has been observed by SEM on specimen cross sections and surfaces (Figs. 2, 3, respectively). In detail, COT100 and COT200 have been proven to consist of fibre bundles with comparable sizes (namely, 180 ± 20 and 220 ± 20 μm, respectively), also visible comparing Fig. 2a with c and Fig. 2b with d, respectively. Increasing the grammage from 100 to 200 g/m2, the bundle number increases as expected. This effect becomes more pronounced in COT400 (Fig. 2e, f), where the disorder level and bundle size are higher (350 ± 50 μm) than those observed in COT100 and COT200. These findings confirm the technical data for the different fabrics reported in Table 1; indeed, the number of threads per unit length is the same in terms of warp, but not in terms of weft (namely, 31, 30 and 52 for COT100, COT200 and COT400, respectively). However, the fibre diameter is almost equal for the three fabrics: namely, 19 ± 5, 18 ± 4 and 18 ± 5 μm for COT100, COT200 and COT400 (Fig. 3a b, c d, e, f respectively).

Fig. 2
figure 2

SEM observations on cross sections of COT100 (a and b), COT200 (c and d) and COT400 (e and f)

Full size image
Fig. 3
figure 3

SEM observations on surfaces of COT100 (a and b), COT200 (c and d) and COT400 (e and f)

Full size image

Fabric combustion

As already mentioned in the Introduction, cotton flammability was extensively studied in this century and its combustion mechanism was understood (Horrocks 1996; Kandola et al. 1996; Kandola and Horrocks 1999; Mamleev et al. 2009; Bourbigot et al. 2002; Price et al. 1997). This phenomenon depends on the pre-ignition step during which cellulose contained within cotton degrades in air at high heating rates. This occurs through the competition between depolymerisation and dehydration (Horrocks 1996; Shafizadeh and Fu 1973; Shafizadeh and Sekiguchi 1983 and Shafizadeh and Sekiguchi 1984). Depolymerisation induces the formation of highly combustible volatile species (mainly levoglucosan, furan and furan derivatives), which are responsible for cotton ignition (Faroq et al. 1991). On the contrary, dehydration produces the formation of a thermally steady char (Morterra and Low 1983, 1984, 1985) having aromatic features (Sekiguchi et al. 1983). The predominance of one of these two processes has recently been proven to be a function of heating rates (Alongi et al. 2013a). Indeed, the higher the heating rate, the more favourable the depolymerisation and thus cotton flammability are. On the other hand, dehydration and thus char formation are favoured when cotton is heated at low heating rates.

In any common standard procedure employed for investigating the combustion of a material, the adopted heating rate is very high, usually hundreds °C/min as in cone calorimetry, flame spread or LOI tests. This does not occur in PCFC, where the heating rate is generally 60 °C/min (Lyon and Walters 2004; Lyon et al. 2009); in addition, and in contrast with the other aforementioned tests, in PCFC only the combustion of pyrolysis products is carried out (Yang and Hu 2011, Yang and Hu 2012; Yang et al. 2010. Hence, these considerations highlight the different scenarios described by the common characterisation techniques and PCFC, which have found remarkable interest in the last decade, in particular concerning the characterisation of coatings derived from nanotechnology such as layer-by-layer assembly (Alongi et al. 2014a). This does not mean that PCFC is a useless instrument, but it is important to point out that the information derived from it does not describe the same scenario as cone calorimetry or flame spread tests. Indeed, this equipment can be very useful when combined with thermogravimetry, as recently demonstrated (Alongi et al. 2014b), as it is able to measure the combustion heat of pyrolysis products, which is not directly possible by using thermogravimetry.

Under 25 kW/m2 heat flux, cotton with different fabric grammages behaves in different ways (Table 2). Indeed, COT100 did not give reproducible data: two out of five specimens did not ignite. The amount of material is probably too low (namely, 1.2 g) and thus the amount of volatile species released upon heating does not reach the flammability limit. Indeed, this does not occur with COT200 and COT400. As a consequence of different grammages, COT200 and COT400 exhibited different TTI, FO and combustion durations: the higher the fabric grammage, the lower the TTI and FO were. Furthermore, COT400 exhibited a higher THR with respect to COT200 (4.9 versus 1.8 MJ/m2) and an apparently higher PHRR (132.0 versus 69.3 kW/m2), as expected.

Table 2 Combustion data of COT100, COT200 and COT400 collected by cone calorimetry
Full size table

The next question arises: is it correct to compare two specimens having such different masses under cone calorimetry? First of all, we have to take into account that the cone calorimeter was not designed for fabrics but for thick materials; indeed the balance present in the standard configuration is not as sensitive for thin specimens (Schartel and Hull 2007). In spite of this, we have already demonstrated that it is possible to use this instrumentation for fabrics (Tata et al. 2011) and fibres (Ceylan et al. 2013), but it is necessary to proceed carefully. In this context, for meaningfully comparing COT200 and COT400, it is necessary to normalise parameters such as THR and PHRR on the basis of initial specimen masses rather than surface area exposed. As reported in Table 2, it is clear that the THR trend is not changed (the higher the fabric grammage, the higher the total heat release during combustion is) but this does not occur for PHRR. Indeed, taking the experimental error into consideration, the PHRRs of COT200 and COT400 are almost equal; this suggests that the heat release rate measured in the cone calorimeter is not a function of cotton grammage, but is a feature typical of the material under investigation.

Another important aspect that should not be neglected is the meaning of THR normalised on mass and expressed in kg; indeed, the cone calorimeter can directly measure the heat release normalised on the specimen mass (this parameter is the effective heat of combustion, EHC), but not in the case of thin materials. Since it is quantified as the ratio between THR and mass at a specific time, the instrumental error for mass would be very high because of the low sensitivity of balance (as already mentioned). Thus, it is possible to overcome this instrumental limit for fabrics by calculating the normalised THR and thus eliminating the mass effect due to samples having a remarkable difference in terms of mass specimen, as in our case.

Furthermore, during combustion, the [CO] yield turned out to be a function of fabric grammage, while the [CO2] yield was almost equal for COT200 and COT400 (Table 2). As reported here, the amount of [CO] yield of COT400 is drastically lower than that of COT200. This may be attributed to a higher formation of char for COT400 than COT200.

The behaviour of these fabrics is ascribed to the different grammages, which can affect the irradiative heat transmission during cone calorimetry tests. Indeed, monitoring the transmitted heat flux, it is possible to observe that the lower the fabric grammage, the higher the heat flux transmission is, as is visible in Fig. 4a. Thus, COT400 absorbs a higher heat flux with respect to COT200 and COT100 (Fig. 5a). On the other hand, calculating the heat absorbed per mass unit in 1 s (Fig. 5b), it is possible to observe that COT100 absorbs more heat than COT200 and COT400; thus, reasonably, its temperature and heating rate are higher than those of the other two samples. As a result, the decomposition kinetics of the three fabrics may be significantly different as cotton degradation is governed by a precise equilibrium between the volatilisation and charring processes (Alongi et al. 2013a). The collected data demonstrated that the higher the absorbed heat (the higher heating rate), the more volatile species are released and the lower the char formed. In conclusion, the higher the fabric grammage, the higher the char formed is.

Fig. 4
figure 4

Transmitted heat flux measurements of COT100, COT200 and COT400 under 25 (a) and 35 kW/m2 (b)

Full size image
Fig. 5
figure 5

Absorbed heat flux measurements of COT100, COT200 and COT400 (a and c) and calculated absorbed heat (b and d)

Full size image

By increasing the heat flux to 35 kW/m2 (Table 2), COT100 gave reproducible data, as did COT200 and COT400. Once again, TTI and FO are non-linearly dependent on fabric grammage, as already observed under 25 kW/m2. Conversely, THR, nTHR and PHRR increase by increasing fabric grammage. Also in this case it was confirmed that the heat release rate is a material characteristic, as demonstrated by nPHRR values, and that the grammage affects the heat transmission (Fig. 4b) and adsorption (absorbed heat flux in kW/m2 and J/gs in Fig. 5c, d respectively).

At the end of these tests, independently of heat flux and fabric grammage, cotton does not leave any residue. In conclusion, fabric grammage may affect the parameters assessed by cone calorimetry, regardless of the adopted heat flux; indeed:

  1. 1.

    the highest TTI was found with the lowest fabric grammage,

  2. 2.

    the lowest THR (and nTHR) was found with the lowest fabric grammage, and

  3. 3.

    only nPHRR should be taken into consideration for fabrics as it should be an intrinsic characteristic of cotton.

The effect of fabric grammage on cotton combustion has also been investigated by horizontal and vertical flame spread tests. These two configurations are drastically different in terms of combustion kinetics (namely, afterflame time and burning rate); indeed, when cotton is exposed to a methane flame in a horizontal configuration, it burns more slowly than in a vertical configuration, as is clear when comparing the snapshots in Figs. 6 and 7, respectively. More specifically, when cotton is horizontally burnt, it is possible to measure the afterflame and afterglow times and the burning rate. This is almost impossible in a vertical configuration; thus, only a qualitative observation of the test was performed in this work.

Fig. 6
figure 6

Snapshots of COT100, COT200 and COT400 at 0, 20, 40 s and the time to reach 12.5 cm during horizontal flame spread tests

Full size image
Fig. 7
figure 7

Snapshots of COT100, COT200 and COT400 at 0, 5, 10 and 115 s during vertical flame spread tests

Full size image

In detail, after removing the ignition source, in a horizontal arrangement COT100, COT200 and COT400 burnt slowly under flaming combustion for 51, 78 and 141 s (afterflame time in Table 3), leaving a very thin residue that undergoes glowing combustion for 9, 11 and 34 s, respectively (afterglow time in Table 3; Fig. 6). As expected, by increasing the fabric grammage, the combustion duration (intended to be the sum of the afterflame and afterglow times) increases, while the burning rate decreases. Indeed, the more dense and compact the fabric, the more slowly it burns, as is visible when comparing the snapshots in Fig. 6. As a consequence, the time for reaching the 12.5 cm line is very different: 57, 78 and 171 s for COT100, COT200 and COT400, respectively.

Table 3 Combustion data of COT100, COT200 and COT400 collected by horizontal flame spread and LOI tests
Full size table

From a visual observation, the same behaviour of COT100 and COT200 was found when fabrics were placed in a vertical arrangement. They burnt completely and vigorously in approximately 15 s; the last 3 s involved the afterglow, which totally consumed the remained specimen, leaving no residue (Fig. 7). On the other hand, COT400 exhibited a different behaviour: indeed, after removing the flame, COT400 combustion was focussed on a smaller area with respect to the other two fabrics, as evidenced when comparing the snapshots at 0 s reported in Fig. 7. After 5 and 10 s, combustion involves a smaller area with respect to that burnt in COT100 and COT200. At 15 s, COT100 and COT200 are totally consumed, while COT400 goes on burning for another approximately 5 s in the presence of flame and another 3 min as glowing combustion, leaving a very thin and impalpable residue, the amount of which is however negligible. As was predictable, the burning rate observed in these tests was significantly higher than that observed in the horizontal configuration, but once again was conversely dependent on fabric grammage: the higher the fabric grammage, the lower the burning rate was. Also in these tests there was a precise dependence of cotton flammability (independently of the adopted configuration) on fabric grammage; indeed:

  1. 1.

    the highest afterflame and afterglow times and the lowest burning rates were found with the highest fabric grammage.

Beyond flame spread tests, the LOI was commonly used for characterising fabrics; indeed, it has long been considered a good flammability index by both industrial and academic researchers, although it is questionable whether the collected results can be correlated with those of any other test and in particular with those of a real fire scenario (Weil et al. 1992). Indeed, it is common opinion that the LOI can only be considered a useful tool for defining candle-like ignition and is of no value for real-world fire safety. However, in our opinion, it is important to measure this value as a function of fabric grammage. As reported in the last column of Table 3, COT100, COT200 and COT400 exhibited an LOI value of 19.5, 20.0 and 21.5 %, respectively. As the LOI represents the minimum concentration of oxygen in a flowing mixture of oxygen and nitrogen that will just support flaming combustion (ASTM D2863), it should represent an intrinsic characteristic of the material that should not depend on the fabric grammage (as studied here) and, as a consequence, on the fabric weave (Table 1). This finding raises some doubts about the real meaning of these tests for fabrics. In conclusion, fabric grammage may affect the cotton LOI value; indeed:

  1. 1.

    the highest LOI was found with the highest fabric grammage.

Referring to the behaviour of cotton when first pyrolysed and further burnt in PCFC, it was possible to observe a completely different behaviour with respect to that previously described (in particular with the cone calorimeter). Table 4 reports the collected data and Fig. 8a plots the HRR curves of COT100, COT200 and COT400.

Table 4 Data of COT100, COT200 and COT400 collected by pyrolysis-combustion flow calorimetry
Full size table
Fig. 8
figure 8

HRR (a), TG (b) and dTG (c) curves (in nitrogen) of COT100, COT200 and COT400

Full size image

First of all, the T initial, THR and PHRR values decreased with fabric grammage increases. This trend was completely opposite to that exhibited by the fabrics under investigation in cone calorimetry. Figure 9 reports the comparison among THR values collected by these two techniques as an example. In addition, also T PHRR and t PHRR decreased by the same trend.

Fig. 9
figure 9

Comparison between THR values curves of COT100, COT200 and COT400 collected by cone calorimetry and PCFC

Full size image

This discrepancy between cone calorimetry and PCFC can be easily explained taking into account the char's former character of cellulose contained in cotton fabrics. Indeed, thermogravimetry (Figs. 8b, c, 10, respectively) showed that COT400 forms more char (23 and 35 % in nitrogen and air, respectively) than those left by COT100 and COT200, which is approximately the same (17 and 25 %). This means that a lower amount of volatile species was released by COT400 if compared with COT100 and COT200. Thus, if COT100 produces more volatiles than COT400, more pyrolysis products will be produced and further oxidised in PCFC; as a consequence COT100 THR is higher than COT400 THR (11.7 vs. 8.8 kJ/g). Here it is not necessary to normalise THR and PHRR as the measured parameters have been directly given in kJ/g and W/g, respectively. Comparing PHRR values, it is important to highlight that these values are not equal as in the cone calorimeter and are a function of cotton grammage, to which the char-forming character is directly connected.

Fig. 10
figure 10

TG (a) and dTG (b) curves of COT100, COT200 and COT400 in air

Full size image

In conclusion, fabric grammage may affect the parameters assessed by PCFC; indeed:

  1. 1.

    the lowest THR and PHRR were found with the highest fabric grammage (assessed by PCFC), and indirectly

  2. 2.

    the highest char formation was found with the highest fabric grammage (assessed by thermogravimetry).

Conclusions

In the present manuscript, cotton fabrics with 100, 200 and 400 g/m2 grammage were tested by cone calorimetry, (horizontal and vertical) flame spread and LOI tests, PCFC and thermogravimetry (in nitrogen and air). It has been demonstrated that cotton combustion behaviour is affected by fabric grammage and the collected results significantly depend on the adopted test method, as graphically reported in Fig. 11. Thus, the present study provides basic and fundamental knowledge for the understanding and proper comparison of cotton combustion data evaluated by different means.

Fig. 11
figure 11

Relationship between cotton grammage and combustion data collected by different characterisation techniques

Full size image