Journal of Hazardous Materials 272 (2014) 137–147

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Photooxidation of cellulose nitrate: New insights into degradation mechanisms Sebastien Berthumeyrie a,c , Steeve Collin a,b , Pierre-Olivier Bussiere a,c , Sandrine Therias a,b,∗ a

Clermont Université, Université Blaise Pascal, Institut de Chimie de Clermont-Ferrand, BP 10448, F-63000 Clermont-Ferrand, France CNRS, UMR 6296, ICCF, BP 80026, F-63171 Aubiere, France c Clermont Université, ENSCCF, Institut de Chimie de Clermont-Ferrand, BP 10448, F-63000 Clermont-Ferrand, France b

h i g h l i g h t s • The photooxidation of cellulose nitrate was focused on the formation of polluting chemical products. • Chemical modifications revealed de-nitration of cellulose nitrate and formation lactone and anhydride. • Correlation between modifications in chemical structure and thermal functional properties.

a r t i c l e

i n f o

Article history: Received 27 December 2013 Received in revised form 21 February 2014 Accepted 23 February 2014 Available online 12 March 2014 Keywords: Photooxidation Cellulose nitrate UV light Propellants

a b s t r a c t Cellulose nitrate (or nitrocellulose) has received considerable interest due to its uses in various applications, such as paints, photographic films and propellants. However, it is considered as one of the primary pollutants in the energetic material industries because it can be degraded to form polluting chemical species. In this work, the UV light degradation of cellulose nitrate films was studied under conditions of artificially accelerated photooxidation. To eliminate the reactivity of nitro groups, the degradation of ethylcellulose was also investigated. Infrared spectroscopy analyses of the chemical modifications caused by the photooxidation of cellulose nitrate films and the resulting formation of volatile products revealed the occurrence of de-nitration and the formation of oxidation photoproducts exhibiting lactone and anhydride functions. The impact of these chemical modifications on the mechanical and thermal properties of cellulose nitrate films includes embrittlement and lower temperatures of ignition when used as a propellant. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Cellulose nitrate, or nitrocellulose (NC), is known as a versatile, widely used polymer with numerous applications. NC is obtained by preparing a nitrate ester of cellulose, a polysaccharide composed of pyranose rings. The maximum nitrogen content in NC is 14.14%, which corresponds to the replacement of the three free hydroxyl groups of cellulose with nitrate groups on each pyranose ring. With nitrogen content of less than 12%, NC is used for manufacturing photographic films, inks, and paints. The primary application of NC with nitrogen contents higher than 12% is in the preparation of explosive

∗ Corresponding author at: ICCF, UMR6296, Equipe Photochimie, 24 avenue des Landais, BP 80026, 63171 Aubière Cedex, France. Tel.: +33 04 73 40 71 43; fax: +33 04 73 40 77 00. E-mail address: [email protected] (S. Therias). http://dx.doi.org/10.1016/j.jhazmat.2014.02.039 0304-3894/© 2014 Elsevier B.V. All rights reserved.

formulations as propellants [1]. NC is considered as one of the main pollutant in energetic materials industries and the presence of nitro compounds materials in the waster effluent causes severe environmental problems, high toxicity. NC has been found to be sensitive to various elements, including heat [2–7] and light [8–10]. Therefore controlling NC degradation is essential to ensuring environmentally friendly waste management [11] as well as the stability of the NC properties [12]. Although the thermal decomposition of NC has been widely investigated in terms of its mechanism and kinetics, the light sensitivity of NC has received considerably less attention. During primary decomposition, NO2 is released and three fully nitrated pyranose rings disintegrate, inducing chain scission reactions [2,4–7,10]. During secondary degradation, NO2 acts as an oxidiser [6,7,10]. The reactions rates of secondary degradation are faster than those of primary decomposition, particularly when UV-light and oxygen are provided because light reduces NO2 to NO and

138

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

singlet oxygen, two highly reactive species [10]. The formation of carbonyl functions [4,5,8] on cellulosic backbones and the occurrence of cross-linking reactions [8] have been reported but were not extensively investigated. Thus, a mechanism for the chemical modifications of NC during thermal- and photodegradation that accounts for the primary oxidation species has yet to be proposed. The main objective of this work was to provide a detailed degradation mechanism for cellulose nitrate (NC) under photoaging that explains the modification of its chemical structure and accounts for the formation of known oxidative species. Within this objective, diethyl cellulose (EC) was studied as a model polymer of NC. EC was chosen to highlight the specific reactivity of the nitrate substituent groups. NC and EC thin films (4–30 ␮m thick) were submitted to artificially accelerated weathering under UV–vis light irradiation ( > 295 nm, 60 ◦ C) in ambient air. The resulting chemical modifications were investigated by IR spectroscopy. The photoproducts were identified by various analytical measurements of the oxidised films combining physical and chemical derivatisation treatments coupled with IR spectroscopy. Identification of the low molecular weight products was performed by head-space solid-phase micro-extraction (HS-SPME) coupled with GC–MS analysis and ion chromatography experiments. Another goal of this work was to elucidate the qualitative and quantitative relationship between the modification of the chemical structure upon irradiation and the degradation of the materials properties of NC, such as its thermal properties, which are particularly important in predicting the fate of the material.

2. Experimental 2.1. Materials 2.1.1. Cellulose nitrate and ethyl cellulose films Nitrocellulose and ethylcellulose (EC) are produced from cellulose. These polymers are composed of d-glucopyranose units linked by glucosidic bonds with a ␤(1–4) conformation. The structural differences from cellulose consist of the groups present on carbons C2, C3 and C6 [13]. These three positions, in the case of nitrocellulose, can be occupied by nitro groups rather than hydroxyl groups, or by ethoxy groups in the case of EC. The chemical structures of these polymers depend on the relative ease of substitution at each site and on the average number of substituted groups in each pyranose unit. Substitution reactions on cellulose are statistical, allowing the monomers to undergo varying degrees of substitution. Cellulose nitrate (NC) films were made from membranes purchased from GE Healthcare (US). Quantitative elemental analysis indicated weight percentages of carbon, hydrogen and nitrogen of 27.20%, 2.93% and 11.76%, respectively. Consequently, this sample should contain 2.2 nitro groups per glucopyranosyl unit. Ethyl cellulose (EC) containing 2.5 ethoxy groups per glucopyranosyl unit was supplied by Scientific Polymer Products. According to the literature [13,14], for a substitution degree higher than 2, the most

(a)

(b) C6 H

O NO 2

C6

O O

C5

H C4 O NO 2 H

C3 H

C2

H

O

C5

O O

H

C1 H

O NO 2

C4

O

C3 H

H

C2 O

Fig. 1. Chemical structures of the polymers (a) NC and (b) EC.

C1 H

representative monomer is tri-substituted. Accordingly, the chemical structures of NC and EC used in this work are depicted in Fig. 1. Films of NC and EC with thicknesses of 10 and 60 ␮m, respectively, were obtained from solutions of THF and chloroform at a maximum concentration of 10 g l−1 . Following solution casting using a coat-master (Erichsen Coatmaster 809 MC) for the NC samples and Petri boxes for the EC samples, the films were allowed to dry in ambient air for a minimum of 24 h. Thin samples (thickness less than 10 ␮m) were obtained by depositing a few droplets of dilute solution on KBr windows. 2.1.2. “Impregnated” diethylcellulose films To identify the degradation products, EC films blended with low molecular weight species that were potential oxidation products of the polymer, such as gluconolactone (Sigma–Aldrich) and glucuronic acid (Sigma–Aldrich), were prepared in methanol and chloroform, respectively. Blends of EC and poly (maleic anhydride alt 1 octadecene) were prepared in chloroform, and free-standing films were obtained after at least 24 h of drying in ambient air. 2.2. Accelerated weathering 2.2.1. UV light irradiation The UV–vis light irradiation ( > 295 nm) of the films was performed in an irradiation device, the SEPAP 12/24 unit (ATLAS) that was designed for studying polymer photodegradation under artificial ageing with medium-accelerated conditions [15]. The chamber consisted of a square reactor equipped with four medium-pressure mercury lamps (Novalamp RVC 400 W) situated vertically at each corner of the chamber. Wavelengths below 295 nm were filtered by the glass envelope of the lamps. In the centre of the chamber, the samples were fixed on a 13-cm diameter rotating carousel that could hold up to 24 samples. In this set of experiments, the temperature at the surface of the samples was fixed at 60 ◦ C. Irradiation in the absence of oxygen (photolysis experiments) was performed on samples introduced into borosilicate tubes and sealed under a vacuum of 10−4 Pa with a diffusion vacuum line. The samples under vacuum in the tubes were then placed in the SEPAP 12/24 device for ageing. 2.3. Characterisation 2.3.1. Infrared analysis IR spectra were recorded in transmission mode with a Nicolet 6700 Fourier transform infrared (FTIR) spectrophotometer operated with the OMNIC software. The spectra were obtained with 32 scan summations at 4 cm−1 resolution. The IR absorption band corresponding to the acetal structure at 1070 cm−1 was used as a reference for the film thickness. IR-ATR (Attenuated Total Reflectance) spectra were recorded in reflection mode with a Nicolet 380-FTIR spectrophotometer equipped with a Thunderdome-ATR (4 cm−1 , 32 scans). The Thunderdome is a single reflection ATR accessory with a diamond crystal (depth analysis ∼2–3 ␮m). 2D FTIR correlation spectroscopy was performed using the commercial software SpectraCorr from Thermo Scientific. The theory and methodology of 2D FTIR spectroscopy are summarised elsewhere [16]. Briefly, IR spectra were recorded at different ageing times, and the spectrum of the un-aged sample was used as a reference and therefore subtracted from each spectrum prior to the calculations. The complex cross correlation intensity was calculated by the software. The real part (synchronous correlation) and imaginary part (asynchronous correlation) are reported in this work.

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

2.3.3. Size exclusion chromatography (SEC) The change in molecular weight of the NC was determined by size exclusion chromatography (SEC) using the Viscotek SEC-TDA equipment. The instrument consisted of a TDA 302 module (triple detector array) that included a column oven and a triple detector consisting of an RI detector, a four-bridge viscometer and an LS detector. The latter consisted of a right-angle light scattering (RALS) detector and an innovative low-angle light scattering (LALS) detector. Two TSK-GEL columns (GMHXL and G3000HXL) were used in series and preceded by a TSK-GEL guard column (HXL-L). The analyses were performed with THF as the eluent at a flow rate of 1 ml min−1 . The NC solutions were prepared in THF (∼15 mg polymer/5 ml solvent) and were filtered prior to injection. The OmniSEC software program was used for the acquisition and analysis of the Viscotek data. The detector constants were determined using a polystyrene standard (Mw ∼ 99 000 Da). A second standard (Mw ∼ 235 000 Da) was used to verify the calibration. 2.3.4. Solid phase micro extraction (SPME) Films of EC and NC were also irradiated in sealed vials to collect the volatile photodegradation products. A Carboxen–PDMS fibre (75 mm) purchased from Supelco (Bellefonte, PA, USA) was used to extract the volatile products. The extraction time was 5 min at 60 ◦ C. The volatile compounds were analysed by gas chromatography/mass spectrometry (GC–MS) with a 6890N Agilent GC coupled to a 5973 Agilent mass detector. The GC was equipped with a PEG Supelcowax® 10 column (30 m × 0.25 mm × 0.25 mm) from Supelco. Splitless injections were used (5 s). The oven temperature was cycled from 35 ◦ C (10 min hold) to 60 ◦ C at 5 ◦ C min−1 and then increased at 10 ◦ C min−1 to 200 ◦ C (15 min hold). Helium was used as the carrier gas with a constant flow rate of 1 ml min−1 . The temperature of the splitless injector was 280 ◦ C. The temperature of the transfer line between the oven and the detector was 280 ◦ C. The electron source temperature was maintained at 230 ◦ C. The mass spectra and reconstructed chromatograms (total ion current, TIC) were acquired under the electron ionisation mode (EI) at 70 eV and recorded from 20 to 400 m/z. The compounds were identified by comparing the retention times and mass spectra with either the standards or the spectral library. 2.3.5. Ionic chromatography analyses (IC) Samples of NC (60 mg) were photo-aged in SPME vials. A 2 ml volume of Milli-Q water was introduced into the vials with a syringe. Each vial was shaken by hand for 30 s. A 1.5 ml volume of this water was then removed from the vials and added to 3.5 ml of water. The resulting solutions were filtered and diluted 20-fold. The column used was a METROSEP A SUPP 4 with a length of 250 mm and a diameter of 4.0 mm. The particle size within the column was 9.0 ␮m, and the elution rate was 1 ml/min. The eluting solvent was an aqueous solution of Na2 CO3 at a concentration of 1.8 mmol l−1 containing NaHCO3 at a concentration of 1.7 mmol l−1 . The ions were detected by conductimetry. 2.3.6. Differential scanning calorimetry (DSC) The thermal decomposition of NC was measured using a Mettler Toledo DSC 822 programmed from 150 to 300 ◦ C at a heating rate of 2 ◦ C min−1 . The analyses were performed on films with thicknesses of ∼150 ␮m that had been placed in a 40-␮L crucible.

2.3.7. Traction analysis Traction tests were performed on rectangular samples (dimensions 35 mm × 8 mm, thickness ∼10 ␮m) using a TA Q800 DMTA apparatus. 2.3.8. AFM nanoscale thermal analysis Atomic force microscopy (AFM) was performed on a Bruker Nanoscope IIIa atomic force microscope. The modification of the degradation temperature at the surface of the NC substrates was monitored using an AFM nanoscale thermal analysis module (Vita). This technique allows the determination of the local degradation temperature of polymers with nanoscale spatial resolution. The measurements were performed on irradiated and non-irradiated NC samples to investigate changes in the thermal properties of the samples during irradiation. Each measurement was performed 6 times to ensure good reproducibility. During these experiments, the probe was heated from 25 ◦ C to 300 ◦ C at a heating rate of 5 ◦ C/s. The Vita probes can also be used to obtain AFM images of the surfaces. Thus, AFM images of the samples were captured after testing using the same Vita probe in contact mode [19]. 3. Results and discussion 3.1. Determination of the mechanism of degradation 3.1.1. Modification of the chemical structure as measured by Infrared spectroscopy To characterise the modification of the chemical structure of cellulose nitrate (NC) under photo-ageing, it is important to first identify the IR bands of the polymer. As explained in the introduction, to highlight the reactivity of nitrate groups, diethyl cellulose (EC) was selected as a model polymer. The IR spectra of (NC) and diethyl cellulose (EC) prior to irradiation are therefore presented in Fig. 2. The main spectral features of the NC and EC polymers are presented in Table 1. Both the EC and NC samples exhibited the characteristic bands of cellulosic compounds [20]. Strong absorption bands associated with the acetal linkages characteristic of carbohydrate structures are observed in the range of 1200–1000 cm−1 . The absorption band at 920 cm−1 corresponds to the pyranose ring vibration. The specific band from the ␤ glucosidic bond is observed at 880 cm−1 in EC. However, this band cannot be observed in the NC spectrum because of the large band at 840 cm−1 due to the nitro groups. Other characteristic and intense absorption bands of the nitro groups are noted in the spectrum of NC at 1650, 1280, 750 and 690 cm−1 . The spectrum of EC presents the IR bands characteristic of ethoxy groups and, in particular, the 3.0

NC EC

2.5 2.0 Absorbance

2.3.2. Chemical derivatisation reaction The oxidation products were identified by performing chemical derivatisation treatments that selectively converted the oxidation products into chemical groups with different IR characteristics [17,18]. For example, ammonia (NH3 ) reacts with carboxylic acids and esters to generate, respectively, carboxylate ions and amides.

139

1.5 1.0 0.5 0.0 4000

3500

3000

2500 2000 -1 Wavelength (cm )

1500

1000

500

Fig. 2. IR spectra of a thin deposit (e ∼ 15 ␮m) of cellulose nitrate (NC) and diethyl cellulose (EC).

140

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

Table 1 Assignments of the main infrared absorption bands of NC and EC [20–23,35]. NC  (cm−1 )

EC  (cm−1 )

Assignment

3500

3480 2975 2930 2900 2870

␯ OH ␯a CH in CH3 from ethoxy groups ␯a CH in CH2 in cellulosic backbone ␯s CH in CH3 from ethoxy groups ␯s CH in CH2 in cellulosic backbone ␯a NO2 ␦ CH in CH3 from ethoxy groups ␦ CH2 in cellulosic backbone ␦ CH in cellulosic backbone ␦ OH ␯s NO2 ␯a COC in ethoxy groups ␯s COC in ethoxy groups C5 OC1 OC4 (acetal structure of polysaccharides)

2965 2905 1650 1460 1375

1485 1460 1375 1355

1280

1160 1120 1060

1110 1100 1160 1110 1060

O R O O

H O R C4 H

C5 H O R

H O R

H

H

O R

H

O O H

C1 H

H

920 840 750 690

920 880

O R

␦ pyranose ring ␤ glucosidic bond ␯ NO ␦ NO2 ␦ NO2

␯, stretching; ␦, deformation; subscript a, anti-symmetric; subscript s, symmetric.

characteristic vibrations of CH3 groups at 2975 cm−1 . One can also note the higher frequency of the stretching vibrations of CH2 in the cellulosic backbone of the NC film relative to the EC film. This behaviour can be attributed to the presence in the ␣ position of the strongly electron attractive group ONO2 . A shoulder can also be observed at approximately 1740 cm−1 in the IR spectrum of the non-irradiated NC film, which may be due to the presence of small amounts of oxidation products. 3.1.1.1. Nitrocellulose photooxidation. The modifications in the IR spectra of the cellulose nitrate samples submitted to photooxidation are reported in Fig. 3. Irradiation leads to the appearance of new absorption bands in the hydroxyl region (a broad band with a maximum at 3480 cm−1 ) and in the carbonyl region (a broad band with a maximum at 1740 cm−1 ). The intensities of the initial bands of the nitro groups at 1650 cm−1 , 1280 cm−1 and 840 cm−1 decrease during irradiation,

Fig. 3. IR spectra of the nitrocellulose (NC) film (e ∼ 4 ␮m) during photooxidation.

revealing the progressive de-nitration of nitrocellulose, consistent with results previously reported in the literature [8]. Moreover, the broad band with a maximum at 1070 cm−1 corresponding to the acetal structure of polysaccharides and the band at 920 cm−1 related to the pyranose ring also decrease with increasing exposure duration. To underline the connections between these spectral features, 2D FTIR correlation spectroscopy, an analytical tool that has previously been applied in the study of polymer degradation [16,24], was used in the range of 1900–750 cm−1 . Synchronous and asynchronous correlation intensities calculated from the IR spectra of the photooxidised NC film are provided in Fig. 4. The synchronous correlation is presented in Fig. 4a. It exhibits autopeaks at 1740, 1650, 1280, 1070 and 850 cm−1 . These data confirm that the main spectral alterations during photooxidation occur in the characteristic bands of the nitrate groups (1650, 1280 and 850 cm−1 ), the acetal structures (broadband centred at 1070 cm−1 ) and in the carbonyl region (broadband centred at 1740 cm−1 ). Negative cross-peaks (off-diagonal positions) are observed between the bands related to the carbonyl region and the nitrate groups, confirming that the absorption of the carbonyl groups increases, while that of the nitrate groups decreases. A positive cross-peak at position [1740, 1210 cm−1 ] is also observed. Although it was not clearly defined in Fig. 3, the increase in absorption at 1210 cm−1 should be directly correlated with the increase in absorption at 1740 cm−1 . This behaviour was expected because the peak at 1210 cm−1 typically corresponds to the C O stretch in carbonyl compounds, such as esters, carboxylic acids and anhydrides. The asynchronous correlation revealed in Fig. 4b provides further insight through broadband decomposition [16]. Crosspeaks can be observed at positions [1660, 1630 cm−1 ] and [1295, 1275 cm−1 ], indicating that these bands should be decomposed into two bands at 1660 cm−1 and 1630 cm−1 and 1295 cm−1 and 1275 cm−1 , respectively. These bands can be unequivocally assigned to nitrate species, and, considering the chemical formula of the NC samples used in this study (as specified in Section 2), the peaks can be attributed to the presence of two different nitrate species. 3.1.1.2. Ethylcellulose photooxidation. As explained in the introduction, EC was used as a model polymer and as a probe to illustrate nitrate group reactivity. The irradiation of the diethylcellulose (EC) film led to several changes in the IR spectrum of EC (Fig. 5). In the hydroxyl region, the absorption intensity and bandwidth increased. A new absorption band with a maximum at 1740 cm−1 appeared in the carbonyl vibration region, as observed in the NC spectrum. An increase in absorbance at 1640 cm−1 was also observed. If this band was present following the photooxidation of NC, it was obscured by the large band from the NO2 groups at 1650 cm−1 . Simultaneously, a band at approximately 1235 cm−1 appeared. As with the NC film, the photooxidation of the EC film led to a decrease in the absorption intensity of the acetal structures, corresponding to the broadband centred at 1100 cm−1 . The decrease in acetal functional groups was correlated with the decreases in absorbance at 920 cm−1 (pyranose rings) and 880 cm−1 (␤ glucosidic bond). This observation suggests that chain scissions and ring opening occurred during photodegradation. Two-dimensional FTIR analysis was also performed. In the synchronous spectrum (not shown), three autopeaks were observed: 1740 cm−1 (assigned to C O bonds), 1235 cm−1 (assigned to C O bonds) and 1110 cm−1 (assigned to acetal structures). The presence of negative cross-peaks confirmed that the absorption intensity at 1740 cm−1 and 1235 cm−1 increased during photooxidation, whereas the intensity at 1110 cm−1 decreased. The asynchronous spectrum displayed a positive cross-peak at [1750, 1715 cm−1 ], indicating that the broad band at 1740 cm−1 could comprise two

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

141

Fig. 4. 2D correlation FTIR spectra from the photooxidation of a NC thin film (e ∼ 4 ␮m) in the 1900–750 cm−1 range. (a) Synchronous spectrum and (b) asynchronous spectrum.

and involved in the formation of carbonyl products and the loss of acetal groups (band at 1060 cm−1 ). Moreover, the band at 920 cm−1 due to the pyranose ring was unaffected by the treatment.

3.1.2. Identification of photoproducts To identify the degradation products detected by IR analysis, additional experiments were performed. The NC film was irradiated in the absence of oxygen to emphasise the specific reactivity of the nitro groups. Finally, a chemical derivatisation treatment using NH3 was performed on the photooxidised NC and EC films to identify carbonyl compounds [17,18].

3.1.2.2. Chemical treatment of photooxidised EC and NC films with NH3 . To identify the carbonylated products that gave rise to the absorption band at 1740 cm−1 in the IR spectra of the photooxidised NC films, NH3 treatment was performed. NH3 reacts with carboxylic acids and esters to produce, respectively, carboxylate and amide groups [17,18]. We first verified that no reaction occurred between NH3 and the NC film prior to photooxidation. However, following photooxidation, NH3 treatment of the NC sample caused a decrease in the broad absorption band at 1740 cm−1 and the emergence of two bands at 1685 cm−1 and 1595 cm−1 (see Fig. 7). These bands could be related to the formation of amide (1685 cm−1 ) and ammonium carboxylate groups (1595 cm−1 ). The formation of amides reveals the presence of an ester or anhydride, while ammonium carboxylate formation stems from carboxylic acid or anhydride moieties. Similar reactivity with NH3 was observed in the photooxidised EC sample (Fig. 7c and d), suggesting that the chemical structures of some of the photodegradation products of EC and NC are similar. Moreover, for the EC film, the subtracted spectrum (before and after NH3 treatment) (Fig. 7d) allowed us to achieve the “chemical deconvolution” of the IR band at 1740 cm−1 by revealing decreases in the 2 maxima at 1765 cm−1 and 1740 cm−1 . This

-1

840cm

-1

1740cm

Absorbance

2,0 1,5

-1

1060cm

2,5

1280cm

0h 20h

-1

3,0

1650cm

3.1.2.1. Irradiation in the absence of oxygen. To investigate the photochemical reactivity of ONO2 functional groups in nitrocellulose, a NC film was submitted to irradiation in the absence of oxygen (photolysis). The IR spectra of the NC film before and after 20 h of photolysis are shown in Fig. 6. Following photolysis, the IR spectrum of the NC exhibited decreases in the absorption bands at 1650, 1280, and 850 cm−1 , corresponding to the loss of NO2 groups under UV-light irradiation. An increase in absorbance in the carbonyl region appeared at 1740 cm−1 , while an increase in absorption at 1210 cm−1 and a decrease in the broadband absorption centred at 1060 cm−1 corresponded to the acetal functional groups. The same experiment was performed on an EC film, and no modifications in the IR spectra were detected. From this experiment, we can conclude that the CO NO2 bond is photochemically reactive

-1

overlapping bands, suggesting the formation of two types of photooxidation products. This phenomenon was not observed during NC photooxidation.

1,0 0,5 0,0 4000

Fig. 5. IR spectra of the EC film (e = 15–20 ␮m) during photooxidation.

3500

3000

2500 2000 -1 Wavelength (cm )

1500

1000

500

Fig. 6. IR spectra of a NC film (e ∼ 30 ␮m) before and after 20 h of irradiation in the absence of oxygen.

142

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

Table 2 IR absorption bands (␯(C O)) of potential oxidation products in an EC film and band shifts following NH3 chemical treatment. Potential oxidation product

Chemical formula

␯(C O) band(s) before treatment

␯(C O) band(s) after NH3 treatment

1740 cm−1

1640 cm−1 1590 cm−1

1710 cm−1

1600 cm−1

1780 cm−1 1860 cm−1

1660 cm−1 1555 cm−1

OH O

HO H H

H

OH

OH

O

H

OH

Gluconolactone

HO

OH H HO H H

O

H OH

OH

Hydrolysis into gluconic acid

O OH O

HO H

H H H

HO OH

OH O

Poly (maleic anhydride alt 1 octadecene)

n

1600

-1 -1

-0.4 1900

1500

-1

0.2 0.0 1900

1800

1700

1500

(d)

1600 -1

Wavelength (cm )

1500

0.2 -1

Absorbance

1600 cm

0.4

1600

-1

-1

0.6

1700

Wavelength (cm ) 0.4

(c)

1680 cm

Absorbance

0.8

1800

-1

Wavelength (cm )

0.0 -0.2 1900

1800

1700

-1

1700

-0.2

1600 cm

1800

0.0

-1

0.2 1900

1740 cm

0.4

(b)

0.2

1765 cm -1 1740 cm

0.6

Absorbance

0.8

1595 cm

0.4

-1

(a)

1740 cm

Absorbance

1.0

-1

O

1685 cm

O

1680 cm

Glucuronic acid

1600

1500

-1

Wavelength (cm )

Fig. 7. IR spectra in the carbonyl region (1900–1500 cm−1 ) before/after NH3 treatment of (a and b) a 50 h photooxidised NC film (e ∼ 3 ␮m) and (c and d) a 40 h photooxidised EC film (e ∼ 15–20 ␮m). (a and c) Raw spectra; (b and d) Subtracted spectrum revealing differences between the IR spectra before and after treatment.

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

observation clearly confirms the presence of two different photooxidation products, in agreement with the results of 2D FTIR. Both products react with NH3 and must therefore contain acidic and/or ester functions. It is also important to note that the band at 1640 cm−1 detected during the photooxidation of EC was not affected by NH3 treatment. Thus, this photoproduct represents a chemical function that is not sensitive to NH3 . 3.1.2.3. Potential oxidation photoproducts. To identify the photoproducts in the carbonyl region, several potential oxidation photoproducts (in the solid state and as a blend with EC) were analysed by IR spectroscopy and then subjected to NH3 treatment. Based on previous studies addressing the oxidative degradation of cellulose [25], the formation of gluconolactone-type compounds containing ester functions should be considered. This product is obtained by the scission of the ␤ glycosidic linkage and the oxidation of the C1 carbon in the pyranose ring. It has also been reported that the oxidation of carbon C6 leads to the formation of a carboxylic acid, namely, a glucuronic acid. Finally, anhydride compounds should also be investigated; a polymer containing anhydride functions, namely, poly (maleic anhydride alt 1 octadecene) was therefore chosen as a reference to represent cyclic anhydride functions. The chemical formulas of these three compounds are reported in Table 2. The results of the identification of these products by IR spectroscopy are also given in Table 2. It is apparent that a band was recorded at 1740 cm−1 for gluconolactone and at 1710 cm−1 for glucuronic acid. Two bands were also observed for the anhydride (1780 cm−1 , 1860 cm−1 ). After NH3 treatment, two bands were observed for gluconolactone (1640 cm−1 and 1590 cm−1 ). In fact, gluconolactone can be easily hydrolysed into gluconic acid, which can react with NH3 , explaining the presence of the band at 1590 cm−1 . From these results, gluconolactone can be identified as the primary degradation product of the photooxidation of NC and EC. For the EC film, the double population observed following chemical treatment is likely correlated to the presence of anhydride groups. In addition, because EC is considered a model polymer, the presence of anhydride during the photo-oxidation of NC should also be considered. 3.1.3. Analysis of low molecular weight products: solid phase micro-extraction (SPME) and ionic chromatography (IC) For the SPME and IC experiments, NC films were irradiated in glass vials sealed under air atmosphere. In the SPME analysis, methyl nitrate was detected as the main molecular weight product formed in the gas phase during photooxidation. We can conclude that the de-nitration of NC films occurs during photooxidation. De-nitration was previously reported in the literature but only produced inorganic nitrates and nitric oxides [10]. In particular, the de-nitration of NC films should induce the liberation of nitrogen dioxide, a reddish gas that can be dissolved in water to form nitric acid. Indeed, when the NC film was photo-aged in a sealed vial for SPME analysis, a reddish vapour was produced. To verify the presence of nitrogen oxide, distinct experiments were performed. The acidic character of the red vapour can be demonstrated by placing pH paper in the vial. It is important to note that the methyl nitrate detected by SPME exhibited a pKa of 10.2 [26] and therefore cannot be the source of any observed acidity. To verify the presence of nitric acid in this vapour phase, IC experiments were performed. Vapours were first dissolved in pure water, and the solution was then analysed by IC. It was determined that nitrate ions were present in the solution and that their concentration increased with ageing time (Table 3). In addition, it is possible to conclude that the nitrate present in the vapour phase

143

Table 3 Changes in the concentration of NO3 − ions and percentage of O NO2 groups present in the film as a function of irradiation time. Irradiation time (h) 0 10 20

Concentration of NO3 − ions in water (mg l−1 ) 0 683 ± 1 1571 ± 4

Percentage of ONO2 groups still present in film (%) 100 86 67

was directly linked to the decrease in the IR band corresponding to the nitrate function and thus to the loss of the ONO2 groups from the NC film. 3.1.4. Mechanism of degradation of cellulose nitrate The experimental results reported above reveal that the degradation products resulting from the photooxidation of cellulose nitrate are predominantly composed of gluconolactone. Using EC as a model polymer, it is also apparent that the photo-ageing of these cellulose derivatives leads to the formation of anhydrides and acids. The following mechanism, which accounts for the identified oxidation photoproducts, can therefore be proposed (Scheme 1) and involves oxidation pathways on three different NC carbon atoms. A first route (pathway (Pho)) for the chemical modifications of cellulose nitrate under UV-irradiation must be taken into account and concerns the direct homo scission of the O NO2 bond that can occur, leading to the formation of an alkoxy macroradical at carbon atom C3 . In addition, during photooxidation, this step can be considered the initiation step. According to the literature, de-nitration by homolytic scission of O NO2 bonds should occur preferentially at the C3 on cellulose nitrate films [13]. Thus, reactivity at this carbon is proposed in Scheme 1. The fixation of hydrogen gives rise to an alcohol group at C3 following the de-nitration of NC. In the secondary degradation step, • NO2 can initiate hydrogen abstraction, similar to chromophoric radical impurities (X• ) [10]. For example, de-nitration followed by hydrogen abstraction at C3 leads to the formation of a cyclic ketone, K, as the intermediate photoproduct. This intermediate can then undergo Norrish reactions, leading to ring opening and the formation of carboxylic acids, but does not involve macromolecular chain scission. It has been shown that the hydrogen atom in the ␣-position of a tertiary carbon in an ether moiety is the preferential site for radical attack. Such behaviour has been clearly demonstrated in the degradation mechanism of several polymers, such as polyethylene oxide [27] and polypropylene [28]. Hydrogen abstraction at the tertiary carbon atom C1 (pathway (C1) in Scheme 1) leads to the formation of a macroalkyl radical, and after subsequent oxygen fixation and hydrogen abstraction, hydroperoxides are formed as the primary oxidation products. The photochemical decomposition of the hydroperoxides leads to an alkoxy radical (R1). A ␤-scission at the C O bond on R1 can occur, producing a gluconolactone derivative (G1) and involving the breakage of the ␤ glycosidic link with a corresponding loss of the acetal structure, as determined by IR spectroscopy. Hydrogen abstraction at the tertiary carbon atom C5 (pathway (C5) in Scheme 1) leads to the formation of hydroperoxides that photochemically decompose to produce alkoxy radicals (R2). The ␤-scission of the C5 –C6 bond on R2 leads to the formation of another cyclic ester, which is also denoted as a gluconolactone derivative (G2), as well as a low molecular weight methyl nitrate product (detected by SPME). However, this ␤-scission does not imply macromolecular chain scission. Both types of gluconolactone photoproducts are detected in the IR spectra by the absorption band at 1740 cm−1 . Another hydrogen abstraction at C5 on G1 can also occur via the ␤-scission previously described between C5 and C6 (pathway (A)), producing a C O function in the gluconolactone

144

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

2

Scheme 1. Photooxidation mechanism of cellulose nitrate.

ring leading to the anhydride product A detected at 1765 cm−1 . Anhydride A can also be formed from the gluconolactone G2. The formation of the anhydride product A through two successive hydrogen abstractions (on C1 and C5 or on C5 and C1 ) involves

the loss of the glycosidic bond and of the acetal structure, as determined by IR spectroscopy. Finally, hydrogen abstraction at carbon C3 (pathway (C3) in Scheme 1) leads to the formation of a radical in the O NO2 group.

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147 Table 4 Molecular weight and s(t) of the NC polymer during photooxidation.

0 3 5 10 20

s(t)

125 000 74 000 56 000 39 000 27 000

0 1.6 2.4 4.4 7.3

Heat (W/g)

8

Mw (g mol−1 )

Time of irradiation (h)

145

0h 10h 20h

6 4 2 0

In the presence of X• species, such as • NO2 , this structure is particularly unstable and can lead to the disintegration of pyranose ring [10].

3.2.1. Loss of mechanical properties The photooxidation mechanism of cellulose nitrate indicates that products, such as gluconolactone and anhydride are formed. The formation of these photoproducts leads to macromolecular chain scissions, which affect the molecular weight distribution of the polymer. To monitor the changes in the molecular weight of NC during photooxidation, Size exclusion chromatography (SEC) was performed. The determination of the dn/dc factor was first achieved. The obtained value of 0.072 mlg is in good agreement with the results provided in the literature [29]. The modifications in the molecular weight of the NC film during photooxidation are reported in Table 4. A random chain scission model was used to calculate the average number of chain scissions in one chain s(t) [30] following Eq. (1).



1 Mw(t)

−

1



(1)

Mw(t=0)

The calculated s(t) values for the different irradiation times are reported in Table 4. As shown in Table 4, the Mw for un-aged NC is 125 000 g mol−1 and gradually decreases during photooxidation, while s(t) increases. Correlating the molecular weight changes with the oxidation data obtained by IR spectroscopy should provide significant mechanistic insights. The variations in the intensities of the absorption bands attributed to the acetal structures (1060 cm−1 ), the pyranose ring (920 cm−1 ) and gluconolactone (1735 cm−1 ) were plotted. The results presented in Fig. 8 reveal a strong correlation between these parameters. The average number of chain scissions induced

10

0.15

S(t) pyranose gluconolactone acetal

9 8

0.10 0.05

7

S(t)

-0.05

5

-0.10

4 3

-0.15

2

-0.20

1

-0.25

0 0

5

10

15

absorbance

0.00

6

200

250

Temperature (°C) Fig. 9. Modification of the heat of NC decomposition during photooxidation: DSC thermograms.

3.2. Relationship between chemical modifications and functional properties

s(t) = 125 000.2

150

-0.30 20

Time (hours) Fig. 8. Average number of chain scissions of NC and absorbance of the IR bands of pyranose, gluconolactone and acetal groups as a function of photooxidation time.

by photodegradation is correlated with the loss of the pyranose and acetal structures and the formation of glucanolactone. These results verify the proposed photodegradation mechanism. These results indicate that NC predominantly undergoes chain scissions under photooxidation, which is likely to impact the mechanical properties of the material. Tensile tests confirmed the embrittlement of the NC films during photooxidation. The Young’s modulus and the elongation at break decreased from 2.5 GPa and 5% to 1.1 GPa and 1.2% after only 10 h of photooxidation. 3.2.2. Loss of thermal properties NC has attracted significant interest as a propellant [1]. This property is directly linked to its heat of decomposition (i.e., its thermal decomposition temperature), which leads to the release of a significant quantity of gas. This property was studied by DSC, and the results obtained are presented in Fig. 9. An un-aged NC film exhibited a dramatic exothermic peak at approximately 190 ◦ C, consistent with previous results [5]. Integration of the peak allows the heat of NC decomposition to be determined. A decrease in this value was observed, indicating that the efficiency of NC as a propellant should also decrease. Rychly et al. obtained similar results for the thermal oxidation of NC films [5]. Because this behaviour is not observed in pure cellulose (lacking O NO2 groups) [31], it is likely that the heat of decomposition is directly related to the amount of O NO2 groups in the sample. Interestingly, a loss of ONO2 groups with increased ageing time was observed in both the ion chromatography and infrared analyses and can be directly correlated to the decrease in the heat of NC decomposition. The decomposition temperature (Td ), which is related to the nitrogen content in the NC sample, was also studied [4]. AFM Vita® thermal analysis was performed to measure changes in the decomposition temperature at the surface. In this AFM mode, the tip is progressively heated until a thermal transition occurs. The variation in the tip deflection directly provides the onset temperature of the transition. The results are displayed in Fig. 10. The data in Fig. 10 reveal that Td decreases under photooxidative conditions down to 15 ◦ C after 50 h of exposure. This result has important implications for the use of NC as a propellant because such a decrease in Td could lead to the unexpected ignition of the propellant. Fig. 10 also shows the cross-correlation between the changes in Td during photooxidation and the loss of the O NO2 group, as measured by infrared spectroscopy. A linear fit is observed, which was not unexpected because the decomposition temperature and the corresponding energy can be directly related to the presence of O NO2 groups. 3.2.3. Correlation To correlate the chemical modifications due to polymer oxidation with the mechanical properties of the polymer, we searched for kinetic information related to the various degradation parameters.

146

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147

Table 5 kinetic values obtained from the photooxidation of the NC film. Wavelength (cm−1 )

Chemical function Pyranose ring Acetal Gluconolactone Anhydride

Kinetic model −kt

First order(X = X∝ + (X0 − X∝ )e First order First order First order

920 1060 1735 1765

)

t1/2 (h)

k (h−1 )

R2

14 23 17 22

0.05 0.03 0.04 0.03

0.988 0.986 0.987 0.993

Properties

Kinetic model

t1/2

k

R2

Molecular weight Young’s modulus

First order First order

24 18

0.03 0.04

0.992 0.984

References

230 -1

840 cm 220 -1 1650 cm -1 1280 cm 210 Td

0.0

200

-0.2

190 -0.3

180

-0.4

170

Td (°C)

absorbance

-0.1

160

-0.5

150

2

y=-1.344x+217 R = 0.987

-0.6 0

10

20

30

40

50

140

Time (hours) Fig. 10. Changes in the onset decomposition temperature (Td ) of a NC film during photooxidation and correlation with the decrease in absorbance at 1650–1280–840 cm−1 (O NO2 groups).

We used an empirical model to describe the kinetics of the processes. The results indicated that the rate determining steps in the degradation mechanism under UV irradiation are zero or firstorder with a degradation rate k [32,33]. Typically, a zero order reaction is the consequence of a primary photochemical reaction (as highlighted for the ONO2 groups), while first-order kinetics are generally the result of photodegradation processes with no diffusion-limited oxidation [34]. The results given in Table 5 indicate that only the disappearance of the ONO2 groups followed zero order kinetics. These results highlight the differences in reactivity of these species and confirm the diverse routes in the degradation mechanism proposed in Scheme 1. The various kinetic parameters that were evaluated in an attempt to establish a correlation between the modification of the chemical structure and the mechanical properties of the material exhibited first-order kinetics with similar k values (the average k value was 3.6 × 10−2 h−1 ). The half-life time was calculated from the kinetics, and the average t1/2 value was 20 h. The good agreement between the degradation rate and the k-values for the various parameters confirms the proposed cross-correlation. 4. Conclusion The photooxidation of cellulose nitrate leads to chemical modifications, which were determined in this study to comprise de-nitration and the formation of oxidation products, such as gluconolactone and anhydrides. Based on the photoproducts detected in the film and in the gas phase of cellulose nitrate (NC) and diethylcellulose (EC), a mechanism of photooxidation of NC was proposed that accounts for macromolecular chain scissions. By comparing the kinetic rates of the chemical modifications with those of the mechanical properties and the thermal stability, a cross-correlation was shown.

[1] L. Li, M. Frey, Preparation and characterization of cellulose nitrate-acetate mixed ester fibers, Polymer 51 (2010) 3774–3783. [2] H.N. Volltrauer, A. Fontijn, Low-temperature pyrolysis studies by chemiluminescence techniques real-time nitrocellulose and PBX 9404 decomposition, Combust. Flame 41 (1981) 313–324. [3] S.M. Pourmortazavi, S.G. Hosseini, M. Rahimi-Nasrabadi, S.S. Hajimirsadeghi, H. Momenian, Effect of nitrate content on thermal decomposition of nitrocellulose, J. Hazard. Mater. 162 (2009) 1141–1144. [4] P.S. Makashir, R.R. Mahajan, J.P. Agrawal, Studies on kinetics and mechanism of initial thermal decomposition of nitrocellulose: isothermal and non-isothermal techniques, J. Therm. Anal. 45 (1995) 501–509. [5] J. Rychly, A. Lattuati-Derieux, L. Matisova-Rychla, K. Csomorova, I. Janigova, B. Lavedrine, Degradation of aged nitrocellulose investigated by thermal analysis and chemiluminescence, J. Therm. Anal. Calorim. 107 (2012) 1267–1276. [6] T.B. Brill, P.E. Gongwer, Thermal decomposition of energetic materials 69. Analysis of the kinetics of nitrocellulose at 50 ◦ C–500 ◦ C, Propell. Explos. Pyrotech. 22 (1997) 38–44. [7] A.L. Kon’kin, B.G. Ershov, Y.M. Kargin, A.A. Chichirov, M.N. Agafonov, Study of the radical products of the thermal decomposition of nitrocellulose, Bull. Acad. Sci. USSR Div. Chem. Sci. 38 (11) (1989) 2426–2428. [8] D.N.-S. Hon, T.L. Gui, Photodegradation of cellulose nitrate, Polym. Photochem. 7 (1986) 299–310. [9] M. Moniruzzaman, J.M. Bellerby, N. Mai, The effect of light on the viscosity and molecular mass of nitrocellulose, Polym. Degrad. Stab. 96 (2011) 929–935. [10] S. Cakic, L. Raskovic, C. Lacnjevac, M. Rajkovic, M. Barac, M. Stojanovic, Physical–mechanical properties of nitrodopes affected by ultra-violet radiation, Sensors 7 (2007) 2139–2156. [11] G. El-Diwani, N.N. El-Ibiari, S.I. Hawash, Treatment of hazardous wastewater contaminated by nitrocellulose, J. Hazard. Mater. 167 (2009) 830–834. [12] A. Quye, D. LittleJohn, R.A. Pethrick, R.A. Stewart, Investigation of inherent degradation in cellulose nitrate museum artefacts, Polym. Degrad. Stab. 96 (2011) 1369–1376. [13] T.K. Wu, Carbon-13 and proton nuclear magnetic resonance studies of cellulose nitrates, Macromolecules 13 (1980) 74–79. [14] Y. Tezuka, K. Imai, M. Oshima, T. Chiba, Determination of substituent distribution in cellulose ethers by means of 13 C NMR study on their acetylated derivatives. 1. Methylcellulose, Macromolecules 20 (1987) 2413–2418. [15] J.L. Philippart, C. Sinturel, J.L. Gardette, Polym. Degrad. Stab. 58 (1997) 261–268. [16] P. Musto, Two-dimensional FTIR spectroscopy studies on the thermal-oxidative degradation of epoxy and epoxy-bis(maleimide) networks, Macromolecules 36 (2003) 3210–3221. [17] C. Wilhelm, J.-L. Gardette, Infrared identification of carboxylic acids formed in polymer photooxidation, J. Appl. Polym. Sci. 51 (8) (1994) 1411–1420. [18] B. Mailhot, J.-L. Gardette, Polystyrene photooxidation. 1. Identification of the IR-absorbing photoproducts formed at short and long wavelengths, Macromolecules 25 (1992) 4119–4126. [19] S. Collin, P.O. Bussière, S. Thérias, J.M. Lambert, J. Perdereau, J.-L. Gardette, Physicochemical and mechanical impacts of photo-ageing on bisphenol a polycarbonate, Polym. Degrad. Stab. 97 (2012) 2284–2293. [20] G. Socrates, Infrared and Raman Characteristic Group Frequencies, Wiley, Chichester, 2001. [21] M. Avram, Gh.D. Mateescu, Spectroscopie Infrarouge: Applications en Chimie Organique, Dunod, Paris, 1970. [22] V.I. Kovalenko, R.M. Mukhamadeeva, L.N. Maklakova, N.G. Gustova, Interpretation of IR spectrum and structure of cellulose nitrate, J. Struct. Chem. 34 (4) (1993) 540–547. [23] V. Suthar, A. Pratap, H. Raval, Studies on poly (hydroxyl alkanoates)/(ethylcellulose) blends, Bull. Mater. Sci. 23 (3) (2000) 215–219. [24] K. Xiang, X. Wang, G. Huang, J. Zheng, J. Huang, G. Li, Thermal ageing behavior of styrene-butadiene random copolymer: a study on the ageing mechanism and relaxation properties, Polym. Degrad. Stab. 97 (9) (2012) 1704–1715. [25] D. Fromageot, N. Pichon, O. Peyron, J. Lemaire, Thermal yellowing sensitized by pre-photo-oxidation of non-deacidified paper, Polym. Degrad. Stab. 91 (2006) 347–357.

S. Berthumeyrie et al. / Journal of Hazardous Materials 272 (2014) 137–147 [26] M. Yamaguchi, Y. Igarashi, R.S. Reddy, T. Shiraishi, M. Hirama, Asymmetric Michael addition of nitroalkanes to prochiral acceptors catalyzed by proline rubidium salts, Tetrahedron 53 (32) (1997)9 11223–11236. [27] S. Morlat, J.L. Gardette, Phototransformation of water-soluble polymers, Part I: Photooxidation of poly(ethylene oxide) in solid state, Polymer 42 (14) (2001) 6071–6079. [28] P. Delprat, X. Duteurtre, J.L. Gardette, Polym. Degrad. Stab. 50 (1995) 1–12. [29] N. Marheineke, Nitrocellulose Measured by GPC, Nitrocellulose Workshop, Cranfield University, 2001. [30] B. Fayolle, E. Richaud, X. Colin, J. Verdu, Review: degradation induces embrittlement in semi-crystalline polymers having their amorphous phase in rubbery state, J. Mater. Sci. 43 (2008) 6999–7012.

147

[31] T.A. Dankovich, Y.-L. Hsieh, Surface modification of cellulose with plant triglycerides for hydrophobicity, Cellulose 14 (2007) 469–480. [32] N. Rose, M. Le Bras, S. Bourbigot, R. Delobel, Thermal oxidative degradation of epoxy resins: evaluation of their heat resistance using invariant kinetic parameters, Polym. Degrad. Stab. 45 (1994) 387–397. [33] P. Musto, G. Ragosta, M. Abbate, G. Scarinzi, Photo-oxidation of high performance epoxy networks: correlation between the molecular mechanisms of degradation and the viscoelastic and mechanical response, Macromolecules 41 (2008) 5729–5743. [34] J.F. Larché, P.O. Bussière, S.S. Therias, J.L. Gardette, Photooxidation of polymers: relating material properties to chemical changes, Polym. Degrad. Stab. 97 (2012) 25–34. [35] V. Suthar, A. Pratap, H. Raval, Studies on poly(hydroxyl alkanoates)/(ethylcellulose) blends, Bull. Mater. Sci. 23 (3) (2000) 215–219.