Durable and high-efficiency casein-derived phosphorus-nitrogen-rich flame retardants for cotton fabrics

A casein derivative (CADP) was synthesized using casein, which is bifunctional containing both –P=O(O−NH4+)2 reactive groups and –P(=O)–O–C– groups, and the durable flame-retardant cotton fabrics were successfully prepared by CADP. The –P=O(O−NH4+)2 reactive groups allowed CADP to be firmly grafted onto cellulose. The –P(=O)–O–C– groups made flame-retardant cotton fabrics more resistant to soaping and improved its durability. The modification by 40% CADP increased the limited oxygen index value of cotton fabric from 17.4 to 41.6%, which maintained at 26.4% after 50 cycles of home machine washes. The results of TG, TG-FTIR, FTIR and SEM indicated that CADP increased the condensed components and decreased the flammable gaseous compounds, resulting the positive effect on char formation of cellulose. The whiteness, softness and tensile strength of cotton fabrics were retained well after modification, and the treated cotton fabrics didn’t have skin irritation.


Introduction
Among all nature textile, cotton fabric is generally applied to clothing, furnishing, automobile, industry and other fields, because of the advantages of good wearability, comfort and dyeability, etc. Pei et al. 2021). However, cotton fabric is highly flammable, which poses a threat to human beings and limits its applications (Xie et al. 2013;Feng et al. 2017). So, the flame-retardant modification is necessary for cotton fabrics.
Over the past years, various flame retardants have been developed to modify the flammability of cotton fabrics, including halogen-containing, phosphoruscontaining (Liu et al. 2012), nitrogen-containing (Xu et al. 2018),sulfur-containing (Xu et al. 2017), boriccontaining (Luo et al. 2020), metal hydroxides , intumescent flame retardants , Si-containing Zhang et al. 2020) and so on. Among them, the most commonly studied flame retardants are phosphorus-containing or phosphorus-nitrogen-containing flame retardants, because they are low toxicity, high efficiency and less smoke produced in the combustion process . At present, Pyrovatex CPÒ and ProbanÒ are widely used as commercial products for the modification of cotton, because they are very efficient and durable (Poon and Kan 2015;Mohamed and Abdel-Mohdy 2006;Yang et al. 2012). However, they have the problem of formaldehyde release (Yasin et al. 2016;Horrocks 2011). Up to now, the flame retardants for cotton fabrics can not meet the requirements of halogen-free, formaldehyde-free, efficient and durable.
As the shortage of oil becoming a global issue, the application of renewable resources has been paid more and more attention. It was considered a promising strategic approach to use different biomass materials to improve the flame retardant property of cotton fabrics, such as nucleic acids (DNAs), plant extracts and proteins (Salmeia et al. 2016;Basak and Ali 2016;Malucelli et al. 2014). By layer-by-layer method or exhaust method, some plant extracts (banana pseudostem and spinach juice) have been used as flame retardants for cotton, because they contain small amounts of phosphorus and metal oxides. These plant extracts could increase the limiting oxygen index (LOI) values of cotton fabrics to 30% (Basak et al. 2015a, b;Basak et al. 2015a, b). DNA contains phosphate groups, deoxyribose units and nitrogencontaining bases, which has been investigated as a natural intumescent flame retardant for textiles (Isarov et al. 2016;Alongi et al. 2014aAlongi et al. , 2013aAlongi et al. , 2015Carosio et al. 2013). It was found that DNA increased the LOI values of cotton fabrics to 28% by layer-by-layer (Alongi et al. 2013b;Bosco et al. 2015). DNA can form phosphoric acid to have the positive effect on char formation and inhibit the continuous combustion of cellulose, but DNA can not be applied on a large scale. Some P-, N-and S-rich proteins like caseins, whey proteins and hydrophobins were also used as ''green'' flame retardants for textiles (Bosco et al. 2013;Alongi et al. 2014b;Carosio et al. 2014;Faheem et al. 2019). By deposition method, these proteins could be homogeneously coated on cotton fabrics, the burning rate of treated fabric was decreased. These biomass materials are widely existing in the nature of various plants and animals, with safe, non-toxic and environmental-friendly characteristics. However, the cotton fabrics treated with these biomass materials have poor flame retardant property and durability, and they can not meet the commercial requirements.
In our previous work, -P = O(O -NH 4 ? ) 2 reactive groups have been introduced into amino acids and proteolytic amino acids, to develop efficient and durable bio-based flame retardants (Xu et al. 2019a(Xu et al. , 2019c. From the results, -P=O(O -NH 4 ? ) 2 reactive groups reacted with the -OH of cellulose to generate P(= O)-O-C groups, and increased the binding between the flame retardants and cellulose Xu et al. 2019b;Jia et al. 2017;Chen et al. 2021). These bio-based P-and N-rich flame retardants could significantly increase the LOI values of cotton fabrics to 40%. Because the covalent bonds between flame retardants and cellulose, the flame retardancy of treated cotton fabrics was durable, which could withstand 50 laundering cycles (LCs) of careful hand washes. However, the flame retardant property was greatly reduced after home machine washes. The unreacted -P=O(O -NH 4 ? ) 2 groups on treated cotton fabrics would react with the metal ions during washing process, leading to a considerable reduction of the flame retardant property.
In this current study, a casein derivative (CADP) was synthesized using casein, which is bifunctional containing both -P=O(O -NH 4 ? ) 2 reactive groups and -P(=O)-O-C-groups. And the durable flame-retardant cotton fabrics were successfully prepared. On the one hand, the reaction between -P=O(O -NH 4 ? ) 2 groups and cellulose during modification can make CADP graft onto the cellulose. On the other hand, the -P(= O)-O-C-groups could reduce the effect of soaping solution on the flame retardant property of cotton fabrics and improved the durability of cotton fabrics treated with flame retardants containing -P=O(O -NH 4 ? ) 2 groups. The flame retardant property, flame retardant mechanisms, thermal decomposition, whiteness, tensile properties and skin irritation of cotton fabrics were studied.

Preparation of flame retardants
First, 22.00 g of casein, 11.00 g (0.1 mol) of dimethyl phosphite and 30 mL of hydrochloric acid were mixed in distilled water (150 mL) in a flask, and the mixture was stirred and heated to reflux. Then, formaldehyde (0.1 mol, 8.11 g) was added into the mixture dropwise, and refluxed for 4 h. Afterwards, H 3 PO 3 (0.12 mol, 9.84 g) was added into the mixture, and formaldehyde (0.12 mol, 9.73 g) was added dropwise, then the mixture was reflux for another 3 h. The pH of the mixture was adjusted to neutral with ammonia water after cooling to room temperature, and then concentrated with rotary evaporator. Finally, the solid product was washed with isopropyl alcohol for 3 times, dried at 60°C, the novel flame retardants, CADP, was obtained with 83.23% yield. The synthesis route is shown in Fig. 1.

Pretreatment of cotton fabrics
Before the modification with CADP, cotton fabrics were treated with sodium hydroxide. Cotton fabrics were soaked in sodium hydroxide solutions (20% w/w) for 5 min. Then the cotton fabrics were washed with distilled water to neutral and dried at 60°C to prepare pretreated cotton (D-cellulose).
Flame-retardant treatment of cotton fabrics Different concentrations of CADP solutions were prepared, and 5% dicyandiamide were dissolved in these solutions as catalyst. Then D-cellulose was soaked in these CADP solutions with a bath ration of 1:20 for 10 min at 90°C. After soaking, cotton fabrics were rolled to achieve a liquid rate of 120%. The cotton fabrics were cured at 185°C in a baking machine for 5 min, and the flame-retardant cotton fabrics (CADP-cotton) were prepared after washed and dried. The reaction between during treatment is shown in Fig. 2.

Characterization
Fourier transform infrared (FTIR) of CADP and cotton fabrics were recorded on a Spectrum GX spectrometer Thermogravimetric (TG) analysis was carried out on a Pyris 1 thermogravimetric analyzer (Perkin-Elmer) in either nitrogen or air atmosphere. The samples were heated from 40 to 700°C with a speed of 10°C/min and a flow rate of 60 mL/min.
The add-ons were calculated using the following formula: where W 0 are the weight of the pure cotton fabrics, W 1 are the weights of CADP-cotton. Limiting oxygen index (LOI) was tested on a M606B digital display oxygen index tester instrument (Qingdao Shanfang Instrument Co. Ltd., Qingdao, China) according to ASTM D2863-2000.
The vertical burning tests were measured on a YG815B vertical fabric flame-retardant tester (Nantong Sansi Electromechanical Science & Technology Co. Ltd., Qidong, China) according to ASTM D6413-99.
The durability of CADP-cotton against gentle hand washes and home machine washes was measured according to AATCC 61-2006: test No. 1A and No. 2A, using a color fastness tester (Roaches International Ltd., Birstall, UK).In test No. 1A, solution of WOB (0.37% w/w) and 10 steel balls were added into the stainless steel canisters, and fabrics were washed for 45 min at 40°C, which was considered to be 5 laundering cycles (LCs) of gentle hand washes. In test No. 2A, solution of WOB standard reference detergent (0.15% w/w) and 50 steel balls were added into the stainless steel canisters, and fabrics were washed for 45 min at 49°C, which was considered to be 5 LCs of home machine washes.
Cone calorimeter tests of cotton fabrics (100 mm 9 100 mm) were carried out by a cone calorimeters (FTT 0007, UK) according to ASTM E 1354 standard under 35 kW/m 2 of irradiative heat flux.
TG-FTIR was analysis on a Nicolet 6700 FTIR spectrometer (Thermo Fisher Scientific., Waltham, MA, US) and a Pyris 1 TG analyzer. The samples were heated from 40to 700°Cat a speed of 10°C/min and a flow rate of 60 ml/min in a nitrogen atmosphere. FTIR was recorded at a range from 400 to 4000 cm -1 with a resolution of 4 cm -1 .
Scanning electron microscopy (SEM) were carried out on a Phenom ProX desktop scanning electron microscope (Phenom-World BV., Eindhoven, Netherlands), and the acceleration voltage was 20 kV. And Fig. 2 The reaction between CADP and D-cellulose the elemental compositions were measured by energydispersive X-ray spectroscopy (EDS).
X-Photoelectron spectroscopy (XPS) analysis was tested using an ESCALab220i-XL electron spectrometer (PerkinElmer, USA) using 300 W Al Ka irradiation.
The tensile tests were performed by an electromechanical universal testing machine (MTS Systema Corp., Eden Prairie, MN, USA) according to ASTM 5035-2006.
The softness of cotton fabrics were tested by a YG (B) 022D fabric stiffness tester (China) according to ASTM D 1388ASTM D -96 (2002 standard test method.
The skin irritations of cotton fabrics were carried out on 3D skin model by reconstruct the in vitro test method of human epidermis model. Cotton fabrics were put into the sterilization box and sterilized at 121°C for 20 min. Then the extract was immersed in DMEM medium containing 10% serum at the ratio of 6 cm 2 /mL and extracted at 37°C for 24 h after filtration. After that, 25 lL extract drops were absorbed and added to the tissue surface of the 3D epidermal model, and three epidermal models were repeated in each group. After 30 min at room temperature, the models were washed with sterile DPBS buffer one by one. Then, the models were cultured in the incubator for 42 h, and the relative tissue viability of each sample was determined by MTT tissue viability test.

FTIR and TG analysis of CADP
The FTIR spectrum of CADP is shown in Fig. 3a. From the result, CADP showed characteristic peaks around 3425 cm -1 attributable to the N-H bond. Intense peaks appeared around 3192 and 1401 cm -1 corresponding to stretching and scissoring vibration of NH 4 ? . The peak at 1702 cm -1 was assigned to the stretching vibration of C=O from -COOH. Peaks at 1667 and 1543 cm -1 were belonged to the stretching vibration of amide I and amide II. Peaks at 1225 and 990 cm -1 were assigned to the absorptions of P=O and -P(= O)-O-C groups. In addition, the peak at 1069 cm -1 corresponded to stretching vibration absorption of C-N. The results suggested that CADP was synthesized.
TG analysis of CADP was carried out in nitrogen, and the curves are illustrated in Fig. 3b. There were three main stages in the pyrolysis of CADP. The first stage was from 100 to 250°C, which caused by the decomposition of CADP to release NH 3 , and the phosphonic acids were formed (Liu et al. 2018). In the second stage, CADP continued to decompose to form phosphonic acids when the temperature exceeded 250°C. In the third stage (400-550°C), the further dehydration and cross-linking of the phosphonic acids were occurred to form polyphosphoric acid, metaphosphate and char, etc.

Flame retardancy and durability
Flame retardant property and durability of cotton fabrics were tested by LOI measurement, vertical burning and cone calorimeter tests. Figure 4 shows the LOI results of control cotton fabric, CAPD-cotton, and CADP-cotton after gentle hand washes and home machine washes. The LOI value of control cotton fabric was only 17.4%, making them highly flammable. After treated with CADP, the flame retardant property of cotton fabrics was remarkably enhanced, the LOI values improved with the increase of CADP concentrations. The LOI value of cotton fabric increased to 31.7% when the concentration of CADP was 10%, and it reached 41.6% when the concentration of CADP increased to 40%. After launder cycles (LCs), the LOI values of CADP-cotton decreased. And the more launder cycles and more intense of washing conditions, the more significant LOI value decreased. However, flame retardant property of treated cotton fabrics still maintained very well. The LOI value of 40% CADP-cotton remained at 30.2% and 26.4% after 50 LCs of gentle hand washes and home machine washes, respectively. After washing process, 40% CADP-cotton still reached the flame retardant standard LOI value of cotton fabric ([ 26%) (Abou-Okeil et al. 2013).
The curves and data from cone calorimeter test of CADP-cotton and CADP-cotton after 50 LCs of home machine washes are shown in Fig. 4 (b)-(d) and Table 1. From Fig. 4b, c, CADP exhibited certain effect in suppressing the heat release of cotton fabrics. The peak of heat release rate (pHRR) of CADP-cotton decreased significantly from 295.95 to 52.81 kW/m 2 . And the pHRR of CADP-cotton after 50 LCs of home machine washes slightly increased to 187.02 kW/m 2 . The total heat release (THR) decreased from 3.4 MJ/ m 2 of control cotton fabric to 2.3 MJ/m 2 of CADPcotton. Additionally, the total smoke productions (TSP) of CADP-cotton and CADP-cotton after 50 LCs of home machine washes were higher than that of control cotton fabric, demonstrating the incomplete combustion of treated cotton fabrics, and thus more smoke is released during combustion. (Ye et al. 2021;Yang et al. 2021) All these results confirm that CADP could suppress the heat release and complete combustion of cotton fabric, and improve the flame retradancy of cotton fabrics. Figure 5 shows the images of control and treated cotton fabrics during vertical burning tests, the corresponding results of vertical burning tests are shown in Table 1. During vertical burning tests, control cotton fabric was quickly ignited and burned vigorously until the sample burned to ashes. The afterflame and after-glow times were 6 and 5 s, respectively. By contrast, 40% CADP-cotton did not ignite during the test, and 69 mm of intact char was left in the ignition area. And 40% CADP-cotton after 50 LCs of home machine washes also did not ignite in air. No after-flame and after-glow were observed, and 93 mm of char was formed in the ignition area.
The results of LOI, vertical burning and cone calorimeter tests indicated that CADP could significantly improve the flame retardant property of cotton fabrics. By forming covalent bond with cellulose, CADP could firmly grafted to cellulose, and the cotton fabrics with high and durable flame retardant property were prepared, which still well preserved after 50 LCs of home machine washes. In addition, these results also verified the efficiency of CADP in hindering combustion, and exhibited promotion effect in the carbonization of cotton fabric. During combustion, CADP could form phosphonic acids to promote the dehydration of cellulose, and the char layer prevented the transfer of heat and mass between gas and condensed phases, enhance the flame retardancy of cotton fabrics (Jia et al. 2017).

Characterization of cotton fabrics
The effects of CADP on the structure of cotton fabric and laundering cycles on the structure of CADP-cotton were studied by FTIR, XPS and SEM-EDS, Fig. 6 shows the FTIR and XPS spectra. For control cotton fabric, the peak presented at 3375 cm -1 was allocated to stretching vibration of O-H, peaks at 2905 and 1366 cm -1 were identified as C-H stretching vibration and deformation vibration, peaks shown at 1162, 1111 and 1060 cm -1 were associated with the adsorption band of C-O-C. It was observed that some new peaks presented in CADP-cotton. The peaks at 3169 and 1401 cm -1 were identified as NH 4 ? stretching and scissoring vibration, which belonged to the unreacted -P=O(O -NH 4 ? ) 2 groups on CADP-cotton. The peaks at 1225 and 991 cm -1 were associated to    of CADP-cotton before washing. However, the peaks of NH 4 ? (3169 and 1401 cm -1 ) were clearly weakened, suggesting the reduction of the unreacted -P=O(O -NH 4 ? ) 2 reactive groups after washing process.
From Fig. 6b, c, the P and N were not found in control cotton fabrics, which detected in CADP-cotton and CADP-cotton after 50 LCs of home machine washes. These results indicated that CADP was still grafted on cellulose after 50 LCs of home machine washes, agreeing well with the results of FTIR. From Fig. 6c, d, CADP-cotton after 50 LCs of home machine washes had a chemical composition similar to that of CADP-cotton, except the peak intensities of NH 4 ? and Na ? . After 50 LCs of home machine washes, the NH 4 ? on CADP-cotton reduced, and the Na ? increased, which suggested that the unreacted -P=O(O -NH 4 ? ) 2 reactive groups reacted with the Na ? in the washing liquid to form -P=O(O -Na ? ) 2 during washing. Figure 7 shows the SEM images, and the chemical compositions (data from XPS and EDS) of cotton fabrics are listed in Table 2. Cotton fibers (Fig. 7a, b) were slightly crimped with some grooves on the surface, free of any other materials, and the fibers consisted of 55.69% C and 44.31% O. Compared with control cotton fabric, the surface morphology and chemical compositions of CADP-cotton were changed. After the treatment with CADP, the fibers were slightly swollen, and a large amount of substances were observed (Fig. 7c, d). Besides, 2.73% P and 12.39 of N were found in CADP-cotton. After 50 LCs of home machine washes, the fibers were still covered with a large amount of flame retardants (Fig. 7e, f), indicating that CADP was indeed firmly grafted onto fibers. During the modification, -P=O(O -NH 4 ? ) 2 reactive groups of CADP reacted with the C6 -OH of cellulose, then the formation of -P(= O)-O-C bonds allowed CADP to be firmly grafted on cellulose. However, 0.86% Na appeared in CADP-cotton after home machine washes, while the content of N decreased to 10.81%, content of P remained at 2.73%.
The results of FTIR, XPS and SEM-EDS indicated that CADP was firmly grafted onto cellulose. But there were some unreacted -P=O(O -NH 4 ? ) 2 groups on CADP-cotton, which reacted with the Na ? in the washing liquid to form -P=O(O -Na ? ) 2 during washing. During combustion, -P=O(O -NH 4 ? ) 2 could form phosphonic acids to promote the dehydration of cellulose, while -P=O(O -Na ? ) 2 could not. The more unreacted -P=O(O -NH 4 ? ) 2 on treated cotton fabrics, the more -P=O(O -Na ? ) 2 were formed during washing, and the more significant the flame retardant property of treated cotton fabrics was reduced. In this ? ) 2 on CADPcotton, so CADP-cotton had excellent durability and could withstand up to 50 LCs of home machine washes. Figure 8 shows TG and DTG curves of control cotton fabric, CADP-cotton and CADP-cotton after washing, and the corresponding data is also marked in Fig. 8. Figure 8a, b show the TG and DTG curves of cotton fabrics in nitrogen atmosphere. For control cotton fabric, the dehydration reaction of cellulose was from 40 to 334°C. Then, the main thermal degradation of cellulose was from 334 to 400°C, which corresponds to the formation of flammable gases, solid residues and volatile liquids, with a maximum mass loss rate of 2.52%/°C at 362°C. When the temperature reached 600°C, only 9.25% of char yield was remained. After the modification with CADP, the charring capacity of CADP-cotton was improved significantly, and the pyrolysis process of CADP-cotton was earlier than that of control cotton fabric. CADP-cotton suffered a Fig. 6 a FTIR spectra of control cotton fabric, 40% CADP-cotton and 40% CADP-cotton after 50 LCs; b P 2p, c N 1 s and d Na 1 s spectra of control cotton fabric, 40% CADP-cotton and 40% CADP-cotton after 50 LCs significant mass loss at 239°C, because the flame retardants on cotton fabrics decomposed to release NH 3 , phosphorus acid and polyphosphoric acid from 100 to 250°C. Then, the phosphorus acid and polyphosphoric acid promoted cellulose dehydration to form char instead of depolymerization, resulting much more char (42.42%) generated at 600°C. After 50 LCs of home machine washes, some unreacted -P=O(O -NH 4 ? ) 2 groups on CADP-cotton reacted with Na ? in the washing liquid to form -P=O(O -Na ? ) 2 during washing, so the phosphorus acid and/or polyphosphoric acid from CADP-cotton decreased. The pyrolysis process of CADP-cotton after washing was later than that of CADP-cotton, and the maximum mass loss rate (V max ) decreased. However, the charring capacity of CADP-cotton after washing still maintained very well.

TG analysis
The results of TG and DTG analysis in air were similar to these in nitrogen. The depolymerization of control cotton fabric occurred from 320 to 360°C, and the further oxidation of the residue occurred from 450 to 530°C in the presence of oxygen. By contrast, CADP promoted cellulose dehydration to form char before the depolymerization. The charring capacity of CADP-cotton before and after washing was improved significantly.

TG-FTIR analysis
The TG-FTIR coupling technique was carried out to identify the volatile gaseous products of control cotton fabric, CADP-cotton and CADP-cotton after washing from thermal degradation. The FIIR spectra of the gaseous products from cotton fabrics at different temperature are shown in Fig. 9. From Fig. 9a, control cotton fabric depolymerized to release a large amount of volatiles from 340 to 360°C. The peaks at 2325 cm -1 and 3554 cm -1 were assigned to the vibration of two non-flammable gaseous: CO 2 and water vapor, respectively. In addition, some flammable gaseous were observed in the volatile gaseous from control cotton fabric: the peaks at 2970 and 2910 cm -1 were allocated the hydrocarbons, the peaks at 1763 and 1064 cm -1 belonged to carbonyl compounds and ethers, respectively. After the modification by CADP, flammable gaseous products were significantly reduced. During thermal degradation, the volatile gaseous products of CADP-cotton were water vapor and CO 2 , and a small amount of hydrocarbons and carbonyl compounds, because CADP could promote cellulose dehydration to form char instead of releasing flammable gaseous. These results also validated the catalytic effect of CADP on cellulose. As shown in Fig. 9c, the flammable gaseous products released from CADPcotton after 50 LCs of home machine washes slightly increased.

Analysises of char residues
To further explore the flame retardant mechanism of CADP in condensed phase, the chemical structure, surface topograhpy and elements composition of char residue from CADP-cotton were studied, and the results is given in Fig. 10. The FTIR spectrum of char residues is showing in Fig. 10 1. After combustion, the char exhibited some P and N containing groups as revealed in the spectra. The peaks at 1561 cm -1 (C-N stretching vibration), 1233 to 1172 cm -1 (P=N and P=O), 1081 cm -1 (P-N) (Jin et al. 2021), 887 cm -1 (P-O-P) and 887 cm -1 (O=P-O) (Zhan et al. 2021) were found. After burning, CADP-cotton formed a large amount of char (Fig. 10b), which still retained the shape of the fibers, indicating the high efficiency of CADP in catalytic char formation of cellulose. Besides, some bubbles were observed on the surface of the char, and there were 4.48% P and 5.84% N in the char of CADP-cotton.
Based on the discussion above, the flame retardant mechanism of CADP could be proposed. During combustion, phosphorus-based acids could be released from CADP, which reacted with cellulose by dehydration and esterification, enhancing the char formation. The char layer acted as a barrier and protected the Fig. 9 FTIR spectra of the volatile gases from the thermal degradation of (a) control cotton fabric, (b) CADP-cotton and (c) CADPcotton after 50 LCs of home machine washes at different temperatures cotton fabric from the attack of radiant heat and oxygen. In addition, the N was conducive to the formation of polymeric species containing P-N bonds, which could phosphorylate cellulose more effectively than P-O bonds (Rosace et al. 2018). In summary, CADP dewatered cellulose to form a char layer preventing the further combustion of cellulose, indicating that CADP mainly played an important role in condensed phase. Figure 11 shows the whiteness, bending length and tensile strength results of cotton fabrics. During the modification, -P=O(OH) groups released from CADP could react with -OH of cellulose, which reduced the whiteness and tensile strength of cotton fabrics. From the results, the whiteness and tensile strength of cotton fabrics slightly decreased after modification with CADP but they still retained well. In this study, -P(= O)-O-C were introduced into casein with dimethyl phosphite, and then -P=O(O -NH 4 ? ) 2 were introduced to form CADP, making CADP more neutral to retain better tensile strength and whiteness for cotton fabric. In addition, the bending lengths of cotton fabrics maintained very well, indicating that CADP has little effect on the softness of cotton fabric.  The 3D skin model was used to investigate the skin irritations of control cotton fabric and 40% CADPcotton, and the results are presented in Table 3. As shown in Table 3, the extracts of control cotton fabric and CADP-cotton had little influence on the tissue activity of the epidermal model. The relative tissue activities of control cotton fabric and CADP-cotton were 92.71% and 81.88%, both of which were more than 50%, indicating that neither control cotton fabric nor CADP-cotton had skin irritation.

Conclusions
In this study, both the -P=O(O -NH 4 ? ) 2 reactive groups and -P(=O)-O-C groups were introduced into casein molecule to synthesize the novel P-N-containing flame retardants, CADP, for cotton fabrics. By the reaction between -P=O(O -NH 4 ? ) 2 groups and -OH of cellulose during modification, CADP was firmly grafted onto cellulose, and -P(= O)-O-C-groups improved the durability of flame-retardant cotton fabrics, which could reduce the effect of soaping solution on the flame retardant property of cotton fabrics. After modification with CADP, the highefficiency and durable flame-retardant cotton fabrics were prepared. The LOI value of CADP-cotton fabric could reach 41.6%, which maintained at 26.4% after 50 times of home machine washes. During combustion, CADP mainly played an important role of condensed phase flame retardant to promote the dehydration of cellulose. In a word, bio-based and highly effective CADP can serve as an excellent substitute for Pyrovatex CP and Proban.