In-situ phosphine oxide physical networks: A facile strategy to achieve durable flame retardant and antimicrobial treatments of cellulose

A facile methodology to durably modify cellulose fibers by forming in-situ phosphine oxide macromolecular physical networks is developed in this work. Such networks were formed by treating the cellulose fibers with a mixture of Michael adducts [trivinyl phosphine oxide (TVPO) and cyclic amines (piperazine and 2,4,6-tri(piper- azine-1-yl)-1,3,5-triazine)] followed by a click crosslinking reaction initiated by dry heat, microwave, or steaming process. The in-situ phosphine oxide network is integrated within the cellulose fibers to produce a durable treatment against laundering. NMR and SEM analysis of the treated fibers verified the in-situ formation of phosphine oxide macromolecules in the bulk of cellulose fiber. The treatment was homogenous and achieved an outstanding phosphorus (P) retention ( > 95%) after even 50 laundry cycles. The treated cotton exhibited excellent fire retardant properties with a limiting oxygen index value > 27% and passed the Swiss vertical flammability test at 2 wt% P. Analysis of evolved gasses during thermal decomposition of treated cellulose supported both the condensed and gas phases mechanism. The flame retardant treatment was further adapted to include antimicrobial properties by in-situ formation of silver nanoparticles within the cellulose fibers using a single-step process to demonstrate the versatility of this novel treatment methodology. The treated fabric exhibited excellent antimicrobial as well as flame retardant properties and was also durable to 50 laundry cycles. Contact angle measurements and moisture management test confirmed the excellent comfort behavior of the treated cellulose fabric.


Introduction
There is a growing demand to develop protective textiles for emergency workers, technical, security, and military personnel, to reduce exposure to biological, chemical, and fire threats [1][2][3]. Increased use of polymeric materials in building construction [3,4] and climate change [5], has led to an increase in fire incidents, which has resulted in the catastrophic wildfire, civilian and deaths, injuries, and property damage. Likewise, the spread of infectious diseases has increased over the past decades, which has great social and economic impacts [6]. Thus, novel multifunctional materials that can aid in minimizing these threats are of great importance to today's society.
Textiles are considered the first line of personal defense against the above-mentioned threats [7]. To address these challenges, innovations in textile construction and material components are of paramount importance. Cellulose is the most abundant polymer globally, and cellulose-based fibers still constitute more than 30% of the world's total fiber consumption. As renewable and biodegradable materials, they are considered sustainable compared to most engineered fibers [8]. However, like most organic materials, they burn easily and require protection against fire hazards. Commercially, cellulose textiles are flame retarded by incorporating flame retardant (FR) additives into their bulk during the fiber manufacturing process (suitable for regenerated cellulose fibers) or by special formaldehyde-based finishing treatments (suitable for cotton cellulose) [9][10][11]. The durability of formaldehydebased FRs is due to the formation of stable acetal linkage between the FR and the -OH group of cellulose, which can easily withstand more than 50 laundry cycles [12,13]. However, formaldehyde, being a carcinogen, requires manufacturing technologies to be heavily regulated and in some instances, banned [14]. As the majority of cellulose textiles are made from cotton, an environmentally friendly manufacturing process for flame retardation of cotton has been the topic of research over the past few decades [10,[15][16][17]. Thus, there is a need to develop formaldehyde-free processes for flame retardation of cellulose with long term durability.
In recent years, researchers have utilized new FR chemistry based on phosphorus (P), nitrogen (N), and silicon (Si) to impart FR properties to cotton cellulose with varying levels of durability to the laundry process [15][16][17]. Some of these approaches involve the use of formaldehyde as a building block for the FR synthesis [18][19][20][21][22][23] or as a crosslinker in the treatment process [24]. FR cellulose with limited durability to laundry can be achieved via the linkage of phosphorus moiety with and without nitrogen to cellulose via a phosphoester bond [19,20,[25][26][27]. In some semi-durable treatments, the phosphorus moiety is linked to cellulose via a "P-O-C bond" [28,29]. Phosphorus moiety, together with methacrylamide has also been grafted to cellulose via radical reactions resulting in treatments stable to 5 laundry cycles [30]. Thus, it is clear that most of these FR treatments on cotton cellulose are either semidurable or they require the use of formaldehyde at some stage of the manufacturing process for achieving long-term durability. Also, such treatments only address the fire safety of the materials. Whereas, in real-world applications, a multifunctional textile substrate would address multiple threats simultaneously. Especially for firefighters, medical responders, military personnel, and other frontline personnel who are exposed to numerous hazards under different situations, textiles exhibiting both flame retardancy and antimicrobial properties could be very relevant. There are limited studies in the literature which integrates both antimicrobial properties and flame retardancy of cellulose fibers in one treatment. Previously, cellulose was treated with phosphorus and quaternary ammonium-based silicon precursors via a sol-gel technique, but the long-term durability was not examined [31].
FR treatments for cellulose fibers require a weight add on of active ingredient over 30% and significantly influence the textile's comfort properties. The comfort of fabrics can be evaluated by moisture management tests to determine the ease of transport of liquid and water vapor away from the body to the atmosphere. The rapid movement of fluid in the textiles inhibits perspiration soaked next to the skin over a long period and preventing the growth of skin-irritating microorganisms. Moreover, fabrics that transport away moisture from the skin while performing vital functions keep the human body capable of cooling itself and contributing to the overall comfort of the textiles [50].
Thus, a multifunctional cellulose fiber via an in-situ methodology to obtain durable flame retardancy and antimicrobial activity in one-step (Scheme 1) was developed in this work. Cotton cellulose was modified by an in-situ Michael addition crosslinking of trivinyl phosphine oxide and different amines to achieve flame retardancy via a straightforward methodology that can be easily adapted for industrial manufacturing. A similar principle of in-situ synthesis of phosphine oxide macromolecules in various substrates has been recently published in the literature [51,52]. The formation of phosphine oxide macromolecules inside cellulose fibers was confirmed via solid-state NMR analysis. The treated fabrics were evaluated for flame retardancy using the Swiss vertical burning test (BKZ-BV) and limiting oxygen index (LOI). The thermal behavior of the treated cellulose was evaluated using thermogravimetric analysis (TGA) and microscale combustion calorimeter (MCC). The long term durability of the treatments was also assessed (i.e. 50 laundry cycles). The evolved gas analysis was performed using the thermogravimetric analysis, coupled with Fourier-transform infrared spectroscopy (TGA-FTIR) and Direct Insertion Probe-Mass Spectrometry (DIP-MS) to identify the new FR's mode of action. Antimicrobial functionality was also incorporated into the fabric via in-situ AgNP formation to demonstrate the versatility of such a method. The surface topography of the treated fabric was evaluated by Scanning Electron Microscopy (SEM) and transmission electron microscopic (TEM) analysis, which confirmed the formation of AgNPs. A combination of FR and antimicrobial  properties using in-situ AgNPs for cellulose has not been reported in the literature. The comfort properties of the treated cellulose fabrics were also evaluated.

NMR analysis
Solid-state 13 C and 31 P CP-MAS NMR spectra were measured at ambient conditions on a Bruker Avance III 400 NMR spectrometer (Bruker BioSpin AG, Fällanden, Switzerland) with a 4 mm CP-MAS probe. Detailed procedure can be found in SI-1.

Cellulose treatments
The conditioned fabric of the required dimension was taken in a petri dish (Fig. 1) and a freshly prepared equimolar aqueous solution of TVPO and piperazine (volume of the solution = twice the weight of cellulose) was applied evenly on the fabric (Fig. 1). The fabric was then allowed to stand for 15 min with occasional inversions to ensure the homogeneity of the treatment. The quantity of TVPO in the sample was calculated using Eq.1. Where P = desired phosphorus % on fabric, Z = Wt. of fabric (g), considering the molar ratio of TVPO : amine as 1:1 for triazine derivative and 2:3 for piperazine. All fabrics were then oven dried at 60 • C for 10 min, except for microwave crosslinked fabrics (dried for 7 min) to retain the minimum moisture content of 20 wt%.

Crosslinking experiments
(i) For Dry crosslinking, the treated cellulose fabrics were cured in a preheated oven at 120 ℃, 140 ℃ and 160 ℃ for 1 h as given in Table 1 (Sample D1-D3 respectively). (ii) In Microwave Crosslinking, the treated fabrics were placed in a glass tube and cured at at 120 ℃, 140 ℃ and 160 ℃ (M1-M3 respectively) for 1 h in a SynthWAVE Microwave reactor from MWS Mikrowellen Systeme GmbH, Switzerland. (iii) During Steam Crosslinking, The monomer containing fabrics were mounted on a frame present inside a pressure vessel ( Fig. SI-1) at 110℃ and operating 0.4 bar above atmospheric pressure were treated for a different duration (Sample S1-S4, Table 1). For comparison, sample S5-S8 were prepared at higher temperature and pressure (117 ℃, 0.8 bar). After the optimization of the steaming condition, sample S9-S12 was prepared with different phosphorous content. Finally, sample S14 and S15 were prepared using TPT, to study the effect of amines on fabric properties. After steaming, samples were rinsed with warm water for 10 min, drying, and conditioning for 24 h at 65% relative humidity (at 20 • C) before analysis. The details of all crosslinking methods are provided in the Sec. SI-3.

2.2.3.1.
In-situ AgNP preparation. The fabrics were treated with an aqueous TVPO-piperazine solution and dried at 60 • C for 5 min (Fig. 1b). The AgNO 3 solution was sprayed over the fabric and kept in the dark for 15 min. The required quantity of AgNO 3 in the solution was maintained to achieve 0.1 wt% (SS1) and 0.75 wt% (SS2) of Ag on the final fabric. It was then dried in an oven at 60 • C for 5 min before steaming. After steaming, the fabrics were washed, conditioned and analyzed (Table SI-4).

Long term durability test (Laundry)
The durability of the treatments was evaluated by the AATCC TM-61-1996 standard, which simulates five home laundering cycles. The details of this experiment can be found in sec. SI-4.

Elemental analysis
Phosphorus content of samples was measured using the inductively coupled plasma optical emission spectrometry method (ICP-OES) using a 5110 ICP-OES apparatus (Agilent AG, Basel, Switzerland). Sample preparation for ICP-OES consisted of mixing 200 mg of a sample with 3 mL HNO 3 , followed by digestion using a microwave reactor (Synthwave from MLS GmbH).
Energy dispersive X-Ray spectroscopy (EDX) mapping experiments were conducted on a Hitachi S-4800 Scanning Electron Microscope (SEM) operating at accelerating voltages of 20 kV. The cellulose fabrics and char were coated with Au/Pd (5 nm) before analysis.

Thermal analysis and fire tests
Thermogravimetric analysis (TGA) was carried in a NETZSCH TG209 F1 Iris instrument. ~4 mg of sample was heated from 25 to 800 • C at a rate of 10 • C/min. The measurements were performed in N 2 and air with a total gas flow of 50 mL/min.
Limiting Oxygen Index (LOI) values of fabrics were measured on the FTT oxygen index apparatus according to ASTM D2863-97 standard using 140 × 50 mm sample size.
Vertical burning tests of cellulose fabrics were performed according to the Swiss Standard (BKZ-VB). The fabrics in a vertical orientation were exposed to a flame at an angle of 45 • for 15 s to record the burn length and time. Five specimens were tested for LOI and vertical burning tests to get an average value. The videos of vertical burning tests of fabrics (blank cellulose, S9, S10, S11, S12, S14, S15, and SS1) are available as Supplementary Information (SI).

Microscale combustion calorimeter (MCC)
Heat release rates (HRR) of cotton cellulose were determined using a microscale combustion calorimeter (Fire Testing Technology Instrument, London, UK) following ASTM D7309. ~7 mg of sample was exposed to a heating rate of 1.0 • C/s from 150 to 750 • C in the pyrolysis zone.

Evolved gas analysis
The TGA instrument, coupled with the Fourier-transform infrared spectrophotometer (Bruker Tensor 27), was used to analyze the gasses evolved during pyrolysis. 10 mg of cotton cellulose was heated from 25 to 800 • C at a rate of 10 • C/min − 1 , and the FTIR spectra of the gases evolved during pyrolysis was recorded with a resolution of 4 cm − 1 ranging from 4000 to 550 cm − 1 .

Direct insertion probe mass spectrometry (DIP-MS) analysis was
performed for a 1-2 μg sample using the ThermoQuest FINNIGAN apparatus (Austin, TX, USA). The sample was heated from 30 • C to 450 • C at a rate of 50 K/min and 10 − 6 mbar pressure.

Transmission electron microscopic (TEM)
Cotton fibers (~5 mm) removed from SS1 and SS2 were placed into molds and embedded in Spurr low viscosity resin. Molds were placed in an oven at 70 • C for 48 h before sectioning in an ultramicrotome (Ultracut F4 Reichert-Jung). First, the block with embedded fibers was trimmed down to an area of interest, followed by cutting ultrathin sections of ~70 nm. Then, ultrathin sections were placed on a Formvar carbon-coated copper TEM grids to record TEM images in a JEOL JEM 2200 fs microscope operating at 200 kV.

Antibacterial assays
The antibacterial property was determined using the model pathogens: Staphylococcus aureus ATCC 6538 and Pseudomonas aeruginosa DSMZ 1117. The fabrics were sterilized for 20 min under UV (254 nm, 100 µW/cm 2 , Kojair Tech Oy, Finland). For the agar diffusion test, 200 µL of exponentially growing bacteria (concentration of 10 5 CFU/mL) was added to Müller-Hinton-Agar (Sigma Aldrich). Sterilized fabrics (5 × 10 mm) were placed on the agar and plates were incubated for 18 h at 37 • C, followed by measurement of the inhibition zone.
For the contact killing test, an exponential growing bacterial culture was prepared in Brain-Heart-Infusion (BHI, Oxoid CM1135) at 37 • C and 160 rpm. The culture was then diluted with phosphate-buffered saline (PBS buffer, Sigma-Aldrich) to 10 6 CFU/mL. 100 µL of bacterial suspension was loaded onto the samples (10 × 10 mm) and incubated at 37 • C for 2 h. The suspension was removed and the samples were washed twice with 450 µL PBS. The removed bacterial suspension and washing solution were mixed, and serial dilutions of the mixture were plated on Plate count agar (Casein-peptone Dextrose Yeast Agar, Sigma Aldrich, 70152). The plates were incubated overnight at 37 • C, followed by colony counting to obtain colony forming units (CFU). Statistical analyses were performed using unpaired and two-tailed Student's t-test.

Fabric appearance.
The change in the sample color after treatment was assessed by measuring reflectance of the treated samples using a Datacolor 550 dual-beam spectrophotometer in a wavelength range between 360 nm and 700 nm.

Fabric comfort.
To evaluate fabric comfort, moisture management properties such as water spreading speed and wettability of fabrics was determined using the Moisture Management Tester (M290 MMT, SDLATLAS). Three fabric samples of 80 × 80mm were prepared and conditioned for at least 24 h in 20 • C and 65% RH before analysis.

Strength and bending length.
Strength of warp and weft yarns ( Fig. SI-20) from untreated and treated fabric were determined using a Zwick/Roell Z100 tensile tester with a test length of 5 cm (length of the sample tested). Details of the strength can be found in Sec. SI-12. Bending length of the fabric samples were determined as per ASTM 3188 standard. 15 × 2.5 cm fabric strips were prepared in both warp and weft direction to determine the bending length (Sec. SI-12 for details).

Cellulose treatments and the principle of phosphine oxide macromolecule formation
The hydroxyl groups present in cellulose provide synthetic access points for various FRs that primarily function in the condensed phase. Dehydration reactions catalyzed by acidic species formed from the decomposition of the FR additives are key to flame retardation of cellulose [10]. For most applications in textiles, the durability of such FR treatment to laundry cycles is essential. Despite tremendous progress over the past decades, a commercially feasible, durable, formaldehydefree FR treatment still eludes the researchers. New FR S12 requires hydrolysis stability and non-leaching behavior to ensure durability throughout laundry treatments. Designing FR moieties to incorporate non-hydrolyzable bonds and network them via a physical or chemical means into the cellulose structure would prevent their leaching.
To fulfill these criteria, a simple strategy of in-situ Michael addition reaction involving phosphorus-based Michael acceptor and nitrogenbased Michael donors was used to synthesize a polymer network within the cellulose (Scheme 1). The cellulose fabrics were treated with a combination of trivinyl phosphine oxide (Michael acceptor), a cyclic amine (Pip or TPT as Michael donors) followed by curing, microwave heating, and steaming to ensure the in-situ formation of phosphine oxide polymeric networks that gets physically trapped within cellulose. A similar principle has previously been demonstrated in the synthesis of phosphine oxide-based pH-sensitive gels [51] and in-situ flame retardant macromolecules in polyamide 6 [52].
Michael addition of TVPO and the amines can be performed at room temperature in aqueous or solvent conditions; however, long curing conditions greater than a few hours [51] are not practical for industrial implementation. Thus, three different crosslinking experiments were performed (i.e., dry heat, microwave irradiation, and steaming) to facilitate faster crosslinking of the Michael adducts inside the cellulose. The detailed procedure of crosslinking experiments is described in Sec. SI-3. Table 1 summarizes details of all cellulose samples prepared using different crosslinking methods. The effect of various conditions, such as crosslinking temperature, time, and pressure was examined. The curing efficacy was estimated by measuring the phosphorus retention (Pretention) of the treated cellulose after standard laundry procedure (5 and 50 cycles). Increasing the curing temperature during dry crosslinking (D1 to D3) improved the P-retention (Table 1) up to 80% at 160 • C. However, high variation in P-retention is possibly due to nonuniform curing. As most of the humidity was removed before the curing process, it may hinder efficient in-situ crosslinking of the Michael adducts, leading to the removal of non crosslinked or partially crosslinked precursors during laundry. As a green processing strategy, microwave-assisted synthesis offers many benefits for its use in the textile industry, such as enhanced reaction rates, higher yields, greater selectivity, and economic. The closed system provided by microwave reactors allowed additional crosslinking experiments to be conducted in humid conditions. Cellulose samples were prepared with a minimum residual humidity (20 wt%) and modified by microwave-crosslinking.
As seen from Table 1 (M1 to M3), a higher P-retention (>90%) could be obtained for all treated cellulose samples, albeit with higher variation similar to the samples prepared by dry crosslinking. Improved P-retention, even at a lower crosslinking temperature (120 • C), is attributed to the residual humidity and microwave irradiation, which facilitate efficient Michael addition of adducts. The partially dried fabric was placed in the glass tube in a vertical orientation for microwave crosslinking. This led to the migration of Michael addition precursors to the bottom of the sample and a P-content variation from the top (lower) to the bottom (higher).
The steam crosslinking was performed in a simple setup (Fig. SI-1) and a high P-retention (>95%) with low-variation was observed for all samples (S1-S16). This demonstrates the uniformity of the steam crosslinking method. In the steam crosslinking procedure, the P-retention was independent of the curing temperature (pressure) and curing time (S1-S4 vs. S5-S8, Table 1). A crosslinking temperature of 110 • C and a curing time of 30 min was optimum to obtain high P-retention, providing reasonable conditions to adopt in an industrial textile process. Only in the case of TPT, ethanol was used as a solvent due to its poor solubility in water and high P-retention was observed, similar to other steam crosslinked samples. The P-retention for cellulose fabrics reported in Table 1 is after 5 laundry cycles. To further examine the durability of the treatments, 50 laundry cycles were performed (S13). Interestingly, 95% P-retention after 50 laundry cycles demonstrated the long term stability of the FR chemistry employed in this work, which is comparable to the commercially available formaldehyde-based FR for cotton fabrics.
To verify the formation of in-situ phosphine oxide macromolecules within the cellulose matrix, a solid-state NMR study was performed on the treated cellulose. For comparison, the crosslinked polymer (Poly-A) was synthesized after solid-state polymerization of TVPO and piperazine in the absence of cellulose (See SI-2 and S1-5). Fig. 2 shows the 31 P NMR data of TVPO, S12, and crosslinked polymer Poly-A. The broad phosphorus resonances centered at δ 49.6 ppm and 49.2 ppm (Fig. 2b) in the 31 P MAS NMR spectra of Poly-A and S12, respectively, can be assigned to polymer-bonded phosphorus atoms. The disappearance of the TVPO phosphorus peak at δ 17.4 ppm in 31 P NMR spectra (in CDCl 3, Fig. 2a) confirms the complete conversion of the starting material [51,52]. The 13 C and 31 P CP-MAS NMR spectra of blank cellulose, poly-A, Poly-B, S12 and S14 are presented in SI5 (Fig. SI-2 to SI-7).

Small scale fire tests
Untreated cellulose-based materials are readily combustible and require the application of flame retardants as a safety measure. These fireproof treatments are often based on various phosphorus chemistries, which have enhanced fire performance [10]. When exposed to fire, phosphorus-based flame retardants in cellulose will decompose to form an acidic phosphorus species that catalyzes char formation [53,54]. The char formed acts as a thermal barrier against heat and thus protects the underlying polymer. This mode of action exhibited by phosphorus-based flame retardants is referred to as the condensed phase activity.
Vertical flammability test (BKZ-VB) and LOI were performed on the treated cellulose fabrics to mimic small scale fires ( Table 2). An increased concentration of phosphorus in the treated cellulose led to an improved fire performance (S9-S12), which is reflected as increased LOI values and reduced char length in the BKZ-BV test. For a P-content >2 wt % a pass in the BKZ-VB test and an LOI value >27% was achieved. Postfire test (BKZ-BV) examination of the remaining residues revealed increased char formation for all treated fabrics (Fig. SI-9). The untreated fabric burned completely (Video1-Cellulose blank), leaving no char residue. Even though S9 and S10 are not self-extinguishing (Video2-S9 and Video3-S10), a significant char remained after the fire test. Concentrations of phosphorus >2 wt% improved the quantity of the char and self-extinguishment of the fabrics (Video4-S11, Video5-S12, and Video7-S15). Likewise, cellulose-containing TPT (S14 and S15) exhibited similar results in the small scale fire tests (Table 2).

Microcombustion calorimetry (MCC)
Micro combustion calorimetry (MCC) was used to estimate functionalized cellulose's fire performance (Fig. 3, Table 3). As shown in Fig. 3 and SI-10, all treated samples displayed a lower peak heat release  a Swiss standard vertical fire test. All Samples for the fire tests were exposed to 5 wash cycles, dried and conditioned before the fire tests. Details of SSI can be found in Table SI-4. rate (pHRR), heat release capacity (HRC), total heat release (THR), and higher residue compared to the untreated cotton (Table 3). Lower temperature for pHRR (T max ) in treated samples is due to the catalyzed decomposition of cellulose by acidic phosphorus species formed during the thermal decomposition of phosphorus compounds [55][56][57]. Increased P-content in cellulose leads to higher catalytic activity and results in lower HRC and T Max , similar to the observation made in previous research [55]. Compared to piperazine containing cellulose (S10 and S11), TPT treated samples (S14 and S15) at a similar P-content exhibited lower pHRR, possibly due to synergistic interaction between the triazine moiety in TPT with the phosphorus moieties [57].

Thermal data and evolved gas analysis
TGA studies of treated cellulose were undertaken to understand the decomposition behavior and gain insights into possible mode of action of the in-situ formed phosphine oxide macromolecules. The relevant TGA data of treated cellulose in inert and oxidative conditions are summarized in Table SI (Table SI-5) but significantly higher residue at 800 • C compared to blank cellulose. Poly-A and Poly-B exhibited a very high thermal stability (~360 • C and ~400 • C respectively) and a high char residue of ~30% (Poly-B). High thermal stability is typical behavior of crosslinked polymer systems [58,59] and higher residue could be due to the formation of thermally stable O-P-N-based inorganic polymer in N 2 and air, which is a typical effect of phosphorus based FR on cellulose [10,55,57]. Poly-B owing to the thermally stable triazine moiety, exhibits higher thermal stability compared to Poly-A.
The decrease in decomposition temperatures with improved char formation in treated cellulose was more pronounced in samples with higher P-content (S9-S12), highlighting the acid-catalyzed decomposition of cellulose. The phosphorus compounds decompose to produce phosphoric acid substructures and catalyzing cellulose decomposition to produce carbonaceous residues. In addition, base-catalyzed decomposition of cellulose is also possible due to amines moieties (i.e., piperazine and triazine moieties) in the phosphine oxide macromolecules [60]. For the treated cellulose, lower residues at 800 • C was observed in the air compared to those in N 2 atmospheres, which may be due to lower oxidative stability of the char [55,56].

Mechanism studies
It is clear from the MCC and TGA data that treated cellulose exhibits higher char formation compared to the untreated cellulose. Besides, in small scale fire tests (BKZ-BV, Fig. SI-9) considerable char formation was observed for the treated cellulose. The chemical composition of such residue can provide useful information on the fate of important elements like P, O, and C after the combustion process. The elemental composition of char obtained after vertical burning tests (BKZ-BV) of five samples (S9, S12, S14, S15, and SS1) was determined via EDX analysis (Table SI-6 and Fig. SI-13). For all residues, C, P, and O were the primary elements detected. With increasing initial P-content, decreased carbon, increased phosphorus, and oxygen contents were observed for treated cellulose. The decomposition of the phosphine oxide macromolecules leads to phosphorus species, which could crosslink to assist in forming a continuous char layer. Interestingly, nitrogen could only be detected for triazine containing amine samples (i.e., S14, S15), clearly indicating the Fig. 3. Heat release rate (HRR) of Cellulose blank, S12, S15 and SS1 were measured using MCC.

Table 3
Heat release profile of treated cotton cellulose.  difference in the condensed phase activity of the triazine based phosphine oxide macromolecules compared to the piperazine based phosphine oxide macromolecules. The morphology of char left after the burning was evaluated and for simplicity, only the microscopic image for S12 is presented in Fig. SI-13. One can see that the charred fibers on the fabric residue are non-voluminous, thus it is unlikely that the FR treatment has intumescent characteristics. The evolved gas analysis was performed using TGA-FTIR and DIP-MS to understand the FR mode of action in treated cellulose. Briefly, the TGA-FTIR data (Fig. SI-14 and 15) of treated cellulose confirmed an increased formation of nonflammable gasses like H 2 O, CO 2 , and reduction in flammable gasses like (carbonyl compounds, hydrocarbons, and CO). Such observation is a characteristic of condensed phase action of phosphorus compounds in cellulose [56,57]. The formation of phosphorus-based volatiles was confirmed by phosphoryl peak at 1268 cm − 1 [61]. Further, DIP-MS of treated cellulose (Fig. SI-16) confirmed the formation of phosphorus (PO*, triethyl phosphine oxide) and nitrogen (piperazine and TPT) species in the gas phase. The formation of volatile phosphorus species indicates their possible gas-phase activities.
Based on these observations, decomposition mechanisms of the treated cellulose is proposed in Scheme 2. Phosphine oxide macromolecules in S12 and S14 decompose to yields triethyl phosphine oxide radical, piperazine (S12) and TPT (S14) in the gas phase. In the next stage, piperazine or TPT decomposes to form ammonia (NH 3 ) [62], nitrogen (N 2 ), H 2 O, CO 2 and alkane. Furthermore, the triethyl phosphine oxide fragment of S12 and S14 decompose to produce alkane and PO • . The PO • could remain in the condensed phase to produce phosphoric acids and polyphosphoric acid, thus reinforcing the condensed phase activity. The PO • can also work in the gas phase by recombining H and OH radicals [63,64]. Amines catalyze acidic phosphate intermediates formation leads to cellulose phosphorylation and yields polyphosphoric acid, which assists in the condensed phase activity. Moreover, the amines can also react with polyphosphoric acids to form phosphorus oxynitride, reinforcing it, and acting as an excellent thermal barrier [57].

Fabric morphology
The phosphine oxide macromolecules are expected to form primarily inside the cellulose without influencing any surface characteristics. Thus, the SEM analysis was performed to get further insight into surface morphology of treated samples. No difference in surface morphology of blank cotton cellulose and S12 was observed in the SEM analysis ( Fig. 5a  and b), which indicates that phosphine oxide macromolecules are formed inside the cellulose fibers due to the diffusion of precursors into the cellulose matrix before crosslinking. The in-situ formation of crosslinked networks is already highlighted by complete P-retention even after 50 laundering cycles and solid-state NMR analysis. On the contrary, coating observed in the SEM image of S15 (Fig. 5c) is due to poor diffusion of TPT (insoluble in water) into the matrix, resulting in the formation of crosslinked networks on and within the fibers.

Fabric comfort, appearance and strength
It is important to examine the effect of FR treatments on the comfort properties of the textiles. Use of FR above 20-30 wt% of the material for efficient fire protection results in a possible change in the fabric's moisture absorption and transport capacity, and affecting the comfort properties. Thus, comfort properties were evaluated by measuring the wettability and moisture transport characteristics of fabric samples. The wettability of the cellulose textiles measured as water contact angle (WCA, Fig. 6a) indicated the hydrophilic behavior of the untreated cotton and S12 (WCA of 0 • ). Whereas, WCA of 90 • in S15 indicates a hydrophobic nature of the treatments (Fig. 6a).
The moisture management properties were investigated to assess critical aspects of wear comfort of protective clothing [50]. The water spreading speed and maximum wetted radius were taken as measures for moisture transport and absorption, respectively (Fig. 6b and c). The maximum wetted radius reflects the moisture spreading distance for fabric and indicates its moisture uptake properties. Good moisture uptake properties are important for moisture absorption from the environment and control the clothing microclimate. This is critical to prevent the growth of skin-irritating bacteria, yeast, and fungus. A fastspreading speed contributes to the quick distribution of liquid in the fabric, contributing to a good drying behavior of the fabric. This is important to keep the water vapor gradient within the garment to maintain the body's evaporative cooling and, on the other side, to remove moisture clothing to maintain protective properties. Moisture management parameters were measured at the top and bottom of the blank and treated cellulose textiles and are classified according to the literature [65]. The higher wetted radius for S12, compared to the blank fabric and S15 can be due to the hydrophilic nature of the Poly-A. S15 showed a medium wetted radius, which might be due to the hydrophobic nature of Poly-B. Furthermore, the liquid spreading speed contributed to quick evaporation and was classified as very fast for S12 (5.3 mm/sec) and pristine cellulose (4.1 mm/sec). The spreading rate on S15 is classified as very slow.
Cellulose fabrics tend to turn yellow when subjected to heat [66] and alkaline conditions [67], adversely impacting their visual aesthetics. To ascertain the effect of the novel FR treatment on the fabric's appearance, the reflectance was measured using a spectrophotometer (Fig SI-19). Interestingly, the reflectance value of S12 was equal to that of untreated fabric within the wavelength range of 500-800 nm, confirming no discoloration during the treatment. On the other hand, the lower reflectance of S15 at all wavelengths ( Fig. SI-19b) indicates a slight yellowing of the fabric. This could be due to small impurities (ppm levels) present in TPT after its synthesis.
From SEM and comfort analysis, it is clear that FR treatment using piperazine (S12) maintains the cotton fabric's desired properties. To further evaluate the potential impact of the FR treatment, mechanical properties of warp and weft yarns of the treated fabrics (S12) were evaluated and presented in Sec. SI-12. As shown in Table SI-7, no adverse effect of FR treatment on the mechanical properties was observed. The bending length of a fabric that indicates its stiffness was thus measured and sample S12 has a slightly higher value than the untreated fabric (Table SI-7), indicating a marginal increase in the fabric's stiffness after FR treatment.

In-situ silver nanoparticle formation and antibacterial property
By exploiting reducing ability of amines to form AgNPs from silver salts [68,69], a multifunctional approach to integrated flame retardant Fig. 8. Antimicrobial activity of (a) S12, (b) SS1 and (c) SS2 determined by agar diffusion test using S. aureus and P. aeruginosa. Contact killing of (d) S. aureus and (e) P. aeruginosa by S12, SS1 and 50WSS1 (SS1 after 50 launderings). Standard deviations derived from three replicates are shown as Error bars. and antimicrobial properties within cellulose was explored. In contrast to conventional methods, this route eliminated the use of any additional reducing agent for preparation of AgNPs [70,71] and enabled simultaneous in-situ formation of TVPO-piperazine network and AgNPs (Sec. 2.3). Cellulose treated with AgNO 3 maintained their white color even after drying (Fig. SI-2), while yellow color after steaming indicated the formation of AgNPs (Fig. SI-2). The actual silver content in SS1 and SS2 were 0.08% and 0.75% respectively (Table SI- 4). The durability of silver treatment was confirmed by subjecting the SS1 to 50 laundering cycles (AATCC . Interestingly, the sample displayed excellent Agretention and P-retention even after 50 home laundering cycles (Table SI-4). The fire performance of the fabric was not affected by the presence of AgNPs (SS1). As shown in Video V8 of SI and Table 2, SS1 exhibited excellent flame retardancy similar to S11.
To confirm the presence of AgNPs in the treated cellulose, SEM and TEM analyses were performed. SEM and TEM analysis of SS1 confirmed the presence of AgNPs on and within the fiber with an average particle size of 17 ± 6 nm 8.5 ± 3 nm, respectively ( Fig. 7a and b). From the TEM image (Fig. 7b), it can be seen that number of AgNPs and their size tends to increase towards the surface of the fiber. It can be attributed to the migration of Ag + ions towards the surface during drying followed by the formation of AgNPs during steaming. This also explains the presence of larger AgNP on the surface of the fibers than in its interior. Increasing the AgNO 3 loading to 0.75% (SS2), an increase in particle size to 11 ± 4 nm was observed (Fig. 7d). Similar to SS1, more number of AgNPs with a larger size were observed near the surface of the fiber in case of SS2.
The antibacterial property of the cotton fabric with an amine-based phosphine oxide polymer and silver nanoparticle was evaluated by agar diffusion test using S. aureus and P. aeruginosa. Bacterial growth on and around S12 fabric samples was observed in both cases (Fig. 8a). Unlike published reports [72,73], no zone of inhibition was observed for P. aeruginosa exposed to fabric samples with AgNPs ( Fig. 8b and c, lower panels) and only slight inhibition of S. aureus growth adjacent to the sample edges ( Fig. 8b and c, upper panels). The absence of a zone of inhibition can be attributed to a lack of sufficient silver release from the samples. A similar observation, even in a sample with high Ag-content (SS2) highlights the stabilizing effect of a crosslinked Poly-A network present in the composite cellulose (Fig. 8c). Very low loss of silver even after 50 laundering cycles also confirmed the stabilizing effect of phosphine oxide macromolecule (Table SI-4).
To elucidate the antibacterial efficacy of fabrics, contact killing of both bacteria was carried out using the sample with 0.1% Ag before (SS1) and after 50 laundry cycles (50WSS1) [74,75]. The bacterial suspension was directly loaded on the SS1 and 50WSS1, followed by incubation at 37 • C for 2 h (Sec. 2.7). Similar to the agar diffusion test, no killing was observed in the case of S12. Interestingly, the complete killing of both bacterial strains by SS1 and 50WSS1 during contact killing tests (Fig. 8d and e) highlights the antibacterial efficiency and its durability. It is worth mentioning that excellent antibacterial property with low Ag-release is highly desirable for protective materials with minimum exposure to nanomaterial pollution.
Finally, flame retardant cellulose prepared in this work is compared with the recently reported state of art FR treated cellulose (Table 4). Properties such as durability of the treatment, fire performance (LOI), whiteness index (%), comfort, and multifunctionality of the treated fabrics are considered for comparison. It can be seen that (Table 4), the novel FR treatment for the cellulose developed in this work can achieve durable fire performance (after 50 laundry cycles) and excellent comfort property without discoloration of the fabric. Additionally, the novel method offers a possibility to impart excellent antimicrobial property along with fire performance in a single step, which is not common among the recently reported state of art methods.

Conclusion
Formaldehyde-free flame retardant treatments for cotton cellulose have been a challenge for researchers over the last few decades with no clear solution which has been commercialized. In this work, a durable flame retardant cellulose was developed via in situ Phospha-Michael addition crosslinking reaction using TVPO and cyclic amines (piperazine or TPT) as precursors. A simple application from an aqueous solution followed by crosslinking under the steam of the treated fabric ensured the formation of phosphine oxide physical networks that are stable to 50 laundry cycles. Cellulose fabrics containing a minimum of 2 wt% phosphorus displayed excellent flame retardant behavior, characterized by LOI of ≥27.0% and after the flame time of 0 sec during the BKZ-VB Test. Based on the TGA-FTIR and DIP-MS analysis, a possible condensed (dominant) and gas (minor) mechanism of flame retardancy for the phosphine oxide polymers was proposed. The new treatment thus developed is not only efficient as a flame retardant but can be adapted to offer multifunctional treatments (i.e., antimicrobial functionality via incorporation of silver nanoparticles). Unlike other flame retardant treatments described in the literature, the new treatment ensures the treated fabric's excellent comfort. In the future, detailed investigations regarding the treatment of colored cellulose fabrics, impact on air permeability, cost evaluation, and optimization of other fabric properties will be performed, which will enable future commercialization of this technology. *30 after washing cycles, a after 5 washing cycle, Not given (NG).