A Polysiloxane Delivery Vehicle of Cyclic N-Halamine for Biocidal Coating of Cellulose in Supercritical CO2

Cyclic N-halamines are highly antimicrobial, very stable, and not susceptible to bacterial resistance. A polysiloxane delivery vehicle was synthesized to deliver cyclic imide N-halamine onto cellulose via a benign and universal procedure that does not require a harmful solvent or chemical bonding. In brief, Knoevenagel condensation between barbituric acid and 4-hydroxybenzaldehyde furnished 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione, whose phenolic O−H was subsequently reacted with the Si−H of poly(methylhydrosiloxane) (PMHS) via silane alcoholysis. The product of silane alcoholysis was interpenetrated into cellulose in supercritical CO2 (scCO2) at 50 °C, to form a continuous modification layer. The thickness of the modification layer positively correlated with interpenetration pressure in the experimental range of 10 to 28 MPa and reached a maximum value of 76.5 nm, which demonstrates the ability for tunable delivery, to control the loading of the imide N−H bond originating from barbituric acid unit. The imide N−H bonds on cellulose with the thickest modifier were then chlorinated into N−Cl counterparts using tert-butyl hypochlorite, to exert a powerful biocidability, providing ~7 log reductions of both S. aureus and E. coli in 20 min. The stability and rechargeability of the biocidability were both very promising, suggesting that the polysiloxane modifier has a satisfactory chemical structure and interlocks firmly with cellulose via scCO2 interpenetration.


Introduction
Antibacterial materials can prevent bacterial infection and hence are very useful in hygiene and daily life [1,2]. It is of great importance to modify materials for durable biocidability, with a powerful biocide and a facile and green procedure [3]. Compared with other commonly used polymers, such as polyolefins, cellulose is more susceptible to bacterial attachment and proliferation, due to its surface hydrophicility, which origates from the hydroxy groups [4]. It is therefore particularly significant to use cellulose as a reprensatitive to evaluate the performace of biocides and modification procedures.
Many biocidal functionalities such as N-halamines (N-chloramines and N-bromamines) [5][6][7], antibiotics [8,9], quaternary ammonia salts [10][11][12], reactive oxygen species [13,14], and heavymetal ions [15][16][17][18] have been incorporated onto cellulose to give an antibacterial ability. Nhalamines, produced by halogenation of imide/amide/amine N−H to N−X (X=Cl, Br), have interpenetration is especially useful for modification of substrates such as polyolefins and cellulose that do not swell in common liquid solvents and do not have the required reactivity. Among the few CO 2 -soluble polymers, polysiloxanes are usually chosen as a delivery vehicle to convey the desired functionalities such as biocidability onto substrates, since they are industrially available, nonhazardous, chemically stable, and resistant to a wide range of temperatures. The synthesis of such a delivery vehicle involves the incorporation of a functional group to a backbone composed of a -Si-O-repeat unit.
Based on the above discussion, using scCO 2 interpenetration to insert a polysiloxane delivery vehicle of cyclic imide N-halamine groups is hypothesized to be a versatile and friendly solution to forming a powerful and durable biocidability on cellulose. To verify this assumption, a polysiloxane with barbituric acid pendants was synthesized and interpenetrated into cellulose in scCO 2 , to form a stable interlocking layer. Subsequent chlorination of N−H bonds of barbituric acid furnished a polysiloxane modifier with cyclic imide N-halamine sites. Briefly, Knoevenagel condensation between barbituric acid and 4-hydroxybenzaldehyde furnished 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione. The phenolic hydroxy group of 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione was reacted with of Si−H of PMHS via silane alcoholysis, to produce a biocidal polysiloxane precursor (referred to as barbituric acid-based polysiloxane). The barbituric acid-based polysiloxane was interpenetrated into cellulose in scCO 2 at 50 • C, the temperature at which polysiloxanes are most soluble [36], to furnish a sample named barbituric acid-based polysiloxane@cellulose. Since the pressure of scCO 2 is a crucial parameter affecting the solubility and hence the thickness of barbituric acid-based polysiloxane, the relationship between interpenetration pressure and thickness of barbituric acid-based polysiloxane was studied. The results showed that the thickness of the barbituric acid-based polysiloxane was able to be conveniently adjusted by variation of the interpenetration pressure, to deliver different amounts of the N-halamine precursor, imide N−H from barbituric acid. Cellulose with the maximum loading of barbituric acid-based polysiloxane was chlorinated to convert N−H bonds using tert-butyl hypochlorite, to create a sample with cyclic imide N-halamine sites on the surface, which is hereafter called cyclic imide N-halamine polysiloxane@cellulose. The overall procedure is shown in Scheme 1.
ifier interlocks firmly with the substrate to provide a new interfacial property. ScCO2 i terpenetration is especially useful for modification of substrates such as polyolefins an cellulose that do not swell in common liquid solvents and do not have the required rea tivity. Among the few CO2-soluble polymers, polysiloxanes are usually chosen as a deli ery vehicle to convey the desired functionalities such as biocidability onto substrates, sin they are industrially available, nonhazardous, chemically stable, and resistant to a wid range of temperatures. The synthesis of such a delivery vehicle involves the incorporatio of a functional group to a backbone composed of a -Si-O-repeat unit.
Based on the above discussion, using scCO2 interpenetration to insert a polysiloxan delivery vehicle of cyclic imide N-halamine groups is hypothesized to be a versatile an friendly solution to forming a powerful and durable biocidability on cellulose. To veri this assumption, a polysiloxane with barbituric acid pendants was synthesized and inte penetrated into cellulose in scCO2, to form a stable interlocking layer. Subsequent chlori ation of N−H bonds of barbituric acid furnished a polysiloxane modifier with cyclic imid N-halamine sites. Briefly, Knoevenagel condensation between barbituric acid and 4-h droxybenzaldehyde furnished 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione. Th phenolic hydroxy group of 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione was reacte with of Si−H of PMHS via silane alcoholysis, to produce a biocidal polysiloxane precurso (referred to as barbituric acid-based polysiloxane). The barbituric acid-based polysiloxan was interpenetrated into cellulose in scCO2 at 50 °C, the temperature at which polysilo anes are most soluble [36], to furnish a sample named barbituric acid-based polysilo ane@cellulose. Since the pressure of scCO2 is a crucial parameter affecting the solubili and hence the thickness of barbituric acid-based polysiloxane, the relationship betwee interpenetration pressure and thickness of barbituric acid-based polysiloxane was stud ied. The results showed that the thickness of the barbituric acid-based polysiloxane w able to be conveniently adjusted by variation of the interpenetration pressure, to deliv different amounts of the N-halamine precursor, imide N−H from barbituric acid. Cellulo with the maximum loading of barbituric acid-based polysiloxane was chlorinated to co vert N−H bonds using tert-butyl hypochlorite, to create a sample with cyclic imide N-ha amine sites on the surface, which is hereafter called cyclic imide N-halamine polysilo ane@cellulose. The overall procedure is shown in Scheme 1. Antibacterial and stability tests showed that the resultant cyclic imide N-halamin polysiloxane@cellulose possessed a super, stable, and rechargeable biocidability, as e pected, confirming the benefits of a cyclic imide N-halamine, polysiloxane delivery veh cle and scCO2 interpenetration. Antibacterial and stability tests showed that the resultant cyclic imide N-halamine polysiloxane@cellulose possessed a super, stable, and rechargeable biocidability, as expected, confirming the benefits of a cyclic imide N-halamine, polysiloxane delivery vehicle and scCO 2 interpenetration.

Instrumentation
Fourier transform infrared spectroscopy (FTIR) spectra were acquired with a Nicolet Magna IR-560 FTIR spectrometer(Madison, WI, USA), using KBr pellet method at a resolution of 0.5 cm −1 in the wavelength range of 400 to 4000 cm −1 .
The types and valences of elements on cellulose were measured with a Thermo Scientific Escalab 250Xi X-ray photoelectron spectrometer (XPS) (Waltham, MA, USA) equipped with a Al Ka radiation source and a low energy electron flood gun. The test pressure and takeoff angle were maintained at 1 × 10 −6 Pa and 45 • , respectively, during spectrum acquisitions. Survey spectra were collected at an analyzer pass energy of 100 eV and a resolution of 1 eV, while high resolution spectra were collected at an analyzer pass energy of 23.5 eV and a resolution of 0.05 eV. Elemental binding energies were referenced to the aliphatic C 1s at 284.7 eV.
Pristine cellulose and cyclic imide N-halamine polysiloxane@cellulose were first coated with platinum under vacuum with a Cressington 108 auto sputter coater and then observed with a FEI Nano SEM450 scanning electron microscope (Waltham, MA, USA) that was equipped with a secondary electron detector. SEM images were collected at an accelerating voltage of 10 kV and a chamber pressure of 1 × 10 −4 Pa.
The density of barbituric acid-based polysiloxane was measured using a density balance (FA2004J) from Shanghai Pingxuan Science Instrument Co., Ltd. (Shanghai, China).

Synthesis of Barbituric Acid-Based Polysiloxane That Bears Cyclic Imide N−H Sites
The synthesis of barbituric acid-based polysiloxane involves two steps, as shown in Scheme 1. First, 2.56 g (0.02 mol) of barbituric acid and 2.44 g of (0.02 mol) 4-hydroxybenzald ehyde were dissolved in 50 mL water in a flask. The solution was stirred at 70 • C for 6 h to complete the Knoevenagel condensation and furnish 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione, which was purified by recrystallization from water.

Interpenetration of Cellulose with Barbituric Acid-Based Polysiloxane in scCO 2
A 100 mL high-pressure stainless-steel cell was separated into an upper and a lower chamber using a removable perforated steel slice. A 5.0 × 5.0 cm 2 cellulose swatch (about Polymers 2022, 14, 5080 5 of 16 0.26 g) was placed in the upper chamber and a glass vial containing 0.4 g of barbituric acid-based polysiloxane was placed in the lower one, to avoid direct contact. The cell was then purged with CO 2 , heated to 50 • C and charged with CO 2 to a desired pressure with a syringe pump. The cell was maintained under these conditions overnight, to complete one interpenetration process. The pressurization and depressurization periods were about 3 and 5 min, respectively. An average of three measurements was reported for each interpenetration condition. A temperature of 50 • C was selected as polysiloxanes have maximum solubility at this temperature [36]. An overnight interpenetration period was chosen since a longer time did not cause a measurable change of the loading of barbituric acid-based polysiloxane on cellulose. It is known that the solubility of polysiloxanes increases with the increase of pressure [38], indicating a facile way to form an interpenetration layer with adjustable thickness, to achieve an improved delivery ability. Herein, the thickness of barbituric acid-based polysiloxane was controlled by varying the interpenetration pressure between 8 and 28 MPa, which in turn decided the amount of cyclic imide N-halamine that was delivered to the final cellulose surface.
The thickness of the barbituric acid-based polysiloxane modifier at a certain pressure can be estimated based on the weight change before and after the interpenetration process using the following Equation (1) [33]: where h is the layer thickness of barbituric acid-based polysiloxane, d is the average diameter of the original cellulose fiber (11.8 µm), W 0 is the weight of the original fibers used for interpenetration, W 1 is the weight of fibers after interpenetrated with barbituric acid-based polysiloxane, ρ cellulose is the density of cellulose (1.54 g/cm 3 ), and ρ polysiloxane is the density of barbituric acid-based polysiloxane (0.99 g/cm 3 ). Each sample was tested in triplicate.
The cellulose with the thickest interpenetration layer was chosen for further modifications and characterizations including chlorination, biocidal assessments, and stability tests.

Chlorination of Cyclic Imide N−H Sites of Barbituric Acid-Based Polysiloxane
For chlorination of polymers including barbituric acid-based polysiloxane, tert-butyl hypochlorite performs better than inorganic sodium hypochlorite, due to the "like dissolves like" principle and hence was used in this study. Tert-butyl hypochlorite was prepared using sodium hypochlorite, glacial acetic acid, and tert-butyl alcohol, according to a reported procedure based on the following reaction (2) [39].
Cellulose fabrics (5.0 × 5.0 cm 2 ) with the thickest interpenetration layer were then immersed in excessive tert-butyl hypochlorite in a flask at room temperature overnight to chlorinate the cyclic imide N−H sites of barbituric acid-based polysiloxane to their N−Cl counterparts. After that, excessive tert-butyl hypochlorite was vacuum dried to obtain cyclic imide N-halamine polysiloxane@cellulose, as shown in Scheme 1.

Titration of Biocidal Chlorine
The loading of biocidal chlorine on cyclic imide N-halamine polysiloxane@cellulose was assessed using iodimetric/thiosulfate titration for correlation with a biocidability that is concentration dependent. Before titration, the cyclic imide N-halamine polysilox-ane@cellulose fabrics were initially washed with DI water, in order to remove free chlorines and then dried in the air. Pristine cellulose fibers were subjected to the same treatments and served as controls. Each sample was tested in triplicate. The loading of biocidal chlorine on cellulose was calculated using the following Equation (3) [40]: where m Cl + % is the weight percentage of biocidal chlorine on cyclic imide N-halamine polysiloxane@cellulose fibers, N is the concentration (mol/L) of Na 2 S 2 O 3 solution for titration, V Cl + and V 0 represent the volumes (L) of Na 2 S 2 O 3 solution used for titrations of cyclic imide N-halamine polysiloxane@cellulose fibers and pristine cellulose fibers, respectively, and W is the weight (g) of cellulose fibers in titrations.

Antibacterial Kinetics of Cyclic Imide N-Halamine Polysiloxane@Cellulose Fabrics
The biocidal kinetics of cyclic imide N-halamine polysiloxane@cellulose fabrics were measured by challenging with Gram-negative E. coli and Gram-positive S. aureus in a "sandwich test", using pristine ones as negative controls [41]. E. coli and S. aureus were first cultured in Luria-Bertani (LB) and Tryptic Soy (TS) broths, respectively, at 37 • C for 24 h. The bacteria were then separated with the aid of centrifugation, washed several times with sterilized PBS solution, and finally adjusted to a concentration of~10 8 CFU/mL with PBS. Then, 25 µL of the suspension was sandwiched between two pieces of identical swatches of 2.54 cm 2 each. A sterilized weight was placed on the top to ensure good contact of the swatches and bacterial suspension. The swatches of different contact periods were soaked in 10 mL of sterile Na 2 S 2 O 3 (0.05 wt%) in a tube, to neutralize biocidal chlorines and vortexed for 2 min, to remove the surface-attached bacteria. Next, 100 µL of each serial dilution of a vortexed solution was placed on an agar plate at 37 • C for 24 h for determination of the number of viable cells. Each sample was tested in triplicate and in three independent experiments, for evaluation of biocidal performance.

Testing of Biocidal Stability and Rechargeability
The biocidal modifier is gradually worn away from the surface and the N-halamine sites are progressively hydrolyzed during usage and storage, causing the loss of antibacterial capacity. Therefore, it was important to evaluate the stability and rechargeability of the biocidal function. The stability and rechargeability biocidal polysiloxane modifier and imide N-halamine sites toward washing cycles were tested using a standard method of the American Association of Textile Chemists and Colorists (2A procedure of AATCC 61−1996), in which 2.5 × 5.1 cm 2 of swatches were placed in a canister with 150 mL of 0.15% AATCC detergent and 50 stainless steel balls. The canister was rotated at 42 rpm and 49 • C for 45 min to complete one washing cycle, which was equivalent to five machine cycles. Washed swatches were subsequently cleaned with distilled water, followed by drying in the air at room temperature. Some of the dried swatches were directly titrated for assessment of the concentration of residual chlorine, while others were first rechlorinated with tert-butyl hypochlorite and then subjected to titration for evaluation of the rechargeability of the lost chlorines.
The stability and rechargeability of biocidal functionality for long-term storage were assessed by storing cyclic imide N-halamine polysiloxane@cellulose samples at 25 • C and 65% relative humidity (RH) for up to 3 months under laboratory light from fluorescent lamps. One set of stored samples was immediately titrated and the other set was rechlorinated and subsequently titrated, to evaluate the retention and regeneration of N-halamine structures.
The stability and rechargeability toward UV irritation (~340 nm) were investigated using an AP-UV Accelerated Weathering Tester (Dongguan Aipei Test Equipment Co., Ltd., Dongguan, China).) at room temperature and 25% RH. After a certain time of UV exposure, the swatches with and without rechlorination were subjected to iodimetric/thiosulfate titration, to estimate the photolytic stability of the polysiloxane modifier and N-halamine sites of cyclic imide N-halamine polysiloxane@cellulose.

Construction and Characterizations of Biocidal Polysiloxane Layer on Cellulose
The relationship between interpenetration pressure and the thickness of barbituric acid-based polysiloxane on cellulose was first studied. The solvent strength of scCO2 increases with the increase of pressure, which leads to a higher solubility of barbituric acidbased polysiloxane and, therefore, a thicker modification layer on cellulose. Therefore, variation of interpenetration pressure is a convenient method to adjust the thickness/delivery ability of barbituric acid-based polysiloxane, for control of the loading of biocidal N-halamine, whose efficacy is content-dependent, and to achieve different killing capaci- In the second step, 5-(4-hydroxybenzylidene)pyrimidine-2,4,6-trione was reacted with PMHS via silane alcoholysis. In the spectrum of PMHS, the stretching and bending peaks of Si−H groups appeared at 2164 and 837 cm −1 , respectively, as shown in Figure 1d, respectively. However, the intensities of these two peaks were greatly reduced in the spectrum of the silane alcoholysis product (Figure 1e), barbituric acid-based polysiloxane, due to the consumption of Si−H groups in the alcoholysis reaction. The peaks from 5-(4hydroxybenzylidene)pyrimidine-2,4,6-trione, including the stretching mode of N−H and C=O, also occur in the spectrum at similar wavenumbers of 3218 and 3080 cm −1 and in the range of 1670 to 1697 cm −1 , respectively. The FTIR characterization proved the successful synthesis of barbituric acid-based polysiloxane.

Construction and Characterizations of Biocidal Polysiloxane Layer on Cellulose
The relationship between interpenetration pressure and the thickness of barbituric acidbased polysiloxane on cellulose was first studied. The solvent strength of scCO 2 increases with the increase of pressure, which leads to a higher solubility of barbituric acid-based polysiloxane and, therefore, a thicker modification layer on cellulose. Therefore, variation of interpenetration pressure is a convenient method to adjust the thickness/delivery ability of barbituric acid-based polysiloxane, for control of the loading of biocidal N-halamine, whose efficacy is content-dependent, and to achieve different killing capacities.
The solubility of a polymer in scCO 2 is very sensitive to pressure. The relationship of the thickness of barbituric acid-based polysiloxane layer and interpenetration pressure is presented in Figure 2. It was observed that the thickness of the modifier increased sharply in the pressure region from 8 to 20 MPa and then more gradually at higher pressures, reaching a maximum thickness of 76.5 nm at 28 MPa. Therefore, it is quite convenient to control the loading of the modifier to deliver different amounts of cyclic imide N-halamine onto cellulose, eventually through the adjustment of interpenetration pressure. The cellulose with the thickest modifier (76.5 nm) was chlorinated to produce cyclic imide N-halamine polysiloxane@cellulose, which was used for following characterizations and tests. The composition changes of cellulose caused by interpenetration and chlorination were first characterized using the FTIR technique. The spectra of unmodified cellulose, barbituric acid-based polysiloxane@cellulose, and cyclic imide N-halamine polysilox-ane@cellulose are shown in Figure 3. Compared with that of pristine sample, the spectrum of cellulose interpenetrated with barbituric acid-based polysiloxane (Figure 3b The composition changes of cellulose caused by interpenetration and chlorination were first characterized using the FTIR technique. The spectra of unmodified cellulose, barbituric acid-based polysiloxane@cellulose, and cyclic imide N-halamine polysiloxane@cellulose are shown in Figure 3. Compared with that of pristine sample, the spectrum of cellulose interpenetrated with barbituric acid-based polysiloxane (Figure 3b) shows characteristic peaks arising from barbituric acid-based polysiloxane, including an imidic N−H stretching vibration centered at 3202 cm −1 , C=O stretching vibrations of NHCONH and NHCOC< modes in the range of 1671 to 1718 cm −1 , a Si-O-C stretching mode at 1261 cm −1 , and Si−O−Si stretching mode at 801 cm −1 . After chlorination, the stretching peak of N−H vanished due to the transformation to N−Cl, as shown in Figure 3c. In addition, the stretching vibrations of different C=O bonds appeared in a higher wavenumber range of 1692 to 1738 cm −1 . This shift of C=O stretching peak was also been observed in other cases and is believed to be caused by the disappearance of hydrogen bonding of N-H-O=C and the increase of atomic mass when using chlorine to replace hydrogen [32,42]. Therefore, FTIR proved the success of the interpenetration and chlorination processes. The loading of the biocidal chlorine on cyclic imide N-halamine polysiloxane@cellulose was calculated to be 0.31 wt% based on data of titration using Equation (3). XPS was subsequently used as a supplement to FTIR, since it merely collects infor mation on the top surface (~5 nm), which avoids signal interference with the cellulos substrate. XPS survey scans of pristine cellulose, barbituric acid-based polysiloxane@cel lulose, and cyclic imide N-halamine polysiloxane@cellulose are shown in Figure 4. Th spectrum of pure cellulose only displays photoelectron peaks of C1s and O1s at approxi mate 285 and 532 eV, as shown in Figure 4a. The spectrum of barbituric acid-based pol ysiloxane@cellulose (Figure 4b) essentially shows photoelectron peaks of elements of th XPS was subsequently used as a supplement to FTIR, since it merely collects information on the top surface (~5 nm), which avoids signal interference with the cellulose substrate. XPS survey scans of pristine cellulose, barbituric acid-based polysiloxane@cellulose, and cyclic imide N-halamine polysiloxane@cellulose are shown in Figure 4. The spectrum of pure cellulose only displays photoelectron peaks of C 1s and O 1s at approximate 285 and 532 eV, as shown in Figure 4a. The spectrum of barbituric acid-based polysiloxane@cellulose (Figure 4b) essentially shows photoelectron peaks of elements of the barbituric acid-based polysiloxane modifier, including Si 2p , Si 2s , C 1s , N 1s , and O 1s at about 102, 152, 285, 400, and 532 eV, respectively. The spectrum of the chlorinated sample, cyclic imide N-halamine polysiloxane@cellulose, exhibited two more peaks of Cl 2p and Cl 2s at around 202 and 272 eV, as illustrated in Figure 4c, respectively, owing to the transformation of N−H to N−Cl. In addition, chlorination caused a change of chemical status of nitrogen. Then, high resolution spectra of N 1s , before and after chlorination, were acquired, as shown in Figure 5, which displays a binding energy shift from 399.9 eV to 400.7 eV, confirming the transformation of the N−H structure to its N−Cl counterpart [26,43].     The morphology of the biocidal polysiloxane modifier on cyclic imide N-halamine polysiloxane@cellulose was observed using SEM, after the verification of its existence with FTIR and XPS. SEM images show that a continuous and slightly coarse film fully covered the original smooth surface of pristine cellulose, comparing Figure 6a,b. A defect-free modifier with the correct chemical composites ensured the desired biocidal performance, which was subsequently confirmed through antibacterial testing. Cellulose fibers swell in scCO 2 but there was no measurable change in diameter after depressurization (image not shown for brevity).   The morphology of the biocidal polysiloxane modifier on cyclic imide N-halamine polysiloxane@cellulose was observed using SEM, after the verification of its existence with FTIR and XPS. SEM images show that a continuous and slightly coarse film fully covered the original smooth surface of pristine cellulose, comparing Figure 6a,b. A defect-free modifier with the correct chemical composites ensured the desired biocidal performance, which was subsequently confirmed through antibacterial testing. Cellulose fibers swell in scCO2 but there was no measurable change in diameter after depressurization (image not shown for brevity).

Antibacterial Testing
The antibacterial function of cyclic imide N-halamine polysiloxane@cellulose was measured against the two model strains Gram-negative E. coli and Gram-positive S. aureus, using pristine cellulose as a negative control. The log reductions of the two bacteria at different contact durations with biocidal swatches and controls are summarized in Table 1.
The controls did not afford any perceptible reduction of both E. coli and S. aureus, even at 20 min of contact, since the slight decreases of E. coli (0.37 log) and S. aureus (0.36 log) were caused by natural mortality and adsorption of microbes onto the cellulose fibers. In contrast, cyclic imide N-halamine polysiloxane@cellulose swatches exerted a powerful biocidability, achieving 3.96, 5.22, 6.78, and 7.11 log reductions of E. coli and 3.42, 4.76,

Antibacterial Testing
The antibacterial function of cyclic imide N-halamine polysiloxane@cellulose was measured against the two model strains Gram-negative E. coli and Gram-positive S. aureus, using pristine cellulose as a negative control. The log reductions of the two bacteria at different contact durations with biocidal swatches and controls are summarized in Table 1. The controls did not afford any perceptible reduction of both E. coli and S. aureus, even at 20 min of contact, since the slight decreases of E. coli (0.37 log) and S. aureus (0.36 log) were caused by natural mortality and adsorption of microbes onto the cellulose fibers. In contrast, cyclic imide N-halamine polysiloxane@cellulose swatches exerted a powerful biocidability, achieving 3.96, 5.22, 6.78, and 7.11 log reductions of E. coli and 3.42, 4.76, 6.12, and 7.08 log reductions of S. aureus at contact time of 5, 10, 15, and 20 min, respectively. The slightly lower efficacy against S. aureus could be attributed to its thicker membrane, which is more resistant to the penetration of biocidal chlorines. The efficacy herein was higher than that of amide and amine N-halamines at a similar chlorine loading, which needed a longer contact time (20-30 min) to kill nearly the same number of bacteria [44,45], which proved the advantage of imide N-halamine, which has the highest dissociation constant and the fastest release of chlorine ions, to provide the most efficient killing. In addition, each barbituric acid unit has two N-halamine sites and hence a higher ratio of biocidal functionality, which benefits the antibacterial performance.

Biocidal Stability and Rechargeability
The biocidal stability and rechargeability of cyclic imide N-halamine polysiloxane@cell ulose are important parameters for the evaluation of the antibacterial group, polysiloxane modifier, and modification procedure. These two properties were therefore studied against repeated washings, UV irradiation, and storage. Polysiloxanes are ideal biocidal carriers, since they are sturdy, hydrophobic, and photodegradation resistant. As shown in Table 2, the loading of biocidal chlorine decreased with the increase of washing cycles, since the polysiloxane modifier was gradually washed away and the N-halamine sites progressively hydrolyzed, from an initial value of 0.31 wt% to 0.10 wt% and to 0.06 wt% after 5 and 10 AATCC washing cycles, respectively. The washing stability of the biocidability was very good, since even after 10 AATCC washing cycles, the chlorine loading was still higher than 0.05 wt%, the threshold value for a total kill of~10 7 CFU bacteria [46]. Although the lost part of the biocidability caused by leaching of the modifier cannot be regained, the rechargeability that was contributed by rechlorination of the hydrolyzed N-halamine sites was still very satisfactory, since the samples after 10 AATCC washing cycles could be recharged to a chlorine loading of 0.10 wt% with tert-butyl hypochlorite. The good washing stability and rechargeability are attributed to several factors. The first is that the polysiloxane backbone is hydrophobic and has high bond energy, so that the modifier is resistant to washing cycles. The second is that a cyclic N-halamine is more stable than its acyclic counterpart. In addition, this also indicates that the biocidal polysiloxane was interpenetrated to a sufficient depth into the cellulose in scCO 2 , forming a strong interlocking structure. The method of scCO 2 interpenetration does not rely on chemistry and hence is transferable to other substrates, including PET and polyolefins. The data of degradation and rechlorination of biocidal chlorines of cyclic imide Nhalamine polysiloxane@cellulose under UV irritation are summarized in Table 3. The chlorine loading declined rapidly within the first two hours, losing 55% of the initial amount. Then, the decrease became slower, reaching chlorine loadings of 0.02 wt% and 0 wt% at 24 h and 7 d, respectively. This observation of a negative acceleration is believed to be due to the polysiloxane backbone on top surface that is photodegraded to inorganic SiO x , which protected the components beneath from rapid decomposition [47]. The partial photodegradation of the polysiloxane modifier is the reason why the biocidability of the irradiated samples cannot be fully regained. The rechargeability of lost chlorines due to UV exposure was promising, since chlorine loadings of 0.19 wt% and 0.17 wt% for the samples irradiated for 24 h and 7 d were obtained, respectively. The imide N-halamine sites underwent hydrolysis to furnish N−H bonds during storage, due to the adsorption of moisture from air. The formed N−H bonds can be rechlorinated to recover the biocidability. However, the hydrolysis of an N-halamine produces an HOCl molecule that can trigger acid-catalyzed decomposition or oxidation of components of the biocidal layer, leading to a permanent loss of chlorine. Since both washing and storage involve hydrolysis of N-halamine sites, it is naturally expected that the washing-resistant biocidability described above can be also maintained during longterm storage. The test results proved this expectation: the loading of chlorine on cyclic imide N-halamine polysiloxane@cellulose was merely reduced by 9.0%, from 0.31 wt% for fresh samples to 0.28 wt% for the samples stored at 25 • C and 65% RH in an airconditioned chamber for 3 months. Almost all of the lost chlorines in this study could be recharged, which indicates that the loss of chlorines was due to the hydrolysis of N−Cl to N−H instead of irreversible chemical changes. This good storage stability and rechargeability are attributed to the hydrophobic delivery vehicle of polysiloxane, which does not adsorb moisture rapidly, and the cyclic N-halamine structure, which is more stable than an acyclic one. However, more than 85% of acyclic N-halamine sites were lost in only 1 month at the same temperature and RH, when using hydrophilic chitosan as a carrier that quickly adsorbs a large quantity of moisture, to facilitate the hydrolysis of less stable, open-chain N−Cl bonds [19]. Therefore, the biocidal chlorines of cyclic imide N-halamine polysiloxane@cellulose can be preserved for a long period, to offer sufficient antibacterial ability. The biocidal modifier was relatively thin in this study, due to the low solubility of barbituric acid-based polysiloxane in scCO 2 , which only has a poor solvent strength. A thicker modifier is desirable, since it is more damage-resistant and ensures a more durable biocidability. An increase of the interpenetration pressure can partially resolve this problem.

Conclusions
A polysiloxane delivery vehicle was designed to carry cyclic imide N-halamine sites onto cellulose using scCO 2 interpenetration, to offer a powerful, durable, and rechargeable biocidability. For this purpose, a barbituric acid-based polysiloxane was first synthesized using a two-step reaction between barbituric acid and 4-hydroxybenzaldehyde, followed by silane alcoholysis of the resultant sample with of PMHS. The barbituric acid-based polysiloxane was interpenetrated into cellulose in scCO 2 at 50 • C and formed a layer whose thickness increased with the increase of pressure, with a maximum value of 76.5 nm at 28 MPa, showing the ability for adjustable delivery of the control of the loadings of imide N−H and therefore N−Cl. After chlorination of cyclic imide N−H bonds of barbituric acid pendants to N-halamine formats, the final cyclic imide N-halamine polysiloxane@cellulose provided powerful biocidability, killing S. aureus and E. coli with populations of~10 7 CFU in 20 min. The biocidal stability and rechargeability towards washing, UV irradiation, and storage were also promising compared with previously reported cases. These features of biocidability are attributed to the sturdy, photo-resistant, and hydrophobic polysiloxane backbone and to the stable and efficient cyclic imide N-halamine. This study also showed that scCO 2 interpenetration is a benign and general surface modification method, since it does not use a polluting solvent or rely on the chemistry of the substrate. The produced antibacterial cellulose has broad applications in both daily life and the healthcare field, due to its ability to prevent bacterial infection.