Antimicrobial properties of viscose yarns ring-spun with integrated amino-functionalized nanocellulose

The chain-like cellulose nanofibrils (CNFs), with diameters in the 10–70 nm range and length of the micrometer scale, were supplied from the University of Maine, The Process Development Center in the USA. The 100% viscose fibers, of 1.3 dtex and 39 mm long produced by Lenzing AG, were prepared as a compact sliver (of 1.2 ktex and 21.5 Nm) by Litija Spinnery Ltd according to the conventional industrial process (including mixing, carding, drawing, and roving steps) and used throughout the study. Hexamethylenediamine (HMDA, 98% purity) and all other chemicals used were purchased from Sigma-Aldrich Co. Ltd. (USA), and used without further purification.

The functionalization of CNFs with HMDA was performed by the following two-step procedure presented in Fig. 2 (Vivod et al. 2019; Jin et al. 2015): Sodium periodate oxidation of CNFs to dialdehyde (CNF-ald) and further reaction with HMDA through a Schiff-based reaction to obtain CNF-ald-HMDA (marked as CNF-HMDA through the paper). Briefly, 100 mL of CNFs (1 wt%) and 1.9 mmol/g amount of sodium periodate were mixed, and stirred for 48 h in the absence of light at room temperature. The residual sodium periodate was then removed by adding 10 mL of ethylene glycol. The product was dialyzed against deionized water for 3 days using a dialysis membrane with a molecular weight cut-off of 12.000–14.000. 200 mL of CNF-ald suspension (0.5 wt%) was ultrasonicated for 5 min, and then 8 mmol of HMDA was added. The mixture was stirred continuously for 6 h at 30 °C, followed by the in-situ reduction of the resulting imine intermediate at room temperature, employing 0.58 g of NaBH₄. After stirring for 3 h, the product CNF-HMDA was dialyzed (MWCO: 12.000–14.000) against deionized water until it reached a neutral pH, and used as prepared.

A skin irritation test was performed on Albino rabbits (4 animals) according to the ISO 10,993–10:2012 (Biological evaluation of medical devices—Part 10: Tests for irritation and skin sensitization) in the Laboratory for the biocompatibility study, Institute for the Application of Nuclear Energy (INEP), Belgrade, Serbia. The study was approved by the Ethical Committee of INEP for Protection and Welfare of Experimental Animals (No: 02–763/4) and additionally decided by the Ministry of Agriculture, Forestry and Water Economy of the Republic of Serbia-Veterinary Directorate (No: 323–07-11,160). The study fully complies with the ISO 10,993–2 Animal welfare requirements. Three animals were used for test samples and negative control, whereas one rabbit was used as a positive control. Briefly, suspensions of 3 wt% CNF-HMD (total 6 test samples) were applied on fur-clipped skin (the back dorsolateral surface) of rabbits (two samples per rabbit). The application area was about 8 mm in diameter. Two application areas on each rabbit served as the places for the negative control (application of physiological saline). Four different application areas on one rabbit served as the positive control (application of concentrated HCl), following the same procedure. The CNF-HMD suspension and control samples were left in contact with the skin for a period of 2 h, after which the application procedure was repeated on the same places and the samples were left for an additional 2 h. After that, the CNF-HMD samples were removed carefully and washed with physiological saline. All application sites were marked with a permanent marker.

The observation of skin lesions was done for 1, 24, 48 and 72 h after the removal of the test preparations. Skin changes were evaluated for each application site and each time interval. The Irritation Score (IS) and Primary Irritation Index (PII) were determined as suggested by ISO 10,993–10:2012: IS = (oedema + erythema)/number of readings, and PII = Irritation Score/ number of application sites. According to the values of PII, the test samples were classified as non-irritating (0.0–0.4), slightly irritating (0.5–1.9), irritating (2.0–4.9), and strongly irritating (5.0–8.0).

An Airbrush Colani gun (Harder & Steenbeck) with nozzle-exchange capability of different nozzle sizes (0.4–1.2 mm) and air pressure of ~ 3 bars was used to study the sprayability of different CNF-HMDA suspensions and their application on the viscose fiber sliver.

Ring-spinning was performed on a laboratory ring-spinning machine RM350 / D45 (Mesdan Spa, Italy) according to the following procedure: 630 rpm of twisting 10.000 rpm of spindles, and 15.9 m/min of drain speed. The following technological settings were used: 65 mm length of pre-expansion field, 45 mm length of main expansion field, and using a 5.15 mm beige spacer on a cradle and travelers C2RF No. 4 (Reiners + Fürst GmbH, Germany).

The Zeta-Potential and PSD values of native and HMDA-modified CNF suspensions in different media (milliQ water, ethanol, aceton) were carried out on a Zetasizer (Nano ZS ZEN360, Malvern Instruments Ltd., UK) at 25 ± 0.1 °C, using the DTS1070 disposable folded capillary cell, applying the following parameters: A material refractive index of 1.47 (cellulose), dispersion refractive index of 1.33 (milliQ), 1.363 (EtOH) and 1.357 (aceton), and viscosity of 0.8872 cP (milliQ), 1.1734 cP (EtOH) and 0.3084 cP (aceton). A field of 150 V was applied across the nominal electrode spacing of 16 mm. The samples were prepared at concentrations of 0.01 w/v% in Milli-Q water and measured over a pH range from 3 to 11, being adjusted using 0.1 M NaOH and 0.1 M HCl, respectively. Ultrasound was applied to get well dispersed particles before being fed to the measuring zone, and the scattered laser light intensity was quantified in order to calculate the PSD resulting from the particle scattering pattern. The average values were calculated from at least four individual measurements.

The IR spectra of CNF-HMDA, as well as corresponding CNF-HMDA functionalized yarns, before and after the washings, were recorded using a Perkin-Elmer spectrum one FTIR spectrometer with a Golden Gate ATR attachment and a diamond crystal. The spectra were carried out at ambient conditions from accumulating 16 scans at a 4 cm⁻¹ resolution over a region of 4000–650 cm⁻¹, and with air spectrum subtraction performed in parallel as a background. The Spectrum 5.0.2 software program was applied for the data analysis. All the measurements were carried out in duplicate.

Potentiometric titration of CNF water dispersions, as well as yarns functionalized with CNF-HMDA was performed for quantifying the processing-dependant surface charge contribution. The titration was carried out using a dual-burette instrument (Mettler Toledo T-70), equipped with a combined glass electrode (Mettler TDG 117). The burettes were filled with 0.1 M HCl (Merck, Titrisol) and 0.1 M KOH (Baker, Dilut-it). Samples were rinsed in low pH (0.01 M HCl) to convert basic and acidic groups into protonated forms, soaked in Milli-Q water and, afterwards, dried at 40 °C). Titration was carried out at room temperature (23 ± 1 °C) in forward and back runs between pH 2 and 11. From the potentiometric titration data, the molar concentration relating to the overall charge of the weak ions was calculated (arising from the side groups of CNF and HMDA). All the reported values are the mean values of duplicate determinations.

The wetting properties of yarn samples were determined according to the capillary rise method, using the Tensiometer K12 (Krüss, GmbH Germany) and a special sample holder filled with 1 g of densely packed sample. The glass vessel, filled with 75 mL of milliQ water as the wetting liquid, was placed into the moving table, which rose until the tested liquid touched the lower edge of the sample holder. The software (Kruss, version 2.5.0.2305, LabDesk User Interface) monitored the increase in mass of the sample cylinder with respect to the capillary action and the time during the measurement. Since the sample was densely packed as a bundle of capillaries, the modified Washburn equation was used to present the linear dependency of the square height penetration of the testing liquid in the cylinder vs. time, according to the following equations:where n presents the number of micro capillaries and their mean radius (\(\overline{r}\)) (mm) in bulk material, m (g) is the mass of absorbed test liquid, φ (⁰) is the Contact Angle (CA) between the tested liquid and solid sample, while ρ (g cm⁻³) is the density, γ (Nm⁻¹) is the surface tension, and η (Pa) is the viscosity of the tested liquid. The constant c includes the number of microcapillaries and their mean radius of the cylinder, and depends on the measured sample and physical properties of the tested liquid. To determine the constant c, a measurement was carried out with an optimally wetting (spreading) liquid (in our case n-hepane), with a CA (φ) of 0° (cos φ = 1). The obtained c value was used for measuring the weight increase vs. time (mass²/t) using milliQ water, as well as to calculate the CA. The obtained slopes of weight gained due to the milliQ water penetration as a function of time, presents the sample's absorbency rate, while the amount of test liquid uptake (g milliQ/g) in equilibrium expressed their absorbent capacities. The measurements with both liquids were performed at least 5 times for each sample, in order to obtain statistically significant results.

The structure of native and HMDA-modified CNFs, as well as their dispersability in different media and corresponding ring-spun viscose yarns, were characterized by using a low-vacuum microscope (FEI Quanta 200 3D).

The Minimal Inhibitory Concentration (MIC) of CNF-HMDA suspensions and antibacterial efficacies of CNF-HMDA integrated viscose yarns were provided by the National Laboratory of Health, Environment and Food (NLZOH), Maribor, Slovenia, according to the standard test method E2149-10 using Gram-negative (G-) bacterium Escherichia coli (E. coli, DSM 1576), Gram-positive (G +) bacterium Staphylococcus aureus (S. aureus, DSM 799), and skin fungus Candida Albicans (DSM 1386) as testing microorganisms.

The CNF-HMDA suspensions (differently concentrated) or CNF-HMDA functionalized viscose yarns (1 g) were added, or soaked in a testing buffer (0.3 mM potassium phosphate buffer of pH 7.2) for one hour, respectively. Bacterial cell suspensions with a final concentration of 1 × 10⁸/1.5 × 10⁷ Colony-Forming Units (CFU)/mL and fungal cell of 1.5 × 10⁵ CFU/mL, both in 0.3 mM potassium phosphate buffer of pH of 7.2, were prepared and added to the CNF-HMDA suspension or viscose yarn to a final volume of 50 mL. Control experiments were run, during which no samples were present within the microorganism culture, and treated in the same way. After the contact period (1 h for suspensions, 1–6 h for yarns), the samples were submitted to serial dilutions and plated in triplicates. In the case of suspensions, microorganism growth (tube opacity) was read; the last dilution in which microbial growth was no longer detectable is the MIC value. In the case of yarn samples, the CFUs were counted after an overnight incubation at 37 ±  °C, and the mean values of percent reduction (R, %) were calculated using the following equation: R = (N₀ – N)/N₀ × 100, where N₀ and N are the numbers of bacteria/fungal colonies on the samples before and after the test, respectively. The reduction factor (Log (N0/N) was also given.

In addition, the agar diffusion test (ISO 20,645) was performed to evaluate the inhibition zone. The yarn samples (rolled in to small balls) were soaked in a testing buffer for one hour, and then laid on nutrient agar plates inoculated with the test bacteria of 1.2 × 10⁶ CFU/mL in the same buffer media as stated above. The paper disks (6 mm in diameter) with 15 uL of 3.2 wt% CNF-HMDA suspension were used in the same way. After incubation for 24 h at 37 ± 0.5 °C, the plates were examined for bacterial growth directly underneath the sample and above, as well as around its edges, by measuring the diameter of the zone of inhibition in different directions (because the samplers were not perfectly round) and in comparison to the control samples. The experiment was performed in duplicate and the mean value was taken.

The working suspensions were prepared of the selected bacteria, E. coli (2 × 10⁸ bacteria/mL) and S. aureus (2 × 10⁷ bacteria/mL). After 24 h of inoculating yarn substrates with bacteria cells using working bacterial suspensions (as described above), a stock solution (1.5 mL) of Live/Dead BacLight staining reagent mixture (6 µM SYTO 9 and 30 µM PI) was mixed with 1.5 mL of each bacterial suspension, which remained after shake flask testing. The solutions were incubated at room temperature in the dark for 15 min and then each pipetted (100 µL) into separate wells of 96-well flat-bottomed micro-plates, and measured using the fluorescence mission spectrum using a Tecan Spectrophotometer. The viable bacteria (live cells) were stained with SYTO 9 to a green fluorescent color, whereas the bacteria with damaged membranes (dead cells) were stained with PI to a red fluorescent color. By using an excitation wavelength centered at 480 nm, the integrated intensities of the green (G; 530 nm) and red (R; 630 nm) emission were acquired, and the green/red fluorescence ratios (Ratio G/R) were calculated for each suspension. The biocidal effects (%) against both bacteria were determined by calculating the relationship between Live (G, x) bacteria and G/R fluorescence ratio (y) as [y₁ = 0.067x + 0.6635] for E. coli and [y₂ = 0.0177x + 2.1017] for S. aureus. Experiments were performed in duplicate, and the data are shown as the mean value, containing the Standard Deviation.

The biocidal efficiencies of the samples were examined further by viability tests, performed by using confocal microscopy and fluorescence spectroscopy. After 24 h of exposure to the media inoculated with E. coli (10⁵ CFU/mL) or S. aureus (10⁵ CFU/mL), the yarn samples were stained with 3.4 µM of SYTO 9 and 6.0 µM of PI, respectively, for 30 min in the dark at room temperature, washed with Milli-Q water, air-dried, and then visualized by an inverted confocal laser scanning microscope Axio Observer Z1 LSM 710 (ZEISS, Germany) with an EC "Plan-Neofluar" 20x/0.50 M27 objective. Fluorophores were excited using an argon laser (488 nm) and the emitted light was collected through SP 545 and LP 605 filters.

The manufactured yarns were determined by mechanical properties on the automatic Statimat CU (Textechno H. Stein GmbH & Co. KG, Germany) machine at 25 ± 1 °C and 45 ± 1% humidity, monitored with the Texcount software. The results are presented as representative data or mean ± SD values of up to 10 independent measurements.

The washing procedure, which consisted of 10 consecutive cycles, was carried out according to the Standard ISO 105-C01:1989 with and without the addition of 5 g/L washing agent, using the LABOMAT device (W. Mathis AG, Oberhasli, Switzerland). The yarn samples were placed into LABOMAT beakers and the necessary amount of pre-heated (~ 40 °C) milliQ water was added to give a liquid:sample ratio of 50:1. The samples in beakers were placed into the machine and washed at 40 rpm for 30 min at 40 ± 2 °C. After each cycle the samples were taken out of the beaker, squeezed, put back in the beaker, and covered with fresh milliQ water. After 10 washing cycles, the samples were opened-out manually, laid on the filter paper and dried out at room temperature.

The testing of 3 wt% CNF-HMD suspension, performed on Albino rabbits (Fig. 1) according to the ISO 10,993–10:2012, can be classified as skin non-irritating (PII = 0.0), identically as the negative control (physiological saline), while positive control (concentrated HCl) has been classified as irritating (PII = 3.2).

The spraying process was first optimized in terms of CNF-HMDA dispersibility using different CNF-HMDA concentrations, suspension mediums and sizes of nozzle openings. The effect of CNF-HMDA dispersibility in fast-volatile mediums was thus analyzed by Zeta-Potential (ZP) and size-distribution values, as well as by SEM imaging of samples sprayed on Aluminum foil.

In order to identify the charge contribution from CNF and HMDA on dispersibility, the CNF suspension was also analyzed by potentiometric titration after different modification steps (Fig. 2), i.e. the oxidation (1st step) with sodium periodate (NaIO₄), forming a pair of aldehyde groups (CNF-ald) by cleaving the C2–C3 bond of anhydro-glucopyranoside ring, and the binding of HMDA (2nd step) through a Schiff-base reaction, yielding CNF-ald-HMDA. The titration curve of native CNF shows one small bend at pH 4.4, given the small quantity (0.12 mmol/g) of negative charge that may be related to the presence of rare anionic (preferably sulphonate and carboxylic) surface groups, formed during the preparation of CNF or resulting from the presence of lignin residues. These groups were reduced to 0.096 mmol/g after the oxidation (CNF-ald), also giving another peak at pH ~ 9.1, which might be related to the aforementioned phenomenon, while they could not be detected anymore after attachment of HMDA (CNF-ald-HMDA), yielding a positive charge of around 5.64 mmol/g at pH ~ 8.1. The titration curve of pure HMDA (inserted chart) shows high positive charge at a wide pH range until ~ 10.7 (being related to its complete deprotonation, pK) with a small band between pH 6.0–7.0 (representing the partial deprotonation stage of diamines, H₂N-(CH₂)₆-NH₂), thus confirming the high contribution of amino (CNF-NH₂) groups. A gradual reduction of the titration curve for the CNF-ald-HMDA sample and shifting of HMDA’s pK value towards lower pH may, thus, be because of its binding to CNF-ald. In order to identify the efficacy of the HMDA attachement on aldehydes, the sequential/second-step oxidation of CNF-ald was also performed with sodium chlorite (NaIO4) (CNF-ald-COOH; Liimatainen et al. 2012) which, however, gave a much smaller amount of carboxylic acid groups (~ 3.44 mmol/g) as would be expected stoichiometrically. This might be due to the incomplete oxidation of aldehydes, because the aldehyde content on CNF-ald determined by hydroxylamine hydrochloride (HONH₂·HCl) according to (Kim et al. 2004) was quantified to be around 6.1 mmol/g. This confirms the relatively high degree of HMDA binding to CNF-ald, which is preferentially a one-side (grafting) and rarely both-sides (crosslinking) reaction, being also supported by the good dispersibility of CNF-ald in aqueous media, although the attached HMDA resulted in the formation of CNF-HMDA aggregates, as revealed by the SEM images (Fig. 2).

The Zeta-Potential (ZP) values and the size distribution of CNF-HMDA in different media, presented in Fig. 3, clearly indicate an increase of ZP from about + 18 mV (in absolute water) to about + 20 to + 30 mV (in the case of ethanol addition from 12 to 50%, respectively) and its inversely proportional behavior in the case of acetone (from about + 35 mV to + 33 mV and + 18 mV by addition of 12, 25 and 50% of acetone, respectively), as well as the consequent increase or decrease of its average size distribution, being related to its aggregation phenomenon, as already discussed above. A better dispersion of the CNF-HMDA in a bipolar medium is attributed to the good compatibility of the hydrophobic ethylene groups from HMDA bound to the CNFs with ethanol/acetone and simultaneously ionized amino groups. However, the acetone, as a ketone without direct OH groups, cannot form hydrogen bonds and, thus, does not have the ability to attach to the functional groups of CNF-HMDA, as compared to the ethanol, which may, thus, also act as a better dispersing medium from the point of view of temporal stability.

The CNF-HMDA aqueous suspension deposition patterns, presented by the SEM imaging in Fig. 4, confirmed these findings by showing the difference in the amount and homogeneity of CNF-HMDA sprayed on aluminum foil by using different water mediums, as well as the ability to use lower nozzle-size openings without their clogging. While 1.5 wt% was the highest concentration for water, suspended CNF-HMDA enabled its spraying with at least 0.8 mm nozzle openings; a bipolar water–acetone or water–ethanol medium enabled its spraying with up to 5 wt% concentrated CNF-HMDA and using nozzles with up to 0.4 mm size openings. As can be seen from the SEM images, the amount of CNF-HMDA deposition could be increased significantly, ensuring good dispersibility with relatively high homogeneity, among which water solutions with the addition of 25–35% ethanol or 12% acetone were selected as the optimal, and were used at the spray deposition of CNF-HMDA to the fiber sliver.

The viscose slivers were sprayed with known amounts of differently concentrated and suspended CNF-HMDA (Table 1), and air-dried before being ring-spun into a functional yarn. The chemical (FTIR), surface (charge, hydrophilicity), wetting and structure-morphological properties, fineness and tenacity of the successfully spun viscose yarns were assessed to get information about the yarns` properties.

As can be seen from the results collected in Table 1 and presented graphically in Fig. 5, the spinning from the slivers sprayed with relatively low deposits (volume and conc.) of highly agglomerated water-suspended CNF-HMDA, and the use of the smallest (0.8 mm) nozzles to spray, resulted in a deposition of up to a max 132.5 mg/g (weight increase of 13.3%) CNF-HMDA (Fig. 6), which, however, decreased the yarn`s tenacity by about 30% (from 19.42 to 13.6 cN/tex) and elongation at break by about 23% (from 14.4 to 11.1%) according to the REF yarn. Comparatively, between 50–100 mg/g of CNF-HMDA can be applied using 1 wt% of water suspended CNF-HMDA, without significant reduction of the yarn`s mechanical properties. In contrast, a relatively higher amount of CNF-HMDA (up to 160 mg/g, resulting in 16% of weight increase) could be integrated with CNF-HMDA suspended in 35% ethanol or acetone. In this case, the tenacity was reduced by about 19.6% (to 15.61 cN/tex) and the elongation at break by about 7% (to 13.39%), given less deviation when 5 wt% CNF-HMDA suspended in 12% of acetone was used. Besides better dispersibility of CNF-HMDA among the fibers, faster drying kinetic of CNF-HMDA suspended in bi-polar mediums (where acetone evaporates even faster than ethanol despite its higher surface tension) during spraying, as compared to the absolute water, is, thus, another parameter reducing viscose fibers` gluing, and enabling smoother spinning, that gives a finer yarn (REF-acetone/ethanol) without affecting its tenacity. The reduction of both of these parameters in the case of CNF-HMDA functionalized yarns are, thus, the result of its concentration, integration and distribution among the fibers, as can be observed from the SEM images presented in Fig. 6.

The potentiometric titration curves of differently prepared yarns (Table 1), performed to follow the charge contribution from CNF-HMDA, follows the same, though much less intense, trend as pure CNF-HMDA (Fig. 2). The values of total and amino functional groups were increasing with increasing of the CNF-HMDA content, given ~ 0.079 and ~ 0.012 mmol/g (at 132.5 mg/g content), and even ~ 0.663 and ~ 0.439 mmol/g (at 160 mg/g content) of charges when sprayed from aqueous and faster evaporating aqueous-acetone mediums, respectively, and using the optimal spraying conditions (i.e. concentration of CNF-HMDA and nozzle size). The effect of washing resistance of such a yarn (Fig. 7), studied by 10 consecutive washings, showed that the use of detergent reduced the charge quantities significantly (~ 72%) (to ~ 0.018 and ~ 0.01 mmol/g), which was even more pronounced when the yarn was washed without it (to ~ 0.084 and ~ 0.024 mmol/g). The results indicate possible, preferably hydrophobic, interactions of surfactant molecules from the detergent with in-yarn surface integrated CNF-HMDA (as already confirmed in our other study, Vivod et al. 2019), which most likely prevented the excretion of CNF-HMDA from the yarn during washing as compared to washing it with pure water.

The FTIR spectra of viscose yarns prepared with and without the addition of CNF-HMDA, as well as after 10-times washing cycles performed with and without the addition of detergent, were performed, to identify its durability (Fig. 7). The spectrum for pure viscose yarn (d) shows a characteristic wide O–H stretching band centered at around 3330 cm⁻¹, C–H stretching centered at around 2893 cm⁻¹ and C–H bending around 1460 and 1370 cm⁻¹, an H–O–H bending band of absorbed water at around 1638 cm^–1, and strong C–O–C stretching vibrations` band of the glucoside ring in the spectral region ranging from 1150 cm⁻¹ to 800 cm⁻¹ representing the fingerprint of the cellulose (Peets et al. 2019). A decrease of intensity of O–H stretching and C–O stretching bands and their shifting to about 3304 cm^–1 and 1032 cm⁻¹, respectively, the splitting of the C–H band to asymmetric and symmetric stretchable CH₂ vibrations (at around 2925 cm^–1 and 2857 cm^–1) with the appearance of symmetrical CH₂ bending and CH₂ wagging vibrations at about 1459 cm⁻1 and 1314 cm^–1, respectively, all corresponding to the methyl chains from HMDA. The appearance of N–H bending vibrations for amino groups in the region of 1540–1580 cm⁻¹, can be also observed in the yarn with integrated CNF-HMDA (a), which supports the fact of CNF-HMDA integration, as well as its interactions with the cellulose units of viscose by the creation of new inter- and intramolecular hydrogen bonds (Vivod et al. 2019). The HMDA assigned bands in the case of washed yarns (b and c) were reduced, while cellulose (O–H, C–H and C–O) related bands were increased, the effect being more pronounced when using detergent (c).

The wettability of differently prepared yarns, studied by the kinetic of milliQ water absorbency and its capacity (Fig. 8), also exhibited a corresponding dependency on the yarns` preparation, correlated well with the results of titration and spectroscopy analysis. It is well-known that the wettability of the fibers is induced by the hydrogen bonding capability with the liquid, thus depending on its surface properties (charge, hydrophobicity), crystallinity structure and density of the fibers, as well as the temperature and pH of the liquid. The absorbency curve of the reference yarn sample (with no CNF-HDMA content) showed the slowest absorbency rate, meaning that this sample needed the longest period for establishing the equilibrium (i.e. 120 s), while the amount of absorbed milliQ was the highest (~ 1.67 g/g of dry sample, inserted Table) compared to the CNF-HDMA integrated yarns. Contrarily, the yarn containing the highest (160 mg/g) amount of CNF-HDMA, showed the highest aborbency velocity (indicated by the steeper curve), since the equilibrium was established in the shortest time (~ 40 s) compared to the other samples containing CNF-HMDA (43–58 s), but it had the lowest value of adsorbency (~ 1.53 g/g). The other samples (containing 33.3, 66.7, 100, and 133 mg/g CNF-HDMA) showed more or less similar wetting behavior, i.e. similar absorbency velocity, while absorbing bewteen 1.66 and 1.53 g/g of milliQ water. On the other hand, the reference sample showed the highest Contact Angle (CA, ~ 88.8º), while the amount of added CNF-HDMA decreased the CA values from ~ 87.8º to ~ 73.3º, being the lowest for the sample containing the highest amount of CNF-HDMA (160 mg/g), and which can be described as the most surface hydrophilic, given the highest aborbency velocity. As the CA depends on the difference between the surface tension of the material and the surface tension of the liquid (in our case water with very high surface tension, ~ 72.8 mN/m), and the adsorption depends on the difference between them (i.e. the greater the difference between surface tensions, the lower is the adsorption), we can concude that the addition of CNF-HDMA (bearing highly polar amino groups) to viscose yarn contributed importantly to its wettability.

It is obvious that CNF-HMDA functionalized viscose yarn absorbs milliQ water kinetically faster compared to the pure viscose yarn, which may be related to the presence of amino functional groups, which, next to the free hydroxylic groups arising from CNF, enchanced the yarn`s surface hydrophilicity (being seen in the relatively lower CA values), and, thus, faster hydration of the surface. However, on the other hand, the water`s absorption ability was decreased quantitatively by increasing the CNF-HMDA content, which can be due to various reasons. Deprotonated CNF-HMDA’s amino groups may mitigate the electrostatic repulsion with the present rare anionized sulpho- and carboxylic- groups at CNF, and with the presence of hydrophobic interactions between the alkyl chains of the HMDA, reducing the reactive groups coming into contact with water molecules and reducing or deswelling the yarn. The effect of the CNF-HMDA present, i.e. its orientation and distribution through the yarn, is also reflected in the yarn`s morphological structure (i.e. a more open yarn structure with less densely packed and oriented viscose fibers, which can be seen on the SEM images presented in Fig. 6), and reduced fineness (Table 1, Fig. 5), which also influenced the CA values, suggesting that pure viscose yarn is less hydrophilic in nature compared to the yarns containing CNF-HMDA.

The antimicrobial activity of CNF-HMDA suspension and CNF-HMDA integrated viscose yarns was assessed using different methods in order to define their efficacy under different conditions.

The Minimal Inhibitory Concentration (MIC) of CNF and CNF-HMDA suspensions was tested first. The CNFs cannot target and disturb the cell wall, cell membrane, or active enzymes of bacterial and fungal strains, so haven`t shown any antimicrobial activity. However, HDMA functionalized CNFs exhibited good antibacterial and antifungal effects against all selected microorganisms associated with inflammatory-related skin diseases: Gram-positive bacterial (G +) S. aureus, Gram-negative (G-) bacteria E. coli, and skin fungus C. albicans, given MIC of ~ 0.012 wt%, ~ 0.193 wt% and ~ 0.024 wt %, respectively. Besides of a hydrophobic tail, the contact of amino groups might, namely, play an essential role in the antimicrobial activity, which is related to the positively charged cationic nature (Fig. 2), leading to an increase in the chance of the interaction with the negatively charged cell walls of the microorganisms.

The ability of CNF-HMDA suspension and (133 mg/g) CNF-HMDA integrated viscose yarns (washed and 10-cycles washed without detergent) to inhibit selected bacterial growth after 24 h of incubation was tested by the agar diffusion method and using both bacteria. The results, presented on Fig. 9, showed a clear zone of inhibition around the highly-concentrated suspension of CNF-HMDA, as well as viscose yarns containing CNF-HMDA, with no bacterial growth on and under the yarns` top surfaces, while the control yarn showed no sign of inhibition. The inhibitory effect of CNF-HMDA suspension was, in diameter, higher against S. aureus (~ 8 mm) compared to E. coli (~ 10 mm), while the yarns showed a slightly different effect (~ 14 mm of inhibition for S. aureus and ~ 16 mm for E. coli) being more pronounced but less effective (~ 13 and ~ 15 mm, respectively) for the 10-cycles washed samples. Although this testing indicated that samples have an antimicrobial activity potential, it only implied that bacteria have been prevented from growing on the yarn surfaces (i.e. by contact), and potentially, also due to the slow release of CNF-HMDA onto the fiber surface and/or from the surface during incubation (i.e. by diffusion).

In order to get information on the surface-adsorption ability effect on the microorganisms` inhibition, the samples were incubated at different time-intervals (1–6 h) and the percent of their reduction (R, %) was evaluated, as well as the reduction factor (Log R). As seen from the Fig. 10 a-c, an increased content of CNF-HMDA in the yarn increased bacteria reduction, although it also gave different effects depending on the microbial profile. The powerful effect of CNF-HMDA functionalized yarn on the Gram-positive S. aureus rather than the Gram-negative E. coli bacterium may be due to the difference in the composition of the cell wall on the one hand, and the effect of the surface charge (amination and overall hydrophilicity) of the yarn on the other. Gram-negative bacteria have a thick layer of phospholipids rather than peptidoglycans when compared to Gram-positive bacteria, which have a thick layer of peptidoglycans. The negative charge on the phospholipids have, thus, enhanced the adhesion power of the polycationic polymer of the cell wall. A slightly smaller effect on the antimicrobial activity for E.coli could, thus, be explained by an increase in the distance between the attached amino groups and the CNF backbone (Mohy Eldin et al. 2012), which, however, still yielded the 80–99% reduction of bacteria within 1–3 h at the highest CNF-HMDA content (133 mg/g), and a given log reduction in between 0.69–2.95. A similar effect on the reduction of S.aureus was obtained by using a yarn containing only 33 mg/g of CNF-HMDA, while its further increase already reduced the log reduction significantly to a value of around 4.0 within 3 h.

The durability of the yarn’s antibacterial effect after washing was also performed using yarn functionalized with 160 mg/g of CNF-HMDA, before and after 10-cycles washing being performed, with and without detergent. As shown in Fig. 10 d, the functionalized yarn still provided up to 80% reduction of E. coli after 10-cycles of washing without detergent, while showing no effect in the case of its washing in the presence of detergent. This may be due to the presence of the residues of non-ionic surfactants on the yarn surface after washing (Kaisersberger-Vincek et al. 2017), which are interacting hydrophobically with the hydrophobic sites of HMDA, thus blocking its functional amino groups. This fact can be supported by the washing of unfunctionalized yarn with the surfactant, showing a small reduction of bacteria, indicating that CNF-HMDA inhibits the growth of bacteria in contact with the surface.

The bactericidal effect, determined after 24 h of yarns` incubation, coincides with the above findings. The data inserted in the graphs in Fig. 11 show an increase in bactericidal effect as a result of the relative decrease in green fluorescence intensity when yarn was functionalized with 133 mg/g of CNF-HMDA. These values for Gram-positive S. aureus were comparatively higher (~ 45% for both unwashed and 10-times washed) than those for Gram-negative E. coli bacterium (given ~ 21% for unwashed sample only), while showing no effect in the case of the reference samples (without CNF-HMDA content). However, as already pointed out in one of our previous studies (Kaisersberger-Vincek et al, 2017), these results need a critical evaluation, due an in-effective staining of some intact Gram-negative bacteria with SYTO9, being related to its permeability problems through the double-membrane layer or its exporting from the bacteria cytoplasm, as well as a decrease in green fluorescence intensity due to the entrance of PI.

The confocal fluorescence microscopy imaging of the yarn samples was also performed, despite the fact that the combined fluorescent staining may lead to results with some deviations, above all as the viscose fibers themselves were also colorized. In Fig. 11, the presented representative fluorescence images after exposure to the bacteria show green labeled cells on the surface of all the tested samples, being more intensive on the reference sample when compared to the yarns containing CNF-HMDA, where the presence of a high density of red bacteria (above all in the case of S. aureus) indicates that CNF-HMDA may interact with bacteria’ cytoplasmic membranes. A diffusion of the CNF-HMDA from the yarn on its surface may be also seen on the washed samples.