Valorisation of Waste Guanidine Thiocyanate into a Guanidinium-Modified Silica Sol/Poly(Vinyl Alcohol) Adhesive: CuCl₂/Cu⁰ Synproportionation Route, Colloidal Stability and Antibacterial Performance

1. Introduction

Waterborne adhesives based on poly(vinyl alcohol) (PVA) and poly(vinyl acetate) (PVAc) dominate non-structural bonding of wood, paper and mineral materials because they are inexpensive, solvent-free and easy to apply [1,2,3]. Their main drawbacks are limited water resistance and the absence of intrinsic biostability: hydrophilic glue lines absorb moisture and can support microbial growth, which is undesirable in furniture, packaging and building elements used in healthcare and food-related environments [4,5]. Two strategies are commonly used to overcome these limitations—crosslinking or hybridisation of the polymer matrix with inorganic nanofillers, and incorporation of biocidal components [6,7,8,9].

Inorganic–organic hybrids of PVA and colloidal silica combine the film-forming ability of the polymer with the stiffness and barrier effect of nanosized SiO₂, which improves strength, thermal stability and transparency [8,9,10,11]. The performance of such hybrids, however, depends critically on the colloidal state of the filler: aggregated silica forms weak boundary layers and micro-defects in the glue line [12,13,14]. Silica sols are electrostatically stabilised at pH 8–10, gel most rapidly at pH 5–7 and are very sensitive to multivalent and organic cations [15,16,17]. Any cationic additive introduced into a silica-containing adhesive therefore alters the charge balance of the particles and, through it, the processability of the formulation [18,19,20,21].

Guanidinium-containing compounds are among the most effective non-leaching antimicrobial agents. The guanidinium cation (pK_a ≈ 13.6) remains protonated over the whole practical pH range [22,23]; it forms multiple hydrogen bonds through six N–H donors [24] and interacts electrostatically with the negatively charged bacterial membrane [25,26,27]. Poly(hexamethylene guanidine) (PHMG) and related polymers have been incorporated into PVA films and hydrogels [28,29,30], acrylic coatings [31], polyester and polyolefin fibres [32,33], wound dressings [34] and dental adhesives [35], typically giving broad-spectrum activity against Gram-positive and Gram-negative bacteria [36,37,38,39,40,41]. Polymerisable guanidinium salts, including guanidinium (meth)acrylates, are attractive because the cation can be anchored to the matrix through the vinyl group [42].

Commercial guanidine derivatives are produced from urea and ammonium nitrate or from cyanamide at elevated temperature. Guanidine thiocyanate (GT), [C(NH₂)₃]SCN, which is formed in thiocyanate-based guanidine routes (e.g., fusion of dicyandiamide with ammonium thiocyanate) and accumulates as a solid residue at guanidine-salt plants, is therefore a potentially cheap source of the guanidinium cation [43]. Its direct use is hindered by the thiocyanate anion: SCN⁻ is toxic to aquatic organisms, persistent in effluents and difficult to destroy biologically or electrochemically [44,45,46]. Moreover, thiocyanate is easily oxidised by nitric acid and by Cu²⁺ ions; under uncontrolled conditions these reactions generate nitrogen oxides, thiocyanogen, hydrogen cyanide and other hazardous products. A valorisation route must therefore remove SCN⁻ quantitatively, under mild conditions and preferably in the form of a useful product. Ion exchange with nitric acid, the most obvious way to replace the anion, is unattractive for this reason: the oxidising acid and thiocyanate have to be kept apart kinetically, and the resulting guanidinium nitrate is an energetic salt. Copper(I) thiocyanate is such a product: it is an insoluble, wide-band-gap p-type semiconductor used as a hole-transport layer in perovskite and organic solar cells [47,48,49] and as a biocide in antifouling paints [50].

Waste-to-adhesive approaches have been reviewed mainly for biomass-derived proteins, lignocellulosics and wood waste [5,51,52]; reports on the use of nitrogen-rich industrial residues as functional adhesive modifiers remain scarce. Related studies have shown that the performance of formulations derived from secondary industrial streams is often governed by the state of the dispersed phase rather than by the chemistry of a single component [53,54]; the present study examines this for a guanidinium-modified silica sol/PVA adhesive. In this work we (i) develop a one-step CuCl₂/Cu⁰ synproportionation route that removes the thiocyanate anion of waste GT as β-CuSCN and leaves the guanidinium cation in solution as guanidinium chloride (GuCl), without nitric acid or other strong oxidants, (ii) characterise the product obtained by reacting GuCl with acrylic acid and critically assess the extent of acrylate exchange by speciation analysis and FTIR, (iii) analyse the colloidal state of silica sol in the PVA-based formulation by ζ-potential, particle size and DLVO calculations, (iv) evaluate the rheological, adhesive, thermal, water-resistance and antibacterial properties of the resulting composite adhesive in comparison with an unmodified PVAc reference and with requirements for commercial adhesives, and (v) establish the material balance and raw-material cost of the process. To the best of our knowledge, this is the first route in which both ions of a guanidine thiocyanate waste are converted into value-added products, the guanidinium cation into a functional adhesive modifier and the thiocyanate anion into β-CuSCN.

2. Materials and Methods

2.1. Materials

Waste guanidine thiocyanate (technical grade, solid residue of guanidine-salt production; Republic of Uzbekistan, Navoiy region, “Navoiyazot” JSC) was used as received after determining moisture by drying; its identity was confirmed by FTIR (Section 3.1). Copper(II) chloride dihydrate (0.1–0.5 M aqueous solutions), metallic copper (turnings/powder, ≥ 99%), acrylic acid (99%, stabilised with hydroquinone monomethyl ether), PVA (degree of hydrolysis 98–99%), epoxy-dian resin ED-20 with polyethylene polyamine (PEPA) hardener, formalin (37% HCHO) and an aqueous colloidal silica sol (30 wt% SiO₂; particle diameter 12–22 nm; specific surface area 124–227 m² g⁻¹; Na⁺-stabilised, pH 9–10) were of technical or analytical grade. Silver nitrate and iron(III) chloride hexahydrate (analytical grade) were used for qualitative tests for chloride and thiocyanate. Distilled water was used throughout; KNO₃ (1 mM) served as the background electrolyte in electrokinetic measurements.

2.2. Two-Stage Preparation of the Guanidinium Modifier

Stage I (synproportionation). Copper turnings were added to a 20 wt% aqueous GT solution in a jacketed reactor, and a 0.1–0.5 M CuCl₂ solution was introduced dropwise at 20–30 °C under stirring at atmospheric pressure until precipitation ceased (molar ratio GT : Cu²⁺ : Cu⁰ = 2 : 1 : 1). Thiocyanate is precipitated as CuSCN, whereas the guanidinium cation remains in solution as GuCl:

2[C(NH₂)₃]SCN + CuCl₂ + Cu → 2CuSCN↓ + 2[C(NH₂)₃]Cl

(1)

Metallic copper reduces Cu²⁺ to Cu⁺, so that thiocyanate is precipitated by synproportionation rather than oxidised, and no acid is formed:

Cu²⁺ + Cu⁰ + 2SCN⁻ → 2CuSCN↓

(2)

The precipitate was separated on a vacuum filter, washed with distilled water until the test for chloride (AgNO₃) was negative, dried at 60–80 °C to constant mass and weighed. Residual thiocyanate in the filtrate was checked with FeCl₃. The GuCl filtrate was transferred directly to stage II.

Stage II (reaction with acrylic acid). Acrylic acid was added to the GuCl filtrate at a molar ratio of 1 : 1, and the mixture was stirred continuously at 20–40 °C without any neutralising agent:

[C(NH₂)₃]Cl + CH₂=CHCOOH ⇄ [C(NH₂)₃][CH₂=CHCOO] + HCl

(3)

The temperature was kept below 40 °C because spontaneous radical polymerisation of acrylic acid with a sharp viscosity rise was observed above this limit. The product of stage II is denoted below as the “GA modifier”; its actual composition is discussed in Section 3.3. The processing scheme is summarised in Figure 1.

2.3. Preparation of the Composite Adhesive

The base formulation (variant A) contained 40 g of GA modifier, 100 g of PVA (dry polymer, dissolved in water), 15 g of 30% silica sol and 5 g of formalin. PVA was dissolved in water at 80–95 °C and cooled to 20–40 °C before the modifier was added, to avoid uncontrolled polymerisation of acrylic species during mixing; the GA modifier was then added, the silica sol was introduced dropwise over 30–40 min, and formalin was added last. The mixture was heated at 15–20 °C h⁻¹ to 50–90 °C and held for 60–90 min at pH 4.5–7.0. The silica sol content was varied from 5 to 25 g and the modifier content from 0 to 60 g. In variant B, ED-20 resin was used as the matrix; the modifier and silica sol were dispersed at 200–250 rpm and PEPA (13–14 parts per 100 parts of resin) was added last. Liquid samples were used for colloidal and rheological measurements; films cast at equal thickness and dried under standard conditions were used for mechanical, thermal and antibacterial tests.

2.4. Characterisation

FTIR spectra were recorded on a Shimadzu FTIR spectrometer (IRTracer-100, Shimadzu Corporation, Kyoto, Japan) in attenuated total reflectance mode (400–4000 cm⁻¹, resolution 4 cm⁻¹); bands were assigned according to Nakamoto [55]. The phase composition of the CuSCN precipitate was determined by X-ray diffraction (Cu K_α radiation, Malvern Panalytical Empyrean, 2θ = 10–75°). DTA/TGA was performed on a Shimadzu thermal analyser (Shimadzu DTG-60H) in nitrogen at ≈10 °C min⁻¹ from 30 to 600 °C (sample mass 4.10 mg).

The hydrodynamic diameter D_h was measured by dynamic light scattering (Malvern Zetasizer Nano ZS). The ζ-potential was determined by moving-boundary macroelectrophoresis in a U-tube: samples were diluted to 0.05% SiO₂ with 1 mM KNO₃, which also served as the side liquid; the voltage U between the electrodes was controlled with a digital voltmeter, and the boundary velocity u was measured with a millimetre scale and a stopwatch. The field strength E = U/L (L is the inter-electrode distance) and the electrophoretic mobility u_E = u/E were calculated, and each measurement was repeated three times with reversed polarity. Because the Smoluchowski condition (κa ≫ 1) is not satisfied for silica particles of radius a ≈ 10 nm in 1 mM electrolyte (κa ≈ 1.04), mobilities were converted to ζ-potentials with Henry’s equation using Ohshima’s approximation for f(κa) [56]; the Smoluchowski equation would underestimate ζ by a factor of 1.45 under these conditions (Table 1) [57,58]. Because the modifier raises the ionic strength of the diluted samples (≈8.5-fold dilution of the formulation; I ≈ 1–42 mM), f(κa) was evaluated for each sample at its actual ionic strength rather than at that of the background electrolyte.

Contact angles of the liquid formulation were measured by the sessile-drop method (5 µL, 23 ± 1 °C, 50 ± 5% RH) [59]; for wood the spreading diameter was also recorded because of penetration and moisture effects [60,61,62]. The surface free energy of solids was resolved into dispersive and polar components by the Owens–Wendt method [63], and the work of adhesion was calculated with the Young–Dupré equation, W_a = γ_L(1 + cos θ). Flow curves were recorded on a rotational rheometer and fitted with the Ostwald–de Waele model; thixotropy was evaluated from the hysteresis-loop area [64,65].

Lap-shear strength was determined according to ISO 4587 [66] on metal, wood, glass and ceramic substrates (at least three specimens per series), and the failure mode was recorded [67,68]. Tensile and compressive strength of cured films were measured on Shimadzu AGS-X (Shimadzu Corporation, Kyoto, Japan). Water resistance was evaluated from the mass gain after 24 h immersion and from the retention of strength. An unmodified PVAc adhesive served as the reference. Antibacterial activity was determined by the agar diffusion method against Staphylococcus aureus ATCC 25923 and Escherichia coli ATCC 25922; the modifier-free formulation served as the control. To assess wash resistance, samples were washed in running water for 2 h and re-tested [30,32].

2.5. Calculations

The acrylate fraction was calculated as α(A⁻) = 1/(1 + 10^{pKa − pH}) with pK_a = 4.25 for acrylic acid [69]. Ionic strengths were calculated for a 1 : 1 electrolyte, and the Debye length as κ⁻¹ = 0.304/√I nm (25 °C). Particle interaction energies were estimated within the DLVO framework [70,71]: the van der Waals term was calculated with the complete Hamaker expression for two spheres (R = 10 nm, A = 0.8·10⁻²⁰ J for SiO₂–water–SiO₂ [72]), and the double-layer term with the linear superposition approximation, the diffuse-layer potential being obtained from the Grahame equation for an effective surface charge of −0.010 C m⁻² [73,74].

3. Results and Discussion

3.1. Stage I: One-Step Removal of Thiocyanate and Formation of Guanidinium Chloride

The FTIR spectrum of the waste GT (Figure 2a) is dominated by an intense, narrow ν(SCN) band at 2051 cm⁻¹ (transmittance down to 28%), accompanied by ν(C–S) at 721 cm⁻¹ and a weaker band at 634 cm⁻¹. The broad N–H stretching region (3363–3165 cm⁻¹), the protonated amine band at 3019 cm⁻¹, ν_as(CN₃)/ν(C=N) at 1666 cm⁻¹ and the N–H/C–N bands at 1412 and 1378 cm⁻¹ confirm the guanidinium cation; the width of the N–H envelope reflects an extensive intermolecular hydrogen-bond network typical of guanidinium salts [24]. Since the 2000–2100 cm⁻¹ region is free from other absorptions, the 2051 cm⁻¹ band is an unambiguous marker of thiocyanate and was used to monitor its removal.

Stage I combines the removal of thiocyanate and the formation of a soluble guanidinium salt in a single operation (Eq. 1, Table 2). On dropwise addition of CuCl₂ to the GT solution containing copper turnings, a white precipitate formed immediately, and the filtrate gave a negative FeCl₃ test for thiocyanate at the end of the addition. Because reaction (1) consumes no proton and releases no acid, the guanidinium cation remains in solution as GuCl and the filtrate can be transferred to stage II directly, without crystallisation of an intermediate salt and without neutralisation of an acidic effluent. This is the principal advantage over ion exchange with nitric acid, in which oxidation of thiocyanate has to be suppressed by cooling and the guanidinium salt has to be recovered by crystallisation from a solution in which it is appreciably soluble.

The FTIR spectrum of the stage II product (Figure 2b) contains no absorption in the 2000–2100 cm⁻¹ region and no ν(C–S) band at 721 cm⁻¹, which, together with the negative FeCl₃ test of the filtrate, confirms that thiocyanate is removed in stage I; the guanidinium cation, in contrast, is retained through all stages (Table 5).

3.2. Thermodynamics of Synproportionation and Phase Composition of β-CuSCN

Direct precipitation of thiocyanate with Cu²⁺ is accompanied by oxidation of SCN⁻, because Cu²⁺ is a sufficiently strong oxidant towards thiocyanate. In the presence of metallic copper the reaction proceeds by synproportionation (Eq. 2). The standard potentials of the Cu²⁺/CuSCN (E° = +0.91 V) and CuSCN/Cu⁰ (E° = −0.23 V) couples [69] give ΔE° = +1.14 V and lg K = 19.3 for reaction (2), so that precipitation is practically complete. Because oxidation of SCN⁻ by Cu²⁺ is itself thermodynamically allowed (E°((SCN)₂/SCN⁻) ≈ +0.77 V < E°(Cu²⁺/CuSCN) = +0.91 V), the role of metallic copper is kinetic: it consumes Cu²⁺ faster than Cu²⁺ can oxidise thiocyanate. The precipitate is white; a white or light-grey colour is characteristic of copper(I), whereas copper(II) compounds are blue or green. The yield of CuSCN was 92% of the theoretical value (94.7 g per 100 g of GT; theoretical 102.9 g).

The XRD pattern of the dried precipitate (Figure 3) shows reflections at 2θ = 16.2 (003), 27.2 (101), 28.8 (012), 32.6 (006), 34.6 (104), 47.1 (110) and 49.9° (113), which correspond to rhombohedral β-CuSCN, the polymorph used in optoelectronic applications [47,48,49]. Dropwise addition of CuCl₂ is essential: at high supersaturation nucleation outruns crystal growth, and a fine, poorly filterable precipitate is formed. The CuSCN washings, which contain small amounts of Cu²⁺ and Cl⁻, are the only aqueous effluent of the process (Section 3.7).

3.3. Stage II: Composition of the GA Modifier

Reaction of the GuCl filtrate with acrylic acid at 20–40 °C gave a colourless to pale-yellow, viscous liquid containing fine white crystals. Its high viscosity already suggested partial polymerisation of acrylic acid. The direction of reaction (3) can be estimated from the acid strengths: K = K_a(AA)/K_a(HCl) ≈ 7·10⁻¹¹ (pK_a(HCl) ≈ −5.9 [93]), which corresponds to ΔG° ≈ +58 kJ mol⁻¹. Because no acid is formed in stage I, the acidity of the medium is determined by acrylic acid itself: for an equimolar 2 M GuCl/acrylic acid solution the calculated pH is 1.98, and only 0.53% of acrylic acid is present as the acrylate anion (Figure 4, Table 3). Removing water facilitates crystallisation but cannot shift the equilibrium substantially, since its position is governed by the acid strengths.

The FTIR spectrum of the stage II product (Figure 2b, Table 4) fully agrees with this estimate. The strongest bands are ν(C=O) of non-ionised COOH at 1709 cm⁻¹ and the C–O/δ(OH) band at ≈1160 cm⁻¹, while δ(CH₂) at 1457 cm⁻¹ and the saturated ν(C–H) bands at 2923/2853 cm⁻¹ are characteristic of a polymer chain. The monomer doublet at 1635/1618 cm⁻¹ and the =CH₂ wagging band at 985 cm⁻¹ are absent, the strong ν_as(COO⁻) band expected at 1550–1570 cm⁻¹ is observed only as a weak band at 1550 cm⁻¹, and the guanidinium N–H bands are weak, because ATR sampling of the heterogeneous product probes mainly the viscous liquid phase rather than the fine salt crystals. Thus, without neutralisation, the stage II product is a mixture of guanidinium salts with predominantly non-ionised, partially polymerised acrylic acid (poly(acrylic acid)-like segments), rather than stoichiometric guanidinium acrylate. Quantitative formation of the acrylate requires neutralisation to pH ≥ 6.3 (α > 99%). For the adhesive application this composition is not a drawback: the COOH/COO⁻ system buffers the formulation in the required pH 4.5–7.0 window, and PVA–poly(acrylic acid) hydrogen bonding is known to produce strong waterborne adhesives [75].

3.4. Structure and Thermal Behaviour of the Composite

In the spectrum of the composite (Figure 2c) a very broad, intense band at 3367 cm⁻¹ results from overlapping N–H and O–H stretching vibrations and indicates an extensive hydrogen-bond network between the guanidinium fragments, the hydroxyl groups of PVA and the silanol groups of silica. The carbonyl band at 1713 cm⁻¹ (non-ionised COOH and residual acetate of PVA) is retained, the band at 1635 cm⁻¹ is due to bound water δ(H–O–H) overlapping with guanidinium δ(NH₂), and the bands at 1192 and 1062 cm⁻¹ belong to ν(C–O) of PVA and ν_as(Si–O–Si) of silica. The absence of the 2051 cm⁻¹ band confirms that no thiocyanate is carried into the adhesive. Table 5 traces the diagnostic bands through all stages: SCN⁻ is removed in stage I, the guanidinium ν(C=N) band is preserved at every stage, and carboxyl groups are present mainly in non-ionised form.

The DTA/TGA curves of the dried composite (Figure 5, Table 6) show practically no mass loss up to ≈130 °C (Δm < 1%); 5 and 10% mass loss occur at ≈140 and ≈150 °C, respectively, so the material is thermally stable up to 100 °C. The endotherm at 103–143 °C (peak 118.3 °C, signal minimum 136.4 °C, ΔH ≈ 101 J g⁻¹) proceeds without mass loss and is attributed to melting of a crystalline guanidinium-salt phase; a contribution of residual GT (m.p. ≈ 118 °C) is excluded by the absence of the ν(SCN) band. The main mass loss (39.7% up to 196 °C) is associated with a strong endotherm at 193.3 °C (ΔH ≈ 192 J g⁻¹), where evaporation of bound water, free acrylic acid (b.p. 141 °C) and residual formaldehyde, dehydration of carboxyl groups and melting of PVA crystallites overlap [76,77]. The second stage (196–348 °C, 27.3%) corresponds to side-chain dehydration of PVA to polyenes and to deamination/condensation of guanidinium fragments, and the third (348–599 °C, 26.1%) to destruction and carbonisation of polyene and poly(acrylic) chains [77,78,79]. No exotherm is observed over the whole range, which shows that polymerisable acrylic monomer is practically absent (radical polymerisation of residual vinyl groups would produce an exotherm at 120–180 °C). Because the route does not pass through guanidinium nitrate, whose decomposition is strongly exothermic [80], the dried adhesive contains no energetic guanidinium salt. The residue at 600 °C (6.93%) exceeds the calculated SiO₂ content of the dry composite (3.07%), the difference (≈3.9%) being carbonised organic matter formed in nitrogen.

3.5. Colloidal State of the Silica Sol in the Formulation

The order of addition proved decisive for homogeneity. When formalin was added last, a homogeneous, translucent and stable system was obtained; when it was added together with or before the other components, gel particles and aggregates appeared. Two processes compete in the system—crosslinking of the PVA matrix and distribution of silica particles throughout its volume—and crosslinking must start only after the filler has been distributed. Likewise, rapid addition of the sol locally raises the particle concentration, shortens inter-particle distances and initiates irreversible siloxane bridging, whereas dropwise addition allows each portion to disperse. The pH window 4.5–7.0 is a compromise between the requirements of the three components: acetalisation of PVA by formaldehyde is efficient only in strongly acidic media (pH < 2–3), the guanidinium fragment should not be exposed to strongly acidic conditions, and silica sol is stable at pH 8–10 but gels fastest at pH 5–7 [15,16,17,81].

Without modifier the silica sol has ζ = −28.6 mV and D_h = 72.4 nm (Table 7, Figure 6). Increasing the GA content progressively decreases the negative potential to −4.8 mV at 30 g, where D_h passes through a maximum (128.5 nm) and slight turbidity appears; at the base content (40 g) the sign of ζ is reversed (+2.5 mV) and the system is homogeneous and stable (D_h = 91.7 nm). At 50–60 g, ζ increases to +8.3 and +14.4 mV, but D_h grows again (118.3 and 146.5 nm) and the film becomes more hydrophilic. This sequence—charge compensation, inversion and restabilisation by excess cations—is the classical response of a negatively charged sol to a cationic additive [19,20,21].

A charge balance shows that the measured ζ-values cannot be explained by complete adsorption of guanidinium. For 15 g of 30% sol (4.5 g SiO₂, total surface 560–1020 m²) and a surface charge density of 0.01–0.05 C m⁻² at pH 6–8 [73], the total negative charge of the particles is 0.06–0.53 mmol, i.e., only 0.01–0.09 g of modifier would suffice to neutralise it, whereas 40 g provides ≈450–4100 times more guanidinium than surface sites. Hence most of the cations remain in solution, and the slow approach of ζ to zero over the 0–40 g range indicates weak adsorption under conditions of strong screening and competition. A Stern–Langmuir analysis with an adsorption constant K of 6·10²–5·10³ M⁻¹ places the charge neutralisation concentration at c₀ = 0.5/K = 0.05–5 mM and predicts ζ ≈ +9 mV at the base-formulation concentration (≈261 mM) irrespective of K, in reasonable agreement with the measured +2.5 mV. Adsorption of PVA, which shifts the shear plane outwards, and the decrease in pH towards the isoelectric point of silica (pH 2–3) also reduce |ζ| [15,18]. A dedicated ζ(c) titration in the micromolar–millimolar range is required to determine K experimentally.

The ionic strength of all practical formulations (20–60 g modifier; 0.13–0.39 mol kg⁻¹) compresses the diffuse layer below 1 nm (Table 7). DLVO calculations (Figure 7) give an energy barrier of ≈48 kT at 1 mM and ≈10 kT at 10 mM; the barrier falls below 15 kT above ≈7 mM and vanishes at ≥ 100 mM. If the diffuse-layer potential is instead identified with the measured ζ-potential of the sol (−28.6 mV), the barrier does not exceed ≈6 kT even at 1 mM (Figure 7b), which reinforces this conclusion. Therefore none of the formulations with 20–60 g modifier is stabilised electrostatically. The model is semi-quantitative (the Derjaguin approximation loses accuracy at κa ≈ 1), and silica is known to be more stable than predicted by DLVO theory owing to hydration forces [82,83,84]. The absence of an electrostatic barrier therefore does not imply fast coagulation; rather, the homogeneity and 120-day shelf life of the base formulation are attributed to steric stabilisation by PVA macromolecules adsorbed on SiO₂ together with hydration repulsion [11,13,14].

The silica content has a strong, non-linear effect on the dispersion (Figure 8, Table 8). Between 5 and 15 g of sol, D_h increases only from 68.4 to 89.6 nm, and the base formulation (15 g; 3.07% SiO₂ in the dry composite) has a shelf life of 120 days. At 20 g, D_h rises to 136.5 nm and the shelf life falls to 30 days because aggregation starts; at 25 g (198.7 nm, 7 days) the films become brittle and non-uniform. In parallel, the dynamic viscosity increases from 420 to 1900 mPa·s, the flow index decreases from 0.82 to 0.48 and the hysteresis area increases from 18.5 to 105.6 Pa s⁻¹, i.e., the formulation becomes more pseudoplastic and thixotropic as a particle network forms [13,64,65]. The base formulation (950 mPa·s, n = 0.68, 43.2 Pa s⁻¹) combines good spreading under shear with resistance to sagging at rest.

3.6. Wetting, Adhesion, Water Resistance and Antibacterial Activity

Contact angles of the formulation on glass (31.4°) and ceramic (28.7°) were markedly lower than on metal (48.5°) and wood (63.2°) (Table 9). The work of adhesion, estimated with the Young–Dupré equation using γ_L = 72.7 mN m⁻¹ because the surface tension of the formulation was not determined, follows the same order: ceramic > glass > metal > wood. Since PVA is surface-active and lowers γ_L, the absolute W_a values are upper-bound estimates, but the ranking is not affected. The better wetting of the high-energy, silanol-rich mineral surfaces (γ_S = 52.7–55.3 mN m⁻¹) is consistent with electrostatic and hydrogen-bond interactions of the cationic guanidinium centres with substrates that are chemically similar to the silica filler [85,86].

The lap-shear strength of the composite exceeded that of the unmodified PVAc reference on all substrates (Figure 9a): 6.8 vs. 4.2 MPa on metal (+61.9%), 5.2 vs. 3.6 MPa on wood (+44.4%), 6.1 vs. 4.0 MPa on glass (+52.5%) and 6.4 vs. 4.1 MPa on ceramic (+56.1%). Failure was cohesive on wood, glass and ceramic, and mixed (mainly cohesive) on metal, indicating that the adhesive–substrate interface is not the weakest element of the joint. The increase in strength is attributed to (i) silica particles acting as physical crosslinks in the matrix, (ii) electrostatic adsorption of guanidinium on the substrate, which strengthens the boundary layer [12], and (iii) hydrogen bonding between the modifier and PVA, reflected by the broad 3367 cm⁻¹ band [75]. The PVA-based composite (variant A) had a compressive strength of 12.5 MPa, a tensile strength of 8.4 MPa and an elongation of 86.0%, whereas the epoxy-based variant B reached 28.6 and 16.8 MPa but only 6.5% elongation. Variant A is thus suited to flexible joints and porous or mineral substrates, and variant B to rigid, highly loaded joints [87,88,89].

After 24 h in water, the composite gained only 1.8% in mass compared with 7.6% for the reference, retained 96.4% of its strength (reference 78.5%) and showed no swelling, cracking or disintegration, whereas the reference swelled and softened slightly. The temperature of 5% mass loss was ≈140 °C for the composite and ≈115 °C for the reference. Formaldehyde modification (hemiformal and partial acetal formation) is expected to reduce the number of free hydroxyl groups and chain mobility, and well-dispersed silica particles lengthen the diffusion path of water [9,10]. Excess modifier has the opposite effect, because the ionic salt is hydrophilic.

The modifier-free formulation showed no antibacterial activity, whereas the composite produced inhibition zones of 12.6, 18.4 and 20.7 mm at 20, 40 and 60 g of modifier (Figure 9b). After washing, the zones were 9.8, 16.9 and 15.4 mm, corresponding to 77.8, 91.8 and 74.4% retained activity. The base formulation (40 g) therefore combines high initial activity with the highest wash resistance, while at 60 g part of the modifier is evidently weakly bound and is washed out. Because the agar-diffusion test detects only the fraction of the biocide that can diffuse into the medium, the retained inhibition zones characterise the wash-resistant but still mobile fraction of guanidinium; the activity of the immobilised fraction has to be quantified by contact-killing tests [36,41]. Retention of activity is attributed to incorporation of acrylic fragments into polymer chains (the FTIR and DTA data show practically no residual vinyl monomer) and to adsorption and hydrogen bonding of guanidinium cations on silica particles [24,30,32,41]. The mechanism of action—adsorption of permanently charged guanidinium groups on the negatively charged bacterial membrane followed by its disruption—is the same as for guanidine polymers [25,26,27].

Table 10 compares the composite with requirements and data for commercial adhesives. Its viscosity (0.95 Pa·s) falls within the range of low-viscosity PVAc dispersions [90], so standard application equipment can be used. However, its strength on wood (5.2 MPa) is below the EN 204 [91] requirement for durability classes D1–D3 under standard conditions (10 N mm⁻², determined on beech specimens according to EN 205), and its strength on steel is well below that of a commercial structural epoxy (22.6 N mm⁻², ISO 4587) [92]. The composite is therefore not intended for load-bearing structural joints; its advantages—wash-resistant antibacterial activity, low water uptake and good adhesion to mineral surfaces—make it suitable for moderately loaded joints and protective coatings on glass, ceramic and metal where biostability is required. Classification according to EN 204 requires additional testing according to EN 205. Because about 57 wt% of the modifier is GuCl, the dry adhesive contains ≈5.8 wt% chloride; for bonding of corrosion-sensitive metals, partial replacement of chloride, e.g., by anion exchange, should be considered.

3.7. Material Balance, Atom Economy and Raw-Material Cost

The material balance was calculated for 100 kg of GT (0.846 kmol) from the stoichiometry of reactions (1) and (3) and the base formulation. Stage I consumes 72.1 kg of CuCl₂·2H₂O and 26.9 kg of copper (GT : Cu²⁺ : Cu⁰ = 2 : 1 : 1) and gives 102.9 kg of CuSCN theoretically (94.7 kg at the experimental yield of 92%), 80.8 kg of GuCl in the filtrate and 15.2 kg of water released from the crystal hydrate; no acid is formed, and no acidic effluent has to be neutralised. With both CuSCN and GuCl counted as products, the atom economy of stage I is 92.3%. Stage II consumes 61.0 kg of acrylic acid and yields 141.8 kg of the GA modifier. For 1 t of the base adhesive formulation (variant A; dry-component basis: PVA 100, modifier 40, silica sol 15 and formalin 5 mass parts), 176.3 kg of waste GT is processed and 166.9 kg of CuSCN is obtained as a by-product. The GT-derived raw materials therefore close the loop for both ions of the waste: the guanidinium cation becomes the functional component of the adhesive, and the thiocyanate anion becomes a marketable copper(I) salt. The only aqueous effluent consists of the CuSCN washings, from which residual copper is precipitated as hydroxide.

A raw-material cost estimate for 1 t of the base formulation, based on market prices of June–October 2026 [94,95,96,97,98] and on zero cost of the waste GT, is given in Table 11. The total raw-material cost is ≈2600 USD t⁻¹; PVA (38.9%), CuCl₂·2H₂O (29.3%) and copper (25.7%) are the main items. Sale of the CuSCN by-product offsets 35.8% of the cost of the copper reagents and lowers the net raw-material cost to ≈2090 USD t⁻¹, i.e., ≈2.3 USD per kg of dry matter (dry-matter content 0.915), compared with retail prices of ≈4.5–5.1 USD per kg of dry matter for commercial PVA wood glues in Uzbekistan [97,98]. The difference is not a profit margin, since energy, labour, depreciation and packaging are not included, but it shows that the process is not penalised by the cost of thiocyanate removal; the use of secondary copper turnings would reduce the cost further.

4. Conclusions

A one-step route for the valorisation of waste guanidine thiocyanate has been developed. Treatment of aqueous GT with CuCl₂ in the presence of metallic copper at 20–30 °C removes thiocyanate as β-CuSCN in 92% yield by Cu²⁺/Cu⁰ synproportionation (ΔE° = +1.14 V, lg K = 19.3), without oxidation of thiocyanate, without nitric acid and without formation of acid; the guanidinium cation remains in solution as GuCl and is used directly in the next stage. FTIR confirms the complete disappearance of the ν(SCN) band at 2051 cm⁻¹.

Reaction of GuCl with acrylic acid without a neutraliser is thermodynamically unfavourable (K ≈ 7·10⁻¹¹; 0.53% acrylate at pH 1.98), and FTIR shows that the product consists of guanidinium salts and partially polymerised, mostly non-ionised acrylic acid. This modifier nevertheless performs well in a formalin-modified PVA/silica sol adhesive, where the COOH/COO⁻ system buffers the formulation at pH 4.5–7.0.

In the formulation, guanidinium reverses the ζ-potential of silica from −28.6 to +2.5 mV at 40 g per 100 g PVA, but the high ionic strength (≈0.26 mol kg⁻¹) eliminates the electrostatic DLVO barrier; the stability of the base formulation (120 days) is due to steric stabilisation by adsorbed PVA. The optimum composition (PVA 100 : modifier 40 : 30% silica sol 15 : formalin 5) gives 950 mPa·s, lap-shear strengths of 5.2–6.8 MPa (+44–62% vs. PVAc), 1.8% water uptake, 96.4% strength retention after water exposure, thermal stability up to 100 °C and wash-resistant antibacterial activity (91.8% retained). The adhesive is intended for moderately loaded joints and coatings requiring biostability; for structural applications and EN 204 classification, further testing is required. The process closes the loop for both ions of the waste: 176 kg of GT per tonne of adhesive is converted, 167 kg of marketable CuSCN is co-produced, and the net raw-material cost is ≈2.1 kUSD t⁻¹. Further work will address the quantitative composition of the modifier (¹H NMR, titration), contact-killing tests according to ISO 22196 and partial replacement of chloride for metal bonding.