An Innovative Wood Fire-Retardant Coating Based on Biocompatible Nanocellulose Surfactant and Expandable Graphite

Abstract

Nanocellulose (CNC) seems to be a promising surfactant, which, together with expandable graphite (EG), forms the essence of an effective natural-based fire-retardant wood coating. In our research, the most suitable composition of the mixture was tested concerning good solubility, dispersion, and consistency. Favorable results were achieved with the formulation composed of a 4% CNC alkaline solution with 80 wt.% of the selected EG. Subsequently, six different types of EG were used to prepare these wood fire-retardant coatings. The effectiveness of treatments was verified using a test with a radiant heat source, where the test samples’ relative weight loss, relative burning rate, and surface temperature during 600 s were evaluated. All prepared formulations can be characterized as more or less equally effective. However, the best results were obtained with the EG of GG 200–100 N, where the mass loss of the sample was 8.10 ± 1.24%. Very good results were also achieved by graphite 25 E + 180 HPH (8.70 ± 0.89%), which is similar to the previous one, even according to the microscopic assessment of the coating as well as the expanded layer. The graphite type 25 K + 180 (8.86 ± 0.65%) shows the expanded layer’s best adhesion, coating uniformity, and ease of application. The results of this work confirmed that the CNC coating itself has significant retardation effects.

1. Introduction

The use of wood is huge, and in the past, it was irreplaceable due to its unique properties, ease of processing, and availability. However, this has partially changed with the advent of new materials, and in many industries, wood has been replaced by materials that eliminate its shortcomings, whether susceptibility to biological degradation or flammability. Recently, when humanity began to realize its impact on the environment, wood as a natural, renewable, and easy-to-produce resource has once again become an important material both for the present and for the future [1,2].

Thanks to technological progress, we are now able not only to use wood more efficiently but also partially eliminate its negative properties. In addition to its energy use, we try to suppress the flammability of wood. For this purpose, retardants are used to reduce the flammability of wood using physical or chemical means, but in practice, both methods are usually combined. The currently used retardants are quite effective, but in many cases, they are toxic and have a negative effect not only on the environment but also on human health. Such retardants must be adequately replaced to maintain their effectiveness, eliminate toxicity, and be economically acceptable [3,4,5].

One of the modern alternatives to fire retardants is nanomaterials. A characteristic feature of nanomaterials is their high proportion of surface area to the entire volume of the material, which gives them different chemical and physical properties compared to compact materials with the same chemical composition. The smaller the dimensions of the nanoparticles, the greater the proportion of surface atoms, which causes an increasing proportion of the surface energy to the total energy of the system made of nanoparticles [6,7,8]. For example, protective coatings based on minerals, clays, and nanocellulose, which are non-toxic but not yet widely available from an economic point of view, appear to be very promising. In addition, if coatings made of the above-mentioned nanomaterials are used separately, their effectiveness is still insufficient compared to the simultaneously used means. Excellent results are achieved when they are combined with conventional fire-retardant formulations [9].

The principle of achieving fire retardancy with nanoparticles is to create a coating on the surface of the wood that prevents the release of flammable gases and the production of smoke. The concentration of combustible gases can be diluted by non-flammable gases produced by the pyrolysis of nanoparticles, thereby terminating the chain reaction of combustion by scavenging highly reactive free radicals generated during the pyrolysis of nanoparticles. A dense charred layer forms on the wood’s surface, creating an insulating barrier [6].

The main idea of this published work is to use nanomaterial in the form of nanocellulose (CNC—cellulose nanocrystals) as a surfactant in an innovative fire-fighting coating on a natural basis with the addition of expandable graphite (EG).

Surfactants are surface-active substances formed from molecules containing polar and non-polar parts. These substances are used for better dispersion of powder materials in a liquid phase. They can form aggregated structures in aqueous or non-aqueous media, including microemulsions in their mixtures. They are also used as emulsifiers in preparing emulsions and as stabilizers in producing foams [10]. These are substances on a natural basis, which are biocompatible and biodegradable, have a good binding potential between the wood surface and fire retardant, and will not introduce any additional chemical substances into the wood–thermographite system.

Nanocrystalline cellulose (CNC) is isolated from cellulose fibrils. Cellulose is known to contain crystalline and amorphous fractions depending on the source. Less stable amorphous parts are removed during the nanocellulose preparation process by acid-catalyzed hydrolysis, while the crystalline parts are preserved [11]. Currently, the largest source of cellulose is biomass, but it occurs naturally in all plants [12]. In addition to wood and plants, the source of cellulose can also be animals, algae, or bacteria [13]. The source of so-called “green” nanocellulose is industrial waste, e.g., from beer production or solid municipal waste [14]. The action of sulfuric acid during hydrolysis also results in the formation of charged groups on the surface of the nanocrystals, thereby promoting electrostatic interactions in the aqueous environment.

The main factors influencing the resulting properties of the nanocellulose obtained in this way are the reaction time, temperature, and acid concentration. This isolation method results in rod-shaped nanocrystals 100–6000 nm long and 4–70 nm wide with a crystallinity index of 54%–88% [13]. Due to the high surface-to-volume ratio and a great number of hydroxyl (–OH) groups, CNC is suitable for many types of surface functionalization. With various modifications, it is possible to introduce positive or negative charges to the CNC surface to improve the dispersibility in solvents [15,16]. Esterification is one of the most common methods of CNC modification, especially using acid anhydride, succinic anhydride, and citric acid as a catalyst. The presence of negative charge carboxyl groups on the CNCs surface greatly improved the stability of CNCs suspension [17,18,19]. In oxidative modification, strong oxidizing agents are used, e.g., TEMPO (2,2,6,6-tetramethyl-1-piperidinyloxy) in the presence of NaBr and NaOCl [20] or sodium periodate [21]. Currently, the most used method of modifying the CNC surface is esterification, which results in increased thermostability of the derivatives and their better solubility in water [17]. Modification of the graphite coating with nanocellulose could lead to an increase in the binding capacity, and thus, it would be possible to control the binding interactions with the wood surface.

CNC is also widely tested as a possible fire retardant, mainly in the form of a composite modified with other substances. For example, Luo and Wang [22] dealt with the preparation of TiO₂-modified CNC composite aerogel and carried out various analyses. The results show that the TiO₂-modified CNC aerogels exhibited a 3D network structure and underwent a decline in crystallinity through TiO₂ doping and modification. The TiO₂/CNC aerogels surpassed the CNC aerogel in both thermal decomposition temperature and the maximum thermal decomposition rate. TiO₂/CNC aerogels had a lower thermal conductivity than the CNC aerogel, and the thermal conductivity of cellulose aerogel gradually decreased with the increase of TiO₂ doping amount. This research confirmed that doping and modification with suitable inorganic particles is an effective way to enhance the fire retardant and thermal insulation properties of cellulose aerogel. In our case, we combine CNC with another prospective material, which, based on the results of our previous research, could very well fulfill the function of a wood-burning retardant. It is a natural material called expandable graphite, which has excellent fire retarding properties and is already widely used as an intumescent retarder, for example, in the automotive industry for fire protection of polymers [23,24,25]. A highly valued property of EG is its ability to increase its volume after exceeding a certain temperature and thus create an effective insulating layer. In addition, compared to other chemically formed intumescent materials, this char layer retains very good heat resistance and expands with sufficient force to allow its use in rigid systems [26]. In order for EG to be successfully used for the thermal protection of wood, it is first necessary to solve its lack of adhesion with the wooden surface or wood structure. Several papers have already been published on this topic [27,28,29], but it is still evident that the bonding mechanism and sufficient anchoring of the graphite particles to the wood structure need to be paid more attention and further investigated and verified.

If it were possible to find a sufficient adhesion of a suitable type of EG with a wood surface, a highly effective, environmentally friendly, and economically available wood fire retardant would be created [30,31]. Promising results were found when combining water glass with EG [28,32,33]. However, the combination of CNC and EG, according to our knowledge, has not yet been published in the scientific literature.

The objective of this work was to design a suitable formulation of expandable graphite with nanocellulose (solvent, procedure, and mixing ratio), which meets all the requirements of a wood coating in terms of good EG dispersion, surfactant solubility, consistency, easy application, non-flowability from vertical surfaces, adhesion with wood even after graphite expansion and uniformity of the coating. Subsequently, six different types of EG were tested in this formulation, and the effectiveness of these fire-retardant coatings on spruce wood was evaluated using a radiant heat source test and microscopic observation. On that occasion, the effectiveness of the CNC surfactant itself was also verified from the point of view of fire retardancy properties.

2. Materials and Methods

2.1. Materials

2.1.1. Wood

Samples for the test with a radiant heat source with dimensions of 10 × 40 × 50 mm³ were prepared from dried Norway spruce wood (Picea abies (L.) H. Karst) without visible defects and damage. The auxiliary samples for the application test were prepared from the same planks as mentioned above and had dimensions of 20 × 20 × 300 mm³. All samples were placed in an air-conditioning chamber, where they were conditioned for 21 days at a temperature of 21 ± 2 °C and 65 ± 5% relative humidity. This achieved their equilibrium moisture content of approximately 12%.

2.1.2. Nanocellulose

Nanocellulose in crystalline form (CNC) fulfills the function of a surfactant in the proposed fire-retardant formulation—it ensures the adhesion of the coating with the surface of wood and graphite, even after it expands when exposed to high temperatures. At the same time, it has a certain fire retarding effect. CNC is transparent and odorless, and it is found in aqueous slurry form. More detailed characteristics of the used nanocellulose are shown in

Table 1. Specification of selected nanocellulose.

Product	Solid Content	Surface Property	Surface Groups	Fiber Dimensions (nm)	Producer (Country)
CNC–COM–regular	6% suspension in water	Hydrophilic	Hydroxyl, sulfonic	Width: 5–20 Length: 100–250	CELLULOSELab (Fredericton, NB, USA)

.

2.1.3. Expandable Thermographites

Commercially available expandable graphites (EG) differ in many parameters: starting temperature, pH value, particle size, carbon content, minimum expansion and possibly also chemicals used for intercalation and neutralization of the product. In this research, six commercially available EGs from two different suppliers were tested. Their characteristics are shown in

Table 2. Specification of selected expandable graphites.

Product	Carbon Content (%)	Expansion Volume (mL/g)	Particle Size (μm)	pH Value	Starting Temperature (°C)	Producer (Country)
40 D + 500 LST	99	400	min. 80% > 500	5–9	180–220	Epinikon a.s. (Vodňany, Czech Republic)
20 K + 300 LST	95	200	min. 70% > 300	5–9	140–170	Epinikon a.s. (Vodňany, Czech Republic)
25 K + 180	95	250	min. 80% > 180	5–9	180–220	Epinikon a.s. (Vodňany, Czech Republic)
25 E + 180 HPH	98	250	min. 80% > 180	8–11	180–220	Epinikon a.s. (Vodňany, Czech Republic)
GG 210–200 N	>93	80	90	5–8.5	210	NeoGraf (Lakewood, OH, USA)
GG 200–100 N	>93	175	150	5–8.5	200	NeoGraf (Lakewood, OH, USA)

.

2.2. Methods

2.2.1. Preparation, Evaluation, and Selection of Formulations

First of all, six formulations of the selected EG representative (with average values of expansion volume, particle size, starting temperature, and normal pH value) were prepared with CNC in order to select the most suitable variant (

Table 3. Overview of prepared formulations of expandable graphite 25 K + 180 (Epinikon) with nanocellulose-based surfactant.

Surfactant Type	CNC
Solution (%)	in Water		in 5% NaOH
Dry matter content of surfactant (%)	2	4	2	2	4	4
EG content in formulation (wt.%)	60	60	60	80	60	80

).

The preparation of the formulations was first carried out in a distilled water solution, but the solubility of the surfactant and subsequent dispersion of the EG particles was not sufficient. Therefore, based on the literature study according to Santos et al. [29], a 5% aqueous solution of NaOH was also used, which had to be pre-cooled to a temperature of approximately 0–5 °C to avoid clumping of added cellulose. Subsequently, a 2% and 4% colloidal solution of CNC was added, which should be mixed gradually in small amounts so that no clumps form and perfect dissolution occurs. The result was a viscous, clear mixture to which EG was subsequently added in amounts of 60 wt.% and 80 wt.%. After thorough mixing, a thick pasty mass was formed, for which the following parameters were evaluated: surfactant solubility, dispersion of EG particles in surfactant, and formulation consistency.

All the prepared formulations were subsequently applied to the horizontal and vertical surfaces of spruce wood, and the ease of application was evaluated—spreadability, coating uniformity, flowability on vertical surfaces, drying time, and adhesion with wood after drying. Based on the evaluation of all monitored parameters, the best of these tested formulations was chosen, according to which other formulations were prepared with selected types of single-piece EG. These formulations were subsequently applied to test specimens for testing with a radiant heat source.

2.2.2. Application on Wood

For applying the surfactant formulation with EG, the method of painting the mixture with a brush with synthetic fibers was chosen. The mixture was relatively easy to spread, but achieving an even coating was more difficult and required a certain skill. Another possible method of applying the mixture is with a metal spatula, thanks to which a fairly even layer of graphite can be achieved on the surface of the wood. A total of 60 samples coated with selected formulations were prepared for the test with a radiant heat source, i.e., 10 samples from each series differing in the type of EG in the formulation. Furthermore, test samples of wood were painted with only surfactant (CNC) to determine the thermal resistance of the surfactant itself. Ten samples of spruce wood remained untreated, creating the so-called reference series. The amount of formulation applied to a given area of the sample, i.e., 40 × 50 mm², was 1.00 ± 0.05 g. The control was carried out by weighing the sample on analytical balances, and emphasis was placed on the uniformity of the coating. All coatings prepared in this way had a thickness of 442–495 μm, as subsequently determined by image analysis obtained using a Nikon Eclipse Ni polarizing microscope at 40× magnification and 0.85 μm/px resolution.

2.2.3. Radiant Heat Source Test, Surface Temperature Measurement and Evaluation of Layer Adhesion after Thermographite Expansion

Non-standard test method—radiant heat source test was chosen to evaluate fire-retardant properties of examined EG-treated wood samples. In this method, a ceramic horizontal thermal infrared heater (Ceramicx, Cork, Ireland) with a constant electric power of 1000 W and Radwag PS 3500.R2 electronic scales (Radom, Poland) were used. The heat flux on the sample surface was 30 kW/m². The duration of action was 600 s, and the distance of the samples from the surface of the heater was 40 mm [34]. The schematic of the arrangement of the experimental equipment is shown in Figure 1. The mass loss was recorded every 10 s (using the RLAB program). Any ignition of the samples was visually checked with a time record from the start of the test using a stopwatch if this phenomenon occurred. Subsequently, the relative mass loss and relative burning rate of wood from the measured values were calculated according to the following Equations (1) and (2) [32]:

$${\mathsf{\delta }}_{\mathrm{m}}\left(\mathsf{\tau }\right)={\displaystyle \frac{\mathrm{m}\left({\mathsf{\tau }}_0\right)-\mathrm{m}\left(\mathsf{\tau }\right)}{\mathrm{m}\left({\mathsf{\tau }}_0\right)}}\times 100\left(\mathrm{\%}\right)$$

(1)

$${\mathsf{\upsilon }}_{\mathrm{r}}={\displaystyle \frac{{\mathsf{\delta }}_{\mathrm{m}}\left(\mathsf{\tau }\right)-{\mathsf{\delta }}_{\mathrm{m}}\left(\mathsf{\tau }+\mathsf{\Delta }\mathsf{\tau }\right)}{\mathsf{\Delta }\mathsf{\tau }}}\times 100\left(\mathrm{\%}\times {\mathrm{s}}^{-1}\right)$$

(2)

where: ${\mathsf{\delta }}_{\mathrm{m}}\left(\mathsf{\tau }\right)$—relative mass loss over time $\left(\mathsf{\tau }\right)\left(\mathrm{\%}\right)$; ${\mathsf{\upsilon }}_{\mathrm{r}}$—relative burning rate $\left(\mathrm{\%}\times {\mathrm{s}}^{-1}\right)$; $\mathrm{m}\left({\mathsf{\tau }}_0\right)$—sample original weight (g); $\mathrm{m}\left(\mathsf{\tau }\right)$—sample weight at time $\left(\mathsf{\tau }\right)\left(\mathrm{g}\right)$; ${\mathsf{\delta }}_{\mathrm{m}}\left(\mathsf{\tau }+\mathsf{\Delta }\mathsf{\tau }\right)$—relative mass loss over time $\left(\mathsf{\tau }+\mathsf{\Delta }\mathsf{\tau }\right)\left(\mathrm{\%}\right)$;${\mathsf{\delta }}_{\mathrm{m}}\left(\mathsf{\tau }\right)$—relative mass loss at time $\left(\mathsf{\tau }\right)\left(\mathrm{\%}\right);$ and $\mathsf{\Delta }\mathsf{\tau }$—the time interval for reading weights (s).

With the thermal camera Fluke RSE600 (Fluke Corporation, Everett, WA, USA), images were taken continuously during the test using the Fluke SmartView R&D software IRSoft2, version 7.0 at selected points—on the top of the samples—in 1 s intervals. The evaluation criteria were ignition time, relative mass loss, relative burning rate, temperature course on the top of the samples, and layer adhesion after EG expansion. The adhesion of the layer after EG expansion was evaluated by simply rotating the sample by 180° after performing the test with a radiant heat source. The amount of protective layer that remained on the surface of the sample and even the fallen part were checked gravimetrically and assessed. The adhesion of the expanded coating layer for individual types of EG was compared.

2.2.4. Microscopic Analysis

Samples were imaged using a Nikon SMZ1270 (Nikon, Tokyo, Japan) stereomicroscope equipped with a Plan Apo 1× objective with a 70 mm working distance and a Nikon DS-Fi3 digital camera (Nikon, Tokyo, Japan). The NIS-Elements AR 5.42.05 software was used for scanning the samples. The stereomicroscope provided excellent optical quality and a wide range of magnification, allowing detailed observation of the fine structures of the samples. Focused composite images were compiled for some of the samples that did not allow for perfect surface focusing.

3. Results and Discussion

3.1. Selection of Suitable Formulation

Testing of all defined formulations of expandable graphite 25 K + 180 (Epinikon, CZ) was performed on test samples of spruce wood. A wide range of criteria were evaluated, from the preparation of the formulation itself through its properties to adhesion with the wood surface (

Table 4. Evaluation of properties of prepared formulations from nanocellulose and expandable graphite 25 K + 180.

Surfactant Type	CNC
Solution (%)	in Water		in 5% NaOH
1. Dry matter content of surfactant (%)	2	4	2	2	4	4
2. EG content in formulation (wt.%)	60	60	60	80	60	80
3. Surfactant solubility	−	−	+ +	+ +	+ +	+ +
4. Dispersion of EG particles in surfactant	−	− −	+	++	+	+
5. Formulation consistency	− − −	− − −	−	−	−	−
6. Application	− −	− −	−	+	− −	− −
7. Coating uniformity	− −	− −	+	+	− −	− −
8. Flowability on vertical surfaces	+	+	+ + +	+ + +	+ + +	+ +
9. Drying time	−	−	+ +	+ +	+ +	+ +
10. Adhesion with wood after drying	−	−	+ + +	+ + +	+ +	+ +
Recommendations for use	no	no	yes	definitely yes	rather yes	rather not

Explanations: from the best (+ + +) to the worst (− − −) evaluated samples in the tested series. Ad. 3, 4: (−) the degree and amount of clusters, (+) easy and quick solubility or dispersion; Ad. 5, 6: (−) pastier, harder to spread formulation, (+) good consistency suitable for application; Ad. 7: (−) visually uneven coatings, (+) smooth uniform coatings; Ad. 8: (+) a degree of stability on a vertical surface; Ad. 9: (−) coating surface visibly wet for more than 30 min, (+) surface dry within 30 min and earlier; Ad. 10: (−) tendency to lose part of the coating even before expansion, (+) sufficient adhesion without significant losses and peeling of the coating within the tested series.

). The evaluation took place on the basis of observation, laboratory experience, and relative comparison within the entire tested group, with the aim of selecting only the best formulation.

After applying all the prepared formulations to the test samples, a 2% CNC surfactant alkaline solution with 80 wt.% EG in the formulation was selected as the most suitable. At this mixing ratio, an excellent solubility of the surfactant in the alkaline NaOH solution and dispersion of the EG particles was achieved. The consistency of the mixture was acceptable, and the spreadability of the coating and its uniformity when applied to wood was good, or the best within the group of tested formulations. The advantage is that the coating does not run off the vertical surfaces of the treated wood samples, dries relatively quickly, and the adhesion with the wood after drying is very good.

It would still be possible to recommend a formulation with 2% and 4% CNC alkaline solution in combination with 60 wt.% EG. However, from an economic point of view, the previous formulation mentioned above was preferred with a lower proportion of the relatively expensive surfactant and with the possibility of higher EG saturation so that the protective layer would be more effective.

Aqueous solutions of the mixture did not work. EG had a tendency to settle, particles did not disperse even during mixing, and the consistency was not optimal; therefore, the coatings on the wood were uneven and flowed down from a vertical surface.

The advantage of an alkaline environment of NaOH is that it ensures easier dissolution of CNC and prevents swelling or clumping. Then cellulose enables the exfoliation of graphite in aqueous solutions, but it also performs as a plasticizing adhesive that allows the thin exfoliated lamellae to reassemble when they are applied on the wood surface [29]. In addition, when the alkaline environment affects the wood, its slight degradation occurs, or fibrillation [35]. This creates a larger specific surface with a higher binding capacity, which can bind the graphite coating saturated with CNC.

A certain, but only temporary, aesthetic disadvantage can be the slight yellowing of the wood surface immediately after applying the coating, which is caused by the alkaline NaOH solution. It is not noticeable after drying.

3.2. The Radiant Heat Source Test

The test with a radiant heat source is a non-standard test method used in model burning tests. Using the mentioned test method, we can relatively quickly demonstrate the material’s behavior when exposed to high temperatures.

The appearance of the prepared samples before and after the test of the radiant heat source with the subsequent test of the adhesion of the expanded layer of EG is shown in Figure 2. The sample of untreated wood (reference) and wood treated only with the corresponding amount of CNC present in the formulation after this test is shown in Figure 3.

The results of the experiments evaluating the response of treated and untreated wood surfaces to the exposure to a certain degree of heat flux of wood are summarized in this section in the graphs titled relative mass loss of tested samples (Figure 4), evolution of relative mass loss and relative burning rate in time (Figure 5), and temperature course on the sample surface when exposed to a radiant heat source (Figure 6).

The calculated relative mass loss of all types of tested samples in the graph in Figure 4 shows that in the case of the untreated (reference) spruce wood sample, the highest decrease of up to 88.67 ± 1.33% of the original weight of the sample occurs after the test with a radiant heat source after 600 s. In the case of treatment of the sample surface with only the CNC surfactant used, this mass loss was reduced by more than half (approximately 56%) compared to the reference sample, and the value of relative mass loss for the CNC-treated sample is, therefore, 32.80 ± 2.87%. In the case of treatment of the wood surface with a formulation consisting of CNC surfactant and EG of six different species, the values of relative mass loss ranged in a very close range of 10.01%–8.10%, which means the suppression of weight loss compared to the reference sample by approximately 78.7%–80.6%. From this point of view, all prepared formulations can be characterized as more or less equally effective. However, the formulation with GG 200–100 N from the US manufacturer GrafGuard stood the best in the evaluation, when there was a weight loss of the sample by 8.10 ± 1.24%, and two formulations using graphite from the Czech company Epinikon 25 E + 180 HPH (8.70 ± 0.89%) and 25 K + 180 (8.86 ± 0.65%). In these three cases, it was always EG with an expansion volume of 175–250 mL/g, a starting temperature of 180–220 °C, and an average particle size from the series tested, i.e., 150–180 μm. The highest decrease in the weight of the samples after the test with a radiant heat source occurred within the tested group EG in the formulation using graphite type 40 D + 500 LST, i.e., by 10.01 ± 0.71%.

Kmeťová et al. [28] tested the formulation of different types of EG and water glass (WG) on spruce wood in their work. The mass loss of untreated wood is identical to our results, as the samples achieved a mass loss of 87.72%. When using a surfactant, in this case, WG, the mass loss was reduced to 19.45%, which is about 12% less than CNC wood processing only. The mass losses of the samples treated with coatings combined with EG and WG are comparable to those with EG and CNC coatings. The best results in both cases were confirmed by samples of the used expandable graphite GG 200–100 N and water glass, or CNC.

Bilbao et al. [36] observed the influence of the type of ignition (piloted or spontaneous) and the distance between the heating source and the sample. Higher values of time to ignition were observed for spontaneous ignition than for piloted ignition and with the increasing distance between the heating source and the sample. The time to ignition of wood depends on several factors, including wood type, radiant panel output, distance, and heat flux [37,38].

In terms of evaluating the adhesion of the carbon layer after graphite expansion, on the other hand, samples with graphite 40 D + 500 LST with the highest expansion volume (400 mL/g) and the largest particle size of 500 μm stood the best, and the expanded layer was only minimally dusted spontaneously from the surface. Similarly, samples with a graphite content of 25 K + 180 had the second-highest expansion volume (250 mL/g) from the tested series of graphites. However, we cannot talk about the connection between the layer adhesion and the expansion volume of thermographite because, for example, the sample with 25 E + 180 HPH with the same expansion volume of 250 mL/g, as mentioned above, showed the lowest layer adhesion after expansion. Most were dusted from the surface of the sample. Insufficient adhesion (i.e., mass loss of the expanded layer more than 20%) was also noted in samples with GG 200–100 N. Conversely, the second type of US graphite GG 210–200 N with the lowest expansion volume of 80 mL/g had relatively good adhesion. Improving the compactness of the layer after expansion could be done in the future by additional coating treatment with the surfactant solution itself or by surface color treatment. However, the adhesion of the expanded coating layer with the wood surface was evaluated as sufficient in all formulations, regardless of the type of EG used.

When evaluating the development of relative mass loss over time in the graphs in Figure 5, it is evident that in the case of all used formulations with different types of EG, not only the total weight loss of the samples compared to the reference but also the trend of gradual decrease over time is similar.

For the sample treated only with CNC surfactant, there is a linear decrease in this quantity over time compared to all other tested samples. It is interesting that the linear section of the weight decrease in the time 0–270 s is almost identical to the course of weight loss in the untreated reference sample. In both cases, there is a weight loss of approximately 15.5%–15.7% in the given time, while for samples treated with formulations with EG, this loss is more than half lower, i.e., 6.4%. In the case of the untreated sample, the highest total weight loss (68.6%) occurs in the time range of 270–500 s during which the sample burns. In the last 100 s of the measurement, the weight loss of the untreated sample is negligible, which indicates the termination of its burning.

When comparing all tested samples based on the relative burning rate parameter, the graphs describing the behavior of samples treated with formulations with EG again show the similarity of trends within this group. This trend can be described by a rapid increase from the beginning of the measurement up to 20 s to a value of around 0.04%·s⁻¹ and a subsequent idealized exponential up to fourfold decrease below 0.01%·s⁻¹. The highest value of relative burning rate (peak height at 20 s) has the samples with graphite 40 D + 500 LST (0.058%·s⁻¹) and then 20 K + 300 LST (0.045%·s⁻¹) and GG 210–200 N (0.042%·s⁻¹). The values of the other samples (GG 200–100 N, 25 E + 180 HPH, 25 K + 180) are below the mentioned limit of 0.04%·s⁻¹ in the range of 0.036–0.039%·s⁻¹ and are, therefore, rated the best in this respect. It should also be noted that for samples GG 200–100 N and 20 K + 300 LST, a change in the value of this parameter is observed between 20 s and 30 s, compared to the others. The drop in relative burning rate below the mentioned 0.01%·s⁻¹ occurs first at US graphite samples, i.e., GG 200–100 N (approx. in the 350th s) and GG 210–200 N (approx. 390th s), then at samples 25 E + 180 HPH and 25 K + 180 (in the 410th s) and no later than in the 460th s from the start of the test for samples with 40 D + 500 LST and 20 K + 300 LST.

For the sample treated only with CNC surfactant, two maxima are observed on the relative burning rate development curve over time, namely in the 70th s (0.077%·s⁻¹) and the 230th s (0.071%·s⁻¹). From the 230th s, there is a gradual decrease to a value of 0.038%·s⁻¹ in the 570th s. Compared to the fourfold reduction of this parameter in the case of formulations with EG, in the case of CNC surfactant use, it is only a twofold reduction during the test.

In the case of the untreated reference sample of spruce wood, the development of the relative burning rate has a completely different trend compared to the treated samples. Up to the 240th s (0.074%·s⁻¹), the parameter values increase gradually; then, there is a steep increase to the first maximum with a value of 0.44%·s⁻¹ at time 290 s. Between 290 s and 330 s, the value decreases to 0.31%·s⁻¹ and subsequently increases again to the second maximum with a value of 0.36%·s⁻¹ at time 410 s. In the next 100 s, the value drops approximately sixfold to a value of 0.058%·s⁻¹ and only slightly decreases by the end of the test.

The graph in Figure 6 shows the temperature evolution on the exposed surface of the examined samples during the entire test. It is possible to observe the connection with the development of the relative burning rate (graphs in Figure 5), which for samples treated with formulations with EG and CNC increases within approximately 20 s of the test, as well as the surface temperature. After this time, an expanded graphite layer is formed, which prevents further heating and burning. The surface temperature of the treated samples in 20 s is in the range of 330–370 °C, while the samples containing 40 D + 500 LST, 25 K + 180, and 25 E + 180 HPH are at the lower limit. The temperature of the sample with 40 D + 500 LST, which can be considered the best from this point of view, subsequently drops even more significantly by up to 100 °C within approximately 150 s and then moves normally, like all treated samples, in the range of 290–320 °C. In the samples treated with GG 200–100 N graphite, the temperature rapidly increased to 356 °C in the initial phase. Still, in 50 s, the temperatures already drop to the level of 25 K + 180, 25 E + 180 HPH, and 20 K + 300 LST, and at 150 s, even below the level of the best-rated sample 40 D + 500 LST. The second sample of US graphite GG 210–200 N has a very similar temperature trend over time, but the temperature values are up to 60 °C higher in certain phases than for GG 200–100 N.

The increase in temperature over time for the untreated reference sample is more gradual compared to samples with EG—it starts from temperatures of 120 °C and reaches a temperature of 330 °C up to 70 s later than the treated samples (at 90 s). After that, temperatures continue to increase up to 656 °C, which the surface of the untreated wood sample reaches in 420 s of the test. This is followed by a slight decrease and maintenance at a constant temperature limit of 600 °C at the end of the test caused by the retardation effect of the carbonated surface layer.

The graph in Figure 6 also shows the effectiveness of the protective coating, as at the end of the test, all samples treated with EG and CNC coating show a surface temperature half lower (ca. 300 °C) than the untreated wood samples (ca. 600 °C).

3.3. Microscopic Analysis

Images from microscopic observation (Figure 7) provide a detailed view of the structure of individual coatings containing different types of EG. The surface generally shows a flat layer of irregular particles predominantly aligned along the surface plane [24]. The uniformity of the coatings is affected by the size and dispersion of individual graphite particles. Graphite GG 210–200 N, with the smallest size of regular shiny particles (90 μm), forms the most homogeneous surface. Even so, it does not achieve such a fire performance (higher weight loss in the radiant heat test by 1%) as graphite GG 200–100 N from the same manufacturer, with a larger particle size (150 μm) of irregular size and dull appearance, which cause occasional cracks due to layering of graphite flakes. When microscopically comparing two types of graphite with the same particle size (180 μm), a big difference is visible at first glance: The 25 E + 180 HPH particles have a very smooth and shiny surface. Their irregular shape causes insufficient continuity and overlap so that they form a homogeneous coating layer without cracks and holes. Graphite 25 K + 180, on the other hand, has regular particles, a uniform layer, and a matte appearance. Despite the imperfection of the coating with a content of 25 E + 180 HPH, this sample, as well as GG 200–100 N, achieves one of the best results when evaluating the relative weight loss of the sample (8.7%). However, there is a greater visual similarity between samples 25 K + 180 and GG 200–100 N (8.10%). The micrographs of the coatings also support the observations about the adhesion of the graphite layers after expansion. It can be seen that the 25 K + 180 and 40 D + 500 LST samples already form a highly compact uniform layer in the coating itself, which retains a high degree of adhesion even after expansion. In the case of graphite 40 D + 500 LST, these are the largest particles (500 μm), and they are of shiny character and irregular dimensions, which fit well together in shape. However, the same cannot be said about the 20 K + 300 LST (300 μm) graphite particles, which create a very irregular surface with many depressions and free holes in the coating. Then heat and fire could transfer to the wood substrate through the cracks and holes of the char structure, resulting in weaker fire resistance, as was also observed by Zhan et al. [39] in their work. During the heating of treated wood, an insulation layer is formed on its surface, which prevents direct contact of the flame with the wood’s surface, preferentially absorbing the flame’s heat and controlling its access to the surface of the wood. This layer has a significant thermal insulation effect and prevents the further spread of the flame. The EG expands and non-flammable gases are released, which reduces the concentration of oxygen and thereby increases the retardation effect [33]. The adhesion of these particles after the expansion is very good. Still, the relative weight loss after the test with a radiant heat source is also the largest within the group for both graphites with the largest particles. The middle and right column in Figure 7 brings further knowledge when studying the structure of the inorganic expansion layer that has an important influence on fire resistance.

4. Conclusions

As part of this research, formulations with different content of CNC surfactant and EG were tested. After selecting the most suitable formulation, six series of coatings were prepared, which differed in the type of EG used by two different manufacturers. Based on the test results with a radiant heat source and microscopic observation of the coatings and the expanded layers, the individual formulations were evaluated, and the best ones were selected and recommended.

The main findings arising from this research are as follows:

• The formulation with the best surfactant solubility, dispersion of EG particles, suitable viscosity, and easy applicability consists of a 2% solution of CNC in 5% NaOH and 80 wt.% EG. At the same time, this coating has a relatively fast drying time, does not run off vertical surfaces, and has very good adhesion to the wood surface even after drying.
• Suitable types of EG to ensure the best fire-retardant properties of the coating in the formulation have a particle size of 150–180 μm, an expansion volume of 175–250 mL/g, and a starting temperature of 180–220 °C.
• If we consider the relative weight loss of the samples after the test with a radiant heat source, all prepared formulations with EG can be more or less equally effective. The values of relative mass loss ranged in a very close range of 10.01%–8.10%, which means the suppression of weight loss compared to the reference sample by approximately 78.7%–80.6%. The lowest value of this parameter was obtained with EG of GG 200–100 N (8.10 ± 1.24%). Relative mass loss of samples treated only with CNC (32.80 ± 2.87%) was reduced by more than half (approximately 56%) compared to the reference sample.
• The best adhesion of EG layers shows samples with 40 D + 500 LST and 25 K + 180.
• The course of the relative burning rate of EG-treated samples again shows the similarity within this EG-tested group. This trend can be described by a rapid increase from the beginning of the measurement up to 20 s to a value of around 0.04%·s⁻¹ when the expanded layer is created and a subsequent idealized exponential up to fourfold decrease below the value of 0.01%·s⁻¹. The drop in relative burning rate below the mentioned 0.01%·s⁻¹ occurs first at GG 200–100 N (approx. in the 350th s).
• The surface temperature of all EG-treated samples in 20 s is the highest and in the range of 330–370 °C (at a lower limit of, e.g., 25 K + 180). After that, the temperature drops, and from approx. 120 s, it ranges between 290 °C and 320 °C until the end of the test. The effectiveness of the protective coatings with EG + CNC is obvious at the end of the test, when these samples show a surface temperature half lower (ca. 300 °C) than the untreated wood samples (ca. 600 °C).
• Microscopic analysis demonstrated the uniformity of the coating using graphite 40 D + 500 LST and 25 K + 180, which is probably also related to the very good adhesion of the graphite after expansion. On the other hand, it is not possible to observe a connection between the uniformity of the coating and the effectiveness against weight loss after the fire test since coatings containing cracks, holes, or unevenly distributed graphite flakes achieve slightly lower weight losses of the samples after the test with a radiant heat source (e.g., GG 200–100 N or 25 E + 180 HPH).

At the current stage of research, following the previous evaluation, it is possible to consider GG 200–100 N (Graf Guard) on the American market and 25 K + 180 (Epinikon) on the European market as the best-rated EG from the series tested. The subject of a follow-up study to this pilot work will be to confirm these results with the help of further analyses and measurements (e.g., limiting oxygen index, vertical burning test, or cone calorimeter test). In the future, it would certainly be advisable to verify the effectiveness of other more available cellulose-based surfactants or a suitable combination of several types of EG.