Arundo donax (AD), an invasive species, and olive stone (OS), a by-product of the olive oil extraction industry, were selected as feedstocks. AD canes were harvested locally in Kampel, Slovenia. Crushed OS was provided by the Oljarna Krozera Franka Marzi sp olive oil extraction company (Srgaši, Slovenia) as by-products in 5 mm-sized dry particles.

The chemical composition of AD, is 21.1%, 37.9%, 34.0% for lignin, cellulose, and hemicellulose, respectively (Suárez et al., 2021).

The chemical composition of OS is 32.10%, 26.9%, and 34.8% for lignin, cellulose, and hemicellulose, respectively (Ferreiro-Cabello et al., 2022).

Technical N₂ with a purity of 99.998% (Grade 4.8) was used for the biomass pyrolysis and proximate analysis. N₂ and CO₂ gases with a purity of 99.999% (Grade 5.0) were used for physisorption analysis. Aqueous formaldehyde solution (37% w/v) was purchased from Carlo Erba reagents (Dasti group, Val de Reuil, France).

AD and OS were manually cleaned, ground with a cutting mill grinder (Pulverisette 25/19, Fritsch, Idar-Oberstein, Germany) using a 1 mm mesh, and dried in an oven for 24 h at 105°C to remove excessive moisture. The AD was then subjected to a demineralization treatment by water washing under stirring at 60°C for 1 h to reduce the ash content. The OS, however, was not treated as our previous research indicated that the demineralization effect was negligible for this type of biomass (Zouari et al., 2023). Details about the characterization of the two biomasses can be found in (Zouari et al., 2023).

BC was prepared by pyrolysis of the two biomasses in a tube furnace (Nabertherm RSRC 120-1000/13, Nabertherm, Lilienthal, Germany). Each biomass, AD and OS, was pyrolyzed separately for 30 min at variable temperatures (300, 400, 500, 600, 700, and 800°C) under N₂ gas flow of 300 L/h and a heating rate of 1500°C/h, which is automatically controlled by the tube furnace and optimized to fit in the configuration of slow pyrolysis and to be able to control the thermal inertia of the furnace.

The pyrolysis process generated 12 specimens which were labelled based on the type of original biomass (AD or OS) and the pyrolysis temperature: AD-BC300, AD-BC400, AD-BC500, AD-BC600, AD-BC700, AD-BC800, OS-BC300, OS-BC400, OS-BC500, OS-BC600, OS-BC700, OS-BC800.

After cooling to room temperature, the BC samples were weighed to determine the pyrolysis yield using Eq. 1. Pyrolysis yield = W f W i × 100 (1)

Where W _i is the weight of the raw biomass (g), and W _f is to the weight of the derived BC (g).

The collected BC powders were then ball milled in distilled water media using a planetary ball miller (Pulverisette 5, Fritsch, Idar-Oberstein, Germany) for 30 min at 400 rpm. For milling, stainless-steel jars and 20 mm in diameter balls were utilized, and the ball: BC ratio was equal to 100. The ball-milled specimens were then dried at 105°C for 24 h to allow water evaporation, and the obtained powders were stored in glass containers until further characterization. The AD-BC800 sample was also processed through an activation phase to compare formaldehyde removal between non-activated and activated BC. AD-BC800 was selected for activation based on the formaldehyde removal results and a specific interest in valorizing AD as an invasive species. The BC activation was performed via physical activation under a CO₂ gas flow of 300 L/h at 800°C for 1 h. The obtained sample was identified as: Activated AD-BC800.

Proximate analysis was performed using a thermogravimetric analyzer (LECO TGA801, LECO corporation, Saint-Joseph, MI, United States). The volatiles, ash, and fixed carbon contents were determined according to the ASTM D7582 standard. Values for moisture, volatiles, ash, and fixed carbon were calculated using Eqs 2–5 detailed as specified in Section 14 of ASTM D7582 (ASTM, 2015). Moisture = W i n i t i a l − W 107 ° C W i n i t i a l × 100 (2) Volatiles = W 107 ° C − W 950 ° C W i n i t i a l × 100 (3) Ash = W 750 ° C W i n i t i a l × 100 (4) Fixed carbon = 100 − M o i s t u r e + V o l a t i l e s + A s h (5)

Where W _initial is the initial weight of the BC sample (g), and W _*°C refers to the weight of the BC sample (g) at the designated * temperature.

All tests were repeated three times, and the average values were reported.

The pH was measured in the leachate of the BC specimen by first mixing 0.5 g of BC with 10 mL of deionized water and agitating the mixture in an incubation shaker (IS-OS 20, Phoenix Instruments, Naperville, Illinois, United States) at 23 °C for 1 h. The solutions were then left to settle for 30 min, and the pH was determined using a pH meter equipped with an IS-68 × 591206-B-VSTAR pH/LogR Module (Orion Versa-Star Pro Meter, Thermo Fischer Scientific, Waltham, Massachusetts, United States).

Functional groups on the BC surface were analyzed using a Fourier transform infrared (FTIR) spectrometer (Alpha FT-IR Spectrometer Bruker, Billerica, MA, United States) connected to an ATR (attenuated total reflection) module. Wavelength ranges from 400 to 4,000 cm⁻¹, and a resolution of 4 cm⁻¹ was set to record the FTIR spectra. For each sample, 64 scans were performed, and ten repetitions were done to obtain accurate results and minimize the effect of the atmospheric noise. Opus software was used to collect average spectra, which were further treated by eliminating CO₂ and atmospheric water vapor effects.

Hydrophobicity was assessed by an optical tensiometer (Attension Theta Flex Auto 4 system, Biolin Scientific, Gothenburg, Sweden) that measured the dynamic water contact angle (WCA). Following the method reported by Bachmann et al. (2000), the BC powders were evenly spread and pressed as a thin layer on a flat glass surface covered with adhesive tape (with the adhesive side exposed to the BC particles for proper fixation). The sessile drop method was used to determine the WCA. Five droplets, 4 µL each, were placed on the BC powders, and the WCA value was taken at 30 s. Averages with standard deviations were reported.

The particle size distribution was determined by laser diffraction technique using a Horiba Scientific LA-960A2 analyzer (HORIBA, Kyoto, Japan). The refractive index was set to 1.92. Before each measurement, 1 min of ultrasonication was done to facilitate homogeneous dispersion of the particles. Three repetitions were performed for each sample, and the average values were reported.

The internal porosity of the BC particles was investigated using a physisorption analyzer (Anton Paar Quantachrome Instruments, Boynton Beach, Florida, United States). N₂ and CO₂ gases were used to evaluate the surface area and pores distribution according to Brunauer-Emmett-Teller (BET) and Barrett, Joyner, and Halenda (BJH), and density functional theory (DFT) models, respectively. Before analysis, samples were degassed under vacuum for 12 h at 250°C to remove impurities trapped in the BC’s cavities. Then, for the BET model, the N₂ adsorption-desorption isotherms were determined over a relative pressure range from 0.005 bar to 1 bar at 77 K to determine SSA and total pores volume, represented by mesopores (2 nm–50 nm in width) and macropores (>50 nm in width). In addition, the BJH method was used to determine the meso and macropores size distribution. For the DFT model, CO₂ adsorption-desorption isotherms were collected at a pressure range from 0.001 bar to 0.03 bar at 273 K to determine microporous surface area and micropores (<2 nm in width) volume.

The formaldehyde removal performance of the different BC samples was evaluated using a batch experiment setup, as shown in Figure 1. Before the tests, the samples were oven-dried overnight at 105 °C to remove any residual moisture. Tests were carried out in a closed glass chamber with a volume of 54 L equipped with an electrical fan. A Petri dish containing 1 g of BC powder sample was placed inside the testing chamber on a glass support, and 35 µL of formaldehyde solution was injected into the chamber through the inlet using a 50 µL micropipette. The chamber was hermetically sealed, and the fan was turned on to ensure a homogeneous distribution of the formaldehyde molecules. The initial formaldehyde concentration stabilized at 4 ppm. The experiment was conducted in standard ambient conditions: 23°C temperature, 50% relative humidity, and atmospheric pressure. The changes in formaldehyde concentration inside the chamber were continuously recorded using a formaldehyde electrochemical detection sensor (Stox-HCHO, EC Sense, Schäftlarn, Germany) with 0.1 ppm resolution and 1 s response time. The sensor was connected to the computer, which enabled direct data acquisition using a TVOC-HCHO logger software (ADDproS, Celje, Slovenia).

The formaldehyde removal percentages were calculated according to Eq. 6 (Do et al., 2022), using the residual pollutant concentration measured in the chamber after 1 h. Formaldehyde removal = C 0 − C 1 h C 0 × 100 (6) Where C ₀ is the initial formaldehyde concentration (ppm) and C _1h is the measured formaldehyde concentration after 1 h of the experiment (ppm).

A blank test was carried out by repeating the same experiment without BC particles. The reduction in pollutant concentration calculated for the blank was subtracted from the values obtained for each sample for results correction. Three repetitions were conducted per BC substrate, and the average values were reported.

The re-usability of the spent adsorbent, AD-BC800, was assessed by determining the formaldehyde removal capacity after five cycles of thermal regeneration. The BC specimen was placed in the oven at 100 °C for 1 h to desorb the formaldehyde molecules captured in the BC pores during the previous cycle.

The Pearson correlation test was used to evaluate the relationship between BC’s properties and the formaldehyde removal capacity. Furthermore, the effect of BC’s properties on the formaldehyde removal capacity was evaluated by performing a multiple regression analysis. For both analyses, variables from all samples (AD-BC and OS-BC) were evaluated to remove the biomass type effect. NCSS software (NCSS, LLC, Version 07.1.21) was used for the statistical analysis. The effect of the independent variable was considered to be significant when the p-value was below 0.05.

The formaldehyde removal capacity of AD-BC and OS-BC was compared by unpaired t-test in Excel. A paired t-test was performed to compare the formaldehyde removal performance of non-activated and activated BC.

The formaldehyde removal percentages calculated after 1 h of experimental time in the presence of 1 g of BC are presented in Figure 5. The BC samples prepared at different temperatures exhibited different formaldehyde removal potentials. Indeed, the variation of pyrolytic temperature influenced the physicochemical and structural properties of the resulting BC samples and consequently changed the sorptive capacities. All BC samples achieved a certain formaldehyde removal activity. The percentage of formaldehyde removal ranged between 27% ± 2% and 62% ± 1% and between 26 ± 3%and 64% ± 2% for AD and OS-derived BC, respectively. Moreover, the amount of captured formaldehyde increased by 131% and 144% when the pyrolysis temperature increased from 300°C to 800 C for AD and OS-derived BC, respectively.

The correlation and multiple regression analysis results are represented in Table 5. The variables’ effect was considered statistically significant when the corresponding p-value was below 0.05.

The correlation coefficients showed a positive correlation between formaldehyde removal capacity and fixed carbon, SSA, and microporous surface area. At the same time, a negative correlation was found between the formaldehyde removal capacity and the volatile content (Table 5). The p values from the correlation test showed a statistically significant relation between formaldehyde removal capacity and the four tested variables (p < 0.05). Results from the multiple regression analysis indicated that only microporous surface area was statistically significant when developing an equation to predict formaldehyde absorption based on BC properties. This result suggested that the occurrence of micropores in the BC’s structure was the key parameter for efficient formaldehyde removal.

The unpaired t-test showed no statistically significant difference between the formaldehyde removal ability of BC prepared from AD and OS (p = 0.311). Therefore, the original feedstock did not influence the removal capacity. The slightly higher formaldehyde removal potential of OS-derived BC samples (Figure 5) was likely related to their larger microporous surface areas (Table 4) as compared to the AD-derived BC samples. Despite its lower SSA, AD-BC800 showed higher adsorption compared to AD-BC600. This result was attributed to the higher microporous surface area of AD-BC800 as compared to AD-BC600 (Table 4).

The presence of ultra-micropores concentrated in size range from 0.35 to 0.60 nm (Supplementary Figure S2) further confirms the beneficial effect of smaller micropores occurrence on the BC’s potential in removing formaldehyde. One possible explanation is the small size of the formaldehyde molecule, which has a diameter of 0.25 nm. The small size of the adsorbate was likely favorable for filling narrow micropores. Indeed, prior research reported that the adsorbate and pore size directly influence the adsorption capacity (Yang et al., 2018). Prior research (Yang et al., 2018) reported that the adsorption was not favored i) if the pore size of the porous material (i.e., adsorbent) is smaller than the target molecule’s (i.e., adsorbate) size and ii) if the pore size is much larger than the molecule’s size. Additionally, Liu et al. (2019) used molecular simulation models to investigate the kinetics of formaldehyde adsorption on carbonaceous materials and reported that ultra-micropores with 0.7 nm in width enabled optimal formaldehyde capture.

The increase in formaldehyde removal capacity by the different BC was not constant (Figure 5). Indeed, the formaldehyde removal increased when the temperature increased from 300°C to 400°C, then dropped at 500°C and increased again gradually from 600°C to 800°C. These results were attributed to two formaldehyde removal mechanisms and the predominance of one or the other. Indeed, Chen et al. (2008) highlighted that the pollutant removal by BC is ensured by either the partitioning effect and/or the adsorption effect. They also stated that the noncarbonized and carbonized fractions and surface and bulk properties of the BC material determine the contribution of either mechanism. In our study, the BC samples prepared at mild pyrolytic temperatures (under 500°C) had a higher noncarbonized organic matter content, reflected by the high volatiles and low fixed carbon content (Table 1). Noncarbonized organic matter likely favored the interaction with formaldehyde molecules and enabled the capture through a partitioning mechanism. During partitioning, the formaldehyde molecules diffused into the noncarbonized organic matter characterized by its amorphous aliphatic feature. This mechanism usually involves polar and non-polar interactions. Therefore, the presence of surface functional groups in BC samples prepared at 300°C and 400°C, observed in the FTIR spectra (Figure 2), likely helped to enable further the interaction forces between the formaldehyde molecules and the BC particles. Prior research by Meng et al. (2019) reported that functional groups on the surface of carbonaceous materials favored the interaction with gaseous VOCs.

For BC prepared at higher pyrolytic temperatures (500°C and above), the formaldehyde capture was assisted by the developed porous structure and large surface area. In this case, pores served as adsorption sites for the formaldehyde molecules and enabled the capture through a physical adsorption mechanism. Indeed, the adsorption phenomenon is mainly governed by Van der Waals forces, pores filling, and electrostatic adsorbent-adsorbate interaction (Vikrant et al., 2020). Therefore, the well-developed porosity, specifically microporosity, of the BC prepared at 500 °C and above (Table 4) was a key parameter in formaldehyde removal by BC, as verified by the multiple regression analysis (Table 5). Zhang et al. (2017) studied the performance of BC prepared from different biomasses at variable carbonization temperatures for air purification. They reported that the VOCs removal was governed by a partitioning mechanism in the case of BC prepared at 450°C or below and by a physical adsorption mechanism in the case of BC prepared at 600°C.

AD-BC800s formaldehyde removal capacity increased by 13% after physical activation (Figure 5). This increase was attributed to the improved microporous surface area and micropore volume (Table 4). The activation treatment enabled further development of the porous structure of BC by creating new internal pores and channels. The paired t-test results indicated that the effect of physical activation on the formaldehyde removal capacity was statistically significant (p = 0.001). A deeper study on the air purification performance of BC with and without activation is required to assess the environmental benefits of choosing non-activated or activated BC because of their pollutant removal capacity.

Figure 6 represents the evolution of formaldehyde concentrations inside the test chamber during 1 h of experimental time.

The decrease in formaldehyde concentrations was fast during the first 20 min of the experiment for AD and OS-derived BC. After approximately 20 min, the adsorption slowed, and the formaldehyde concentrations stabilized (Figure 5). Indeed, at the experiment’s beginning, the BC structure pores were free and served as sites for the formaldehyde molecules. When most pores were occupied and the equilibrium time was reached, the adsorption process gradually decreased and stabilized, suggesting that the BC samples became saturated. The equilibrium time was slightly more delayed for samples prepared at temperatures above 500°C given that the formaldehyde capture mainly depended on pores filling. Similarly, Aziz et al. (2017) found that the adsorption capacity of zeolite-based adsorbents towards aromatic VOCs gradually decreased after reaching the respective equilibrium time (approximately 25 min).

Figure 7 represents the formaldehyde removal percentages of AD-BC800 after five cycles of thermal regeneration, which consists of using heat transfer forces to release the adsorbed formaldehyde molecules from the BC pores.

It was observed that the AD-BC800 sample maintained a comparable formaldehyde removal capacity up to the fourth cycle, indicating that it was possible to desorb formaldehyde molecules. After heating, BC’s pores were freed and available for a new adsorption cycle. However, the removal capacity decreased by 13% after the fifth cycle, indicating that the re-usability of BC adsorbent tended to decline after several regeneration cycles. This result was attributed to the changes in structure and surface properties of the BC that likely occurred after repeated heating cycles. Prior research (Ahn et al., 2021) found that the formaldehyde adsorption capacity onto metal-BC decreased by about 50% and 65% after the first and second cycles of thermal regeneration at 80°C for 1 day, respectively. They explained that changes might have occurred in the pores structure and surface chemistry of the metal-BC, contributing to a partial loss in the adsorption performances. Moreover, Yang et al. (2018) investigated the adsorption-desorption cycles of toluene onto activated carbon samples with variable pore sizes. They reported that the pollutants’ molecules accumulated in the activated carbon’s narrow micropores, making it challenging to desorb at low regeneration temperatures. Hence, other methods should be pursued to free the BC pores and allow a long-lasting removal capacity. In this goal, follow-up research is conducted to dope BC particles with active photocatalysts and investigate the integrated adsorption-photocatalytic degradation potential.

The AD-BC800 sample was further tested using FTIR spectroscopy after the fifth adsorption and regeneration cycle to investigate eventual changes in the functional groups’ composition. The spectra collected for the fresh and re-used samples (Supplementary Figure S3) were similar. Hence, the second derivatives were also collected to enhance peak separation and reveal differences between the spectra. No obvious changes in the functional groups’ composition were observed after the re-use and regeneration cycles except that the same peaks were slightly intensified. This result suggests that thermal regeneration effectively desorbs the formaldehyde from the spent material, especially since formaldehyde is characterized by high volatility. Prior research (Ahn et al., 2021) compared FTIR spectra of metallic biochar before and after interaction with formaldehyde. They observed new peaks at 2,900, 2,520, and 2,220 cm⁻¹, attributed to intermolecular hydrogen bonds of formate, carboxylic acid OH stretch, and alkyne stretching, respectively. However, they utilized metallic biochar, which facilitated the conversion of formaldehyde after adsorption and led to the appearance of intermediate molecules. Moreover, they collected the FTIR spectra directly after adsorption without performing a thermal regeneration step, which was not the case in the present study. Similarly (Chen et al., 2013), utilized Ag-containing catalysts for formaldehyde conversion, and they could identify several functional groups related to the occurrence of formaldehyde intermediates, such as formate species (HCOO) after interaction of the catalytic material with the formaldehyde gas. The investigation of variation in the adsorbent properties after several re-use cycles could be interesting to identify the cause of the decrease in formaldehyde removal efficiency after re-using, which will be considered in further research.