Application of design of experiments (DoE) for optimised production of micro- and mesoporous Norway spruce bark activated carbons

In this work, Norway spruce (Picea abies (Karst) L.) bark was employed as a precursor to prepare activated carbon using zinc chloride (ZnCl2) as a chemical activator. The purpose of this study was to determine optimal activated carbon (AC) preparation variables by the response surface methodology using a Box–Behnken design (BBD) to obtain AC with high specific surface area (SBET), mesopore surface area (SMESO), and micropore surface area (SMICR). Variables and levels used in the design were pyrolysis temperature (700, 800, and 900 °C), holding time (1, 2, and 3 h), and bark/ZnCl2 impregnation ratio (1, 1.5, and 2). The optimal conditions for achieving the highest SBET were as follows: a pyrolysis temperature of 700 °C, a holding time of 1 h, and a spruce bark/ZnCl2 ratio of 1.5, which yielded an SBET value of 1374 m2 g−1. For maximised mesopore area, the optimal condition was at a pyrolysis temperature of 700 °C, a holding time of 2 h, and a bark/ZnCl2 ratio of 2, which yielded a SMESO area of 1311 m2 g−1, where mesopores (SMESO%) comprised 97.4% of total SBET. Correspondingly, for micropore formation, the highest micropore area was found at a pyrolysis temperature of 800 °C, a holding time of 3 h, and a bark/ZnCl2 ratio of 2, corresponding to 1117 m2 g−1, with 94.3% of the total SBET consisting of micropores (SMICRO%). The bark/ZnCl2 ratio and pyrolysis temperature had the strongest impact on the SBET, while the interaction between temperature and bark/ZnCl2 ratio was the most significant factor for SMESO. For the SMICRO, holding time was the most important factor. In general, the spruce bark AC showed predominantly mesoporous structures. All activated carbons had high carbon and low ash contents. Chemical characterisation indicated that the ACs presented disordered carbon structures with oxygen functional groups on the ACs’ surfaces. Well-developed porosity and a large surface area combined with favourable chemical composition render the activated carbons from Norway spruce bark with interesting physicochemical properties. The ACs were successfully tested to adsorb sodium diclofenac from aqueous solutions showing to be attractive products to use as adsorbents to tackle polluted waters.


Introduction
Activated carbon is a porous carbon material that has been subjected to reaction with gases and/or chemicals (e.g. ZnCl 2 ) before, during, or after pyrolysis to obtain beneficial physicochemical and adsorptive properties [1]. Activated carbon (AC) can be produced with specific properties to meet the needs for low-cost and high-performance materials for more sustainable technologies [2][3][4]; it is one of the most common materials used as adsorbent [5][6][7][8] and catalyst and electrode material in environmental, chemical, and energy storage applications [9,10].
However, it is not trivial to elucidate how and to which extent these factors affect the AC production process and resulting characteristics [11,13]. Using the design of experiments (DoE) methodology enables identifying significant factors for specific processes and how they affect the resulting product properties [11,13,14]. DoE also provides resource efficiency by creating experimental spaces that maximise the output of relevant data from a minimum number of experiments for building empirical models that correlate responses with the experimental factors [14,15].
DoE has been applied successfully to optimise the experimental conditions for AC preparation from precursors such as sewage sludge [11], coconut shell [16], bamboo [17], seed pods [18], jatropha hull [19], polycarbonate [20], and Turkish lignite [21]. Two factors were analysed when Karacan et al. [21] employed DoE to produce highly porous AC from lignite: chemical impregnation ratio (ranging from 0 to 4) and activation temperature (ranging from 500 to 900 °C). The authors found that the optimum condition to prepare AC with the highest S BET was at a temperature of 800 °C with a chemical impregnation ratio of 2.05. Ayyalusamy and Mishra [22] targeted the optimal conditions for the preparation of ACs from polyethylene terephthalate by evaluating the effect of three main factors (activation temperature, holding time, and chemical impregnation ratio) on two responses (high surface area and yield). Through DoE analysis, impregnation ratio and activation time were found to have the most significant effects on S BET values of the ACs. The optimised experimental response values were 537 m 2 g −1 and 12.57% for S BET and AC yield, respectively, from optimal preparation conditions of 37.63% ratio of chemical activator, 600 °C, and 30 min.
In this work, Norway spruce (Picea abies (Karst.) L.) bark was used as a biomass precursor to produce activated carbons with highly developed porosities. Norway spruce covers large areas of Europe, accounting for more than 30 million ha [23], and has substantial economic importance for the wood market in Scandinavia and Europe. Yearly, only in Sweden, 90 Mm 3 standing volumes are harvested for industrial utilisation. Around 10-15% of this volume consists of bark that falls out as a low-value side-product at the sawmill and pulp industries. Spruce bark is rich in phenolic compounds, such as condensed tannins [24], and its main components comprise cellulose, hemicellulose, and lignin-making it very suitable as a precursor for AC [25].
In this work, Norway spruce bark ACs were produced by chemical activation using ZnCl 2 as activator reagent. The literature reports that ZnCl 2 acts as a superior activating agent compared to the others, producing AC with much higher developed porosity [8,18]. As a result, ZnCl 2 provides ACs with high surface area and mass yield values [8]. In addition, ZnCl 2 is well known for developing AC mesoporosity which is very suitable for solid-liquid separation [18]. The activation with ZnCl 2 works as a standard method to make mesoporous AC. In addition, the ZnCl 2 has the advantage of using lower activation temperatures [8], making the AC preparation process cheaper when compared to other chemical activators, e.g. KOH and H 3 PO 4 . Combining all these factors can be extremely attractive to industry interests, making the process more economically feasible.
One potential drawback of using ZnCl 2 is its potential toxicity if it is not managed correctly in the activation process; however, the acid leaching procedure that the pyrolysed materials are subjected using 6 M HCl at 80 °C under reflux eliminates practically all ZnCl 2 compounds from the carbon matrix [5][6][7][8]18]. This procedure eliminates the toxicity of the AC prepared with ZnCl 2 [5][6][7][8]18].
To the best of our knowledge, only two studies with spruce bark as the main precursor for the preparation of AC have previously been reported [12,26]. Besides, no optimisation study for the production of ACs from spruce bark using DoE has been reported yet. Furthermore, the effects of pyrolysis temperature, holding time, and ZnCl 2 /bark ratio on specific surface area (S BET ), mesopore area (S MESO ), and micropore area (S MICRO ) were investigated and optimised using DoE. In addition, the resulting ACs were fully characterised by elemental analysis, X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, hydrophobicity index, and surface morphologies (SEM). To evaluate a possible suitable and efficient application for the prepared ACs, they were tested to adsorb sodium diclofenac (DFC) from aqueous solutions.

Raw materials
The spruce bark was delivered from a pulp and paper mill and prepared at the Biomass Technology Centre (BTC), Swedish University of Agricultural Sciences, Umeå, Sweden. The wet bark was dried in a custom-made plane drier at 40 °C, shredded with a screen size of 15 mm (Lindner Micromat 2000, Lindner-Recyclingtech GmbH, Spittal an der Drau, Austria), hammer-milled with a screen size of 4 mm (Bühler DFZK 1, Bühler Group, Uzwil, Switzerland), representatively sampled according to ISO 18,135:2017, and cutting-milled with a screen size of 200 µm using a Fritsch Pulverisette 14 mill equipment. The zinc chloride, ZnCl 2 , was acquired from Sigma-Aldrich, and tap water was used throughout the AC preparation.

AC preparation
The ACs were prepared through a one-step pyrolysis activation process according to a procedure previously reported in [5][6][7][8]11]. A total of 15.0 g of spruce bark was blended with the ZnCl 2 , and about 30.0 mL of water was added during mixing to form a homogeneous paste [5][6][7][8]11]. The paste was dried at 105 °C for 24 h, was then put in a metallic crucible, and was heated with a fixed heating rate of 10 °C min −1 under nitrogen flow of 600 mL min −1 in a conventional high-temperature oven. After the set temperature was reached, the sample was treated for a specific holding time. After treatment, the reactor was cooled under an N 2 flow until the temperature reached 200 °C before the sample was taken out. To remove residual ZnCl 2 , samples were boiled for 2 h at 75 °C with a 6.0 M HCl solution under reflux conditions and washed with deionised water until the washing fluid obtained a stable pH.

Experimental design
The spruce bark pyrolysis and activation were performed according to a Box-Behnken design (BBD). BBD is a factorial combination of a minimum of three factors with incomplete block designs. In each block, one factor is held at the central point (0), while the others vary according to four different combination values for the upper (+ 1) and lower (− 1) limits [27,28]. Our experimental design comprised 15 experiments which included 12 factorial points and three centre points for the factors pyrolysis temperature (°C), pyrolysis holding time (h), and ZnCl 2 /bark dry matter mass ratio ( −), as listed in Table 1. Based on the literature and preliminary experiments, these three factors and their ranges were chosen [11,[13][14][15][16][17]. The running order of the experiments was randomised to minimise the effects of the uncontrolled factors.
Minitab software (version 20, Minitab Inc., USA) was used for the BBD analysis to elucidate factor influences on the responses (S BET , S MESO , and S MICRO ) and generate factor values for optimal responses.

AC characterisation
The mass yield (%) was calculated from the dry matter quota after and before activation. Analyses of surface areas (S BET , S MESO , and S MICRO ) and pore volume were carried out by N 2 sorption/desorption analysis (Tri Star 3000 apparatus, Micrometrics Instrument Corp.). Before the analysis, samples were degassed at 180 °C for 3 h in an N 2 atmosphere. The surface areas, pore size distribution, and pore volume were calculated by multipoint nitrogen gas sorptiometry according to the Brunauer-Emmett-Teller (BET) principle. BET and t-plot equations were used to calculate the S BET , S MESO , and S MICRO .
The elemental analysis was performed using an elemental analyser (EA-IsoLink, Thermo Fisher Scientific). Shortly, 0.05 g oven-dried samples were used to determine total carbon (C), nitrogen (N), oxygen (O), and hydrogen (H) contents. The ash mass fraction was determined by subtracting the C, N, O, and H mass fractions from the total mass of the sample.
XPS spectra were collected using a Kratos Axis Ultra DLD electron spectrometer using a monochromated Al K α source operated at 150 W. An analyser pass energy of 160 eV for acquiring survey spectra and a pass energy of 20 eV for individual photoelectron lines were used. The samples were gently hand-pressed using a clean Ni spatula into the powder sample holder. Because activated carbon is conductive, no charge neutralisation system was used. The binding energy (BE) scale was calibrated following the ASTM E2108 and ISO 15,472 standards. Processing of the spectra was accomplished with the Kratos software.
Raman spectra were collected using a Bruker Bravo spectrometer (Bruker, Ettlingen, Germany) attached to a docking measuring station. Shortly, 0.5 g of AC samples were manually ground using an agate mortar and pestle, placed in 2.5-mL glass vials, and scanned in the 300-3200 cm −1 spectral range at 4 cm −1 resolution for 256 scans. Min-Max normalisation over the 1000-2000 cm −1 region and smoothing (9 points) was done using the built-in functions of the OPUS software (version 7, Bruker Optik GmbH, Ettlingen, Germany). No baseline correction was needed.
The H 2 O vapour adsorption isotherms were determined by dynamic vapour sorption (DVS Advantage, Surface Measurement Systems). Shortly, 0.02 g was used to get the isotherms. Equilibrium moisture contents (EMCs) were collected by recording the sample weight at each 1% RH in the 0-10% range, each 2.5% RH in the 10-20% range, each 5% in the 20-50% range, and each 10% in the 50-95% range during adsorption, and at each 20% during desorption.
The hydrophobicity/hydrophilicity index (HI) was calculated according to a method previously reported in the literature [22]: 0.3 g of each AC was placed into 5-mL beakers and inserted into plugged 1.5-L E-flasks with saturated atmosphere solvent vapour (water or n-heptane) using 80 mL of each solvent. The beakers were placed in the centre of the E-flasks to avoid contact with the flask walls. After 24 h, the beakers were removed and weighed. The weight gained was used to calculate the maximum H 2 O vapour adsorption.

Diclofenac (DCF) batch adsorption tests
The DCF solution used for the adsorption tests was at an initial concentration of 2000 mg L −1 and a pH of 6.0. Aliquots of 20.00 mL of DFC were added to 50.0-mL Falcon flat tubes containing 30 mg of each AC. The Falcon tubes containing DCF and ACs were agitated in a shaker model TE-240 for 4 h. Afterwards, to separate the DFC and ACs, the flasks were centrifuged. After adsorption, the residual solution of DCF was quantified using a UV-Visible spectrophotometer (Shimadzu 1800) at a maximum wavelength of 285 nm. The amount of DCF adsorbed by the ACs was calculated using Eq. (1): Table 2 shows S BET , S MESO , S MICRO , S MESO% , S MICRO% , pore volume, and mass yield of the prepared ACs. The mass yield of the produced ACs ranged from 30.3 to 41%. The lowest yield was obtained for AC15 (900 °C, 3 h, and ZnCl 2 /bark ratio 1.5), and the highest for AC6 (700 °C, 2 h, and ZnCl 2 / bark ratio 2). It has previously been reported that a higher pyrolysis temperature and longer holding time reduce the mass yield [29,30] due to more volatile compounds exiting the biomass during the pyrolysis.

Yield and textural characteristics of the ACs
For instance, Bergna et al. [29] produced ACs from birch and spruce wood chips at two different activation holding times: 2 and 4 h, with reported mass yield values of 10.4% and 5.5% for birch and 15.2% and 11.5% for spruce, respectively. Sulaiman et al. [30] pyrolysed cassava stem to produce AC with yields varying from 3.8 to 25.7%. In another work [31], ACs were prepared by using CO 2 and MgCl 2 as activation agents. The yield of the AC prepared with CO 2 was 6.54%, while MgCl 2 activation resulted in a yield of 18.1%. The abovementioned studies presented a substantially lower yield compared to the ACs prepared in this work. Generally, ZnCl 2 activation gives higher yields than activation with other chemical reagents [29,30]. Impregnation with ZnCl 2 results in degradation of the cellulosic material, and carbonisation produces dehydration that results in charring and aromatisation of the carbon skeleton, thereby avoiding huge losses [6,8].
All produced ACs, regardless of experimental conditions, exhibited high S BET values ( Table 2). The highest S BET value of 1374 m 2 g −1 was obtained through pyrolysis at 700 °C for 1 h and a ZnCl 2 /bark ratio of 1.5 (AC14). This AC preparation condition corresponded to the lowest temperature and holding time and the mid-level for the ZnCl 2 /bark ratio in the experimental design. Also, ACs with S BET above 1300 m 2 g −1 were generated at three other experimental conditions (see Table 2), indicating that spruce bark is a promising and feasible precursor for producing ACs with high specific surface areas and well-developed porosities using different conditions.
As indicated by S MESO and S MESO% values, the spruce bark ACs showed mesoporous structures; for instance, the percentage of mesopores (with relation to S BET values) of AC6, AC12, and AC2 was 97.4%, 96.1%, and 93.4%, respectively. The presence of mesopores is highly desirable in ACs for adsorption and energy storage applications because they ensure wetting and liquid transport throughout the bulk of the AC [2,3,9,10]. In contrast, AC8 (prepared at 800 °C, 3 h, ZnCl 2 /bark ratio of 1) and AC15 (prepared at 900 °C, 3 h, ZnCl 2 /bark ratio of 1.5) exhibited highly microporous structures: 94.3% and 80.8% of the total S BET , respectively.
Comparing the obtained results in Table 2 with international literature, Dos Reis et al. [11] produced ACs from sewage sludge and reported S BET values up to 679 m 2 g −1 .
Danish et al. [32] used Acacia mangium wood to make AC and reported an S BET of 1767 m 2 g −1 , with 95% of the total surface area as mesopores. In another work, Li et al. [20] prepared ACs from Chinese chestnut burs; the highest S BET was 1254.5 m 2 g −1 with a microporosity ratio of 87.2%. Bouchelta et al. [33] produced ACs from biomass of Algerian date pits with high-developed microporosity. The AC with the highest S BET (1467 m 2 g −1 ) was obtained at 700 °C and 4 h of holding time. They concluded that increasing the holding time favoured the development of microporosity and led to high S BET values. Mistar et al. [34] synthesised ACs from yellow bamboo (Bambusa vulgaris "Striata"), and all prepared ACs presented predominantly microporous structures. The AC's microporosity increased significantly with pyrolysis temperature and chemical activator ratio.
Galiatsatou et al. [35] used olive pulp and peach stones as precursors for AC preparation and concluded that extended holding times favoured mesopore development. Hu et al. [36] reported that high chemical ratios, long holding times, and high temperatures increase the mesoporosity in ACs. They suggested that a high ZnCl 2 /biomass ratio creates new mesopores by widening the micropores.

Response surface plots
The response-contour surface plot is a valuable and visual tool for understanding how selected factors affect the studied responses by providing a "map" of the fitted model's response values at all possible settings within the range of the experimental design. However, it is important to remember that the visuals represent model data, not experimental values, to avoid overinterpretation of results and underestimating model errors. Figure 1A-I show trends for the factors' effects and their interactions on the specified responses for the ACs. In addition, optimum conditions can be found within the region with the highest response values. ACs with the highest S BET are obtained at shorter activation holding times, low pyrolysis temperatures, and high ZnCl 2 /bark ratios (see Fig. 1A-C). In more detail, ACs with the highest S BET values are obtained at holding times no longer than 2.5 h, pyrolysis temperatures no higher than 800 °C, and a ZnCl 2 /bark ratio up to 2.0.
The highest mesopore surface area (S MESO ) is obtained at pyrolysis temperatures up to 800 °C, holding times up to 2 h, and ZnCl 2 /bark ratios around 1.5-2 (see Fig. 1D-F). High micropore surface areas (S MICRO ) require longer activation holding times (close to 3 h) and a pyrolysis temperature up to 800 °C. Here, the effect of the ZnCl 2 /bark ratio is more challenging to interpret due to possible interaction effects (see Fig. 1G-I).
Surface contour plots show the behaviour of the studied factors on the S BET , S MESO , and S MICRO responses; however, elucidating factor and interaction importance need additional analysis. For this purpose, Pareto charts and normal plots are used and analysed in the below section.

Pareto charts and normal plots
From experimental results, the effects of the factors and their interactions on the S BET , S MESO , and S MICRO can be statistically evaluated by use of a Pareto chart; it aims to graphically demonstrate the significance and the relationship of the studied factors on the related responses [11,13,15]. The bars visualise the factors and their interactions [11]. The dotted line is associated with the p-value [11]; the factors or interactions that exceed this line have a statistically significant effect on the studied response [11]. Therefore, for the S BET response, the Pareto chart ( Fig. 2A) shows that the ZnCl 2 /bark ratio (C), followed by pyrolysis temperature (A), is statistically significant (α = 0.05) factors while holding time (B) in the studied range does not influence the S BET values significantly. The square for the factors ZnCl 2 /bark ratio (CC) and holding time (BB) and the interaction between holding time and ZnCl 2 /bark ratio (BC) have significant but weak effects on the S BET . For the S MESO , the interaction between holding time and ZnCl 2 /bark ratio (BC) is the most critical factor (Fig. 2B), while for S MICRO , holding time (B) is the most influential. All other factors with t-test values that do not cross the reference line are statistically insignificant (α = 0.05) [11,13].
Morali U et al. [36] optimised the preparation of AC from sunflower seed extracted meal using DoE. They concluded that the ZnCl 2 /biomass ratio was one of the most influential factors for the S BET values of the ACs. In another work, dos Reis et al. [11] used DoE to optimise the AC textural properties and similarly found that the ratio of ZnCl 2 played a crucial role for the S BET values. In carbon activation, ZnCl 2 is intercalated into the carbon matrix to produce pores at temperatures above its melting point, which act as templates to create and develop porosity [35,36]. The melted ZnCl 2 expands the existing micropores into mesopores leading to a higher S BET [35,36]. Moreover, in this activation process, reactions between the carbon atoms in the precursor material and ZnCl 2 in extended carbon interlayers contribute to creating pores [9,35]. In addition, ZnCl 2 chemical activation increases the carbon content by forming an aromatic graphitic structure, leading to a higher S BET .
The Pareto chart helps to identify the factors and interactions most decisive for a selected response [11]. However, it does not show if the effect is positive or negative; a normal plot provides such information (see Fig. 3). In the normal plot, the significant factors are labelled with red symbols; those that negatively affect the response are situated far to the left of the fit line and vice versa [11]. Consequently, high temperature decreases the S BET (Fig. 3A), and a long holding time increases the micropore area (Fig. 3C). A high ZnCl 2 /bark ratio increases the S BET values (Fig. 3A), and interactions between holding time and the ZnCl 2 /bark ratio increase the mesopore area. The remaining factors and their interactions have no statistically significant effects (α = 0.05) on the responses under the studied conditions.
Our results agree with previous literature, showing that the concentration of ZnCl 2 during activation substantially affects the S BET values of the ACs [5-8, 37, 38]. This result is reasonable because when the ZnCl 2 /precursor ratio increases, more metallic ions are complexed with the functional groups present in the biomass during the mixing step (between biomass and ZnCl 2 ) [39,40]. After pyrolysis, the inorganic compounds occupy a considerable volume of the carbonaceous matrix. When these inorganics are leached out during the HCl washing step, the resulting AC materials display well-developed pore network structures and high S BET [39,40].
In addition, the negative effect of the temperature on the S BET is also reported in the literature [11,41]. The authors concluded that increased pyrolysis temperatures induce shrinkage of the carbon structure, resulting in a reduction in S BET and pore volume. This finding is also reported by several researchers, which showed a negative effect of pyrolysis temperature on the S BET [3,8,21,25].

AC preparation: comparison with literature
Any precursor rich in carbon can be converted to activated carbon with tailored properties targeting different applications such as adsorbents, catalysts, and anodes for energy storage devices. However, to produce an efficient AC, careful considerations of pyrolysis conditions must be made since the conditions play an essential role in the final AC's characteristics. Moreover, robust optimisation studies are required to make the production process sustainable and environmentally friendly. Fig. 3 Normal plots of the standardised effects of pyrolysis temperature, holding time, and the ratio of ZnCl 2 on A S BET , B S MESO , and C S MICRO for the AC preparation (at a minimum statistically significant level of 95% confidence). Negative and positive effects are separated at the 50% level DoE has been used to optimise AC preparation in several previous studies (see Table 3). Comparing the results obtained in this work with international literature gives a more accurate evaluation of whether the DoE was applied and appropriately conducted and how effective it can be to optimise the preparation of porous AC. For an efficient comparison, the optimisation of the production of ACs, using several different DoE, is displayed in Table 3. Table 3 shows that the pyrolysis temperature, holding time, and chemical/precursor ratio are the most studied factors, and S BET is the most evaluated response. A large variety of carbon precursors have been treated under different preparation conditions. Table 4 shows a considerable variation in response values (especially for S BET ) with biomass precursors and preparation conditions.
Abioye et al. [42] employed a DoE Box-Behnken design (with 15 runs in total) to optimise oil palm shell ACs' textural properties. Three factors were evaluated: pyrolysis temperature: 800, 850, 900 °C; holding time: 20, 30, 40 min; and CO 2 flow rate: 200, 300, 400 cm 3 min −1 . Two responses were analysed: S BET and micropore volume, S MICRO . The authors found that the holding time had the most profound effects on S BET and S MICRO values [42]. Also, the S BET values ranged from 291 to 574 m 2 g −1 . The optimum condition was identified as 900 °C, 40 min, and CO 2 flow rate: 400 cm 3 min −1 .
dos Reis et al. [11] used DoE (a full factorial design with 12 runs in total) to optimise the production of ACs from sewage sludge through conventional and microwave heating methods. The authors evaluated three factors with S BET as the response. The factors were pyrolysis temperature or microwave power, holding time, and chemical activator proportion. Factor conditions were temperature (500, 650, 800 °C), microwave power (700, 840, 980 W), holding time (15, 37.5, and 60 min for conventional pyrolysis and 8, 10, and 12 min for the microwave methods), and the ZnCl 2 :sludge ratio (0.5, 1.0, and 1.5). The optimum factor combination for producing AC with the highest surface area was found at 500 °C, 15 min, and a ZnCl 2 :sludge ratio of 0.5 for conventional pyrolysis, while it was 980 W, 12 min, and a ZnCl 2 :sludge ratio of 0.5 for microwave pyrolysis. The optimal conditions generated ACs with S BET values of 679 m 2 g −1 and 501 m 2 g −1 for conventional and microwave pyrolysis, respectively. The most critical factors for conventional pyrolysis were temperature and the interaction between pyrolysis temperature, holding time, and chemical activator. The interaction between radiation power, holding time, and the chemical activator ratio was the most influential for microwave pyrolysis.
Comparing our results with the literature data (Table 3) reveals that ACs with very high S BET and well-developed mesoporosity can be produced from Norway spruce bark under optimal pyrolysis conditions. The literature on AC production covers a plethora of precursors and experimental matrices of pyrolysis conditions and activation methods. It is not always possible to compare the results from different studies, but it is safe to conclude that the optimum methodology varies with the precursor properties. Also, further systematic studies are needed to increase the knowledge on how suitable combinations of biomass precursors, pyrolysis conditions, and chemical activation procedures to obtain ACs with textural and functional properties for specific applications. Table 4 shows the results for the elemental composition of the prepared ACs. For comparison, a typical commercial AC has 88% of C, 0.5% of H, 0.5% of N, and 3-4% ash [47]. All ACs prepared from spruce bark through ZnCl 2 activation had a C content above 93% (Table 5), which can be considered a high value compared with the literature. For example, Correa et al. [47] employed a range of different biomasses, based on alpha-cellulose, xylan, kraft lignin, to produce ACs, and their C content varied from 76.9 to 87.8%. In another work [48], coconut shell was used to prepare ACs, and the highest C content was 82.66%. High carbon content reflects good AC characteristics due to the carbon's physicochemical, electrical, and thermal properties. Also, an elevated C content is often a sign of low ash content, and ashes in AC are often responsible for reducing S BET and surface functionalities. Surface functionality is vital for many AC applications, such as adsorption of pollutants in waters and as electrode materials in energy storage systems. The prepared ACs had relatively high oxygen content. This feature may reflect a high hydrophilicity index, a good quality for AC applications in solid/liquid systems [47]. In addition, the ACs had high C/O ratios, which indicates the formation of hydroxyl groups on the AC surfaces (this was analysed by XPS analysis and discussed in the following section). Also, high C/O ratios indicate that the spruce bark AC structures are similar to near-perfect graphene layers. Sample AC15 had the highest (57.3) and AC14 the lowest (33.7) C/O values.

X-ray photoelectron spectroscopy (XPS)
The elemental composition and chemical state of the ACs were analysed with XPS. It was employed to understand AC surface chemistry and assess the effects of the ZnCl 2 treatment and the relationship between spruce bark composition and pyrolysis conditions. Four AC samples (AC1, AC9, AC13, and AC14) were studied (Fig. 4). The samples were   [46] selected based on their S BET values (from lowest (AC13) to highest (AC14) and two in between (AC1 and AC9). As can be seen, Fig. 4 shows spectra related to C1s and O1s, carbon and oxygen bonds, respectively [14,49,50]. The deconvolution of the C1s spectrum indicates that all samples contain graphitic/aromatic C = C or hydrocarbon C-C (~ 284.3 eV), C-O in phenolic hydroxyl or ether groups (~ 286.5 eV), carbonyl C-O (~ 287.5 eV), and ester O = C-O (~ 289.1 eV), graphitic carbon being the major component [49].
O1s spectra were deconvoluted to three chemical oxygen states with binding energies at around 530.8-531.2 eV, 532.6-533.1 eV, and 534.8-536.3 eV. These binding energies could correspond to oxygen singly bonded to carbon in aromatic rings, in phenols and ethers (533.2-533.8 eV), or to oxygen double-bonded with carbon in carbonyl and quinone-like structures (530.8-531.2 eV) [51], confirming the presence of some functional groups on the ACs' surfaces.
The quantitative information from XPS is shown in Table 5. The surface C content of all four ACs did not vary significantly (from 94.5 to 94.8%). In contrast, the O content showed a larger variation (from 4.4 to 5.2%), and the values match the values presented by elementary analysis (see Table 4). The highest O1s content (5.2%) was found in the AC14 sample, suggesting that this sample had more functionalities when compared to the others, especially for functional groups related to C = O, C-OH, and C-O-C (see Table 5). The AC14 sample had the highest S BET and amount of functional groups, suggesting high efficiency as an adsorbent for water decontamination and electrode material for energy storage systems [2,9,18,37].

Raman spectroscopy
Raman spectroscopy analysis was performed to evaluate the degree of graphitisation of the prepared ACs. Using Raman, it was possible to determine the ratio of the I D / I G bands (see Fig. 5). The lower I D /I G value indicates that ACs exhibit more perfect and orderly graphite structures with a high graphitisation degree [52][53][54], while a higher I D /I G peak intensity ratio reveals more structural defects in carbon materials. Moreover, I D /I G values also serve to identify the size of the sp 2 domain, related to graphene structure, in the AC structure [53,55]. Figure 5 shows that AC6 (700 °C, 2 h, 2) had the highest graphitisation degree, followed by AC14 (700 °C, 1 h, 1.5)-presenting I D /I G values of 0.85 and 0.88, respectively. These values indicate that both samples (AC6 and AC14) had highly disordered structures. Interestingly, these samples also presented S BET values among the highest in the experimental design: 1343

Hydrophobic-hydrophilic analysis (HI)
The surface characteristics of the prepared ACs were evaluated by solvent vapours of different polarities using n-heptane (less polar) and water (more polar) [18,56,57]. The hydrophobic-hydrophilic ratio is an important parameter to estimate the tendency of the material to adsorb compounds that are organic or water-based [56]. Figure 6 displays the calculated mass ratio of n-heptane to water uptake by all ACs. It is observed that all the AC surfaces are predominantly hydrophilic (see Fig. 6) since HI values are lower than 1.0. Hydrophilic groups on AC surfaces are related to H-bonding and oxygen groups (already shown in Fig. 4) known to be present on biomass AC surfaces [18]. However, AC2, AC8, and AC12 presented HI values above 1.0, which means they have a hydrophobic behaviour. Interestingly, these three samples were made with a ZnCl 2 :bark ratio of 2 and at higher temperatures (800 and 900 °C); it seems that a high ZnCl 2 concentration and high temperature lead to more hydrophobic surfaces. Leite et al. [18] prepared AC from avocado seed biomass and stated that an increase in pyrolysis temperature increased the HI values of the ACs. However, the hydrophobic/hydrophilic behaviour of ACs is not only influenced by the experimental pyrolysis condition and activation method. It also depends on the number of functional groups and types present on the AC's surface, the AC's aromatisation rate, and the chemical nature of the biomass precursor [35,49,57].

Surface morphology (SEM analysis)
SEM was employed to examine the effects of pyrolysis and activation conditions on the surface morphology characteristics of the AC samples. All SEM micrographs (Fig. 7) show rugosity, an irregular structure, holes, and cavities to a lesser or greater extent that agrees with the S BET and pore structure analyses. The SEM pictures reveal differences that indicate that the pyrolysis conditions and the ratio of ZnCl 2 did influence the surface characteristics of the ACs.
Macropores are present in all samples. Macropores are very important if the AC is used as an adsorbent to remove water pollutants or electrodes in energy storage systems. They serve as vectors for the solution passage through the pores until it attains the micropores and mesopores. Depending on the preparation conditions, the AC external surfaces exhibited macropores of different sizes and shapes, reflecting their S BET , S MICRO , and S MESO values.
The ACs were found to be very porous regardless of the preparation condition. During the pyrolysis, it seems that the cavities and holes on the AC's surfaces resulted from the ZnCl 2 state transformation-its low melting point at 290 °C and the boiling point at 732 °C are fused into the biomass matrix, thereby creating a denser structure and well-developed pore network [5,6,11,18].

Water vapour sorption
Water vapour sorption isotherms for the AC1, AC4, AC9, AC13, AC14, and AC15 are shown in Fig. 8. According to the IUPAC classification [58], all isotherms are very close to type V, characterised by little water uptake at low relative pressures and the presence of a hysteresis loop over a big part of the pressure range. Although the isotherms exhibited similar types and shapes, the hysteresis forms differ, indicating that the pyrolysis conditions influence the AC characteristics as previously discussed in the manuscript.
The H 2 O isotherms were not influenced by the AC's textural properties, such as S BET , because the AC with the highest S BET presented one of the lowest H 2 O uptake values. Hence, the AC water vapour sorption may, contrary to the N 2 adsorption, be very dependent on the surface chemistry, such as the presence of functional groups on the AC surface.
Water adsorption isotherms for most porous carbons exhibit a hysteresis loop. The size and shape of the loop varied between ACs with different pore sizes and pore structures. From the isotherms displayed in Fig. 8, it is possible to see that AC8-the most microporous sample (94.6%)exhibited the longest and widest hysteresis. Gallego-Gomez reported that the completion of the micropore filling delays the onset of multilayer adsorption, leading to hysteresis at high relative pressure [59].

DFC adsorption
In this work, Norway spruce bark activated carbons were tested to remove DCF from aqueous solutions to evaluate their suitability for the adsorption process (see Fig. 9). The ACs displayed very high adsorption capacity values (272-417.4 mg g −1 ). Their physicochemical properties can explain the high efficiency of the AC in removing DFC. All ACs presented very high S BET and well-developed pore structure in the range of mesoporosity with few samples mainly composed of micropores. Both micro-and mesoporous ACs are highly efficient to adsorb DFC molecules with a small molecule size (1.015 nm) [62]. The DCF molecule can easily be accommodated in pores bigger than its size, which is the case of micropores up to 2.0 nm and mesopores up to 50.0 nm. The results exhibited in Fig. 9 indicate that the prepared ACs could be successfully employed to remove DFC and organic emerging compounds from aqueous solutions.
Comparing our results with the literature data, the adsorption capacity (q) of the best performing AC (AC14) is comparable and even higher to the adsorption capacities of many different adsorbents reported in the literature (see Table 6). Compared to ACs produced in other works, the high diclofenac adsorption capacity efficiency can be attributed to the small particle size and high developed porosity and the functional groups present on the surface of the ACs.  Table 6 shows that Norway spruce bark AC (AC14) had the second-highest adsorption capacity (417.4 mg g −1 ) among all presented adsorbents. The highest q value (444.44 mg g −1 ) was reached by polyethyleneimine-functionalised sodium alginate/cellulose nanocrystal/polyvinyl alcohol core-shell microspheres (PVA/SA/CNC)@ PEI). Although PVA/SA/CNC)@PEI presented a higher q value than AC14, its production cost is incredibly much higher when compared to that of AC14. Consequently, if the production cost is added to the desirable properties, ZnCl 2 -activated Norway spruce AC could be classified as an excellent adsorbent to remove DFC from aqueous solutions.

Conclusion
Norway spruce bark can be a very suitable feedstock for the production of activated carbons with well-developed porosity. Response surface methodology based on Box-Behnken design was confirmed to be effective in optimising activated carbons' preparation. Three responses were studied: S BET , mesoporosity (S MESO ), and microporosity (S MICRO ). The maximum values for the responses were obtained at the following conditions: S BET : 700 °C, 1 h, and 1.5:1 (sample AC14), S MESO : 700 °C, 2 h, and 2:1, and S MICRO : 800 °C, 3 h, and 2:1 ZnCl 2 :bark ratio.  The highest S BET , S MESO , and S MICRO were 1374, 1311, and 1117 m 2 g −1 , respectively. The ZnCl 2 ratio and pyrolysis temperature were the most critical factors for the S BET , while the interaction between temperature and ZnCl 2 ratio was the most significant factor for S MESO . For the S MICRO , holding time was the most important factor. The spruce bark AC showed predominant mesoporous structures and relatively high yield compared to the literature. Based on I D /I G values, the AC samples with the highest S BET values presented much more disordered structures. The water vapour results suggested that the H 2 O adsorption was not controlled by the textural properties of AC but rather by their chemical properties and the availability of functional sites. The DFC adsorption data showed that the AC displayed very high adsorption capacity values (272-417.4 mg g −1 ), and the sample with the highest S BET (AC14) also exhibited the highest q value (417.4 mg g −1 ) for DFC removal.
The large surface area, interesting chemical structure, and high DFC adsorption capacity of the ACs in this study show a great potential of Norway spruce bark residues as a precursor material for the production of AC with good adsorption properties.
Funding Open access funding provided by Swedish University of Agricultural Sciences. This research was funded by the Treesearch Postdoctoral program, Bio4Energy-a Strategic Research Environment appointed by the Swedish government, and the Swedish University of Agricultural Sciences. Ms. Mathieu thanks ERASMUS for the financial support.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.