1 Introduction

Production of top platform chemicals, such as levulinic acid and 5-hydroxymethylfurfural, from cellulosic biomass acid-catalyzed hydrolysis has attracted much attention [1]. Both levulinic acid and 5-hydroxymethylfurfural can be converted into value-added chemical intermediates, fuels, polymers, solvents, and resins [1]. Levulinic acid was even regarded as a chemical bridge connecting biomass and petroleum processing, due to its importance in producing a variety of liquid biofuels (e.g., valerate esters, γ-valerolactone, liquid alkenes) [2, 3]. Humins, the solid wastes, are inevitably generated during the biomass acid hydrolysis process, leading to serious decreases of biomass utilization efficiency [4]. The humin yields reached 21–36 wt% from glucose acid hydrolysis for production of levulinic acid and/or 5-hydroxymethylfurfural [5,6,7,8,9], and as high as 30–50 C% during levulinic acid production from cellulose hydrothermal conversion [10]. It is meaningful to develop high value-added products from waste humins for the whole biorefineries. Several works have focused on conversion of humins to bio-oils [11], syngas [12], organic acids [13], and composites and polymeric materials [14, 15]. Importantly, the humins are stable polymer containing 50–66% carbon [9, 13, 16, 17]. The carbon rich in property and polymer structure made humins a potential feedstock for production of porous carbon materials, e.g., activated carbons [18]. For comparison, terrestrial higher plants are routine materials for activated carbon production, while their average carbon content is only about 44% [19]. Activated carbons are widely applied in manufacturer and our daily life, such as wastewater treatment, air clean, and detoxification agent for biomass hydrolysate [20, 21]. They could be prepared through several technologies, in which chemical activation is the most conventional one with alkali (e.g., KOH), salt (e.g., ZnCl2), and acid (H3PO4) [22, 23] as the activating agents.

As humins are formed from acid-catalyzed polymerization reactions, the use of acidic activating agent (e.g., H3PO4) might avoid excessive corrosion of humins carbon structure, which would give a high yield of activated carbon. In addition, a part of the mineral acid used for acid hydrolysis could be adsorbed on raw humins [5]. The use of acidic activating agent does not require the neutralization reaction or excessive water washing to remove acids before the activation process.

To make waste humins profitable, H3PO4 activation and pyrolysis process were originally determined for activated carbon production from humins. The structure and properties of the activated carbons were systematically investigated, and the performances were elaborated by adsorption of methylene blue (MB), a common dye contaminant.

2 Materials and methods

2.1 Materials

Humins were obtained by a routine H2SO4 (0.05 mol/L) catalytic hydrolysis process [13]. In brief, 80 mL of acidic glucose solution (1 mol/L) was put in a 100-mL unstirred polytetrafluoroethylene reactor, and then the reactor was sealed and heated in an oven. After 1 h of heating time, the reactor reached 160 °C and then kept at the temperature for 6 h. After reaction and cooling down, the solids were filtrated, washed, and dried at 100 °C to obtain humins. Methylene blue (MB) was purchased from Macklin Reagent, Shanghai, China.

2.2 Activation treatments

In the activation process, 3.0 g of humins was mixed with 20 mL 40% H3PO4 solution and stirred for 12 h, and the slurry was further oven dried at 120 °C for 12 h. The dried sample was heated to a preset temperature (300–700 °C) with a heating rate of 10 °C/min in a horizontal cylindrical furnace, and then kept at the preset temperature for 2 h with nitrogen protection (100 mL/min nitrogen flushing rate). The prepared samples were washed repeatedly with DI water until the pH level is back to 7. The sample was dried at 120 °C for 12 h to obtain activated carbon and, labeled as ACX according to each pyrolysis temperature X. For example, the activated carbon prepared at 400 °C is labeled as AC400.

2.3 Sample characteristics

The morphology of the samples was analyzed by a JEOL JSM-6701F environmental scanning electron microscopy (SEM) system (Tokyo, Japan). The functional groups were analyzed by Fourier transform infrared spectroscopy (FT-IR) (Bruker Tensor 27, Karlsruhe, Germany) using KBr as an interference-free matrix. The samples were mixed and ground with KBr by a ratio of 1:10, and the scanning region ranged from 1000 to 3600 cm−1. X-ray diffraction (XRD) was analyzed by a Rigaku D/max-IIIA X-ray diffractometer (TX, USA). Thermogravimetric (TG) analyses were conducted by a Netzsch 209F3 (NETZSCH-Gerätebau GmbH, Germany). The ACs were placed in a sample pan under air and heated from room temperature to 600 °C with a heating rate of 10 °C/min. The nitrogen adsorption–desorption isotherms were analyzed by a Micromeritics Instruments TriStar II (GA, USA) to evaluate the Brunauer–Emmett–Teller (BET) surface area and total Barrett, Joyner, and Halenda (BJH) pore volume with pores from 1.7 to 300 nm. AC400 was also analyzed by an Autosorb-iQ-C chemisorption–physisorption analyzer (Quantachrome, USA) in order to measure the micropores with diameter < 1.7 nm. All the texture characteristics of the ACs were determined by low-temperature (77.4 K) nitrogen adsorption with the addition of 0.07–0.1 g sample in the quartz testing tube. The total acid groups were determined by the Boehm titration method [24]. Briefly, the ACs were added into 0.02 mol L−1 NaOH solution in a glass bottle, and the bottle was shaken in a shaking bed for 24 h. After shaking and filtration, a pre-prepared amount of HCl solution (0.03 mol/L) was added to the solution. Then, the solution was back-titrated with NaOH solution (0.02 mol L−1) to neutral using phenolphthalein as an indicator.

2.4 Adsorption and desorption process

Adsorption of MB was performed in 50-mL sealed glass bottles. Thirty milliliters solution with initial concentration of 1000–2000 mg/L was placed without any pH adjustment. Fifty-milligram humins-based activated carbon was added into each bottle and placed in an isothermal shaker at 200 rpm and 30 °C for 24 h. After each adsorption process, the MB solution was separated from the activated carbon by filtration, and the MB concentration was analyzed by UV-vis spectrum analysis at 663 nm.

The regeneration of used AC400 was conducted by organic solvent extraction in a 60-mL Soxhlet extractor. The AC400 (50 mg) was first shaken with 2000 mg/L MB solution for 24 h. The MB-adsorbed AC400 was put in the Soxhlet extractor, and then extracted by three common organic solvents, i.e., methanol, acetone, dichloromethane (DCM) for 24 h. The volatilization temperature of the methanol, acetone, and DCM in the bottom of the Soxhlet extractor was 75 °C, 70 °C, and 55 °C, respectively. After the extraction process, the treated AC400 was dried at 45 °C to evaporate the organic solvents to obtain the regenerated AC400. The regenerated AC400 was reused for MB adsorption as described above.

2.5 Simulation and calculation

Two generally used kinetic models, pseudo-first-order and pseudo-second-order, were used for the kinetic dynamic simulation. The linear form of pseudo-first-order rate is shown in Eq. (1):

$$ \ln \left({q}_{\mathrm{e}}-{q}_{\mathrm{t}}\right)=\ln {q}_{\mathrm{e}}-{k}_1t $$
(1)

The linear form of pseudo-second-order rate is shown in Eq. (2):

$$ \frac{t}{q_{\mathrm{t}}}=\frac{1}{k_2{q}_{\mathrm{e}}^2}+\frac{t}{q_{\mathrm{e}}} $$
(2)

where qt and qe is the amount of MB adsorbed on the sorbent at time (min) and at equilibrium (mg/g), respectively. k1 is the first-order rate constant (h−1) and k2 is the pseudo-second-order rate constant (g/mg min).

The kinetics parameters and correlation coefficients (R2) were determined by linear regression. The applicability of both kinetic models was verified by normalized standard deviation △q (%) as shown in Eq. (3):

$$ \triangle q\ \left(\%\right)=100\times \sqrt{\sum \frac{{\left[\frac{q_{\mathrm{exp}}-{q}_{\mathrm{cal}}}{q_{\mathrm{exp}}}\right]}^2}{\left(n-1\right)}} $$
(3)

where the subscripts “exp” and “cal” refer to the experimental and calculated values, respectively, and n is the number of data points.

Two adsorption models, Langmuir and Freundlich models, were used to simulate the adsorption process [25]. The linearized Langmuir isotherm is shown as Eq. (4):

$$ \frac{C_{\mathrm{e}}}{q_{\mathrm{e}}}=\frac{1}{q_{\mathrm{m}}{K}_L}+\frac{1}{q_{\mathrm{m}}{C}_e} $$
(4)

where KL is the Langmuir constant (L/mg).

The linearized Freundlich isotherm is shown as Eq. (5):

$$ \ln {q}_{\mathrm{e}}=\ln {K}_{\mathrm{F}}+\frac{1}{n_F\ln {C}_{\mathrm{e}}} $$
(5)

here the nF is used to indicate whether the adsorption is linear, or a typical chemical or a favorable physical process, and the KF is the Freundlich constant.

An equilibrium parameter RL (Eq. (6)) was used to interpret the adsorption process, whether it is “favorable” or not.

$$ {R}_{\mathrm{L}}=\frac{1}{1+{K}_{\mathrm{L}}{C}_0} $$
(6)

3 Results and discussion

3.1 Characteristics of activated carbons

The humins were solid residues cumulated by microspheres with a diameter of 2–10 nm (see Fig. 1a), with a very low BET surface area (6 m2/g) and negligible pore volume (see Table 1). Direct pyrolysis of humins at 400 °C only resulted in a carbon material with a still low BET surface area of 39 m2/g and pore volume (see Table 1), indicating low contents of pores were formed. As expected, the H3PO4 activation promoted the formation of pores, and a high BET surface area of 2375 m2/g was measured on the AC400. The H3PO4 would apply as an acid catalyst promoting bond cleavage and crosslinks formation, and form phosphate and polyphosphate bridges connecting and crosslinking polymer fragments [26, 27]. Because of the dilation process caused by the addition and/or insertion of phosphate groups, pore structures were formed in the matrix in an expanded state after the removal of acid. However, phosphate linkages is thermally unstable at temperature above 450 °C, which lead to form an activated carbon with more densely packed and less porous structure [26, 27]. In addition, H3PO4 is instable over high temperatures of 213 °C (http://www.ilo.org/dyn/icsc/showcard.display?p_version=2&p_card_id=1008, http://erplan.net/msds/7664382msds.htm). With the increase of temperature, H3PO4 would lose water to form pyrophosphoric acid (e.g., 200–300 °C) and further lose water to form triphosphoric acid and even partial phosphoric acid (e.g., > 300 °C) [28, 29]. The loss of water in the H3PO4 body with the increase of temperature might affect the acid catalytic activity. Therefore, the properties of activated carbons should be affected by both the H3PO4 activation effects and pyrolysis temperature. As shown in Fig. 1, the microspheres of activated carbons were destroyed on the AC400. Interestingly, microspheres were still on AC500, though a small destruction of these microspheres was observed. However, the destruction of the microspheres was observed again on the AC700. In addition, the AC500 had relative higher yields than AC400 and AC700, and had similar microspheres to that of humins, indicating both H3PO4 activation effects and direct pyrolysis were relatively weak at 500 °C. It seemed that the H3PO4 activation effects caused a most serious destruction of humins at 400 °C, while high-temperature pyrolysis resulted in microsphere destruction at 700 °C. Since H3PO4 activation effects were key factors for the formation of pores, a high pyrolysis temperature may not be favored due to the instability of H3PO4 and phosphate linkages. This is confirmed by BET surface area and total pore volume data as listed in Table 1. For example, the trend of surface area is AC400 (2375 m2/g) > AC500 (2005 m2/g) > AC600 (1774 m2/g) > AC700 (1673 m2/g). It is noted a very low pyrolysis temperature of 300 °C could not effectively promote the pore formation, as the AC300 has a relative low surface area of 1358 m2/g.

Fig. 1
figure 1

SEM images of humins (a), AC400 (b), AC500 (c), AC700 (d), and MB-adsorbed AC400 (e)

Full size image
Table 1 Properties of humins and activated carbons
Full size table

The N2 adsorption isotherms at 77 K for AC300–AC700 are shown in Fig. 2. All the isotherms belong to IUPAC type I isotherm, and a sharp curve of increased adsorption volume was observed at relatively low pressures (P/P0 < 0.2), indicating the adsorption was mainly contributed by small pores. The pore size distribution of AC400 is shown in Fig. 3, which confirms the major surface area and pore volumes are attributed to the pores with diameter ≤ 3 nm. Actually, the average diameter of AC400 was 1.97 nm.

Fig. 2
figure 2

Nitrogen adsorption isotherms for various activated carbons at 77 K

Full size image
Fig. 3
figure 3

Pore size (0.5–100 nm) distribution of AC400

Full size image

It is very common to use activated carbon in the presence of air. The thermal stability investigation of AC400–AC700 under air atmosphere is shown in Fig. 4. The TG inflection temperature and the highest DTA temperature peak of AC400, AC500, AC600, and AC700 were around 405 and 500 °C, 470 and 539 °C, 483 and 557 °C, 489 and 566 °C, respectively. Therefore, all the AC400–AC700 might be used below 400 °C, and a higher activation temperature benefits for the thermal stability. The humins and all the AC400–AC700 are amorphous carbon, based on broad diffraction peaks located between 2θ = 15° and 30° in the XRD spectra (see Fig. 5) [24]. The FT-IR spectra (Fig. 6) indicated the formation of aromatic groups in the pyrolysis process, since the AC400 has much stronger peaks around 1540 and 1590 cm−1 as compared with the humins. The AC400 also has a hydroxyl group (around 3440 cm−1) and C=O bond (around 1820 cm−1), which are important acidic groups in favor of adsorption of alkaline solutes (e.g., MB). Actually, the acidic groups of the AC300–AC700 were between 3.3 and 3.7 mmol/g as listed in Table 1.

Fig. 4
figure 4

Thermal stability of AC400–AC700 in air atmosphere

Full size image
Fig. 5
figure 5

XRD spectra of humins and AC400–AC700

Full size image
Fig. 6
figure 6

FT-IR spectra of humins, MB, AC400, MB-adsorbed AC400, and DCM-regenerated AC400

Full size image

3.2 Adsorption

3.2.1 Adsorption comparison and mechanism

To compare the adsorption capability, all the AC300–AC700 were used to adsorb MB with an initial concentration of 2000 mg/L for 24 h, and the AC400 exhibited the highest MB adsorption capability (1085 mg/g) as listed in Table 1. This highest adsorption capability of AC400 can be attributed to its highest surface area and total pore volumes (see Table 1). The BET surface area and total BJH pore volume with pore size 1.7–300 nm on AC400 decreased from 2375 m2/g and 0.88 cm3/g to 862 m2/g and 0.27 cm3/g, respectively, after the adsorption process. Since the pore volume of MB is negligible, the decrease of total BJH pore volume means that about 0.317 (=0.88−(1 + 1.085) × 0.27) cm3 of the initial pore volume with aperture1.7–300 nm per gram AC400 was filled by the MB. The three-dimension of MB is 1.41-nm width, 0.55-nm depth, and 0.16-nm thickness, respectively [30], and the molecular volume of MB is 0.297 nm3. Therefore, about 1.07 × 1021 MB molecules (or 566 mg) were adsorbed by the pores with 1.7–300 nm. It was reported an accessible pore size for dye adsorption should be at least 1.3–1.8 times greater than the dye molecule width [31]. Therefore, the adsorption of MB on micropores with aperture < 1.7 nm might be negligible in considering the width of MB (1.41 nm). The adsorption of the rest 47.8% (=\( \frac{1085-566}{1085}\times 100\%\Big) \)of MB (519 mg) per gram AC400 was mainly attributed to the macropores with aperture > 300 nm and the outer surface, in which the high acidic groups (3.6 meq g−1) of AC400 favored the adsorption of alkaline MB. The adsorption of MB on the outer surface was confirmed by the SEM spectra, as the surface of MB-adsorbed AC400 was much smoother as compared with that of AC400 (see Fig. 1). The adsorption of MB on the AC400 was also qualitatively reflected by the FT-IR analysis (see Fig. 6). Compared with the AC400, the MB-adsorbed AC400 had additional peaks at 1380 and 1320 cm−1, related to the methyl groups of MB.

3.2.2 Adsorption kinetics and isotherm

After an integrated consideration of the activated carbon yield, pyrolysis temperature, BET surface area, pore volume, and MB adsorption capacity, AC400 was selected as the preferred one and it was used for further adsorption kinetics and isotherm studies. Information of kinetics is very important for modeling the adsorption process, as well as for both economic and industrial applications. Since AC400 has a high adsorption capability, relative high initial MB concentrations, i.e., 1000, 1500, and 2000 mg/L, were used for the kinetics studies. The influence of contact time on MB adsorption is shown in Fig. 7. The saturation curves rise sharply at initial stages and the curves level off with contact time, and a plateau of all curves occurs when the MB adsorption reached equilibrium. Longer contact time for equilibrium was required for MB solutions with higher initial concentration. It seems that contact time for the equilibrium required approximately 3 h, 12 h, and 24 h for MB solutions with initial concentrations of 1000 mg/L, 1500 mg/L, and 2000 mg/L, respectively. The adsorption capacity at equilibrium (qe) increased from 592 to1085 mg/g when the initial MB concentrations increased from 1000 to 2000 mg/L.

Fig. 7
figure 7

Adsorption isotherms of MB with various concentrations

Full size image

Table 2 listed the data of pseudo-first-order and pseudo-second-order kinetic models. The pseudo-second-order model indicates the adsorption is a chemisorption process through sharing or exchanging of electron between the solvent and the sorbate. The pseudo-first-order model means the adsorption rate based on the adsorption capacity [32,33,34]. The correlation coefficient R2 and △q were 1.000 and 6–10 for pseudo-second-order kinetic model, and 0.7769–0.9572 and 84–93 for pseudo-first-order kinetic model, respectively. In addition, the qe,cal calculated by the pseudo-second-order model is in good agreement with qe,exp, while the qe,cal calculated from the pseudo-first-order kinetic model is significantly different from the qe,exp. All these R2, △q, and the difference between qe,cal and qe,exp confirmed that the MB adsorptions are better simulated by the pseudo-second-order kinetic model.

Table 2 The parameters for the pseudo-first-order and pseudo-second-order kinetic models
Full size table

Langmuir and Freundlich models are used to investigate the adsorption process, and the parameters are listed in Table 3. Generally, the Langmuir model describes a homogenous surface adsorption, and Freundlich isotherm describes heterogeneous adsorption systems considering multilayer adsorption and interactions between the molecules adsorbed [25, 35].

Table 3 Isotherm parameters of Langmuir and Freundlich models
Full size table

The nF in the Langmuir isotherm equation can be used to determine whether the adsorption is a typical chemical (nF < 1) or a favorable physical process (nF > 1) [32]. The nF is 0.212, indicating that adsorption process tends to be the former. The correlation coefficients R2 of Langmuir isotherm and Freundlich isotherm are 1.000 and 0.8696, respectively, indicating that the adsorption behavior complies with the Langmuir isotherm model better. The adsorption process is also interpreted by RL, as the adsorption isotherm can be classified by unfavorable (RL > 1), favorable (0 < RL < 1), linear (RL = 1), and irreversible (RL = 0). The RL is between 3.64 × 10−3 and 7.25 × 10−3 (see Table 3) when the initial MB concentration is 1000–2000 mg L−1, indicating the adsorption process is favorable. Based on the calculation by Langmuir isotherm, the maximum adsorption capacity of the MB was 1125 mg/g. This adsorption capacity is quite high as compared with the MB adsorption capacity reported by previous works [22, 23, 32, 34].

3.3 Desorption

A conventional method, extraction of MB by low-boiling point organic solvents (methanol, acetone, and DCM) in a Soxhlet extractor, was employed for a preliminary desorption study. After the first time extraction by methanol, acetone, and DCM, the adsorption capacity of regenerated AC400 was 883 mg/g, 805 mg/g, and 981 mg/g, respectively (see Fig. 8). It is therefore suggested that DCM was the best extraction solvent, and the regeneration resulted in decrease of adsorption capacity. The decrease of adsorption capacity after the regeneration can be explained by the FT-IR analysis (Fig. 6). The adsorbed AC400 had two obvious peaks at 1320 and 1380 cm−1, owing to the methyl groups of MB. The two peaks were still observed on the regenerated AC400, though the intensity of the two peaks was much weaker than that of MB-adsorbed AC400, indicating that a small part of MB remained on the regenerated AC400. Although the adsorption capacity of regenerated AC400 decreased with the increased reuse times, a relative high adsorption capacity of 680 mg/g was observed in reusing the AC400 for five times (see Fig. 8).

Fig. 8
figure 8

Adsorption capacity of fresh AC400 and regenerated AC400. The used AC400 was regenerated by solvent extraction with a Soxhlet extractor for 24 h. A, fresh AC400; B, regenerated AC400 by methanol extraction for the 1st time; C, regenerated AC400 by acetone extraction for the 1st time; D, regenerated AC400 by DCM extraction for the 1st time; E, regenerated AC400 by DCM extraction for the 2nd time; F, regenerated AC400 by DCM extraction for the 3rd time; G, regenerated AC400 by DCM extraction for the 4th time

Full size image

4 Conclusion

In this work, value-added application of biomass hydrolysis-derived humins for activated carbon production was achieved by H3PO4 activation method. The activated carbons owned amorphous carbon and aromatic structures, with the yield, total acidity, BET surface area, and BJH pore volume of 51.9–73.5 wt%, 3.3–3.7 meq g−1, 1358–2375 m2/g, and 0.25–0.88 cm3/g, respectively. Pyrolysis temperature is an important factor affecting the activated carbons’ properties, and 400 °C was found as an optimum temperature for the preparation of high-performance AC400, exhibiting highest BET surface area, BJH pore volume, and MB adsorption capacity. The adsorption of MB on AC400 complied with the pseudo-second-order kinetics, and an outstanding Langmuir adsorption capacity of 1125 mg/g was calculated. Moreover, the used AC400 could be simply regenerated by organic solvent Soxhlet extraction method, and a comparable adsorption capacity of 680 mg/g was maintained for fifth reusing. The good adsorption and desorption properties of the AC400 confirmed that the valorization of humins by phosphoric acid activation process is an effective and promising technology.