Design of biomass-based composite photocatalysts for wastewater treatment: a review over the past decade and future prospects

This investigation applied a systematic review approach on publications covering primary data during 2012–2022 with a focus on photocatalytic degradation of pollutants in aqueous solution by composite materials synthesized with biomass and, at least, TiO2 and/or ZnO semiconductors to form biomass-based composite photocatalysts (BCPs). After applying a set of eligibility criteria, 107 studies including 832 observations/entries were analyzed. The average removal efficiency and degradation kinetic rate reported for all model pollutants and BCPs were 77.5 ± 21.5% and 0.064 ± 0.174 min−1, respectively. Principal component analysis (PCA) was applied to analyze BCPs synthesis methods, experimental conditions, and BCPs’ characteristics correlated with the removal efficiency and photodegradation kinetics. The relevance of adsorption processes on the pollutants’ removal efficiency was highlighted by PCA applied to all categories of pollutants (PCA_pol). The PCA applied to textile dyes (PCA_dyes) and pharmaceutical compounds (PCA_pharma) also indicate the influence of variables related to the composite synthesis (i.e., thermal treatment and time spent on BCPs synthesis) and photocatalytic experimental parameters (catalyst concentration, pollutant concentration, and irradiation time) on the degradation kinetic accomplished by BCPs. Furthermore, the multivariate analysis (PCA_pol) revealed that the specific surface area and the narrow band gap are key characteristics for BCPs to serve as a competitive photocatalyst. The effect of scavengers on pollutants’ degradation and the recyclability of BCPs are also discussed, as necessary aspects for scalability trends. Further investigations are recommended to compare the performance of BCPs and commercial catalysts, as well as to assess the costs to treat real wastewater.


Introduction
Water resources all over the world are currently threatened by the advancement of technology and industrial growth (Bhattacharya 2017). The major human-related sources of water contamination are agriculture (zootechnical farms and agrochemicals), industries (compliant landfills, raw material landfills and accidental discharge), and human agglomeration (wastewater and compliant landfills), among others (Pal et al. 2010;Mokarram et al. 2020;Ulrich et al. 2021;Dunea 2022). Apart from these compounds, there are several inorganic substances (fluoride, arsenic, lead, chromium, mercury) that can get into the water system from natural sources, industrial processes, as well as from plumbing systems (US EPA 2006). To mitigate water contamination, water/wastewater treatment techniques currently applied worldwide are sedimentation or settling, biological treatment, boiling/ distillation, chemical treatment (precipitation/coagulation/ adsorbents), disinfection, and filtration (Chan et al. 2009;Margot et al. 2013;Luján-Facudo et al, 2019). However, conventional wastewater treatment techniques are inefficient for the removal of organic and inorganic contaminants due to mainly the physicochemical properties of a large variety of substances (Cunha et al. 2019b;Gesels et al. 2021). Additionally, the transformation of contaminants into intermediary chemical compounds can make the treated wastewater even more hazardous (Alfred et al. 2020).
Innovative treatment alternatives such as advanced oxidation processes (AOPs) rely on the generation of strong nonselective oxidation species capable of degrading and/ or mineralizing a large range of contaminants in CO 2 , H 2 O, and inorganics, with significant advantages, when used as a complementary treatment over the conventional methods (Omorogie and Ofomaja 2018;Iervolino et al. 2020).
Since the discovery of the photoelectrochemical water splitting reaction by Fujishima and Honda (1972), photocatalysis using semiconductors as an AOP has attracted increasing attention (Sillanpää et al. 2018;Jabbar and Esmail Ebrahim 2022). Heterogeneous photocatalysis (HP) applied to water/wastewater treatment has been reported for decades (Malato et al. 2009). The number of publications on HP has increased considerably since 2010, reaching 4,719 publications in January 2022 (Fig. 1).
ZnO and TiO 2 are two metal-oxides with similar properties, with a band gap around 3.2 eV, and light absorption properties in the near ultraviolet region (Ishchenko et al. 2016). In addition, they have similar conduction and valence bands position. However, some important drawbacks are their quick recombination of electron-hole pairs (e − /h + ) and their very low performance in the visible range of the solar spectrum irradiation (Abarna et al. 2019;Djellabi et al. 2019;Fazal et al. 2020). In practice, photocatalytic reactions are performed under UV light as small amounts of photons can be absorbed in the visible range, which increases the overall costs of the process (Colmenares et al. 2013;Peñas-Garzón et al. 2019;Alfred et al. 2020).
Therefore, the development of new photocatalysts comes into sight to be a necessary step to push forward the current photocatalytic performance (Nemiwal et al. 2021). An increasing number of investigations are currently conducted to find out new materials and/or improve the properties of the current ones to get efficient visible-light-driven photocatalysts, including doping with different metals (Pd 2+ , Cu 2+ , Cd 2+ , Cr 3+ , Fe 3+ ) and non-metals, such as activated carbon, graphene, biomass carbonaceous materials, dye sensitization, and development of heterostructure semiconductors (Malato et al. 2009;Devi and Kavitha 2013;Park et al. 2013;Schneider et al. 2014;Zhang et al. 2017b;Cunha et al. 2019a). In the case of carbonaceous materials-dopedphotocatalysts such as TiO 2 and ZnO, numerous investigations have focused on increasing surface area and hence the active sites on the surface of photocatalysts (El-Salamony et al. 2017;El Mouchtari et al. 2020;He et al. 2021;Loo et al. 2021). Biomass-based materials may also narrow the bandgap of photocatalytic nanomaterials due to the formation of a localized electron trapping site below the conduction band of photocatalyst that prevents recombination and the effective charge transport by the metal-oxygen-carbon linkage between biomass and semiconductor photocatalysts (Colmenares et al. 2016;Lisowski et al. 2017;Liu et al. 2018). Several studies have reported cheap, chemically stable, and easily recoverable biomass-based composite photocatalysts (BCPs) supported by multiple waste biomasses feedstocks employed for the removal of different pollutants (Zhang et al. 2014;Luo et al. 2015;El-Zawahry et al. 2016;Lisowski et al. 2017;Chen et al. 2019;Lu et al. 2019;Peng et al. 2019;Silvestri et al. 2019;El Mouchtari et al. 2020;Leichtweis et al. 2020).
Recent literature reviews (basically in narrative format) on BCPs have focused on their synthesis, characterization, photocatalytic mechanisms, and applications in wastewater treatment (Tan et al. 2016;Zhao et al. 2019;Huang et al. 2019;Cui et al. 2020;Minh et al. 2020;Ahmaruzzaman 2021;Ambaye et al. 2021;Qiu et al. 2021;Fito et al. 2022;Rangarajan et al. 2022;Sutar et al. 2022). Traditional narrative reviews are mostly useful for stablishing the state of the art, obtaining a broad perspective about the topic. However, they cannot quantitatively synthesize and interpret large datasets and the results across studies. In contrast, systematic reviews are aimed to answer specific research questions using an explicit methodology to identify, select (according to eligibility criteria), aggregate, and critically evaluate the results of the selected publications (Wright et al. 2007). In addition, multivariate statistical techniques, such as principal component analysis (PCA), can be applied to analyze datasets in which observations are described by several inter-correlated quantitative dependent variables, extracting important information from the dataset, to represent it as a set of new orthogonal variables called principal components (PCs) (Abdi and Williams 2010). Furthermore, PCA is a versatile technique that can be applied to almost all data to look for overall trends, for initial data inspection and to explore a specific effect of interest (Buvé et al. 2022).
To the best of the authors' knowledge, no attempts have been reported so far concerning the synthesis, characterization, and applications of BCPs in wastewater treatment from a systematic review perspective including meta-analysis. Therefore, in the present investigation, a systematic review was carried out, covering references from January 2012 to January 2022 to analyze the BCPs synergistic adsorption and degradation/reduction efficiency, and photodegradation kinetics of distinct classes of pollutants. Publications were selected according to the eligibility criteria previously established. The database compiled includes biomass-based composite photocatalysts (BCPs) formed by, at least, the photocatalysts TiO 2 and/or ZnO and different biomasses as feedstocks, with a focus on synthesis methods, photocatalytic experimental conditions, and model pollutants. The multivariate analysis by principal component analysis (PCA) was applied to the compiled data as a first step to obtain an insight into the datasets and to extract information out of many response variables (i.e., pollutants adsorption and degradation/ reduction efficiency, and photodegradation kinetics). Additionally, the intrinsic structure and optoelectronic properties of BCPs are critically discussed. The BCPs' performance is addressed in terms of the reactive species produced during photodegradation, and the composites stability and recyclability, which is a necessary step to move from the bench/experimental scale to full-scale applications.

Sources, search strategy, and study selection
The survey covered peer-reviewed scientific publications within the period from August/2021 to January/2022 in the Journal Citation Reports (JCR) indexed journals. According to the first requirement, each article had to report original, quantitative data on pollutants degradation/ reduction and/or photodegradation kinetics from an aqueous matrix using a binary composite material consisting of, at the least, a biomass-based material, and the photocatalysts TiO 2 and/or ZnO. Studies on BCPs composed by graphitic carbon nitride (g-C 3 N 4 ), graphene, commercial activated carbon, or an unidentified mixture of biomasses were excluded.
It was decided to analyze BCPs performance in terms of adsorption (ADS), photodegradation/reduction (RE) and/or degradation kinetics expressed by the rate constant (Kphdeg) of organic and inorganic pollutants. In some cases, it was necessary to infer the removal performance of catalysts based on graphical information. Studies not providing results as photodegradation/reduction or degradation kinetics or not reporting enough details to perform this interpretation were excluded from the analysis.
A better overview was gained through a flowchart adapted from the PRISMA diagram (Page et al. 2021) (Fig. 2). The diagram shows the number of publications identified, included and excluded. Potentially relevant studies (1,440) were identified, and the complete selection process (see below) yielded 107 articles and 832 experimental observations or experimental runs.

Data abstraction and descriptive analysis
From the 107 eligible papers (according to the eligibility criteria), information/data was collected and organized in a database. The workflow used is found in Online Resource 1 (Fig. S1). For each selected study, we gathered a range of metadata related to the details of every biomass-based composite photocatalysts, in terms of composition and synthesis (biomass feedstock and pre-treatment, semiconductor and synthesis methodology applied). Additionally, the main experimental variables and their respective levels applied were identified and included (i.e., pollutant and catalyst concentration, reaction time, pH, temperature and irradiation source). Then, the characterization of the novel composites was further collected.
Finally, the data related to BCPs efficiency including adsorption (%), degradation/reduction (%) and degradation kinetics (min −1 ) were extracted from each individual observation. Based on applicability and number of data entries, 16 out of 43 variables were selected for statistical analysis (variables identified by an asterisk (*) in Fig. S1. Other variables were addressed on a descriptive basis.  (Page et al. 2021) In addition, Table S1 (in Online Resource 1) includes the model pollutant categories (dyes, pharmaceutical compounds, phenolic compounds, metals, and others), Chemical Abstracts Service (CAS) number, molecular formula and molecular weight (Mw) of each substance investigated.
When available, the following information was also registered: reusability and stability of BCPs; efficiency of pollutant removal from "real water" (wastewater, surface water or synthetic water/wastewater, i.e., deionized water containing multiple pollutants); commercial catalyst comparison and costs assessment.
The complete database and references are found in Online Resource 2.

Principal component analysis (PCA)
Before conducting the analyses, the dataset was checked to meet the Kaiser-Meyer-Olkin (KMO) criteria measurement for sample adequacy (MSA), by which the minimum value is 0.5 for the validity of PCA application. KMO was conducted following the procedure previously demonstrated  using the REdaS package in the R software (Team 2013).
PCA was applied to investigate associations among the synthesis methodology applied, experimental conditions, and BCPs' characterization related to the variables of interest: removal by adsorption (%), synergistic removal by adsorption and photocatalysis (%) and degradation kinetic rate (min −1 ) of pollutants performed by BCPs.
Prior to modeling, data were scaled using Z transformation, which consists in transforming each variable subtracting the mean value of the parameter and dividing by its standard deviation (Meloun and Militký 2011). The PCA results were computed using FactoMineR and Factoshiny, the R packages for multivariate analysis (Lê et al 2008) and visualized using statistical tools for high-throughput data analysis.
Confidence ellipses were defined as regions containing 95% of all samples that can be drawn from the underlying Gaussian distribution, and they were used to visualize whether categorical groups were significantly different from each other.
The principal components (PCs) with eigenvalues > 1.0 were accounted for. Eigenvalues indicate the significance of the PC, e.g., PC with higher eigenvalues is considered to be more significant (Gulgundi and Shetty 2018). The eigenvalues were extracted from the covariance matrix of the original variables (Chabukdhara and Nema 2012). The first PC accounted for the most significant variance in the dataset followed by the second PC and so on (Liu and Zhang 2009).
Additionally, Pearson's correlation coefficient was calculated to estimate the correlation between the variables studied (Schober and Schwarte 2018).

Descriptive analysis and categories definition
Following our selection criteria, a total of 832 observations were collected from 107 different selected publications (Online Resource 2). More than one entry was added to the database for 94 out of 107 publications due to more than one experimental observation (e.g., one BCP tested for more than one pollutant or different BCPs tested for the same pollutant) (Online Resource 2).
Biomass feedstocks were classified into three categories (Fig. 3a), according to the source and general characteristics of the organic material, as follows: (i) agricultural and forestry residues (n = 688 observations); (ii) industrial by-products and solid waste (n = 68 observations); and (iii) non-conventional materials (n = 76 observations) (Huang et al. 2019).
Agricultural and forest waste residue were the most commonly used biomass, representing 83% of the observations, probably because of the high content of lignin and cellulose, providing powerful security to produce low-cost adsorbent supports (Huang et al. 2019). This type of biomass support includes a large variety of agricultural wastes, such as fruit peels, rests of agriculture crops, seeds, stems, shells from nuts, and fibers, among others El-Salamony et al. 2017;Lara-López et al. 2017;Abarna et al. 2019;Wang et al. 2019;Fazal et al. 2020). Agricultural wastes represent a sustainable option for wastewater treatment due to their chemical properties, abundant functional groups, easy accessibility, and low cost (Sutar et al. 2022).
Some studies used biomass materials less investigated, including solid waste and non-conventional materials (8% and 9% of the total observations, respectively). The nonconventional materials derived from biomass found are lignin, glucose, and Starbon800® purchased from laboratories and local industries (Colmenares et al. 2013;Zhou et al. 2021). Lignin is produced from natural sources of lignocellulosic biomass, while Starbon800® materials are derived from different polysaccharides, presenting interesting properties such as mechanical and thermal stability (200-1000 °C), high mesoporous (Vmeso > 0.3 cm 3 g −1 ), and high specific surface area (> 500 m 2 g −1 ) (Colmenares et al. 2013;Colmenares et al. 2016;Khan et al. 2018).
Sewage sludge is a solid waste material suitable for the production of BCPs with increasing availability due to the world population growth and increasing amounts generated at wastewater treatment plants. However, the material suffers from low surface adsorptive functional groups and high metal concentrations (Mian and Liu 2019). Hence, blending biomasses can help to improve the surface properties of the BCPs and to produce better photocatalytic materials.
The synthesis of BCPs with desirable characteristics from extensively available biomass is a fundamental requisite to build up photocatalysts to treat large-scale wastewater systems (Sutar et al. 2022). For the preparation of BCPs, synthesis in one-step and multi-steps was applied. (i) in onestep synthesis, the raw biomass is added for the synthesis of the semiconductor (TiO 2 or/and ZnO) and only one thermal treatment is applied to synthesize the composite (n = 257 observations); and (ii) in multi-step synthesis, the biomass receives thermal and/or chemical treatment before the synthesis with the semiconductors (TiO 2 or ZnO); in this case, two or more thermal treatment steps are applied to synthesize the composite (n = 575 observations) (Fig. 3b). In the multi-step synthesis, the thermal treatment of biomass can occur by carbonization to produce carbonaceous biomass called char or biochar (Yaashikaa et al. 2020). This carbonization involves techniques such as pyrolysis, calcination, hydrothermal, and quaternization (Fig. 3c). Pyrolysis of biomass represented 24% of the studies and is differentiated by the reaction time of the treatment (slow, rapid, and "flash") and the heating method (burning of fuels, electrical heating, or microwaves) with the absence of oxygen (Zhang and Lu 2018;Lu et al. 2019;Fazal et al. 2020). Instead, calcination is the thermal decomposition of biomass occurring in the presence of oxygen (n = 169 observations). Hydrothermal pretreatment can be also referred to as hydrothermal carbonization (HTC) of biomass and is a thermochemical process performed by submerging biomass in water and then sealed in a confined system to be heated for several hours under saturated pressure (Donar et al. 2018). One of the advantages of HTC is the greater energy efficiency, due to higher solid yields without the need for intensive drying compared to dry pyrolysis . Flash carbonization was applied in only one study and consists of a process at medium to high temperatures (~ 600 °C) during short periods (between 5 and 10 min) in the presence of air (Kim and Kan 2016).

Fig. 3 (a)
Percentage of data/entries per category of biomass feedstock included in the review (832 observations); (b) percentage of data/entries per method of BCPs synthesis included in this review (832 observations); (c) types of thermal treatment applied on pretreatment of biomass feedstock (107 scientific papers). Note that "not applied" refers to raw biomasses before synthesis with the semicon-ductors (i.e., one-step synthesis); (d) types of chemical treatment applied on pretreatment of biomass feedstock (107 scientific papers and 9 different chemicals); (e) number of data/entries per type of synthesis method applied to produce BCPs (832 observations and 9 methods found in 107 scientific papers); (f) amount of data per model pollutant found in 107 scientific papers whereas the chemical and thermal treatment are more versatile, because they allow the synthesis of biomass with an adjustable porosity made up of micropores and mesopores (Lisowski et al. 2018a).
It is important to highlight that among 575 studies applying multi-step synthesis to produce BCPs, 13.7% did not report the technique applied for the biomass pre-treatment.
For the synthesis of BCPs, the following variables were extracted: the photocatalyst (e.g. TiO 2 and/or ZnO); the biomass feedstock; and the BCP synthesis method (i.e., hydrothermal, impregnation, infiltration, sol-gel, one-pot in situ, US-assisted or thermal decomposition) (Fig. 3e). Mechanical mixing was collectively classified as an impregnation synthesis method. The synthesis of BCPs has been extensively explored through several reviews (Cui et al. 2020;Iervolino et al. 2020;Minh et al. 2020;Ahmaruzzaman 2021). Details on various preparation methods, including highlights on the most influential advancements can be found in these papers. For this reason, information regarding the BCP synthesis method was addressed on PCA with consideration of these parameters on BCP efficiency.
The evaluation of BCPs' efficiency was assessed through adsorption (ADS), synergistic adsorption and photodegradation/reduction (RE) and degradation kinetics (Kphdeg) of model pollutants in water. Although dark adsorption measurement is always required as a control, only 39% of the total observations presented such information.
In general, the pseudo-first-order kinetics model fits well with the rate of degradation of organic contaminants using BCPs for water and wastewater treatment (Wu et al. 2015;de Moraes et al. 2019;Guan et al. 2020). The photocatalytic reduction of metals such as Cr(VI) was modeled using the Langmuir-Hinshelwood mechanism (Yang et al. 2014;Matsena and Chirwa 2021). Furthermore, the degradation of humic acid (HA) by BCPs was also in agreement with the apparent first-order kinetics rate constant, and HA degradation was achieved through a synergistic mechanism of adsorption and photocatalysis ).

Categories of pollutants
In total, 47 organic model pollutants were grouped into five categories: (i) dyes; (ii) pharmaceutical compounds; (iii) phenolic compounds; (iv) pesticides; and (v) humic acid (Fig. 3f). Besides, the inorganic compounds were individually identified as (i) metal Cr (VI); and (ii) ammonia nitrogen. Dyes are the most studied model pollutant (47.8% of 832 observations) and consist of organic compounds with large molecular weights that contain extended systems of conjugated double bonds with pi-electron systems and can be excited by visible light (González-García 2018). Pharmaceutical compounds represent 20.8% of the observations and this category includes antibiotics, analgesics, antiepileptics, and gadolinium-based ionic compounds. Organic compounds such as phenol and phenolic compounds were classified as phenolic compounds (6.4% of the total observations) and humic acid (HA) had 15 observations (1.8% of the total observations). Inorganic pollutants include the groups metal Cr (VI) (n = 21) and ammonia nitrogen (NH 3 -N) (n = 54 meaning 6.5% of the total observations). Methylene blue dye was the most evaluated model pollutant, representing 20% of total observation (n = 166), followed by NH 3 -N (n = 54 or 6.5%), and orange G dye (n = 50 or 6.0%).

Principal component analysis (PCA)
The PCA was employed to assess the overall trend relationship between BCP synthesis method, experimental conditions, and catalyst characteristics. The visualization of scores and loadings through PCA enabled to explore the relationship among variables and identify parameters driving the removal performance of BCPs. Three PCAs were performed using the following selected criteria: a) PCA_pol: focus on the BCPs adsorption (ADS) and synergistic adsorption and photodegradation/ reduction (RE) for all categories of pollutants, in terms of the BCP synthesis methods, photocatalysis experimental parameters, and characterization; b) PCA_dyes: focus on BCP degradation kinetics (Kphdeg) of textile dye pollutant, synthesis methods, and experimental parameters. BCP characterization was not included in the PCA due to the lack of information in the selected studies; c) PCA_pharma: focus on BCP degradation kinetics (Kphdeg) of pharmaceutical compounds, synthesis methods, and experimental parameters. Due to the same reason given above for PCA_dyes, BPC characterization was not included.
Pearson's correlation was applied to the datasets used on each PCA (i.e., PCA_pol, PCA_dyes and PCA_pharma) to find the linearity of two continuous variables, such as the relationship between the BCPs removal and photodegradation rate with any of the parameters, such as pollutant concentration, irradiation time, and others. Pearson's correlation between two variables ranges from -1 to + 1, where − 1 indicates the strongest negative correlation and + 1 indicates the strongest positive correlation. The terms "very strong," "strong," "moderate," "weak," and "very weak" as applied to variables, refer to absolute loading values (> 0.80, PCA_pol PCA_pol ( Fig. 4) was applied to fifteen variables (for quantitative and qualitative variables, see Table S2b) to identify the factors involved in the adsorption (ADS) and photodegradation/reduction (RE) of all classes of pollutants by BCPs. It is worth mentioning that PCA does not accept missing data and the use of estimation methods was not an option, since each observation came from an individual study. For this reason, PCA_pol analysis with a focus on BCP ADS and RE was limited to part of the database (n = 159), which did not have missing data. The selected quantitative variables for analysis were as follows: BCP synthesis calcination temperature (BCP_temp); pollutant concentration (POL_conc); catalyst concentration (BCP_conc); reaction time (R_time); BCP specific surface area (BET); BCP bandgap (Bandgap); pollutant adsorption (ADS); and pollutant adsorption and photodegradation/reduction (RE). The categorical variables were biomass classification (Bclass); semiconductor (Sem); BCP synthesis by one-step or multi-step (BCP_S); BCP synthesis method (BCP_M); pollutant class (POL_class); and experimental light source (Light). To evaluate the suitability of the principal component analysis, Kaiser-Meyer-Olkin (KMO) was applied, indicating that PCA could result in a significant dimensionality reduction (KMO factor adequacy = 0.59; Table S2). In this analysis, 65% of the variation was summarized in three components: PC1, PC2, and PC3. Eigenvalues, percentages of variance, and cumulative percentages of each PC are shown in Supplementary Material (Table S2a).
In Fig. 4, it is possible to see the vectors that contributed with the highest weight for each component. The variables POL_conc, BET, ADS, BCP_conc, BCP_temp, RE, R_time, and Bandgap participated with 34.4% for PC1, while RE, ADS, Bandgap, BET, POL_conc, and R_time with 16.7% for PC2. In Fig. S2, R_time, RE, ADS, Bandgap, and BCP_ temp favored PC3 by 13.9% (Table S2a). High correlations are observed between ADS, POL_conc, and BET in PC1 (r = 0.64, 0.75 and 0.77, respectively); and between RE and ADS in PC2 (r = 0.80 and 0.53, respectively) ( Fig. 4; Table S2b). Some vectors showed close collaboration in more than one component (ADS: PC1 and PC2; BCP_temp: PC1 and PC3; R_time: PC2 and PC3), which indicates the strong influence of these parameters on pollutants removal by BCPs. In PC1 (Fig. 4), ADS (mean = 32.3 ± 24.2%) was highly correlated with the BCP characterization BET, which suggests that the adsorption of pollutants by BCPs may vary according to the BCP specific surface area. From the results of BET (mean = 167.8 ± 127.6 m 2 g −1 ), it is possible to reinforce the conclusion that an increase in the surface area of the composites results in better adsorption capacity (Luo et al. 2015). Also, the initial concentration of pollutants can play a role in the adsorption performance of BCPs up to a certain value, when higher concentrations can increase the adsorption on the surface of BCPs (Zhang and Lu 2018). However, increasing the concentration of the pollutant can result in an increased amount of mineralized inorganic by-products in the solution, which compete with the contaminant molecules for the active sites on the BCP surface (Alfred et al. 2020). In the same way, BCP_temp (r = 0.51) has a moderate positive correlation in PC1, reinforcing that an increase in the pyrolysis temperature may favor the adsorption capacity of the BCPs by increasing their pore size properties (Gonçalves et al. 2020). Liu et al. (2018) also reported an increase in specific surface area and pore volume of BCPs by raising the temperature during synthesis, but expanding the time of synthesis promoted the aggregation of TiO 2 nanocrystals into small particles, thus causing the reduction of specific surface area. With less relevance, the RE (r = 0.25) of the pollutants also participates in PC1 and presented an inverse correlation with Bandgap. Narrowing the bandgap energy of BCPs can grant the absorption of visible light and thus promote the removal of pollutants through the degradation performed by the photogenerated electrons in the conduction band of the semiconductors (Djellabi et al. 2019).
On PC2, it is also possible to note that RE was mainly associated with the ADS of pollutants, a variable that favored PC1 (Fig. 4). The interaction between adsorption and removal on biomass-based photocatalysts is understandable, since cooperative effects between support and active components must enhance removal efficiencies (Minh et al. 2020;Chakhtouna et al. 2021). Besides, the high surface area and highly oxidative reactive species can enhance the adsorption capacity during the photodegradation process (Ambaye et al. 2021). The Pearson's correlation coefficients indicate a moderate correlation between RE and ADS (r = 0.47), ADS and BET (r = 0.40) and BET and POL_conc (r = 0.52); an inversely moderate correlation between Bandgap and POL_conc (r = − 0.52), and the BET and Bandgap (− 0.52) (Table S3), which emphasize the results reported on PCA_pol.
Among the categorical variables, biomass classification (Bclass) is the best variable to illustrate the distance between individuals on PC1 and PC2, according to Wilks's test p-value (p < 0.005; Table S2b). Based on 95% confidence intervals (visualized as shaded ellipses in Fig. 4), agricultural and forestry residues (AF) and non-conventional materials (NC) were significantly different from each other (industrial by-products and solid waste "IM" and AF were not significantly different). It is possible to conclude that the non-conventional materials are correlated with the BET and adsorption of the catalysts and can be good candidates for the synthesis of novel and unexplored biomass-based composites photocatalysts.
The results obtained in PCA_pol allow us to conclude that the adsorption of pollutants by the BCPs, as well as the surface area of the material, has a fundamental role in the performance of the composite for the removal of contaminants from water. In this case, the possible mechanism for photodegradation of pollutants by BCPs relies on the migration and diffusion of contaminant molecules from the liquid solution to the surface of the BCPs, followed by the adsorption on the surface of the composites photocatalysts and degradation on the site of the photocatalyst (i.e., TiO 2 or ZnO), enhancing photocatalytic degradation activity (Chakhtouna et al. 2021). However, the temperature for synthesis calcination of BCP and the experimental parameters of BCP concentration and the pollutant concentration influences the removal mechanism and must be considered for the synthesis of the composite, as well as for the optimization of experimental parameters.

PCA_dyes
In PCA_dyes (Fig. 5 and Fig. S3), the missing data criteria were also applied and the analysis with a focus on BCP degradation kinetics (Kphdeg) of textile dyes model pollutant was limited to 75 entries in the database. The values of KMO in PCA_dyes were equal to 0.55, which is ≥ 0.5, confirming the validity of applying PCA (Table S4).
PC2 was mainly correlated with R_time (r = 0.78) and inversely correlated with Kphdeg (r = − 0.69) and BCP_ temp (r = − 0.62). The strong correlation between Kphdeg and BCP_temp points out that the photodegradation is greatly affected by the increase in calcination temperature and can be underpinned by the better crystallization of the semiconductor and the oxidation of residual organics in the biomass-composites photocatalysts after calcination (Khraisheh et al. 2013;Liu et al. 2018). Also rising the calcination temperature during BCPs synthesis can boost the pore size of the composite, which could favor the adsorption process and the photodegradation of dyes (Gonçalves et al. 2020). Some assumptions regarding the inverse correlation between degradation kinetic and reaction time can be done with caution.
In general terms, there are two possible explanations for the way kinetic occur: (i) the evaluation of degradation kinetic strictly after the contact of the catalyst and the pollutant. In this case, the initial reaction is driven by the capacity of the composites to adsorb and degrade the pollutant. After some time, the active sites of the material become saturated, and the reaction tends to slow down, being mainly controlled by the degradation of the pollutant, and (ii) the occurrence of a two-step mechanism if the model pollutant is converted to an intermediate and then mineralized to carbon dioxide (Choo 2018). On this assumption, the photocatalytic reactions can be written as follows: where P is the pollutant (initial substrate); I is the intermediate; K´ is the first-step adsorption coefficient; k´ is the first-step rate constant; K´´ is the second-step adsorption coefficient; and k´´ is the second-step rate constant. Hence, The PC3 was governed by strong correlation of BCP_min (r = 0.65) and moderate correlations of Kphdeg (r = 0.43), POL_conc (r = 0.40), and BCP_temp (r = − 0.44) (Fig. S3 and Table S4b).
Some variables participate in more than one component (POL_conc: PC1 and PC3; BCP_temp: PC1 and PC2; Kphdeg and BCP_min: PC2 and PC3), suggesting the significant contribution of these parameters on the degradation kinetics of dyes by BCPs. These results are in line with PCA_pol and demonstrate the relevance of synthesis conditions and experimental parameters on the photodegradation kinetics of dyes. Regarding Pearson's correlation, a very strong positive correlation was observed between BCP_conc and POL_conc (0.82), a negative strong correlation between Kphdeg and R_time (− 0.61), a moderate positive correlation between POL_conc and POL_Mw (r = 0.41) and an negative moderate correlation between BCP_min and BCP_temp (r = − 0.52) (Table S5). In general terms, catalyst and pollutant concentration are two of the most important parameters affecting photocatalytic efficiency ). In theory, higher concentrations of BCP could enhance photocatalysis, as a result of more generation of • OH (Le et al 2021). However, an excess of BCP can turn it difficult to diffract UV rays in the solution, thus decreasing the degradation efficiency (Lazarotto et al. 2020). Therefore, optimization of the BCPs concentration can lead to an increase in the degradation efficiency of the process ).
On the categorical variable, Kphdeg accounted for 26.8% of the PC2 variability alone and BCP_M suggests the positive relationship of synthesis methods hydrothermal and impregnation for the BCPs tested (Table S4b). Based on 95% confidence intervals (visualized as shaded ellipses in Fig. 5), hydrothermal and impregnation synthesis techniques were better related to BCP photodegradation rate (Kphdeg) and significantly different from the sol-gel method. Impregnation and hydrothermal synthesis of BCPs were not significantly different. In sum, the hydrothermal method, which is typically carried out in steel vessels (autoclaves) with temperatures around 200 °C, can be highly effective for incorporating biomass and the crystalline structure of TiO 2 or ZnO (Upneja et al. 2016). This method was applied in 17% of the studies and has drawn interest among the scientific community due to the high photocatalytic activity achieved by the controlled synthesis of hollow particles (Upneja et al. 2016;Li et al. 2018;Peñas-Garzón et al. 2019). Also, the impregnation method can be attractive because of its technical simplicity, low cost, and the limited amount of waste generated (Silvestri et al. 2019).

PCA_pharma
In PCA_pharma (Fig. 6 and Fig. S4), 73 individual observations were selected based on the missing data criteria. The BCP degradation kinetics (Kphdeg) for pharmaceutical compounds was evaluated and the KMO values were equal to 0.55 (≥ 0.5) ( Table S6). The first two dimensions (PC1 and PC2) explained 60.4% of the total variance. The PC3 was not considered in the analysis since it presented an eigenvalue of 0.94, which is lower than 1.0 (Table S6a).
Kphdeg was positively correlated with PC1 (40% of the total variance), loading with BCP synthesis temperature (BCP_temp) and BCP catalyst concentration (BCP_conc) an additional contribution of 29.3% and 21.3% to the variance, respectively ( Fig. 6 and Table S6b). The very strong and strong correlations between BCP_temp, BCP_conc and Kphdeg (r = 0.84, 0.71 and 0.68, respectively; p < 0.0001; Table S6b) corroborate with previously observed evidence that photodegradation kinetics increase on BCPs calcined under high temperatures and can lead to the degradation of pharmaceutical compounds Gonçalves et al. 2020). In addition, the results point out that increasing the catalyst concentration resulted in the enhanced removal of pharmaceutical compounds which is corroborated by several studies (Khraisheh et al. 2014;Le et al. 2021;Zhou et al. 2021). The inverse correlation between Kphdeg and R_time observed in PC1 (r = 0.68 and − 0.66, respectively) reinforces the observations on PCA_dyes, which the adsorption can play a role at the beginning of the reactions, influencing the performance of BCPs degradation kinetics.
The PC2 accounted for 20.4% of the total variance on PCA_pharma and it was densely weighted by BCP_min and POL_conc, contributing with 45.8% and 35.7% to the total variance of the second component, respectively. The contribution of these variables to PC2 can be considered as a result of the synthesis processes and the dependence of the experimental conditions on the photodegradation rate of the pharmaceutical compounds. As for the Pearson's correlation coefficient, it was possible to observe a strong correlation between BCP_conc and BCP_temp (r = 0.63) and a negative moderate correlation between Kphdeg and R_time (r = − 0.52) and between BCP_min and BCP_temp (r = − 0.42) ( Table S7).
The best qualitative variable to illustrate the distance between individuals on PCA_pharma (plane expressed by PC1 and PC2) is light source irradiation (Light). Based on 95% confidence intervals (visualized as shaded ellipses in Fig. 6), UV light irradiation (UV) and solar irradiation (SUN) were better related to Kphdeg and significantly different from sun-simulation (SUN-S) and Visible light irradiation (VIS). SUN-S and VIS were not significantly different. Based on this observation, the BCP degradation kinetics exhibited superior efficiency under UV than under VIS and SUN-S, which is related to the band-gap energy of the semiconductors TiO 2 and ZnO, which need higher energy photons to be excited (Bhanvase et al. 2017). However, a part of the energy could be used from the solar irradiation (SUN) to motivate the catalysts generating radical species and degrading the pollutants (Peng et al. 2019). Still, for BCPs catalysts, the production of radical species seems to be better under UV light, resulting in higher removal efficiency of pollutants, when compared to solar simulated light.
The PCA_dyes and PCA_pharma indicated that the degradation kinetics of textile dyes and pharmaceutical compounds were influenced by intrinsic interactions between the synthesis method applied to produce the composite and the experimental conditions. However, other factors which could reasonably impact the photocatalytic activity and mechanism were not included as parameters, such as (i) the intrinsic characteristics of photocatalysts; (ii) the effect of pollutant categories and their specificities; and (iii) the effect of some system parameters (e.g., different wavelength).

BCPs characterization
Adequate and powerful techniques for material characterization are crucial to unraveling the intrinsic structure and optoelectronic properties of photocatalysts (Qiu et al. 2021). Besides, the use of multiple characterization techniques, such as X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy (SEM), and transmission electron microscopic (TEM) help to confirm the stability of BCPs. Table S8 presents the results of some characterization techniques applied to different BCPs used in the selected studies along with some descriptive statistics, depending on the number of observations. The study of BCPs morphology reveals the presence and distribution of photocatalysts (e.g.: TiO 2 or ZnO) on the biomass surface. Of the 832 experimental observations, 80% employed at least one technique (e.g., SEM and/or TEM) to examine the material surface topography (Table S8). The SEM image generated from multidirectional secondary electrons imaging allows not only to visualize the three-dimensional (3D) morphology of materials, but also to measure their size, dispersion state, and in some cases, even the size and distribution of pores. However, TEM technique has higher resolution and can provide more information regarding the internal microstructure of materials. Fazal et al. (2020) reported that TiO 2 nanoparticles were successfully attached to the surface of algae-based biochar, exhibiting uniform distribution with small size variation, leading to improved performance of the photocatalyst system for the removal of methylene blue dye from an aqueous medium. On the contrary, Chekem et al. (2017) observed that CAT/ SB.37 catalysts had negligible photoactivity towards phenol removal due to the presence of aggregated titania inside the carbonaceous matrix.
The energy-dispersive X-ray microanalysis (i.e., EDX, EDS, EDAX) was reported by 52.2% of the studies and is another important technique applied in conjunction with SEM for the determination of the composition or orientation of individual crystals or features of BCPs (Table S8).
Advanced spectroscopy technologies, including FTIR (n = 502), XPS (n = 213), and Raman spectroscopy (n = 148) have been used to infer about BCPs elemental composition, chemical statement, structural defect, and surface functional groups, and can be further utilized for investigation over their photo-and chemical stability during reactions (Qiu et al. 2021). For example, the FTIR analysis of N20Z catalyst indicated that all peaks and initial functional groups were present in the reused composite, demonstrating its stability even after several recycles (Leichtweis et al. 2020).
In addition, XRD was reported by 87.3% of the selected studies and it is the best technique to determine the crystallographic structure of the photocatalyst (Table S8). The knowledge of the crystal phases is crucial, since the combination of two or more of these phases in a composite can enhance the photoactivity in the visible-light region (Feng et al. 2015). For example, Aeroxide Degussa P25 (TiO 2 ) titania has been widely used as a reference photocatalyst and consists of 30% rutile and 70% anatase (Wang et al. 2012). Luo et al. (2015) measured the XRD patterns of BCPs and the results showed superior photocatalytic degradation of BPA by TiO 2 -WC-500 (58.5 wt% anatase and 41.5 wt% rutile) over commercial TiO 2 (P25), since adsorbed pollutant by wood charcoal biomass speed up the mass transfer to the decomposition center of TiO 2 in TiO 2 -WC.
The average surface area (n = 491), total pore volume (n = 342), and mean pore diameter (n = 219) of the BCPs compiled on the studies were 185.97 ± 202.24 m 2 g −1 , 0.19 ± 0.19 cm 3 g −1 , and 15.27 ± 21.62 nm, respectively (Table S8). The properties of the BCPs can be influenced by the biomass source (Gonçalves et al. 2020) but also by the synthesis methodologies applied Huang et al. 2019). High specific surface areas are normally related to more exposed reaction sites, contributing to highly efficient surface adsorption capacity and catalytic performance of BCPs (Le et al. 2012).  synthesized TCNSP composites via acid catalyzed sol-gel method with a high specific surface area (454 m 2 g −1 ) and great porosity (66.9%) along with a pore size of less than 5 μm. In addition, Khraisheh et al. (2013) associated the high BET surface area of TCNSP100 with increasing calcination temperature due to increased polymerization reactions. Furthermore, the porosity of BCPs can display a similar trend with the pore volume of the composites, i.e., with increasing calcination temperature, the porosity of the BCPs increase; however, it decreased after determined temperature and time of synthesis reaction . The effect of the specific surface area of BCPs on the removal efficiency is in line with PCA results (Fig. 4 and Table S2b), which suggest that a higher BET surface area is advantageous for the removal of pollutants from water.
The average BCP pH pzc (isoelectric point) was found to be 6.58 ± 1.00 (n = 102). The results indicate that the photocatalyst process is more efficient at near neutral pHs with more OH-ions being absorbed on the surface and increasing radical hydroxyl production at pH values lower than pH pzc (Asgharzadeh et al. 2020).
Also, many scientific studies have been directed towards broadening the visible light band of semiconductors for their effective utilization under solar light irradiation (Dávila-Jiménez et al. 2018;Peñas-Garzón et al. 2021). The average band gap of BCPs reported by 421 observations in the studies was 2.84 ± 0.50 eV. This value is lower than the band gap of bare TiO 2 (Eg ≈ 3.2 eV) and ZnO (Eg ≈ 3.37 eV), which indicates the combination of carbonaceous materials with semiconductors might lead to narrowing the band gap. An interesting literature review by  reports the major techniques tried for reducing the band gap on TiO 2 biochar-supported photocatalysts.

Mechanisms associated to pollutants degradation by BCPs
The study of the pollutant category and their remediation mechanism is a necessary and imperative burning issue. On BCPs mechanism, efforts should be made to determine whether adsorption and photocatalysis is the dominant factor for the pollutant removal without/with light irradiation. Adsorption interactions between the inorganic contaminants (metals) and the biomass-based material can occur via: (i) ion exchange between target metal and exchangeable metal in biochar; (ii) electrostatic attraction of anionic metal and cationic metal; and (iii) precipitation (Ahmad et al. 2014). The postulated adsorption mechanisms of biomass-based materials with organic contaminants include: (i) Partition onto non-carbonized area; (ii) electrostatic attraction (anionic and cationic); (iii) hydrogen bonding; (iv) surface precipitation, among others, such as physical adsorption   (Fig. 7a). The organic molecules should be assisted by functional groups (hydroxyl, carboxyl, carbonyl, amine), which favors their adsorption on the biomass-based surface material (Ambaye et al. 2021).
In the photocatalytic reaction, one of the most essential components of photocatalysis is charge separation caused by the e − /h + pair. Semiconductor photocatalysts are generally photoexcited when a photon (light exposure) with an energy of hv matches or exceeds the bandgap energy (Eg) of the semiconductor (Fig. 7b). Their next pathway follow the excited state conduction-band (CB) electrons and valenceband (VB) holes that can recombine and dissipate the input energy as heat, get trapped in metastable surface states, or react with electron donors and electron acceptors adsorbed on the semiconductor surface or within the surrounding electrical double layer of the charged particles (Hoffmann et al. 1995). The biomass-based materials may function as a suitable material available to trap the e − or h + , preventing rapid recombination during photocatalysis   (Fig. 7c). Biomass-based materials can also present a larger surface area where nanoparticles can scatter more uniformly on their surface. Well-dispersed surface nanoparticles can boost light scattering and upsurge the number of active sites, which improves the photo-degradation of adsorbed pollutants (Khan et al. 2020). Additionally, adequate surface functional groups facilitate the adsorption of various pollutants, which is advantageous for photocatalysis (Khan et al. 2021). Because of the direct attack of generated reactive species (RS), synchronous adsorption and photodegradation accomplish faster with higher organic pollutant degradation (Fig. 7b).
Identification of specific RS functions during photocatalysis using various BCPs is essential due to the particularities of distinct pollutant degradation. A comprehensive list of different RS-function identification processes for organic and inorganic model pollutant photodegradation by BCPs, compilated in this study, is summarized in Table S9. Isopropanol and tert-butyl alcohol (58% and 27% of the total observation, respectively) are the most used scavengers for hydroxyl radicals ( • OH). EDTA (55%), ammonium oxalate (22%) and potassium iodide (15%) for hole (h + ), and benzoquinone (90%) for superoxide radical (O 2 •¯) . Only 4 studies   (Table S9), for both dyes and pharmaceutical compounds. The differences between the studies observations are explained considering that the photodegradation mechanism can be affected by several characteristics of the material: in particular, features such as morphology, dimension/shape of the crystals and possible presence of other phases (Quarta et al. 2019). It is worthwhile to mention that only 33 out of 107 compiled studies reported the role of specific reactive species on photodegradation. Therefore, further investigations of the BCPs reactions must be performed to better understand the action of the different active species during the pollutant degradation reaction.

Reusability and stability
The stability and reusability of BCPs are very important factors for practical applications on a field/real scale (Ahmadi and Ganjidoust 2021). To assess the reusability of the sample, reuse tests should be conducted under the same reaction conditions. In total, only 60 (7.2%) studies evaluated the BCPs removal efficiencies before and after reusability (Table 1). Almost half of these studies (n = 29) reported the reusability of BCPs for the degradation of dyes, being methylene blue (MB) the most investigated one (n = 11). In the TiO 2 -6%GC system, the authors indicated no reduction of BCP performance for MB removal, even after four repeated runs (Wu et al. 2015). For other dyes, such as acid orange 7, only a slight decrease in performance was observed and even though, it remained above 80% after six runs for MMa catalysts (Silvestri et al. 2019). The crystal violet removal decreased by only 5% during the fourth cycle of reuse and 15% in the fifth cycle, which was assigned to the loss of material during the recovery process and the decrease in active sites available on the material surface (Abarna et al. 2019). Likewise, in the photocatalytic oxidation process, no remarkable activity decline was observed for the LC/ZnO-2 catalyst towards methyl orange (MO) degradation after three cycles, suggesting that the photo-corrosion of ZnO nanoparticles can be efficiently inhibited by mixing LC into the composite (Wang et al. 2017). The average loss performance in the removal of dyes between virgin and reused BCPs was 12.1 ± 11.5% after the cycles (average repeated cycles of 5.3 ± 5.0) ( Table 1).
When aiming to remove pharmaceutical compounds, BCPs performance has resulted in similar trends. The reusability results are based on 21 studies and the average difference in removal of pharmaceuticals between fresh and reused BCPs was 10.3 ± 10.6% with the average repeated cycles of 5.0 ± 1.7 (Table 1). For example, the degradation of gemifloxacin decreased only slightly with increasing cycle number, maintaining above 85% within 130 min after five reuse cycles in Zn-Co-LDH@BC . The TCT-375-500 catalyst degraded 99% of tetracycline hydrochloride (TH) within 30 min, even after four repeated runs under simulated sunlight irradiation . For ZnO/ZnS@BC catalyst, a slight but progressive performance decrease was observed when the photodegradation efficiency of norfloxacin decreased from 95 to 79% after five consecutive cycles at 180 min under ultraviolet light (Liu et al. 2020).
The stability and reusability of BCPs on successive cycles of phenol degradation were also evaluated by few studies (n = 6). Makrigianni et al. (2015) reported that the BCP, named as CT 0.2/2, presented practically negligible loss of adsorption capacity and photocatalytic activity after three catalytic cycles of reuse. On the other hand, the BCPs produced from Pine Tar 773, commercial TiO 2 (P25), and lignin (25 wt.% TiO 2 /secondary char + LIGNIN) or softwood pellets biochar (25 wt.% TiO 2 /secondary char + SWP700) showed a decrease on the degradation of phenol from 35.7% to 0.5%, and from 51.6% to 13.4% after five cycles, respectively, which represented a significant loss of their activities (Lisowski et al. 2018b). The authors pointed out three possible reasons that could be responsible for this decrease in phenol degradation: (i) the synthesis temperature may not been high enough to yield stable char; (ii) small mineral particles on the surface of carbon materials tended to surface agglomeration and accumulation on or inside the catalyst surface; and (iii) the high pH observed on the suspension may be repelled the phenolate species away from the photocatalyst surface, thereby opposing adsorption of contaminant molecules and decreasing the photocatalytic activity (Lisowski et al. 2018b).
In the standard protocol used to regenerate a photocatalyst, washing and drying the powder after treatment are commonly used due to limited adsorption for most materials (Quarta et al. 2019). However, for materials with high surface areas like BCPs, where adsorption can considerably influence the removal performance, different regeneration protocols are employed, including thermal or chemical treatments (Rocha et al. 2017). For example, the degradation efficiency of methyl orange decreased constantly for 30 subsequent cycles by TiO 2 /BC catalyst, while after treatment with calcination under N 2 the BCP restored the catalytic activity after any build-up of organic intermediates (Zhang et al. 2014). Similarly, the catalytic activity of TiO 2 -6%GC could be effectively recovered by treating with anhydrous ethanol and deionized water after each cycle, and the MB removal efficiency had no distinct activity decay after four cycles (100%) (Wu et al. 2015).
The overall average of decrease in photodegradation efficiency of pollutants by all types of BCPs after reuse was 12.0 ± 11.4% and the average of reuse cycles was 5.2 ± 3.7. Qiu et al. (2021) indicate that at least five running cycles are necessary to evaluate the recyclability of photocatalysts. The results indicate that the use of biomass can promote the stability of biomass-based photocatalysts with the advantage of maintaining satisfactory catalytic performance over multiple successive cycles.

Lack of information, limitations, and prospects
It is clear that the number of experiments investigating the influence of experimental parameters and synthesis methods on the photodegradation performance of BCPs has grown during the last decade. However, valuable information is missing in most investigations related to photodegradation kinetics, particularly for other pollutants rather than dyes. The increasing number of micropollutants of environmental and human health concerns found in water matrices demand urgent solutions. Optimization of the photocatalysis process for contaminants removal is a promising way to go. Likewise, the gathered results presented huge variability, which accentuates the demand for additional data based on accurate experimental methods that can help reducing current uncertainties.
Another critical issue that restrained the analysis was the scarcity of information regarding the pre-treatment of biomass feedstocks that precedes the photocatalyst composite synthesis. In some experiments, only the name of the biomass had been provided without elemental composition or any characterization. In other cases, even worse, the pretreated biomass was generally referred to as "charcoal," with no details about the thermal treatment applied. Due to uncertainties, we did not include these studies in our analysis. Other studies were kept out of the analyses due to some important missing information, which otherwise could have provided a valuable contribution to the multivariate data analysis. In addition to these limitations, the dataset was reduced due to a lack of comparative experiments with conventionally used catalysts, e.g., TiO 2 or ZnO. We found that only 5.3% of the observations (n = 42) performed sideby-side comparisons. Therefore, for comparison purposes, it is strongly recommended to include a commercial catalyst (i.e., Aeroxide Degussa P25 TiO 2 ) as a control treatment in future investigations on BCPs photocatalytic performance.
Most of the studies make use of the traditional onefactor-at-a-time (OFAT) approach, examining the effect of individual parameters, such as initial concentration of target compound, degradation time, catalyst dose, and  (Sakkas et al. 2010). The chemometric experimental design (DoE) in BCPs photocatalytic processes ought to be adopted to reach the optimum catalytic reaction. Response surface methodology (RSM) and a central composite design models (CCD) is highly recommended since it can be efficiently applied to investigate the optimum conditions to synthesize the biomass-based catalysts (Khraisheh et al. 2013), as well as to interpret several experimental parameters and the interaction effects between them (Sleiman et al. 2007).
In addition to the thorough investigation of BCPs synthesis and experimental conditions under laboratory scale, BCPs applicability must also be evaluated in real water and full-scale conditions. Removal from real water was only assessed in 5% of the observations (n = 42), and its effect (i.e., percent reduction in pollutant removal) was scarcely quantified. We vigorously reassure that future studies must test how effective BCPs are in removing pollutants from real water (e.g., complex matrices such as municipal or industrial wastewater at environmentally relevant concentrations), and must investigate pollutants that reflect current water treatment demands (Tan et al. 2016;Cui et al. 2020;Minh et al. 2020).
Most of the current studies have employed synthetic dyes as a model pollutant. However, this should be avoided, since several dyes have the ability, when photoexcited, to inject an electron into the conduction band of a semiconductor (Grätzel 2009) becoming themselves a source of photocatalytic activity. Therefore, sensitized dye degradation is a confounding factor in the assessment of photocatalytic activity because, in the irradiated suspension, degradation could be due to either a photocatalytic process (the genuine effect one wants to highlight), a dye sensitization or both (Barbero and Vione 2016).
Most of the reports focus on enhancing photocatalytic activities, whereas little attention has been paid to the recovery and reusability of the BCPs photocatalysts. In addition, when the photocatalyst is used in suspension (powder form), a further step is required to remove the catalyst from the liquid and recover it for reuse. These issues are critical to the feasibility of scaling up this technique for practical applications in water/wastewater treatment (Tang et al. 2013). Therefore, reusability remains a challenging aspect for the development of BCPs materials with higher photocatalytic performance in full-scale conditions. In this regard, BCPs can be immobilized onto different support materials, such as glass, steel, and membrane, to overcome the removal step of the composite after its use and allow reuse (Cunha et al. 2018).
Finally, it must be highlighted that very few studies have addressed economic aspects associated with the new materials (n = 3), which is of paramount importance for acceptance of any technology by the water treatment industry sector Iervolino et al. 2020).
In general terms, we recommend that future studies on the application of photocatalytic treatment should focus on micropollutants of increasing concern, having in common, an experimental protocol including reaction kinetics, and the reaction mechanisms involved in the process, as well as the identification of major transient intermediates and a control with commercial photocatalyst. Likewise, we strongly reinforce the systematic provision of some important basic information, such as (i) biomass characteristics (physicochemical characterization); (ii) biomass-based photocatalyst synthesis procedures; (iii) experimental conditions (pollutant concentration, catalyst concentration, temperature profile, pH, experiment duration, and irradiation source); and (iii) catalyst characterization before and after photocatalysis application.
While the systematic review has allowed us to explore some of the variables that regulate the removal of pollutants by BCPs, we could not analyze the influence of several other factors that may be important, due to an insufficient number of studies to establish statistical relationships. Nonetheless, the use of standardized experimentation protocols can set guidelines comparison of novel photocatalysts (Parrino et al. 2019), contributing to the scaling up of this attractive technique in a near future.

Conclusion
This systematic review focuses on BCPs composites/nanocomposites, formed by, at least, the photocatalysts TiO 2 or/and ZnO and carbon-rich renewable sources of biomass originated from agricultural wastes, forest residues, solid wastes, industrial by-products, and non-conventional materials. The valorization of biomass residues which are discarded as bio-wastes and eventually landfilled, have gained considerable weightage for BCPs synthesis, as a green technology strategy. The majority of BCPs evaluated through multivariate analysis have been synthesized through traditional methods (e.g., impregnation, hydrothermal and sol-gel), but also by non-conventional methods such as infiltration and thermal decomposition, which have been explored during the last decade. Future research may delve into less explored biomasses and straightforward methodologies to design BCPs to be innovative, efficient, and sustainable in view of economic aspects. Common types of organic and inorganic pollutants were listed and several experimental parameters, which have a crucial impact on the BCPs photocatalytic degradation have been critically discussed. The efficiency of BCPs is affected by the synthesizing method. Synthesis temperature and time play a fundamental role in the surface properties, impacting the performance of the composite for the removal of contaminants from water. The surface area of the BCPs is generally higher than the corresponding photocatalyst, and the band gap energy show potential to be minimized. The presence of biomass on BCPs composites may stabilize the material to be reused efficiently, which increase the possibilities for technology scaling up. However, more investigations should endorse standardized photodegradation protocols, including the characterization of materials to enable an effective comparison of catalysts performance. The present groundbreaking study provides a comparative assessment of biomass-based photocatalysts applications for the removal and photodegradation of target pollutants in the aqueous phase and is consequently a noteworthy improvement on the route to establish effective options for environmental pollution management.