Unified machine learning dataset for hydrogen yield prediction from microwave-assisted pyrolysis of waste

Dataset Summary This dataset supports the development and benchmarking of machine learning models for predicting hydrogen yield from microwave-assisted pyrolysis of diverse waste feedstocks. It comprises 241 experimental datapoints compiled from 16 peer-reviewed studies and includes 25 input features spanning feedstock properties, microwave operating conditions, microwave absorber dielectric properties, and catalyst characteristics. No publicly available dataset currently exists that combines all four categories of input features for hydrogen yield prediction from microwave-assisted pyrolysis in a single unified framework. 1. Scope 1.1 Context, Purpose, and Motivation Global hydrogen demand is approaching 100 million tonnes per year, driven by widespread use in petroleum refining, ammonia synthesis, and methanol production. Approximately 96% of current hydrogen production relies on fossil fuels, with steam methane reforming contributing approximately 920 Mt CO2 annually. Thermochemical conversion of waste feedstocks represents a promising low-carbon alternative that simultaneously addresses waste management challenges and supports circular economy objectives. Microwave-assisted pyrolysis is an emerging thermochemical conversion pathway that offers advantages over conventional pyrolysis, including lower energy consumption, reduced secondary reaction time of volatile intermediates, and increased gas-phase product yields. Machine learning has been increasingly applied to predict microwave-assisted pyrolysis product yields. Existing machine learning studies, however, have focused mostly on bio-oil, biochar, and total gas yields. Hydrogen production differs from bio-oil, biochar, and syngas production. Hydrogen generation depends not only on feedstock decomposition but also on secondary conversion reactions such as reforming and cracking, which are usually catalyst driven. Furthermore, when plastic feedstocks are used, a microwave absorber is required, since plastics do not absorb microwave energy. Existing studies do not include catalyst properties and microwave absorber dielectric properties as input features, despite their known influence on catalytic cracking efficiency and microwave energy transfer. Thus, no unified machine learning dataset exists for hydrogen yield prediction across diverse feedstock types in microwave-assisted pyrolysis systems incorporating catalyst and absorber dielectric properties. This dataset fills that gap by providing a structured and fully documented experimental dataset for machine learning prediction of hydrogen yield from microwave-assisted pyrolysis across diverse feedstock types. The dataset is specifically designed to support: 1. Training and evaluation of machine learning models for hydrogen yield prediction from microwave-assisted pyrolysis of diverse feedstocks 2. SHAP-based interpretability analysis of input feature contributions to hydrogen yield Researchers working in thermochemical conversion, waste valorization, and machine learning for engineering applications will find this dataset applicable. Users should be aware that the dataset is compiled from laboratory scale experiments across diverse research groups and reactor configurations. Before extending model predictions to industrial settings, users should consider differences in reactor scale, feedstock variability, heating uniformity, and operating conditions that may not be represented in the compiled dataset. 1.2 Requirements The dataset compilation was guided by the following three requirements: Feedstock diversity - The dataset covers biomass, plastic waste, and mixed municipal solid waste feedstocks across 16 experimental studies, rather than a single feedstock type. This was necessary to develop a generalizable model for hydrogen yield prediction from microwave-assisted pyrolysis and motivated the snowball sampling approach used during data compilation. Feature completeness - Hydrogen yield from microwave-assisted pyrolysis is governed by the combined effects of feedstock composition, catalyst/absorber activity, and microwave heating behaviour. The dataset therefore incorporates 25 input features covering four categories (feedstock properties, microwave operating parameters, microwave absorber properties, and catalyst properties) to capture these combined effects. Incomplete datapoints were retained using missing value encoding rather than discarded. Missing value transparency - Assembling data from multiple experimental studies introduced two types of missing values. Features that are not applicable to the experiment are indicated as MISSING_NA, and features that exist but were not reported are indicated as MISSING_NR in the dataset. These two types carry different physical meanings and were coded separately to prevent physically meaningless values from being introduced into the model. 2. Ethicality and Reflexivity 2.1 Ethicality This dataset contains no data about human subjects. All data was extracted from peer-reviewed publications reporting laboratory-scale pyrolysis experiments. Informed consent is therefore not applicable, and no ethical review processes involving human subjects were required or conducted. The primary benefit of this dataset is to assist research on data driven hydrogen yield prediction from microwave-assisted pyrolysis of waste feedstocks. The potential harms of releasing this dataset are minimal. The dataset contains only experimental measurements from published studies and does not include proprietary process parameters, confidential industrial data, or information that could facilitate harm. An alternative approach to constructing this dataset would have been to generate new experimental data in a single laboratory under controlled conditions. This approach was not pursued for two reasons. First, generating sufficient experimental datapoints to cover the range of feedstock types, catalyst/absorber systems, and operating conditions represented in the literature would require a large-scale multi-year experimental programme beyond the scope of this study. Second, the value of a unified dataset lies in its ability to capture the diversity of conditions reported across multiple research groups. A single-laboratory dataset would not achieve this objective. Overall, the benefits of enabling research in data driven microwave-assisted pyrolysis hydrogen yield prediction are considered to outweigh the potential risks, provided that users perform experimental validation before extending model predictions to specific industrial or process design applications. 2.2 Domain Knowledge and Data Practices Creating this dataset required expertise across several areas. Understanding microwave-assisted pyrolysis was needed to interpret how feedstock composition, operating conditions, microwave absorbers, and catalysts affect product distribution. This understanding also helped distinguish between not-applicable and not-reported missing values. Interpreting feedstock composition, catalyst characterization, and dielectric property measurements was needed to correctly assign values to input features and catch inconsistencies in reported data. WebPlotDigitizer was used to extract data from figures. Unit conversion and standardization across studies was an important part of this process. Python and pandas were used to assemble and validate the dataset. Proficiency in scikit-learn, XGBoost, and SHAP was needed for the modelling and interpretability analysis in the accompanying paper. 2.3 Positionality This dataset was created by a research team with expertise in chemical engineering, pyrolysis, and machine learning. The background of the team shaped how the dataset was built, including which features were included and how missing data was handled. The dataset was created to support the development of machine learning models for predicting hydrogen yield from microwave-assisted pyrolysis of diverse waste feedstocks. Researchers with different goals, such as those focused on social or economic aspects of energy systems, may find this dataset less useful for their needs. The dataset compilation process reflects two different perspectives. From a chemical engineering perspective, features that are physically meaningful were prioritized, such as absorber dielectric properties, catalyst pore diameter, surface area, and metal loading, even though some of these properties were only available for a small portion of the data. From a machine learning perspective, having enough data and diversity to train a model that works well was also important, which is why data was combined from 16 studies and incomplete datapoints were kept instead of discarded. These choices affected the dataset structure. A different research team might have made different choices, producing a different dataset. Users applying this dataset to their own work should keep this in mind. 2.4 Environmental Footprint The environmental footprint of this dataset relates to the computational resources used for model training and evaluation, not for data generation (which was performed by the original experimental studies). Model training was performed on a standard laptop computer (Apple M-series processor, estimated average power draw of 15W). The full six-model pipeline with 10-fold cross-validation and 50 RandomizedSearchCV iterations required approximately 20 minutes of computation time. Energy consumption = 20 min × (1 hr/60 min) × (1 kW/1000 W) × 15W = 0.005 kWh Using an Alberta grid emission intensity of approximately 0.47 kg CO2 e/kWh (Alberta grid, 2023): Carbon footprint = 0.005 kWh × 0.47 kg CO2 e/kWh ≈ 0.0024 kg CO2 e The computational footprint of this dataset is negligible. The dataset itself (241 rows, 26 columns) is a small tabular file with no significant storage or transmission footprint. The dataset was designed to be intentionally compact, compiled from existing literature rather than generated through new computational simulation or experimental work. This keeps the overall environmental footprint minimal relative to the potential research impact of enabling more efficient microwave-assisted pyrolysis process design. 3. Data Pipeline A dataset of 241 datapoints was assembled from peer-reviewed literature using a snowball sampling approach. Review articles were first identified through a Scopus search using the keywords "microwave-assisted pyrolysis" AND "hydrogen production". A total of 16 experimental studies reporting hydrogen yield were selected from relevant review articles. Data extraction was performed manually by the dataset creator. 3.1 Dataset Composition The dataset contains 241 instances, each representing a single microwave-assisted pyrolysis experimental run. Each instance consists of 25 input features and one output feature (hydrogen yield). The input features are divided into four categories: 1. Feedstock properties (10 features): particle size, carbon content, hydrogen content, nitrogen content, oxygen content, sulfur content, moisture content, volatile matter, fixed carbon, ash content. 2. Microwave operating parameters (4 features): pyrolysis temperature, heating rate, microwave power, isothermal time. 3. Microwave absorber properties (5 features): absorber identity, absorber particle size, dielectric constant (ε′), dielectric loss tangent (tan δ), feedstock to absorber ratio. 4. Catalyst properties (6 features): catalyst identity, catalyst surface area, catalyst pore diameter, metal loading, feedstock to catalyst ratio, catalyst particle size. 3.2 Source Studies The 16 experimental studies contributing data to this dataset are cited in the accompanying dataset file. Each instance in the dataset includes the digital object identifier (DOI) of the source publication from which the experimental data was extracted. Studies were selected based on the following criteria: 1. Reports hydrogen yield directly, or indirectly through gas composition and total gas yield that allows hydrogen yield to be calculated (mmol H2/gfeedstock or equivalent convertible units). 2. Uses microwave heating as the energy source for pyrolysis. 3. Reports at least some of the 25 input features alongside hydrogen yield. 4. Published in a peer-reviewed journal. Studies reporting only total gas yield without hydrogen-specific data were excluded. Studies reporting hydrogen yield from conventional thermal pyrolysis of waste feedstocks were excluded. All experimental data was compiled from microwave-assisted pyrolysis conducted without externally introduced steam or oxidising agents, so the dataset does not represent steam-assisted or oxidative pyrolysis. 3.3 Missing Value Encoding Two types of missing values were identified and indicated separately in the dataset. The first type of missing value was not applicable, where an input feature was absent because it was irrelevant to the experimental study. For instance, catalyst properties such as catalyst surface area, pore diameter, metal loading, and particle size were absent when no catalyst was used. Imputing a numerical value for such input features would introduce physically meaningless information into the model. These values are denoted as MISSING_NA in the dataset. The second type of missing value was not reported, where an input feature existed physically but was not reported in the original experimental study. For instance, a study may have used a catalyst without reporting its surface area or operated at a specific heating rate without documenting it. These values are denoted as MISSING_NR in the dataset. 3.4 Data Processing, Wrangling, and Annotation Data processing for this dataset consisted of the following steps: 1. Numerical values reported in the dataset were taken directly from the experimental studies. Values reported in figures were extracted using WebPlotDigitizer (version 4.6). 2. All features were converted to consistent units across all instances (for example, particle sizes to mm, surface areas to m2/g, pore diameters to nm, hydrogen yield to mmol H2/gfeedstock). 3. In cases where hydrogen yield was not reported directly, it was calculated from reported gas composition (vol%) and total gas yield (L/g or similar units). Calculated datapoints are indicated in green in the attached dataset file. 4. Missing entries were labelled in the dataset as either MISSING_NA or MISSING_NR, depending on whether the input feature was applicable to the experiment. 5. All extracted values were cross-checked against the original publications to verify accuracy. Unit conversions were independently verified. No normalization, imputation, or feature engineering was applied to the raw dataset. These steps are implemented in the accompanying Python code for modelling purposes and are applied after the train-test split to prevent data leakage. 4. Data Quality 4.1 Suitability This dataset was designed to support machine learning prediction of hydrogen yield from microwave-assisted pyrolysis, addressing key gaps in the existing literature. Existing datasets focus mostly on bio-oil, biochar, and biogas yields. Only one existing study targets hydrogen yield, and it covers a single feedstock type (municipal sludge, the solid waste separated from wastewater during urban sewage treatment) with only two input features (feedstock proximate analysis and microwave power). Most existing datasets use only feedstock composition and operating parameters as inputs, leaving out catalyst properties and microwave absorber dielectric properties, even though these are known to affect catalytic cracking and microwave energy conversion. This dataset fills the above mentioned gap and includes all four types of factors, feedstock composition, operating conditions, absorber properties, and catalyst characteristics. Most existing studies focus on a single feedstock type, such as only biomass or only plastic waste. This dataset covers biomass, plastic waste, and mixed municipal solid waste feedstocks from 16 studies, giving enough variety to train models that work across different feedstock types. Beyond its primary purpose, the dataset is suitable for other research tasks such as: 1. Benchmarking machine learning models for regression tasks with missing values. 2. Fine-tuning models trained on this dataset using new experimental data. 3. Sensitivity analysis and uncertainty quantification for hydrogen yield prediction. 4. Reviewing experimental conditions reported across the microwave-assisted pyrolysis literature. 4.2 Representativeness The dataset contains 241 experimental runs from 16 studies on microwave-assisted pyrolysis. While it covers a range of feedstock types, operating conditions, absorber properties, and catalyst characteristics, certain limitations should be considered. Several input features are correlated with catalyst type and cannot be independently varied in analyses. Dielectric properties are available for only 32% of instances, limiting generalizability for absorber materials. All experimental data was compiled from microwave-assisted pyrolysis conducted without externally introduced steam or oxidising agents, so the dataset does not represent steam-assisted or oxidative pyrolysis. The dataset is also static and reflects the literature only at the time of compilation. The dataset is dominated by plastic waste (LDPE, HDPE) and lignocellulosic biomass, with municipal solid waste and mixed feedstocks under-represented. Iron-based catalysts and zeolites account for most catalyst-based datapoints, while other catalyst families such as metal oxides are less represented. The snowball sampling approach favours studies cited in review articles, which may skew towards more widely cited or earlier published work. Studies with poor reporting practices were effectively excluded, which may bias the dataset towards more rigorously conducted experiments. The 16 peer-reviewed studies originate predominantly from research groups in Asia and Europe, so experimental procedures and reactor designs from other regions may not be well represented. 4.3 Authenticity and Reliability Every value in this dataset is taken directly from peer-reviewed publications. Each instance includes a DOI, so the original publication can be found, and the dataset values can be checked against it. No changes were made to the original experimental values, except for converting units to a consistent format (fully documented) and calculating hydrogen yield from gas composition data where it was not reported directly (these calculated values are shown in green in the attached dataset file). Values reported in tables were copied directly from the original publications. For values extracted from figures using WebPlotDigitizer, there is an error of ± 2-5% depending on figure resolution. Hydrogen yield values calculated from reported gas composition and total gas yield data introduce additional uncertainty. The accuracy of any datapoint can be checked by finding the source publication using the DOI and comparing it to the original tables or figures. If any discrepancies are found, please report them to the dataset maintainer using the contact details in Section 5.3. 4.4 Structured Documentation The dataset card was developed following the structure recommended by the “Evaluating Datasets for ML: Toolkit”. The dataset card is accompanied by a “Datasheets for Datasets” following Gebru et al. (2021) framework. The dataset Excel file is also attached, containing the raw data with calculated datapoints indicated in green and DOIs included. 5. Data Management 5.1 Findability This dataset is publicly available on Zenodo. Upon publication of the accompanying paper, users should cite the DOI rather than the Zenodo URL, as the DOI provides a permanent link that will remain valid over time. 5.2 Accessibility and Interoperability The dataset is publicly available without registration or login requirements. It can be accessed and downloaded directly from the Zenodo page. No data sharing agreement is required. The dataset is provided as an Excel file (.xlsx format). This is the format used by the accompanying Python code. The file includes colour coding to distinguish calculated datapoints from directly reported values. The accompanying modelling code is available at https://github.com/RoshniSK9/MAP-Hydrogen-Yield-ML and requires Python 3.12.11 with the following libraries, scikit-learn (v1.7.1), XGBoost (v3.0.1), SHAP (v0.47.2), NumPy (v2.0.0), pandas (v2.3.1), matplotlib (v3.10.0), scipy (v1.16.0), and openpyxl (v3.1.5). Column names follow standard chemical engineering and pyrolysis terminology. The dataset includes a summary of the input and output features (Table 1 of the accompanying paper) that defines each feature, its unit, data type, and value range. 5.3 Reusability Provenance - 1. Where - Data compiled from 16 peer-reviewed publications identified through Scopus search and snowball sampling. 2. Who - Compiled by Dr. Roshni Sajiv Kumar at University of Calgary, supervised by Dr. Josephine M. Hill. 3. When - The dataset was compiled between November 2025 and January 2026. The experimental data within the dataset spans publications from approximately 2018 to 2025. License: Creative Commons Attribution 4.0 International (CC BY 4.0). Users are free to share and adapt the dataset for any purpose, provided appropriate credit is given to the dataset creators and the accompanying paper is cited. Citation - [Citation to be added upon paper acceptance] The preprocessing pipeline used for modelling is described in the methods section of the accompanying paper and in the datasheet. The full implementation is available in the accompanying Python code at https://github.com/RoshniSK9/MAP-Hydrogen-Yield-ML. Maintenance - The dataset will be managed by Dr. Josephine M. Hill at University of Calgary. Any corrections identified after publication will be communicated to the contact below and incorporated into updated versions on Zenodo. Updates will be versioned to ensure reproducibility of results reported in the accompanying paper. Contact - [e-mail] 6. References Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna Wallach, Hal Daumé III, Kate Crawford, Datasheets for Datasets, Communications of the ACM, Volume 64, Issue 12, 2021, 86–92. https://doi.org/10.1145/3458723 [Accompanying paper citation - to be added upon acceptance] Review articles identified through Snowball sampling: Zhicheng Jiang, Prof. Dr. Yuan Liang, Fenfen Guo, Yuxuan Wang, Ruikai Li, Aoyi Tang, Youjing Tu, Xingyu Zhang, Prof. Dr. Junxia Wang, Prof. Dr. Shenggang Li, Prof. Dr. Lingzhao Kong, Microwave-Assisted Pyrolysis-A New Way for the Sustainable Recycling and Upgrading of Plastic and Biomass: A Review, ChemSusChem, Volume 17, 2024, e202400129. https://doi.org/10.1002/cssc.202400129 Amani Hussein, Raihan Mahirah Ramli, Suriati Sufian, Najib Al-mahbashi, Siti Shawalliah Idris, Abid Salam Farooqi, Microwave-assisted pyrolysis of solid waste for the production of high-value carbon nanomaterials and hydrogen gas: a review, Carbon Resources Conversion, Volume 9, Issue 2, 2026, 100355. https://doi.org/10.1016/J.CRCON.2025.100355 Xuesong Zhang, Kishore Rajagopalan, Hanwu Lei, Roger Ruan, Brajendra K. Sharma, An overview of a novel concept in biomass pyrolysis: microwave irradiation, Sustainable Energy Fuels, Volume 1, Issue 8, 2017, 1664–99. https://doi.org/10.1039/C7SE00254H Yafei Shen, Microwave-assisted pyrolysis of biomass and plastic wastes for hydrogen production, Green Chemistry, Volume 27, Issue 35, 2025, 10402–22. https://doi.org/10.1039/d5gc03030g Veluru Sridevi, Dadi Venkata Surya, Busigari Rajasekhar Reddy, Manan Shah, Ribhu Gautam, Tanneru Hemanth Kumar, Harish Puppala, Kocherlakota Satya Pritam, Tanmay Basak, Challenges and opportunities in the production of sustainable hydrogen from lignocellulosic biomass using microwave-assisted pyrolysis: A review, International Journal of Hydrogen Energy, Volume 52, Part A, 2024, 507-531 https://doi.org/10.1016/j.ijhydene.2023.06.186 Yafei Shen, Hydrogen production via microwave-assisted pyrolysis – A review, Renewable and Sustainable Energy Reviews, Volume 239, 2026, 117155. https://doi.org/10.1016/j.rser.2026.117155 16 peer-reviewed papers selected from the review articles forming the basis of the dataset: Qiang Cao, Hui-Chao Dai, Jing-Hui He, Cheng-Liang Wang, Chang Zhou, Xue-Feng Cheng, Jian-Mei Lu, Microwave-initiated MAX Ti3AlC2-catalyzed upcycling of polyolefin plastic wastes: Selective conversion to hydrogen and carbon nanofibers for sodium-ion battery, Applied Catalysis B: Environmental, Volume 318, 2022, 121828.https://doi.org/10.1016/j.apcatb.2022.121828 Md Arafat Hossain, J. Jewaratnam, P. Ganesan, J.N. Sahu, S. Ramesh, S.C. Poh, Microwave pyrolysis of oil palm fiber (OPF) for hydrogen production: Parametric investigation, Energy Conversion and Management, Volume 115, 2016, 232-243. https://doi.org/10.1016/j.enconman.2016.02.058 Kaushal R. Parmar, Vishal Tuli, Ashley Caiola, Jianli Hu, Yuxin Wang, Sustainable production of hydrogen and carbon nanotubes/nanofibers from plastic waste through microwave degradation, International Journal of Hydrogen Energy, Volume 51, Part B, 2024, 488-498. https://doi.org/10.1016/j.ijhydene.2023.08.224 Ashak Mahmud Parvez, Tao Wu, Muhammad T. Afzal, Sannia Mareta, Tianbiao He, Ming Zhai, Conventional and microwave-assisted pyrolysis of gumwood: A comparison study using thermodynamic evaluation and hydrogen production, Fuel Processing Technology, Volume 184, 2019, 1-11, https://doi.org/10.1016/j.fuproc.2018.11.007 Xiangyu Jie, Weisong Li, Daniel Slocombe, Yige Gao, Ira Banerjee, Sergio Gonzalez-Cortes, Benzhen Yao, Hamid AlMegren, Saeed Alshihri, Jonathan Dilworth, John Thomas, Tiancun Xiao, Peter Edwards, Microwave-initiated catalytic deconstruction of plastic waste into hydrogen and high-value carbons, Nature Catalysis, Volume 3, 2020, 902–912. https://doi.org/10.1038/s41929-020-00518-5 Bowen Zhang, Hui Wang, Yiyun Yang, Yanping Zhou, Bing Zhang, Kama Huang, Microwave-carbon fiber cloth co-ignited catalytic degradation of waste plastic into high-yield hydrogen and carbon nanotubes, Journal of Environmental Chemical Engineering, Volume 11, Issue 3, 2023, 109710. https://doi.org/10.1016/j.jece.2023.109710 Liansheng Yao, Baokui Yi, Xiqiang Zhao, Wenlong Wang, Yanpeng Mao, Jing Sun, Zhanlong Song, Microwave-assisted decomposition of waste plastic over Fe/FeAl2O4 to produce hydrogen and carbon nanotubes, Journal of Analytical and Applied Pyrolysis, Volume 165, 2022, 105577. https://doi.org/10.1016/j.jaap.2022.105577 Yuqing Dong, Bonan Liu, Minyi He, Zijian Zhang, Chao Tang, Jun Zhao, Hanyu Wu, Yu Fan, Wangjing Ma, Microwave-coupled recycling of plastic waste into hydrogen and carbon nanotubes over economical iron-based catalyst, International Journal of Hydrogen Energy, Volume 115, 2025, 24-36. https://doi.org/10.1016/j.ijhydene.2025.03.072 Meng Xie, Hai Xu, YiXuan Wang, Helin Pan, Dengle Duan, Bo Niu, Donghui Long, Yayun Zhang, Constructing bifunctional porous nanosheets for efficient conversion of waste plastics into valuable hydrogen and carbons, Chemical Engineering Journal, Volume 471, 2023, 144460. https://doi.org/10.1016/j.cej.2023.144460 Bilal Shoukat, Hammad Hussain, Muhammad Yasin Naz, Ahmed Ahmed Ibrahim, Shazia Shukrullah, Yasin Khan, Yaning Zhang, Microwave-Assisted Catalytic Deconstruction of Plastics Waste into Nanostructured Carbon and Hydrogen Fuel Using Composite Magnetic Ferrite Catalysts, Scientifica, 2024, 3318047. https://doi.org/10.1155/2024/3318047 Wenchang Qin, Hailong Yu, Chaoqian Wang, Shuting Qin, Xiaolong Li; Experimental Study on Optimizing Hydrogen Production from Sludge by Microwave Catalytic Pyrolysis Using Response Surface Methodology, ACS Omega, Volume 9, Issue 44, 2024, 44474–44486. https://doi.org/10.1021/acsomega.4c06104 Rohit R. Mishra, Paresh H. Rana, Parimal A. Parikh, Microwave-assisted pyrolysis of low-density polyethylene (LDPE)to value-added products over HY, HZSM-5 and Hβ catalyst, Materials Today Communications, Volume 41, 2024, 110829. https://doi.org/10.1016/j.mtcomm.2024.110829 Zhengdong Zhang, Kai Huang, Chuang Mao, Jiaming Huang, Qingli Xu, Lifang Liao, Rui Wang, Shoutao Chen, Pize Li, Chenyang Zhang, Microwave assisted catalytic pyrolysis of bagasse to produce hydrogen, International Journal of Hydrogen Energy, Volume 47, Issue 84, 2022, 35626-35634. https://doi.org/10.1016/j.ijhydene.2022.08.162 Mingli Yue, Jingxin Cheng, Qiuhang Jiang, Guoqiang Xu, Jing Wang, Ying Fu, Fei Ye, Enhanced hydrogen production from straws using microwave-assisted pyrolysis with NiO/C based catalyst/absorbent, International Journal of Hydrogen Energy, Volume 59, 2024, 535-550. https://doi.org/10.1016/j.ijhydene.2024.01.253 Zhe Zhang, Huan Chen, Wenheng Hu, Meng Xie, Yukun Pan, Bo Niu, Dengle Duan, Lu Ding, Donghui Long, Yayun Zhang, Temperature-Dependent NiFeAl catalysts for efficient microwave-assisted catalytic pyrolysis of polyethylene into value-added hydrogen and carbon nanotubes, Fuel, Volume 366, 2024, 131390. https://doi.org/10.1016/j.fuel.2024.131390 Chen Zeng, ZhiWei Jiang, Yongjian Zeng, Suyu Zhang, Rafael Luque, Kai Yan, Microwave-assisted pyrolysis of biomass for efficient H2-rich syngas production promoted by calcium oxide, Chemical Engineering Journal, Volume 506, 2025, 159905. https://doi.org/10.1016/j.cej.2025.159905