Home Journals Articles Updates About
CN
image
收藏切换
A dataset for electrocatalytic hydrogen evolution reaction performance of non-noble transition metal phosphides
收藏切换
PDF
Teng ZHU1, 2, 3, Huanhuan WANG4, Yuming LI4, *
Journal of Fuel Chemistry and Technology | 2026, 54(7) : 20250390-1430 - 20250390-1436
Less
收藏切换
Journal of Fuel Chemistry and Technology | 2026, 54(7): 20250390-1430-20250390-1436
• DATA PAPER •
A dataset for electrocatalytic hydrogen evolution reaction performance of non-noble transition metal phosphides
Full
Teng ZHU1, 2, 3, Huanhuan WANG4, Yuming LI4, *
Affiliations
  • 1School of Rescue and Command, China People’s Police University, Langfang 065000, China
  • 2National Engineering Research Center for Fire and Emergency Rescue, Shenyang 110000, China
  • 3Hebei Key Laboratory of Emergency Rescue Technology, China People’s Police University, Langfang 065000, China
  • 4State Key Laboratory of Heavy Oil Processing, China University of Petroleum (Beijing); Beijing 102249, China
Published: 2026-07-10 doi: 10.1016/S1872-5813(26)60693-7
Outline
收藏切换

This dataset compiles the HER performance data of 203 non-noble transition metal phosphide (TMP) catalysts, covering detailed information on catalyst preparation (e.g., phosphating temperature, precursor, synthesis method), chemical composition (mass fractions of elements such as Ni, Co, Fe, P, Mo, W and Zn), and testing conditions (e.g., electrolyte type and concentration, electrode substrate). The key parameters for catalytic performance include the overpotential at 10 mA/cm2 (η10) and the Tafel slope. This dataset has been rigorously extracted, cleaned, and standardized to ensure a high degree of structure and machine readability. This provides a reliable data foundation for data-driven methods, such as machine learning and statistical modeling, enabling rapid screening and design of high-performance HER catalysts, supporting performance prediction, in-depth structure-activity analysis and the rational development of novel catalysts.

water electrolysis for hydrogen evolution  /  transition metal phosphides  /  dataset  /  catalyst  /  machine learning
朱腾, 王焕焕, 李宇明. 非贵金属基过渡金属磷化物电解水析氢性能数据集. 燃料化学学报（中英文）, 2026 , 54 (7) : 20250390-1430 -20250390-1436 . DOI: 10.1016/S1872-5813(26)60693-7
Teng ZHU, Huanhuan WANG, Yuming LI. A dataset for electrocatalytic hydrogen evolution reaction performance of non-noble transition metal phosphides[J]. Journal of Fuel Chemistry and Technology, 2026 , 54 (7) : 20250390-1430 -20250390-1436 . DOI: 10.1016/S1872-5813(26)60693-7
Full Text
Less
Hydrogen energy, recognized for its high energy density and zero-carbon emissions, which is widely regarded as a key clean energy source for achieving global energy transition and carbon-neutrality goals[1−2]. Among various hydrogen-production technologies, water electrolysis for hydrogen evolution reaction (HER) can directly convert renewable electricity into chemical energy, offering high-purity hydrogen with a pollution-free process, thereby representing an ideal route for green hydrogen production[3−4]. However, the high energy consumption caused by slow reaction kinetics hinders its large-scale commercial application. The solution to this problem relies on high-performance electrocatalysts. Although noble-metal catalysts such as platinum (Pt) remain the performance benchmark, their extreme scarcity and high cost limit their widespread application. Therefore, developing non-noble metal-based catalysts that combine high activity, high stability, and low cost has become a challenge and research focus for promoting the large-scale application of HER[5].
Among the many non-noble metal-based catalysts that have been developed, transition metal phosphides (TMPs) have become highly promising catalysts for the HER in recent years[6−7]. During HER process, these catalysts can induce unique electronic restructuring, generating electron-rich P sites and electron-deficient metal sites, thereby synergistically optimizing the hydrogen adsorption free energy and improving intrinsic activity[8−9]. Furthermore, TMPs exhibit metallic-like conductivity, excellent mechanical strength, and corrosion resistance in acidic, alkaline, and even neutral electrolytes. Through strategies such as elemental doping (such as Fe, Co, Mo)[10−12], structural engineering (including the construction of porous, hollow, or core-shell structures)[13], and interface engineering (e.g., the construction of heterojunctions)[14−15], researchers have developed a series of TMP-based catalysts with catalytic performance approaching that of commercial Pt/C. For instance, Zheng et al.[16] used an in situ hydrothermal method to grow Ni2P nanocrystals on black phosphorus (BP) nanosheets, constructing a Ni2P/BP heterojunction. This resulted the formation of built-in electric field from BP to Ni2P can promote interfacial charge transfer and optimize the d-band center of Ni active sites, leading to markedly improved HER activity in alkaline electrolyte. The catalyst requires only 85 mV overpotential at a current density of 10 mA/cm2 and can operate stably for 125 h, showing promising application prospects. Furthermore, the synergistic strategy of rare earth doping and carbon coating provides a new approach to improve the overall performance of TMPs. For example, Su et al.[17] constructed cerium-doped and carbon-coated NiCoP nanorod arrays on nickel foam (Ce-NiCoP@C/NF). The as-introduced Ce species can not only induce lattice distortion and enhance structural stability of NiCoP, but also significantly promote strong bonding between the carbon layer and the phosphide surface, effectively preventing carbon detachment. This catalyst achieved an HER overpotential of only 37 mV at 10 mA/cm2 in alkaline medium and required a cell voltage of merely 1.51 V for overall water splitting, outperforming the commercial RuO2||Pt/C system. However, TMP-based catalysts still face challenges such as limited numbers of active sites, insufficient long-term operational stability, and difficulty in predicting the optimal stoichiometry, which require more systematic and in-depth research into their structure-activity relationship.
To address these challenges and accelerate the rational design of TMP catalysts, data-driven machine-learning (ML) techniques have shown unprecedented potential[18−19]. ML can learn and uncover complex hidden patterns from massive, multi-dimensional experimental data, establishing accurate mapping models from material features (for example, elemental composition, synthesis parameters, and morphology) to target properties (including overpotential and Tafel slope), thereby enabling effective performance prediction and high-throughput screening of new materials[20−21]. However, the accuracy and generalizability of ML models strongly depend on high-quality, standardized, and large-scale datasets. Currently, experimental data in this field are scattered across various publications, with inconsistent formats and testing standards, which hinders the effective application of ML in the HER. Therefore, this work aims to construct a standardized dataset for the HER performance of transition-metal phosphides in water electrolysis, as shown in Figure 1. This dataset systematically integrates catalyst preparation information, multi-element composition, testing conditions, and key catalytic parameters, aiming to establish a reliable data benchmark for this field. This provides a solid data foundation for subsequent ML-assisted catalyst design, in-depth structure-activity relationship analysis, and rapid discovery of new materials. Unlike traditional research papers that use datasets for ML training, this data paper focuses primarily on describing and validating the dataset, elucidating its generation, composition, and inherent limitations. Thus, this dataset helps promote data sharing, which provides valuable data resources for understanding the design and development strategies of water electrolysis catalysts.
1　Dataset description and discussion
Less
1.1　Data value
The basic information of this dataset is described in Table 1. The core value of this dataset lies in providing a structured benchmark for the study of non-noble transition metal phosphide (TMP) electrocatalysts in HER. Through the integration of TMP catalysts, the dataset offers insights into revealing macroscopic performance trends and structure-activity relationships. The violin plots of overpotential and Tafel slope (Figure 2(a), (b)) visually illustrate the overall performance levels of existing TMP catalysts. The results show that the overpotential (η10) is densely distributed in the range of 50–200 mV, while the Tafel slope is concentrated between 50–100 mV/dec, which provides a clear reference for evaluating the competitiveness of newly developed catalysts. As shown in Figure 2(c), the relationship between overpotential and Tafel slope further demonstrates a certain positive correlation. Some of these TMP catalysts exhibit both low overpotential and low Tafel slope, representing a direction for high-performance optimization, which indicates the potential of TMP catalysts to replace noble metal Pt-based catalysts for HER. Furthermore, it can be clearly seen from Figure 2(c) that the pH environment of the electrolyte has a significant impact on the catalyst performance, which is relatively low in acidic electrolytes.
From the preparation parameters and composition information in the dataset, the violin plot of phosphidation temperature (Figure 3(a)) clearly shows that the phosphidation temperatures in the dataset are mainly concentrated between 200–400 °C, providing foundational data for the key process parameters in TMP catalyst preparation.
Statistical analysis of element occurrence frequency further validates the importance of the dataset from another dimension. As shown in Figure 3(b)−(d), Ni, Co, Fe and P are the most frequently used elements in the dataset, reflecting the current consensus and preference for elements and their combinations in TMP catalysts. Synthetic method analysis (Figures 3(e), (f)) shows that for TMP catalysts, the preparation of metal precursors and phosphating methods are mainly based on hydrothermal synthesis, which highlights the importance of expanding the synthetic methods in this field. Combining this data-driven approach with feature importance and statistical analysis promises to guide experiments, optimizing research directions and element selection strategies. Furthermore, it can reduce the blind spots inherent in traditional trial-and-error methods.
1.2　Data generation method or process
This dataset was constructed following a rigorous data engineering workflow, comprising three key stages: data collection, extraction, and cleaning and standardization. All raw data were derived from publicly available, high-quality academic journals and underwent systematic literature review to ensure the authority and representativeness of the data sources. From each publication, three major categories of information were extracted, and those are synthesis information (including metal-precursor preparation method, phosphidation method, phosphidation temperature, phosphorus precursor, and electrode substrate), composition information (element type, mass fraction, etc.), and catalytic evaluation information (electrolyte type and concentration). In the data processing phase, the extracted data were first standardized. For elemental composition data, a data-pivoting and grouping-summation approach was used to unify dispersed element contents into normalized mass fractions. Categorical variables such as synthesis methods and electrolyte types were converted into a machine-readable numerical format by one-hot encoding. Regarding catalytic performance, the overpotential at 10 mA/cm2 (η10) and the Tafel slope were adopted as the output variables, because η10 is explicitly reported in most publications and serves as a key indicator for cross-study comparison. Although the onset potential is also an important parameter for HER, it was not included as a catalytic performance indicator in this dataset due to inconsistencies in related literature and unclear definitions. This systematic data-processing workflow ensures internal consistency and machine readability of the dataset, which lays a solid foundation for subsequent data analysis and modeling.
1.3　Data sample description
This dataset is organized in a structured table format, where each row corresponds to an independent catalyst sample and each column represents a specific feature or catalytic performance. A typical data record includes the following information, including the major elemental composition distribution of the catalyst, such as phosphorus content (~10%–40%), nickel content (~20%–60%), and cobalt content (~10%–40%); preparation conditions including phosphidation temperature (predominantly within 300–400 °C), phosphidation method (mainly gas-solid phase method), and electrode substrate (including carbon cloth, nickel foam, glassy carbon electrode, etc.); as well as electrochemical performance parameters such as overpotential (50–200 mV) and Tafel slope (40–100 mV/dec) at a current density of 10 mA/cm2. As shown in Figure 3, the violin plot of the frequency distribution of element content and element combination types helps in understanding the data. The results of data analysis indicate that despite the diverse combinations of metal elements, several multi-component systems, such as Ni-Co, Co-Fe and Ni-Co-Fe, are currently the mainstream in TMP-based HER catalyst research. These statistical trends can not only reflect the differences in catalyst composition and preparation conditions, but also relate to the catalytic performance, thereby providing a visual analysis for understanding composition-performance relationships.
1.4　Data quality
To ensure the reliability and accuracy of the data, each data point is annotated with its original literature source, establishing a complete data-lineage record. During the data-extraction stage, strict selective standard was applied, and only samples that provided full preparation details, elemental composition, and key catalytic performance data were collected, while records with incomplete information were excluded.
To address potential discrepancies in test conditions across the literature, we implemented data standardization to maximize data comparability. For example, we used the overpotential at a current density of 10 mA/cm2 as the core catalytic performance indicator, which represents different testing conditions in a uniform normalized manner. Regarding missing values, samples lacking critical features such as elemental content and/or overpotential value were removed, and this can ensure that all final data possess a high degree of information completeness. In addition, we conducted consistency tests such as correlation heatmap analysis to examine the intrinsic relationships between variables, thereby effectively identifying and handling potential outliers or contradictory data, and ensuring the overall quality and reliability of the dataset.
Some catalyst parameters, such as catalyst particle size, are important factors influencing the HER performance. However, the available data on these parameters are scattered and incomplete, resulting in low data quality. Therefore, these parameters have not been included in the current version of the dataset. It is recommended that future studies systematically report such information to facilitate more efficient and comprehensive machine-learning analysis.
1.5　Data usage methods and suggestion
This dataset supports a variety of application areas and analytical methods. For machine learning research (Figure 4), the dataset can be used for regression tasks (predicting overpotential or Tafel slope). Before modeling, researchers can refer to the feature importance histograms provided in this study (e.g., SHAP analysis results) to determine the priority of key features and perform appropriate feature engineering. Preliminary modeling results indicate that for the overpotential, the random forest (RF) model achieved an R2 of 0.72 on the training set and 0.437 on the test set, while for the Tafel slope, the XGBoost model yielded an R2 of 0.722 (training set) and 0.307 (test set). The corresponding root-mean-square errors (RMSE) were 0.0018 (training set) and 0.0037 (test set) for the RF-predicted overpotential and 0.125 (training set) and 0.126 (test set) for the XGBoost-predicted Tafel slope. Therefore, it is recommended that RF and XGBoost algorithms be used as the starting point for predictive modeling of this dataset.
For the field of catalysis, the dataset can be directly employed for in-depth statistical analysis. Scatter plots between elemental composition and catalytic performance (e.g., the influence of P or Ni content on overpotential) provide an intuitive way to explore univariate trends. Cluster analysis combined with violin plots of catalytic performance distributions enables rapid assessment of the potential HER activity of different active components. A particularly valuable approach is the integration of model-interpretability tools such as SHAP, which can quantitatively uncover key factors governing catalytic performance from trained machine-learning models. For instance, SHAP analysis clarifies intrinsic determinants of HER performance. It suggested that lower phosphorus content tends to lower the overpotential, which can be due to the modulation of electronic structure of metal sites with an optimal amount of phosphorus. Besides, the combination of low Ni and high Fe contents would lead to the increase of overpotential, highlighting the need for precise tuning of bimetallic synergy. Moreover, lower phosphidation temperatures favor the formation of highly active catalysts. For Tafel slope, higher phosphorus content would reduce its value, implying the improved reaction pathways and charge-transfer kinetics. These findings reveal the complexity of performance-governing factors. For example, the effects of phosphorus content on thermodynamic parameters (overpotential) and kinetic parameters (Tafel slope) differ, indicating a need to find the optimal balance point. These key features and mechanisms can serve as important descriptors for constructing catalyst performance prediction models, which helps in the rational design of high-performance catalysts. When developing novel catalysts, key focus should be placed on the preferred metal element combinations (e.g., Ni−Co and Co−Fe) and optimized phosphating temperature ranges revealed by the dataset.
Furthermore, the performance prediction model based on this dataset can be used for preliminary screening of candidate formulations before experiments. In addition, scatter plots can be used to compare predicted and actual values to evaluate the reliability of the model, which significantly improves research efficiency and reduces experimental costs and time.
In the field of HER, the onset potential is also an important performance parameter. However, due to inconsistent reporting or large variations in test conditions across the collected literature, this parameter was not included as a catalytic performance parameter in this dataset. Therefore, it is recommended that future studies uniformly report onset potential to facilitate more comprehensive analysis.
2　Conclusion
Less
This study successfully constructed a dataset on the hydrogen evolution reaction (HER) performance over transition metal phosphides (TMPs) based catalysts. This dataset integrates preparation details, multi-element compositions, testing conditions, and key catalytic performance of catalysts reported in multiple published studies, forming a structured and standardized data resource. Through statistical analysis and visualization, this dataset reveals the macroscopic trends in the catalytic performance, elemental composition, and process selection of current TMP catalysts, providing a benchmark for catalyst evaluation and design. Furthermore, the effectiveness of the dataset in predicting catalytic performance was validated using multiple machine learning algorithms. Moreover, the application of SHAP model interpretation quantitatively identified phosphorus content, nickel content, and phosphidation temperature as the most critical features influencing HER performance, offering clear directions for rational catalyst design.
This dataset provides the first systematic database of TMP-based catalysts in the HER field, demonstrating the significant potential of data-driven methods in accelerating the screening and development of high-performance catalysts. Through data sharing and model collaboration, this dataset is expected to promote the rapid advancement of green hydrogen production technologies.
References
Less
1
PATEL K B, PARMAR B, RAVI K, et al. Metal-organic framework derived core-shell nanoparticles as high performance bifunctional electrocatalysts for HER and OER[J]. Appl Surf Sci, 2023, 616: 156499.
2
MENG X Y, CHEN M Y, GU A L, et al. China’s hydrogen development strategy in the context of double carbon targets[J]. Nat Gas Ind B, 2022, 9: 521−547.
3
LI X P, HUANG C, HAN W K, et al. Transition metal-based electrocatalysts for overall water splitting[J]. Chin Chem Lett, 2021, 32: 2597−2616.
4
CAO S, WANG C J, FU W F, et al. Metal phosphides as Co-catalysts for photocatalytic and photoelectrocatalytic water splitting[J]. ChemSusChem, 2017, 10: 4306−4323.
5
CUI Z B, JIAO W S, HUANG Z Y, et al. Design and synthesis of noble metal-based alloy electrocatalysts and their application in hydrogen evolution reaction[J]. Small, 2023, 19: 2301465.
6
LIU P, RODRIGUEZ J A. Catalysts for hydrogen evolution from the [NiFe] hydrogenase to the Ni2P(001) surface: The importance of ensemble effect[J]. J Am Chem Soc, 2005, 127: 14871−14878.
7
LIU Q, TIAN J Q, JIANG P, et al. Carbon nanotubes decorated with CoP nanocrystals: A highly active non-noble-metal nanohybrid electrocatalyst for hydrogen evolution[J]. Angew Chem Int Ed, 2014, 53: 6710−6714.
8
LIU X Y, WU X. Transition metal phosphide electrocatalysts for hydrogen generation from water[J]. Chin Chem Lett, 2025, online. doi: 10.1016/j.cclet.2025.111801.
9
LIN Y, WANG Z H, LI J Y, et al. Strengthening theelectrocatalytic hydrogen evolution process by structuredesign and atomic-level regulation of transition metalphosphides[J]. Small, 2026, 22: e09589.
10
DU H T, XIA L, ZHU S Y, et al. Al-doped Ni2P nanosheet array: A superior and durable electrocatalyst for alkaline hydrogen evolution[J]. Chem Commun, 2018, 54: 2894−2897.
11
CHU S J, CHEN W, CHEN G L, et al. Holey Ni-Cu phosphide nanosheets as a highly efficient and stable electrocatalyst for hydrogen evolution[J]. Appl Catal B: Environ, 2019, 243: 537−545.
12
RAO Y, WANG S W, ZHANG R Y, et al. Nanoporous V-doped Ni5P4 microsphere: A highly efficient electrocatalyst for hydrogen evolution reaction at all pH[J]. ACS Appl Mater Interfaces, 2020, 12: 37092−37099.
13
YAN L T, DAI P C, WANG Y, et al. In situ synthesis strategy for hierarchically porous Ni2P polyhedrons from MOFs templates with enhanced electrochemical properties for hydrogen evolution[J]. ACS Appl Mater Interfaces, 2017, 9: 11642−11650.
14
WANG C D, DING T, SUN Y, et al. Ni12P5 nanoparticles decorated on carbon nanotubes with enhanced electrocatalytic and lithium storage properties[J]. Nanoscale, 2015, 7: 19241−19249.
15
LAI S J, LV C X, CHEN S, et al. Ultrathin nickel phosphide nanosheet aerogel electrocatalysts derived from Ni-alginate for hydrogen evolution reaction[J]. J Alloy Compd, 2020, 817: 152727.
16
ZHENG Y, SONG C H, ZOU J, et al. In-situ growth of Ni2P nanocrystals on black phosphorus nanosheets endows an effective built-in electric field for enhanced H2 production in anion exchange membrane water electrolyzers[J]. Chem Eng J, 2025, 524: 169189.
17
SU J, JIANG N, WANG Y Y, et al. Understanding the synergistic effect of Ce-doping and carbon-encapsulating in NiCoP for enhanced stability and activity for overall water splitting[J]. Energy, 2025, 318: 134905.
18
ZHENG J N, SUN X, HU J X, et al. Symbolic transformer accelerating machine learning screening of hydrogen and deuterium evolution reaction catalysts in MA2Z4 materials[J]. ACS Appl Mater Interfaces, 2021, 13: 50878−50891.
19
CAO S Y, LUO Y H, LI T H, et al. Machine learning assisted screening of doped metals phosphides electrocatalyst towards efficient hydrogen evolution reaction[J]. Mol Catal, 2023, 551: 113625.
20
JIANG X, WANG Y, JIA B R, et al. Prediction of oxygen evolution activity for NiCoFe oxide catalysts via machine learning[J]. ACS Omega, 2022, 7: 14160−14164.
21
BAGHBAN A, HABIBZADEH S, ASHTIANI F Z. On the evaluation of hydrogen evolution reaction performance of metal-nitrogen-doped carbon electrocatalysts using machine learning technique[J]. Sci Rep, 2021, 11: 21911.
Appendix
Less
Year 2026 volume 54 Issue 7
PDF
139
53
Cite this Article
BibTeX
Article Info
doi: 10.1016/S1872-5813(26)60693-7
  • Receive Date：2025-12-05
  • Online Date：2026-09-23
  • Published：2026-07-10
Article Data
Affiliations
History
  • Received：2025-12-05
  • Revised：2026-01-08
  • Accepted：2026-01-30
Affiliations
    1School of Rescue and Command, China People’s Police University, Langfang 065000, China
    2National Engineering Research Center for Fire and Emergency Rescue, Shenyang 110000, China
    3Hebei Key Laboratory of Emergency Rescue Technology, China People’s Police University, Langfang 065000, China
    4State Key Laboratory of Heavy Oil Processing, China University of Petroleum (Beijing); Beijing 102249, China

Corresponding:

Tel: 010-89739125, E-mail:
References
Share
https://castjournals.cast.org.cn/joweb/rlhxxb/EN/10.1016/S1872-5813(26)60693-7
Share to
QR

Scan QR to access full text

Cite this article
BibTeX
Citations
表12种不同金属材料的力学参数
科
Family		 属数
Number of
genus		 种数
Number of
species		 占总种数比例
Percentage of
total species (%)		 属
Genus		 种数
Number of
species		 占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
  • BibTeX
  • EndNote
  • RefWorks
  • TxT
Links
Articles
Latest Articles
Most Read
Collections
Updates
Events
News
Multimedia
About
About Us
Contact
No. 86 Xueyuan South Road, Haidian District, Beijing
100081
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn

Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House