Skip to main content
v2026.11,986 entries · CC-BY 4.0

Battery Modeling: Methods, Physics & Lab Guide

A technical guide to battery modeling, comparing equivalent circuit and electrochemical DFN approaches, HPPC parameter extraction, and lab simulation.

Written and maintained by CASRAI Editorial Board

Published

On this page

Battery modeling is the mathematical and computational representation of electrochemical, electrical, and thermal dynamics within battery cells and packs to predict performance, manage state estimates, and assess degradation. Spanning empirical equivalent circuit models for real-time control to physics-based Doyle-Fuller-Newman electrochemical formulations, modeling enables laboratories and engineers to optimize electrode architecture, track state of charge, and prevent accelerated aging without destructive cycling.

Definition and Scope of Battery Modeling

Battery modeling formalizes the multi-physics phenomena governing secondary electrochemical cells into predictive mathematical structures. At its core, a battery is a coupled, non-linear system where chemical species transport, interfacial charge transfer kinetics, solid-state diffusion, and resistive electron flow interact across microscopic particle boundaries and macroscopic cell geometries. A robust battery model transforms raw operating observables—specifically terminal cell voltage, applied current, and surface temperature—into unobservable state variables including state of charge (SOC), state of health (SOH), local overpotentials (η), and dynamic internal resistance (R0).

Within the broader battery energy storage discipline, modeling serves three fundamental functions: cell design optimization, operational control, and lifetime prognostic assessment. During material screening and cell engineering, models simulate how modifications in electrode thickness, particle size distribution, active material loading, and electrolyte conductivity impact energy density and transport limitations. In operational systems, battery management systems (BMS) execute embedded models to enforce safety boundaries, calculate state of power (SOP) limits, and maintain cell balancing. In degradation analysis, modeling separates reversible thermodynamic losses from irreversible kinetic degradation, providing quantitative insight into capacity fade and impedance growth.

The operational scope of battery modeling is defined by the tension between physical fidelity and computational overhead. High-fidelity electrochemical models resolve spatially distributed partial differential equations (PDEs) across the solid active particles and liquid electrolyte phase, capturing local lithium concentrations and reaction rates. Conversely, control-oriented equivalent circuit models (ECMs) condense these distributed physical phenomena into lumped electrical networks composed of resistors and capacitors governed by ordinary differential equations (ODEs). Choosing an appropriate modeling paradigm requires matching the mathematical architecture to the specific research question, computational budget, and available experimental characterization data.

How Battery Modeling Works in Practice: Protocols and Experimental Parameterisation

Every battery model requires empirical validation and parameter extraction derived from rigorous laboratory testing protocols. A model is only as reliable as the experimental datasets used to identify its internal parameters. In practice, establishing a calibrated model involves a structured sequence of galvanostatic pulsing, frequency-domain impedance testing, and thermal characterization under strictly controlled environmental chamber conditions.

The foundational protocol for parameterising equivalent circuit models is the Hybrid Pulse Power Characterization (HPPC) profile, established under the United States Advanced Battery Consortium (USABC) and Idaho National Laboratory electric vehicle battery testing standards (such as INL/EXT-15-34184). The HPPC protocol systematically evaluates dynamic discharge power, regenerative charge acceptance, and internal resistance across the full operating range from 100% down to 0% SOC at discrete 10% intervals. Following an extended thermal soaking and rest period to achieve thermodynamic equilibrium, the cell is subjected to a calibrated discharge current pulse (typically 1C to 5C for 10 to 30 seconds), followed by a short rest, and a corresponding regenerative charge pulse. By measuring the immediate step voltage drop (ohmic response) and the subsequent transient relaxation curve (polarization response), laboratory engineers extract the ohmic resistance and capacitive time constants characteristic of the cell.

To capture frequency-dependent interfacial phenomena, researchers utilize Electrochemical Impedance Spectroscopy (EIS). EIS applies a small-amplitude sinusoidal potential or current perturbation across a frequency spectrum ranging from high frequencies (100 kHz) to ultra-low frequencies (10 mHz or lower). The resulting Nyquist and Bode representations decouple distinct physical processes based on their characteristic relaxation times: high-frequency inductive cable and geometry effects, bulk electrolyte and separator resistance (intercept on the real axis), intermediate-frequency solid electrolyte interphase (SEI) layer and charge-transfer resistance (depressed semicircles), and low-frequency Warburg diffusion impedance within the porous solid electrodes (sloped 45-degree tail).

For physics-based models, parameter extraction demands specialized diagnostic techniques beyond standard cycling. Galvanostatic Intermittent Titration Technique (GITT) applies incremental current pulses followed by long relaxation intervals (often several hours) to determine true equilibrium open-circuit voltage (OCV) curves and solid-state lithium diffusion coefficients (Ds) as functions of stoichiometry. Parameter optimization frameworks, including open-source parameter estimation packages such as PyBOP and PyBaMM, integrate optimization algorithms—such as Covariance Matrix Adaptation Evolution Strategy (CMA-ES) or Levenberg-Marquardt non-linear least squares—to fit complex parameter sets against experimental voltage and temperature profiles while accounting for parameter identifiability and correlation constraints.

Types and Approaches: From Equivalent Circuits to Multi-Scale Physics

Battery modeling strategies are broadly categorized into four primary paradigms, spanning from empirical electrical analogies to fully coupled three-dimensional electrochemical-thermal simulations. Each paradigm represents a specific engineering compromise between physical insight, parameter complexity, and execution speed.

1. Empirical and Data-Driven Models: These formulations treat the battery as a black-box or gray-box system, relying on direct mathematical fitting, polynomial regressions, or machine learning architectures (such as Long Short-Term Memory networks or Gaussian Process Regression). While computationally trivial and effective for interpolation under standard drive cycles, empirical models fail to extrapolate under unseen operating temperatures or dynamic C-rate shifts, offering zero insight into internal physical degradation.

2. Equivalent Circuit Models (ECMs): ECMs represent the standard choice for embedded automotive and stationary storage BMS implementations. The simplest variant, the Rint model, consists solely of an ideal OCV source connected in series with an ohmic internal resistor. To account for diffusion polarization and electrochemical relaxation, the Thevenin (1-RC) and Dual Polarization (2-RC) models introduce parallel resistor-capacitor (RC) networks in series with the ohmic resistance. The first RC pair captures fast charge-transfer kinetics and electrical double-layer capacitance, while the second RC pair captures slower mass transport and concentration polarization dynamics in the electrolyte.

3. Single Particle Model (SPM) and SPMe: The Single Particle Model represents the first tier of physics-based modeling. It approximates each porous electrode as a single representative spherical active material particle, neglecting spatial concentration gradients and potential drops across the liquid electrolyte phase. The SPM solves Fickian diffusion equations for lithium concentration within the particle spheres and applies Butler-Volmer kinetics at the particle surfaces. While computationally efficient (running as coupled ODEs), the SPM breaks down at moderate to high C-rates (exceeding 1C to 2C) where electrolyte transport limitations become prominent. To resolve this boundary, the Single Particle Model with Electrolyte (SPMe) asymptotically re-incorporates first-order electrolyte concentration and potential gradients, dramatically improving accuracy under dynamic automotive loading.

4. Doyle-Fuller-Newman (DFN / P2D) Models: Based on the seminal porous electrode and concentrated solution theory established by Doyle, Fuller, and Newman in 1993, the DFN framework (also termed Pseudo-Two-Dimensional or P2D) is the benchmark standard for electrochemical modeling. The DFN model discretizes the cell across two spatial coordinates: a macro-scale dimension through the thickness of the negative electrode, separator, and positive electrode (x-axis), and a micro-scale radial dimension within the active particles at each macroscopic position (r-axis). It couples four non-linear partial differential equations: solid-phase lithium diffusion (Fick’s second law), liquid-phase lithium ion transport and diffusion, solid-phase electronic charge conservation (Ohm’s law), and liquid-phase ionic charge transport. Because it resolves local overpotentials and phase concentrations, the DFN model directly simulates degradation mechanisms such as SEI growth, transition metal dissolution, and lithium plating during fast charging.

Sourced Comparison of Battery Modeling Paradigms

The table below summarizes the core architectural differences, governing physical equations, computational burdens, and primary deployment environments across standard battery modeling methodologies, traced to published electrochemical and engineering literature.

Static algebraic polynomial fits

3 to 8 coefficients

Extremely Fast (under 0.1 ms)

Preliminary sizing, lookup-table SOC estimation

1 linear ODE + algebraic OCV

4 to 6 parameters

Real-Time Baseline (1.0x)

Production automotive BMS, microcontrollers

2 coupled linear ODEs

6 to 10 parameters

Real-Time (1.5x to 2.0x)

Hardware-in-the-loop (HIL), drive-cycle simulation

Radial 1D diffusion ODEs

15 to 22 physical parameters

Fast Physics (10x to 20x)

Low-rate cycling, fast material comparison

DAE system (solid + electrolyte)

22 to 30 physical parameters

Moderate Physics (30x to 50x)

Physics-based BMS estimators, degradation tracking

4 coupled non-linear PDEs

35 to 50 physical parameters

Computationally Heavy (over 500x)

Electrode design, fast charging, degradation physics

Model Paradigm Governing Equations Typical Parameters Relative Speed Primary Use Case
Empirical / Polynomial
1-RC Thevenin (ECM)
2-RC Dual Polarization
Single Particle Model (SPM)
SPM with Electrolyte (SPMe)
Doyle-Fuller-Newman (DFN)

In addition to cell-level formulations, testing facilities also evaluate flow battery systems and commercial grid packs where fluid mechanics and external reactant circulation introduce hydrodynamic modeling requirements distinct from enclosed intercalation cells.

Worked Example: Parameterising a 1-RC Thevenin Model from HPPC Pulse Data

To illustrate how raw cycler test logs are translated into functional model parameters, consider an illustrative composite test of a commercial 3.7 V, 5.0 Ah lithium nickel manganese cobalt oxide (NMC811) cylindrical cell undergoing an HPPC characterization pulse in a temperature chamber regulated at 25°C. The target is to extract the ohmic resistance (R0), polarization resistance (R1), and polarization capacitance (C1) at exactly 50% SOC.

The cell has been discharged to 50% SOC and rested at open-circuit for two hours. The baseline equilibrium voltage before pulse initiation is recorded as V0 = 3.820 V. At time t = 0, a constant discharge current of I = 5.0 A (1C rate) is applied for exactly 10.0 seconds.

Step 1: Extract Ohmic Resistance (R0)
High-speed data acquisition (sampled at 100 Hz) captures the instantaneous voltage step occurring within the first 10 milliseconds of current onset. The voltage drops abruptly from V0 = 3.820 V to V1 = 3.785 V. This drop represents the purely ohmic resistance of the metallic current collectors, tabs, active material particle contacts, and bulk electrolyte solution:

ΔVohmic = V0 – V1 = 3.820 V – 3.785 V = 0.035 V (35 mV)

R0 = ΔVohmic / I = 0.035 V / 5.0 A = 0.0070 Ω = 7.0 mΩ

Step 2: Extract Polarization Resistance (R1)
Over the remainder of the 10-second pulse, the voltage decays non-linearly as electrochemical charge transfer polarization and solid-state concentration gradients develop across the active material. At t = 10.0 s, immediately prior to current interruption, the measured terminal voltage reaches V2 = 3.745 V. The total transient polarization overpotential is:

ΔVpolarization = V1 – V2 = 3.785 V – 3.745 V = 0.040 V (40 mV)

Assuming the 10-second duration is sufficient for the first-order RC polarization branch to reach near steady-state equilibrium:

R1 = ΔVpolarization / I = 0.040 V / 5.0 A = 0.0080 Ω = 8.0 mΩ

Step 3: Extract Polarization Capacitance (C1) and Time Constant (τ)
Upon current interruption at t = 10.0 s, the current steps to zero. The terminal voltage immediately jumps upward by ΔVohmic = 35 mV, then undergoes an exponential relaxation back toward open-circuit equilibrium. The characteristic time constant τ = R1 × C1 represents the time required for the polarization voltage to relax by 63.2% (1 – 1/e) of its transient amplitude. In the recorded test log, the voltage recovers 63.2% of the 40 mV polarization drop (25.3 mV) in exactly 5.0 seconds. Therefore:

τ = 5.0 s

C1 = τ / R1 = 5.0 s / 0.0080 Ω = 625.0 F

Through this standardized calculation, the laboratory parameterises the cell at 50% SOC as a Thevenin equivalent circuit with R0 = 7.0 mΩ, R1 = 8.0 mΩ, and C1 = 625 F. By repeating this procedure across SOC steps from 10% to 90% and across temperatures (-10°C, 0°C, 25°C, 45°C), engineers compile multi-dimensional lookup tables that populate real-time BMS state observers.

Where Battery Modeling Matters for Laboratories, Testing Facilities, and Engineering Decisions

Battery modeling is not merely a theoretical exercise; it directly governs equipment specification, facility operations, and risk mitigation across industrial and academic research settings. In advanced testing environments, modeling guides the design of experimental matrices. Because long-term cycling to 80% retained capacity (end of life) can require 12 to 24 months of continuous testing, laboratories employ physics-based degradation models to design accelerated aging protocols. Models predict the optimal combination of elevated ambient temperatures (35°C to 45°C), high charging C-rates, and elevated upper cut-off voltages that will accelerate specific degradation pathways—such as mechanical particle cracking or SEI growth—without triggering unrepresentative failure modes like sudden lithium plating.

In facility design, accurate thermal-electrochemical modeling dictates laboratory infrastructure requirements. Large-format pouch and prismatic cells undergoing high-rate pulse testing generate substantial heat dissipation that must be actively managed. Thermal models calculate heat generation rates (coupling irreversible Joule heating, charge transfer overpotential heating, and reversible entropic heat coefficients), enabling facility engineers to dimension liquid cooling chillers, thermal chambers, and specialized dry room facilities where moisture levels are maintained below dew points of -40°C to prevent hydrolysis of hexafluorophosphate salts into hydrofluoric acid.

From a compliance and system safety perspective, modeling underpins certification against stationary energy storage standards such as UL 1973. Standards require comprehensive failure mode and effects analysis (FMEA), thermal runaway containment verification, and overcharge protection validation. Battery models allow developers to simulate propagation risks, evaluating whether thermal runaway in a single failing cell will cascade across adjacent cells within a module before expensive destructive physical testing is initiated.

Furthermore, testing laboratories utilize modeling to establish a quantitative lab equipment lifecycle assessment. By estimating channel utilization, cycler calibration drift, and environmental chamber duty cycles against simulation needs, lab managers optimize capital investment and ensure instrumentation accuracy throughout multi-year testing campaigns.

Common Misunderstandings in Battery Modeling

Several persistent technical misconceptions recur in battery modeling literature and testing facilities, often resulting in inaccurate state estimation, unexpected thermal excursions, or premature cell failure:

1. Assuming Equivalent Circuit Parameters Are Invariant Across Temperature: A frequent mistake in basic BMS development is fitting an ECM at 25°C and applying those parameters across broader operating conditions. Electrochemical reaction rates and ion mobility follow Arrhenius temperature dependencies. Ohmic resistance (R0) and charge transfer resistance (R1) can increase by an order of magnitude when operating temperatures drop from 25°C to -10°C, leading to severe voltage undershoots and premature cut-off tripping if the model lacks dynamic Arrhenius scaling.

2. Treating Open-Circuit Voltage as a Static Single Curve: Many models implement OCV as a single lookup table indexed solely by SOC. In reality, commercial lithium-ion chemistries—especially lithium iron phosphate (LFP) and high-silicon anodes—exhibit significant thermodynamic voltage hysteresis. The equilibrium OCV during charge is systematically higher than during discharge at identical SOC levels due to mechanical microscopic stress and multi-phase transition pathways. Neglecting hysteresis introduces SOC estimation errors exceeding 10% to 15% in flat-plateau regions.

3. Using the Single Particle Model for Fast Charging Analysis: Because the SPM is computationally rapid, researchers frequently attempt to simulate fast charging (exceeding 2C or 3C). However, the SPM fundamentally assumes uniform electrolyte concentration and infinite electrolyte conductivity. Under high charging currents, steep salt concentration gradients develop across the liquid electrolyte phase, driving significant liquid-phase diffusion overpotentials and local negative electrode polarization. Using an SPM under fast charging obscures the exact physical threshold where negative electrode potential drops below 0 V vs. Li/Li+, completely missing the onset of dangerous metallic lithium plating.

4. Overparameterising Complex Models Without Identifiability Checks: The Doyle-Fuller-Newman model incorporates over 35 physicochemical parameters, including solid-state diffusion coefficients, reaction rate constants, transfer coefficients, particle radii, and Bruggeman tortuosity exponents. Attempting to fit all 35 parameters simultaneously from simple terminal voltage data results in ill-conditioned mathematical optimization where distinct parameter combinations yield virtually identical terminal voltage profiles. Without decoupled material-level characterization (e.g., half-cell GITT, four-point probe conductivity, mercury intrusion porosimetry), full-cell parameter fitting produces unphysical numerical artifacts.

5. Confusing Reversible Capacity Loss With Irreversible Degradation: Degradation models must rigorously differentiate between true active material loss (loss of active material / LAM, loss of lithium inventory / LLI) and apparent capacity loss caused by increased cell impedance. As internal resistance grows over cycling, the cell reaches its terminal voltage cut-off earlier during constant-current discharge, leaving usable lithium unextracted. High-fidelity modeling employs differential voltage analysis (dV/dQ) and incremental capacity analysis (dQ/dV) to separate kinetic polarization truncation from thermodynamic active mass degradation.

Frequently asked questions

What is the difference between an Equivalent Circuit Model and a physics-based electrochemical model?

An Equivalent Circuit Model (ECM) approximates cell behavior using electrical components such as resistors, capacitors, and open-circuit voltage sources. It captures external voltage-current dynamics with minimal computational burden, making it ideal for real-time microcontrollers. In contrast, a physics-based electrochemical model (such as the Doyle-Fuller-Newman model) solves coupled partial differential equations governing species transport, solid-state diffusion, and electrochemical reaction kinetics. Physics-based models require extensive parameterisation and higher computational power, but they reveal internal physical variables like local overpotentials, lithium plating risks, and degradation mechanisms.

How many RC branches are needed for an accurate real-time ECM?

For the majority of automotive and stationary battery management systems, a 2-RC Dual Polarization model provides the optimal balance of accuracy and computational efficiency. The first RC branch (with a time constant typically between 0.5 and 5 seconds) captures rapid charge-transfer kinetics and double-layer capacitance, while the second RC branch (with a time constant between 10 and 100 seconds) captures slower electrolyte diffusion and concentration polarization. Adding a third RC branch increases computational complexity and parameter estimation noise without yielding statistically significant improvements in state of charge estimation accuracy.

When does the Single Particle Model fail in battery simulations?

The Single Particle Model (SPM) assumes infinite ionic conductivity and uniform salt concentration in the liquid electrolyte, solving diffusion only within representative solid active particles. Consequently, the SPM fails under high C-rate operation (typically exceeding 1C to 2C), thick electrode architectures, or low-temperature environments where electrolyte transport limitations dominate. Under these conditions, the Single Particle Model with Electrolyte (SPMe) or the full Doyle-Fuller-Newman (DFN) model must be used to prevent substantial voltage prediction errors.

What laboratory tests are required to extract parameters for a DFN model?

Parameterising a DFN model requires comprehensive physical, chemical, and electrochemical testing across cell components. Key tests include: geometric and physical measurements (electrode thickness, coating loading, active material density, and porosity via mercury intrusion porosimetry); half-cell GITT and slow C-rate cycling (C/50 or C/100) on harvested or fabricated single electrodes to measure equilibrium open-circuit potential curves; Electrochemical Impedance Spectroscopy (EIS) across temperatures to determine charge transfer kinetics; and liquid electrolyte characterization to quantify ionic conductivity, diffusion coefficients, and transference numbers across salt concentrations.

How does battery modeling account for cell degradation and capacity fade?

Degradation modeling incorporates specific kinetic and mechanical damage equations into the governing cell equations. The most common formulations model Solid Electrolyte Interphase (SEI) growth as a continuous side reaction consuming cyclable lithium inventory via solvent diffusion through the growing layer. Advanced lifetime models also incorporate particle mechanical stress equations to track fatigue cracking, transition metal dissolution kinetics, and kinetic overpotential conditions that trigger metallic lithium plating during fast charging.

Can battery models predict thermal runaway in energy storage systems?

Yes, coupled multi-physics models predict thermal runaway initiation and propagation by incorporating exothermic decomposition reaction kinetics into three-dimensional energy balance equations. These models simulate the successive stages of thermal runaway: SEI layer breakdown (80°C to 120°C), anode-electrolyte reaction, separator melting, cathode active material decomposition with oxygen release (above 180°C to 220°C), and violent electrolyte combustion. Modeling allows engineers to design cell spacing, thermal barriers, and phase-change materials to prevent single-cell failures from cascading across entire battery modules.

Sources

  • Doyle, M., Fuller, T. F., and Newman, J. (1993). Modeling of Galvanostatic Charge and Discharge of the Lithium/Polymer/Insertion Cell. Journal of The Electrochemical Society, 140(6), 1526–1533. https://doi.org/10.1149/1.2221597 (accessed 2026-11-16).
  • Idaho National Laboratory (INL). (2015). Battery Test Manual for Electric Vehicles (Revision 3, INL/EXT-15-34184). U.S. Department of Energy. https://inldigitallibrary.inl.gov/sites/sti/sti/6847846.pdf (accessed 2026-11-16).
  • Marquis, S. G., Sulzer, V., Please, C. P., and Chapman, S. J. (2019). An Asymptotic Derivation of a Popular Electrochemical Model for Lithium-Ion Batteries. Journal of The Electrochemical Society, 166(15), A3693–A3706. https://doi.org/10.1149/2.0341915jes (accessed 2026-11-16).
  • National Renewable Energy Laboratory (NREL). (2024). Battery Lifetime Analysis and Simulation Tool (BLAST) Suite. NREL Transportation and Hydrogen Systems Center. https://www.nrel.gov/transportation/blast.html (accessed 2026-11-16).
  • Sulzer, V., Marquis, S. G., Timms, R., Please, C. P., and Chapman, S. J. (2021). Python Battery Mathematical Modelling (PyBaMM). Journal of Open Research Software, 9(1), 14. https://doi.org/10.5334/jors.309 (accessed 2026-11-16).
  • International Electrotechnical Commission (IEC). (2018). Secondary lithium-ion cells for the propulsion of electric road vehicles – Part 1: Performance testing (IEC 62660-1:2018). https://webstore.iec.ch/publication/34293 (accessed 2026-11-16).

Data: Comparative Profile of Battery Modeling Paradigms

A quantitative compilation evaluating computational overhead, typical parameter counts, mathematical formulation class, and spatial resolution across foundational battery modeling methodologies, derived from US Department of Energy battery testing manuals and electrochemical literature.

Period covered: 1993/2026. Coverage: United States and International.

Comparative Profile of Battery Modeling Paradigms
Model ParadigmMathematical ClassParameter CountRelative Execution Time (normalized)Spatial ResolutionPrimary Application
Rint Internal ResistanceZero-state algebraic equation20.2Lumped 0D with no spatial gradientPreliminary cell sizing and rapid EV power loss estimation
1-RC Thevenin ECMFirst-order linear ODE41Lumped 0D with single time constantProduction automotive BMS microcontrollers and SOC estimation
2-RC Dual PolarizationSecond-order coupled linear ODEs61.8Lumped 0D with dual time constantsHardware-in-the-loop simulation and dynamic drive-cycle analysis
Single Particle Model (SPM)Coupled 1D solid diffusion ODEs18121D radial particle with lumped electrolytePhysics-based microcontroller estimators and low-rate cycling
SPM with Electrolyte (SPMe)Asymptotically coupled DAE system24351D radial particle plus 1D electrolyteFast-charge state estimation and cell degradation tracking
Doyle-Fuller-Newman (DFN)Coupled non-linear PDE-DAE system38850Pseudo-2D across electrode thickness and radiusElectrode design, fast charging limits, and degradation physics

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · full access

Ask about Battery Modeling: Methods, Physics & Lab Guide

Full access: large monthly usage, up to 10 source passages behind each answer, and MCP and API access for Claude Code, Codex and other CLIs. Cancel any time.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →