Quantifying Cell Degradation Using Thermally Corrected Differential Capacity Spectra

Signal

Differential Capacity Derivation and Voltage Resolution Limits

Low-rate C/20 cycling yields continuous potential profiles that produce sharp electrochemical markers when differentiated against terminal voltage. Differentiating these curves turns flat open-circuit voltage plateaus into distinct dQ/dV peaks corresponding to phase transitions in cathode and anode host structures. Microvolt-level resolution isolates precise stoichiometric phase boundaries, from stage 4 down to stage 1 graphite intercalation, as well as nickel-rich cathode phase changes like the H1-H2 and H2-H3 transitions in NMC-811 formulations.

Low C-rate measurements isolate thermodynamic behavior by minimizing ohmic IR drop across cell terminals. Testing channels equipped with 24-bit analog-to-digital converters capture voltage steps under 50 microvolts, preserving peak shape during numerical differentiation. Broadening or flattening in these peaks indicates rising kinetic impedance, whereas a shift along the voltage axis points to shifting electrode stoichiometry or growing internal polarization.

High-resolution hardware detects these subtle shifts long before overall capacity fade appears on standard amp-hour integration logs.

Without filtering, voltage noise destroys the dQ/dV derivative: calculating dQ over tiny dV increments turns single-bit converter noise into large high-frequency spikes that obscure phase signatures. Moving-distance voltage interval methods or standard Savitzky-Golay filters smooth the raw data while maintaining peak heights and centroid positions without shifting integrated peak areas. Selecting the proper numerical window width preserves subtle low-amplitude features, such as the initial solid electrolyte interphase reconstruction signature near 3.55 volts in blended graphite-silicon anodes.

Differential capacity peaks degrade into unquantifiable signal noise when voltage acquisition resolution falls below sixteen bits during low C-rate cycling.

Thermal Distortion Mechanisms during Low-Rate Spectrum Acquisition

Temperature variations across the test environment distort differential capacity spectra by altering reaction kinetics and shifting phase transition potentials. Room temperature drifts of just 3 degrees Celsius produce false peaks and artificial peak splitting in dQ/dV plots, making HVAC cycling look like structural phase transformations in the active materials. Depending on state of charge, open-circuit voltage temperature coefficients range from -0.4 millivolts to +0.3 millivolts per kelvin, causing potential shifts that mask actual degradation.

Internal heat generated during low C-rate cycling creates dynamic thermal gradients between the core and casing of commercial cylindrical and pouch cells. Even at C/20, a 21700 cell with an internal resistance of 45 milliohms generates enough localized Joule heat to raise core temperatures above surface thermocouple readings. This internal lag alters local lithium-ion diffusion rates through the electrolyte and particle boundaries, shifting measured dQ/dV peaks toward lower potentials on discharge.

Thermal drift routinely throws off automated peak-tracking algorithms during batch diagnostics. Chamber temperature fluctuations shift peak centroids and trigger false structural degradation signals, making temporary thermal shifts look like permanent active material loss. Isolating real electrochemical fade requires testing in environmental chambers stabilized within plus or minus 0.1 degrees Celsius, or post-processing raw voltage data into isothermal differential capacity spectra using simultaneous core-temperature logs.

What specific baseline signal filtering threshold balances high-frequency noise rejection against phase transition peak preservation in high-capacity pouch cells?

Shift

Entropic Voltage Drift and Open-Circuit Potential Coefficients

Electrochemical potential inside a lithium-ion cell varies with temperature according to the Gibbs-Helmholtz relation, where the entropic temperature coefficient dE/dT dictates open-circuit potential drift. This entropic coefficient changes sign and magnitude across the state of charge, reflecting order and disorder within the insertion host lattices. In a graphite anode, dE/dT shifts from positive values during stage 1 intercalation to negative values in diluted stage 3 structures, driving non-linear voltage movement during ambient temperature swings.

Cathode chemistry dictates how severely temperature alters voltage measurements. Lithium iron phosphate exhibits minimal entropic drift across its central flat plateau, remaining steady under minor thermal shifts, whereas high-nickel layered oxides show pronounced dE/dT swings exceeding 0.2 millivolts per kelvin near full charge. These entropic offsets shift differential capacity peaks along the potential axis regardless of actual degradation.

Temperature shifts distort dQ/dV peak alignment when comparing cycles over long test windows. A 5 kelvin bump during baseline testing shifts the primary graphite stage 2 peak by up to 1.5 millivolts ~ roughly the same offset caused by 50 deep cycles of solid electrolyte interphase growth. Without precise thermal correction across the full state of charge spectrum, diagnostic software routinely mistakes thermal offsets for structural degradation in the graphite anode.

Ambient temperature variation of five degrees Celsius shifts differential capacity peak centroids by an amount equivalent to fifty deep degradation cycles.

Arrhenius Kinetics and Charge-Transfer Impedance Masking

Charge-transfer resistance at the electrode-electrolyte interface follows Arrhenius temperature dependence, changing kinetic voltage losses during current flow. When cell temperature drops, charge-transfer resistance increases exponentially. The resulting overpotential pushes discharge dQ/dV peaks down in voltage and charge peaks up.

This symmetric peak separation mimics the impedance growth caused by a thickening solid electrolyte interphase layer, obscuring real physical degradation when uncorrected.

Lower temperatures also increase electrolyte viscosity, suppressing ionic conductivity within the liquid-filled pores of thick calendered electrodes. Reduced bulk conductivity creates steep salt gradients across the electrode thickness, broadening differential capacity peaks and lowering their maximum amplitudes. Without correction, analysts evaluating low-temperature dQ/dV curves easily mistake kinetic broadening for active material loss or uneven current distribution in the jelly-roll.

Solid-state diffusion inside active material particles slows significantly in the cold, restricting lithium-ion transport within host crystal structures. Slower intra-particle diffusion delays electrochemical equilibrium, shifting peak positions well beyond thermodynamic open-circuit locations even at C/20. Disentangling differential capacity measurements requires decoupled kinetic algorithms that separate temperature-dependent Arrhenius impedance growth from structural lattice damage.

A drop in ambient temperature doubles charge transfer resistance for every eight degrees Celsius decrease, masking structural active material loss behind heavy kinetic polarization.

Filter

Mathematical Transformation and Temperature-Compensated Spectra

Extracting thermally corrected differential capacity requires simultaneous recording of terminal voltage, current, casing temperature, and calculated core temperature. Converting raw, temperature-distorted data into isothermal spectra works by subtracting entropic voltage variations and kinetic overpotential drops from measured terminal voltage. Transformation models integrate the entropic coefficient array dE(SOC)/dT to reset measured voltages to a standard reference temperature, typically 25 degrees Celsius.

Dynamic thermal correction models isolate kinetic overpotential by continuously recalculating internal resistance against instantaneous temperature and state of charge. The primary transformation adjusts measured terminal voltage V_meas to reference voltage V_ref through explicit temperature compensation:

V_ref = V_meas – (T_cell – T_ref) (dE/dT) – I R_int(T_cell, SOC)

Applying this transformation eliminates pseudo-peaks caused by ambient temperature swings during multi-hour C/20 test runs. Running this correction converts skewed dQ/dV profiles into aligned isothermal spectra suitable for peak area integration and automated curve fitting.

How do temperature fluctuations alter kinetic differential capacity peaks?

Temperature fluctuations shift peak centroids along the voltage axis, broaden peak widths, and distort calculated peak areas through variable overpotentials. As cell temperature changes during low C-rate cycling, internal resistance shifts according to Arrhenius kinetics while open-circuit potential moves along the entropic coefficient curve. These combined shifts alter local overpotential, compressing or stretching discharge peaks along the voltage axis depending on whether ambient temperatures rise or fall during phase transitions.

Temperature shifts during phase transitions alter lithium-ion charge transfer rates and solid-state diffusion speeds within active material particles. A temporary temperature rise reduces local overpotential, causing terminal voltage to linger near a phase transition plateau and creating an artificially tall dQ/dV peak. Conversely, a sudden temperature drop increases overpotential, accelerating voltage drop and flattening the peak.

These kinetic artifacts obscure real degradation, making thermal noise look like active material loss or lithium inventory depletion.

Running an isothermal correction workflow eliminates these thermal artifacts by normalizing raw voltage profiles against continuous thermal and state-of-charge measurements. The execution routine follows five steps:

1. Log continuous cell voltage, current, surface temperature, and ambient temperature at high sample rates during C/20 charge and discharge cycles.
2. Calculate internal core temperature using a two-state lumped thermal network model driven by surface temperature readings and instantaneous Joule heating.
3. Look up instantaneous entropic coefficients dE/dT and charge-transfer resistance values from a calibrated baseline matrix mapped across state of charge and temperature.
4. Subtract entropic potential offsets and kinetic overpotential losses from raw voltage profiles to compute isothermal voltage vectors referenced to 25 degrees Celsius.
5. Differentiate isothermal capacity data against isothermal voltage vectors using a voltage-distance adaptive filtering algorithm to produce thermally compensated dQ/dV spectra.

Applying this dynamic thermal compensation framework across 3.5 Ah NMC-811 pouch cells exposed to ambient temperature swings between 15 and 25 degrees Celsius during C/20 discharge cycles clarifies raw spectral signals. Raw dQ/dV processing showed erratic peak splitting near 3.65 volts and an uncorrected primary peak area loss of 14.2 percent. Applying isothermal transformation eliminated false peak splitting entirely and reduced measured peak area loss to 3.8 percent, matching teardown measurements of active lithium inventory loss.

Warranty claims built on uncompensated differential capacity spectra fail legal validation when baseline temperature drift exceeds the standard laboratory tolerance defined in IEC 62619.

Spline Interpolation and Adaptive Filtering Algorithms

Unequal voltage spacing from high-speed digital sampling requires adaptive spline interpolation before computing numerical derivatives. Standard discrete differentiation yields erratic dQ/dV values when dV step sizes fluctuate from converter quantization. Fitting raw voltage-capacity pairs with monotonic cubic Hermite splines establishes continuous, differentiable functions that generate smooth differential spectra without introducing artificial peak distortions.

Effective filtering algorithms use fixed voltage steps dV rather than uniform time intervals dt during differential calculations. Differentiating on uniform time intervals creates excessive point density along flat OCV plateaus where voltage moves slowly, while undersampling steep potential drops between phase transitions. Resampling at constant voltage steps maintains a uniform signal-to-noise ratio across the entire state of charge, giving low-amplitude high-voltage cathode transitions the same resolution as dominant low-voltage anode peaks.

Edge effects near charge and discharge cut-offs ruin end-point peak detection unless smoothing algorithms are properly bounded. Polynomial filtering windows extending past terminal voltage limits create false slope reversals, generating synthetic dQ/dV peaks near 2.8 volts and 4.2 volts. Truncating filtering windows or applying asymmetrical endpoint weights prevents false diagnostic peaks near maximum and minimum states of charge.

Applying static linear filtering to non-isothermal voltage profiles distorts peak area integration, resulting in false state-of-health assessments that invalidate battery asset valuations.

Audit

Decoupling Loss of Lithium Inventory from Active Material Degraded States

Thermally corrected dQ/dV spectra allow mathematical separation of primary aging mechanisms, distinguishing loss of lithium inventory from loss of active negative or positive material. Lithium inventory loss shows up as a uniform horizontal shift of anode capacity features relative to cathode features along the cell voltage axis. Isolating this peak movement quantifies cyclable lithium without cell teardown or coin-cell reconstruction.

Loss of active negative material shrinks the total envelope of graphite-specific peak features, particularly the low-potential stage 1 and stage 2 intercalation peaks. When graphite degrades through particle cracking, isolation, or exfoliation, corresponding dQ/dV peak areas contract proportionally. Comparing normalized anode peak areas against initial baseline spectra isolates negative electrode damage independently of SEI lithium consumption.

Positive electrode degradation alters high-voltage spectral features, lowering cathode phase transition peak amplitudes and broadening peak widths. High-nickel cathodes experience microcracking along grain boundaries during H2-H3 phase transformations at high states of charge, causing rapid attenuation of dQ/dV peak areas above 4.0 volts. Quantifying individual peak area loss maps structural damage directly to specific degradation modes within the positive electrode matrix.

Uncorrected differential capacity profiles lead to severe analytical errors when diagnosing field-returned cells exposed to varying temperatures. Common analytical breakdowns include:

• False Lithium Loss Identification when temperature-induced entropic shifts move anode peak positions, causing diagnostic software to record false lithium inventory reduction.
• Misdiagnosed Active Material Loss from kinetic peak broadening under cold testing conditions, mistaking low diffusion rates for physical electrode isolation.
• Unregistered Cathode Degradation caused by high ambient temperatures sharpening high-voltage dQ/dV peaks, masking real intragranular microcracking and structural capacity loss.
• Corrupted Impedance Metrics arising from fluctuating internal resistance during non-isothermal cycling, rendering calculated peak broadening useless for state-of-health tracking.
• Flawed Remaining Useful Life Projections caused by integrating thermally distorted peak areas into baseline degradation models, leading to unwarranted pack replacements.

Feeding non-isothermal dQ/dV curves into diagnostic models distorts degradation tracking, leading pack integrators to reject healthy modules or keep severely degraded assets in service past their real operating limits.

Cell manufacturers frequently attribute field-returned capacity fade to customer thermal abuse rather than manufacturing defects when uncorrected dQ/dV peaks show generalized impedance growth.

Margin

Commercial Impact on Cell Qualification and Warranty Validation

Automated degradation tracking using thermally corrected dQ/dV spectra redefines cell qualification during incoming shipment audits. Sourcing teams use high-precision differential capacity analysis to verify whether supplier life-cycle data reflects genuine cell durability or artificial life flattery achieved in low-stress lab environments. Thermally corrected audits isolate early-life SEI growth rates within the first 50 cycles, predicting long-term capacity retention long before standard 1,000-cycle aging tests finish.

Validating cell warranty claims depends on proving specific degradation mechanisms under defined operating conditions. Energy storage warranties exclude capacity loss caused by thermal abuse, placing the burden of proof on pack operators. Thermally compensated dQ/dV analysis provides legally defensible evidence showing whether observed fade stems from manufacturing defects like cathode cracking or operational abuse like sustained high-temperature running.

Second-life battery valuation relies on clear non-destructive diagnostic records to price used modules taken from electric vehicles. Storage integrators buying secondary packs need verifiable degradation signatures confirming that incoming modules retain structural active material integrity. Isothermal differential capacity spectra form the core of digital battery passports required under EU Battery Regulation, establishing historical ratios of active material loss versus lithium inventory consumption.

Assessing cell degradation with thermally corrected dQ/dV spectra requires buyers and cell manufacturers to agree contractually on baseline testing parameters. Sourcing teams enforce standardized auditing terms in supply contracts to ensure fair warranty enforcement and transparent batch acceptance:

• Isothermal Reference Temperature establishing 25.0 degrees Celsius plus or minus 0.2 degrees Celsius as the mandatory mathematical reference state for all spectrum processing.
• Data Acquisition Resolution specifying a minimum 24-bit voltage logging channel requirement with absolute accuracy within plus or minus 100 microvolts across full range.
• Mandatory Filtering Protocol defining Savitzky-Golay cubic polynomial filtering with a fixed 5-millivolt voltage step window to standardize peak height calculation.
• Degradation Mode Allocation Thresholds setting explicit contract limits for lithium inventory loss rates versus cathode structural degradation rates during warranty periods.

Supply agreements operating under UN 38.3 transport guidelines incorporate thermally corrected dQ/dV degradation criteria to define lot rejection rights when incoming cell batches show premature active material loss during initial quality control testing.