Thermal Desorption Spectroscopy¶

A programmed-temperature surface-analysis method that converts the release of adsorbed species into mass-resolved desorption-rate curves, using peak position, shape, area, and coverage dependence to infer surface populations and desorption kinetics.

Version: v1 · 2026-08-30 · History
Domain-specific #: 2964
Origin domain: surface science
Subdomain: adsorption and desorption kinetics
Aliases: Temperature-programmed desorption, TPD, Thermal desorption spectrometry, TDS

Core Idea¶

Thermal Desorption Spectroscopy (TDS), usually called Temperature-Programmed Desorption (TPD), measures how an adsorbed population leaves a surface while the surface follows a controlled temperature program. The experiment prepares an adsorbate on a substrate, heats that substrate—commonly at an approximately constant rate—and records the flux or partial pressure of released species as a function of temperature or time. A mass spectrometer often separates the outgoing products by mass-to-charge ratio. The result is not merely a list of temperatures: it is a family of desorption-rate curves whose peak position, shape, integrated area, mass channel, and response to starting coverage jointly constrain what occupied the surface and how it was bound.^[1]

The locked identity is prepared adsorbate–surface population + declared temperature ramp + thermally activated release + species-sensitive detection -> desorption signal versus temperature/time -> inference about populations, states, and kinetics. Its theoretical center is the Polanyi–Wigner rate law, often written r = -dθ/dt = ν(θ) θ^n exp[-E_des(θ)/(RT)], together with the heating program β = dT/dt. Coverage θ, kinetic order n, activation energy E_des, prefactor ν, heating rate, pumping, readsorption, and detector response can all affect the curve. Consequently, a peak temperature is not a context-free binding-energy meter. Inference requires a declared kinetic model, suitable calibration, or richer analyses across coverages and heating rates.^[2]

TDS deserves a domain-specific node because the existing Measurement prime does not entail programmed thermal perturbation of an adsorbed population, mass-resolved escape detection, or the associated kinetic inverse problem. Adsorption describes population of a surface and an adsorption isotherm describes equilibrium loading at fixed temperature; TDS instead follows a deliberately driven, non-equilibrium release trajectory.

Structural Signature¶

a prepared substrate — a solid surface whose material, orientation, cleanliness, area, and defect structure are controlled or characterized;
an adsorbed initial population — one or more surface species with a declared dose, exposure history, and starting coverage;
distinguishable binding states — terraces, steps, defects, phases, chemical forms, or interaction environments that may yield different release behavior;
a temperature program — temperature as a measured function of time, often a linear ramp with heating rate β;
thermally activated desorption — surface species cross a kinetic barrier and enter the gas phase;
a transport and vacuum environment — pumping speed, chamber volume, background pressure, and geometry mediate the relation between emitted flux and detected pressure;
species-sensitive detection — typically a quadrupole mass spectrometer monitoring selected mass-to-charge channels, but other calibrated detectors may instantiate the role;
a desorption trace — signal intensity plotted against temperature or time for each monitored channel;
peak location and shape — observables influenced by barrier, prefactor, kinetic order, coverage, lateral interactions, ramp, and instrumental response;
integrated signal — after calibration and fragmentation correction, a proxy for the amount released;
a kinetic model — the Polanyi–Wigner relation or an explicitly more complex reaction network connecting surface state to release;
coverage series or ramp series — repeated traces used to distinguish order, interaction, and parameter effects;
background and blank controls — measurements that separate sample desorption from chamber, holder, cracking-pattern, and contamination signals;
an inverse inference — attribution of peaks to species or states and estimation of kinetic quantities with stated assumptions;
a validity envelope — regimes in which readsorption, heat gradients, mass-transfer limitation, overlapping channels, or detector saturation do not dominate.

All essential roles must compose. Heating a bulk specimen and observing its mass loss without resolving a surface population is not automatically TDS, nor is measuring an equilibrium adsorption curve at several fixed temperatures.

What It Is Not¶

Not desorption itself. Desorption is the physical event; TDS is a controlled measurement protocol for observing it.
Not adsorption equilibrium. An isotherm relates equilibrium loading to pressure or concentration at fixed temperature, whereas TDS drives the system through temperature and records release.
Not thermogravimetric analysis by default. TGA measures total sample mass change and need not identify desorbing species or isolate surface kinetics.
Not differential scanning calorimetry. Heat-flow anomalies do not directly constitute a mass-resolved desorption flux.
Not a single peak-to-energy lookup. Redhead-style estimates depend on assumed order, prefactor, linear heating, and negligible readsorption.
Not intrinsically ultra-high-vacuum-only. UHV single-crystal TPD is canonical, but the structural method can be instantiated with other controlled surfaces and detection arrangements if transport effects are accounted for.
Not proof of a unique adsorption site. Different kinetic models can generate similar curves, and one state may contribute to more than one product channel.
Not ordinary mass spectrometry. The discriminating intervention is a controlled temperature program applied to an adsorbate–surface system.

Scope of Application¶

TDS is central to surface science, heterogeneous catalysis, corrosion and semiconductor studies, hydrogen storage, thin films, and characterization of porous or reactive materials. On a clean single crystal, a coverage series may separate terrace and step adsorption states. In catalysis, product channels can reveal whether an adsorbed precursor desorbs intact or reacts before release. In materials work, water, solvents, hydrogen, carbon monoxide, or decomposition products can be followed during heating.

The method spans simple first-order molecular desorption, second-order recombinative desorption, zero-order multilayer loss, coverage-dependent barriers, interacting adsorbates, and coupled surface-reaction networks. These are variants of the analysis problem, not interchangeable assumptions. For a first-order process with a coverage-independent barrier and prefactor, peak temperature is approximately independent of initial coverage. Recombinative second-order behavior commonly shifts with coverage. Zero-order multilayer desorption can show a shared leading edge and coverage-dependent termination. Those signatures are diagnostic clues rather than proofs because distributions of sites, lateral interactions, and instrumental broadening can imitate them.^[3]

Common analysis families include Redhead peak analysis, leading-edge analysis, complete analysis across coverages, inversion or simulation of the full trace, and global fitting of reaction networks. The appropriate choice depends on whether kinetic order, prefactor, energy distribution, and reaction coupling are known. Full-curve methods carry more information but also expose identifiability and objective-function choices.^[4]

Clarity¶

The word “spectroscopy” can mislead: the primary independent coordinate is programmed temperature, not photon frequency, and the detector frequently measures partial pressure by mass channel. “Thermal desorption spectrometry” emphasizes the measurement, while “temperature-programmed desorption” emphasizes the intervention. Within surface science these labels name the same structural family, although particular laboratories may reserve them for narrower apparatus configurations.

Peak temperature is an emergent observable. Increasing the barrier tends to move release to higher temperature, but changing the prefactor, ramp rate, order, starting coverage, lateral interactions, or readsorption can also move or distort a peak. Peak area tracks amount only after detector sensitivity, fragmentation patterns, pumping response, transmission, and baseline are handled. A mass channel identifies an ion fragment, not necessarily a unique neutral parent molecule.

The surface temperature must be the temperature relevant to the adsorbate. Thermocouple placement, radiative gradients, current heating, thermal lag, and calibration can separate the reported temperature from the active surface. Likewise, the measured chamber pressure is a convolution of emission, flight geometry, pumping, and detector response. Reference-grade interpretation makes both mappings explicit.

Manages Complexity¶

A surface may contain many species and binding environments whose populations cannot be directly enumerated during reaction. TDS compresses that hidden state into reproducible, species-indexed release curves under a known perturbation. Repeating the experiment across doses, heating rates, isotopes, surface preparations, or coadsorbates turns a one-dimensional trace into a comparative constraint system.

This compression is powerful precisely because it is incomplete. Multiple states can overlap, and a kinetic compensation between E_des and ν can make different parameter pairs fit the same data. The abstraction manages that ambiguity by requiring model assumptions, controlled variations, and residual checks to accompany numerical parameter claims.

Abstract Reasoning¶

If two peaks occupy different temperature ranges under otherwise matched conditions, at least one kinetic or binding-state parameter differs, but peak separation alone does not identify which one.
If a mass channel appears, the released neutral must be inferred through its ionization fragmentation pattern; a channel is not an unambiguous molecule label.
If the heating rate increases, a kinetic peak generally shifts because less time is available at each temperature; comparisons require β to be known.
If the integrated calibrated signal doubles while response is linear and products are conserved, the released amount doubles even if peak shape changes.
If readsorption is significant, chamber pressure no longer directly tracks the one-way surface escape rate.
If starting coverage changes the peak position, order, lateral interactions, site filling, or coverage-dependent kinetic parameters may be implicated.
If two parameter sets reproduce one trace, data at other coverages or ramp rates can break—or reveal—the identifiability failure.
If a surface reaction precedes desorption, the measured product peak can be controlled by reaction kinetics rather than the product’s own desorption barrier.
If temperature calibration has a systematic offset, derived activation energies inherit a systematic bias.
If a proposed model fits peaks but not leading edges or integrated mass balance, it has not explained the full measurement.

Knowledge Transfer¶

The exact abstraction transfers among single-crystal UHV studies, catalyst surfaces, semiconductor processing, thin films, and adsorbate-storage materials when a controlled temperature program releases a prepared population and a detector records species-resolved flux. Apparatus and pressure regime may change, but the roles remain recognizable.

Temperature-programmed reaction and reduction share the programmed-perturbation architecture, yet add chemical reaction or oxidation-state transformation as the target process. Pyrolysis mass spectrometry and evolved-gas analysis may overlap operationally but need not begin with a surface-bound population. The transferable prime-level core is Measurement: instrument and procedure map a target property onto a trace with calibration and uncertainty. TDS’s surface-kinetic accent should not be promoted to a prime.

Examples¶

CO on a stepped metal: separate peaks vary with defect density, supporting assignment to terrace and step environments;
hydrogen recombination: adsorbed atoms combine before molecular hydrogen desorbs, producing second-order behavior under a simplified model;
water multilayer: a thick condensed layer shows a zero-order-like leading edge until the layer is exhausted;
coverage series: shifting and broadening peaks reveal lateral attraction, repulsion, or heterogeneous energy distributions;
heating-rate series: moving peak temperatures constrain kinetic parameters beyond one Redhead estimate;
isotope labeling: mass channels separate products and test a proposed surface-reaction pathway;
blank holder run: a peak persisting without the prepared substrate is assigned to apparatus background;
failure—uncorrected cracking: two mass channels attributed to two species are actually fragments of one desorbing molecule;
failure—thermal lag: the reported thermocouple temperature differs from surface temperature, distorting the inferred barrier;
non-example—adsorption isotherm: equilibrium uptake is plotted against pressure at constant temperature.

Structural Tensions¶

energy inference vs. prefactor assumption — peak location constrains combinations of kinetic parameters more directly than one unique barrier;
species resolution vs. fragmentation overlap — mass analysis distinguishes products while ionization can make their signals non-unique;
controlled heating vs. thermal gradients — faster ramps improve throughput but magnify lag and spatial nonuniformity;
vacuum sensitivity vs. realistic conditions — clean UHV measurements isolate surface states while application environments introduce transport and coadsorbates;
simple model vs. heterogeneous surface — tractable kinetic orders compress distributions of sites and interactions;
signal strength vs. perturbation — higher coverage improves detection but can change the very kinetics being inferred;
full-curve information vs. inverse non-identifiability — richer fits use more data while allowing more compensating parameters.

Structural–Framed Character¶

Thermal Desorption Spectroscopy is predominantly structural. Its identity follows from a programmed intervention, an evolving surface population, a species-sensitive measurement chain, and an inverse kinetic relation. Laboratory conventions matter to reliability but do not supply the abstraction’s ontology.

Structural Core vs. Domain Accent¶

The structural core is controlled perturbation + instrumented response trace -> inference about hidden state and transition rates. The domain accent is adsorption on physical surfaces, thermally activated escape, Polanyi–Wigner kinetics, vacuum transport, and mass-resolved detection. Removing that accent leaves Measurement or system identification, not TDS.

Measurement — the apparatus and protocol map evolving desorption flux onto calibrated temperature-indexed signals.
Activation Energy — kinetic barriers shape the temperature dependence of release.
Arrhenius Equation — exponential thermal dependence underlies the standard rate law.
Inverse Problem — observed traces constrain hidden populations and kinetic parameters non-uniquely.
Signal and Noise — background, fragmentation, pumping, and detector response condition inference.

The minimal prospective DAG placement is strict subsumption under prime:measurement.

Relationships to Other Abstractions¶

Current abstraction Thermal Desorption Spectroscopy Domain-specific

Parents (1) — more general patterns this builds on

Thermal Desorption Spectroscopy is a kind of Measurement Prime

the apparatus and protocol map evolving desorption flux onto calibrated temperature-indexed signals.

Hierarchy path (1) — routes to 1 parentless root

Thermal Desorption Spectroscopy → Measurement

Neighborhood in Abstraction Space¶

Thermal Desorption Spectroscopy sits in a sparse region of the domain-specific corpus (92^nd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

adsorption or an adsorption isotherm;
thermogravimetric analysis;
differential scanning calorimetry;
temperature-programmed reaction or reduction;
pyrolysis mass spectrometry without an adsorbed surface population;
desorption ionization methods whose heating program and surface-kinetic inference differ;
a context-free conversion from peak temperature to binding energy.

References¶

[1] P. A. Redhead, “Thermal Desorption of Gases,” Vacuum 12(4) (1962), 203–211, https://doi.org/10.1016/0042-207X(62)90978-8. registry ↩

[2] David A. King, “Thermal Desorption from Metal Surfaces: A Review,” Surface Science 47(1) (1975), 384–402, https://doi.org/10.1016/0039-6028(75)90302-7. registry ↩

[3] A. M. de Jong and J. W. Niemantsverdriet, “Thermal Desorption Analysis: Comparative Test of Ten Commonly Applied Procedures,” Surface Science 233(3) (1990), 355–365, https://doi.org/10.1016/0039-6028(90)90649-S. registry ↩

[4] Aditya Savara, “Simulation and Fitting of Complex Reaction Network TPR: The Key Is the Objective Function,” Surface Science 653 (2016), 169–180, https://doi.org/10.1016/j.susc.2016.07.001. registry ↩

[5] E. Habenschaden and J. Küppers, “Evaluation of Flash Desorption Spectra,” Surface Science 138(1) (1984), L147–L150, https://doi.org/10.1016/0039-6028(84)90488-6. registry

[6] “Thermal desorption spectroscopy,” Wikipedia, frozen evidence packet, https://en.wikipedia.org/wiki/Thermal_desorption_spectroscopy. registry