Measuring Oxygen Nonstoichiometry in Functional Oxides: TGA, Coulometric Titration, and Defect Analysis
Oxygen nonstoichiometry is often written as \(\delta\) in a composition such as \(ABO_{3-\delta}\). It is not merely a small deviation from an ideal chemical formula: it is a thermodynamic state variable that can control transition-metal valence, electronic carrier concentration, oxygen-ion transport, chemical expansion, and the stability of functional oxides.
A defect-chemistry model may predict how \(\delta\) should vary with temperature and oxygen partial pressure, but the model must be tested against experimental composition data. This article focuses on how oxygen nonstoichiometry is measured, how the measured signal is converted into \(\delta\), and how to avoid over-interpreting the result.
The central distinction is between measuring oxygen exchange and identifying the defects responsible for it. Gravimetry and coulometry can quantify oxygen uptake or release under suitable conditions. They do not, by themselves, establish the charge state, spatial distribution, or mobility of every defect.
1. What does oxygen nonstoichiometry mean?
For an oxygen-deficient perovskite, the composition is commonly represented as:
\[ ABO_{3-\delta} \]
where \(\delta\) is the number of missing oxygen atoms per formula unit relative to the chosen \(ABO_3\) reference composition. Thus:
- \(\delta=0\) corresponds to the selected stoichiometric reference.
- \(\delta>0\) represents oxygen deficiency.
- A decrease in \(\delta\) represents oxygen uptake or oxidation.
- An increase in \(\delta\) represents oxygen loss or reduction.
The reference composition must be defined carefully. Some oxides accommodate oxygen excess and are better represented as \(ABO_{3+\delta}\) or with a more general composition variable. Layered oxides may contain multiple crystallographic oxygen sites, and their oxygen defects need not be distributed uniformly.
The symbol \(\delta\) is therefore a composition descriptor, not a direct count of freely mobile oxygen vacancies. Defect association, ordering, and site selectivity can make the relationship between total oxygen deficiency and transport more complicated.
2. Why measure oxygen nonstoichiometry directly?
A measured value of \(\delta(T,p_{O_2})\) provides an experimental constraint for defect models. It can help answer several questions:
- How much oxygen is released when the oxide is reduced?
- How does the equilibrium oxygen content respond to temperature?
- How strongly does the composition depend on oxygen partial pressure?
- Is the response reversible when the atmosphere is switched back?
- Does a proposed defect model satisfy both composition data and charge neutrality?
For an idealized oxygen-loss reaction:
\[ O_O^x \rightleftharpoons V_O^{\bullet\bullet} +\frac{1}{2}O_2(g)+2e' \]
one oxygen atom leaves the lattice, one oxygen vacancy is formed, and two electrons are generated. The electrons may be localized on cations or occupy electronic states; they should not automatically be treated as mobile conduction electrons.
The measured oxygen content constrains the total reduction, while complementary measurements are needed to determine how the associated charge is accommodated.
3. Thermogravimetric analysis under controlled atmosphere
Thermogravimetric analysis (TGA) records the mass of a sample as a function of temperature or time. For an oxide that exchanges oxygen with the surrounding gas, oxygen uptake or release changes the sample mass.
A useful experiment controls:
- Temperature and heating/cooling rate.
- Oxygen partial pressure, \(p_{O_2}\).
- Total gas flow and gas composition.
- Water-vapour content or dew point.
- Sample mass, geometry, and pan material.
- Time allowed for equilibration after each gas or temperature change.
For equilibrium nonstoichiometry, isothermal measurements at a series of oxygen partial pressures are often more informative than a single continuous heating ramp. A continuous ramp can mix kinetic limitations with equilibrium thermodynamics.
3.1 Converting mass change into oxygen deficiency
Consider a sample with a known initial composition \(ABO_{3-\delta_i}\). Let:
- \(m_i\) be the initial sample mass at the reference state.
- \(\Delta m=m_f-m_i\) be the measured mass change.
- \(M_i\) be the molar mass of the initial formula unit, in g mol\(^{-1}\).
- \(M_O\) be the molar mass of oxygen, approximately 15.999 g mol\(^{-1}\).
If oxygen is the only species exchanged and the number of formula units remains constant, the change in oxygen deficiency is:
\[ \Delta\delta =\delta_f-\delta_i =-\frac{\Delta m\,M_i}{M_O\,m_i} \]
Therefore:
\[ \boxed{ \delta_f=\delta_i- \frac{\Delta m\,M_i}{M_O\,m_i} } \]
The sign is important. During reduction, oxygen leaves the sample and \(\Delta m\) is negative; the calculated \(\delta\) increases. During oxidation, the sample gains oxygen, \(\Delta m\) is positive, and \(\delta\) decreases.
This expression assumes that the mass change is caused exclusively by oxygen exchange. It is not valid without modification if the sample loses volatile cations, decomposes, absorbs water or carbon dioxide, reacts with the sample holder, or changes its number of formula units through another chemical process.
3.2 Worked calculation
Suppose a 0.5000 g oxide sample has an initial formula-unit molar mass of 200.0 g mol\(^{-1}\) and a known initial oxygen deficiency of \(\delta_i=0.020\). After equilibration in a reducing atmosphere, its mass decreases by 0.00160 g.
Using the oxygen-only exchange approximation:
\[ \Delta\delta =-\frac{(-0.00160)(200.0)} {15.999(0.5000)} \]
\[ \Delta\delta\approx 0.0400 \]
The final oxygen deficiency is:
\[ \delta_f=0.020+0.040=0.060 \]
The final composition is therefore approximately \(ABO_{2.94}\), provided the assumed reference formula is \(ABO_3\) and the material remains a single phase with no other mass-changing reaction.
The arithmetic may be straightforward, but the reliability of the result depends on the quality of the initial reference composition and the validity of the oxygen-only assumption.
3.3 Choosing a reference state
Absolute oxygen nonstoichiometry requires a reference state whose oxygen content is known or independently established. If only the mass difference between two atmospheres is measured, the experiment determines \(\Delta\delta\), not necessarily the absolute value of \(\delta\).
Possible reference strategies include:
- A well-characterized oxidized composition.
- A validated reduction endpoint with known final phases and stoichiometry.
- Independent chemical analysis of the initial or final material.
- A coulometric titration reference state.
- A composition standard established by a suitable redox method.
Do not label a mass-change-derived value as an absolute \(\delta\) unless the reference oxygen content is defensible.
4. Experimental corrections and sources of error in TGA
Small oxygen exchanges can correspond to very small mass changes. Several instrumental and sample-related effects must therefore be considered.
4.1 Buoyancy and gas-density effects
Changing the gas composition or temperature can change the buoyancy force acting on the sample and suspension. This can produce an apparent mass shift even when the sample composition is unchanged.
A blank measurement using an appropriate inert reference or empty pan can help identify the instrument response. The correction procedure should match the gas sequence, flow, temperature, and pan configuration used for the sample.
4.2 Baseline drift
Long high-temperature experiments can show balance drift. A stable baseline, blank correction, and repeated reference measurements are important when the oxygen-related signal is small.
4.3 Gas switching and equilibration
The gas reaching the sample does not necessarily change instantaneously when a mass-flow controller setting is changed. Tubing volume, furnace volume, flow rate, and gas mixing create a finite response time.
A mass plateau should not be accepted as equilibrium solely because the displayed trace appears flat. Check whether the gas composition has stabilized and whether the sample response is reversible.
4.4 Sample geometry and kinetic limitations
Large or densely packed samples may develop oxygen-chemical-potential gradients. Powder particle size, porosity, gas access, and oxygen diffusion can affect the time required to equilibrate.
Repeat selected conditions with a smaller sample mass or a different particle-size fraction. If the final mass depends on sample geometry or dwell time, equilibrium may not have been reached.
4.5 Volatile species and secondary reactions
Mass loss is not necessarily oxygen loss. Carbonates, adsorbed water, residual organics, volatile cations, and reactions with the crucible can all alter the mass.
Use pre-treatment and blank experiments where appropriate. Confirm phase stability before and after the TGA sequence using XRD or another suitable method.
5. Mapping oxygen nonstoichiometry against oxygen partial pressure
At a fixed temperature, the equilibrium oxygen deficiency can be measured as a function of oxygen partial pressure:
\[ \delta=\delta(p_{O_2})\big|_T \]
A series of measurements at several temperatures produces a family of isotherms. These data can reveal changes in the dominant defect regime and provide constraints for thermodynamic models.
The oxygen chemical potential is related to oxygen partial pressure by:
\[ \mu_{O_2}(T,p_{O_2}) =\mu_{O_2}^{\circ}(T) +RT\ln\left(\frac{p_{O_2}}{p^{\circ}}\right) \]
where \(p^{\circ}\) is the standard pressure and \(R\) is the gas constant. For defect reactions written per mole of oxygen atoms rather than oxygen molecules, the corresponding chemical-potential expression contains the appropriate factor of one-half.
A log–log plot of \(\delta\) against \(p_{O_2}\) can be useful, but a straight-line segment does not uniquely identify a defect species. The slope must be interpreted together with the proposed mass-action expressions, charge neutrality, and the assumptions used in the model.
6. Coulometric titration
Coulometric titration determines oxygen exchange by measuring the electrical charge passed through an electrochemical cell. A solid electrolyte that conducts oxygen ions can be used to pump oxygen into or out of a sample while the applied current and time are recorded.
The method is particularly useful when oxygen exchange is too small for reliable gravimetric detection or when a controlled change in oxygen content is required.
For transfer of oxygen ions, two electrons are involved per oxygen atom. If \(Q\) is the magnitude of the corrected charge and \(n_f\) is the number of moles of sample formula units, the magnitude of the oxygen-content change is:
\[ |\Delta\delta|=\frac{|Q|}{2F n_f} \]
where \(F\) is the Faraday constant. The sign of \(\Delta\delta\) depends on whether the cell operation oxidizes or reduces the sample.
This relation assumes that the measured charge is used quantitatively for oxygen transfer. Parasitic currents, leakage, electronic conduction through the electrolyte, side reactions, and incomplete Faradaic efficiency must be assessed.
6.1 Practical considerations
A coulometric titration experiment requires careful control of:
- Electrolyte composition and oxygen-ion conductivity.
- Electrode material and electrode/electrolyte contact.
- Cell sealing and gas leakage.
- Current integration and baseline correction.
- Temperature uniformity.
- The sample’s equilibration time after each charge increment.
A stable cell voltage does not by itself prove that the sample has reached chemical equilibrium. It is necessary to distinguish the electrochemical response of the cell from the oxygen-transport and equilibration response of the oxide under study.
6.2 Strengths and limitations
Coulometric titration can provide sensitive oxygen-content changes and can be used to control the oxygen chemical state of a sample. However, it requires a suitable electrochemical cell and a reliable accounting of the Faradaic charge.
For thin films, conventional gravimetry is often impractical because the film mass is extremely small. Electrochemical approaches, chemical capacitance, or coulometric methods adapted to thin-film cells may be more suitable. Their interpretation still depends on a defensible relationship between the measured electrical response and oxygen content.
7. Chemical redox analysis
Chemical titration methods can determine the average oxidation state of redox-active cations. If the cation composition and charge balance are known, the average oxygen content can then be calculated.
For a nominal composition containing known cation valences, electroneutrality provides the relationship between oxygen content and average cation oxidation state. For example, for a simple \(ABO_{3-\delta}\) composition:
\[ z_A+z_B=2(3-\delta) \]
where \(z_A\) and \(z_B\) represent the formal oxidation-state contributions per formula unit. In a real multication compound, the sum must include every cation and its stoichiometric coefficient.
Iodometric titration, cerimetric titration, and other redox methods can be useful for selected transition-metal oxides. The method must be validated for the particular composition because incomplete dissolution, side reactions, mixed valence, and endpoint uncertainty can bias the inferred oxygen content.
A chemical redox result is an average bulk composition measurement. It does not reveal whether oxygen defects occupy particular crystallographic sites or form ordered defect structures.
8. Comparing the main measurement methods
| Method | Primary measured quantity | Main advantage | Main limitation |
|---|---|---|---|
| Controlled-atmosphere TGA | Mass change | Direct, versatile oxygen-exchange measurement for bulk samples | Requires a reliable reference state and corrections for non-oxygen mass changes |
| Coulometric titration | Electrical charge | Sensitive control and quantification of oxygen transfer | Requires a suitable cell and Faradaic-efficiency assessment |
| Chemical redox titration | Redox equivalents | Independent bulk chemical analysis | Depends on dissolution chemistry and a valid redox reaction |
| Chemical capacitance | Differential electrical response | Useful for thin films and small oxygen-content changes | Requires a model or calibration to convert response into composition |
| Neutron or X-ray diffraction | Structural occupancies and lattice response | Can provide structural information | Occupancy refinement may be correlated with other parameters and is not always sufficiently sensitive |
No technique is universally best. A strong study selects methods according to sample mass, expected oxygen-exchange magnitude, temperature range, atmosphere, and the required distinction between absolute composition and relative changes.
9. From measured composition to a defect-chemical model
Once \(\delta\) has been measured, it can be compared with a defect model. For a simplified oxide in which oxygen vacancies and electrons dominate reduction:
\[ O_O^x \rightleftharpoons V_O^{\bullet\bullet} +\frac{1}{2}O_2(g)+2e' \]
The corresponding mass-action expression is:
\[ K_{red}= \frac{a_{V_O^{\bullet\bullet}}a_{e'}^2 (p_{O_2}/p^{\circ})^{1/2}} {a_{O_O^x}} \]
The model must also satisfy charge neutrality. In an idealized intrinsic reduction regime:
\[ [e']=2[V_O^{\bullet\bullet}] \]
If oxygen sites are nearly fully occupied and dilute-defect approximations are valid, the model gives the limiting dependence:
\[ [V_O^{\bullet\bullet}]\propto p_{O_2}^{-1/6} \]
This exponent is conditional, not universal. In acceptor-doped oxides, fixed dopant charge can control vacancy concentration over part of the oxygen-pressure range. At more reducing conditions, electronic carriers and other defects may become important. Defect association and non-ideal activities can also change the apparent slope.
A good analysis therefore does not fit a power law first and assign defects afterward. It specifies candidate defect reactions, derives the expected dependencies, and tests whether one physically consistent model explains the composition and other measurements over the full experimental range.
10. Thermodynamic quantities from oxygen nonstoichiometry
Composition isotherms can be used to investigate oxygen chemical potential and partial molar thermodynamic properties. At a fixed temperature, the dependence of oxygen chemical potential on composition is related to the oxygen partial pressure at equilibrium.
For oxygen molecules in the gas phase:
\[ \mu_{O_2}-\mu_{O_2}^{\circ} =RT\ln(p_{O_2}/p^{\circ}) \]
The partial molar quantities of oxygen in the solid can be obtained from the equilibrium relation with the gas phase, using a consistent convention for oxygen atoms or molecules. Temperature-dependent measurements can then be analyzed using thermodynamic relations such as the Gibbs–Helmholtz equation.
Care is required when comparing reported enthalpies and entropies: some authors define quantities per mole of oxygen atoms, while others use a mole of oxygen molecules or a reaction as written. Always check the stoichiometric convention and standard state before comparing values.
11. Connecting oxygen content with structure and transport
Oxygen nonstoichiometry is only one part of the material response. It should be connected to independent structural and functional measurements.
11.1 X-ray diffraction
XRD can identify phase changes and track lattice parameters as oxygen content changes. A lattice shift may be consistent with reduction, but it is not a direct measurement of \(\delta\). Cation valence, local relaxation, thermal expansion, stress, and defect ordering can all affect the lattice.
Use consistent refinement models and verify whether the phase symmetry remains unchanged throughout the atmosphere sequence.
11.2 Electrical conductivity and impedance
Conductivity measurements can show how the electronic and ionic response changes with oxygen chemical potential. However, conductivity does not directly equal vacancy concentration. Carrier mobility, defect association, and electronic compensation must also be considered.
EIS can help separate bulk, grain-boundary, and electrode contributions when their characteristic responses are distinguishable. A decrease in resistance under a reducing atmosphere should not automatically be described as an increase in oxygen-ion conductivity.
11.3 Chemical expansion
When oxygen content changes, cation valence and local bonding may change, producing chemical strain. Combining TGA with dilatometry or in-situ XRD can relate oxygen exchange to dimensional or lattice response.
For the mechanical implications and definitions of chemical-expansion coefficients, see Chemical Expansion and Thermochemical Strain in Functional Oxides.
12. A recommended experimental workflow
A practical measurement campaign can follow this sequence:
- Confirm composition and phase purity. Characterize the starting powder or ceramic and document its thermal history.
- Define the oxygen reference state. Establish how the initial or final oxygen content is known.
- Select the measurement method. Match TGA, coulometry, or chemical analysis to the sample mass and expected oxygen exchange.
- Calibrate and run blank experiments. Check buoyancy, baseline drift, gas switching, and cell background.
- Establish equilibrium criteria. Define a mass or voltage stability criterion and a minimum dwell time; verify selected points by extending the dwell.
- Measure isotherms. At each temperature, vary \(p_{O_2}\) over a documented range and allow the sample to equilibrate.
- Test reversibility. Return to earlier gas conditions and check whether the original mass or charge state is recovered.
- Check phase stability. Use post-experiment XRD or complementary analysis to detect decomposition or secondary phases.
- Calculate \(\delta\) with uncertainty. Propagate uncertainties in mass, reference composition, formula mass, and calibration.
- Fit a defect model. Apply mass action and charge neutrality together; report the assumptions and the range over which the model is valid.
For a robust report, include sample mass, particle size or geometry, pan or cell materials, gas purity, flow rate, water content, temperature calibration, dwell time, \(p_{O_2}\) calibration, blank correction, and the reference oxygen composition.
13. Common errors in reporting oxygen nonstoichiometry
- Reporting an absolute \(\delta\) from a relative mass change without defining the reference oxygen content.
- Assuming all measured mass loss is oxygen loss.
- Ignoring buoyancy changes when switching gases.
- Treating a transient mass plateau as proof of equilibrium without checking gas-delivery and sample kinetics.
- Applying the TGA conversion to a composition that decomposes or loses volatile cations.
- Ignoring the distinction between oxygen atoms and oxygen molecules in coulometric calculations.
- Comparing thermodynamic quantities that use different oxygen reference conventions.
- Treating a fitted \(p_{O_2}\) exponent as unique proof of a defect species.
- Equating total oxygen deficiency with the concentration of mobile oxygen vacancies.
- Using XRD lattice parameters or XPS peak fitting as a substitute for a validated bulk oxygen-content measurement.
14. Key takeaways
- Oxygen nonstoichiometry is a measurable composition variable that constrains defect-chemical models.
- Controlled-atmosphere TGA converts oxygen-related mass changes into \(\Delta\delta\), provided oxygen is the only exchanged species and a valid reference state is known.
- Coulometric titration quantifies oxygen transfer through Faradaic charge and can be valuable when gravimetric sensitivity is insufficient.
- Chemical redox analysis provides an independent route for suitable redox-active oxides.
- Absolute oxygen content and relative oxygen-content changes are not interchangeable.
- Equilibrium, reversibility, phase stability, and instrumental corrections are essential to reliable measurements.
- The strongest interpretation combines composition data with defect equilibria, structural characterization, and transport measurements.
References
- M. Oishi, K. Yashiro, and J. Mizusaki, “High Temperature Microbalance Technique for the Determination of the Metal Oxides Nonstoichiometry under Controlled Atmosphere,” Netsu Sokutei 34 (2004), 206. Journal article
- M. Karppinen, M. Matvejeff, K. Salomäki, and H. Yamauchi, “Oxygen content analysis of functional perovskite-derived cobalt oxides,” Journal of Materials Chemistry 12 (2002), 1761–1764. https://doi.org/10.1039/B200770N
- D.-K. Lee, J.-I. Jeon, M.-H. Kim, W. Choi, and H.-I. Yoo, “Oxygen nonstoichiometry (\(\delta\)) of TiO\(_{2-\delta}\)-revisited,” Journal of Solid State Chemistry 178 (2005), 185–193. https://doi.org/10.1016/j.jssc.2004.07.034
- Y. Ling, F. Wang, R. A. Budiman, T. Nakamura, and K. Amezawa, “Oxygen nonstoichiometry, the defect equilibrium model and thermodynamic quantities of the Ruddlesden-Popper oxide Sr\(_3\)Fe\(_2\)O\(_{7-\delta}\),” Physical Chemistry Chemical Physics 17 (2015), 7489–7497. https://doi.org/10.1039/C4CP05719H
- Y. Zhao et al., “Measurement and control of oxygen non-stoichiometry in praseodymium-cerium oxide thin films by coulometric titration,” Journal of Electroceramics 51 (2023), 28. https://doi.org/10.1007/s10832-023-00309-x