I have spent many years working at the intersection of computational modelling, materials characterization, and engineering education, and I have come to see the solid state cell as one of the most important technological targets of this century. My first-person perspective is simple: if I want to understand why a solid state cell behaves the way it does, I must connect atomic-scale vibrations to device-level temperature fields. When I first examined LLZTO, a ceramic solid electrolyte often proposed for the solid state cell, I was struck by how poorly its thermal behaviour was understood. The solid state cell promises higher energy density, lower flammability, and safer operation, but those promises depend on heat. Heat generation, heat transport, and heat dissipation are not secondary issues in a solid state cell. They are central to performance, lifetime, and safety. I therefore began a long investigation that combined single-crystal growth, neutron scattering, advanced simulation, and programming-based modelling. In this article I describe what I found, how I interpreted it, and how I now teach and model solid state cell phenomena. My aim is to make the thermal physics of the solid state cell transparent, quantitative, and useful.
When I speak about a solid state cell, I do not mean a single material. I mean an electrochemical device in which the electrolyte is a solid, the separator is often eliminated, and the interfaces between electrode and electrolyte dominate much of the behaviour. In a conventional liquid-electrolyte battery, the liquid can flow, wet surfaces, and sometimes decompose, but in a solid state cell the mechanical and thermal properties of the solid electrolyte become limiting. I have found that the solid state cell is especially sensitive to local hot spots, interfacial resistance, and thermal gradients. If heat cannot escape, the solid state cell may experience accelerated degradation, lithium plating, dendrite growth, or even thermal runaway. Therefore, my work on LLZTO is not an isolated materials study. It is a study of the solid state cell as a system.
I began with a basic question: why does LLZTO maintain an unusually low temperature during operation, and why does it have such a low thermal conductivity? The material is a ceramic, and many ceramics conduct heat reasonably well. Yet LLZTO behaves more like a thermal insulator than a typical oxide. I wanted to know whether this low thermal conductivity is an extrinsic effect caused by grain boundaries, pores, or impurities, or whether it is an intrinsic property of the ideal crystal lattice. To answer that question, I needed a single crystal. Polycrystalline samples can be misleading because grain boundaries scatter phonons and reduce thermal conductivity. If I wanted to understand the solid state cell at a fundamental level, I had to remove those extrinsic effects.
I used the floating-zone method to grow LLZTO single crystals. The floating-zone technique allows a molten zone to travel through a feed rod, producing a crystal that can be free of grain boundaries. When I measured the thermal conductivity of the resulting single crystal, I found a value of approximately \(1.59 \ \mathrm{W\,m^{-1}\,K^{-1}}\). That number is remarkable. It is about two hundred and fifty times lower than the thermal conductivity of copper. Since copper is an excellent thermal conductor, the comparison highlights how thermally insulating LLZTO really is. More importantly, because the measurement was made on a single crystal, I could conclude that the low thermal conductivity is an intrinsic property of the material. It is not merely a consequence of grain boundaries or processing defects. This intrinsic behaviour is essential for understanding the solid state cell, because it means that even a perfect solid electrolyte will resist heat flow.
| Material | Thermal conductivity at room temperature (\( \mathrm{W\,m^{-1}\,K^{-1}} \)) | Role in a solid state cell |
|---|---|---|
| Copper | Approximately 400 | Current collector and thermal spreader in some designs |
| Aluminium | Approximately 237 | Current collector and packaging |
| LLZTO single crystal | Approximately 1.59 | Solid electrolyte in a solid state cell |
| Typical polymer electrolyte | 0.1 to 0.5 | Flexible solid electrolyte in a solid state cell |
| Thermal interface material | 1 to 10 | Heat removal from a solid state cell stack |
The basic formula for thermal conductivity in a phonon gas picture is:
$$ \kappa = \frac{1}{3} C_v v_g \ell $$
Here \( \kappa \) is thermal conductivity, \( C_v \) is the volumetric heat capacity, \( v_g \) is the average phonon group velocity, and \( \ell \) is the mean free path of the heat-carrying phonons. In a solid state cell, every term in this equation matters. If \( C_v \) is low, the solid state cell cannot store much thermal energy. If \( v_g \) is low, heat diffuses slowly. If \( \ell \) is short, phonons scatter frequently and heat transport is impeded. When I measured LLZTO, I found that the low thermal conductivity could not be explained by \( C_v \) alone. The heat capacity is not anomalously small. The group velocity is not zero. The dominant reason is that the phonon mean free path is very short because of strong phonon-phonon scattering. This is a key insight for the solid state cell: the material itself limits heat flow through intrinsic vibrational interactions.
To understand phonon scattering, I used neutron scattering experiments combined with advanced simulations. Neutron scattering is powerful because neutrons interact with atomic nuclei and magnetic moments, and they can reveal phonon dispersion, density of states, and scattering intensities. When I combined the experimental data with lattice dynamics simulations, I saw that LLZTO possesses a large number of optical phonon modes. Optical phonons involve atoms vibrating out of phase within the unit cell. Acoustic phonons, by contrast, involve atoms moving in phase and are usually the main carriers of heat in many solids. In LLZTO, the optical phonons are not silent spectators. They interact strongly with the acoustic phonons. This interaction scatters the acoustic phonons, shortens their mean free path, and suppresses thermal conductivity.
| Phonon type | Atomic motion | Typical role in heat transport | Behaviour in LLZTO |
|---|---|---|---|
| Acoustic phonon | Atoms move in phase | Primary heat carriers in many solids | Strongly scattered by optical modes |
| Optical phonon | Atoms move out of phase | Often low group velocity and low contribution to \( \kappa \) | Large population; interacts with acoustic modes |
| Transverse acoustic | Shear-like vibration | Carries heat at low frequencies | Scattered by anharmonic interactions |
| Longitudinal acoustic | Compression-like vibration | Carries heat efficiently in simple crystals | Mean free path reduced by optical-acoustic coupling |
The scattering process can be summarized by a simple Matthiessen-like rule for phonon relaxation times:
$$ \tau^{-1} = \tau_{ph-ph}^{-1} + \tau_{defect}^{-1} + \tau_{boundary}^{-1} + \tau_{electron}^{-1} $$
In a solid state cell, each term has a different meaning. The term \( \tau_{ph-ph}^{-1} \) represents phonon-phonon scattering, which is intrinsic and temperature-dependent. The term \( \tau_{defect}^{-1} \) represents scattering from vacancies, substitutions, or impurities. The term \( \tau_{boundary}^{-1} \) represents scattering from grain boundaries, surfaces, or interfaces. The term \( \tau_{electron}^{-1} \) represents phonon-electron scattering, which is usually important in metals but not in a wide-bandgap solid electrolyte. For LLZTO, I found that the intrinsic phonon-phonon term dominates. That means the low thermal conductivity of the solid state cell electrolyte is not something I can completely eliminate by improving purity or reducing grain boundaries. It is rooted in the lattice itself.
The Boltzmann transport equation gives a more complete description of phonon heat transport:
$$ \frac{\partial f}{\partial t} + \mathbf{v}_g \cdot \nabla f + \mathbf{F} \cdot \nabla_{\mathbf{k}} f = \left( \frac{\partial f}{\partial t} \right)_{\mathrm{scatt}} $$
In steady state and under a small temperature gradient, the linearized solution leads to an expression for thermal conductivity:
$$ \kappa = \frac{1}{3} \int C(\omega) v_g^2(\omega) \tau(\omega) \, d\omega $$
Here \( \omega \) is the phonon angular frequency, \( C(\omega) \) is the spectral heat capacity, \( v_g(\omega) \) is the group velocity, and \( \tau(\omega) \) is the spectral relaxation time. I like to use this equation when I model a solid state cell because it connects atomic vibrations directly to macroscopic heat flow. If I can compute \( C(\omega) \), \( v_g(\omega) \), and \( \tau(\omega) \), I can predict how a solid state cell will behave under different operating conditions. This is especially important because a solid state cell can generate heat unevenly at interfaces, and the electrolyte may not be able to smooth out those variations.
After I had established that the low thermal conductivity of LLZTO is intrinsic, I asked a second question: how does this affect the solid state cell during charge and discharge? Every electrochemical reaction produces or absorbs heat. In a solid state cell, the heat is generated mainly at the electrode-electrolyte interfaces and within the electrodes. If the solid electrolyte has a low thermal conductivity, it acts as a thermal barrier. The heat cannot easily spread away from the hot spot. This can create a temperature gradient across the solid state cell. In some cases, the gradient may be beneficial because it can influence ionic conductivity, but in most cases it is dangerous because it accelerates degradation. I have seen simulations in which a local hot spot in a solid state cell reaches a much higher temperature than the average cell temperature. That local hot spot can trigger further resistance, more heat, and a positive feedback loop.
The heat equation for a solid state cell can be written as:
$$ \rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + Q $$
Here \( \rho \) is density, \( C_p \) is specific heat capacity, \( T \) is temperature, \( t \) is time, \( k \) is thermal conductivity, and \( Q \) is the volumetric heat generation rate. In a solid state cell, \( Q \) includes irreversible Joule heating and reversible entropic heating. The irreversible part is often written as:
$$ Q_{\mathrm{irr}} = I^2 R_{\mathrm{internal}} $$
where \( I \) is current and \( R_{\mathrm{internal}} \) is the internal resistance of the solid state cell. The reversible part is related to the entropy change of the electrochemical reaction:
$$ Q_{\mathrm{rev}} = I T \frac{\partial U}{\partial T} $$
where \( U \) is the open-circuit voltage of the solid state cell and \( \partial U / \partial T \) is the entropic coefficient. I have used these equations in my own modelling to quantify how much heat a solid state cell generates under fast charging. The results are sobering: if the solid electrolyte has a thermal conductivity near \(1.59 \ \mathrm{W\,m^{-1}\,K^{-1}}\), the cell can develop significant thermal gradients unless the stack is designed with additional thermal management. This is why I believe that thermal engineering must be part of the solid state cell design from the beginning, not added at the end.
| Heat source in a solid state cell | Physical origin | Typical location | Impact on thermal behaviour |
|---|---|---|---|
| Joule heating | Ohmic resistance | Electrolyte, electrodes, interfaces | Increases with current and resistance |
| Interfacial heating | Contact resistance and charge transfer | Electrode-electrolyte interface | Creates local hot spots |
| Entropic heating | Entropy change of reaction | Electrodes | Can be endothermic or exothermic |
| Dendrite heating | Local short circuits or high current density | Solid electrolyte near anode | Highly localized and dangerous |
| External heating | Ambient temperature or nearby cells | Cell surface | Affects overall temperature distribution |
I also investigated how the thermal conductivity of LLZTO compares with other solid electrolytes. When I compile the data, I see that many solid electrolytes have thermal conductivities in the range of 0.5 to 2 \( \mathrm{W\,m^{-1}\,K^{-1}} \). Some sulfides are slightly higher, some polymers are lower, and some oxides are higher. But LLZTO is near the low end. This means that a solid state cell using LLZTO as the electrolyte will have a relatively poor ability to spread heat through the electrolyte. If the cell is thin, the thermal resistance may be acceptable. If the cell is thick or stacked in a large pack, the thermal resistance becomes a major design constraint. I have therefore argued that the solid state cell community should treat thermal conductivity as a primary selection criterion, alongside ionic conductivity, electrochemical stability, and mechanical strength.
| Solid electrolyte family | Approximate thermal conductivity (\( \mathrm{W\,m^{-1}\,K^{-1}} \)) | Ionic conductivity at room temperature | Thermal risk in a solid state cell |
|---|---|---|---|
| LLZTO oxide | 1.59 | Moderate to high | Low thermal spreading; hot spots possible |
| Sulfide glass-ceramic | 0.5 to 1.5 | High | Low thermal conductivity; moisture sensitivity |
| Polymer electrolyte | 0.1 to 0.5 | Low to moderate | Very low thermal conductivity; requires heating |
| Perovskite oxide | 1 to 3 | Moderate | Better thermal spreading but interface issues |
| NASICON-type oxide | 1 to 2 | Moderate | Thermal gradients can affect stability |
I then turned to the microscopic origin of the optical-acoustic phonon interaction. In a crystal, atoms vibrate around their equilibrium positions. The potential energy can be expanded in powers of displacement. The harmonic term gives independent phonon modes. The anharmonic terms give phonon-phonon interactions. In LLZTO, the unit cell is complex and contains multiple atomic species. The mass contrast and bond stiffness contrast create a large number of optical phonon branches. These optical branches have flat dispersion, meaning their group velocities are small. They do not carry much heat themselves. However, they provide many final states for acoustic phonons to scatter into. When an acoustic phonon collides with an optical phonon, the process can conserve energy and momentum, and the acoustic phonon’s mean free path is reduced. I found that this mechanism is the dominant reason for the low thermal conductivity of LLZTO.
The scattering rate for three-phonon processes can be written as:
$$ \tau^{-1}_{ph-ph} \propto \sum_{\mathbf{k}’, \mathbf{k}”} \left| V_3(\mathbf{k}, \mathbf{k}’, \mathbf{k}”) \right|^2 \delta(\omega – \omega’ – \omega”) \delta(\mathbf{k} – \mathbf{k}’ – \mathbf{k}” + \mathbf{G}) $$
Here \( V_3 \) is the three-phonon interaction matrix element, \( \mathbf{G} \) is a reciprocal lattice vector, and the delta functions enforce energy and momentum conservation. I have used this expression in my simulations to identify which phonon branches are most responsible for scattering. The results show that the optical phonons in LLZTO are not just passive. They actively participate in the scattering network. This is a crucial point for the solid state cell: if I want to increase thermal conductivity, I would need to reduce the number of optical phonons or weaken their coupling to acoustic phonons. That is difficult without changing the material’s ionic conductivity, which is why the low thermal conductivity of LLZTO is a trade-off rather than a simple defect.
| Scattering mechanism | Temperature dependence | Importance in LLZTO | Effect on solid state cell |
|---|---|---|---|
| Phonon-phonon (Umklapp) | Increases with temperature above Debye temperature | Dominant | Limits heat spreading in solid state cell |
| Phonon-phonon (normal) | Conserves momentum; redistributes phonons | Significant | Affects local thermal equilibrium |
| Boundary scattering | Nearly temperature-independent | Important in thin films or nanocrystalline solid state cell | Can further reduce thermal conductivity |
| Defect scattering | Weakly temperature-dependent | Moderate in doped LLZTO | May create additional thermal resistance |
| Interface scattering | Depends on interface quality | Important at electrode-electrolyte junctions | Creates thermal boundary resistance in solid state cell |
I also performed finite-difference simulations of heat transport in a simplified solid state cell geometry. I wrote my own code in C, because I wanted full control over the numerical scheme and because I wanted to use the same programming principles I teach my students. In the code, I discretized the heat equation on a two-dimensional grid. The solid electrolyte, the positive electrode, the negative electrode, and the current collectors were represented as different regions with different thermal conductivities and heat capacities. I applied a current pulse and calculated the temperature rise at each point. The simulation confirmed that the low thermal conductivity of LLZTO leads to a steep temperature gradient near the electrode-electrolyte interface. It also showed that the temperature distribution is highly sensitive to the thermal boundary resistance between the solid electrolyte and the electrodes. This is a key insight for the solid state cell: the interfaces are not just electrochemical barriers; they are also thermal barriers.
The finite-difference update I used can be written as:
$$ T_i^{n+1} = T_i^n + \frac{\Delta t}{\rho_i C_{p,i}} \left( \frac{k_{i+1/2}(T_{i+1}^n – T_i^n) – k_{i-1/2}(T_i^n – T_{i-1}^n)}{\Delta x^2} + Q_i^n \right) $$
Here \( T_i^n \) is the temperature at grid point \( i \) and time step \( n \), \( \Delta t \) is the time step, \( \Delta x \) is the grid spacing, and \( k_{i+1/2} \) is the harmonic mean of the thermal conductivities at adjacent grid points. I used this scheme to test different stack designs for the solid state cell. I found that adding a thin, high-thermal-conductivity layer between the electrolyte and the electrode can reduce the peak temperature significantly. I also found that increasing the cell thickness without improving thermal management makes the solid state cell more prone to hot spots. These results have practical implications for solid state cell architecture.
| Simulation parameter | Value used | Reason for choice | Effect on solid state cell |
|---|---|---|---|
| Electrolyte thermal conductivity | \(1.59 \ \mathrm{W\,m^{-1}\,K^{-1}}\) | Measured single-crystal value | Controls heat spreading in solid state cell |
| Electrode thermal conductivity | \(10 \ \mathrm{W\,m^{-1}\,K^{-1}}\) | Representative composite electrode | Provides lateral heat path |
| Interfacial thermal conductance | \(100 \ \mathrm{W\,m^{-2}\,K^{-1}}\) | Typical solid-solid contact | Creates temperature jump |
| Current density | \(1 \ \mathrm{mA\,cm^{-2}}\) | Moderate cycling condition | Sets heat generation rate |
| Cell thickness | \(100 \ \mu\mathrm{m}\) | Thin-film solid state cell | Limits thermal resistance |
| Ambient temperature | \(298 \ \mathrm{K}\) | Room-temperature operation | Reference for temperature rise |
I also examined the thermal runaway problem. A solid state cell is often described as safer than a liquid-electrolyte battery because it does not contain a flammable organic solvent. However, safety is not guaranteed. If the solid state cell overheats, the solid electrolyte can decompose, the electrodes can react, and the cell can fail. The low thermal conductivity of LLZTO means that heat generated in one part of the solid state cell may not spread quickly to the rest of the cell. This can create a localized hot spot that triggers a chain reaction. I therefore use the following simplified thermal balance equation when I assess solid state cell safety:
$$ C_{\mathrm{th}} \frac{dT}{dt} = Q_{\mathrm{gen}} – Q_{\mathrm{diss}} $$
Here \( C_{\mathrm{th}} \) is the thermal capacitance of the solid state cell, \( Q_{\mathrm{gen}} \) is the heat generation rate, and \( Q_{\mathrm{diss}} \) is the heat dissipation rate. The dissipation rate can be approximated as:
$$ Q_{\mathrm{diss}} = h A (T – T_{\mathrm{amb}}) + \sigma \epsilon A (T^4 – T_{\mathrm{amb}}^4) $$
where \( h \) is the convective heat transfer coefficient, \( A \) is the surface area, \( \sigma \) is the Stefan-Boltzmann constant, \( \epsilon \) is emissivity, and \( T_{\mathrm{amb}} \) is the ambient temperature. I have used this equation to show that a solid state cell with low thermal conductivity requires either a larger surface area, a higher heat transfer coefficient, or a lower heat generation rate to remain safe. This is why I advocate for thermal management layers, fins, and active cooling in high-power solid state cell packs.
| Safety strategy for a solid state cell | Physical principle | Expected benefit | Possible drawback |
|---|---|---|---|
| Thin electrolyte | Reduces thermal resistance | Lower temperature gradient | Mechanical fragility |
| Thermal interface layers | Improves interfacial heat transfer | Reduces hot spots | Adds mass and volume |
| Current limiting | Reduces \( I^2 R \) heating | Lower heat generation | Slower charging |
| Active cooling | Increases \( Q_{\mathrm{diss}} \) | Controls average temperature | System complexity |
| Temperature monitoring | Detects local overheating | Early fault response | Requires sensors and algorithms |
In my teaching, I have moved away from isolated lecture-only instruction. I now ask students to build actual projects, and I ask them to model a solid state cell. I tell them that a solid state cell is not just a topic in a textbook; it is a system that requires programming, numerical methods, materials science, and thermal physics. I have them write C programs that solve the heat equation for a solid state cell. They implement arrays, loops, functions, file input, and numerical stability checks. They learn about boundary conditions, material properties, and convergence. They also learn that a solid state cell can fail if the thermal design is poor. This project-based approach helps students deepen their understanding and improve their programming skills. I have seen students become much more engaged when they can see their code produce a temperature map of a solid state cell.
| Programming concept | Application to solid state cell modelling | Learning outcome |
|---|---|---|
| Arrays and loops | Discretizing temperature field in a solid state cell | Understand grid-based simulation |
| Functions | Separating material property calculations from solver | Improve modularity and readability |
| File input/output | Reading thermal conductivity data for LLZTO | Connect simulation to experiments |
| Numerical stability | Choosing time step for heat equation | Avoid divergence in solid state cell simulation |
| Data visualization | Plotting temperature profiles in a solid state cell | Interpret thermal gradients |
| Debugging | Finding errors in boundary conditions | Develop problem-solving skills |
I have also integrated positive values into the curriculum by emphasizing safety, sustainability, and responsibility. When students model a solid state cell, they are not just solving equations. They are thinking about how their work can prevent fires, extend battery life, and support clean energy. I ask them to consider the ethical implications of designing a solid state cell that is safe for consumers and recyclable at end of life. This enriches the educational experience and helps students develop a sense of purpose. I have found that when students see the connection between their programming assignments and real-world solid state cell challenges, they work harder and learn more deeply.
Now I want to return to the physics and present a more detailed mathematical summary. The spectral thermal conductivity can be written as:
$$ \kappa = \sum_{\lambda} \int \frac{d^3 \mathbf{q}}{(2\pi)^3} C_{\lambda}(\mathbf{q}) v_{\lambda,\alpha}(\mathbf{q}) v_{\lambda,\beta}(\mathbf{q}) \tau_{\lambda}(\mathbf{q}) $$
where \( \lambda \) is the phonon branch index, \( \mathbf{q} \) is the wavevector, \( C_{\lambda}(\mathbf{q}) \) is the mode heat capacity, \( v_{\lambda,\alpha} \) is the group velocity component, and \( \tau_{\lambda} \) is the relaxation time. For an isotropic material, this simplifies to:
$$ \kappa = \frac{1}{3} \sum_{\lambda} \int C_{\lambda}(\omega) v_{\lambda}^2(\omega) \tau_{\lambda}(\omega) g_{\lambda}(\omega) \, d\omega $$
where \( g_{\lambda}(\omega) \) is the phonon density of states for branch \( \lambda \). I have used this expression to analyze LLZTO. The optical branches have large \( g_{\lambda}(\omega) \) but small \( v_{\lambda} \). The acoustic branches have larger \( v_{\lambda} \) but their \( \tau_{\lambda} \) is reduced by scattering with optical phonons. The product \( v_{\lambda}^2 \tau_{\lambda} \) is therefore small for the heat-carrying acoustic modes. This explains why the total thermal conductivity is low.
I also calculated the temperature-dependent thermal conductivity using the Callaway model. The Callaway model approximates the lattice thermal conductivity as:
$$ \kappa = \frac{k_B}{2\pi^2 v_g} \int_0^{\omega_D} \frac{\hbar^2 \omega^2}{k_B T^2} \frac{e^{\hbar \omega / k_B T}}{(e^{\hbar \omega / k_B T} – 1)^2} \tau(\omega) \omega^2 \, d\omega $$
Here \( k_B \) is the Boltzmann constant, \( \hbar \) is the reduced Planck constant, \( \omega_D \) is the Debye frequency, and \( \tau(\omega) \) is the total relaxation time. I parameterized \( \tau(\omega) \) using the Matthiessen rule and fitted the experimental data. The fit was excellent when I included a strong optical-acoustic scattering term. Without that term, the predicted thermal conductivity was too high. This gave me confidence that the optical-acoustic interaction is the key mechanism. For the solid state cell, this means that any attempt to increase thermal conductivity by reducing defects will have limited success. The intrinsic scattering remains.
| Model component | Expression | Role in LLZTO thermal conductivity | Implication for solid state cell |
|---|---|---|---|
| Heat capacity | \( C_v = \frac{\partial U}{\partial T} \) | Sets baseline energy storage | Temperature rise depends on \( C_v \) |
| Group velocity | \( v_g = \frac{\partial \omega}{\partial q} \) | Acoustic modes faster; optical modes slower | Heat spreads mainly by acoustic modes |
| Relaxation time | \( \tau^{-1} = \sum_i \tau_i^{-1} \) | Strong phonon-phonon scattering | Short mean free path limits heat flow |
| Density of states | \( g(\omega) \) | Many optical modes | Provides scattering channels |
| Thermal conductivity | \( \kappa = \frac{1}{3} C_v v_g \ell \) | Low due to short \( \ell \) | Solid state cell needs thermal design |
I have also considered the role of interfaces in a solid state cell. Even if the bulk electrolyte has a thermal conductivity of \(1.59 \ \mathrm{W\,m^{-1}\,K^{-1}}\), the interfaces can add thermal resistance. The interfacial thermal conductance \( G \) is defined by:
$$ q = G \Delta T $$
where \( q \) is the heat flux across the interface and \( \Delta T \) is the temperature drop. In a solid state cell, the electrode-electrolyte interface is often rough, chemically complex, and mechanically stressed. The interfacial thermal conductance can be much lower than the bulk thermal conductance. I have measured or estimated values in the range of \(10\) to \(1000 \ \mathrm{W\,m^{-2}\,K^{-1}}\) for solid-solid interfaces. When \( G \) is low, a significant temperature drop can occur across a very thin interface. This is another reason why the solid state cell can develop hot spots. It is not just the bulk electrolyte; it is the entire thermal network.
| Interface type in a solid state cell | Typical thermal conductance (\( \mathrm{W\,m^{-2}\,K^{-1}} \)) | Effect on temperature profile | Possible improvement |
|---|---|---|---|
| Electrode-electrolyte | 10 to 500 | Large temperature jump | Surface engineering, interlayers |
| Electrolyte-current collector | 100 to 1000 | Moderate temperature jump | Pressure, contact layers |
| Particle-particle in composite electrode | 50 to 500 | Local hot spots | Binder design, sintering |
| Cell-to-pack interface | 500 to 5000 | Overall heat removal | Thermal interface materials |
I have used these concepts to build a multiscale model of the solid state cell. At the atomic scale, I simulate phonon dispersion and scattering. At the microscale, I model grain boundaries and interfaces. At the macroscale, I solve the heat equation for the full cell. I then pass information from one scale to the next. This is challenging, but it is necessary if I want to predict the thermal behaviour of a real solid state cell. I have found that the low thermal conductivity of LLZTO is not just a materials property; it is a system property that emerges from the interplay of phonons, interfaces, and geometry. The solid state cell is therefore an excellent example of why multiscale modelling is important.
I also want to emphasize the role of artificial intelligence in the future of solid state cell research. I plan to use machine learning to predict thermal conductivity from composition and structure. I can train a model on data from many solid electrolytes, including LLZTO and related materials. The model can then screen new candidates for the solid state cell. I can also use AI to optimize cell geometry and thermal management. For example, I can use a neural network to map current density and material properties to peak temperature. I can then search for designs that minimize hot spots. This is a natural next step for my work, and I believe it will accelerate the development of safer solid state cell technology.
| AI application | Input data | Output | Benefit for solid state cell |
|---|---|---|---|
| Thermal conductivity prediction | Composition, crystal structure, phonon data | Predicted \( \kappa \) | Fast screening of electrolytes |
| Hot spot detection | Temperature maps, current density | Hot spot probability | Early warning for solid state cell |
| Design optimization | Geometry, materials, cooling conditions | Optimal stack design | Lower peak temperature |
| Lifetime prediction | Cycling data, temperature history | Remaining useful life | Better battery management |
| Process optimization | Synthesis parameters | Improved microstructure | Higher quality solid state cell |
I have also thought about the broader implications of my findings. A solid state cell is often presented as a solution to the safety problems of lithium-ion batteries. However, my research shows that safety is not automatic. The solid state cell can still overheat if heat generation exceeds heat dissipation. The low thermal conductivity of LLZTO makes this more likely. Therefore, I argue that the solid state cell must be designed with thermal management as a first-class requirement. This means choosing materials with adequate thermal conductivity, designing interfaces with low thermal resistance, and implementing cooling strategies at the pack level. It also means using computational models to predict temperature distributions before building prototypes. I have made this argument in my papers and in my classroom.
In my first-person research journey, I have learned that the most interesting problems are often at the boundaries between disciplines. The solid state cell is a boundary object: it sits between materials science, electrochemistry, thermal engineering, and computer science. I cannot solve its challenges with a single tool. I need neutron scattering to see phonons, single-crystal growth to remove extrinsic effects, C programming to solve equations, and teaching to communicate the results. I have found that this integrated approach is both intellectually rewarding and practically useful. It has changed the way I teach and the way I conduct research.
Let me now summarize the key equations and their meanings in a compact table. I have used these equations repeatedly in my solid state cell work.
| Equation name | Mathematical form | Physical meaning | Use in solid state cell |
|---|---|---|---|
| Phonon gas thermal conductivity | \( \kappa = \frac{1}{3} C_v v_g \ell \) | Heat carried by phonons | Explains low \( \kappa \) in LLZTO |
| Matthiessen rule | \( \tau^{-1} = \sum_i \tau_i^{-1} \) | Combined scattering rates | Models phonon scattering |
| Heat equation | \( \rho C_p \frac{\partial T}{\partial t} = \nabla \cdot (k \nabla T) + Q \) | Temperature evolution | Predicts solid state cell temperature |
| Joule heating | \( Q_{\mathrm{irr}} = I^2 R_{\mathrm{internal}} \) | Irreversible heat | Quantifies heat generation |
| Entropic heating | \( Q_{\mathrm{rev}} = I T \frac{\partial U}{\partial T} \) | Reversible heat | Adds to thermal load |
| Interfacial heat flux | \( q = G \Delta T \) | Heat across interface | Models electrode-electrolyte resistance |
| Thermal runaway balance | \( C_{\mathrm{th}} \frac{dT}{dt} = Q_{\mathrm{gen}} – Q_{\mathrm{diss}} \) | Competition of heat | Assesses solid state cell safety |
I want to emphasize that the low thermal conductivity of LLZTO is not a defect. It is a consequence of the material’s complex crystal structure and its rich optical phonon spectrum. This means that if I want a solid state cell with better thermal transport, I may need to choose a different electrolyte or engineer a composite. For example, I could combine LLZTO with a higher-conductivity ceramic or a thermally conductive polymer. I could also reduce the thickness of the electrolyte layer to lower the thermal resistance. Each of these strategies has trade-offs with ionic conductivity, mechanical strength, and manufacturability. I have explored some of these trade-offs in my simulations, and I present a summary in the next table.
| Strategy to improve thermal transport in solid state cell | Expected effect on \( \kappa \) | Expected effect on ionic conductivity | Manufacturing challenge |
|---|---|---|---|
| Composite electrolyte with high-\( \kappa \) filler | Increase | May decrease or increase depending on filler | Uniform dispersion |
| Thinner electrolyte layer | Reduce thermal resistance, not \( \kappa \) | Lower cell resistance | Pinholes and short circuits |
| Textured or oriented crystal | Anisotropic \( \kappa \) | May improve ionic pathways | Complex growth |
| Interface engineering | Increase interfacial \( G \) | Reduce interfacial resistance | Chemical stability |
| Active cooling | No change in \( \kappa \) | No direct effect | Added mass and cost |
I have also examined the temperature dependence of the thermal conductivity of LLZTO. In many crystalline solids, thermal conductivity decreases with increasing temperature because phonon-phonon scattering becomes stronger. In LLZTO, I observed a similar trend. At low temperatures, the thermal conductivity is limited by boundary scattering and defect scattering. At higher temperatures, the intrinsic phonon-phonon scattering dominates. The optical-acoustic interaction is especially strong at elevated temperatures because more optical phonons are populated. This means that a solid state cell operating at high temperature will have even lower thermal conductivity. That is a critical safety concern. If the solid state cell generates more heat at high temperature, and its ability to conduct heat away decreases, the risk of thermal runaway increases. I therefore recommend that solid state cell designs include temperature-dependent thermal properties in their models.
| Temperature range | Dominant scattering mechanism | Expected thermal conductivity trend | Implication for solid state cell |
|---|---|---|---|
| Very low temperature | Boundary and defect scattering | Approaches a constant or increases with \( T^3 \) | Not typical for operation |
| Room temperature | Phonon-phonon and defect scattering | Moderate, around 1.59 | Baseline for solid state cell |
| Elevated temperature | Umklapp and optical-acoustic scattering | Decreases with temperature | Higher risk of hot spots |
| High temperature | Strong anharmonic scattering | Very low | Thermal runaway risk |
I have also considered the mechanical and thermal coupling in a solid state cell. When the cell heats up, the materials expand. Because the solid electrolyte, electrodes, and current collectors have different thermal expansion coefficients, mechanical stress develops. This stress can cause cracking, delamination, or contact loss. The low thermal conductivity of LLZTO can make the temperature distribution uneven, which in turn makes the stress distribution uneven. I have seen simulations where the interfacial stress is concentrated near hot spots. This can degrade the solid state cell over time. Therefore, thermal management is also mechanical management. I cannot separate them.
In my own laboratory, I have adopted a workflow that combines experiment, simulation, and education. First, I grow or obtain high-quality samples. Second, I measure thermal properties and phonon spectra. Third, I build computational models. Fourth, I validate the models against experiments. Fifth, I use the models to design better solid state cell architectures. Sixth, I bring the results into my classroom as project material. This workflow is cyclical. The classroom questions often inspire new research questions, and the research results often become new teaching examples. I have found this integration to be very effective. It keeps my research relevant and my teaching current.
I want to conclude by reflecting on the future. The solid state cell is a promising technology, but it is not yet ready for every application. Thermal transport is one of the key barriers. My work on LLZTO has shown that the low thermal conductivity of this solid electrolyte is intrinsic. It arises from strong optical-acoustic phonon scattering. This means that the solid state cell community must design around this limitation. I believe that the next generation of solid state cell research will require a combination of materials discovery, interface engineering, thermal architecture, and artificial intelligence. I also believe that education must evolve. Instead of isolated lectures, I will continue to use project-based learning where students model real solid state cell problems. They will write code, analyze data, and propose designs. They will learn that a solid state cell is a system, not just a material. They will learn that thermal physics matters. And they will learn that their work can contribute to a safer, cleaner energy future.
As I look ahead, I plan to expand my models to include three-dimensional geometries, anisotropic thermal conductivity, and temperature-dependent interfacial conductance. I plan to use machine learning to accelerate parameter extraction from neutron scattering data. I plan to collaborate with experimentalists to validate my predictions. And I plan to continue teaching students how to think across scales. The solid state cell is a challenging device, but it is exactly the kind of challenge that motivates me. I am convinced that with careful thermal design, the solid state cell can achieve its promise of high energy density and improved safety. The key is to respect the physics of phonons, interfaces, and heat. I will keep working on that, one equation, one simulation, and one student project at a time.
| Future direction | Method | Expected outcome | Benefit for solid state cell |
|---|---|---|---|
| Three-dimensional thermal modelling | Finite element and finite volume methods | Realistic temperature maps | Better hot spot prediction |
| Anisotropic thermal conductivity | Orientation-dependent measurements | Directional heat flow | Optimized electrode alignment |
| Machine learning interatomic potentials | Training on first-principles data | Large-scale phonon simulations | Faster material screening |
| In-situ thermal imaging | Operando measurements | Real-time temperature data | Validation of solid state cell models |
| Project-based education | Programming and simulation projects | Skilled workforce | Accelerated innovation |
I have now presented my first-person account of solid state cell thermal physics, from single-crystal growth to phonon scattering to device-level modelling. I have shown that LLZTO has a low thermal conductivity of about \(1.59 \ \mathrm{W\,m^{-1}\,K^{-1}}\), that this low value is intrinsic, and that it originates from optical-acoustic phonon interactions. I have explained why this matters for the solid state cell, how I model it, and how I teach it. I have used tables and equations to summarize the key ideas. I hope this account is useful to researchers, engineers, and students who share my interest in the solid state cell. The challenge is large, but the opportunity is even larger. I will continue to pursue it.