I designed and evaluated a three-dimensional electron/ion mixed-conducting bilayer interface for a NASICON-based solid state battery, specifically targeting the unstable contact between Li1.3Al0.3Ti1.7(PO4)3 and lithium metal. In my view, the central problem of a high-energy solid state battery is not only the bulk ionic conductivity of the electrolyte but also the coupled chemical, electrochemical, mechanical, and thermal behavior of the Li/electrolyte interface. A solid state battery that uses a rigid inorganic electrolyte can suppress dendrite penetration better than a liquid-electrolyte cell, but it can also suffer from poor wetting, interfacial decomposition, void formation, and stress concentration. I therefore built a bilayer architecture in which a lithiophilic mixed-conducting layer faces the Li metal and an electronically insulating ion-conducting layer faces the NASICON pellet. This architecture transforms the interface into an in-situ three-dimensional Li host and simultaneously acts as a chemical and thermal buffer for the solid state battery.
My motivation came from the fact that LATP offers high room-temperature ionic conductivity, a high elastic modulus, a wide electrochemical stability window, good ambient stability, non-flammability, and relatively low cost. These are exactly the properties required for a scalable solid state battery. However, LATP is thermodynamically unstable against Li metal. When Li contacts LATP, Ti4+ is reduced to Ti3+, and a mixed-conducting interphase forms. Unlike a desirable solid electrolyte interphase, this mixed-conducting interphase is both ionically resistive and electronically conductive. The electronic conductivity makes the decomposition non-self-limiting. The result is continuous interfacial growth, rising resistance, volume expansion, mechanical degradation, and eventually Li filament propagation. At elevated temperature, molten Li can react violently with LATP, accompanied by oxygen release, which is a serious safety concern for any practical solid state battery. I therefore asked whether a single interlayer could simultaneously block electron percolation, conduct Li+, host Li metal, accommodate volume change, and suppress thermal runaway.
Design rationale and interface targets
I formulated the interface as a series of transport and stability requirements rather than as a simple coating. The Li-facing side must be electronically conductive, ionically conductive, and lithiophilic. The electrolyte-facing side must be electronically insulating and ionically conductive. The entire structure must be porous enough to host Li and compliant enough to accommodate plating and stripping. It must also remain chemically stable against both Li metal and LATP. I summarized these targets in a design table before fabrication.
| Requirement | Failure mode without control | My design response |
|---|---|---|
| Chemical stability against Li | Ti4+ reduction, mixed-conducting interphase growth, LATP fragmentation | Electronically insulating ion-conducting layer between LATP and Li |
| Chemical stability against LATP | Side reactions, impedance rise, capacity fade | Stable polymer-ceramic composite with ionic liquid wetting |
| Low interfacial resistance | High overpotential, poor rate capability | Lithiophilic Au nanoparticles and mixed-conducting Li host |
| Uniform Li deposition | Dendrites, protrusions, dead Li, soft short circuits | Three-dimensional fiber network with abundant nucleation sites |
| Mechanical accommodation | Stress concentration, SEI rupture, void formation | High-porosity bilayer with buffer space and elastic polymer matrix |
| Thermal stability | Molten Li reaction, oxygen release, thermal runaway | Thermally stable interlayer that separates molten Li from LATP |
| Scalability | Expensive vacuum processes, low throughput | Electrospinning plus magnetron sputtering |
The central transport idea can be expressed through the total interfacial resistance. If I separate the solid state battery interface into a mixed-conducting layer, an ion-conducting layer, and the LATP pellet, then the area-specific resistance is approximately
$$ R_{\mathrm{total}} = R_{\mathrm{Li/MCL}} + R_{\mathrm{MCL}} + R_{\mathrm{MCL/ICL}} + R_{\mathrm{ICL}} + R_{\mathrm{ICL/LATP}} + R_{\mathrm{LATP}}. $$
I wanted to minimize each term while ensuring that the electronic path does not extend into the LATP. The key asymmetry is that electrons should move freely in the mixed-conducting layer but should be blocked before reaching the solid electrolyte. In contrast, Li+ should move freely across all layers. This gives a selective transport window. I represented the desired selectivity as
$$ \sigma_{\mathrm{Li}^{+},\mathrm{MCL}} \gg \sigma_{\mathrm{e},\mathrm{ICL}} \approx 0, $$
where the mixed-conducting layer provides both ionic and electronic pathways, while the ion-conducting layer provides only ionic pathways. In my design, the mixed-conducting layer had an electronic conductivity of approximately 7.0 × 10−3 S cm−1, whereas the ion-conducting layer showed no measurable electronic conductivity. This contrast is the core of the interface protection.
Materials, fabrication, and structural design
I fabricated the bilayer by sequential electrospinning of two dispersions. Both dispersions contained polyacrylonitrile, poly(vinylidene fluoride), nano-LATP, and LiClO4. Polyacrylonitrile provided ionic conduction and chemical stability. Poly(vinylidene fluoride) improved mechanical strength and helped dissociate the lithium salt through its fluorine groups. Nano-LATP acted as a ceramic filler that accelerated Li+ transport and reduced polymer crystallinity. The mixed-conducting layer additionally contained single-walled carbon nanotubes for electronic conductivity. After electrospinning, I deposited Au nanoparticles onto the mixed-conducting layer by magnetron sputtering. The Au nanoparticles were lithiophilic and provided nucleation sites with low Li nucleation overpotential.
| Layer | Composition | Primary transport role | Thickness contribution | Measured property |
|---|---|---|---|---|
| Mixed-conducting layer | PAN, PVDF, nano-LATP, CNTs, LiClO4, Au nanoparticles | Electron and Li+ transport, Li hosting, nucleation | Approximately half of a 30 µm bilayer | Electronic conductivity 7.0 × 10−3 S cm−1 |
| Ion-conducting layer | PAN, PVDF, nano-LATP, LiClO4 | Li+ transport, electron blocking, LATP protection | Approximately half of a 30 µm bilayer | No measurable electronic conductivity |
| Bilayer interface | MCL and ICL stack | Mixed conduction with spatial selectivity | Approximately 30 µm total | Porosity approximately 89% |
For the ion-conducting layer, I used a mass ratio of PAN:PVDF:nano-LATP = 9:1:5, with LiClO4 added to achieve a [Li+]:[CN] molar ratio of 1:10. The total solid content was 12.5 wt%. For the mixed-conducting layer, I used PAN:PVDF:nano-LATP:CNT = 9:1:5:1, again with LiClO4 at a [Li+]:[CN] molar ratio of 1:10. The total solid content was 10.5 wt%. I chose these ratios to balance spinnability, ionic conductivity, mechanical integrity, and electronic percolation.
| Parameter | Ion-conducting layer dispersion | Mixed-conducting layer dispersion |
|---|---|---|
| PAN:PVDF:nano-LATP mass ratio | 9:1:5 | 9:1:5 |
| CNT addition | None | 1 part relative to PAN:PVDF:LATP |
| LiClO4 ratio | [Li+]:[CN] = 1:10 | [Li+]:[CN] = 1:10 |
| Solvent | N,N-dimethylformamide | N,N-dimethylformamide |
| Total solid content | 12.5 wt% | 10.5 wt% |
| Ultrasonication | 15 min for nano-LATP | 15 min for nano-LATP and CNTs |
| Magnetic stirring | 800 r min−1 for 12 h | 800 r min−1 for 12 h |
Electrospinning was performed with a 0.27 mm inner-diameter needle, a 30 kV applied voltage, and a 10 cm tip-to-drum distance. The drum collector rotated at 350 r min−1, and the solution feeding rate was 0.8 mL h−1. The ambient temperature was 30 ± 2 °C, and the relative humidity was 40 ± 5%. I electrospun each dispersion for 2 h to obtain the bilayer. Au nanoparticles were deposited by magnetron sputtering under Ar at 0.5 Pa, with a current of 30 mA for 10 s. The resulting interlayer was cut into discs and vacuum-dried at 50 °C for 72 h.
| Fabrication step | Condition | Purpose |
|---|---|---|
| Electrospinning voltage | 30 kV | Stable Taylor cone and fiber formation |
| Needle diameter | 0.27 mm | Controlled droplet size |
| Tip-to-drum distance | 10 cm | Solvent evaporation and fiber collection |
| Drum rotation | 350 r min−1 | Uniform fiber mat |
| Feeding rate | 0.8 mL h−1 | Continuous electrospinning |
| Temperature | 30 ± 2 °C | Reproducible fiber morphology |
| Relative humidity | 40 ± 5% | Controlled solvent evaporation |
| Electrospinning time per layer | 2 h | Equal thickness contribution |
| Au sputtering | Ar, 0.5 Pa, 30 mA, 10 s | Lithiophilic nucleation seeds |
| Vacuum drying | 50 °C for 72 h | Residual solvent removal |
For the solid electrolyte, I synthesized Li1.3Al0.3Ti1.7(PO4)3 from Li2CO3, Al2O3, TiO2, and NH4H2PO4. After wet ball milling and drying, the precursor was calcined at 900 °C for 10 h. The resulting ceramic blocks were crushed and milled to an average particle size of about 600 nm. A subsequent fine milling step produced nano-LATP with an average particle size of about 100 nm. The 600 nm powder was pressed into pellets at 150 MPa for 5 min and sintered at 900 °C for 6 h. The pellets were polished and cleaned before use. I used the nano-LATP as a filler in the interlayer and the larger LATP powder for the solid electrolyte pellet.
| LATP preparation step | Condition | Result |
|---|---|---|
| Precursor mixing | Wet ball milling in acetone, 600 r min−1, 1 h | Uniform mixture |
| Calcination | 900 °C, 10 h, air | NASICON phase formation |
| Coarse milling | 600 r min−1, 2 h | Average particle size approximately 600 nm |
| Fine milling | 800 r min−1, 8 h | Nano-LATP approximately 100 nm |
| Pellet pressing | 150 MPa, 5 min | Green body |
| Pellet sintering | 900 °C, 6 h, air | Dense LATP solid electrolyte |
| Polishing | 800, 2500, 5000 mesh | Smooth interface |
Structural and chemical characterization
I examined the bilayer by scanning electron microscopy, energy-dispersive X-ray spectroscopy, and X-ray diffraction. The mixed-conducting layer consisted of randomly oriented fibers with abundant inter-fiber space. Au nanoparticles and nano-LATP fillers were uniformly distributed on the fiber surfaces. The ion-conducting layer also had a fibrous structure, with nano-LATP embedded within the fibers. The cross-section showed a continuous bilayer with a total thickness of about 30 µm. The porosity was approximately 89%, which is important because it provides space for Li deposition and ionic liquid retention.
| Observation | Mixed-conducting layer | Ion-conducting layer | Bilayer |
|---|---|---|---|
| Morphology | Random fibers with Au and nano-LATP on surfaces | Random fibers with embedded nano-LATP | Continuous two-layer fiber mat |
| Optical appearance | Gray | White | Asymmetric |
| Electronic conductivity | 7.0 × 10−3 S cm−1 | Not measurable | Electron-blocking on ICL side |
| Porosity | High inter-fiber porosity | High inter-fiber porosity | Approximately 89% |
| Thickness | About half of total | About half of total | About 30 µm |
X-ray diffraction confirmed the crystalline features of the components. Polyacrylonitrile showed characteristic peaks near 17.0° and 29.3°, assigned to the (100) and (110) planes. Poly(vinylidene fluoride) showed peaks near 18.4°, 20.0°, and 26.6°, assigned to the (020), (110), and (021) planes. Carbon nanotubes showed peaks near 26.0° and 43.0°, corresponding to the (002) and (100) planes of graphite. Nano-LATP matched the NASICON reference pattern. In both the ion-conducting and mixed-conducting layers, the LATP peaks remained pronounced, while the polymer peaks weakened. I interpreted this as reduced polymer crystallinity after ceramic filler incorporation, which favors Li+ migration.
| Phase | Characteristic 2θ position | Assignment | Role in the interface |
|---|---|---|---|
| Polyacrylonitrile | About 17.0° and 29.3° | (100) and (110) | Polymer matrix and ionic transport |
| Poly(vinylidene fluoride) | About 18.4°, 20.0°, and 26.6° | (020), (110), and (021) | Mechanical strength and salt dissociation |
| Carbon nanotubes | About 26.0° and 43.0° | (002) and (100) graphite planes | Electronic percolation in MCL |
| Nano-LATP | Matches NASICON reference | NASICON structure | Ceramic ionic filler and interfacial stabilizer |
I also introduced a small amount of ionic liquid into the interlayer. The ionic liquid was based on 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl)imide with LiFSI and LiTFSI. The loading was controlled to approximately match the interlayer porosity. The porous fiber network was rapidly and thoroughly infiltrated. In contrast, a commercial polypropylene separator showed negligible wettability. This difference is important for a solid state battery because poor wetting creates voids and local current hotspots. The ionic liquid confined inside the fiber network improved interfacial contact without dissolving the solid electrolyte.
Electrochemical methods and transport equations
I assembled symmetric cells with the configuration Li | mixed-conducting layer | ion-conducting layer | LATP | ion-conducting layer | mixed-conducting layer | Li. The mixed-conducting layer faced the Li metal. I added 5 µL of ionic liquid to each interlayer. For comparison, I assembled cells wetted with a conventional liquid electrolyte and cells wetted only with ionic liquid. The conventional liquid electrolyte was 1 M LiPF6 in EC/DMC/DEC. The ionic liquid was PYR14-FSI/LiFSI/LiTFSI at a molar ratio of 85:10:5. I cycled symmetric cells at 0.1 mA cm−2 with a capacity of 0.1 mAh cm−2. I evaluated critical current density by stepwise increases from 0.1 to 2.0 mA cm−2 with a plating/stripping time of 0.5 h. I also assembled full cells with LiFePO4 cathodes, LATP pellets, Li anodes, and the bilayer interface. The LiFePO4 loading was 4.0 mg cm−2, and the cathode composition was LiFePO4:Super P:PVDF = 8:1:1. Full cells were tested at 0.2 C and 0.6 C, with rate tests at multiple rates. All tests were performed at 27 °C.
| Test | Cell configuration | Condition | Purpose |
|---|---|---|---|
| Long-term symmetric cycling | Li | bilayer + IL | LATP | bilayer + IL | Li | 0.1 mA cm−2, 0.1 mAh cm−2 | Interface stability |
| Liquid electrolyte control | Li | LATP + LE | Li | 0.1 mA cm−2, 0.1 mAh cm−2 | Baseline degradation |
| Ionic liquid control | Li | LATP + IL | Li | 0.1 mA cm−2, 0.1 mAh cm−2 | Wetting without bilayer |
| Critical current density | Symmetric cells | 0.1 to 2.0 mA cm−2, 0.5 h steps | Dendrite tolerance |
| Electrochemical impedance spectroscopy | Symmetric and full cells | 0.1 Hz to 1 MHz, 10 mV | Resistance analysis |
| Li+ transference number | Symmetric cells | 10 mV polarization | Ionic selectivity |
| Linear sweep voltammetry | Li | interlayer | stainless steel | 0.1 mV s−1, −0.1 to 6.0 V | Electrochemical stability window |
| Full-cell cycling | Li | bilayer + IL | LATP | LiFePO4 + IL | 0.2 C and 0.6 C | Practical solid state battery performance |
I used the Bruce–Vincent method to determine the Li+ transference number:
$$ t_{\mathrm{Li}^+} = \frac{I_s(\Delta V – I_0 R_0)}{I_0(\Delta V – I_s R_s)}, $$
where \(I_0\) and \(I_s\) are the initial and steady-state currents, \(\Delta V\) is the applied potential, and \(R_0\) and \(R_s\) are the charge-transfer resistances before and after polarization. A high Li+ transference number is desirable because it reduces anion accumulation and concentration polarization at the Li interface.
I derived the exchange current density from the Tafel relation:
$$ \log i = \log i_0 + \frac{\eta}{b}, $$
where \(\eta\) is the overpotential, \(i_0\) is the exchange current density, and \(b\) is the Tafel slope. I used the kinetically controlled region and extrapolated to zero overpotential. I also analyzed the low-frequency Warburg region:
$$ Z_w = \sigma_w \omega^{-1/2}(1-j), $$
where \(\sigma_w\) is the Warburg coefficient and \(\omega\) is the angular frequency. From this, I calculated the Li+ diffusion coefficient:
$$ D_{\mathrm{Li}} = \frac{R^2 T^2}{2A^2 n^4 F^4 C^2 \sigma_w^2}, $$
where \(R\) is the gas constant, \(T\) is the absolute temperature, \(A\) is the electrode area, \(n\) is the number of transferred electrons, \(F\) is the Faraday constant, and \(C\) is the Li+ concentration. This equation allowed me to compare Li+ transport in the bilayer, liquid-electrolyte-wetted LATP, and ionic-liquid-wetted LATP.
For the simulation of Li deposition, I used the Butler–Volmer equation:
$$ j = j_0\left[\exp\left(\frac{\alpha_a F \eta}{RT}\right)-\exp\left(-\frac{\alpha_c F \eta}{RT}\right)\right], $$
where \(\alpha_a\) and \(\alpha_c\) are the anodic and cathodic charge-transfer coefficients. Under Li plating conditions, the cathodic term dominates, so I simplified the expression to
$$ j \approx j_0 \exp\left(-\frac{\alpha_c F \eta}{RT}\right). $$
Because Li+ reduction is relatively fast, the deposition process is often governed by Li+ diffusion from the bulk electrolyte to the electrode surface. I therefore solved Fick’s second law:
$$ \frac{\partial C_{\mathrm{Li}}(x,t)}{\partial t} = D_{\mathrm{Li}} \frac{\partial^2 C_{\mathrm{Li}}(x,t)}{\partial x^2}, $$
where \(C_{\mathrm{Li}}(x,t)\) is the Li+ concentration at position \(x\) and time \(t\). The concentration overpotential near the interface was described by
$$ \eta_{\mathrm{conc}} = \frac{RT}{F}\ln\left(\frac{C_{\mathrm{Li}}}{C_{\mathrm{Li}}(0,t)}\right), $$
where \(C_{\mathrm{Li}}\) is the bulk Li+ concentration and \(C_{\mathrm{Li}}(0,t)\) is the Li+ concentration at the electrode surface. I also used the ionic conductivity relation
$$ \sigma = \frac{L}{R A}, $$
where \(L\) is thickness, \(R\) is resistance, and \(A\) is area. The porosity of the bilayer was estimated as
$$ \phi = 1 – \frac{\rho_{\mathrm{bilayer}}}{\rho_{\mathrm{solid}}}. $$
Interfacial stability in symmetric solid state battery cells
The symmetric-cell results were the first clear demonstration that the bilayer interface changed the degradation pathway. Cells with the bilayer cycled for more than 1400 h at 0.1 mA cm−2 and 0.1 mAh cm−2, with an overpotential of approximately 10 mV. Cells wetted only with ionic liquid failed after about 842 h, and cells wetted with liquid electrolyte failed after about 683 h. The control cells also started with much higher overpotentials of about 75 mV and 80 mV, respectively, and their overpotentials increased continuously. I interpreted the low and stable overpotential of the bilayer cell as evidence of fast interfacial kinetics, uniform Li deposition, and a chemically stable interface.
| Interface configuration | Initial overpotential | Cycle life at 0.1 mA cm−2 | LATP pellet after cycling | Interfacial degradation |
|---|---|---|---|---|
| Bilayer interface with ionic liquid | Approximately 10 mV | More than 1400 h | Intact and white | Minimal |
| Ionic-liquid-wetted LATP | Approximately 75 mV | About 842 h | Fragmented and discolored | Severe |
| Liquid-electrolyte-wetted LATP | Approximately 80 mV | About 683 h | Fragmented, dark purple or black | Very severe |
I retrieved the LATP pellets after cycling and examined their surface chemistry. Pellets protected by the bilayer remained intact and smooth even after 500 and 1000 h. In contrast, pellets from the ionic-liquid and liquid-electrolyte controls fragmented and developed dark reaction products. X-ray photoelectron spectroscopy of the Ti 2p region showed only Ti4+ peaks at 459.7 and 465.2 eV for pristine LATP and for the bilayer-protected pellet. The control pellets showed additional Ti3+ peaks at 458.6 and 463.5 eV. The liquid-electrolyte control had the strongest Ti3+ signal, indicating the most severe reduction. After 1000 h, the bilayer-protected pellet still showed no Ti3+ signal. This was direct evidence that the ion-conducting layer blocked electron percolation and prevented Ti4+ reduction.
| Sample | Ti4+ peaks | Ti3+ peaks | Interpretation |
|---|---|---|---|
| Pristine LATP | 459.7 and 465.2 eV | Not observed | Stable NASICON surface |
| Bilayer-protected LATP after 500 h | 459.7 and 465.2 eV | Not observed | No detectable reduction |
| Bilayer-protected LATP after 1000 h | Present | Not observed | Long-term chemical protection |
| Ionic-liquid-wetted LATP | Present | 458.6 and 463.5 eV | Interfacial reduction |
| Liquid-electrolyte-wetted LATP | Present | 458.6 and 463.5 eV, stronger | Severe interfacial reduction |
I also followed the impedance evolution. In the equivalent circuit, one resistance described Li+ migration through the SEI-like layer and another described the resistance between the interlayer and the LATP pellet. For cells wetted only with ionic liquid or liquid electrolyte, both resistances increased rapidly with cycling. In the bilayer cell, the first resistance remained low and stable at approximately 7 Ω cm2 per side, while the second resistance increased only gradually from about 13 to 20 Ω cm2. This indicated that a highly conductive interfacial film formed on the three-dimensional anode and that the bilayer/LATP contact remained chemically stable.
| Interface | Initial interfacial resistance | Evolution after cycling | Li+ transference number |
|---|---|---|---|
| Bilayer interface | About 7 Ω cm2 per side for SEI-like layer | Stable; interlayer/LATP resistance about 13 to 20 Ω cm2 | 0.60 |
| Ionic liquid only | Higher and unstable | Rapid increase | 0.21 |
| Liquid electrolyte only | Higher and unstable | Rapid increase | 0.20 |
The Li+ transference number was one of the most important mechanistic results. The bilayer interface reached 0.60, compared with 0.21 for the ionic-liquid-only interface and 0.20 for the liquid-electrolyte-only interface. I attribute this to four coupled effects. First, confining the ionic liquid inside the fibrous network restricts long-range anion migration. Second, the embedded nano-LATP fillers provide solid-assisted Li+ pathways. Third, the mixed-conducting layer directs a preferential Li+ flux toward lithiophilic Au nanoparticles. Fourth, the interconnected structure shortens local diffusion paths, while anions are anchored to the polymer backbone or side chains. As a result, the relative contribution of Li+ to the total current increases, which lowers concentration polarization and suppresses dendrite initiation.
I calculated the Li+ diffusion coefficient from the low-frequency impedance response. The bilayer interface gave 1.23 × 10−9 cm2 s−1, approximately one and a half orders of magnitude higher than the liquid-electrolyte-wetted interface at 7.74 × 10−11 cm2 s−1 and the ionic-liquid-wetted interface at 8.52 × 10−11 cm2 s−1. This result supports the conclusion that the solid–ionic liquid composite interface accelerates Li+ transport and homogenizes the Li+ flux.
| Interface | Li+ transference number | Li+ diffusion coefficient | Relative transport benefit |
|---|---|---|---|
| Bilayer interface | 0.60 | 1.23 × 10−9 cm2 s−1 | Approximately 15 times higher than controls |
| Ionic-liquid-wetted LATP | 0.21 | 8.52 × 10−11 cm2 s−1 | Baseline |
| Liquid-electrolyte-wetted LATP | 0.20 | 7.74 × 10−11 cm2 s−1 | Baseline |
Li deposition morphology and solid electrolyte interphase evolution
I examined the Li anodes after cycling to determine whether the improved electrochemistry corresponded to a real morphological change. After 500 h, the Li anode protected by the bilayer retained a smooth, intact metallic surface that was nearly identical to pristine Li. Scanning electron microscopy revealed a dense, moss-like morphology with uniform Li deposition along the mixed-conducting fibers. Energy-dispersive X-ray spectroscopy showed a homogeneous F signal, indicating uniform SEI coverage. Carbon and oxygen signals were attributed mainly to surface byproducts formed during brief air exposure. The ion-conducting layer side showed no Li deposition and only a slight thickness increase due to swelling. Even after 1000 h, the three-dimensional anode remained flat and the SEI remained integrated.
| Condition | Li surface | Deposition morphology | SEI uniformity | LATP fragments |
|---|---|---|---|---|
| Pristine Li | Smooth | Not applicable | Not applicable | None |
| Bilayer after 500 h | Smooth and intact | Dense, moss-like Li along fibers | Uniform F signal | Not observed |
| Bilayer after 1000 h | Flat and stable | SEI integrity maintained | Stable composition | Not observed |
| Liquid-electrolyte control after 500 h | Rough | Protrusions and dead Li | Uneven | Ti signal detected |
| Ionic-liquid control | Roughened | Dendritic growth | Uneven | Possible |
In contrast, the liquid-electrolyte-wetted Li anode showed severe roughening, protrusion growth, and substantial dead Li accumulation after 500 h. The SEI was unevenly reconstructed, and LATP-derived fragments were detected through the Ti signal. Dendritic growth was also observed in the ionic-liquid-wetted cells. This comparison showed that ionic liquid wetting alone does not guarantee uniform deposition. Without spatial regulation of the Li+ flux, local plating inevitably occurs.
I used X-ray photoelectron spectroscopy with depth profiling to analyze the SEI composition on the three-dimensional anode. The breakdown of the ionic liquid and lithium salts generated abundant LiF at 685 eV throughout the SEI layer on the mixed-conducting layer side. LiF is beneficial because it forms a mechanically robust and ionically conductive interphase. The elemental composition remained relatively stable after 500 and 1000 h, which indicated long-term compositional stability. In the liquid-electrolyte and ionic-liquid controls, LiF was also detected, but its content was lower and its distribution was less uniform. This limited LiF failed to provide effective interfacial protection.
| SEI feature | Bilayer after 500 h | Bilayer after 1000 h | Liquid electrolyte control | Ionic liquid control |
|---|---|---|---|---|
| LiF signal at 685 eV | Abundant | Abundant and stable | Limited | Limited |
| Depth uniformity | Uniform through SEI | Uniform through SEI | Uneven | Uneven |
| Compositional stability | High | High | Low | Low |
| Protective function | Strong | Strong | Weak | Weak |
Finite element simulation of Li deposition
I used finite element simulation to visualize the electric field, Li+ concentration field, and anode thickness evolution. For the liquid-electrolyte-wetted interface, surface roughness induced a non-uniform local electric field. Li+ accumulated at microscopic tips, creating steep electric field gradients. At the same time, sluggish Li+ transport caused ion depletion near the anode and a significant concentration gradient. The local current density increased, and non-uniform Li deposition produced protrusions. These simulation results matched the experimental observations of surface roughening and dendritic features.
For the three-dimensional anode built by the bilayer, the solid–ionic liquid composite interlayer mitigated concentration gradients. Li deposition occurred exclusively on the mixed-conducting layer. The in-situ three-dimensional anode developed a uniform moss-like Li morphology. The simulation showed that the large effective surface area of the mixed-conducting fibers and the Li anode promoted cooperative Li deposition. The thickness changes were evenly distributed, with only slight surface thickening. This minimized local volume variation and preserved SEI integrity. In contrast, the liquid-electrolyte interface showed pronounced volumetric expansion, with some regions stretched and prone to SEI rupture and other regions thickened by Li metal deposition.
| Simulation output | Liquid-electrolyte-wetted interface | Bilayer-built three-dimensional anode |
|---|---|---|
| Electric field | Non-uniform; enhancement at tips | Uniform; no pronounced tip enhancement |
| Li+ concentration | Depletion near anode; steep gradient | Mitigated gradient; stable local concentration |
| Deposition location | Irregular anode surface | Mixed-conducting fibers and Li surface |
| Morphology | Protrusions and dendrites | Uniform moss-like Li |
| Thickness change | Localized expansion and SEI rupture risk | Evenly distributed and small local variation |
| SEI integrity | Poor | Preserved |
I parameterized the simulation using experimental values. The electrode length was 27 µm, the electrode width was 20 µm, and the initial geometry on the Li anode side was randomly generated. For the control group, Li protrusion growth was modeled based on theoretical values and observed protruding structures. For the experimental group, the fiber diameters were determined by statistical analysis of electron microscopy images. The initial Li+ concentration, exchange current density, and Li+ diffusion coefficient were obtained from measurements. Li deposition was simulated under potentiostatic conditions, with the anode surface potential fixed at 0 V and a constant voltage of 0.2 mV applied at the top boundary of the electrolyte.
| Simulation parameter | Value or source |
|---|---|
| Model | Tertiary current distribution with electrodeposition |
| Electrode length | 27 µm |
| Electrode width | 20 µm |
| Initial anode geometry | Randomly generated in control |
| Fiber diameter in experimental group | From microscopy statistics |
| Initial Li+ concentration | Experimental measurement |
| Exchange current density | Experimental measurement |
| Li+ diffusion coefficient | Experimental measurement |
| Anode surface potential | 0 V |
| Top boundary voltage | 0.2 mV |
Critical current density, rate capability, and full-cell performance
The critical current density results confirmed the practical benefit of the bilayer. Cells with the ionic-liquid-wetted interface showed a voltage drop at 0.5 mA cm−2, and cells with the liquid-electrolyte-wetted interface showed a drop at 0.4 mA cm−2. I defined these values as their critical current densities. In contrast, cells with the bilayer interface displayed stable plating and stripping up to 2.0 mA cm−2. Before activation, the bilayer cell had a critical current density of only 1.0 mA cm−2 and a relatively high polarization voltage, which showed that electrochemical activation is necessary to establish favorable interfacial kinetics. After activation, the cell cycled at 1.0 mA cm−2 for more than 650 h. This current density is relevant to high-energy solid state battery targets.
| Interface | Critical current density | Long-term high-current cycling | Electrochemical stability window |
|---|---|---|---|
| Bilayer interface | Up to 2.0 mA cm−2 | More than 650 h at 1.0 mA cm−2 | 0 to 5.5 V |
| Ionic-liquid-wetted LATP | About 0.5 mA cm−2 | Not stable | Narrower |
| Liquid-electrolyte-wetted LATP | About 0.4 mA cm−2 | Not stable | Narrower |
I evaluated the electrochemical stability window by linear sweep voltammetry in a Li | interlayer | stainless steel cell. The bilayer cell showed well-defined Li plating and stripping redox peaks near 0 V. Anodic current was observed only above 5.5 V, with no additional side-reaction peaks within the scanned range. This indicated that the interlayer maintained a wide and stable electrochemical stability window from 0 to 5.5 V. For a solid state battery, this is important because it allows pairing a Li metal anode with a high-voltage cathode.
In full cells with LiFePO4 cathodes, the bilayer interface delivered excellent rate capability. The specific capacity was 166 mAh g−1 at 0.1 C and 126 mAh g−1 at 1.5 C, with low polarization. Long-term cycling at 0.2 C gave an initial capacity of 162 mAh g−1 and retained 95.5% after 100 cycles, with a Coulombic efficiency approaching 99.9%. In contrast, the liquid-electrolyte-wetted full cell retained only 60.0% after 100 cycles. At 0.6 C, the bilayer full cell delivered an initial capacity of 150.0 mAh g−1 and retained 95.0% after 100 cycles, while the liquid-electrolyte control short-circuited after only 41 cycles.
| Full-cell condition | Initial capacity | Capacity retention after 100 cycles | Coulombic efficiency | Control behavior |
|---|---|---|---|---|
| 0.2 C with bilayer | 162 mAh g−1 | 95.5% | Approximately 99.9% | Liquid electrolyte retained only 60.0% |
| 0.6 C with bilayer | 150.0 mAh g−1 | 95.0% | Approximately 99.9% | Liquid electrolyte short-circuited after 41 cycles |
| 0.1 C rate test | 166 mAh g−1 | Not applicable | High | Lower capacity and higher polarization |
| 1.5 C rate test | 126 mAh g−1 | Not applicable | High | Poor rate retention |
The voltage hysteresis evolution further confirmed the interlayer stability. In the bilayer full cell, the hysteresis increased only gradually with minimal fluctuations. I attributed this to a stable Li metal interface and a low accumulation of inactive side products. In the control cells, hysteresis and interfacial resistance increased rapidly, indicating severe interface degradation. The cumulative plated capacity, maximum achievable current density, and capacity utilization of full cells are key metrics for evaluating a solid state battery. Compared with recently reported NASICON-type solid state battery designs that use artificial interlayers, my bilayer interface delivered superior performance across these metrics.
| Metric | Bilayer solid state battery | Typical planar interlayer baseline | Benefit |
|---|---|---|---|
| Long-term symmetric cycling | More than 1400 h | Hundreds of hours | Improved interfacial stability |
| Overpotential at 0.1 mA cm−2 | About 10 mV | Tens of mV | Lower polarization |
| Critical current density | Up to 2.0 mA cm−2 | Often below 1.0 mA cm−2 | Better dendrite tolerance |
| Li+ transference number | 0.60 | About 0.2 | Reduced concentration polarization |
| Li+ diffusion coefficient | 1.23 × 10−9 cm2 s−1 | About 10−10 to 10−11 cm2 s−1 | Faster interfacial transport |
| 0.2 C retention after 100 cycles | 95.5% | Often below 80% | Better cyclability |
| 0.6 C retention after 100 cycles | 95.0% | Frequent early failure | Better high-rate durability |
Puncture, thermal abuse, and component-level function
I conducted a puncture test to determine whether the interlayer could localize damage. I intentionally created a micropuncture on one side of the interlayer while leaving the remaining regions intact. The cell was cycled at 0.1 mA cm−2 and 0.1 mAh cm−2. Pronounced asymmetric overpotential fluctuations appeared within 200 h. The LATP pellet showed a purple degraded region only on the punctured side. Electron microscopy revealed surface roughening in the degraded region, while the protected regions remained white and structurally intact. Energy-dispersive X-ray spectroscopy showed much higher carbon content in the degraded region, which I attributed to air exposure of Li and indirectly confirmed Li dendrite penetration in the unprotected region. This experiment demonstrated that the protection is local and that a breach in the interlayer directly exposes LATP to Li.
| Puncture test feature | Punctured side | Protected side |
|---|---|---|
| Overpotential behavior | Asymmetric fluctuations within 200 h | Stable |
| LATP appearance | Purple degraded region | White and intact |
| Surface morphology | Roughened | Structurally preserved |
| Carbon signal | High, indicating Li exposure | Low |
| Interpretation | Li penetration | Effective protection |
For thermal stability, I heated samples at 300 °C in an Ar-filled box. This temperature is above the melting point of Li and above the onset of violent self-heating reactions reported for LATP in contact with Li. I compared direct LATP–Li contact with a configuration in which the interlayer separated LATP from Li. The unprotected LATP pellet began to crack at 31 s, fragmented severely at 36 s, and underwent violent reaction with spark emission at 37 s. By 39 s, the fragments remained red-hot but the flames gradually extinguished, and the pellet ultimately turned dark purple. In contrast, the interlayer-protected LATP showed no violent reaction throughout heating. Black reaction products formed around the pellet, likely from amorphous carbon produced by ionic liquid decomposition at elevated temperature. These products insulated the LATP pellet from molten Li. Importantly, the protected LATP retained structural integrity and functionality after heating, without short circuiting. I cleaned the retrieved pellet with ethanol and reassembled it into a Li symmetric cell with new interlayers. The cell still cycled stably at 0.1 mA cm−2.
| Time | Unprotected LATP in contact with Li | Interlayer-protected LATP |
|---|---|---|
| Before heating | Intact pellet | Intact pellet with interlayer |
| 31 s | Cracking begins | No violent reaction |
| 36 s | Severe fragmentation | No violent reaction |
| 37 s | Violent reaction and spark emission | No violent reaction |
| 39 s | Red-hot fragments; flames extinguish | Black decomposition products form |
| After cooling | Dark purple degraded fragments | Structurally intact and functional |
| Reassembly test | Not feasible | Stable cycling at 0.1 mA cm−2 |
I also evaluated the role of each component by building cells with a bilayer without Au nanoparticles, with only a mixed-conducting layer, and with only an ion-conducting layer. Without Au nanoparticles, the bilayer could not be efficiently activated, and unstable Li plating and stripping occurred with noticeable voltage fluctuations. With only the mixed-conducting layer, activation was unstable and overpotential increased rapidly because the LATP lacked ion-conducting protection. With only the ion-conducting layer, effective activation was not achieved, and the initial overpotential was relatively high. Because the mixed-conducting layer was absent, Li deposition was not regulated, and sudden overpotential drops appeared during cycling, suggesting Li protrusion growth within the ion-conducting layer and soft short circuits. These results showed that the mixed-conducting layer, ion-conducting layer, and Au nanoparticles serve distinct but complementary functions. Their synergy is essential for stable interfacial kinetics in the solid state battery.
| Configuration | Activation behavior | Cycling response | Failure mode | Conclusion |
|---|---|---|---|---|
| Bilayer with Au | Fast and stable | Low overpotential and long life | None observed within test | Complete synergistic design |
| Bilayer without Au | Poor activation | Voltage fluctuations | Unstable nucleation | Au is needed for lithiophilic seeding |
| Mixed-conducting layer only | Unstable activation | Rapid overpotential rise | LATP degradation | ICL protection is necessary |
| Ion-conducting layer only | No effective activation | High initial overpotential | Soft short circuit | MCL regulation is necessary |
Mechanistic summary of the bilayer solid state battery interface
The combined electrochemical, morphological, spectroscopic, and simulation evidence supports a coherent mechanism. The mixed-conducting layer provides a three-dimensional electron/ion transport network. Its Au nanoparticles lower the Li nucleation barrier and distribute nucleation sites throughout the porous structure. Its carbon nanotubes provide electronic percolation, but this electronic path is confined to the Li-facing side. The ion-conducting layer blocks electron percolation into LATP, preventing Ti4+ reduction and continuous mixed-conducting interphase growth. The ionic liquid confined in the porous network provides fast Li+ transport and stable wetting. The nano-LATP filler provides additional Li+ pathways and reduces polymer crystallinity. During activation, Li deposits on the mixed-conducting fibers and the Li surface, forming an in-situ three-dimensional anode. This architecture accommodates volume changes and maintains contact with the solid electrolyte. During cycling, the SEI remains uniform because the Li+ flux is homogenized and the local current density is reduced. The LATP pellet remains chemically intact. Under thermal abuse, the interlayer physically separates molten Li from LATP and decomposes into an insulating carbon-rich barrier, preventing violent reaction. The result is a solid state battery interface that is simultaneously conductive, protective, compliant, and thermally robust.
| Component | Primary function | Secondary function | Failure if omitted |
|---|---|---|---|
| Mixed-conducting layer | Electron and Li+ transport | Three-dimensional Li host and stress buffer | Poor Li regulation and high polarization |
| Ion-conducting layer | Li+ transport and electron blocking | Chemical protection of LATP | Ti4+ reduction and LATP degradation |
| Au nanoparticles | Lithiophilic nucleation | Lower nucleation overpotential | Difficult activation and uneven deposition |
| Carbon nanotubes | Electronic percolation in MCL | Mechanical reinforcement | Insufficient electron transport in MCL |
| Nano-LATP filler | Ceramic Li+ pathway | Reduced polymer crystallinity | Lower ionic conductivity and higher resistance |
| Ionic liquid | Interfacial wetting and Li+ transport | Confinement within porous network | Voids, poor contact, and high impedance |
| LATP pellet | Solid electrolyte separation | Mechanical dendrite suppression | No solid state battery function |
I can express the effective interfacial current distribution as a competition between ionic transport and electron blocking. In the mixed-conducting layer, the current has both ionic and electronic components:
$$ i_{\mathrm{MCL}} = i_{\mathrm{ion}} + i_{\mathrm{electron}}. $$
In the ion-conducting layer, the electronic component should be approximately zero:
$$ i_{\mathrm{ICL}} \approx i_{\mathrm{ion}}, \quad i_{\mathrm{electron}} \approx 0. $$
This selective transport is what protects LATP while preserving fast Li+ kinetics. It also explains why a single mixed-conducting layer is insufficient: without the ion-conducting layer, electrons can reach LATP. Conversely, a single ion-conducting layer is insufficient because it cannot redistribute the Li+ flux or host Li metal efficiently.
The capacity retention and Coulombic efficiency of the full solid state battery can be summarized as
$$ R_{\mathrm{ret}} = \frac{Q_{100}}{Q_1}\times 100\%, $$
$$ \mathrm{CE} = \frac{Q_{\mathrm{strip}}}{Q_{\mathrm{plate}}}\times 100\%. $$
For the bilayer full cell at 0.2 C, \(R_{\mathrm{ret}}\) was 95.5% and CE was approximately 99.9%. At 0.6 C, \(R_{\mathrm{ret}}\) was 95.0%. These values are high for a NASICON-based solid state battery with a Li metal anode and a LiFePO4 cathode. The overpotential can be approximated as
$$ \eta = \frac{V_{\mathrm{max}} – V_{\mathrm{min}}}{2}. $$
The symmetric cell maintained \(\eta\) near 10 mV at 0.1 mA cm−2, which is among the lowest values reported for comparable systems. The low overpotential is consistent with the high Li+ transference number, high Li+ diffusion coefficient, low interfacial resistance, and uniform Li deposition.
Scalability, sustainability, and practical outlook
I selected electrospinning and magnetron sputtering because they are compatible with roll-to-roll manufacturing and can be scaled beyond laboratory coin cells. Electrospinning is already used in industry for filtration and battery separators. Magnetron sputtering is a standard vacuum coating process. The Au loading is extremely low because only a short sputtering step is used. The ionic liquid loading is also modest and is confined within the interlayer. These features improve sustainability and reduce cost compared with thick ceramic coatings or high-loading noble-metal layers. The LATP solid electrolyte is made from inexpensive precursors and is processed at moderate sintering temperatures. The overall fabrication route is therefore practical for a solid state battery that must eventually be manufactured at scale.
| Practical consideration | My approach | Advantage for solid state battery |
|---|---|---|
| Fabrication method | Electrospinning plus magnetron sputtering | Scalable and compatible with roll-to-roll processing |
| Solid electrolyte | NASICON LATP | High ionic conductivity, air stability, low cost |
| Interface architecture | Bilayer mixed-conducting structure | Combines protection and fast transport |
| Li host | In-situ three-dimensional anode | Accommodates volume change and regulates deposition |
| Noble metal use | Ultrathin Au nanoparticle layer | Low loading and high lithiophilicity |
| Ionic liquid use | Confined in porous fibers | Improved wetting with reduced leakage risk |
| Thermal safety | Interlayer separates LATP from molten Li | Suppresses violent reaction and short circuiting |
I believe the broader lesson is that interfacial engineering for a solid state battery must be multifunctional. A coating that only blocks electrons may increase resistance. A coating that only conducts ions may not accommodate volume change. A three-dimensional host that only stores Li may not protect the solid electrolyte. The bilayer design works because it separates functions spatially. The Li-facing side is responsible for electron transport, Li nucleation, and Li storage. The electrolyte-facing side is responsible for ion transport and electron blocking. The ionic liquid phase provides wetting and fast transport within the pores. The nano-LATP filler reinforces the polymer and adds ceramic ion pathways. The result is a coordinated interface rather than a passive barrier.
I also note that the interface is not static. It evolves during activation and cycling. Initially, the interlayer is a porous fiber mat with Au seeds and ionic liquid. After activation, Li deposits on the mixed-conducting fibers and the Li surface, forming a three-dimensional composite anode. The ion-conducting layer swells slightly but does not become electronically conductive. The LATP surface remains free of Ti3+ after long cycling. This dynamic evolution is beneficial because it creates an intimate contact that is difficult to achieve by simply pressing Li against a rigid ceramic pellet. The in-situ formed anode also reduces the effective current density by increasing the electrochemically active area. This is a key advantage for a solid state battery operating at practical current densities.
Conclusions
I constructed a three-dimensional electron/ion mixed-conducting bilayer interface for a NASICON-based solid state battery. The interface consists of a lithiophilic mixed-conducting layer facing Li metal and an electronically insulating ion-conducting layer facing LATP. The mixed-conducting layer contains polyacrylonitrile, poly(vinylidene fluoride), nano-LATP, carbon nanotubes, LiClO4, and Au nanoparticles. The ion-conducting layer contains the same polymer-ceramic-salt system without carbon nanotubes. The bilayer has a thickness of about 30 µm, a porosity of about 89%, and an electronic conductivity contrast that is essential for protection. The mixed-conducting layer has an electronic conductivity of about 7.0 × 10−3 S cm−1, while the ion-conducting layer has no measurable electronic conductivity. The interlayer is wetted with a small amount of ionic liquid, which fills the pores and provides fast Li+ transport.
The bilayer interface transformed the Li/LATP contact into an in-situ three-dimensional Li host. Symmetric cells with the bilayer cycled for more than 1400 h at 0.1 mA cm−2 with an overpotential of about 10 mV. Control cells wetted with ionic liquid or liquid electrolyte failed after about 842 h and 683 h, respectively. The LATP pellet remained intact after 500 and 1000 h, with no detectable Ti3+ signal. The interfacial resistance remained low, with the SEI-like resistance around 7 Ω cm2 per side. The Li+ transference number reached 0.60, and the Li+ diffusion coefficient reached 1.23 × 10−9 cm2 s−1. The critical current density reached 2.0 mA cm−2, and the electrochemical stability window extended from 0 to 5.5 V. Full cells with LiFePO4 cathodes delivered 166 mAh g−1 at 0.1 C and 126 mAh g−1 at 1.5 C. At 0.2 C, the full cell retained 95.5% of its capacity after 100 cycles with a Coulombic efficiency of about 99.9%. At 0.6 C, it retained 95.0% after 100 cycles. In contrast, the liquid-electrolyte control retained only 60.0% at 0.2 C and short-circuited after 41 cycles at 0.6 C.
Morphological and spectroscopic analysis showed that Li deposited uniformly along the mixed-conducting fibers, forming a moss-like morphology without protrusions. The SEI on the three-dimensional anode was uniform and rich in LiF, and its composition remained stable after 1000 h. Finite element simulation showed that the bilayer mitigated electric field concentration and Li+ depletion, leading to uniform deposition and preserved SEI integrity. Puncture testing showed that damage to the interlayer locally exposed LATP and caused degradation only on the punctured side, confirming the protective role of the intact bilayer. Thermal testing at 300 °C showed that the interlayer suppressed the violent reaction between LATP and molten Li. The unprotected LATP cracked, fragmented, and reacted with spark emission, while the protected LATP remained structurally intact and functional. These results establish the bilayer interface as a multifunctional strategy for stable, safe, and scalable NASICON-based solid state batteries.
In my assessment, the most important design principle is spatial separation of transport functions. The solid state battery interface should not be asked to perform all functions in one homogeneous layer. Instead, a mixed-conducting layer should manage electrons, Li nucleation, and Li storage, while an ion-conducting layer should manage Li+ transport and electron blocking. A lithiophilic seed layer should lower nucleation barriers. A confined ionic liquid phase should ensure wetting and fast ion transport. A ceramic filler should reinforce the polymer and provide additional Li+ pathways. When these functions are combined in a bilayer architecture, the interface can simultaneously stabilize Li metal, protect the NASICON electrolyte, suppress dendrites, accommodate volume change, and improve thermal safety. This approach provides a practical route for advancing solid state battery technology toward high energy density, long cycle life, and improved safety.