Future lunar surface architectures, specifically in the polar regions, must simultaneously satisfy three tightly coupled constraints: (i) power availability under highly anisotropic, seasonally modulated illumination; (ii) thermal management in the presence of steep radiative gradients and permanently shadowed regions (PSRs); and (iii) mechanical robustness and constructability using in situ resources. Long-baseline simulations with Lunar Orbiter Laser Altimeter (LOLA) topography have shown that only a small fraction of polar terrain achieves near-continuous or quasi-continuous illumination over a precessional cycle [1,2], while radiometry from the Diviner Lunar Radiometer Experiment (Diviner) reveals strong spatial variability in regolith thermophysical properties that governs diurnal and seasonal temperature swings [3]. These constraints are particularly acute along narrow ridges and crater rims where illumination, access to PSRs, and engineering feasibility must be traded against each other. In parallel, the National Aeronautics and Space Administration (NASA) Open Science Data Repository (OSDR) has emerged as an open-science model for spaceflight biology, curating space-flown and ground-based analog datasets and demonstrating how multiomics, physiological, and imaging data from low-sample-size missions can be integrated to advance biofilm biology and enable artificial intelligence (AI) and machine learning (ML) analysis pipelines [4].

In situ resource utilization (ISRU) has progressed from conceptual studies to experimental demonstrations of solar-driven sintering and additive manufacturing with lunar regolith simulants. Concentrated sunlight has been used to fabricate three-dimensional bricks and structural elements by directly sintering regolith layers, without imported binders [5]. Recent reviews of lunar regolith formation technologies emphasize that sintered or cast regolith could provide radiation shielding, structural shells, and thermal mass for surface infrastructure, provided that process windows and mechanical properties can be reliably controlled under lunar conditions [6]. These developments motivate microhabitat concepts in which thin, sintered shells act as structural and radiative envelopes, while internal liners and inserts host functional subsystems.

Against this background, microbial life is not an optional consideration but an intrinsic component of any long-duration human presence in space. Biofilms (surface-associated, matrix-embedded microbial communities) are widely recognized as a dominant mode of microbial life on Earth and an important contributor to the early fossil record [4]. In the context of spaceflight, environmental surveys and longitudinal studies at the International Space Station (ISS) have shown that confined habitats develop characteristic microbiomes shaped by crew, materials, and operational patterns [7,8]. Reviews focused on biofilms in space highlight that biofilm formation can compromise hardware, alter corrosion and fouling of stainless-steel components, and modulate health risks to crew, while also offering potential benefits for bioprocesses and life-support functions [9,10].

Targeted spaceflight experiments have begun to probe biofilm formation under microgravity and spacecraft-relevant conditions. The Space Biofilms project and the Bacterial Adhesion and Corrosion (BAC) investigations examine how mono- and polymicrobial biofilms grow on stainless steel, respond to microgravity, and interact with candidate biocides and surface treatments, using organisms such as Pseudomonas aeruginosa and Penicillium rubens as model systems [11,12]. These datasets provide quantitative constraints on attachment rates, extracellular polymeric substance (EPS) production, stress-response pathways, and material-specific differences in colonization that are directly relevant to engineered surfaces in space habitats.

The present work explores a deliberately constrained problem at the intersection of these domains: the design space of sintered-regolith microhabitats whose internal surfaces serve as controlled interfaces for biofilm formation and suppression. Rather than proposing a full-scale habitat architecture, we focus on a single class of geometries (spherical-cap sandwich shells) anchored to polar ridge sites with favorable illumination and regolith properties. We then couple: (i) a generative geometric model that produces a family of shells parameterized by radius, height, and thickness; (ii) a one-dimensional radial conduction model that evaluates internal climate under polar forcing; and (iii) a reduced biofilm model that maps an illumination-coupled, engineered internal ultraviolet (UV) control variable and material properties into a finite-horizon biofilm response and a scalar controllability index.

The goal is not to produce operational design recommendations, but to construct a computational scaffold that links open planetary and microbiological measurements to architecturally legible variables. The framework is therefore evaluated as a conceptual and exploratory screening system. Its outputs are benchmarked against published ranges and qualitative material ordering, and its sensitivity to the principal assumptions is quantified; however, no claim of experimental calibration, structural qualification, or mission readiness is made. By explicitly tracing the mapping from LOLA, Diviner, and Lunar Reconnaissance Orbiter Camera (LROC) products and ISS biofilm experiments into effective biofilm and thermophysical parameters, the analysis identifies which combinations of site, shell geometry, liner treatment, and controlled internal UV exposure warrant higher-fidelity simulation or laboratory validation.

2.1. Geometry and Generative Model

The microhabitat is represented as a spherical-cap sandwich shell with base radius R, rise h, structural thickness t, rim width w, and rim rise $w_r$ (Figure 1). This abstraction is consistent with classical thin-shell formulations and recent finite-element verification studies of spherical domes [13,14]; no structural qualification is claimed here. The radius of curvature of the outer cap is

and the cap terminates at

The inner surface is generated by the reduced dimensions $R_i=R-t$ and $h_i=h-t$ , giving

Here, R denotes the base radius, whereas S_o and S_i denote the outer and inner radii of curvature, respectively. The outer and inner meshes are sampled over meridional and azimuthal coordinates and exported as a watertight triangulated object for inspection and downstream analysis. Local stiffeners, apertures, seals, tie-down hardware, and pressure-vessel details are excluded because the model represents a robotic experimental microhabitat rather than a crew-rated habitat.

For thin shells, the cap area and first-order shell volume are

The corresponding sintering energy scale as

where $E_s$ is the effective specific process energy and ${\rho }_{\mathrm{reg}}$ is the processed-regolith density. Equation (6) is used only for normalized geometry sensitivity; detailed process planning would require toolpath, heat-loss, porosity, and rejected-material terms.

2.2. Site Screening and Horizon-Based Polar Illumination

Let $z\left(\mathbf{x}\right)$ denote a local LOLA digital elevation model. For a candidate point ${\mathbf{x}}_0$ , the azimuthal horizon elevation is

where ${\widehat{\mathbf{e}}}_{\psi }=(\cos \psi ,\sin \psi )$ . The radial maximum-slope construction follows established horizon-based polar illumination methods [1,17,18,19,20].

Near the lunar poles, the reduced solar-elevation track is represented as

where $\epsilon =1.{54}^{\circ }$ and ${\theta }_q$ samples the slow precessional phase. Visibility is

with a safety margin $\delta =0.1°$ . Fractional illumination is

The reduced solar track omits libration, finite-disc integration, detailed ephemerides, and time-dependent terrain shadowing. It is used for relative site ranking, not for operational power prediction.

Candidate pixels are pre-filtered by slope and local relief. A typical evaluation uses $N_{\psi }=720$ , a radial search of up to 10 km, and eight precessional phases. The computational complexity is $\mathcal{O}\left(N_{\psi }{N}_r\right)$ per site. LOLA, Diviner, and LROC products provide topographic, thermophysical, and local-context constraints, respectively [3,19,20].

2.3. Radial Thermal Model and Internal Climate

The shell is reduced to L concentric spherical layers representing sintered regolith, bulk regolith, liner, and internal protective layers, where L denotes the total number of layers in the thermal model. In layer i,

where ${\alpha }_i\left(T\right)=k_i\left(T\right)/\left({\rho }_i{c}_{p,i}\left(T\right)\right)$ . The factor $2/r$ is required for spherical symmetry. Temperature and heat flux are continuous at every material interface. The interior and exterior boundary conditions are

where ϵrad is the effective exterior hemispherical emissivity and σ is the Stefan–Boltzmann constant. The subscript distinguishes emissivity from the lunar-pole angular parameter ε defined in Section 2.2. The absorbed external flux is

where ${\alpha }_{\mathrm{bol}}$ is bolometric albedo, S_⊙ is the incident solar irradiance (solar flux) before albedo losses, and v_⊙(t) is the time-dependent solar-visibility factor. Exterior radiation may be linearized around Tref for implicit time integration. The radial discretization produces a tridiagonal system solved by an implicit or Crank–Nicolson scheme. Material properties are constrained by Diviner-derived behavior and laboratory measurements of regolith and sintered simulants [3,21,22,23,24].

For first-order sensitivity, the thin-layer thermal resistance is

so the conductive temperature drop obeys $\Delta T\propto qt/k$ . This scaling is used to quantify the direction and magnitude of uncertainty before three-dimensional analysis.

2.4. Sintering Energy and Autonomous Construction

For processed mass m, total energy is

Published microwave and solar-sintering energy estimates span approximately $3\times {10}^3$ –$2\times {10}^4$ J g⁻¹, depending on peak temperature, coupling efficiency, porosity, and process path [5,25,26,27,28,29]. Available power is represented as

where A_aperture is the effective collector or concentrator aperture area, and the efficiency factors describe conversion, pointing, and dust derating. These factors are held constant in the present screening analysis. Robotic emplacement, excavation, feedstock grading, layer bonding, dust accumulation, maintenance, and repeated thermal cycling are not dynamically modeled.

2.5. Biofilm Attachment and Growth Under Controlled Radiative Stress

The biological subsystem is represented by two state variables:

where A is attached biomass, B is biofilm biomass, $C_b$ is a fixed planktonic-source proxy, $k_a$ is a baseline attachment coefficient, $k_d$ is an effective detachment/death rate, ${\Phi }_0$ is a material-dependent attachment factor, ${\mu }_{\max }$ is a maximum specific growth rate, K_s is the Monod half-saturation constant with respect to attached biomass A (with the same biomass units as A), and ${\beta }_{\varphi }{D}_{\mathrm{uv}}$ is the effective radiative-damage rate. The model follows low-order attachment and Monod-type formulations but intentionally omits substrate depletion, carrying capacity, multispecies interactions, flow, EPS mechanics, mutation, and spatial gradients [30,31,32,33].

The normalized internal dose is

where ${\chi }_{\mathrm{uv}}$ combines engineered UV delivery, duty cycle, optical attenuation, shielding, and geometry. The baseline scenario uses ${\chi }_{\mathrm{uv}}=1$ . Equation (20) does not imply that exterior lunar UV directly irradiates the internal liner; the microhabitat is assumed to contain a controlled UV source or an optically managed exposure path. Consequently, $D_{\mathrm{uv}}$ is dimensionless and cannot be compared directly with a laboratory fluence in mJ cm⁻² without independent radiometric calibration.

For fixed $D_{\mathrm{uv}}$ , the attached biomass converges to

The asymptotic net growth exponent for B is

Because Equation (19) is linear in B, it has no finite positive biomass equilibrium when ${\lambda }_B>0$ . The term “steady-state biomass” is therefore not applicable in this regime; instead, we report the finite-horizon terminal window metric

where $T=240$ h, $t_n=n\Delta t$ , and $\Delta t=0.1$ h. Thus, the reported terminal-window value is the arithmetic mean of the 241 numerical samples from 216.0 to 240.0 h, inclusive. The same definition is used for ${\overline{A}}_T$ . This finite-horizon formulation preserves the model dynamics while preventing transient, finite-time exponential growth from being misidentified as a biological steady state. The critical dose $D_c$ is defined by ${\lambda }_B\left(D_c\right)=0$ .

2.6. Translation of OSD-554 Data into Biological Priors

The OSD-554 normalized-count tables for stainless-steel-like (SS) and liquid-infused-surface (LIS) coupons are separated into flight and ground replicates [11]. For each material, the pipeline computes gene-wise log₂ fold changes between flight and ground means, selects the ten genes with the highest absolute fold changes, and defines a condition score $s_c$ as the mean normalized expression of those genes. Scores are standardized across the four gravity–material conditions:

The standardized score is mapped monotonically to effective parameters:

The rates are expressed in h⁻¹ where applicable. Equation (28) is a prior heuristic rank-preserving transformation, not a statistical or mechanistic calibration. It encodes the relative ordering present in the selected transcriptomic score and the experimentally observed tendency of surfaces similar to LIS to reduce attachment relative to stainless steel [10,12].

2.7. Trade-Off Indices and Sensitivity Protocol

The physical screening variables are

where τ is the reference processing-time scale (with units of time) used to nondimensionalize the construction energy to- power requirement. The biofilm controllability index is defined comparatively as

where $B_{\mathrm{ref}}$ is the largest ${\overline{B}}_T$ within the explicitly stated comparison set. Because this normalization can saturate when all compared biomasses are very small, ${\log }_{10}{\overline{B}}_T$ and the critical dose $D_c$ are reported alongside ${\mathcal{C}}_{\mathrm{bio}}$ .

A composite score may be formed as

but no universal weights are asserted. This study emphasizes Pareto structure and one-at-a-time sensitivity analysis rather than a single preferred ranking. Biological parameters are perturbed by $\pm 20\%$ , and the resulting $D_c$ is recomputed. The thermal scaling is evaluated for $\pm 25\%$ perturbations in thickness, conductivity, and imposed heat flux of the shell. Geometry-energy sensitivity follows Equation (6).

2.8. Verification Hierarchy and Scope of Validation

The framework uses four levels of verification. First, the geometry is checked against the analytical spherical-cap area and volume relations. Second, illumination outputs are compared with the range and spatial scarcity of highly illuminated polar ridges reported by studies based on LOLA [1,2,16,18]. Third, the radial thermal solver is checked against analytical steady-state spherical conduction and energy-balance limits before application to temperature-dependent properties. Fourth, biological outputs are checked for numerical convergence, analytical consistency with Equations (22)–(23), and qualitative SS–LIS ordering reported in ISS-relevant studies [10,11,12].

These checks establish internal consistency and correspondence with the literature, not experimental validation under simulated lunar conditions. A definitive validation campaign would require controlled measurements of SS and LIS coupons under specified nutrient, humidity, temperature, microgravity analog, and UV-fluence, followed by estimation of $k_a$ , $k_d$ , ${\mu }_{\max }$ , and ${\beta }_{\varphi }$ from replicated time-series data.

2.9. Software, Reproducibility, and Data

The complete Hypertext Markup Language (HTML) for the parametric shell renderer and Python source code (LOLA preprocessing, horizon and illumination analysis, OSD-554 prior construction, finite-horizon biofilm integration, controllability-index calculation, and dose-response sweep), execution sequence, and module descriptions are available in the external reproducibility archive [34].

3.1. Benchmark Correspondence

The selected polar candidate is the grid cell at row 197 and column 75, with $f_{\mathrm{illum}}=0.755556$ . This value lies in the high-illumination regime used to screen narrow polar ridges and is consistent with the conclusion of long-baseline LOLA studies that such conditions occur only over a small fraction of polar terrain [1,2,16]. This comparison is a correspondence benchmark; the reduced solar-track model does not replace full ephemeris-based illumination analysis.

The biological parameter transformation reproduces the expected material ordering: the SS-like condition has a larger attachment factor and smaller effective detachment/damage coefficients than the LIS-like condition (Table 1). This is consistent with the direction of reported differences between conventional stainless steel and low-adhesion or liquid-infused surfaces, but it does not constitute kinetic calibration [10,11,12].

3.2. Finite-Horizon Biofilm Response at the Selected Site

With the baseline coupling $D_{\mathrm{uv}}=f_{\mathrm{illum}}=0.755556$ and $T=240$ h, the terminal-window metrics are presented in Table 2:

Both cases are in a negative-net-growth regime. The value ${\mathcal{C}}_{\mathrm{bio}}=0$ for micro_SS does not indicate uncontrolled biological accumulation; it indicates that micro_SS is the larger of two already vanishing finite-horizon responses under the pairwise normalization. The log-scale biomass and the critical-dose analysis are therefore more informative than the bounded index in this regime.

3.3. Critical-Dose and Biological Sensitivity Analysis

Solving ${\lambda }_B\left(D_c\right)=0$ gives

The selected-site value $D_{\mathrm{uv}}=0.755556$ exceeds both thresholds (Table 3). Figure 2 also shows that the sharp transition in the dose-response sweep is an analytical consequence of the sign change in ${\lambda }_B$ , not evidence of a measured sterilization threshold.

For micro_SS, ${\beta }_{\varphi }$ produces the largest one-at-a-time threshold excursion, followed by ${\mu }_{\max }$ and ${\Phi }_0$ ; $k_d$ is less influential around the baseline. For micro_LIS, all four parameters materially affect the low threshold, with ${\mu }_{\max }$ , $k_d$ , and ${\beta }_{\varphi }$ producing comparable relative shifts. The separation between the two baseline thresholds remains substantially larger than the perturbation intervals, preserving the relative SS–LIS ordering under this local sensitivity test.

3.4. Geometry-Energy and Thermal Sensitivities

At fixed aspect ratio $\kappa =h/R$ , density, and specific process energy, Equation (6) gives

Consequently, a 25% increase in base radius raises energy by 56.25%, whereas a 25% increase in thickness raises energy by 25%. Figure 3 shows this normalized trade-off without imposing an unsupported process efficiency or absolute shell size.

For the first-order thermal-resistance relation $\Delta T\propto qt/k$ , the parameter sensitivities are presented in Table 4:

This scaling identifies conductivity uncertainty as asymmetric in its effect on thermal drop. It does not resolve transient three-dimensional heat bridges, penetrations, internal equipment, or radiative exchange among interior surfaces.

The biological output is interpreted using the governing ordinary differential equations (ODEs) without modification. Large biomass values in the ${\lambda }_B>0$ regime represent finite-time exponential responses rather than steady states, whereas extremely small values occur when radiative damage makes ${\lambda }_B<0$ . Reporting ${\overline{B}}_T$ , ${\lambda }_B$ , and $D_c$ distinguishes numerical horizon effects from equilibrium behavior. The critical-dose analysis shows a materially wider suppression margin for the LIS-like prior than for the SS-like prior, and the one-at-a-time perturbations do not reverse that ordering.

The dimensionless dose remains the largest limitation in connecting site illumination to biofilm response. Exterior illumination determines the available electrical or optical power, but internal microbial exposure depends on source efficiency, duty cycle, attenuation, shielding, surface orientation, dust, and operational constraints. The coupling coefficient ${\chi }_{\mathrm{uv}}$ makes this assumption explicit. A site with $f_{\mathrm{illum}}=0.756$ does not automatically impose $D_{\mathrm{uv}}=0.756$ on an internal liner; that equality is the baseline engineered-control scenario. Direct comparison with published ultraviolet-C (UVC) or ultraviolet-C light-emitting diode (UV-LED) inactivation experiments requires calibration of ${\chi }_{\mathrm{uv}}$ against wavelength-resolved irradiance and exposure time [35,36,37,38,39,40].

The geometry-energy sensitivity is quantitative but normalized. It establishes that base radius is a stronger first-order energy driver than thickness because of the $R^{2}t$ scaling, while the absolute construction budget remains dominated by $E_s$ , density, porosity, heat loss, concentrator efficiency, and toolpath. Similarly, the thermal sensitivity table is a resistance-level diagnostic rather than a substitute for three-dimensional analysis. The radial model cannot represent local heat bridges at penetrations, asymmetric illumination, contact resistance, internal equipment, multilayer delamination, or azimuthal heat spreading.

Operational challenges are not represented by the resolved state variables. Robotic emplacement and excavation are assumed feasible; dust mitigation is represented only by a constant derating factor; and liner degradation, abrasion, radiation aging, leakage, cleaning cycles, and surface-property drift are not modeled. The biological priors represent fixed surface states. Any degradation that changes roughness, wettability, lubricant retention, or optical transmission would require time-dependent ${\Phi }_0$ and ${\beta }_{\varphi }$ .

The OSD-554 transformation is transparent but weakly constrained. Selecting the ten largest absolute flight-to-ground fold changes and averaging normalized counts does not identify causal adhesion, growth, detachment, or UV-response parameters. The monotonic mapping supplies reproducible priors for comparative computation only. Experimental validation should therefore estimate parameters from replicated time-series measurements rather than from transcript abundance alone. A suitable campaign would cross material class, gravity analog, temperature, nutrient level, humidity, and calibrated UV fluence, with independent fitting and held-out validation.

Within these limitations, the framework provides a reproducible foundation for eliminating clearly unfavorable combinations and defining the experiments or high-fidelity simulations needed next. Its defensible outputs are relative rankings, normalized scaling laws, uncertainty intervals, and explicit assumptions. Mission-level claims about structural safety, thermal habitability, sterilization, construction duration, or microbial risk are outside its present evidentiary scope.

The framework integrates spherical-cap geometry, polar illumination screening, radial thermal analysis, sintering-energy scaling, and finite-horizon biofilm dynamics without embedding source-code listings in the manuscript. The code is instead provided through an external reproducibility archive [34].

Two methodological features are particularly important. The thermal equation includes the spherical $2/r$ term, and absorbed solar flux is expressed with $(1-{\alpha }_{\mathrm{bol}})$ . The biological output is characterized using a 240 h terminal window biomass metric, the asymptotic net-growth exponent, and the critical normalized dose.

For the reported priors, $D_c=0.396127$ for the SS-like liner and $D_c=0.052941$ for the LIS-like liner. Under one-at-a-time parameter perturbations of $\pm 20\%$ , the threshold intervals remain separated, preserving the qualitative ordering of the materials. Normalized construction scaling gives $E_{\mathrm{sint}}\propto R^{2}t$ at fixed aspect ratio, while first-order thermal resistance gives $\Delta T\propto qt/k$ .

These results support exploratory comparison and experimental design. They do not provide validated lunar-habitat performance, calibrated UV requirements, or operational biofilm-risk predictions. The next required step is joint physical and biological validation under controlled lunar-analog boundary conditions.

The author was solely responsible for Conceptualization, Methodology, Software, Validation, Formal Analysis, Investigation, Data Curation, Visualization, Writing—Original Draft Preparation, Writing—Review and Editing, and Project Administration. The author has read and agreed to the published version of the manuscript.

The data products and code used in this study are publicly available as follows: LOLA Topography: The LOLA dataset provides the gridded digital-elevation and polar-topography products used for horizon and illumination screening [41]. Diviner Thermophysical Data: The NASA PDS dataset provides the gridded thermal and thermophysical products used to constrain the thermal model [42]. LROC Imaging: The NASA PDS bundle provides calibrated LROC images and local site context [43]. OSD-554 Biological Data: The NASA Open Science Data Repository (OSDR) study provides the normalized-count tables and study metadata used to construct the SS- and LIS-like biological priors [11]. Code Archive: The external reproducibility archive contains the parametric shell renderer, planetary-data preprocessing, horizon and illumination analysis, biological-prior construction, biofilm integration, controllability-index calculation, dose-response sweep, module descriptions, and execution sequence [34]. All NASA records are freely accessible. No access restrictions were encountered; reuse remains subject to the applicable repository terms and record-specific metadata.

The author declares no conflicts of interest.

The study did not receive any external funding and was conducted using only institutional resources.

The author acknowledges the NASA Open Science Data Repository and NASA planetary-data archives for providing publicly accessible datasets used in this study. NASA had no involvement in the design, analysis, interpretation, or conclusions of this work.

During the revision of this manuscript, the author used OpenAI ChatGPT (GPT-5.6 Thinking) for English language editing. The author reviewed and verified all AI-assisted content and takes full responsibility for the accuracy, originality, integrity, and scientific content of the manuscript.