Ambient carbon dioxide concentration correlates with SARS-CoV-2 aerosolability and infection risk

The study investigates how environmental factors—particularly ambient carbon dioxide concentration and relative humidity—affect the infectivity (aerosolability) of aerosolized SARS-CoV-2 and the consequent risk of COVID-19 transmission. Transmission via inhalation of infectious respiratory aerosol requires sufficient airborne viral dose; therefore, environmental controls on aerosolized viral load over time are critical. Prior work indicates that aerosol pH, governed by gas-particle partitioning of bicarbonate/CO₂, drives rapid loss of infectivity after aerosol generation. The authors aim to quantify how moderate increases in ambient CO₂ alter aerosol pH and viral survival across variants (Delta vs Omicron BA.2), and to translate measured decay dynamics into infection risk using a Wells–Riley framework, underscoring the role of ventilation and indoor CO₂ levels.

Background reports identify inhalation of aerosolized SARS-CoV-2 as an important transmission route, with minimal infectious dose depending on immunity and prior exposure. Non-pharmaceutical interventions (ventilation, air cleaning) target the removal or dilution of infectious aerosol. Previous experimental work showed that as SARS-CoV-2 variants evolved to Delta, they became more sensitive to alkaline conditions and less aerosolizable, consistent with rapid infectivity loss associated with aerosol-phase processes (e.g., efflorescence at low RH and pH elevation driven by CO₂/bicarbonate partitioning). Observational evidence from super-spreader events suggested potential for long-range airborne transmission, while mitigation effectiveness indicated near-field controls also matter. Conventional risk models often assume a long half-life (~1.1 h), but emerging measurements reveal complex, multi-phase decay immediately after aerosolization, motivating refined modeling of transmission risk that incorporates short-time decay dynamics and environmental drivers (CO₂, RH).

Virus and cells: Vero E6 cells stably expressing TMPRSS2 and Vero E6 cells expressing human ACE2 and TMPRSS2 (VAT) were cultured in DMEM (high glucose) with 10% FBS at 37 °C, 5% CO₂, plus antibiotics and L-glutamine. Viral stocks of SARS-CoV-2 B.1.617.2 (Delta; GISAID: EPI_ISL_731370) and Omicron BA.2 were prepared using VAT cells (MOI 0.01 or 0.1), incubated (Delta 24 h; BA.2 72 h), clarified (250 g, 10 min), filtered (0.22 µm), aliquoted and stored at −80 °C. Delta achieved higher titer (TCID50/ml 3.4×10^7) than BA.2 (3.1×10^6).

Aerosol infectivity measurements (CELEB5/CELEBS): Aerosolized droplets were generated on-demand (~25 µm initial radius; final 5–10 µm depending on composition and RH) and electrodynamically levitated. A laminar airstream of controlled composition, RH, temperature, and CO₂ flowed over the droplets. Air sources included CO₂-free compressed air (<0 ppm CO₂) or laboratory air; RH was monitored (Honeywell HHI-4602-C2), and CO₂ measured (GSS E4500 Mini HD, ±200 ppm). Typical levitation times were under 5 min. Infectivity per droplet was quantified post-levitation by infecting Vero E6/TMPRSS2 cells in 96-well plates and scoring cytopathic effect after 3–5 days. Infectivity was normalized to droplets levitated for <5 s under conditions with no measurable loss (reference: 40% RH and near-baseline CO₂), using: Infectivity = Virus per Droplet(T,RH,CO₂) / Virus per Droplet(<5 s, 90%, 500 ppm).

Environmental conditions: Experiments spanned RH values of 40%, 80%, and 90%, and a range of ambient CO₂ concentrations representative of indoor environments (e.g., 500, 800, 3000, 6500 ppm), as well as CO₂-free conditions for mechanistic probing. MEM was used as a droplet surrogate for saliva based on matched physicochemical behavior (salt and bicarbonate content, pH rise upon aerosolization, phase state across RH).

Bulk-phase stability: Variant stability at high pH (pH 11) was assessed in DMEM with 2% FBS at 20 °C using TCID50 assays and immunostaining. After incubation at pH 11, samples were neutralized (diluted into pH 7 DMEM) prior to infection and quantification.

Risk modeling: Measured short-time decay profiles from CELEBS were incorporated into a Wells–Riley model to estimate indoor transmission risk in a well-mixed room (e.g., a 300 m² classroom) under different ventilation rates (0.5 vs 8 air changes per hour). The model considered the time evolution of infectious quanta considering ventilation removal and variant-specific viability decay, breathing rate (7.5 L/min), room occupancy, and ambient CO₂.

Statistics: Error bars represent standard error. Two-sample t-tests (equal variance, one-sided where specified) assessed differences across conditions; when multiple variables were evaluated, two-way ANOVA examined interactions (e.g., CO₂ × RH). Reported p-values include significance at p ≤ 0.05/0.005 thresholds for select comparisons.

• Variant differences: Omicron BA.2 exhibited slower decay than Delta at high RH (90%). At 5 min, BA.2 maintained ~17-fold greater viable aerosolized viral load than Delta. Below the droplet efflorescence threshold (low RH), both variants showed an immediate ~50% loss of infectivity associated with efflorescence, followed by slower decay.
• Bulk-phase pH sensitivity: In alkaline bulk solution (pH 11), BA.2 was more resistant than Delta, consistent with higher aerosol stability at elevated pH. Linear fits indicated significantly different decay rates between BA.2 and Delta (e.g., p = 0.00020 for one assay; p = 0.027 with n = 30 for another), supporting variant-specific pH sensitivity as a driver of aerosolability.
• CO₂ effects on aerosol infectivity: Elevating ambient CO₂ significantly increased aerosol infectivity across variants at 80% RH (measured at 120 s). Reported p-values for comparisons across CO₂ levels included 0.035, 0.025, 0.032, and 0.001 for contrasts among 500, 800, 3000, and 6500 ppm conditions, indicating a strong CO₂ dependence of aerosolability. At RH < 80%, a moderate CO₂ increase approximately doubled the remaining aerosolized viral load at 15 s; these differences were statistically significant by two-sample t-tests. At 90% RH, no decay was detected at 15 s (droplets still evaporating), so CO₂ effects at that moment were minimal.
• Comparative importance of CO₂ vs RH: Depending on variant pH sensitivity and solute composition, ambient CO₂ can modulate viral aerosolability more strongly than RH, particularly in the early post-generation period.
• Indoor concentrations: Poorly ventilated indoor environments commonly reach >2000 ppm CO₂ and can exceed 5000 ppm, implying that real-world CO₂ elevations can materially increase viral aerosol survival.
• Risk modeling: Incorporating measured decay dynamics into a Wells–Riley framework showed that higher ambient CO₂ shortens the time until there is a 50% chance that at least one susceptible individual becomes infected, particularly in poorly ventilated spaces. Increasing CO₂ approximately halved the time to a 50% transmission probability in modeled non–well-ventilated scenarios, whereas increased ventilation (e.g., 8 ACH) markedly reduced risk. Ventilation mitigates risk by both lowering indoor CO₂ (promoting faster viral decay) and physically removing infectious aerosol.

Findings support the hypothesis that aerosol pH dynamics—controlled by gas–particle partitioning of bicarbonate and CO₂—govern early-time SARS-CoV-2 viability in aerosols. Elevated ambient CO₂ drives droplet pH toward neutral, suppressing the alkaline rise that inactivates virus, thereby increasing aerosolability. This effect can outweigh RH in the early post-generation period and is variant dependent, with BA.2 showing higher resilience. Translating these dynamics through Wells–Riley modeling demonstrates that CO₂ not only serves as a proxy for ventilation but also directly modulates infectious aerosol survival, compounding transmission risk in high-CO₂ indoor environments. The results emphasize ventilation as a dual-action control: lowering CO₂ (accelerating viral decay) and expediting aerosol removal. The work also reframes seasonal patterns of respiratory infections: seasonal indoor CO₂ variations, alongside RH, may contribute to population-level seasonality. Additionally, the study reveals a triphasic aerosol decay profile (Lag, Dynamic, Slow phases), with high CO₂ truncating the Dynamic phase and sustaining infectivity longer, offering a mechanistic basis for observed transmission patterns and informing more accurate risk models.

This study demonstrates that moderate increases in ambient CO₂ (e.g., from 500 to 800 ppm) can substantially increase the aerosolability of SARS-CoV-2, often more than changes in RH, by altering aerosol pH dynamics. Omicron BA.2 is more aerosol-stable than Delta, consistent with enhanced resistance to alkaline conditions in bulk assays. When incorporated into a Wells–Riley framework, these measurements show that elevated indoor CO₂ materially increases infection risk, especially in poorly ventilated spaces, while increased ventilation strongly mitigates risk by both lowering CO₂ and removing infectious aerosol. The findings underscore the critical importance of maintaining low indoor CO₂ through effective ventilation and inform interpretation of prior aerosol studies where bicarbonate/CO₂ chemistry may have influenced results. Future research should: (i) quantify droplet size–dependent neutralization and decay dynamics; (ii) extend measurements to diverse respiratory viruses and fluid compositions; (iii) refine risk models to include short-range transmission and triphasic decay; and (iv) elucidate the microbiological mechanisms of pH sensitivity and the slow decay phase.

• Modeling assumptions: The Wells–Riley analysis assumes a well-mixed environment and focuses on small aerosol modes for quanta uniformity; CELEBS measurements used droplets initially in the oral mode (~50 µm), introducing scale and mixing uncertainties. Deposition of larger droplets was not explicitly modeled.
• Limited droplet size exploration: The effect of droplet size on neutralization and viability was only explored to a limited extent; broader size-resolved measurements are needed.
• Short-time window: Aerosol infectivity measurements emphasized early times (≤5 min), capturing critical rapid decay but not full long-term behavior under all conditions.
• Proxy fluid: MEM was used as a saliva surrogate; while physicochemical properties are similar, biological differences may remain.
• Environmental control precision: CO₂ sensor accuracy (±200 ppm) and environmental set-point uncertainties may influence exact quantitative thresholds.
• Generalizability: Results focus on Delta and Omicron BA.2; other variants and pathogens may exhibit different pH sensitivities and responses to CO₂/RH.
• Risk underestimation near-field: Conventional Wells–Riley may underestimate short-range (near-field) transmission where rapid early decay competes with high local concentrations; this was not explicitly modeled.