Indoor Air Quality in Charcoal-Grilling Restaurants of Baghdad: Short-Term Monitoring of PM2.5, PM10, CO2, and CO

2.1 Study locations

The gaseous pollutant measurements of PM, carbon monoxide (CO), carbon dioxide (CO₂) were examined and corresponding micro-climatic conditions like temperature and relative humidity were obtained from 5 different restaurants ranging from the time frames of December 03 to December 09 with approximately 3 days in between each measuring point (Table 2) located within the Al-Karkh District of Baghdad where most of the food flavourings derive from wood being used for cooking due to the large concentration of restaurants which serve foods cooked using charcoal in this densely populated urban commercial area that is reflective of what is found in most areas within Baghdad's food service industry. All five restaurants rely exclusively on charcoal as the cooking fuel and operate traditional open-flame grilling without mechanical ventilation systems.

Table 2. Study sites with geographic coordinates and facility descriptions

2.2 Instrumentation

Three portable multi-sensor instruments were employed to measure the target indicators (Table 3). All instruments were manufactured in China and sourced from the same supplier batch to minimize inter-instrument calibration variability. Prior to the field campaign, the devices were checked on 2 December 2024 (one day before the first sampling event). For the CO₂ meter (KT-600), the instrument was allowed to equilibrate in outdoor ambient air (assumed near-background conditions) before each sampling session, and the zero/baseline reading was verified according to the manufacturer’s instructions. For the CO monitor (GX-2009), a fresh-air zero check was performed before each sampling session; where applicable, the instrument’s auto-zero function was used after warm-up. A formal span-gas calibration was not performed in the field because certified calibration gases were not available during the campaign; therefore, instrument performance relied on the manufacturer’s factory calibration and pre-use verification checks. For the PM monitor (AQ-9901), zero verification was performed by operating the instrument in clean outdoor air after stabilization and confirming a stable low baseline prior to indoor measurements.

Table 3. Instruments used for measuring gaseous and particulate pollutants

The nominal sensor specifications reported by the manufacturer were as follows: CO₂ (KT-600): range 400–5,000 ppm, accuracy ± (50 ppm + 5% of reading), response time ≤ 60 s; CO (GX-2009): range 0–500 ppm, accuracy ±5% of reading (or ± 5 ppm, whichever is greater), response time ≤ 30 s; PM₂.₅/PM₁₀ (AQ-9901): range 0–999 μg/m³, resolution 1 μg/m³, response time ≤ 10 s. The detection principles are as follows: the CO₂ sensor operates on non-dispersive infrared (NDIR) absorption; the CO sensor uses an electrochemical detection cell; and the PM sensor uses laser particle-counting (optical light-scattering). It is acknowledged that consumer-grade optical PM sensors can overestimate mass concentrations under high-humidity conditions or in the presence of cooking aerosols with high liquid-organic content; this limitation is discussed in Section 4.4.

2.3 Air quality assessment protocol

Sampling was conducted at nine measurement dates: December 3, 6, 9, 12, 15, 18, 21, 24, and 27, 2024, yielding nine discrete sampling events with approximately three-day intervals. On each date, two measurement sessions were conducted per restaurant: a morning peak session (corresponding to the busiest morning service period, approximately 09:00–11:00 h) and an evening peak session (corresponding to the busiest dinner service period, approximately 19:00–21:00 h).

Instruments were positioned at a sampling height of 1.25–1.50 m above floor level, in accordance with manufacturer guidelines, representing the breathing zone of standing adults. All instruments were powered on and allowed a stabilization period of no less than 5 minutes prior to recording data, to ensure that all sensors had reached operational equilibrium. Readings were then logged continuously for 30 minutes at each session, and both the maximum (peak) and minimum concentrations observed during that window were recorded. The reported value for each session is the arithmetic mean of the five stations’ readings, and individual station values represent the range across the sensor’s measurement window. Results are expressed as the mean ± standard error (SE) of the five station measurements recorded on each sampling date. Each restaurant on each sampling date and time period (morning or evening) was treated as one site-session observation; thus, the dataset consisted of repeated measurements across 5 restaurants, 9 dates, and 2 daily sampling periods.

2.4 Statistical analysis

Data were organized by sampling date, restaurant, and time period (morning and evening). Because morning and evening measurements were obtained from the same restaurants on the same sampling dates, these observations were treated as paired repeated measurements. Accordingly, morning–evening differences for each environmental parameter were evaluated using paired comparisons (paired-samples t-test when the normality assumption was considered acceptable, or the Wilcoxon signed-rank test when appropriate). Descriptive statistics, including mean, standard deviation, minimum, and maximum values, were calculated for all measured variables. Temporal and site-related variation was primarily assessed descriptively because the same restaurants were repeatedly measured across multiple dates, and the limited number of restaurants (n = 5) restricts the robustness of formal site-wise inferential comparisons. Therefore, site-level differences were interpreted cautiously, with emphasis placed on descriptive spatial and temporal patterns rather than strong inferential claims. Statistical significance was considered at p < 0.05. All analyses were performed using SPSS version 26 (IBM Corp., Armonk, NY, USA).

4.1 Pollution sources and combustion characteristics

The presence of CO, CO₂, PM_2.5, and PM₁₀ at the same time indicates that indoor pollution is a result of combustion processes, including those that take place when cooking with charcoal. Because of this, it was expected that a strong correlation between the different types of pollution would be found; indeed, in this study, the collected source fingerprints at the various locations within this study confirm this. The solid fuel (charcoal) is not as completely burned as other fuels (like propane or natural gas), resulting in a much greater amount of emissions of both CO and PM (and also PAHs), when measured on a per-unit-basis. Charcoal grilling produces a complex aerosol plume consisting of a mixture of particle sizes from very small to much larger than would typically be found with other combustion sources (i.e., gas grilling). The largest particles found in the collection of smoke from charcoal grilling consist primarily of EC and OC, whereas the particles produced by cooking with gas consist almost exclusively of CO₂ and water vapor [1]. Charcoal-cooked fatty meats produce pyrolytic lipids from the cooking processes of frying and broiling, such as acrolein, formaldehyde, and different types of aldehydes, in addition to creating PAHs from partially burned fat drippings on the surface of the charcoal. There are differences between charcoal-cooked fatty meats and gas grills (where gas grill PM emissions are lower, but CO₂ emissions are average). Gas grills also have gases with lower PM compared to wood-burning kitchen stoves; additionally, they have higher CO₂ emission rates than conventional stovetop methods of food preparation, thereby representing combustion sources (fired through wood) of significantly higher CO₂ emissions, resulting in dissimilar PM chemical patterns due to different pyrolitic lignin-derived wood components.

Across all nine sampling days, peak pollutants observed in the evening samples were consistently higher than those observed in the morning samples. The main reasons why evening pollutants are greater than morning pollutants are: (1) increased customer traffic and, therefore, more intensive and longer grilling activity compared to dinner hours; (2) build-up of combustion pollutants in the restaurant's indoor air during the operational day, as it pertains to insufficient ventilation systems to achieve stable PM concentrations; and (3) the average amount of charcoal loaded and re-ignited is generally higher in busier evening periods. The particularly elevated concentrations recorded on 27 December evening (PM_2.5: 211 μg/m³; CO: 83 ppm) likely reflect the compounding effect of peak pre-holiday demand and sustained charcoal burning with minimal ventilation during a relatively cold December night, when windows and doors would be kept closed. These findings are consistent with Taner et al. [15], who documented similarly elevated PM_2.5 levels in 14 charcoal-grill restaurants in Turkey, and Kim et al. [16], who identified charcoal barbecue grilling as a primary source of PM₁₀ in urban Korean restaurants.

4.2 Health risk implications

The pollutant concentrations documented in this study carry substantial health risk implications, particularly for kitchen staff who are occupationally exposed for six to twelve hours daily. Chronic exposure to PM_2.5 at the levels documented here—routinely 4–14 times the WHO 24-hour guideline (Figure 1)—is associated with a well-characterized exposure-response relationship for cardiopulmonary outcomes. A dose-response analysis using the WHO’s published concentration-response functions for PM_2.5 suggests that sustained exposure at mean evening values of 100–211 μg/m³ corresponds to substantially elevated risk of developing chronic obstructive pulmonary disease (COPD), lung cancer, ischemic heart disease, and stroke compared to populations exposed at WHO guideline levels. Epidemiological evidence from biomass cooking cohorts indicates a relative risk of COPD approximately 2.4–3.5-fold higher in individuals with chronic solid-fuel cooking smoke exposure compared to those using clean fuels [17]. PM_2.5 particles—by virtue of their sub-2.5 μm aerodynamic diameter—penetrate the conducting airways and deposit in the alveolar region, where they trigger oxidative stress, pro-inflammatory cytokine release, and endothelial dysfunction [2, 18].

Figure 1. Average pollutant concentrations (mean ± SE) at five monitoring stations during morning and evening peak periods, December 2024

Error bars represent ± standard error (SE) across nine sampling days (3–27 December 2024). Dashed reference lines indicate WHO and US EPA guideline values. Statistical significance between morning and evening sessions was evaluated using paired comparisons based on matched site-date observations (paired-samples t-test when assumptions were met, or Wilcoxon signed-rank test when appropriate): * p < 0.05; ** p < 0.01.

CO concentrations exceeding the WHO 1-hour guideline of 30 ppm were documented on the majority of sampling days at most stations (Figure 1). At the peak evening concentrations observed (83 ppm on 27 December), exposure over a 2-hour dinner service would result in carboxyhemoglobin (COHb) levels sufficient to cause headache, dizziness, fatigue, and impaired cognitive and motor function in otherwise healthy individuals [19]. For kitchen staff exposed across a full evening service, the cumulative CO dose may approach or exceed occupational threshold limit values. Differential exposure between staff—who remain in the kitchen throughout service—and customers—whose exposure duration is typically 30–90 minutes per visit—means that the occupational health risk to restaurant workers is substantially greater than that to patrons, even at identical ambient concentrations. Future studies should characterize this differential exposure using personal monitoring approaches.

The measured CO₂ concentrations in these restaurants should be interpreted primarily as an indicator of ventilation adequacy rather than as evidence of acute CO₂ toxicity at the levels observed in this study. Mean values approaching or modestly exceeding 1,000 ppm, particularly during evening peak periods, indicate insufficient air exchange relative to occupant density and combustion activity. At these concentrations, CO₂ is best viewed as a practical proxy for ventilation performance and pollutant accumulation, rather than an immediate health hazard in itself. Nevertheless, several studies suggest that indoor CO₂ concentrations in the range of approximately 1,000–2,500 ppm may be associated with reduced perceived air quality and measurable decrements in some aspects of cognitive performance in enclosed work environments [20-22]. In the present study, the elevated CO₂ levels therefore support the interpretation that ventilation was inadequate to effectively dilute co-emitted combustion pollutants such as PM and CO, thereby indirectly increasing occupational exposure risk for restaurant staff.

4.3 Thermal environment and compounding factors

All five stations had an acceptable indoor thermal rating according to ASHRAE Standard 55 (18-24 ℃), with temperatures ranging from 18.3 ℃ to 22.5 ℃. However, all five stations had a mean RH at or above 60%, creating a latent risk; according to both the US EPA [12] and the WHO [14] guidelines, this is indicative of a potential threshold for microbial proliferation (mould, dust mite activity, etc.). While the current values are not extremely high, they persistently exceed the 60% threshold and, combined with an already contaminated air environment, create conditions favourable for condensation of moisture on surfaces and the potential for biological growth on building materials and ventilation components. This will further degrade indoor air quality beyond what would be expected due to the combustion products that could be directly measured. Therefore, active humidity control measures (e.g., mechanical dehumidification, improved air exchanges) are recommended to lower the relative humidity to a range of 50-55%.

4.4 Limitations

There are a number of methodological limitations to this study that need to be recognized. First, the portable optical PM sensors used in the current study provided both high temporal resolution and practicality for operation; however, there is the potential for PM mass concentrations to be overestimated, particularly in environments with a high liquid-organic aerosol content (such as cooking oil mist). Conversely, particle density and optical properties are assumed to be standardized in the light-scattering algorithm. The absence of cross-validation with gravimetric (filter-based) reference methods was another limitation of this study, but it remains a priority for future studies. Second, the sampling period consisted of only December 2024 (a single month in the colder climate of the study area), and therefore, the possibility for seasonal variability of ventilation behaviours (such as more open-window ventilation in the summer months) could greatly affect the IAQ within the various seasons. Third, a limitation of the present study is that repeated observations were collected from a relatively small number of restaurants (n = 5) across multiple dates. Accordingly, while temporal and spatial patterns were evident, site-level inferential comparisons should be interpreted cautiously and are best considered indicative rather than definitive. Finally, personal monitoring of individual workers' exposure was not conducted. The area concentration estimates reported in this study are not equivalent to personal doses but provide an estimate of the area concentration at the worker location within the sampled area. Additionally, these limitations must also be considered before future studies can be truly robust.