Evaluating the Impact of Increased Heavy Oil Consumption on Urban Pollution Levels through Isotope (δ¹³C, δ³⁴S, ¹⁴C) Composition

Abstract

The impact of heavy fuel oil (HFO) on the chemical and isotopic composition of submicron particulate matter (PM₁) was investigated. For this purpose, we conducted an analysis of water-soluble inorganic ions (WSIIs) and multiple isotopes (δ³⁴S, δ¹³C, ¹⁴C) of PM₁ and SO₂ collected during two heating periods: before (2021–2022) and during the use of HFO (2022–2023) in Vilnius, Lithuania. The results showed that the combustion of HFO increased the concentrations of SO₂ (by 94%) and PM₁-related sulfate (by 30%). It also altered the chemical composition of PM₁, with sulfate becoming the predominant component (~40%) of WSIIs. The stable sulfur isotope ratios of SO₂ (δ³⁴S_SO2) and sulfate (δ³⁴S_PM1) shifted significantly to more negative values (δ³⁴S_SO2 = 0.4‰, δ³⁴S_PM1 = −0.3‰) compared to the previous heating period. Anticorrelation between δ¹³C and δ³⁴S values indicated increased contributions of ¹³C-enriched fossil fuel sources (coal and HFO) in EC, although the share of fossil fuels in elemental carbon (EC) slightly decreased during the HFO period. The combustion of HFO affected the concentrations of PM₁ chemical components and substantially impacted the isotopic composition and source contributions of sulfate and EC.

1. Introduction

In recent decades, concerns about air quality have attained global significance due to the rapid industrialization observed in many countries [1,2]. Submicron particulate matter (PM₁) is one of the most significant pollutants, with substantial variability in concentration and chemical composition [3] and whose diverse constituents can significantly affect environmental and climatic processes [4,5]. PM₁ mainly comprises carbonaceous materials, including organic and elemental carbon (OC and EC), and water-soluble inorganic ions (WSIIs) such as sulfate (SO₄²⁻), nitrate (NO₃⁻), and ammonium (NH₄⁺) [6,7,8]. Organic carbon encompasses a wide range of organic compounds [9], including those directly emitted from natural and anthropogenic sources (primary OC [10,11,12]) and those formed in the atmosphere through chemical reactions (secondary OC [5,13,14,15]), whereas EC is released directly from incomplete combustion processes such as fossil fuel combustion and biomass burning. EC can persist in the atmosphere for several days due to its chemical inertness, significantly contributing to atmospheric warming [16,17]. Secondary inorganic ions (SO₄²⁻ and NO₃⁻) predominantly form in the atmosphere through chemical reactions from their gaseous precursors, sulfur dioxide (SO₂) and nitrogen oxides (NO_x). Water-soluble inorganic ions (WSIIs) influence the hygroscopic properties [18] and acidity of PM [19]. They also significantly contribute to visibility degradation [20] and enhance the formation of haze [21,22,23].

Numerous control measures have been implemented in Europe’s energy, manufacturing, industry, and transport sectors to achieve substantial improvements in air quality. The EURO vehicle standards have led to a 50% reduction in PM_2.5 emissions from global road transport exhaust [24]. Fuel quality directives, which regulate sulfur content and promote the use of less carbon-intensive fuels such as natural gas instead of coal or fuel oil, have resulted in decreased emissions of SO₂ and PM [24,25].

The Russian invasion of Ukraine in February 2022 triggered a significant crisis in the European gas market. The substantial reduction in Russian gas supply led to unprecedented increases in European gas prices, reaching record highs in 2022 [26]. The Lithuanian government has authorized the use of low-sulfur (0.9%) heavy fuel oil (HFO) for central heating as a substitute for natural gas, maintaining stable heating prices for consumers. Heavy fuel oil releases significant quantities of particulate matter [27,28], sulfur dioxide [29,30], nitrogen oxides [31], and heavy metals [32,33], making it a crucial factor influencing urban air quality [34]. During the 2022–2023 heating period, heavy fuel oil comprised approximately 36% of the fuel utilized for heat production in the Vilnius power station [35]. With significant changes in emission sources during the recent heating season, there was a clear need for a comprehensive study on the impact of HFO combustion on urban air quality.

To assess the impact of HFO usage on the air quality of Vilnius, we compared the period of HFO usage to the conventional heating (CH) period of 2021–2022, when HFO was not used in thermal power stations. Studies of this kind are rare because of the unique opportunity to compare two distinct periods: one characterized by increased heavy fuel oil usage and the other without it. Stable isotope analysis provides a valuable tool for identifying the origin of aerosols and their formation processes [36,37,38,39]. Furthermore, the combination of stable carbon and radiocarbon isotope (δ¹³C and ¹⁴C) methods allows for even more detailed characterization of the particles formed during the combustion of coal, liquid fossil fuels, and biomass [40,41]. Although radiocarbon analysis provides an unambiguous distinction between fossil and non-fossil sources [12,40,42], partitioning sources of distinct fossil fuels oftentimes poses inherent difficulties due to overlapping stable isotope value ranges [43,44,45]. On the other hand, sulfur isotope analysis is useful for tracking secondary sulfate-forming reactions [46,47,48] and has been applied to investigate their sources [49,50,51,52,53]. Numerous previous studies either applied carbon isotopes to describe fossil fuel combustion contributions [40,54,55,56] or applied sulfur isotopes to characterize distinct anthropogenic sources and sometimes the fraction of oil combustion itself [48,52,53]. Therefore, the application of combined analysis (δ³⁴S, δ¹³C, and ¹⁴C measurements) can provide a detailed description of pollution sources and atmospheric reactions and can help elucidate the impact heavy fuel oil usage had on local air quality in Vilnius. This study provides wintertime concentration levels of water-soluble inorganic ions (WSIIs) in PM₁ in Vilnius, Lithuania, both before (2021–2022) and during (2022–2023) the introduction of heavy fuel oil. The analysis of PM₁’s chemical and isotopic (δ¹³C, δ³⁴S, ¹⁴C) composition was performed to characterize the contribution of fossil fuel sources to carbonaceous (EC) and sulfur-containing PM₁ during selected heating periods.

2. Materials and Methods

2.1. Sampling Site

The sampling site was located in the southeastern part of Lithuania, specifically in the capital city of Vilnius (54.72 N, 25.32 E). The sampling station for PM₁ sulfate and SO₂ gas samples was on the rooftop of a four-story building (158 m a.s.l., 25 m a.g.l.), within the Center for Physical Sciences and Technology (Figure 1, marked by a star). The site is positioned in a forested environment surrounded by a combination of residential and private properties. The site is situated in a relatively low-traffic zone, located more than 1 km east of a bustling highway. Furthermore, pollution advecting from the more urbanized parts of the city (southwest direction) influences air quality at the sampling site.

The thermal power station (TPS) in Vilnius is the largest of its kind in the city, located approximately 8.9 km southwest of the sampling site (Figure 1, marked by a chimney). It boasts a substantial heating capacity of 913 MW and generates 24 MW of electric power. This facility plays a crucial role in providing heating and electricity to Vilnius, particularly during the colder months.

2.2. Sample Collection

Weekly samples of PM₁ and SO₂ were collected simultaneously using a tandem filter setup. Initially, a quartz fiber filter (Whatman QM-A) 0.15 m in diameter was used to capture deposited PM₁ samples. Positioned beneath it, a K₂CO₃-impregnated glass fiber filter (Sartorius MG 227/1/60, Bohemia, NY, USA) was employed to collect SO₂ gasses. A high-volume PM₁ sampler (DIGITEL DH-77, Volketswil, Switzerland) operating at a flow rate of 0.5 m³/min facilitated sample collection. Filters underwent pre-treatment at 500 °C for 8 h prior to sampling. Twenty-three samples of submicron particulate matter (PM₁) and twenty-one SO₂ gas samples were collected during two heating periods (2021–2022 and 2022–2023) spanning from 28 October to 31 March. After sampling, the filters were individually separated, wrapped in pre-combusted aluminum foil, sealed in plastic bags, and stored at −20 °C in a freezer until further analysis. Ultimately, the samples collected on quartz fiber filters were used for stable isotope measurements of elemental carbon and sulfate, as well as for radiocarbon and water-soluble inorganic ion concentration measurements. K₂CO₃-impregnated glass fiber filters were used only for the stable isotope measurements of SO₂ gasses.

2.3. Stable Isotope Measurements

For the analysis of their isotopic composition, PM₁ and SO₂ samples required pre-treatment for both sulfur and elemental carbon measurements. The elemental carbon fraction was separated from organic carbon by combusting a single punch (2.84 cm²) of quartz fiber filter material with collected PM₁ at 375 °C [57,58]. Following this, the filter material was pretreated with 37% HCl vapors in a glass desiccator for 12 h to remove the carbonate fraction [59,60]. Then, HCl was replaced with NaOH pellets to neutralize any remaining acid from the samples. After these procedures, the filter punch was packed in a tin capsule for stable carbon isotope measurements.

For the isotopic composition measurements of sulfur, half of the quartz fiber filter (76.97 cm²) was used for sulfate extraction, while the entire impregnated glass fiber filter was utilized for SO₂ pretreatment. The following methods of sulfate extraction were nearly identical for SO₂ and PM₁. The filter material was shredded and immersed in 100 mL ultrapure water. For SO₂ samples, 30% hydrogen peroxide was added to oxidize SO₂ to SO₄²⁻. The samples were ultrasonicated and left overnight. The next day, they were filtered using 0.22 μm syringe filters, and the filtered solution was acidified to a pH between 2 and 3 with HCl. Then, 5 mL of 1 mol/L of BaCl₂ was added to precipitate the sulfate as BaSO₄. The following day, the precipitate was collected on 0.2 μm cellulose acetate filters and any remaining Cl⁻ was washed away with ultrapure water. The samples were dried, and the filter material was combusted at 500 °C, leaving only the BaSO₄ behind. For the isotope analysis, 0.50 mg of BaSO₄ was packed with vanadium pentoxide (V₂O₅) powder within tin capsules.

Stable isotope ratio values were measured using an elemental analyzer (EA, Thermo Flash EA 1112, Marlton, NJ, USA) connected to an isotope ratio mass spectrometer (IRMS, Thermo Delta V Advantage, Sheffield, UK). Stable isotope values of carbon and sulfur were measured in per mille (‰) and are expressed as follows:

$$\mathsf{\delta }\mathrm{X}=\left[{\displaystyle \frac{{\mathrm{R}}_{\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}}-{\mathrm{R}}_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}}}{{\mathrm{R}}_{\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}}}}\right]·1000‰$$

(1)

where δX is the δ value of a particular element X (C or S), which denotes the relative difference between the isotope ratios of the measured sample and the reference material. R_sample is the isotope ratio of ¹³C/¹²C (or ³⁴S/³²S) in the sample, and R_reference is the isotope ratio of the reference material Pee Dee Belemnite (PDB) for carbon and Vienna Canon Diablo Triolite (VCDT) for sulfur. For scale normalization, caffeine IAEA-600 and graphite USGS24 reference materials were used for carbon isotope analysis, and IAEA-S-1 and NBS-127 reference materials were used for sulfur isotope analysis. Generally, measurement precision was better than 0.2‰ for carbon and 0.3‰ for sulfur. When possible, multiple measurements of each sample were made, and blank correction was applied to eliminate the influence of filter materials.

2.4. Radiocarbon Measurements

For radiocarbon measurements, aerosol samples were first graphitized using an Automated Graphitization System coupled with an Elemental Analyzer (EA-AGE-3, Ionplus AG, Dietikon, Switzerland). Due to very low elemental carbon amounts in samples, the graphitization of low-carbon aerosol filters (PM₁) involved employing the sample dilution method and using the mass balance equation to estimate ¹⁴C values, as described in detail in a study by Butkus et al. [61]. Following this method, aerosol samples with as little as 40 µg of carbon could be graphitized. With the EA-AGE-3 system, multiple pieces or punches of a single sample could be combusted separately, and the resulting CO₂ gas was captured by a zeolite trap and subsequently transferred into a single reactor. Thus, 10 punches (28.35 cm²) of the filter sample containing low carbon content were combusted, while the remaining carbon needed for graphitization was supplemented with ¹⁴C-free reference material (phthalic anhydride, PhA, 0 pMC). The ¹⁴C values were then calculated using a mass balance equation:

$${\mathrm{p}\mathrm{M}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}}={\mathrm{p}\mathrm{M}\mathrm{C}}_{\left(\mathrm{s}\right)}{\mathrm{n}}_{\left(\mathrm{s}\right)}+{\mathrm{p}\mathrm{M}\mathrm{C}}_{\mathrm{P}\mathrm{h}\mathrm{A}}{\mathrm{n}}_{\mathrm{P}\mathrm{h}\mathrm{A}}$$

(2)

$${\mathrm{p}\mathrm{M}\mathrm{C}}_{\left(\mathrm{s}\right)}={\displaystyle \frac{{\mathrm{p}\mathrm{M}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}}-{\mathrm{p}\mathrm{M}\mathrm{C}}_{\mathrm{P}\mathrm{h}\mathrm{A}}{\mathrm{n}}_{\mathrm{P}\mathrm{h}\mathrm{A}}}{{\mathrm{n}}_{\left(\mathrm{s}\right)}}}$$

(3)

where pMC_(mixture)—measured pMC of the mixture, pMC_(s)—pMC of the sample, n_(s) is the partial contribution of the sample; pMC_(PhA)—pMC of phthalic anhydride, n_(PhA) is the partial contribution of the phthalic anhydride.

Radiocarbon analysis was performed using a Single-Stage Accelerator Mass Spectrometer (SSAMS, National Electrostatics Corp., Middleton, WI, USA). The background of measurements using phthalic anhydride (Alfa Aesar) was estimated at 0.25 pMC (percent of modern carbon), the NIST OXII (134.06 pMC) standard was used as reference material, and the ¹⁴C/¹²C ratio was measured with an accuracy better than 0.3%. Isotopic fractionation was corrected with the ratio of ¹³C to ¹²C.

In this study, the fraction of modern carbon is used and expressed as [62]

$$f_{\mathrm{M} }={\displaystyle \frac{{({}^{14}\mathrm{C}/{}^{12}\mathrm{C})}_{\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}}}{0.749{({}^{14}\mathrm{C}/{}^{12}\mathrm{C})}_{\mathrm{O}\mathrm{x}\mathrm{I}\mathrm{I}}}}$$

(4)

where (¹⁴C/¹²C)_sample is the ¹⁴C/¹²C ratio of the sample and (¹⁴C/¹²C)_OxII is the ratio of the isotope reference material (OxII).

We can calculate the fraction of non-fossil carbon (f_nf), considering that f_M for fossil fuel sources is equal to 0, and using a f_M value of 1.02 [63,64,65] for contemporary biomass burning sources. Fraction f_nf estimates that the biomass used for domestic heating primarily originates from contemporary sources [40,66]. The non-fossil carbon fraction can be determined as follows:

$$f_{\mathrm{n}\mathrm{f}}={\displaystyle \frac{f_{\mathrm{M}}}{1.02}}$$

(5)

Additionally, we can calculate the fraction of fossil carbon as follows:

$$f_{\mathrm{f}}=1-f_{\mathrm{n}\mathrm{f}}$$

(6)

2.5. Water Soluble Inorganic Ions and SO₂ Concentration Measurements

First, a punch (0.20 cm²) of quartz filter material was placed into a tube, and 10 mL of high-purity water was added to dissolve water-soluble inorganic ions (WSIIs). The sample tubes were then ultrasonicated for 30 min, and the solution was then filtered through a 0.22 μm syringe filter. The concentrations of WSIIs were then measured using an ion chromatograph (850 Professional IC, Metrohm, Herisau, Switzerland). Inorganic anions were separated using a Metrosep A Supp 7–250/4.0 column. The eluent used was 3.6 mM sodium carbonate, and 20 mM H₂SO₄ was used as a regenerant. Ambient SO₂ concentration data were obtained from a local monitoring station located 8.5 km away from the sampling location.

2.6. Meteorological Data

Meteorological observation data, including temperature, humidity, wind speed, and mixed layer depth, were obtained from publicly accessible data provided by the National Oceanic and Atmospheric Administration (NOAA) [67].

NOAA HYSPLIT model was used to model backward air mass trajectories for the corresponding sampling periods [68]. To assess the influence of distant air masses, they were categorized as ‘clean’ or ‘polluted’ [69]. Air masses originating from the western regions of Europe, such as Germany and Poland, as well as from the northern and eastern regions, including Latvia, Estonia, Belarus, Ukraine, and Russia, were considered moderately polluted. Southwestern and southern regions were labeled as more polluted directions (Southern Poland, Germany and Czechia). The northwestern direction (Baltic Sea, Scandinavia) was classified as clean. Examples of prevailing air masses are given in Figure S1.

3. Results and Discussion

3.1. Concentration of Water-Soluble Inorganic Ions and SO₂

The winter period of 2021–2022 was classified as the conventional heating (CH) period during which biomass and natural gas contributed 61% and 39%, respectively, to the total fuel utilized for thermal energy generation at the Vilnius thermal power station [70]. During the 2022–2023 heating season, low-sulfur (0.9%) heavy fuel oil largely replaced natural gas at the Vilnius thermal power station. This period is referred to as the heavy fuel oil (HFO) period in the rest of this study. During this period, the primary fuel for heating generation was biomass (56%), followed by heavy fuel oil (36%), with smaller contributions from natural gas (7%) and diesel (1%) [35]. Both periods exhibited similar seasonal meteorological conditions. During the CH period, the averages of temperature, relative humidity, mixed layer depth, and wind speed were 0.6 ± 2.8 °C, 82 ± 14%, 503 ± 124 m, and 16.1 ± 4.4 km/h, respectively. The prevailing air mass directions were from the southwest, west, and northwest. During the HFO period, the averages of temperature, relative humidity, mixed layer depth, and wind speed were −0.3 ± 3.4 °C, 90 ± 5%, 430 ± 118 m, and 15.4 ± 3.5 km/h respectively. The prevailing air mass directions were also similar to the previous period (southeast, southwest, and west).

A time series of the concentrations of eight water-soluble inorganic ions (Na⁺, NH₄⁺, K⁺, Mg²⁺, Ca²⁺, Cl⁻, NO₃⁻, and SO₄²⁻) during both periods is presented in Figure 2. Three out of eight ions (SO₄²⁻, K⁺ and NO₃⁻) were the most abundant, accounting for about 70% of WSIIs during both periods (Figure 2). The CH period was characterized by concentration levels of WSIIs following the order: K⁺ > SO₄²⁻ > NO₃⁻ > Na⁺ > NH₄⁺ > Ca²⁺ > Cl⁻ > Mg²⁺. The concentrations of total WSIIs (sum of Na⁺, NH₄⁺, K⁺, Mg²⁺, Ca²⁺, Cl⁻, NO₃⁻, and SO₄²⁻) varied from 3.17 to 6.42 μg/m³ during the CH period, with an average of 5.12 μg/m³. The average concentrations of potassium, sulfate, and nitrate were 1.61 ± 0.48, 1.16 ± 0.31, and 1.05 ± 0.44 μg/m³, respectively. K⁺ was the predominant ion during this period, accounting for 27% of total WSII, largely due to significant emissions from biomass burning (61%) [70].

During the HFO period, concentrations of WSIIs ranged between 2.79 and 7.21 μg/m³, with an average value of 4.03 μg/m³. The average concentrations of WSIIs in PM₁ followed a different trend, namely SO₄²⁻ > K⁺>NO₃⁻ > Na⁺ > NH₄⁺ > Ca²⁺ > Cl⁻ > Mg²⁺, during this period. The concentration levels of NO₃^- and K⁺ were lower, with average values of 0.47 ± 0.37 and 1.04 ± 0.58 μg/m³, respectively. In contrast, sulfate concentrations increased modestly by a factor of 1.3, with an average value of 1.53 ± 0.38 μg/m³. The chemical composition of PM₁-related WSIIs was similar during both (CH and HFO) periods, with the exception of sulfate and nitrate. The share of sulfate increased from 22% in the 2021–2022 heating season to 37% in the 2022–2023 heating season, while the relative contribution of nitrate decreased by 10%, reaching only 11% during the HFO period. Additionally, the use of HFO had a significant impact on SO₂ concentrations and, as shown in Figure 2, concentrations doubled from 1.30 ± 0.39 μg/m³ during the CH period to 2.54 ± 2.13 μg/m³ during the HFO period.

During the CH period, the concentrations of sulfate (1.16 μg/m³) were on similar level to SO₂ concentrations (1.3 μg/m³) due to the absence of intense local sulfur pollution sources. Sulfate showed a moderate correlation with SO₂ (r = −0.55, p < 0.05), indicating an increased contribution of distant sources to sulfate concentrations in Vilnius during the 2021–2022 heating season. Sulfate correlated significantly only with ammonium (r = 0.77, p < 0.05), suggesting that the main form of sulfate in PM₁ could be (NH₄)₂SO₄. The higher NO₃⁻ concentrations can be attributed not only to intense emission sources, but also to various chemical factors, including the concentrations of gaseous precursors (SO₂, NO_x and NH₃) and the thermodynamic equilibrium between gas and particulate phases [71,72]. The neutralization of sulfuric acid by ammonia and the subsequent formation of particulate ammonium sulfate usually outweigh the formation of ammonium nitrate. Ammonium nitrate is formed only in the presence of an excess of ammonium [73,74]. The high [NH₄⁺]/[SO₄²⁻] molar ratio (1.6 ± 0.5) together with the significantly increased NO₃⁻/SO₄²⁻ ratios (0.90 ± 0.78) suggest that lower SO₂ concentrations during the CH period could contribute to higher nitrate formation (Figure 3). In contrast, during the HFO period, high SO₂ concentrations with a low [NH₄⁺]/[SO₄²⁻] molar ratio (0.98 ± 0.52) and small NO₃⁻/SO₄²⁻ ratio (0.34 ± 0.31) were observed (Figure 3), indicating a limited formation of NH₄NO₃.

A strong correlation between sulfate and K⁺ was found (Figure S2b, r = 0.72, p < 0.05), suggesting that K₂SO₄ was the predominant form of PM₁-related sulfate during the HFO period. The relationship between K⁺ and sulfate in secondary particulate matter affected by biomass burning has been established in several studies [75,76,77]. The formation of K₂SO₄ could result from rapid photochemical oxidation or aqueous phase reactions between KCl (emitted during biomass burning) and SO₂ [78]. In this study, the S/K ratio of 0.6 was closely aligned with the S/K values observed near biomass burning sources [75,79], indicating local sulfate formation. These results highlight the significant influence of heavy fuel oil usage on both SO₂ concentration levels and the subsequent pathways of PM₁-related sulfate formation.

3.2. Carbon and Sulfur Isotope Analysis

Stable isotope values of SO₂ (δ³⁴S_SO2) and PM₁ sulfate (δ³⁴S_PM1), along with stable isotope values of EC (δ¹³C_EC) and the fraction of fossil fuel-derived EC (f_f) are given for both periods of 2021–2022 (CH) (Figure 4a) and 2022–2023 (HFO) (Figure 4b). During the CH period, δ³⁴S_SO2 values were relatively stable and varied from 4.8‰ to 6.1‰ with an average of 5.4 ± 0.6‰. Measured stable isotope values of SO₂ reflect the influence of local sulfur emission sources due to the absence of large SO₂ emission sources around Vilnius and due to its short atmospheric lifetime (~12 h) [80]. Measured values are similar to δ³⁴S values of biomass burning (7.3–9.1‰ [81]) and traffic emissions (4.0–8.0‰ [82]) indicative of their increased contributions. Thus, during the CH period, δ³⁴S_SO2 values suggest that local emission sources remained constant. On the other hand, the δ³⁴S_PM1 values of sulfate were more negative and varied considerably, ranging from -2.0‰ to 4.4‰ with an average of 2.0 ± 2.1‰. Thus, δ³⁴S_PM1 values suggested a fluctuating influence of sulfur pollution sources. The lowest recorded values on 17 January 2021 (−0.4‰), 7 February 2021 (1.9‰), and 18 February 2021 (−2‰) are close to or are even lower than those typically associated with coal combustion sources (−1.0 to 4.4‰) [45]. Previous studies have reported [83] that air masses coming from the southwestern direction significantly affect sulfur pollution levels in Vilnius. During the days with the most negative δ³⁴S_PM1 values, air masses originated from neighboring countries to the southwestern direction (Figure S1), where coal is extensively used for domestic heating and energy production [84,85]. Overall, during CH period, δ³⁴S_SO2 values represented local sulfur pollution sources, whilst δ³⁴S_PM1 values indicated the influence of distant pollution sources.

Elemental carbon values of δ¹³C_EC ranged from −27.4‰ to −29.3‰, with an average of −28.4 ± 0.6‰ (Figure 4a). These values fall in the range of biomass burning values of the EC fraction (EC: −29.9‰ to −25.4‰ [44,86,87]) and are close to traffic emission values (TC: −31.6 to −29.89‰ [88]) but are outside the typical coal combustion values (δ¹³C_EC: −24.7–−23.3‰ [86,89,90]). Additionally, during the CH period, fossil fuel fraction varied between 76 and 91% (Figure 4a), averaging 85 ± 6%, representing the combined contributions of coal combustion and traffic emissions. Thus, the non-fossil fuel (biomass burning) fraction contributed ~15%, accordingly, which is comparable to previous studies [54,55,91,92] and supports our hypothesis regarding various pollution sources (local and distant).

The HFO period (Figure 4b) displays significant differences in δ³⁴S values for both SO₂ and sulfate in comparison to the CH period. During the HFO period, the previously stable background of low SO₂ pollution (CH period) and its consistent δ³⁴S_SO2 values are disrupted by increased HFO emissions and a considerable variability in δ³⁴S_SO2 values is observed. δ³⁴S_SO2 values show a gradual shift towards more negative values, ranging from −4.9‰ to 4.6‰ and, on average, being equal to 0.4 ± 3.2‰. Similarly, δ³⁴S_PM1 values displayed an average of −0.3 ± 2.4‰, with a minimum-recorded value of −4.8‰. During the HFO period, both δ³⁴S_SO2 and δ³⁴S_PM1 exhibited unusually negative values, which fall outside the δ³⁴S range typical of sulfur pollution sources such as biomass burning, coal combustion, or traffic emissions observed in Lithuania [83]. These negative δ³⁴S values may therefore indicate the influence of an additional source, possibly that of heavy fuel oil utilized in the thermal power station. Unfortunately, the δ³⁴S value of HFO used in Vilnius TPS during this time is unknown; thus, it cannot be directly compared to measured emission values. To our knowledge, no studies in recent decades have measured the δ³⁴S values of HFO emissions in Eastern or Northern Europe. However, previous studies have shown that fuel oil can exhibit a wide range of δ³⁴S values, depending on the fuel origin [51]. The δ³⁴S values measured during the HFO period could possibly be of Middle Eastern origin (δ³⁴S values vary from −11.5‰ to −2.7‰) [93]. Moreover, during the HFO usage period, the δ³⁴S_PM1 values approach those of δ³⁴S_SO2, with an average difference of 0.7‰, while during the CH and pre-HFO period, the difference between δ³⁴S_PM1 and δ³⁴S_SO2 was 2.23‰ and 3.31‰, accordingly. Thus, during the HFO period, characterized by increased SO₂ concentrations, the influence of local pollution sources predominates over distant sources. Consequently, both δ³⁴S_SO2 and δ³⁴S_PM1 values predominantly indicate local sulfur emissions, in contrast to the CH period, when distant sources took precedence.

δ¹³C_EC values during the HFO period show a gradual shift towards more positive values, starting from an average of −28.8 ± 0.3‰ (pre-HFO period) and reaching an average of −27.5 ± 0.8‰ (HFO period) with a highest recorded value of −25.8‰ (Figure 4b). δ¹³C_EC values could possibly indicate a rising impact of HFO emissions (~−25.5% in particles [43]). Widory [43] found that fractionation around 3‰ occurs during the combustion of fuel oil. In the case of current measurements in Vilnius, the initial δ¹³C value of fuel oil is unknown. However, particles enriched with more ¹³C during the HFO period may indicate that the carbon delta values may have been influenced by fuel oil during this period. Furthermore, during this period, the observed trend towards more negative values of both δ³⁴S_SO2 and δ³⁴S_PM1, along with a coincident shift of δ¹³C_EC towards more positive values, suggests a common pollution source. This is also evidenced by a significant negative correlation between δ³⁴S_PM1 and δ¹³C_EC (Figure S3, r = −0.6, p < 0.05).

During the HFO period, the fossil fuel emission fraction of total EC (average 75 ± 7%) slightly decreased in comparison to the CH period (Figure 4b). A boxplot comparison between the periods is given in Supplementary Material (Figure S4). The decrease in f_f during the HFO period could have been caused by increased emissions from biomass burning, with recorded δ¹³C_EC values falling within the range typical of biomass burning values. Although, as previously stated, δ¹³C_EC values become more positive, indicating an influence of additional source, possibly of HFO combustion.

In summary, the isotopic composition results of SO₂ (δ³⁴S_SO2), sulfate (δ³⁴S_PM1), and elemental carbon (δ¹³C_EC) reveal clear trends caused by increased heavy fuel oil usage in the Vilnius thermal power station during the 2022–2023 heating season. However, the fraction of fossil fuel emissions slightly decreased during the HFO period, attributable to varying contributions of fossil and non-fossil fuel combustion to the aerosol budget.

4. Conclusions

To evaluate the impact of heavy fuel oil (HFO) usage on air quality in Vilnius, Lithuania, a comparative analysis was conducted on the wintertime concentration levels of water-soluble inorganic ions (WSIIs) in PM₁ during two periods: the conventional heating period (CH period: 2021–2022) and the period of HFO introduction (HFO period: 2022–2023). Significant impacts of HFO usage on the pollution levels and source contributions of sulfur-containing species in Vilnius were observed, evidenced by changes in stable isotope values of sulfur dioxide (SO₂) and particulate matter sulfate (PM₁). The combustion of HFO led to a 94% increase in SO₂ concentrations and a 30% increase in PM₁-related sulfate. This process also altered the chemical composition of PM₁, making sulfate the predominant component (~40%) of water-soluble inorganic ions. During the 2021–2022 period, characterized by the absence of HFO, δ³⁴S_SO2 values were relatively stable (5.4 ± 0.6‰). This stability is indicative of a mixture of sulfur emissions originating from local biomass burning and traffic sources. In contrast, δ³⁴S_PM1 values during the same period exhibited significant variation (2.0 ± 2.1‰), influenced by the origin of the air masses. This variation accentuates the substantial influence of distant pollution sources on the sulfate budget in Vilnius during the 2021–2022 period. Conversely, during the 2022–2023 period, when HFO was used, both δ³⁴S_SO2 and δ³⁴S_PM11 values shifted towards more negative values (δ³⁴S_SO2 = 0.4 ± 3.2‰, δ³⁴S_PM1 = −0.3 ± 2.4‰), indicating a predominant influence of local HFO emissions over distant sources. The overall fraction of fossil fuel contribution in EC marginally decreased during the HFO period (75 ± 7%) compared to the CH period (85 ± 6%). Concurrently, δ¹³C_EC values became more positive, indicating a shift in the EC isotope composition towards ¹³C-enriched fossil fuel emission sources (possible HFO combustion). Ultimately, this study underscores the influence of fuel type changes on local air quality and the isotopic signatures of particulate matter and sulfur dioxide.