The carbon components in indoor and outdoor PM2.5 in winter of Tianjin

To study the carbon components in indoor and outdoor PM2.5, the samples of PM2.5 were collected from Nankai University in December 2015. The contents of eight carbon components were analyzed to use the thermo-optical reflection method. The results indicated that organic carbon (OC) mass concentration was 17.01, 19.48 and 18.92 µg/m3 in outdoor, dormitory and laboratory; elemental carbon (EC) mass concentration was 7.97, 3.56 and 3.53 µg/m3 in outdoor, dormitory and laboratory; and the total carbon aerosol was the proportion of more than 23% of PM2.5 samples. Lower wind speed and higher relative humidity were helpful to the accumulation of PM2.5. The ratio of OC/EC was > 2, and the SOC/OC ratio was > 30%, indicating that SOC was a crucial component indoors and outdoors. About 72% and 85% of the outdoor OC entering dormitory and laboratory environment, and about 59% and 71% of the outdoor EC entering dormitory and laboratory environment. Factor analysis of the eight carbon fractions indicated that the sources of OC and EC in outdoor, dormitory and laboratory is different.

Indoor and outdoor concentrations of OC and EC in PM 2.5 were analyzed for five buildings located near roadsides (an office and a classroom with mechanical ventilation (MV) and three residences with natural ventilation (NV)). The average I/O ratios of OC and EC were 1.02 and 0.80, respectively. The major source of indoor EC, OC and PM 2.5 appears to be penetration of outdoor air 30 . The indoor-outdoor characteristics of PM 2.5 carbonaceous species in six residences were evaluated in Hong Kong during March and April 2004. The average I/O ratios of 24 h PM 2.5 , OC and EC were 1.4, 1.8, and 1.2, respectively. A simple model implied that about twothirds of carbonaceous particles in indoor air are originated from outdoor sources 31 . Indoor and outdoor emission measurements to quantify the carbonaceous matters were performed from the real-world biomass burning in rural households. Fractions of fugitive emissions of the total reached as high as 44-48%. Fugitive emissions would result in very high peak concentrations of approximately tens of mg/m 332 . Concurrent indoor-outdoor PM 2.5 measurements were conducted at urban, suburban, and rural sites in Harbin, China. OC/EC and potassium ion to elemental carbon (K + /EC) ratios verified that biomass was an important source. The highest SOC/ OC ratio was found at urban sites, up to 38.3% for indoors. SOC/OC ratios of indoor environments were higher, which is attributed to the conducive condition of forming the secondary pollutants during the heating period 33 . Studies have indicated that people spend 85%-90% of their lives indoors 34 . Few studies have focused on student dormitories and laboratories, which have distinct management and usage patterns. Unfavorable conditions in dormitory and laboratory environments can harm the health of the occupants. Therefore, the PM 2.5 samplings were conducted in outdoors and dormitory and laboratory in the Jinnan campus of Nankai University. The PM 2.5 concentration, OC and EC correlation, SOC generation, and carbon pollution sources for outdoor, dormitory, and laboratory samples were analyzed and discussed. The results can provide reasonable suggestions for the prevention and control of atmospheric pollution.

Materials and methods
Sample collection. The Jinnan campus of Nankai University was operational in September 2015. The outdoor sampling point was selected from Nankai University Library. The indoor sampling points were the laboratory and student dormitory. The sampling site was at an altitude of 30 m above ground level for indoor (laboratory and dormitory) and outdoor sampling (Library). The sampling point was 1.5 m above the floor for indoor and outdoor sampling. Sampling time is about 9 am-9 pm per day on December 3-21, 2015. The sampling sites in this study are shown in Fig. 1.
Before all the samples were measured, the parallel experiment of different instruments for different samplers indoor and outdoor were performed. The relative standard deviation of quality control of PM 2.5 , OC and EC concentration were less than 2.8%, 3.6% and 4.3%. For the collection of PM 2.5 from the outdoor environment, a sampler (Wuhan Tianhong Company) with a flow rate of 100 L/min was used. The particle sampler equipped with a quartz filter was set at an outdoor monitoring point. The diameter of the filter was 90 mm. The flow rate of the sampler was calibrated before sampling to maintain the flow rate error within the acceptable range. For indoor PM 2.5 sample collection, the LP-5 personal sampling pump (BUCK company, USA) and SKC personal exposed sampling cutting head (cut particle size: 2.5 μm) were used. A quartz filter with a diameter of 37 mm was used. The flow rate of the sampling pump was calibrated using the mini-buck soap film flow meter (BUCK company, USA).
Before sampling, the quartz filter was placed in a muffle furnace at 800 °C for 6 h to remove organic components 35 . Before weighing, the sample was placed in a constant temperature of 25 °C and relative humidity of 50% for 72 h. Subsequently, each sample was weighed twice to obtain an average value with an error of < 5 μg. After weighing 36 , the samples were preserved in a refrigerator (4 °C) for further analysis. OC and EC were analyzed by using the thermal/optical carbon analyzer, and the analysis method was thermooptical reflection (TOR) 12,37,38 . A 0.210-cm 2 sample was obtained from the filter and sent to the thermo-optic analyzer. In a pure helium atmosphere, the temperature was increased to 120, 250, 450, and 550 ℃, and the particulate carbon on the filter was converted into CO 2 to obtain four components of OC (OC1, OC2, OC3, and OC4). Then, the sample was gradually heated at 550, 700, and 800 °C in the helium atmosphere containing 2% (volume fraction) oxygen to obtain EC (EC1, EC2, and EC3). As the temperature increases in the inert helium, some of the OC forms optical pyrolized carbon (OPC) by the reflectance of 633 nm light from a He-Ne laser.
Element analysis. K, Zn, As, and Pb were used as tracer elements to analyze the carbon component sources of PM 2.5 . These elements are measured using inductively coupled plasma-mass spectroscopy (ICP-MS) (Agilent 7500a, Agilent Co. USA).
Indoor-outdoor relationship. I/O represents the ratio of the indoor and outdoor concentration of particulate matter or of a compound. Researchs showed that simple linear correlation equation of indoor and outdoor concentration of particulate matter or of a compound can identify the contribution made by the two sources. The linear equation is used by many researchers 40 .
where, C in refers to the particulate matter or compound concentration in indoor environment; F INF refers to the filtration factor, the fraction of outdoor particulate matter or compound that comes indoor environment; C out refers to the particulate matter or compound concentration in outdoor environment; C ig refers to the concentration of particle or compound generated from indoor sources.
Meteorological data. Local meteorological data included temperature, relative humidity (RH), precipitation and wind speed were obtained from the website https:// darks ky. net/ detai ls/ 39. 202,117. 263. The meteorological data were collected each day during the sampling period.

Results and discussion
Concentration of PM 2.5 , OC, and EC. As shown in Fig. 2 19.77 μg/m 3 for laboratory samples, respectively. The indoor PM 2.5 and OC concentrations were considerably higher than those of the outdoor environment during the three days, which indicated that the indoor PM 2.5 and OC are produced by indoor sources, such as personal activities and cleaning 21 . The concentration of PM 2.5 , OC, and EC increased on December 08, 14, and 21 mainly due to the heavy haze pollution, higher air relative humidity (93%, 82% and 84%), and lower wind speed (0.8 m/s, 1.6 m/s and 0.7 m/s). The average outdoor, dormitory, and laboratory OC concentrations were 17.01, 19.48, and 18.92 μg/m 3 , respectively. The average OC concentration in indoor was higher than that in outdoor, and the OC concentrations in indoor and outdoor appeared as an asynchronous change in hazy weather. This may have caused by the pollutants accumulating indoors due to the indoor confinement and the low ventilation rate (windows and doors closed during in hazy weather, however windows and doors opened about 1-2 h during nonhazy weather). The   44 and considerably higher than that observed in Xiamen (9.87 μg/m 3 ) 45 , Shanghai (7.77 μg/m 3 ) 46 , and Taiwan (10.40 μg/m 3 ) 8 . The EC concentration observed was lower than that observed in Wuhan (22.20 µg/m 3 ) 43 only, mainly because a large amount of coal was burned during winter, and the pollutants did not easily spread in temperature inversion. The OC and EC concentrations were considerably higher than those in Japan (3.75 µg/m 3 and 1.63 µg/m 3 ) 17 , Milan (14.00 µg/m 3 and 1.60 µg/m 3 ) 47 , and Greece (8.44 µg/m 3 and 5.29 µg/m 3 ) 48 , indicating relatively high carbon concentration in China. The carbon component concentration was high during winter in Tianjin, and the EC pollution was relatively serious. Therefore, it is necessary to control the primary carbon component emissions in Tianjin.
Meteorological conditions during the period of this study were shown in Fig. 3. Figure 3a-d indicated that the wind speed and relative humidity, wind speed and temperature are negative correlation. Lower wind speed and higher relative humidity were helpful to the accumulation of PM 2.5 . However, PM 2.5 accelerated the diffusion with lower temperature and wind speed increasing. From the above analysis, the meteorological conditions including wind speed, relative humidity and temperature have important influence on the formation and diffusion of PM 2.5 .
Indoor-outdoor relationship analysis. The regression analysis of OC concentration for dormitory and outdoor, laboratory and outdoor were shown in Fig. 4. The regression analysis of EC concentration for dormitory and outdoor, laboratory and outdoor were shown in Fig. 5.
According to Eq. (1), the filtration factors of the dormitory and outdoor, the laboratory and outdoor for OC were 0.72 and 0.85, which stated about 72% and 85% of the outdoor OC entering dormitory and laboratory environment, respectively. The filtration factors of the dormitory and outdoor, the laboratory and outdoor for EC were 0.59 and 0.71, which stated about 59% and 71% of the outdoor EC entering dormitory and laboratory environment, repectively.
OC/EC ratios and SOC. SOC (secondary organic carbon) is a crucial source of carbon in PM 2.5 . Chow (2001) 39 indicated that EC was relatively stable and can be used as a tracer of primary aerosol in the atmosphere 49 . The ratio of OC to EC was often used as the identification parameter for the existence of SOC. The existence of SOC was proven when OC/EC > 2. Table 2 showed that the average outdoor, dormitory, and laboratory OC/EC ratio was 2.25, 4.92, and 5.36, respectively, indicating that the proportion of SOC to OC was considerably large, and the indoor SOC pollution was more serious than the outdoor SOC pollution. The outdoor OC/EC ratio was lower than that in Xiamen (5.28) 45 , Beijing (4.47) 42 , Xining (4.71) 44 , Shanghai (5.84) 46 , Milan (8.75) 47 , Japan (2.30) 17 , and Taiwan (2.60) 8 and was higher than that in Wuhan (1.05) 43 and Greece (1.86) 48 , which indicated that SOC pollution was relatively serious in China.
To quantitatively describe SOC, Turpin (1995) proposed the following empirical equation: where, OC tot stands for the total organic carbon, EC stands for the elemental carbon, and (OC/EC) min stands for the minimum of OC/EC.
(2) SOC = OC tot − EC × OC EC min  53.47% and 37.77% of OC, respectively. These results indicated that SOC in this study was relatively active. The SOC concentration in the dormitory was higher than the outdoor and laboratory SOC concentrations, most likely because volatile organic compounds emitted from some sources 8,50 , the temperature in the dormitory was higher than that outdoors and in the laboratory during heating, which generated SOC 9,51 . Meanwhile, poor indoor air circulation contributes to the high concentration of SOC 30 . Turpin (1995) 52 found that the correlation between OC and EC concentration can be used to qualitatively analyze the source of atmospheric carbon aerosols. The correlation between OC and EC concentration can be expressed by a linear regression equation,    Figure 6 showed that the correlation between OC and EC (0.9070) in outdoor samples was the strongest, indicating that OC and EC may have similar sources. Related research has shown that coal combustion and vehicle exhaust emissions accounted for approximately 83% of the annual emissions of OC and EC in China 9,51,53-55 ; therefore, coal burning and vehicle exhaust emission may be the main sources of pollution. The correlation between OC and EC in dormitory and laboratory samples were 0.7293 and 0.6422, which were lower than that in outdoor samples, mainly due to photochemical reactions and indoor OC emission sources. Source analysis. Tracer elements. K, Zn, As, and Pb were used as tracer elements to analyze the carbon component sources of PM 2.5 . These elements are measured using inductively coupled plasma-mass spectroscopy (ICP-MS) (Agilent 7500a, Agilent Co. USA). K indicated biomass burning and waste combustion. Zn and Pb were tracer elements of vehicle emissions. As indicated the contribution of coal combustion sources 50,56 . The correlation between OC, EC, and tracer elements is presented in Table 3.

Correlations Between OC and EC.
As shown in Table 3, the correlations of OC-K was highest outdoors, followed by OC-Zn and OC-Pb, which indicated OC outdoor was mainly from biomass burning and vehicle emission. For EC outdoor was mainly from vehicle emission. The correlations of OC-Zn and OC-Pb were the highest in the dormitory, which showed that OC in the dormitory was mainly from the vehicle exhaust, for EC in the dormitory was mainly from the vehicle exhaust. The correlations of OC-Zn, OC-K and OC-Pb were the highest in the laboratory, which showed that OC in the laboratory was mainly from the vehicle exhaust and biomass burning, for EC in the laboratory was mainly from the vehicle exhaust and biomass burning. The recent study has suggested that residential coal burning in the rural areas is an important source of organic matters in north China in winter 57 .
Factor analysis. The principle of TOR analysis of carbon aerosol components was the separation of eight carbon components from each sample by using different temperature gradients. Cao et al. 58 observed that OC1 and OPC (optical pyrolized carbon) were abundantly produced during biomass combustion; OC2, OC3, OC4, and EC1 were abundantly produced during coal combustion and vehicle exhaust; EC2 and EC3 were abundantly produced during diesel vehicle exhaust. Therefore, the abundance distribution of eight carbon components can be used to analyze the source of carbon aerosol 59 . Figure 7 illustrates the abundance distribution of the carbon components observed in outdoor, dormitory, and laboratory samples during the sampling period. The results revealed higher abundance of EC1, OC2, OC3, OC4 and OC1 indoor and outdoor.
To quantify the main sources of carbon pollution in indoor and outdoor PM 2.5 , the eight carbon components was evaluated by factor analysis. The factor analysis for the carbon component in indoor and outdoor samples are presented in Table 4. The outdoor high load components for factor 1 were EC1 (0.926), OC3 (0.851), OC1 (0.833) and OC2 (0.832), which accounted for 74.61% of the carbon component and mainly represented OC was from coal combustion, gasoline vehicle exhaust emissions and biomass burning, EC was from coal combustion and gasoline vehicle exhaust emissions. The high load components in the dormitory for factor 1 were

Conclusion
In this study, the carbon components in indoor and outdoor PM 2.5 in winter of Tianjin were investigated. The mass concentration of OC from outdoor, dormitory, and laboratory samples were 17.01, 19.48, and 18.92 µg/m 3 , the mass concentrations of EC were 7.97, 3.56, and 3.53 μg/m 3 , and the ratios of TC to PM 2.5 were 23.6%, 29.8%, 30.5%, respectively. This result indicated that carbonaceous aerosol was a crucial component in PM 2.5 , and the indoor pollution source certainly influenced the concentration of OC. Compared with major cities in China and abroad, the concentration of carbon components in this study was higher during winter; therefore, the control of carbon component emissions should be strengthened. Lower wind speed and higher  www.nature.com/scientificreports/ relative humidity were helpful to the accumulation of PM 2.5 . However, PM 2.5 accelerated the diffusion with lower temperature and wind speed increasing. The filtration factors of the dormitory and outdoor, the laboratory and outdoor for OC were 0.72 and 0.85. The filtration factors of the dormitory and outdoor, the laboratory and outdoor for EC were 0.59 and 0.71. The average OC/EC for outdoor, dormitory, and laboratory samples were 3.58, 5.78, and 7.84, respectively. Moreover, the SOC/OC for outdoor, dormitory, and laboratory samples were 54.14%, 53.47%, and 37.77%, respectively. This finding indicated that SOC was actively produced in winter, and attention should be paid to SOC pollution.
The correlation coefficients of OC and EC for outdoor, dormitory, and laboratory samples were 0.9070, 0.7293, and 0.6422, respectively. The correlation between OC and EC in outdoor samples was higher than that in dormitory and laboratory samples, which indicated similarity between the pollution sources of OC and EC in outdoor samples. The elemental tracers and factor analysis method were used to analyze the quantitative contribution of carbon pollution in PM 2.5 . The results showed that the carbon aerosol indoors and outdoors was mainly from coal burning, vehicle exhaust, and biomass burning.

Data availability
The datasets generated during or analysed during the current study are available from the corresponding author on reasonable request.