Source Profiles of Volatile Organic Compounds from Biomass Burning in Yangtze River Delta , China

The volatile organic compounds (VOCs) associated with biomass burning were characterized in the Yangtze River Delta of China, including two types of burning conditions (stove burning and field burning) and five typical kinds of biomass (straws of rice, wheat, bean and rape, and wood). According to the results, the VOC emission factors of straw burning ranged from 2.08 g/kg to 6.99 g/kg with an average of (4.89 ± 1.70) g/kg, compared to 0.98 g/kg for wood burning. Some differences in VOC composition were observed with the burning of different biomasses. Oxygenated VOC (o-VOC) were the largest contributors to the mass concentration of measured VOCs from straw burning, with a proportion of 49.4%, followed by alkenes 21.4%, aromatics 13.5%, alkanes 10.6% and halogenated VOC (x-VOC) 5.0%. More aromatics and x-VOC were emitted from wood burning compared with straw burning. Field burning emitted more o-VOC due to more air being supplied during the burning test compared with stove burning. Further examination of the detailed VOC species showed the most abundant VOC species from biomass burning were o-VOC, C2–C3 alkenes and C6–C7 aromatics. The ozone formation potential (OFP) of VOCs from straw burning was in the range of 13.92–33.24 g/kg, which was much higher than that of wood burning (4.30 g/kg). Alkenes and o-VOC were the largest contributors to OFP of VOCs from biomass burning. The top five contributors of OFP were ethene, n-hexanal, propylene, acetaldehyde and methyl vinyl ketone, the sum of which accounted for 77% of total OFP. The ratio of ethylbenzene to m,p-xylenes from biomass burning was significantly higher than those from other VOC sources, and thus this could be seen as the fingerprint of biomass burning.


INTRODUCTION
Biomass was widely used as a biofuel in the rural of China.It was reported that more than 550 million tons of biomass was burned in China in year of 2005 (National Bureau of Statistics, 2006).Biomass burning is known as a significant source of gaseous and particulate pollutants to the atmosphere, which causes serious local and regional air pollution (Levine et al., 1995;Andreae et al., 2005;Koppmann et al., 2005;Bo et al., 2008;Li et al., 2008;Wei et al., 2008;Zhang et al., 2011).Moreover, some of the emissions with hazardous pollutants have adverse impacts on human health (Johnson et al., 2005).
The YRD region locates in the southeast of China, and the meteorological conditions and the types of biomass are different from those in northern China (National Bureau of Statistics, 2006).Furthermore, Jiangsu and Anhui, two large provinces in YRD, are both among the top five provinces which have abundant biomass burning (Li et al., 2009b).However, there have been few reports about the characterization of VOCs from biomass burning in YRD.It is necessary to get better understanding on the emission characteristics of VOCs from local biomass burning in YRD.In this study, two types of burning tests, namely field burning and stove burning, were conducted to study the characterization of VOCs from typical biomass burning in YRD, as discussed in section Emission factors and Source profiles.The ozone formation potential (OFP) of VOCs from biomass burning was estimated in section The OFP of VOCs from biomass burning.The characterization of VOCs from biomass burning in different studies was compared in section Comparison with other studies.In section Comparison with other VOC sources, the characterization of VOCs from biomass burning was compared with those from other VOC sources and the potential fingerprint of biomass burning was identified.The uncertainty of VOC characterization from biomass burning was discussed in section Uncertainty analysis.

Biomass Burning Tests and Sampling
A burning chamber was employed to simulate the field biomass burning, which included a smoke collecting unit, a smoke aging unit, and a sampling unit, as shown in Fig. 1 (left).The pre-weighted biomass was directly burnt on the farmland, and the flue gas was drawn into the collecting unit by the ventilator settled at the end of the burning chamber.Notably, the flow of the ventilator might influence the normal burning if the air velocity at the chamber inlet was larger than the indoor air velocity, namely as 0.25 m/s (Guo, 2011).Thus, the chamber inlet should be large enough.In this study, the air velocity at the chamber inlet was designed as about 0.12 m/s.The weather was windless (< 2 m/s) with the temperature of about 25-30°C during the field test, which has limited impact on the burning process.
After short aging process, the flue gas was sampled into the dilution system (Dekati Fine Particle Sampling (FPS) 4000) by an isokinetic sampling inlet in the sampling unit.The dilution rate was set ~15, and under this condition the temperature of flue gas could be decreased to the ambient temperature and no condensed water could be observed.A gas sampler was connected to the dilution system.The flue gas from the dilution system went through a Teflon filter to remove the particles, and then was pumped into a clean and vacuumed Teflon bag (20 L) by a membrane pump at controlled flow rates (10-100 L/h).PFA tubes were used to connect each part as its low adsorption.After sampling was completed, a portion of the flue gas was immediately transferred from the Teflon sample bag to a 6-L stainless steel, vacuumed SUMMA canisters (Entech Inc.) for the stable storage of gaseous VOCs.Carbon monoxide (CO) in the flue gas was measured by a CO analyzer (Ecotech9830B) simultaneously with a particle-filter removing the particles.
For stove burning, the biomass was directly burnt in stove.The experiment was similar to the "Water Boiling Test (WBT)".In WBT, the testers use a pre-weighed bundle of biomass to raise the temperature of a specific quantity of water from room temperature to boil, and then keep the water boiling for approximately 40 minutes (Wang et al., 2009).The sampling inlet was inserted to the chimney, and the sampling process was similar to that of field burning, as shown in Fig. 1 (right).
For quality assurance and quality control (QA/QC), all the Teflon bags and SUMMA canisters used in each test were cleaned with high-purity nitrogen (> 99.999%) at least three times and evacuated in the laboratory.All the Teflon bags were solely used once to prevent cross contamination.The zero air from the dilution system was also collected as blank samples.The weight of the biomass for each test was kept at about several kilograms and the tests duration ranged from 45 to 60 minutes.After each test, all the char and ash remaining was cleaned.Tests were repeated twice for each type of biomass burning to get reliable results.In total, 14 biomass/stove combinations were tested.After each test, a small bundle of biomass was sealed and brought to the laboratory for proximate analysis.The proximate analyses of the biomass were conducted according to the Chinese national standard methodologies (GB/T 212-2008) for proximate analysis of solid fuels.
The biomass selected for this study included straws of rice, wheat, bean and rape, and wood.Straws of rice, wheat, and bean were very important biofuels used for household energy in China (National bureau of Statistics, 2006).In particular, straws of rice, wheat, corn, rape, and bean were the five most abundant crop wastes in YRD region (Han et al., 2002;Zhang et al., 2008).In this study, maize straw was not included because it was unavailable in summer when the present study was conducted.More work is needed for the characterization of maize straw burning in future.In addition, one types of wood was selected for comparison with straws.

VOCs Analysis
VOC samples were analyzed by one Gas Chromatograph with a Mass Spectrometer and a Flame Ionization Detector (GC-MS/FID).Firstly, VOC samples were pumped into a cryogenic pre-concentrator (TH_PKU-300, Tianhong, China), and were concentrated at -150°C by two traps, respectively.The concentrated VOCs were desorbed at 100°C and were injected into the gas chromatograph (GC2010, Shimadzu, Janpan).The C2-C5 hydrocarbons were separated on a PLOT capillary column ( 0.32 mm × 15 m, Dikma, USA) and were quantified by the FID.The C5-C10 hydrocarbons, halogenated VOCs (x-VOC), and oxygenated VOCs (o-VOC) were separated on a DB-624 ( 0.25 mm × 60 m, Agilent, USA) and were quantified using a quadrupole mass spectrometer (GCMS-QP2010E, Shimadzu, Japan).The source temperature in the MS was 200°C, with a scan mass ranging from 30 to 300 amu.
The VOC species were identified by their retention time and mass spectrums.A commercial standard gas (Spectra, USA) containing PAMS (Photochemical Assessment Monitoring System), o-VOC, and x-VOC was used to confirm compounds' retention time and identify compounds.In this study, 103 species including 28 alkanes, 11 alkenes, 16 aromatics, 16 o-VOC, 31 x-VOC and acetylene were identified, as listed in Table 1.The target species were quantified by using multipoint external calibration method.Calibration curves for all species were made before and after the analysis and had good linear regression (R 2 > 0.99).For every batch of 6 samples, a high-purity nitrogen sample and a standard sample (1 ppb) were spiked to assure the peak time and signal intensity.The method detection limits (MDLs) of the various VOC species ranged from 2 to 70 pptv (Liu et al., 2009;Yuan et al., 2012).

Determination of Emission Factors and Source Profiles
The emission factor was typically calculated by the carbon mass balance method (Zhang et al., 2000;Dhammapala et al., 2006;Wang et al., 2009;Li et al., 2011).In this method, all carbon was assumed to be emitted into the atmosphere as carbonaceous particles and carbonaceous gases, which required the complete measurement of the carbon both in the emissions and in the biomass and ash (Zhang et al., 2000).
In this study, the emission factor of VOCs was determined by the measurement of the total VOC emission amount and the dry weight of biomass.The total VOC emission amount was calculated through multiplying the VOC concentration in the flue gas by the total volume of the flue gas which was equal to the product of the air velocity and the cross section area of chimney (or the burning chamber).The dry weight of the biomass was obtained according to the weight of the biomass and its moisture content.
For the comparison of VOC composition from different biomass burning, VOC source profiles of different biomass burning were obtained according to the mass percentage of each VOC group.

Emission Factors
VOC and CO emission factors of different biomass burning were calculated by the method mentioned in section Determination of emission factors and source profiles, as shown in Fig. 2. For field biomass burning, VOC emission factors of straws were rape (2.90 ± 0.39) g/kg, rice (2.74 ± 0.10) g/kg, bean 2.55 g/kg, and wheat (1.98 ± 0.01) g/kg, respectively.In terms of stove burning, VOC emission factors were rice (6.98 ± 0.10) g/kg, rape (4.17 ± 0.33) g/kg, bean (3.06 ± 0.98) g/kg, and wood 0.98 g/kg respectively.CO emission factors of biomass burning showed similar variation trend with those of VOCs but were larger by one order of magnitude.CO emission factors of straw burning ranged from 17.47 to 66.32 g/kg, and the value of wood burning was 21.09 g/kg.Emission factors of VOCs were positively correlated with those of CO with the R 2 = 0.57 (n = 14), as shown in Fig. 3 (right).From Fig. 2, we can see that the emission factors of stove burning were a little larger than those of field burning, probably due to the incomplete collection of the flue gas by the burning chamber.We can also see that the emission factor of wood burning was significantly lower than that of straw burning (t-test, p < 0.001) because the dense structure of wood slowed down its emission of the volatile components and finally led to a more complete burning of the volatile components during the test (Li et al., 2011).For emission factors of straw burning, those of rice and rape straws were generally higher than those of bean and wheat straws (t-test, p < 0.001).
Many factors might influence the VOC emission factor, such as the biomass types or the burning conditions as mentioned above.Additionally, the calorific value of the biomass might also influence the VOC emission factors.As indicated in Fig. 3 (left), VOC emission factors were negatively correlated with calorific values of the biomass with a correlation coefficient over 0.9 (R 2 = 0.9213, n = 4) for stove burning.However, no significant correlations were observed between VOC emission factors and the calorific values in field burning.Probably, the underestimation of VOC emission factors in field burning due to the incomplete collection of flue gas might be the major reason of the poor correlations.

Source Profiles
The composition of VOCs from different biomass burning displayed some similarity as shown in Fig. 4. o-VOC were    the primary contributors to VOCs from straw burning in the field, with a contribution of 49.4%, followed by alkenes (21.4%), aromatics (13.5%), alkanes (10.6%), x-VOC (5.0%).
Compared with VOC composition of field straw burning, the contribution of o-VOC from stove burning decreased to 42.1%, and by contrast the contribution of alkenes and aromatics was relative high.This was mainly because of the more complete oxidation of volatile components with more air supplied in field burning.Some variations were identified between VOC compositions from different biomass burning.
Taking the stove burning as an example, alkenes accounted for 19.3% of VOCs in wood burning, compared to 26.9% in straw burning, while the contribution of aromatics and x-VOC was relative high in wood burning, as indicated in Fig. 4.
In summary, there were no significant differences among VOC source profiles of biomass burning in the same burning condition.o-VOC, C2-C3 hydrocarbons and C6-C7 aromatics were the major VOC species from biomass burning.The contribution of o-VOC was larger in source profiles of field burning than those of stove burning.Our results displayed some similarity with the previous studies.As reported by Wang et al. (2009), C6-C7 aromatics, propylene, C2-C3 o-VOC were the major VOC species from typical biomass burning in northern China.Of the hydrocarbons, C2-C3 hydrocarbons and C6-C7 aromatics were abundant both in Liu et al. (2008) and Tsai et al. (2003) study.Nevertheless, significant variations of the specific proportions of VOC species were identified between this study and previous studies.For example, proportions of benzene and propylene were the largest with the values of (17.3 ± 8.1)% and (11.3 ± 3.5)% in Wang et al. (2009) study, respectively, compared to (6.4 ± 2.9)% and (4.8 ± 1.0)% in this study.Variations of measured VOC species in different studies played an important role in the variations of VOC source profiles of biomass burning.

The OFP of VOCs from Biomass Burning
The OFP of VOCs from different biomass burning was calculated based on VOC emission factors and source profiles combining with the maximum incremental reactivity (MIR) of each species by Eq. ( 1) below: where, OFP is the ozone formation potential of VOCs emitted from per unit of biomass burning, g/kg; p i is the mass percentage of VOC i in total VOCs, %; MIR i is the MIR of VOC i , g O 3 /g VOC i ; EF is the VOC emission factor, g/kg.The MIR values can be cited in Carter (2008).
The OFP of VOCs from different biomass burning, and the contribution of each chemical group to the total OFP were shown in Fig. 6.Generally, the OFP could be categorized into three groups.The first group was the OFP of field straw burning whose values ranged from 8.06 to 12.20 g/kg, and these values were slightly higher than that of Li et al. (2009b) with an average of 7.0 ± 0.85 g/kg.The reason might be that the o-VOC and x-VOC were not included in the study of Li et al. (2009b).The second group was the OFP of stove straw burning which was 13.92-33.24g/kg.The third group was the OFP of wood burning with the lowest value of 4.30 g/kg.The OFP of different biomass burning was mostly influenced by the various VOC emission factors.
From Fig. 6, we can see that alkenes and o-VOCs were the largest contributing species to the OFP, and their percentages were 41-61% and 28-42%, respectively, followed by aromatics 7-14%.The contribution of both alkanes and x-VOC to the OFP was lower than 2%.Further examination showed that the top five abundant species of the OFP were ethylene (36%), n-hexanal (17%), propene (13%), acetaldehyde (6.4%), and methyl vinyl ketone (3.8%).Of the aromatics, toluene was the most important species of the OFP with the percentage of 2.3%.

Comparison with Other Studies
VOC emission factors of biomass burning in different studies were summarized in Fig. 7.For straw burning, VOC emission factors ranged from 1.98 to 9.62 g/kg in different studies with an average of (4.02 ± 2.21) g/kg, except those performed in Jan and April by Wang et al. (2009).As reported by Wang et al. (2009), VOC emission factors of straw burning were greatly influenced by the ambient temperature, and the values were 0.23-1.48g/kg in Jan and 7.36-26.57g/kg in April, respectively, which were significantly different from those in other studies (t-test, p < 0.001).In terms of the results from the other studies, there were no significant differences among the emission factors of different straws burning (Kruskal-Wallis test, p > 0.05).The present study was conducted in late June and early July which was the typical harvest season in YRD.The other two studies did  Li et al., 2011;c, Zhang et al., 2000;d, Wang et al., 2009; stove-imp, improved stove; Jan, result in January; Apr, result in April; kang, a commonly used stove in northern China for heating bed).not report the season or the ambient temperature of the experiments (Zhang et al., 2000;Li et al., 2011).The effect of the ambient temperature on VOC emission factors of straw burning needs further more studies.
VOC emission factors of wood burning were significantly lower than those of straw burning (t-test, p < 0.001), ranging from 0.08 to 4.95 g/kg with the average of (1.76 ± 1.41) g/kg, as shown in Fig. 7. Our value was 0.98 g/kg and fell in the range.VOC emission factors of wood burning ranged over one order of magnitude in the studies listed in Fig. 7.This large variation might result from different types of wood studied in those studies.
In addition to the differences of emission factors, variations of VOC composition were also observed among different studies.As shown in Fig. 8, the percentages of o-VOC and alkenes were much larger than those of aromatics and alkanes in the present study.While, according to the study of Wang et al. (2009) and her group (Li et al., 2011), the contribution of aromatics to total VOCs was dominant.It should be pointed out that specific VOC species measured in those studies were not always same, as shown in Table 2, which was probably one important reason of variations of the VOC composition.

Comparison with Other VOC Sources
The ratios of specific VOC pairs were usually used as the fingerprint for distinguishing or identifying VOC emission sources.For example, the ratio of toluene to benzene (T/B) was recognized as the characteristic of vehicle emissions with a value of 1.5-2.0(Schauer et al., 2002).The ratios of T/B and ethylbenzene to m,p-xylenes (E/X) in major VOC combustion sources were summarized in Fig. 9 (Liu et al., 2008;Wang et al., 2009;Li et al., 2011;Qiao et al., 2012).For the ratio of T/B, gasoline vehicle exhausts had the highest value, followed by the gasoline evaporation and diesel vehicle exhausts.The ratios of T/B were the lowest Fig. 8.Comparison of the VOC composition from biomass burning among different studies (a, this study; b, Li et al., 2011;d, Wang et al., 2009).for the emissions from coal burning and biomass burning, the difference between which was not significant.Thus, the ratio of T/B was not suitable to distinguish coal burning emissions from biomass burning emissions.
For the ratio of E/X, all VOC combustion sources except biomass burning showed similar values of (0.49 ± 0.12) which were much lower than that of biomass burning with the average of (1.73 ± 1.06).Accordingly, we could easily identify VOC emissions of biomass burning from the other VOC combustion sources by the ratio of E/X.It should be pointed out that the ratio of E/X was usually used for estimating the photochemical age of the air mass (Mckeen et al., 1996;Shao et al., 2011), because ethylbenzene and m,p-xylenes had similar sources but different chemical reactivity in the atmosphere.Consequently, the possible influence of emissions from biomass burning should be taken into account on the estimation of the air mass photochemical age based on the ratio of E/X.

Uncertainty Analysis
Up to now, many studies were carried out to explore the characterization of VOCs from biomass burning.It seemed that the VOC emission factor of the straw burning was around several g/kg.Nevertheless, significant discrepancies of VOC emission factors were identified among different studies (compared in section Comparison with other studies) due to various burning conditions and measurement methods, as shown in Table 2.The discrepancy of VOC emission factors of wood burning among different studies was even more significant.Therefore, there might be two major sources of the uncertainty of VOC characterization from biomass burning.
One reason was the variability of emissions from different burning tests, largely associated with the variations in fire tending behavior (Tsai et al., 2003).The burning cycle usually began with the ignition which was characterized by incomplete combustion, followed by a stable and complete combustion, and ended with combustion of the remaining charcoal.VOC characterization was different with firing processes (Wang et al., 2009).Unfortunately, one entire burning cycle with three processes was almost impossible to be completely repeated by each test, not to mention conducted by different testers.This may be the explanation for the discrepancy among different studies.In addition, the types of biomass also had influence on VOC emissions.Thus, more detailed combustion process should be conducted for better understanding VOC emissions from biomass burning.The other reason of the uncertainty of VOC characterization might come from the measurements.In previous studies, VOC emission factors were usually determined by the carbon mass balance method, in which the accurate measurement of each part of carbon mass was very important.In this study, the emission amount of VOCs was obtained through multiplying the VOC concentration in flue gas by the total flue gas volume.Thus, the measurement of the flue gas volume was very crucial for this method and obviously this method underestimated emission factors of the field burning as the incomplete collection of the flue gas according to our results.In addition, the normative measurement of VOCs was also essential to study VOC characterization of different biomass burning.As shown in Table 2, detailed VOC species measured in different studies were not very similar.For example, in some studies o-VOC were not included, which were recognized as important VOC species from biomass burning.Even in some studies, VOCs were measured as the total non-methane hydrogen carbon (TNMHC).Accordingly, total VOCs were different as different VOC species were measured in those studies.It would be helpful to apply a similar list of the VOC species in future studies of VOC characterization of different sources, which would be also useful to identify the possible fingerprint of different VOC sources.

CONCLUSIONS
In this study, the characterization of VOCs emitted from biomass burning was investigated in YRD, including two types of burning conditions and five types of biomass.Based on the results, VOC emission factors of stove straw burning ranged from 2.08 to 6.99 g/kg with the average of (4.89 ± 1.70) g/kg and the OFP of VOCs emitted from per kilogram straw burning was 13.92-33.24g/kg.The VOC emission factor was 0.98 g/kg for wood burning in stove and its OFP was 4.30 g/kg.Emission factors of field burning were easily underestimated due to the incomplete collection of the flue gas.VOC emission factors of wood burning ranged over one order of magnitude in different studies, and more studies are necessary in the future to investigate VOC emission factors of wood burning.Burning conditions and measurement methods should be paid much attention to because both of them might cause large uncertainty of VOC characterization of biomass burning.Some differences of VOC composition were observed among different biomass burning.For field straw burning, the contribution of o-VOC was 49.4%, followed by alkenes (21.4%) > aromatics (13.5%) > alkanes (10.6%) > x-VOC (5.0%).The percentage of o-VOC from stove burning was lower than that from field burning due to more air supplied in field burning.Wood burning emitted more aromatics and x-VOC compared to straw burning.In terms of detailed VOC species, there were no large differences among the major contributing VOC species from different biomass burning.The most abundant groups were o-VOC, C2-C3 hydrocarbons, and C6-C7 aromatics in VOCs from biomass burning.Alkenes and o-VOC were the largest contributors of OFP, accounting for 41-61% and 28-42%, respectively.Ethylene, n-hexanal, propene, acetaldehyde, and methyl vinyl ketone were the top five contributing species of OFP, proportioning 77% of total OFP.VOC source profiles of biomass burning had significant large ratio of E/X with the value of (1.73 ± 1.06) which could be recognized as the fingerprint of VOCs from biomass burning.

Fig. 1 .
Fig. 1.The schematic of the sampling system for field burning (left) and stove burning (right).

Fig. 2 .
Fig. 2. Emission factors of VOCs and CO from different biomass burning.

Fig. 3 .
Fig. 3. Correlations between emission factors of VOCs and calorific values of the biomass (left) and emission factors of CO (right).

Fig. 4 .
Fig. 4. The composition of VOCs from different biomass burning.

Fig. 5
Fig. 5(a).Source profiles of VOCs from different biomass burning in the field.

Fig. 5
Fig. 5(b).Source profiles of VOCs from different biomass burning in stove.

Fig. 6 .
Fig. 6.OFP of VOCs from different biomass burning and the contribution of each VOC group to total OFP.

Fig. 7 .
Fig. 7. Comparison of VOC emission factors of biomass burning in different studies.(a, this study; b, Li et al., 2011; c, Zhang et al., 2000; d, Wang et al., 2009; stove-imp, improved stove; Jan, result in January; Apr, result in April; kang, a commonly used stove in northern China for heating bed).

Table 1 .
VOC species obtained in this study.