Article pubs.acs.org/est
Semivolatile and Volatile Organic Compound Emissions from WoodFired Hydronic Heaters Johanna Aurell,† Brian K. Gullett,‡,* Dennis Tabor,‡ Abderrahmane Touati,§ and Lukas Oudejans∥ †
National Research Council Post Doctoral Fellow to the U.S. Environmental Protection Agency Office of Research and Development, National Risk Management Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, United States § ARCADIS U.S., Inc., 4915 Prospectus Drive, Durham, North Carolina 27713, United States ∥ Office of Research and Development, National Homeland Security Research Center, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, United States ‡
S Supporting Information *
ABSTRACT: Emissions including polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs), polyaromatic hydrocarbons (PAHs), and volatile organic compounds (VOCs), were sampled from different wood-fired hydronic heater (HH) technologies. Four commercially available HH technologies were studied: a single-stage conventional combustor with natural updraft, a three-stage downdraft combustion system, a bottom-fed pellet burner, and a two-stage heater with both a combustion and gasification chamber. The fuel consisted of three wood types (red oak, white pine, and white ash), one hardwood pellet brand, and one fuel mixture containing 95% red oak and 5% residential refuse by weight. The various HHs and fuel combinations were tested in a realistic homeowner fuel-charging scenario. Differences in emission levels were found between HH technologies and fuel types. PCDD/PCDF emissions ranged from 0.004 to 0.098 ng toxic equivalency/MJinput and PAHs from 0.49 to 54 mg/MJinput. The former was increased by the presence of 5% by weight refuse. The white pine fuel had the highest PAH emission factor, while the bottom fed pellet burner had the lowest. The major VOCs emitted were benzene, acetylene, and propylene. The highest emissions of PAHs, VOCs, and PCDDs/PCDFs were observed with the conventional unit, likely due to the rapid changes in combustion conditions effected by the damper opening and closing.
■
INTRODUCTION The prevalence of wood-fired hydronic heaters (HHs) for residential heat and hot water supply has increased in the U.S., especially in response to recent increases in fossil fuel costs. The increased use of HHs is not without controversy, as some HHs can have significant visible smoke pollution and odor, resulting in localized residential impacts. Despite these concerns, emissions from the various HH technologies have been minimally characterized. Conventional HHs consist of a single-stage combustion chamber process while more modern HHs can have two or three combustion stages. These units burn a variety of fuels, including split logs, manufactured pellets, or wood chips. HHs can run in an “on-demand” heat mode where the opening and closing of an air damper regulates the combustion to satisfy the heat load requirement, or in a heat storage mode where the combustion is continuous, and the heat is collected in a heat storage tank. The variety of technologies and modes of operation lead to disparate combustion quality and emissions. Organic pollutants anticipated include semivolatile organic compounds (SVOCs) such as polyaromatic hydrocarbons © 2012 American Chemical Society
(PAHs), polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and volatile organic compounds (VOCs), all of which are products of incomplete combustion (PICs). PAHs are of concern because some of them have been reported to have mutagenic and carcinogenic properties.1 The PCDDs/PCDFs are recognized as toxic, bioaccumulative, and persistent in the environment.2 Imperfect combustion conditions, anticipated during transient supply of air to damper-regulated HHs, are likely to lead to elevated CO concentrations and have been shown to be associated with increases in PCDD/PCDF emission levels in the presence of chlorine.3 VOCs, also associated with marginal combustion, are comprised of a large group of compounds. Some of the VOCs such as benzene are toxic and some, such as acetylene, styrene, xylene, toluene, ethylbenzene, and 1,2,4- and 1,3,5-trimethylReceived: Revised: Accepted: Published: 7898
March 27, 2012 June 8, 2012 June 13, 2012 July 5, 2012 dx.doi.org/10.1021/es301197d | Environ. Sci. Technol. 2012, 46, 7898−7904
Environmental Science & Technology
■
benzene, are ozone precursors, contributing to formation of ground level ozone. There are only a few reports of organic emissions from residential wood-fired hydronic heater tests in the literature. PCDD/PCDF emission factors have been shown to range from 0.04 to 47 ng toxicity equivalents (TEQ)/kg fuel input with different HH technologies, fuels, operating modes, and unit condition.4−6 The lowest emission factor was obtained from a pellet-fired HH (0.04 ng TEQ/kg) 5 operated in a realistic residential use scenario. This value is a hundred-fold lower than the value that was found from a laboratory pellet burn study by Hedman et al.4 (11 ng TEQ/kg). The highest emission factor (47 ng TEQ/kg) was obtained from a conventional HH in a neglected maintenance condition during a realistic residential use scenario operated in an on-demand heat mode.5 Conventional HHs in good maintenance condition from the same study ranged from 0.53 to 1.5 ng TEQ/kg 5 indicating the importance of unit condition in effecting combustion quality to minimize pollutant formation. Reported PCDD/PCDF emission factors from modern HHs running in on-demand heat modes (0.32−0.37 ng TEQ/kg) 5 were higher than emission factors from modern HHs operating in heat storage mode (0.06−0.13 ng TEQ/kg),6 suggesting that organic emissions are a function of operating mode (on-demand or heat storage mode). PAHs have also been shown to be affected by unit type; conventional HHs have shown up to 90 times higher PAH emission factors than modern HHs.6,7 Benzene emission factors have also been shown to be higher for conventional HHs (608−1730 mg/kg) than modern HHs (3.0−110 mg/kg).7 Fuel composition is also expected to have an effect on emissions. As homeowners may be expected to include residential waste with biomass fuels, Hedman et al. 4 showed that mixing chlorine-containing plastic with wood logs can increase PCDD/PCDF emission factors. Another potential reason for the range of these reported emission factors is the variety of sampling procedures used. For example, in order to ensure consistency in determining unit to unit comparisons, fuel variability is minimized in ASTM Method E2618 8 through the use of stacked crib wood with specified wood type, dimensions, and moisture content. Emissions are determined as a weighted average from results at different operating loads. In contrast, the European method (EN303−5) 9 uses a full operating load, without periodic measurements, to determine an average emission level. The differences in these standardized methods are likely to result in differences in the resulting emission factors. Further, because these methods were not intended to portray the fuel types, loading procedures, and operating methods that might be representative of typical homeowner operation, published values may not reflect actual emissions. Overall, these studies indicate that emission factors depend on HH technology, unit condition, mode of operation (ondemand or continuous), sampling protocol, and fuel type.4−7 To characterize organic emissions from HHs more fully under conditions representative of homeowner operation and with a variety of technologies, both common and new, this study investigates four different HH technologies with multiple fuel types. The emission factors can be used to estimate the impact of these technologies on localized exposure and to enhance inventories that are used for energy and emissions planning.
Article
EXPERIMENTAL SECTION
Hydronic Heater Units. Four different commercially available technologies were investigated: a conventional single stage updraft HH (with a maximum heat output of 18 kW), a three-stage downdraft HH (50 kW), a European two-stage bottom fed pellet burner (40 kW), and a U.S. two-stage downdraft heater (44 kW). The rate of heat supply was monitored and controlled by the water temperature in the recirculation loop for all of the units, except for the U.S. downdraft system, which was operated in full heat mode. The dampers on the conventional and three-stage HHs were opened and closed when the lower and upper temperature limits in the water recirculation loop were reached, respectively. The European pellet burner’s fuel screw feeder was regulated by the temperature in the water recirculation loop. The U.S. downdraft system consisted of both combustion and gasification chambers, where air was continuously added to the firebox and blown downward through the wood logs. Supporting Information (SI) Figure S1 shows illustrations of the different technologies. Fuel. Three split wood types, red oak (Quercus rubra), white pine (Pinus strobes) and white ash (Fraxinus americana), and one hardwood pellet brand were studied representing a range of fuels available in the U.S. Northeast. One common wood, red oak, was selected to compare across technologies. SI Table S1 shows the properties of the fuels. The 6 mm diameter commercial pellet fuel was a made out of sawdust from different wood processing industries and consisted of a blend of hardwood (no bark), mostly oak. Another fuel mixture consisted of a mixture of red oak and 5% by weight refuse. The composition of the refuse was based on a typical residential waste from the New York state area described by Lemieux et al.10 Experimental Setup. The HH units were located outdoors, allowing for testing and sampling under ambient conditions. The duct work was designed to follow the American Society for Testing and Materials Method E2515 11 specifications for duct bends, duct length, and dilution hood dimensions. A conical hood cone was placed above the outlet of the stack to collect and convey the emissions, diluted by entrainment of ambient air, through a stainless steel duct to the facility’s air pollution treatment system (SI Figure S2). To ensure collection of emissions without wind gusts influencing the system draft, a canvas shroud was hung from the hood (not shown). The duct system moved ∼20 m3/min (dry) of air, which correlated to a dilution ratio of 5−10 to 1 from the heater stack. Flows and pressures were controlled by an induction draft fan. Experimental Runs. The heat load profile was derived from a simulation program for heat demand (Energy-10 simulation, National Renewable Energy Laboratory [NREL]) for a 232 m2 home in Syracuse, NY using an averaged hour-perhour heat load for the first two weeks of January, averaged over 25 years. The 24 h heat load demand was simulated by cooling the heated output water of the unit by a chilled water heat exchanger (SI Figure S2) and controlling the flow to meet the heat load profile (SI Figure S3). A heat storage tank was simulated for the U.S. downdraft unit by adding a water/air heat exchanger in series to the water/water heat exchanger. The combustion runs for each of the technologies were different in length due to differences in their firing rates and operation procedures, but each satisfied the same load demand. A full 7899
dx.doi.org/10.1021/es301197d | Environ. Sci. Technol. 2012, 46, 7898−7904
Environmental Science & Technology
Article
charge of logs for the conventional HH (115 ± 7 kg) and the three-stage HH (91 ± 1 kg) lasted for 12 h while a full load of the U.S. downdraft HH (28−30 kg) lasted 3.5 h. Because the pellet-based European unit operated in an on-demand fuel mode, the 12 h heat load was compressed to 6 h without compromising the simulation. Before each test, the conventional HH and three-stage HH heaters were preheated by burning a load of 45 kg, whereas the U.S. downdraft was preheated with a load of 3−4 kg. The experiments and emission samples are shown in Table 1.
spectrometry (HRGC/HRMS) using a Hewlett-Packard gas chromatograph 6890 Series coupled to a Micromass Premier (Waters Inc., UK) with a DB-Dioxin 60 m × 0.25 mm × 0.15 μm column (Agilent/J&W Scientific). The 16 U.S. EPA PAHs were sampled isokinetically via U.S. EPA Method 001013 from the indoor dilution duct (SI Figure S2). The PAHs were analyzed according to U.S. EPA Method 8270D14 using a Hewlett-Packard gas chromatograph 5890 Series II with a 60 m × 0.25 mm × 0.15 μm column (Agilent/ J&W Scientific). The isotope dilution method was used for quantifying PCDD/PCDF concentrations. All PCDD/PCDF surrogate recoveries except for one sample (CL5036, red oak + refuse) were within the standard method criteria, 70−130%. The PAH samples were spiked preanalysis according to U.S. EPA Method 8270D14 but no recovery information could be obtained due to very high PAH levels in the collected samples resulting in high dilution levels in the sample extracts. The PCDDs/PCDFs and PAHs were normalized to 1 atm, 21 °C. PCDD/PCDF values are reported as Totals (tetra- to octachlorinated) as well as Toxicity Equivalent (TEQ) using recent toxicity equivalency factors (TEFs).15 VOCs were sampled over a period of about 3 min from the indoor dilution duct using a Summa canister (6 L) according to U.S. EPA Method TO-1516 and were analyzed by gas chromatography/mass spectrometry (GC/MS). Naphthalene and benzene were measured in real time from the indoor dilution duct with a resonant enhanced multiphoton ionizationtime-of-flight mass spectrometer (REMPI-TOFMS) instrument, described elsewhere.17 Emission factors for PAHs and VOCs were normalized by fuel energy input. Carbon monoxide (CO) was measured continuously in the stack and the dilution duct; CO2 was measured only at the stack. The CO measurements were conducted using U.S. EPA Method 1018 and the CO2 measurements were performed according to U.S. EPA Method 3A.19 CO was measured continuously in the stack and the dilution duct by a Thermo Electron model 48 gas filter correlation analyzer and a Rosemount 880A analyzer, respectively. CO2 was measured at the stack by a California Analytical ZRH analyzer. Emission Factor Calculations. The total carbon emitted (from CO and CO2) and the fuel carbon fraction was used for calculating the mass of fuel burned for the emission factor calculations. The emissions are presented as ng or mg per kg fuel input (eq 1), or ng or mg MJ fuel input (eq 2).
Table 1. Experimental Matrix samples per run
conventional HH
three-stage HH U.S. Downdraft European pellet
fuel
runs
VOC
REMPI/ TOFMS continuous benzene/ naphthalene
red oak
3
1
5a
3b
1b
red oak + refuse white pine white ash red oak red oak
3
1
4a
0
1b
3
1
4a
0
0
2
1
1a
0
0
3 3
1 1
a
2 1a
b
4 3b
1b 1b
pellet
3
1c
1a
0
1b
PCDD/ PCDF PAH
a
Samples from one run only were analyzed. bSamples collected from one run only. cOne combined sample from three runs.
Sampling Methods and Analyses. PCDDs/PCDFs were isokinetically sampled from the stack during damper openings following procedures in U.S. EPA Method 23 12 (SI Figure S2). The damper openings and closings on the conventional and the three-stage HH led to several filter changes per run, mostly due to increased smoldering due to poor combustion when the damper was closed. The PCDD/PCDF sample cleanup and analyses were performed according to EPA Method 23.12 Sample cleanup was automated using a Power-Prep from Fluid Management Systems. The PCDDs/PCDFs were analyzed by high resolution gas chromatography/high resolution mass
Figure 1. PCDD/PCDF TEQ (ND = 0, DL) and Total (tetra- to octachlorinated) emission factors for seven different heater/fuel combinations. Data labels indicate values. 7900
dx.doi.org/10.1021/es301197d | Environ. Sci. Technol. 2012, 46, 7898−7904
Environmental Science & Technology
Article
Figure 2. Total PAH (16 EPA PAHs) emission factors for seven heater/fuel combinations.
Mass of Emission Mass of fuel burned
(1)
EFMass Heating value of fuel
(2)
EFMass =
EFMJ =
■
90%, respectively, higher than with the conventional singlestage HH and likely attributable to their two- and three-stage combustion chamber designs. PCDD/PCDF. The EFs derived in this study varied 25-fold from 0.004 to 0.098 ng TEQ/MJinput (Figure 1, SI Table S2). Of the 17 TEF-weighted congeners for each sample, between 0 and 11 congeners were nondetects (ND) for the 17 unit/fuel test combinations (SI Table S2). The average ratio of the TEQ value with ND = 0 and TEQ with ND = detection limit (DL) was 0.70, suggesting that the ND congeners did not significantly affect the overall TEQ determination (Figure 1, SI Table S2). The lowest ND = 0/ND = DL ratios, 0.38 and 0.29, were determined for the conventional HH unit burning white pine and the U.S. downdraft unit with red oak, respectively, likely due to their relatively low PCDD/PCDF emission levels in combination with the small sample volume size. The EFs derived in this present study, 0.080 to 1.9 ng TEQ/ kginput, are consistent with biomass burn emission factors of 0.91 to 2.3 ng TEQ/kginput,20 woodstove/fireplace values of 0.25−2.4 ng TEQ/kg,21 pellet and wood heater values of 0.04− 2.0 ng TEQ/kg,4,5 and wood stoves and heaters of 0.3−45 ng TEQ/kg.4,5,22 The 2,3,7,8-Cl-substituted PCDD/PCDF congener pattern for the conventional HH, three-stage HH, and the U.S. downdraft units were all similar; the 1,2,3,7,8-PeCDD and 2,3,4,7,8-PeCDF congeners contributed the most to the TEQ value (SI Table S3). This congener pattern and the pattern from wood stoves using oak fuel21 are quite alike. The PCDD/PCDF congener pattern for the European pellet burner was different from the other units. For this unit, the two most toxic congeners (2,3,7,8-TeCDF and 2,3,7,8-TeCDD) contributed the most to the TEQ value: 44 and 25%, respectively. Total PCDD/PCDF (tetra- to octachlorinated) emission factors ranged from about 0.10 to 3.8 ng/MJ of fuel input (Figure 1). Despite the ∼35-fold range in total emission factors, the relative standard deviation for the unit average emission factors was less than 0.3. The total PCDD/PCDF values averaged about 40 times higher than their respective TEQ values. This ratio is quite lower than other values (ca. 100) from open biomass combustion.23 Despite some wide ranges in emission factors from the limited testing, both device and fuel type were statistically
RESULTS AND DISCUSSION Twenty combustion runs with seven unit/fuel combinations were tested. Measurements on the European pellet burner and
Figure 3. Time-resolved benzene and CO concentrations in the dilution duct over one fuel charge cycle. Red oak in the conventional HH. Peaks correspond to damper openings.
the conventional HH burning red oak had the highest and lowest modified combustion efficiencies (MCEs), (CO2/ (CO2+CO)), of all seven unit/fuel combinations, 98% and 74%, respectively. Burning white pine and white ash in the conventional HH unit increased the MCE to 82% and 86%, respectively, over the red oak. The difference may be due to differences in the moisture content which, for these fuels, was about two times higher for red oak than for white pine and white ash (SI Table S1). Mixing refuse with the red oak also increased the MCE to 88%. The three-stage HH and U.S. twostage downdraft units burning red oak had MCEs of 86% and 7901
dx.doi.org/10.1021/es301197d | Environ. Sci. Technol. 2012, 46, 7898−7904
Environmental Science & Technology
Article
Table 2. Comparison of Benzene and Naphthalene Emission Factors Determined by REMPI-TOFMS and Summa Canister/Gas Chromatography or Filter/XAD Method.a unit conventional HH conventional HH conventional HH conventional HH conventional HH three-stage HH three-stage HH European pellet U.S. downdraft a
sample no.
fuel
REMPI-TOFMS benzene mg/MJ
Summa canister benzene mg/MJ
REMPI-TOFMS naphthalene mg/MJ
filter/XAD- Method 0010 naphthalene mg/MJ
CR1
red oak
NM
37
NM
NA
CR2
red oak
97
NM
131
NA
CR3
red oak
99
NM
138
13
CRR1
74
NM
80
6.9
CP1
red oak + refuse pine
NM
NM
NM
34
TR1 TR2 ER1
red oak red oak pellet
32 NM 0.68
NM 22 NM
27 NM 0.12
2.9 NA 0.21
UR1
red oak
99
23
10
6.9
NM, not measured. NA, not analyzed.
Figure 4. Typical REMPI-TOFMS concentration trace of benzene from the U.S. downdraft unit. Inset: comparison of REMPI-TOFMS and Summa canister concentrations.
observed from the barrel burn studies is consistent with the absence of potentially catalytic metals in the added refuse and an improved combustion environment in the presence of 95% by mass wood. Further, these historical barrel burn values will be even lower when normalized to the basis of initial refuse mass used in this work, rather than mass burned, and, thus, closer to the value of 19 ng TEQ/kg “refuse” determined here. PAHs. The white pine fuel on the conventional HH consistently had the highest PAH emissions, whereas the European pellet burner with hardwood pellets had the lowest (Figure 2). PAH levels seem to be a clear function of fuel type (compare white pine with red oak on the conventional HH) and unit type (compare red oak on the conventional HH, three-stage HH, and U.S. downdraft units). Individual PAH emission factors for all seven unit/fuel combinations are shown in SI Table S4 and S5. Eighty percent of the total PAH (16 EPA PAHs) constituted of naphthalene, acenaphthylene, and phenanthrene. Naphthalene had the highest emissions of all PAHs for all unit/fuel combinations and varied 50 fold between the units. The average naphthalene emission factor for the seven unit/fuel combinations in the present study was 9.8 mg/ MJinput, in the same range as compiled emission factors for
shown to play a role in determining PCDD/PCDF emissions. However, this conclusion is tempered by the limited sample size, n varied from n = 1 to n = 3. For all fuel types, the conventional HH unit had higher emissions than the other three units. The low combustion efficiency of the conventional HH unit (74%) suggests that combustion quality also plays a role in determining the total PCDD/PCDF emission levels, consistent with current theory on PCDD/PCDF formation mechanisms.3 Among the conventional HH results, the addition of refuse increased emissions over red oak alone, and the red oak + refuse values were distinctly higher than those from the white pine and white ash. The contribution of the five mass percent refuse toward the PCDD/PCDF TEQ in the red oak + refuse emission factor can be calculated using the red oak-only emission factor. The refuse would have to have an emission factor of 19 ng TEQ/kg “refuse” to reach the red oak + refuse emission factor. This value is consistent with earlier determinations 24,25 using this same approximate refuse composition that resulted in an average baseline value of 79 ng WHO1998-TEQ/kg “burned” and a range of 9−148 ng WHO1998-TEQ/kg waste “burned” when combusted in a barrel. A refuse emission factor for this work on the lower end of those 7902
dx.doi.org/10.1021/es301197d | Environ. Sci. Technol. 2012, 46, 7898−7904
Environmental Science & Technology
Article
asynchronous sampling. It is difficult to ensure simultaneous sampling of the methods, which becomes important during widely variant concentrations caused by the opening and closing of dampers. This could be a factor on the cyclical units where, in fact, the higher differences were observed (naphthalene, Table 2). This point is illustrated in Figure 4 in which the REMPI-TOFMs measured benzene concentration from the relatively “stable” U.S. downdraft HH is compared against the summa canister sampling time. Even a slight asynchronizaton of the sampling periods could induce significant variation in concentration. For the damper units with greater fluctuations in concentration, this phenomenon would be further exacerbated. Lastly, the Summa canisters “sample” or fill nonlinearly with time, their fill rate dropping as the pressure increases, biasing their concentration toward the beginning of the canister fill, whereas the REMPI-TOFMS value is a 3 min average of sampling/measurements with a frequency of 0.1 s−1. In addition, the REMPI-TOFMS was calibrated prior to sampling against a standard mix of 14 aromatic compounds (including benzene) at 100 ppb each, concentrations about 200 000 times lower than observed from these units. If the instrument responds nonlinearly to the high field concentrations, this would explain the higher REMPITOFMS values. However, past results have shown linearity for over 4 orders of magnitude in concentration,29 lending confidence to the REMPI-TOFMS emission factors. Discrepancies in the naphthalene values between Method 0010 and REMPI-TOFMS may be attributed to uncertainty in the Method 0010 determination, as excessively high levels of dilution were necessary and isotope standards could not be used to assess extraction and cleanup recovery values. VOCs. Table 3 shows all of the Summa canister emission factor results. Differences are observed in some VOC emission factor values from the three units despite use of the same fuel, red oak, particularly noticeable for the U.S. downdraft HH concentrations of propylene and styrene which are considerably lower than the conventional HH and three stage HH units. The benzene values from the conventional HH (22−37 mg/MJ) are similar to those determined on cord wood (32 and 91 mg/MJ) using an older, up-draft boiler.7 Benzene emission factors from 12 woodstoves and fireplaces compiled in an EPA data set27,28,30 averaged 36 mg/MJinput, consistent with Table 3 results. Testing on four different technology HH units showed unitand fuel-specific differences in organic emissions. The highest organic emission levels from the four units tested arose from the conventional single-stage updraft HH, which also had the lowest CE. The European two-stage bottom fed pellet burner had the highest MCE and the lowest PAH and benzene emissions of all three units. The three-stage HH and the U.S. two-stage downdraft HH had the lowest PCDD/PCDF emission levels. The PAH and PCDD/PCDF emission levels depended on fuel type, and the addition of refuse to the wood showed the highest PCDD/PCDF emission levels. The highest emissions of PAHs, VOCs, and PCDDs/PCDFs were observed with the conventional unit, likely due to the rapid changes in combustion conditions effected by the damper opening and closing.
Table 3. VOCs from Summa Canister Sampling Burning Red Oak emission
conventional HH mg/MJinput
three-stage HH mg/ MJinput
U.S. downdraft mg/MJinput
benzene acetylene propylene acrolein 1,3-butadiene methyl ethyl ketone styrene toluene acetonitrile m,p-xylene o-xylene chloromethane ethylbenzene acrylonitrile 1,2,4-trimethylbenzene 1,3,5-trimethylbenzene trichlorotrifluoroethane vinyl chloride
37 33 68 26 14 9.4 8.4 18 2.8 2.6 1.0 2.0 1.9 0.70 0.53 0.17 0.10 0.031
22 35 29 10 6.6 2.1 1.2 5.0 0.57 0.45 0.22 0.12 0.71 0.25 0.056 0.014
