ARTICLE pubs.acs.org/est
PAH Molecular Diagnostic Ratios Applied to Atmospheric Sources: A Critical Evaluation Using Two Decades of Source Inventory and Air Concentration Data from the UK Athanasios Katsoyiannis,* Andrew J. Sweetman, and Kevin C. Jones Lancaster Environment Centre, Lancaster University, Lancaster, LA1 4YQ, U.K.
bS Supporting Information ABSTRACT: Molecular diagnostic ratios (MDRs)—the ratios of defined pairs of individual compounds—have been widely used as markers of different source categories of polycyclic aromatic hydrocarbons (PAHs). However, it is well-known that variations in combustion conditions and environmental degradation processes can cause substantial variability in the emission and degradation of individual compounds, potentially undermining the application of MDRs as reliable source apportionment tools. The United Kingdom produces a national inventory of atmospheric emissions of PAHs, and has an ambient air monitoring program at a range of rural, semirural, urban, and industrial sites. The inventory and the monitoring data are available over the past 20 years (1990 2010), a time frame that has seen known changes in combustion type and source. Here we assess 5 MDRs that have been used in the literature as source markers. We examine the spatial and temporal variability in the ratios and consider whether they are responsive to known differences in source strength and types between sites (on rural urban gradients) and to underlying changes in national emissions since 1990. We conclude that the use of these 5 MDRs produces contradictory results and that they do not respond to known differences (in time and space) in atmospheric emission sources. For example, at a site near a motorway and far from other evident emission sources, the use of MDRs suggests “non-traffic” emissions. The ANT/(ANT + PHE) ratio is strongly seasonal at some sites; it is the most susceptible MDR to atmospheric processes, so these results illustrate how weathering in the environment will undermine the effectiveness of MDRs as markers of source(s). We conclude that PAH MDRs can exhibit spatial and temporal differences, but they are not valid markers of known differences in source categories and type. Atmospheric sources of PAHs in the UK are probably not dominated by any single clear and strong source type, so the mixture of PAHs in air is quickly “blended” away from the influence of the few major point sources which exist and further weathered in the environment by atmospheric reactions and selective loss processes.
’ INTRODUCTION Polycyclic aromatic hydrocarbons (PAHs) are a class of ubiquitous organic chemicals which enter the environment primarily from incomplete combustion or pyrolysis of organic material. Some of these combustion processes are natural (e.g., volcanic activity; forest fires), while others are anthropogenic. Combustion conditions affect the efficiency of burning; “inefficient” low-temperature processes generally generate more PAHs. Major anthropogenic sources include low-temperature burning of coal and wood for domestic space heating, hightemperature burning for power generation, vehicle emissions, cigarette smoking, and the uncontrolled burning of wastes and accidental fires.1 4 The aforementioned sources are believed to emit hundreds of gigagrams of PAHs per year (520 Gg y 1 in 20045). Due to their persistence in the environment and semivolatile character, past PAH emissions that have deposited to soils, vegetation, and water bodies6 can be re-emitted to atmosphere. There is also growing evidence that some low molecular weight PAHs can be formed naturally in soils and volatilize to the atmosphere.7,8 Numerous point and diffuse r 2011 American Chemical Society
sources of PAHs can therefore impact on ambient levels. Many PAHs are mutagenic and some are known carcinogens.9 11 Consequently, there are human health concerns about elevated concentrations of PAHs in the environment. Indeed, air quality standards have been proposed for PAHs in the European Union12 and international actions aim to reduce atmospheric emissions. For all these reasons, there is a need to understand the relative importance of different sources—locally, nationally, regionally, and globally13—so that source reduction measures can be efficient and appropriately targeted. A number of techniques have been developed to improve source apportionment. One of the most common is the use of the molecular diagnostic ratios (MDR). It is argued that some PAHs are emitted in reasonably constant proportions from some of the most important sources of PAHs and that their Received: July 2, 2011 Accepted: August 22, 2011 Revised: August 18, 2011 Published: August 22, 2011 8897
dx.doi.org/10.1021/es202277u | Environ. Sci. Technol. 2011, 45, 8897–8906
Environmental Science & Technology
ARTICLE
Table 1. Characteristic PAH Molecular Diagnostic Ratios14,15 petrogenic
pyrogenic
ANT/(ANT + PHE)
0.1
BaA/(BaA + CHR) FLT/(FLT + PYR)
0.4
IPY/(IPY + BPE)
0.2
a
fuel combustion
grass/coal/wood combustion
FLT/(FLT + PYR)
0.4 0.5
>0.5
IPY/(IPY + BPE)
0.2 0.5
>0.5
BaP/BPE
nontraffic
traffic
0.6
a
ANT: Anthracene; PHE: Phenanthrene; BaA; Benzo[a]anthracene; CHR: Chrysene; FLT: Fluoranthene; PYR: Pyrene; IPY: Indeno[1,2,3-cd]pyrene; BPE: Benzo[g,h,i]perylene; BaP: Benzo[a]pyrene. concentrations (thus the ratios) remain constant between the source and the receptor. An example of a commonly used ratio is “ANT/(ANT + PHE)” (ANT: anthracene; PHE: phenanthrene). This is said to be indicative of unburned fossil fuel when the ratio is 0.1.14 Other ratios have been used (see Table 1), often unquestioningly, to identify pyrogenic (combustion sources) or petrogenic (petroleum input) origin, fuel or wood combustion, or traffic-related sources.14 22 However, other studies have highlighted the wide range in emission factors and compound ratios from given source categories, which raises concerns about the use of MDRs for source identification. Additional complications with source apportionment based on MDRs are that atmospheric sources can be mixed and complex in real world settings, and that individual PAH compounds have different atmospheric residence times and reactivities.23 25 These factors will inevitably result in varied and weathered mixtures and ratios in ambient atmospheres. Here we wanted to use a well-established national emissions inventory and data from a long-term ambient monitoring program to test the idea that MDRs respond to spatial and temporal changes in sources. MDRs were derived from a range of rural, semirural, urban, and industrial sites (Figure 1). Some of these sites have been in operation for around 20 years.26 Five MDRs were used, namely the following: 1. ANT (anthracene)/ ANT + PHE (phenanthere); 2. [BaA (benzo[a]anthracene)/ BaA + CHR (chrysene)]; 3. [FLT (fluoranthene)/FLT + PYR (pyrene)]; 4. [IPY (indeno[1,2,3-cd] pyrene)/IPY + BPE (benzo[g,h,i]perylene)]; and 5. BaP (benzo[a]pyrene)/BPE. It has been proposed that MDRs 1 4 can give information about petrogenic or pyrogenic origin, and that 3 and 4 may highlight whether the combustion material is fuel, grass, coal, or wood.14 The fifth MDR has been proposed as an indicator of traffic.14 Their use as diagnostic tools is addressed here using information from the UK inventory and local site knowledge about potential sources. This study is the first to our knowledge to use large atmospheric time-series to examine MDRs together with a 20-year emission inventory. Other studies in which big data sets were used were from Yunker et al.,14 Brandli et al.,15 and Dvorska et al.16
Figure 1. UK map featuring the sampling sites and the population densities.
’ METHODOLOGY In the present study, the data collected by two long monitoring programs (TOMPS and PAHs) and the emission inventory for PAHs compiled in detail for the last two decades are used. Details and information about these programs are presented below. Air Monitoring Data. The UK Department for Environment, Food and Rural Affairs (DEFRA) has supported a longterm program to monitor Toxic Organic Micro Pollutants (TOMPS), including PAHs at a number of rural and urban locations. Some of these sites have been running since 1990 and provide the longest data sets available in Europe. Details of the TOMPS program have been published previously.26 28 Of relevance here is that data are available quarterly (Q1 4, January March, April June, July September, October December). In addition, a wide range of other sites were established later on the PAH Monitoring Network (1999 and onward). Currently about 30 sites are operated. Details of this program are also available.29 Until 2007 the networks were using an Andersen GPS-1 ambient air sampler, although since then the PAHs network samplers were changed to Digitel DHA-80 high volume PM10 aerosol samplers to comply with the requirements of the European Commission’s Fourth Air Quality daughter directive.12 For direct comparability purposes, in the present study, only samples collected with Andersen GPS-1 ambient air samplers are used and discussed. Table S1 in the Supporting Information presents the PAH compounds that are monitored through the PAH monitoring network. 8898
dx.doi.org/10.1021/es202277u |Environ. Sci. Technol. 2011, 45, 8897–8906
Environmental Science & Technology
ARTICLE
Table 2. Descriptive Statistics of the PAH Concentrations (ng m 3) during the TOMPS and the PAHs Monitoring Networksa HAZ
SF
MID
BEL
BIR
average
155
19.5
72.0
41.4
38.1
max
777
134
340
87.8
62.5
min
6.40
6.10
8.30
5.61
12.3
median
124
15.0
57.3
35.4
34.8
EDI
GLA
MAN
LIV
HM
LON
HOL
14.4
27.3
63.0
30.2
12.5
55.4
27.7
25.8
64.7
237
64.9
50.3
267
47.4
9.10
3.51
4.6
18.9
3.0
4.5
12.4
12.5
23.3
45.7
29.3
9.6
39.8
27.6
a
HAZ: Hazelrigg; SF: Stoke Ferry; MID: Middlesborough; BEL: Belfast; BIR; Birmingham; EDI: Edinburgh; GLA: Glasgow; MAN: Manchester; LIV: Liverpool; HM: High Muffles; LON: London; HOL: Holyhead.
Figure 2. (a) Emission inventory estimates per category of emission sources for the years 1990 2009. (b) ΣPAH concentrations at the 6 TOMPS sites versus the emission inventory estimates.
In the present paper, data were compiled from the commencement of monitoring until 2007 2008 and used to derive the MDRs. A range of sites were selected to give long-term trend data [TOMPS sites; see Meijer et al.26 for analysis] and a rural semirural urban industrial gradient. The following were used: Urban [London (LON), Manchester (MAN), Middlesborough (MID), Liverpool (LIV), Birmingham (BIR), Edinburgh (EDI), Glasgow (GLA), and Belfast (BEL)]; rural/semirural [Holyhead (HOL), High Muffles (HM), Stoke Ferry (SF), and Hazelrigg (HAZ)]. The definitions of rural urban are subjective, based on knowledge of the location and surrounding areas. Figure 1 shows the site locations overlaid on a map showing the UK population
density. Using the average total PAH concentration as a guide, the gradient of sites is: HAZ > MID > MAN > LON > BEL > BIR > LIV > HOL > GLA > SF > EDI > HM (Table 2). The evident anomalous site here is HAZ, close to Lancaster. This is semirural, but has yielded concentrations that are routinely higher than urban/industrial locations. The National Emissions Inventory. The UK’s National Atmospheric Emissions Inventory, 1 also funded by DEFRA, compiles data for a range of air pollutants, including PAHs. PAH estimates are available from 1990. The inventory divides sources into a range of categories as highlighted in Figure 2a. Apparently, certain source types (e.g., metal 8899
dx.doi.org/10.1021/es202277u |Environ. Sci. Technol. 2011, 45, 8897–8906
Environmental Science & Technology
ARTICLE
Figure 3. Descriptive statistics for the five MDRs at all sampling sites: (a) ANT/(ANT + PHE), (b) (BaA/(BaA + CHR), (c) FLT/(FLT + PYR), (d) IPY/(IPY + BPE), (e) B[a]P/BPE].
smelting, iron and steel, and aluminum production) are dominated by few local point sources, but the data are compiled as a national estimate of total emissions. Other source categories, such as traffic or space heating, may be broadly linked to population density. In UK urban areas, space heating may typically rely on gas or electricity, while air concentrations in rural areas may be influenced by domestic burning of coal or wood. Previous studies have highlighted that the latter have much higher PAH emission factors than the former and can impact local air quality.30 32 This signal is strongly seasonal, reflecting more domestic burning for space heating in the winter.26 Attempts have been made in recent years to combine the UK emissions inventory with the measurements programs, often in combination with modeling.26 28 This is to check whether the inventory and models are broadly capturing the real trends apparent in ambient measurements. This can give confidence/ support for the inventory, or identify apparent discrepancies and
inform on the need for further emissions estimates, source identification, or model reparameterization.
’ RESULTS AND DISCUSSION Figure 2b overlays the general trends in UK PAH air concentrations for the TOMPS sites on the inventory and Table S2 presents the correlation coefficients of the total PAH concentrations of the various sites with the emission inventory estimates. The latter correlate strongly (p < 0.0005) with the ΣPAHs in LON and MID, but with all the other sites there is a poor correlation. The comparison between inventory estimates and measurements can highlight interesting differences. For example, in 2004 and 2006, an increase in measured PAH concentrations was seen at various sites (HAZ, MID, and MAN being the most characteristic), but this was not predicted in the inventory. 8900
dx.doi.org/10.1021/es202277u |Environ. Sci. Technol. 2011, 45, 8897–8906
Environmental Science & Technology
ARTICLE
Table 3. Between-Sites Correlation Analysis of the ANT/(ANT + PHE) MDR MAN
SF
LON
LIV
MID
BIR
BEL
EDI
GLA
HM
HH
MAN
1.00
SF
0.23
1.00
LON
0.45
0.21
LIV
0.14
0.31
0.02
1.00
MID
0.10
0.03
0.22
0.17
1.00
BIR
0.23
0.08
-0.40
0.40
0.50
1.00
BEL
0.30
0.19
0.02
0.37
0.32
0.50
1.00
EDI GLA
0.45 0.10
0.40 0.08
0.05 0.05
0.83 0.02
0.44 0.30
0.30 0.24
0.86 0.27
1.00 0.30
HM
0.20
0.01
0.05
0.02
0.32
0.02
0.29
0.10
0.00
1.00
HH
0.12
0.27
0.18
0.36
0.08
0.42
0.33
1.00
0.07
0.00
1.00
HAZ
0.18
0.02
0.25
0.00
0.05
0.36
0.17
0.00
0.19
0.16
0.04
HAZ
1.00
1.00
1.00
a
Values in bold and underlined are statistically significant at the 99.95% level of confidence. Values in bold are statistically significant at the 99% level of confidence. Underlined values are statistically significant at the 95% level of confidence.
MDR Results. The detailed presentation of the ratios for all sites and sampling periods is given as Supporting Information (SI), and the descriptive statistics are given in Figure 3a e. In Figure 3a, the ANT/(ANT + PHE) ratio is quite uniform, consistently 0.1 (Figure S1), showing a very different behavior from the other sites. Although the impression given by Figure 3a is that there is consistency between sites, correlation analysis (Table 3) indicates that there is a strong correlation (p < 0.0005) only between the pairs LON MAN, EDI LIV, and EDI BEL. The rural/semirural sites, HAZ, HOL, HM, and SF do not exhibit any correlations with other sites. The BaA/(BaA + CHR) ratios are usually in the range of 0.2 0.4 (Figure 3b). According to Table 1, this is the “neutral” range that does not show clear dominance of petrogenic or pyrogenic sources. The between-site results are similar, although again MID is an outlier, for which the average/ highest values are affected by the unusually high values of the period 1995 1998. HAZ also exhibits a wide variation in this MDR. For the FLT/(FLT + PYR) MDR, the value is consistently >0.5. The interpretation from Table 1 again suggests pyrogenic origins. Again, MID is unusual, due to its behavior in the period 1995 1998 and 2007 2008. This MDR appears quite consistent between sites, being almost always >0.5; according to Table 1 this indicates vegetation, coal, and wood combustion. The IPY/(IPY + BPE) MDR is said to give the same indications as FLT/(FLT + PYR). As seen in Table 1, values >0.2 suggest pyrogenic origins, when 0.2 < MDR < 0.5, this suggests fuel combustion, and when MDR > 0.5, the pyrogenic origin is attributed to grass/coal/wood combustion. As seen from Figure 3d, the mean and average values of all sites (except MID) fall in the “pyrogenic origin” due to fuel combustion, something that is in disagreement with the aforementioned results of the FLT/(FLT + PYR) ratio. The BaP/BPE MDR (Figure 3e) is the ratio that reportedly shows the effect of traffic. For most sites, the conclusion is mainly “non-traffic” emissions; sites MID, HM, and SF are, according to this ratio, the most influenced by traffic. However, from
local knowledge this is inaccurate. In summary, the MDRs give mixed and contradictory messages about site/source differences. As mentioned, the sites selected are urban, rural, or semirural, but feature also other characteristics, such as being close to industrial zones, close to motorways, ports, etc. The specific emissions that each sampling site is receiving suggest that PAHs would exhibit spatial differences, in concentrations and in profiles. For example the HAZ sampling station is 6 km from the city of Lancaster, within 400 m of a busy motorway and far from residences. Previous studies indicate that emissions close to HAZ are due to vehicles and domestic burning of coal and wood. Major conurbations such as LON or MAN are expected to have contributions from domestic combustion/heating, traffic, and industrial/point source emissions. The background sites of HM and SF are expected to reflect the regional background signal, including biogenic emissions;7 neither is influenced by strong local sources. The MID site is influenced by urban and industrial emissions, while HOL is a semirural site, but also an important port, potentially impacted also by the emissions related to maritime traffic. The aforementioned site-related information about known sources is often not confirmed by the MDRs. For example, the B[a]P/BPE ratio at HAZ has only occasionally exceeded the value of 0.6, suggesting “non-traffic PAHs”. Indeed, its values are generally among the lowest observed (see SI), even though the site has a nearby traffic source. The B[a]P/BPE ratio is also unexpected at MID, because the ratio implies a strong influence of traffic, even though industrial sources are known to be nearby (Figure 3e). This could be attributed to the traffic of heavy vehicles, tracks, and machineries associated with industrial activities. In general, results in Figure 3 do not show major spatial differences, even though known source differences exist. If the ratios are truly reflective of source type (Table 1), the results of the ratios BaA/(BaA + CHR), FLT/(FLT + PYR), and IPY/(IPY + BPE) are all demonstrating pyrolytic or mixed sources and there are only a few cases where one of these suggested a petrogenic source of PAHs. Such examples were the BaA/(BaA + CHR) ratio in Manchester from 1994 (Q1) to 1995 (Q3) and the FLT/(FLT + PYR) in MID from 1996 (Q2) to 1997 (Q3) (SI). The ANT/(ANT + PHE) ratio is in most 8901
dx.doi.org/10.1021/es202277u |Environ. Sci. Technol. 2011, 45, 8897–8906
89 3920
36 1570
4 °C) (h)i
0.010 570 31
0.20 29 6.90
6.35 7
7.47 10
0.01
276 193-39-5
276 191-24-2
252 50-32-8
IPY
BPE
BaP
495
202 129-00-00 PYR
550
375 202 206-44-0 FLT
530
448 CHR 228 218-01-9
0.01
e
0.05 (0.046 )
8
1.37 10
11.77c
46 6.51
5.30 8.8
3
1.33 10
11.56c
4
3.33 10
400 228 56-55-3 BaA
404
1.16 (1.21e)
10.8
>1000
>1000
5.08
5.08
2.5
0.14
0.012
0.11
0.014
11 0.015 0.17
0.013 >1000
690 38
44 5.20
5.61
4
6.67 10 0.66 (0.89 )
8.9
5
8.40 10
340 178 85-01-8 PHE
0.11
340 ANT 178 120-12-7b
0.10
e
6
2.93 10
10.4
5.91
38
650
21.3
2.5
0.11
27 0.011 0.19
0.011 0.17
0.15 1.5
3.4 18.9
6.33 310
>1000 49
48 4.54
4.46 7.6 2
9.07 10 2.59 (4.29e)
7.3 3
2.27 10 1.79 (5.63e)
9.5
190
s 1)h 12
( 10 (s 1)g particles) (s 1)g with OH aire (h) (h)d (h)d KOAb KOWb (pa)a (pa-/m3 mol)a (°c)a registry PAH MW
Values taken from http://www.toronto.ca/health/pdf/cr_appendix_b_pah.pdf.35 b Values taken from: Huckins et al.36 c Values taken from Odabasi et al.37 d Behymer and Hites38 e Kwamena et al.39 f Bunce and Dryfhout.40 g Esteve et al..41 h Brubaker and Hites.42 i Halsall et al.43 a
7 23
6 19
19 °C) (h)i
ARTICLE
10
(gas phase
reactions at 37
OH (gas phase OH (gas phase
on diesel particles) reactions at 298 K) reactions at
OH (heterogeneous (heterogeneous
on graphite reaction black carbon) T1/2 in fly ash) pressure constant point
boiling
Law
vapor
log
log
T1/2 (on T1/2 (on
T1/2 in air f (h) photolysis photolysis Henry’s
Table 4. Physical Chemical Properties of MDR-Related PAHs
CAS
T1/2 in air; T1/2 in air;
reaction with OH
reaction with
reaction with
reaction with
reaction with OH
Environmental Science & Technology
cases
