nitrogen isotope fractionation during gas-to-particle ... · ted noxand thus provides valuable...

$: Nitrogen isotope fractionation during gas-to-particle ... · ted NOxand thus provides valuable information on the par-titioning of the NOxsources (Morin et al., 2008). Such a N isotope$
Atmos Chem Phys 18 11647ndash11661 2018httpsdoiorg105194acp-18-11647-2018copy Author(s) 2018 This work is distributed underthe Creative Commons Attribution 40 License

Nitrogen isotope fractionation during gas-to-particle conversion ofNOx to NOminus

3 in the atmosphere ndash implications for isotope-basedNOx source apportionmentYunhua Chang123 Yanlin Zhang123 Chongguo Tian4 Shichun Zhang5 Xiaoyan Ma6 Fang Cao123Xiaoyan Liu123 Wenqi Zhang123 Thomas Kuhn7 and Moritz F Lehmann7

1YalendashNUIST Center on Atmospheric Environment International Joint Laboratory on Climate and Environment Change(ILCEC) Nanjing University of Information Science amp Technology Nanjing 210044 China2Key Laboratory of Meteorological Disaster Ministry of Education (KLME) Collaborative Innovation Center on Forecastand Evaluation of Meteorological Disasters (CIC-FEMD) Nanjing University of Information Science amp TechnologyNanjing 210044 China3Jiangsu Provincial Key Laboratory of Agricultural Meteorology College of Applied MeteorologyNanjing University of Information Science amp Technology Nanjing 210044 China4Key Laboratory of Coastal Environmental Processes and Ecological Remediation Yantai Institute of Coastal Zone ResearchChinese Academy of Sciences Yantai 264003 China5Northeast Institute of Geography and Agroecology Chinese Academy of Sciences 4888 Shengbei RoadChangchun 130102 China6Key Laboratory for Aerosol-Cloud-Precipitation of China Meteorological Administration Earth System Modeling CenterNanjing University of Information Science and Technology Nanjing 10044 China7Aquatic and Isotope Biogeochemistry Department of Environmental Sciences University of Basel 4056 Basel Switzerland

Correspondence Yanlin Zhang (dryanlinzhangoutlookcom dryanlinzhangnuisteducn)

Received 14 April 2018 ndash Discussion started 8 May 2018Revised 31 July 2018 ndash Accepted 5 August 2018 ndash Published 16 August 2018

Abstract Atmospheric fine-particle (PM25) pollution is fre-quently associated with the formation of particulate nitrate(pNOminus3 ) the end product of the oxidation of NOx gases(NO+NO2) in the upper troposphere The application of sta-ble nitrogen (N) (and oxygen) isotope analyses of pNOminus3 toconstrain NOx source partitioning in the atmosphere requiresknowledge of the isotope fractionation during the reactionsleading to nitrate formation Here we determined the δ15Nvalues of fresh pNOminus3 (δ15NndashpNOminus3 ) in PM25 at a ruralsite in northern China where atmospheric pNOminus3 can be at-tributed exclusively to biomass burning The observed δ15NndashpNOminus3 (1217plusmn 155 permil n= 8) was much higher than theN isotopic source signature of NOx from biomass burning(104plusmn 413 permil) The large difference between δ15NndashpNOminus3and δ15NndashNOx (1(δ15N)) can be reconciled by the net N iso-tope effect (εN) associated with the gasndashparticle conversionfrom NOx to NOminus3 For the biomass burning site a meanεN(asymp1(δ

15N)) of 1099plusmn 074 permil was assessed through a

newly developed computational quantum chemistry (CQC)module εN depends on the relative importance of the twodominant N isotope exchange reactions involved (NO2 re-action with OH versus hydrolysis of dinitrogen pentoxide(N2O5) with H2O) and varies between regions and on adiurnal basis A second slightly higher CQC-based meanvalue for εN (1533plusmn 490 permil) was estimated for an urbansite with intense traffic in eastern China and integrated ina Bayesian isotope mixing model to make isotope-basedsource apportionment estimates for NOx at this site Basedon the δ15N values (1093plusmn 332 permil n= 43) of ambientpNOminus3 determined for the urban site and considering thelocation-specific estimate for εN our results reveal that therelative contribution of coal combustion and road traffic tourban NOx is 32 plusmn 11 and 68 plusmn 11 respectivelyThis finding agrees well with a regional bottom-up emis-sion inventory of NOx Moreover the variation pattern of OHcontribution to ambient pNOminus3 formation calculated by the

Published by Copernicus Publications on behalf of the European Geosciences Union

11648 Y Chang et al Nitrogen isotope fractionation during NOx to pNOminus3 formation

CQC module is consistent with that simulated by the WeatherResearch and Forecasting model coupled with Chemistry(WRF-Chem) further confirming the robustness of our esti-mates Our investigations also show that without the consid-eration of the N isotope effect during pNOminus3 formation theobserved δ15NndashpNOminus3 at the study site would erroneouslyimply that NOx is derived almost entirely from coal com-bustion Similarly reanalysis of reported δ15NndashNOminus3 datathroughout China and its neighboring areas suggests thatNOx emissions from coal combustion may be substantivelyoverestimated (by gt 30 ) when the N isotope fractionationduring atmospheric pNOminus3 formation is neglected

1 Introduction

Nitrogen oxides (NOx = NO + NO2) are among the mostimportant molecules in tropospheric chemistry They are in-volved in the formation of secondary aerosols and atmo-spheric oxidants such as ozone (O3) and hydroxyl radicals(OH) which control the self-cleansing capacity of the at-mosphere (Galloway et al 2003 Seinfeld and Pandis 2012Solomon et al 2007) The sources of NOx include both an-thropogenic and natural origins with more than half of theglobal burden (sim 40 Tg N yrminus1) currently attributed to fos-sil fuel burning (224ndash261 Tg N yrminus1) and the rest primar-ily derived from nitrificationdenitrification in soils (includ-ing wetlands 89plusmn 19 Tg N yrminus1) biomass burning (58plusmn18 Tg N yrminus1) lightning (2ndash6 Tg N yrminus1) and oxidation ofN2O in the stratosphere (01ndash06 Tg N yrminus1) (Jaegle et al2005 Richter et al 2005 Lamsal et al 2011 Price et al1997 Yienger and Levy 1995 Miyazaki et al 2017 Dun-can et al 2016 Anenberg et al 2017 Levy et al 1996)The mainultimate sinks for NOx in the troposphere arethe oxidation to nitric acid (HNO3(g)) and the formation ofaerosol-phase particulate nitrate (pNOminus3 ) (Seinfeld and Pan-dis 2012) the partitioning of which may vary on diurnal andseasonal timescales (Morino et al 2006)

Emissions of NOx occur mostly in the form of NO (Se-infeld and Pandis 2012 Leighton 1961) During daytimetransformation from NO to NO2 is rapid (few minutes) andproceeds in a photochemical steady state controlled by theoxidation of NO by O3 to NO2 and the photolysis of NO2back to NO (Leighton 1961)

NO +O3 minusrarr NO2+O2 (R1)NO2+hv minusrarr NO+O (R2)

O+O2Mminusrarr O3 (R3)

where M is any non-reactive species that can take up the en-ergy released to stabilize O NOx oxidation to HNO3 is gov-erned by the following equations During daytime

NO2+OHMminusrarr HNO3 (R4)

During nighttime

NO2+O3 minusrarr NO3+O2 (R5)

NO3+NO2Mminusrarr N2O5 (R6)

N2O5+H2O(surface)aerosolminusrarr 2HNO3 (R7)

HNO3 then reacts with gas-phase NH3 to form ammoniumnitrate (NH4NO3) aerosols If the ambient relative humid-ity (RH) is lower than the efflorescence relative humid-ity (ERH) or crystallization relative humidity (CRH) solid-phase NH4NO3(s) is formed (Smith et al 2012 Ling andChan 2007)

NH4NO3 HNO3 (g)+NH3 (g) (R8a)

If ambient RH exceeds the ERH or CRH HNO3 and NH3dissolve into the aqueous phase (aq) (Smith et al 2012 Lingand Chan 2007)

HNO3 (g)+NH3 (g) NOminus3 (aq)+NH+4 (aq) (R8b)

While global NOx emissions are well constrained individualsource attribution and their local or regional role in particu-late nitrate formation are difficult to assess due to the shortlifetime of NOx (typically less than 24 h) and the high degreeof spatiotemporal heterogeneity with regards to the ratio be-tween gas-phase HNO3 and particulate NOminus3 (pNOminus3 ) (Dun-can et al 2016 Lu et al 2015 Zong et al 2017 Zhang etal 2003) Given the conservation of the nitrogen (N) atombetween NOx sources and sinks the N isotopic compositionof pNOminus3 can be related to the different origins of the emit-ted NOx and thus provides valuable information on the par-titioning of the NOx sources (Morin et al 2008) Such a Nisotope balance approach works best if the N isotopic com-position of various NOx sources display distinct 15N14N ra-

tios (reported as δ15N =(15N14N

)sampleminus(

15N14N)N2

(15N14N)N2times 1000)

The δ15NndashNOx of coal-fired power plant (+10 permil to+25 permil)(Felix et al 2012 2013 Heaton 1990) vehicle (+37 permilto +57 permil) (Heaton 1990 Walters et al 2015 Felix andElliott 2014 Felix et al 2013 Wojtal et al 2016) andbiomass burning (minus7 permil to +12 permil) emissions (Fibiger andHastings 2016) for example is generally higher than thatof lightning (minus05 permil to +14 permil) (Hoering 1957) and bio-genic soil (minus489 permil to minus199 permil) emissions (Li and Wang2008 Felix and Elliott 2014 Felix et al 2013) allowingthe use of isotope mixing models to gain insight on the NOxsource apportionment for gases aerosols and the resultingnitrate deposition (minus15 permil to +15 permil) (Elliott et al 20072009 Zong et al 2017 Savarino et al 2007 Morin etal 2008 Park et al 2018 Altieri et al 2013 Gobel etal 2013) In addition because of mass-independent frac-tionation during its formation (Thiemens 1999 Thiemensand Heidenreich 1983) ozone possesses a strong isotopeanomaly (117Oasymp δ17Ominus 052timesδ18O) which is propagated

Atmos Chem Phys 18 11647ndash11661 2018 wwwatmos-chem-physnet18116472018

Y Chang et al Nitrogen isotope fractionation during NOx to pNOminus3 formation 11649

into the most short-lived oxygen-bearing species includingNOx and nitrate Therefore the oxygen isotopic compositionof nitrate (δ18O 117O) can provide information on the oxi-dants involved in the conversion of NOx to nitrate (Michalskiet al 2003 Geng et al 2017) Knopf et al (2006 2011)and Shiraiwa et al (2012) have shown that NO3 can betaken up efficiently by organic (eg levoglucosan) aerosoland may dominate oxidation of aerosol in the polluted urbannighttime (Kaiser et al 2011) Globally theoretical model-ing results show that nearly 76 18 and 4 of annualinorganic nitrate are formed via pathwaysreactions involv-ing OH N2O5 and dimethyl sulfide or hydrocarbons re-spectively (eg Alexander et al 2009) The stable O iso-topic composition of atmospheric nitrate is a powerful proxyfor assessing which oxidation pathways are important forconverting NOx into nitrate under changing environmentalconditions (eg polluted volcanic events climate change)Along the same lines in this study the average δ18O valueof pNOminus3 in Nanjing was 830plusmn112 permil (see Discussion sec-tion) suggesting that pNOminus3 formation is dominated by thepathways of ldquoOH + NO2rdquo and the heterogeneous hydrolysisof N2O5δ15N-based source apportionment of NOx requires knowl-

edge of how kinetic and equilibrium isotope fractionationmay impact δ15N values during the conversion of NOx tonitrate (Freyer 1978 Walters et al 2016) If these isotopeeffects are considerable they may greatly limit the use ofδ15N values of pNOminus3 for NOx source partition (Walters etal 2016) Previous studies did not take into account the po-tentially biasing effect of N isotope fractionation becausethey assumed that changes in the δ15N values during theconversion of NOx to nitrate are minor (without detailed ex-planation) (Kendall et al 2007 Morin et al 2008 Elliottet al 2007) or relatively small (eg +3 permil) (Felix and El-liott 2014 Freyer 2017) However a field study by Freyeret al (1993) has indicated that N isotope exchange mayhave a strong influence on the observed δ15N values in at-mospheric NO and NO2 implying that isotope equilibriumfractionation may play a significant role in shaping the δ15Nof NOy species (the family of oxidized nitrogen moleculesin the atmosphere including NOx NO3 NOminus3 peroxyacetylnitrate etc) The transformation of NOx to nitrate is a com-plex process that involves several different reaction pathways(Walters et al 2016) To date few fractionation factors forthis conversion have been determined Recently Walters andMichalski (2015) and Walters et al (2016) used computa-tional quantum chemistry methods to calculate N isotopeequilibrium fractionation factors for the exchange betweenmajor NOy molecules and confirmed theoretical predictionsthat 15N isotopes get enriched in the more oxidized form ofNOy and that the transformation of NOx to atmospheric ni-trate (HNO3 NO3(aq) NO3(g)) continuously increases theδ15N in the residual NOx pool

As a consequence of its severe atmospheric particle pollu-tion during the cold season China has made great efforts to-

ward reducing NOx emissions from on-road traffic (eg im-proving emission standards higher gasoline quality vehicletravel restrictions) (Li et al 2017) Moreover China has con-tinuously implemented denitrogenation technologies (egselective catalytic reduction) in the coal-fired power plantssector since the mid-2000s and has been phasing out smallinefficient units (Liu et al 2015) Monitoring and assess-ing the efficiency of such mitigation measures and optimiz-ing policy efforts to further reduce NOx emissions requireknowledge of the vehicle- and power-plant-emitted NOx toparticulate nitrate in urban China (Ji et al 2015 Fu et al2013 Zong et al 2017) In this study the chemical compo-nents of ambient fine particles (PM25) were quantified andthe isotopic composition of particulate nitrate (δ15NndashNOminus3 δ18OndashNOminus3 ) was assessed in order to elucidate ambient NOxsources in the city of Nanjing in eastern China We also in-vestigated the potential isotope effect during the formationof nitrate aerosols from NOx and evaluated how disregardof such N isotope fractionation can bias N-isotope-mixing-model-based estimates on the NOx source apportionment fornitrate deposition

2 Methods

21 Field sampling

In this study PM25 aerosol samples were collected on pre-combusted (450 C for 6 h) quartz filters (25times 20 cm) on adayndashnight basis using high-volume air samplers at a flowrate of 105 m3 minminus1 in Sanjiang and Nanjing (Fig 1) Aftersampling the filters were wrapped in aluminum foil packedin air-tight polyethylene bags and stored at minus20 C prior tofurther processing and analysis Four blank filters were alsocollected They were exposed for 10 min to ambient air (iewithout active sampling) PM25 mass concentration was an-alyzed gravimetrically (Sartorius MC5 electronic microbal-ance) with a plusmn1 microg precision before and after sampling (at25 C and 45plusmn 5 during weighing)

The Sanjiang campaign was performed during a periodof intensive burning of agricultural residues between 8 and18 October 2013 to examine if there is any significant dif-ference between the δ15N values of pNOminus3 and NOx emit-ted from biomass burning The Sanjiang site (in the fol-lowing abbreviated as SJ 4735 N 13331 E) is located atan ecological experimental station affiliated with the Chi-nese Academy of Sciences located in the Sanjiang Plaina major agricultural area predominantly run by state farmsin northeastern China (Fig 1) Surrounded by vast farmfields and bordering far-eastern Russia SJ is situated in aremote and sparsely populated region with a harsh climateand rather poorly industrialized economy The annual meantemperature at SJ is close to the freezing point with dailyminima ranging between minus31 and minus15 C in the coldestmonth January As a consequence of the relatively low tem-

wwwatmos-chem-physnet18116472018 Atmos Chem Phys 18 11647ndash11661 2018


Figure 1 Location of the sampling sites Sanjiang and Nanjing Theblack dots indicate the location of sampling sites (sites are locatedin the area of mainland China and the Yellow East China and SouthChina seas) with δ15NndashNOminus3 data from the literature (see also Ta-ble S4)

peratures (also during summer) biogenic production of NOxthrough soil microbial processes is rather weak SJ is there-fore an excellent environment in which to collect biomass-burning-emitted aerosols with only minor influence fromother sources

The Nanjing campaign was conducted between 17 De-cember 2014 and 8 January 2015 with the main objective toexamine whether N isotope measurements can be used as atool to elucidate NOx source contributions to ambient pNOminus3during times of severe haze Situated in the lower YangtzeRiver region Nanjing is after Shanghai the second-largestcity in eastern China The aerosol sampler was placed at therooftop of a building on the Nanjing University of Informa-tion Science and Technology campus (in the following abbre-viated as NJ 18 m agl 3221 N 11872 E Fig 1) whereNOx emissions derive from both industrial and transportationsources

211 Laboratory analysis

The mass concentrations of inorganic ions (including SO2minus4

NOminus3 Clminus NH+4 K+ Ca2+ Mg2+ and Na+) carbonaceouscomponents (organic carbon or OC elemental carbon orEC) and water-soluble organic carbon were determined usingan ion chromatograph (761 Compact IC Metrohm Switzer-land) a thermal-optical OCndashEC analyzer (RT-4 model Sun-

set Laboratory Inc USA) and a total organic carbon analyzer(Shimadzu TOC-VCSH Japan) respectively Importantlylevoglucosan a molecular marker for the biomass combus-tion aerosols was detected using a Dionextrade ICS-5000+ sys-tem (Thermo Fisher Scientific Sunnyvale USA) Chemicalaerosol analyses including sample pre-treatment analyticalprocedures protocol adaption detection limits and experi-mental uncertainty were described in detail in our previouswork (Cao et al 2016 2017)

For isotopic analyses of aerosol nitrate aerosol subsam-ples were generated by punching 14 cm disks out of the fil-ters In order to extract the NOminus3 sample discs were placedin acid-washed glass vials with 10 ml deionized water andplaced in an ultra-sonic water bath for 30 min Between oneand four disks were used for NOx extraction dependent onthe aerosol NOminus3 content of the filters which was determinedindependently The extracts were then filtered (022 microm) andanalyzed the next day N and O isotope analyses of the ex-tracteddissolved aerosol nitrate (15N14N 18O16O) wereperformed using the denitrifier method (Sigman et al 2001Casciotti et al 2002) Briefly sample NOminus3 is converted tonitrous oxide (N2O) by denitrifying bacteria that lack N2Oreductase activity (Pseudomonas chlororaphis ATCC 13985formerly Pseudomonas aureofaciens referred to below assuch) N2O is extracted purified and analyzed for its N andO isotopic composition using a continuous-flow isotope ra-tio mass spectrometer (Thermo Finnigan Delta+ BremenGerman) Nitrate N and O isotope ratios are reported in theconventional δ notation with respect to atmospheric N2 andVienna standard mean ocean water respectively Analysesare calibrated using the international nitrate isotope standardIAEA-N3 with a δ15N value of 47 permil and a δ18O value of256 permil (Boumlhlke et al 2003) The blank contribution wasgenerally lower than 02 nmol (as compared to 20 nmol ofsample N) Based on replicate measurements of standardsand samples the analytical precision for δ15N and δ18O wasgenerally better thanplusmn02 permil andplusmn03 permil (1σ ) respectively

The denitrifier method generates δ15N and δ18O values ofthe combined pool of NOminus3 and NOminus2 The presence of sub-stantial amounts of NOminus2 in NOminus3 samples may lead to errorswith regards to the analysis of δ18O (Wankel et al 2010) Werefrained from including a nitrite-removal step because ni-trite concentrations in our samples were always lt 1 of theNOminus3 concentrations In the following δ15NNOx and δ18ONOxare thus referred to as nitrate δ15N and δ18O (or δ15NNO3 andδ18ONO3 )

In the case of atmospheric or aerosol nitrate samples withcomparatively high δ18O values δ15N values tend to be over-estimated by 1ndash2 permil (Hastings et al 2003) if the contribu-tion of 14N14N17O to the N2O mass 45 signal is not ac-counted for during isotope ratio analysis For most naturalsamples the mass-dependent relationship can be approxi-mated as δ17Oasymp 052times δ18O and the δ18O can be usedfor the 17O correction Atmospheric NOminus3 does not followthis relationship but inhabits a mass-independent component



Thus we adopted a correction factor of 08 instead of 052 forthe 17O-to-18O linearity (Hastings et al 2003)

22 Calculation of N isotope fractionation value (εN)

As we described above the transformation process of NOx toHNO3 or NOminus3 involves multiple reaction pathways (see alsoFig S1 in the Supplement) and is likely to undergo isotopeequilibrium exchange reactions The measured δ15NndashNOminus3values of aerosol samples are thus reflective of the combinedN isotope signatures of various NOx sources (δ15NndashNOx)plus any given N isotope fractionation Recently Walter andMichalski (2015) used a computational quantum chemistryapproach to calculate isotope exchange fractionation fac-tors for atmospherically relevant NOy molecules based onthis approach Zong et al (2017) estimated the N isotopefractionation during the transformation of NOx to pNOminus3at a regional background site in China Here we adoptedand slightly modify the approach by Walter and Michal-ski (2015) and Zong et al (2017) and assumed that thenet N isotope effect εN (for equilibrium processes Aharr BεAharrB =

((heavy isotopelight isotope)A(heavy isotopelight isotope)Bminus1

)middot 1000permil εN refers to

εN(NOxharrpNOminus3 )

in this study unless otherwise specified) during

the gas-to-particle conversion from NOx to pNOminus3 formation(1(δ15N

)pNOminus3 minusNOx

= δ15NndashpNOminus3 minusδ15NndashNOx asymp εN) can

be considered a hybrid of the isotope effects of two dominantN isotopic exchange reactions

εN = γ times εN(NOxharrpNOminus3 )OH

+ (1minus γ )times εN(NOxharrpNOminus3 )H2O

= γ times εN(NOxharrHNO3)OH+ (1minus γ )times εN(NOxharrHNO3)H2O

(1)

where γ represents the contribution from isotope frac-tionation by the reaction of NOx and photochemicallyproduced OH to form HNO3 (and pNOminus3 ) as shownby εN(NOxharrHNO3)OH

(εN(NOxharrpNOminus3 )OH

) The remainder is

formed by the hydrolysis of N2O5 with aerosol waterto generate HNO3 (and pNOminus3 ) namely εN(NOxharrHNO3)H2O

(εN(NOxharrpNOminus3 )H2O

) Assuming that kinetic N isotope frac-

tionation associated with the reaction between NOx and OHis negligible εN

(NOxharrpNOminus3 )OHcan be calculated based on

mass-balance considerations

εN(NOxharrpNOminus3 )OH

= εN(NOxharrHNO3)OH= εN(NO2harrHNO3)OH

= 1000times

[ (15αNO2NOminus1)(

1minus fNO2

)(1minus fNO2

)+(

15αNO2NOtimes fNO2

)] (2)

where 15αNO2NO is the temperature-dependent (see Eq 7and Table S1 in the Supplement) equilibrium N isotope frac-tionation factor between NO2 and NO and fNO2 is the frac-tion of NO2 in the total NOx fNO2 ranges from 02 to 095(Walters and Michalski 2015) Similarly assuming a negli-gible kinetic isotope fractionation associated with the reac-

tion N2O5 + H2O + aerosol rarr 2HNO3 εN(NOxharrpNOminus3 )H2O

can be computed from the following equation

εN(NOxharrpNOminus3 )H2O

= εN(NOxharrHNO3)H2O=

εN(NOxharrN2O5)H2O= 1000times

(15αN2O5NO2minus1) (3)

where 15αN2O5NO2 is the equilibrium isotope fractionationfactor between N2O5 and NO2 which also is temperature-dependent (see Eq 7 and Table S1)

Following Walter and Michalski (2015) and Zhong etal (2017) γ can then be approximated based on the O iso-tope fractionation during the conversion of NOx to pNOminus3

εO(NOxharrpNOminus3 )

= γ times εO(NOxharrpNOminus3 )OH

+(1minus γ )times εO(NOxharrpNOminus3 )H2O

= γ times εO(NOxharrHNO3)OH+ (1minus γ )times εO(NOxharrHNO3)H2O

(4)

where εO(NOxharrpNOminus3 )OH

and εO(NOxharrpNOminus3 )H2O

represent the O

isotope effects associated with pNOminus3 generation through thereaction of NOx and OH to form HNO3 and the hydrolysisof N2O5 on a wetted surface to form HNO3 respectivelyεO(NOxharrpNOminus3 )OH

can be further expressed as

εO(NOxharrpNOminus3 )OH

= εO(NOxharrHNO3)OH=

23εO(NO2harrHNO3)OH

+13εO(NOharrHNO3)OH

=23

[1000

(18αNO2NOminus1)(

1minus fNO2

)(1minus fNO2

)+(

18αNO2NOtimes fNO2

) + (δ18OminusNOx)]+

13

[(δ18OminusH2O

)+ 1000

(18αOHH2Ominus 1

)]

(5)


can be determined as follows

εO(NOxharrpNOminus3 )H2O

= εO(NOxharrHNO3)H2O=

56

(δ18OminusN2O5

)+

16

(δ18OminusH2O

) (6)

where 18αNO2NO and 18αOHH2O represent the equilibrium Oisotope fractionation factors between NO2 and NO betweenand OH and H2O respectively The range of δ18OndashH2O canbe approximated using an estimated tropospheric water vaporδ18O range of minus25 permil to 0 permil The δ18O values for NO2 andN2O5 range from 90 permil to 122 permil (Zong et al 2017)

15αNO2NO 15αN2O5NO2 18αNO2NO and 18αOHH2O inthese equations are dependent on the temperature which canbe expressed as



1000(mαXY minus 1

)=A

T 4 times 1010+B

T 3 times 108+C

T 2 times 106

+D

Ttimes 104 (7)

where A B C and D are experimental constants (Table S1in the Supplement) over the temperature range of 150ndash450 K(Walters and Michalski 2015 Walters et al 2016 Waltersand Michalski 2016 Zong et al 2017)

Based on Eqs (4)ndash(7) and measured values for δ18OndashpNOminus3 of ambient PM25 a Monte Carlo simulation was per-formed to generate 10 000 feasible solutions The error be-tween predicted and measured δ18O was less than 05 permil Therange (maximum and minimum) of computed contributionratios (γ ) was then integrated in Eq (1) to generate an esti-mate range for the nitrogen isotope effect εN (using Eqs 2ndash3) δ15NndashpNOminus3 values can be calculated based on εN and theestimated δ15N range for atmospheric NOx (see Sect 24)

23 Bayesian isotope mixing model

Isotopic mixing models allow estimating the relative contri-bution of multiple sources (eg emission sources of NOx)within a mixed pool (eg ambient pNOminus3 ) By explicitlyconsidering the uncertainty associated with the isotopic sig-natures of any given source as well as isotope fractionationduring the formation of various components of a mixturethe application of Bayesian methods to stable isotope mixingmodels generates robust probability estimates of source pro-portions and is often more appropriate when targeting natu-ral systems than simple linear mixing models (Chang et al2016a) Here the Bayesian model MixSIR (a stable isotopemixing model using sampling importance and resampling)was used to disentangle multiple NOx sources by generatingpotential solutions of source apportionment as true probabil-ity distributions which has been widely applied in a num-ber of fields (eg Parnell et al 2013 Phillips et al 2014Zong et al 2017) Details on the model frame and comput-ing methods are given in Sect S1 in the Supplement

Here coal combustion (1372plusmn 457 permil) transportation(minus371plusmn 1040 permil) biomass burning (104plusmn 413 permil) andbiogenic emissions from soils (minus3377plusmn1216 permil) were con-sidered to be the most relevant contributors of NOx (Table S2and Sect S2) The δ15N of atmospheric NOx is unknownHowever it can be assumed that its range in the atmosphere isconstrained by the δ15N of the NOx sources and the δ15N ofpNOminus3 after equilibrium fractionation conditions have beenreached Following Zong et al (2017) δ15NndashNOx in theatmosphere was determined by performing iterative modelsimulations with a simulation step of 001 times the equilib-rium fractionation value based on the δ15NndashNOx values ofthe emission sources (mean and standard deviation) and themeasured δ15NndashpNOminus3 of ambient PM25 (Fig S2)

3 Results

31 Sanjiang in northern China

The δ15NndashpNOminus3 and δ18OndashpNOminus3 values of the eight sam-ples collected from the Sanjiang biomass burning field ex-periment ranged from 954 permil to 1377 permil (mean 1217 permil)and 5717 permil to 7509 permil (mean 6357 permil) respectively Inthis study atmospheric concentrations of levoglucosan quan-tified from PM25 samples collected near the sites of biomassburning in Sanjiang vary between 40 and 205 microg mminus3 2to 5 orders of magnitude higher than those measured dur-ing non-biomass-burning season (Cao et al 2017 2016)Levoglucosan is an anhydrosugar formed during pyrolysisof cellulose at temperatures above 300 C (Simoneit 2002)Due to its specificity for cellulose combustion it has beenwidely used as a molecular tracer for biomass burning (Si-moneit et al 1999 D Liu et al 2013 Jedynska et al2014 Liu et al 2014) Indeed the concentrations of lev-oglucosan and aerosol nitrate in Sanjiang were highly cor-related (R2

= 064 Fig 2a) providing compelling evidencethat particulate nitrate measured during our study period waspredominately derived from biomass burning emissions

32 Nanjing in eastern China

The mass concentrations(meanmax

min plusmn 1σn= 43)

of PM25and pNOminus3 measured in Nanjing were 12212278

390 plusmn 479 and178452

40 plusmn 103 microg mminus3 respectively All PM25 concentra-tions exceeded the Chinese Air Quality Standards for dailyPM25 (35 microg mminus3) suggesting severe haze pollution duringthe sampling period The corresponding δ15NndashpNOminus3 val-ues (raw data without correction) ranged between 539 permiland 1799 permil indicating significant enrichment in 15N rel-ative to rural and coastal marine atmospheric NOminus3 sources(Table S4) This may be due to the prominent contribution offossil-fuel-related NOx emissions with higher δ15N values inurban areas (Elliott et al 2007 Park et al 2018)

4 Discussion

41 Sanjiang campaign theoretical calculation andfield validation of N isotope fractionation duringpNOminus

3 formation

To be used as a quantitative tracer of biomass-combustion-generated aerosols levoglucosan must be conserved duringits transport from its source without partial removal by re-actions in the atmosphere (Hennigan et al 2010) The massconcentrations of non-sea-salt potassium (nss-K+ = K+ minus00355timesNa+) is considered as an independentadditionalindicator of biomass burning (Fig 2b) The association ofelevated levels of levoglucosan with high nss-K+ concen-trations underscores that the two compounds derived fromthe same proximate sources and thus that aerosol levoglu-



Figure 2 (a) Correlation analysis between the mass concentrations of levoglucosan and aerosol nitrate during the Sanjiang sampling cam-paign (b) variation of fractions of various inorganic species (MSAminus stands for methyl sulphonate) during dayndashnight samplings at Sanjiangbetween 8 and 18 October 2013 (sample ID 1 to 8) The higher relative abundances of nss-K+ and Clminus are indicative of a biomass-burning-dominated source For sample ID information and exact sampling dates refer to Table S3

cosan in Sanjiang was indeed pristine and represented a re-liable source indicator that is unbiased by altering processesin the atmosphere Moreover in our previous work (Cao etal 2017) we observed that there was a much greater en-hancement of atmospheric NOminus3 compared to SO2minus

4 (a typicalcoal-related pollutant) This additionally points to biomassburning and not coal-combustion as the dominant pNOminus3source in the study area making SJ an ideal ldquoquasi-single-sourcerdquo environment for calibrating the N isotope effect dur-ing pNOminus3 formation

Our δ18OndashpNOminus3 values are well within the broad rangeof values in previous reports (Zong et al 2017 Geng et al2017 Walters and Michalski 2016) However as depicted inFig 3 the δ15N values of biomass-burning-emitted NOminus3 fallwithin the range of δ15NndashNOx values typically reported foremissions from coal combustion whereas they are signifi-cantly higher than the well-established values for δ15NndashNOxemitted from the burning of various types of biomass (mean104plusmn 413 permil range minus7 to +12 permil) (Fibiger and Hastings2016) Turekian et al (1998) conducted laboratory tests in-volving the burning of eucalyptus and African grasses anddetermined that the δ15N of pNOminus3 (around 23 permil) was 66 permilhigher than the δ15N of the burned biomass This implies sig-nificant N isotope partitioning during biomass burning In thecase of complete biomass combustion by mass balance thefirst gaseous products (ie NOx) have the same δ15N as thebiomass Hence any discrepancy between the pNOminus3 and theδ15N of the biomass can be attributed to the N isotope frac-tionation associated with the partial conversion of gaseousNOx to aerosol NOminus3 Based on the computational quantumchemistry (CQC) module calculations the N isotope frac-tionation εN

(meanmax

min plusmn 1σ)

determined from the Sanjiangdata was 10991254

1030plusmn 074 permil After correcting the primaryδ15NndashpNOminus3 values under the consideration of εN the re-sulting mean δ15N of 117298

minus189plusmn195 permil is very close to theN isotopic signature expected for biomass-burning-emittedNOx (104plusmn 413 permil) (Fig 3) (Fibiger and Hastings 2016)

Figure 3 Original δ15N values (δ15Nini) for pNOminus3 calculatedvalues for the N isotope fractionation (εN) associated with theconversion of gaseous NOx to pNOminus3 and corrected δ15N values(δ15Ncorr 15Nini minus εN) of pNOminus3 for each sample collectedduring the Sanjiang sampling campaign The colored bands repre-sent the variation range of δ15N values for different NOx sourcesbased on reports from the literature (Table S2) See Table S3 for theinformation regarding sample ID

The much higher δ15NndashpNOminus3 values in our study comparedto reported δ15NndashNOx values for biomass burning can easilybe reconciled when including N isotope fractionation dur-ing the conversion of NOx to NOminus3 Put another way giventhat Sanjiang is an environment where we can essentially ex-clude NOx sources other than biomass burning at the timeof sampling the data nicely validate our CQC-module-basedapproach to estimating εN



42 Source apportionment of NOx in an urban settingusing a Bayesian isotopic mixing model

Due to its high population density and intensive industrialproduction the Nanjing atmosphere was expected to havehigh NOx concentrations derived from road traffic and coalcombustion (Zhao et al 2015) However the raw δ15NndashpNOminus3 values (1093plusmn 332 permil) fell well within the varia-tion range of coal-emitted δ15NndashNOx (Fig 3) It is tempt-ing to conclude that coal combustion is the main or evensole pNOminus3 source (given the equivalent δ15N values) yetthis is very unlikely The data rather confirm that significantisotope fractionation occurred during the conversion of NOxto NOminus3 and that without consideration of the N isotope ef-fect traffic-related NOx emissions will be markedly under-estimated

In the atmosphere the oxygen atoms of NOx rapidly ex-changed with O3 in the ldquoNOndashNO2rdquo cycle (see Reactions R1ndashR3) (Hastings et al 2003) and the δ18OndashpNOminus3 values aredetermined by its production pathways (Reactions R4ndashR7)rather than the sources of NOx (Hastings et al 2003) Thusδ18OndashpNOminus3 can be used to gain information on the pathwayof conversion of NOx to nitrate in the atmosphere (Fang etal 2011) In the computational quantum chemistry moduleused here to calculate isotope fractionation we assumed thattwo-thirds of the oxygen atoms in NOminus3 derive from O3 andone-third from qOH in the qOH generation pathway (Reac-tion R4) (Hastings et al 2003) correspondingly five-sixthsof the oxygen atoms then derived from O3 and one-sixth fromqOH in the ldquoO3ndashH2Ordquo pathway (Reactions R5ndashR7) The as-sumed range for δ18OndashO3 and δ18OndashH2O values were 90 permilndash122 permil and minus25 permilndash0 permil respectively (Zong et al 2017)The partitioning between the two possible pathways was thenassessed through Monte Carlo simulation (Zong et al 2017)The estimated range was rather broad given the wide rangeof δ18OndashO3 and δ18OndashH2O values used Nevertheless thetheoretical calculation of the average contribution ratio (γ )for nitrate formation in Nanjing via the reaction of NO2and qOH is consistent with the results from simulations us-ing the Weather Research and Forecasting model coupledwith Chemistry (WRF-Chem) (Fig 4 see Sect S3 for de-tails) A clear diurnal cycle of the mass concentration of ni-trate formed through qOH oxidation of NO2 can be observed(Fig S3) with much higher concentrations between 1200and 1800 This indicates the importance of photochemicallyproduced qOH during daytime Yet throughout our sam-pling period in Nanjing the average pNOminus3 formation bythe heterogeneous hydrolysis of N2O5 (126 microg mmminus3) ex-ceeded pNOminus3 formation by the reaction of NO2 and qOH(48 microg mmminus3) even during daytime consistent with recentobservations during peak pollution periods in Beijing (Wanget al 2017) Given the production rates of N2O5 in the at-mosphere are governed by ambient O3 concentrations re-ducing atmospheric O3 levels appears to be one of the most

Figure 4 Comparison between the theoretical calculation andWRF-Chem simulation of the average contribution ratio (γ ) for ni-trate formation in Nanjing via the reaction of NO2 and photochem-ically produced qOH

important measures to take for mitigating pNOminus3 pollution inChinarsquos urban atmospheres

In Nanjing dependent on the time-dependent dominantpNOminus3 formation pathway the average N isotope fraction-ation value calculated using the computational quantumchemistry module varied between 1077 permil and 1934 permil(1533 permil on average) Using the Bayesian model MixSIRthe contribution of each source can be estimated based onthe mixed-source isotope data under the consideration ofprior information at the site (see Sect S1 for detailed infor-mation regarding model frame and computing method) Asdescribed above theoretically there are four major sourcespotentially contributing to ambient NOx road traffic coalcombustion biomass burning and biogenic soil As a startwe tentatively integrated all four sources into MixSIR (datanot shown) The relative contribution of biomass burningto the ambient NOx (median value) ranged from 28 to70 (average 42 ) representing the most important sourceThe primary reason for such apparently high contribution bybiomass burning is that the corrected δ15NndashpNOminus3 values ofminus429042

minus1032plusmn 366 permil are relatively close to the N isotopicsignature of biomass-burning-emitted NOx (104plusmn 413 permil)compared to the other possible sources Based on δ15N alonethe isotope approach can be ambiguous if there are more thantwo sources The N isotope signature of NOx from biomassburning falls right in between the spectrum of plausible val-ues with highest δ15N for emissions from coal combustionon the one end and much lower values for automotive and soilemissions on the other and will be similar to a mixed signa-ture from coal combustion and NOx emissions from traffic



Figure 5 (a) Time-series variation of coal combustion and road traffic contribution to the mass concentrations of ambient pNOminus3 in Nan-jing as estimated through MixSIR (b) correlation analysis between the mass concentrations of coal-combustion-related pNOminus3 and SO2(c) correlation analysis between the mass concentrations of road-traffic-related pNOminus3 and CO

We can make several evidence-based presumptions to bet-ter constrain the emission sources in the mixing model anal-ysis (1) when sampling at a typical urban site in a major in-dustrial city in China we can assume that the sources of roadtraffic and coal combustion are dominant while the contri-bution of biogenic soil to ambient NOx should have mini-mal impact or can be largely neglected (Zhao et al 2015)(2) there is no crop harvest activity in eastern China duringthe winter season Furthermore deforestation and combus-tion of fuelwood have been discontinued in Chinarsquos majorcities (Chang et al 2016a) Therefore the contribution ofbiomass-burning-emitted NOx during the sampling periodshould also be minor Indeed Fig S4 shows that the massconcentration of biomass-burning-related pNOminus3 is not cor-related with the fraction of levoglucosan that contributes toOC confirming a weak impact of biomass burning on thevariation of pNOminus3 concentration during our study period

In a second alternative and more realistic scenario we ex-cluded biomass burning and soil as a potential source of NOxin MixSIR (see above) As illustrated in Fig 5a assumingthat NOx emissions in urban Nanjing during our study periodoriginated solely from road traffic and coal combustion theirrelative contribution to the mass concentration of pNOminus3 is125plusmn 91 microg mminus3 (or 68plusmn 11 ) and 49plusmn 25 microg mminus3 (or32plusmn 11 ) respectively These numbers agree well with acity-scale NOx emission inventory established for Nanjingrecently (Zhao et al 2015) Nevertheless on a nation-wide

level relatively large uncertainties with regards to the over-all fossil fuel consumption and fuel types propagate intolarge uncertainties of NOx concentration estimates and pre-dictions of longer-term emission trends (Li et al 2017) Cur-rent emission-inventory estimates (Jaegle et al 2005 Zhanget al 2012 Liu et al 2015 Zhao et al 2013) suggest that in2010 NOx emissions from coal-fired power plants in Chinawere about 30 higher than those from transportation How-ever our isotope-based source apportionment of NOx clearlyshows that in 2014 the contribution from road traffic to NOxemissions at least in Nanjing (a city that can be consideredrepresentative for most densely populated areas in China)is twice that of coal combustion In fact due to changingeconomic activities emission sources of air pollutants inChina are changing rapidly For example over the past sev-eral years China has implemented an extended portfolio ofplans to phase out its old-fashioned and small power plantsand to raise the standards for reducing industrial pollutantemissions (Chang 2012) On the other hand China con-tinuously experienced double-digit annual growth in termsof auto sales during the 2000s and in 2009 it became theworldrsquos largest automobile market (X Liu et al 2013 Changet al 2017 2016b) Recent satellite-based studies have suc-cessfully analyzed the NOx vertical column concentrationratios for megacities in eastern China and highlighted theimportance of transportation-related NOx emissions (Reuteret al 2014 Gu et al 2014 Duncan et al 2016 Jin et



Figure 6 Estimates of the relative importance of single NOx sources (meanplusmn1σ ) throughout China based on the original δ15NndashNOminus3values extracted from the literature (εN = 0 permil) and under consideration of significant N isotope fractionation during NOx transformation(εN = 5 permil 10 permil 15 permil or 20 permil)

al 2017) Moreover long-term measurements of the ratioof NOminus3 to non-sea-salt SO2minus

4 in precipitation and aerosoljointly revealed a continuously increasing trend in easternChina throughout the latest decade suggesting decreasingemissions from coal combustion (X Liu et al 2013 Itahashiet al 2018) Both coal-combustion- and road-traffic-relatedpNOminus3 concentrations are highly correlated with their corre-sponding tracers (ie SO2 and CO respectively) confirmingthe validity of our MixSIR modeling results With justifiedconfidence in our Bayesian isotopic model results we con-clude that previous estimates of NOx emissions from auto-motivetransportation sources in China based on bottom-upemission inventories may be too low

43 Previous δ15NndashNOminus3 -based estimates on NOx

sources

Stable nitrogen isotope ratios of nitrate have been used toidentify nitrogen sources in various environments in Chinaoften without large differences in δ15N between rainwaterand aerosol NOminus3 (Kojima et al 2011) In previous work noconsideration was given to potential N isotope fractionationduring atmospheric pNOminus3 formation Here we reevaluated700 data points of δ15NndashNOminus3 in aerosol (minus077plusmn 452 permiln= 308) and rainwater (379plusmn 614 permil n= 392) from 13sites that are located in the area of mainland China and theYellow East China and South China seas (Fig 1) extractedfrom the literature (see Table S4 for details) To verify the po-tentially biasing effects of neglecting N isotope fractionation(ie testing the sensitivity of ambient NOx source contribu-tion estimates to the effect of N isotope fractionation) theBayesian isotopic mixing model was applied (a) to the origi-nal NOminus3 isotope data set and (b) to the corrected nitrate iso-tope data set accounting for the N isotope fractionation dur-

ing NOx transformation All 13 sampling sites are located innon-urban areas therefore apart from coal combustion andon-road traffic the contributions of biomass burning and bio-genic soil to nitrate need to be taken into account

Although most of the sites are located in rural and coastalenvironments when the original data set is used without theconsideration of N isotope fractionation in the Bayesian iso-topic mixing model fossil-fuel-related NOx emissions (coalcombustion and on-road traffic) appear to be the largest con-tributor at all the sites (data are not shown) This is partic-ularly true for coal combustion everywhere except for thesites of Dongshan islands and Mt Lumin NOx emissionsseem to be dominated by coal combustion Very high con-tribution from coal combustion (on the order of 40 ndash60 )particularly in northern China may be plausible and can beattributed to a much larger consumption of coal Yet ratherunlikely the highest estimated contribution of coal combus-tion (83 ) was calculated for Beihuang Island (a full-yearsampling on a costal island that is 65 km north of ShandongPeninsula and 185 km east of the BeijingndashTianjinndashHebei re-gion) and not for mainland China While Beihuang may bean extreme example we argue that collectively the contri-bution of coal combustion to ambient NOx in China as cal-culated on the basis of isotopic analyses in previous studieswithout the consideration of N isotope fractionation repre-sents overestimates

As a first step towards a more realistic assessment of theactual partitioning of NOx sources in China in general (andcoal-combustion-emitted NOx in particular) it is imperativeto determine the location-specific values for εN Unfortu-nately without δ18OndashNOminus3 data on hand or data on mete-orological parameters that correspond to the 700 δ15NndashNOminus3values used in our meta-analysis it is not possible to esti-



mate the εN values through the abovementioned CQC mod-ule As a viable alternative we adopted the approximatevalues for εN as estimated in Sanjiang (1099 permil) and Nan-jing (1533plusmn 490 permil) As indicated in Fig 6 the estimatesof the source partitioning are sensitive to the choice of εNWhereas with increasing εN estimates on the relative con-tribution of on-road traffic and biomass burning remainedrelatively stable estimates for coal combustion and biogenicsoil changed significantly in opposite directions More pre-cisely depending on εN the average estimate of the frac-tional contribution of coal combustion decreased drasticallyfrom 43 (εN = 0 permil) to 5 (εN = 20 permil) (Fig 6) while thecontribution from biogenic soil to NOx emissions increasedin a complementary way Given the lack of better constraintson εN for the 13 sampling sites it cannot be our goal hereto provide a robust revised estimate on the partitioning ofNOx sources throughout China and its neighboring areasBut we have very good reasons to assume that disregard ofN isotope fractionation during pNOminus3 formation in previousisotope-based source apportionment studies has likely led tooverestimates of the relative contribution of coal combustionto total NOx emissions in China For what we would con-sider the most conservative estimate ie lowest calculatedvalue for the N isotope fractionation during the transforma-tion of NOx to pNOminus3 (εN = 5 permil) the approximate contribu-tion from coal combustion to the NOx pool would be 28 more than 30 less than N-isotope-mixing-model-based es-timates would yield without consideration of the N isotopefractionation (ie εN = 0 permil) (Fig 6)

5 Conclusion and outlook

Consistent with theoretical predictions δ15NndashpNOminus3 datafrom a field experiment where atmospheric pNOminus3 forma-tion could be attributed reliably to NOx solely from biomassburning revealed that the conversion of NOx to pNOminus3 isassociated with a significant net N isotope effect (εN) It isimperative that future studies making use of isotope mixingmodels to gain conclusive constraints on the source partition-ing of atmospheric NOx consider this N isotope fractiona-tion The latter will change with time and space dependingon the distribution of ozone and OH radicals in the atmo-sphere and the predominant NOx chemistry The O isotopesignatures of pNOminus3 is mostly chemistry (and not source)driven (modulated by O isotope exchange reactions in theatmosphere) and thus O isotope measurements do not allowaddressing the ambiguities with regards to the NOx sourcethat may remain when just looking at δ15N values aloneHowever δ18O in pNOminus3 will help in assessing the relativeimportance of the dominant pNOminus3 formation pathway Si-multaneous δ15N and δ18O measurements of atmosphericnitrate thus allow reliable information on εN and in turnon the relative importance of single NOx sources For ex-ample for Nanjing which can be considered representative

for other large cities in China dual-isotopic and chemical-tracer evidence suggest that on-road traffic and coal-firedpower plants rather than biomass burning are the predomi-nant sources during high-haze pollution periods Given thatthe increasing frequency of nitrate-driven haze episodes inChina our findings are critically important in terms of guid-ing the use of stable nitrate isotope measurements to eval-uate the relative importance of single NOx sources on re-gional scales and for adapting suitable mitigation measuresFuture assessments of NOx emissions in China (and else-where) should involve simultaneous δ15N and δ18O measure-ments of atmospheric nitrate and NOx at high spatiotemporalresolution allowing us to more quantitatively reevaluate for-mer N-isotope-based NOx source partitioning estimates

Data availability Data are available from the corresponding authoron request We prefer not to publish the software for calculating thenitrogen isotope fractionation factor and estimating nitrate sourceattribution at the present stage in order to avoid compromising thefuture of ongoing software registration Readers can download thesoftware through the website httpwwwatmosgeochemcom (lastaccess 1 August 2018) after the completion of software registra-tion

The Supplement related to this article is availableonline at httpsdoiorg105194acp-18-11647-2018-supplement

Author contributions YZ conceived the study YZ ML and YC de-signed the experimental strategy YC CT XM FC SZ ML and TKperformed the geochemical and isotope measurements analyzed theexperimental data and constructed the model YC and YZ proposedthe hypotheses XL and WZ assisted with the laboratory work YCwrote the manuscript with ML and YZ all other co-authors con-tributed to the data interpretation and writing

Competing interests The authors declare that they have no conflictof interest

Special issue statement This article is part of the special issueldquoMultiphase chemistry of secondary aerosol formation under severehazerdquo It is not associated with a conference

Acknowledgements This study was supported by the National KeyResearch and Development Program of China (2017YFC0210101)the National Natural Science Foundation of China (grant nos91644103 41705100 and 41575129) the Provincial NaturalScience Foundation of Jiangsu (BK20170946) the UniversityScience Research Project of Jiangsu Province (17KJB170011) theInternational Joint Laboratory on Climate and Environment Change(ILCEC) and the Collaborative Innovation Center on Forecastand Evaluation of Meteorological Disasters (CIC-FEMD) through



NUIST and through University of Basel research funds

Edited by Daniel KnopfReviewed by two anonymous referees

References

Alexander B Hastings M G Allman D J Dachs J Thorn-ton J A and Kunasek S A Quantifying atmospheric nitrateformation pathways based on a global model of the oxygen iso-topic composition (117O) of atmospheric nitrate Atmos ChemPhys 9 5043ndash5056 httpsdoiorg105194acp-9-5043-20092009

Altieri K E Hastings M G Gobel A R Peters A J andSigman D M Isotopic composition of rainwater nitrate atBermuda the influence of air mass source and chemistry in themarine boundary layer J Geophys Res 118 11304ndash311316httpsdoiorg101002jgrd50829 2013

Anenberg S C Miller J Minjares R Du L Henze D KLacey F Malley C S Emberson L Franco V Klimont Zand Heyes C Impacts and mitigation of excess diesel-relatedNOx emissions in 11 major vehicle markets Nature 545 467ndash471 httpsdoiorg101038nature22086 2017

Boumlhlke J K Mroczkowski S J and Coplen T BOxygen isotopes in nitrate new reference materials for18O17O16O measurements and observations on nitrate-waterequilibration Rapid Commun Mass Sp 17 1835ndash1846httpsdoiorg101002rcm1123 2003

Cao F Zhang S C Kawamura K and Zhang Y LInorganic markers carbonaceous components and sta-ble carbon isotope from biomass burning aerosols inNortheast China Sci Total Environ 572 1244ndash1251httpsdoiorg101016jscitotenv201509099 2016

Cao F Zhang S C Kawamura K Liu X Yang C Xu ZFan M Zhang W Bao M Chang Y Song W Liu S LeeX Li J Zhang G and Zhang Y L Chemical characteristicsof dicarboxylic acids and related organic compounds in PM25during biomass-burning and non-biomass-burning seasons at arural site of Northeast China Environ Pollut 231 654ndash662httpsdoiorg101016jenvpol201708045 2017

Casciotti K L Sigman D M Hastings M G BoumlhlkeJ K and Hilkert A Measurement of the oxygen iso-topic composition of nitrate in seawater and freshwater us-ing the denitrifier method Anal Chem 74 4905ndash4912httpsdoiorg101021ac020113w 2002

Chang Y Liu X Deng C Dore A J and Zhuang GSource apportionment of atmospheric ammonia before duringand after the 2014 APEC summit in Beijing using stable nitro-gen isotope signatures Atmos Chem Phys 16 11635ndash11647httpsdoiorg105194acp-16-11635-2016 2016a

Chang Y Zou Z Deng C Huang K Collett J L Lin Jand Zhuang G The importance of vehicle emissions as asource of atmospheric ammonia in the megacity of Shanghai At-mos Chem Phys 16 3577ndash3594 httpsdoiorg105194acp-16-3577-2016 2016b

Chang Y Deng C Cao F Cao C Zou Z Liu S LeeX Li J Zhang G and Zhang Y Assessment of carbona-ceous aerosols in Shanghai China ndash Part 1 long-term evo-

lution seasonal variations and meteorological effects AtmosChem Phys 17 9945ndash9964 httpsdoiorg105194acp-17-9945-2017 2017

Chang Y H China needs a tighter PM25 limit and achange in priorities Environ Sci Technol 46 7069ndash7070httpsdoiorg101021es3022705 2012

Duncan B N Lamsal L N Thompson A M Yoshida Y LuZ Streets D G Hurwitz M M and Pickering K E A space-based high-resolution view of notable changes in urban NOxpollution around the world (2005ndash2014) J Geophys Res 121976ndash996 httpsdoiorg1010022015JD024121 2016

Elliott E Kendall C Wankel S D Burns D BoyerE Harlin K Bain D and Butler T Nitrogen iso-topes as indicators of NOx source contributions to atmo-spheric nitrate deposition across the midwestern and north-eastern United States Environ Sci Technol 41 7661ndash7667httpsdoiorg101021es070898t 2007

Elliott E M Kendall C Boyer E W Burns D A LearG G Golden H E Harlin K Bytnerowicz A ButlerT J and Glatz R Dual nitrate isotopes in dry deposi-tion Utility for partitioning NOx source contributions to land-scape nitrogen deposition J Geophys Res 114 G04020httpsdoiorg1010292008jg000889 2009

Fang Y T Koba K Wang X M Wen D Z Li J Take-bayashi Y Liu X Y and Yoh M Anthropogenic imprintson nitrogen and oxygen isotopic composition of precipitationnitrate in a nitrogen-polluted city in southern China AtmosChem Phys 11 1313ndash1325 httpsdoiorg105194acp-11-1313-2011 2011

Felix J D and Elliott E M Isotopic composition of pas-sively collected nitrogen dioxide emissions Vehicle soil andlivestock source signatures Atmos Environ 92 359ndash366httpsdoiorg101016jatmosenv201404005 2014

Felix J D Elliott E M and Shaw S L Nitrogen iso-topic composition of coal-fired power plant NOx Influ-ence of emission controls and implications for global emis-sion inventories Environ Sci Technol 46 3528ndash3535httpsdoiorg101021es203355v 2012

Felix J D Elliott E M Gish T J McConnell L L andShaw S L Characterizing the isotopic composition of atmo-spheric ammonia emission sources using passive samplers and acombined oxidation-bacterial denitrifier approach Rapid CommMass Sp 27 2239ndash2246 httpsdoiorg101002rcm66792013

Fibiger D L and Hastings M G First measurementsof the nitrogen isotopic composition of NOx frombiomass burning Environ Sci Technol 50 11569ndash11574httpsdoiorg101021acsest6b03510 2016

Freyer H D Seasonal trends of NH+4 and NOminus3 nitrogen isotopecomposition in rain collected at Juumllich Germany Tellus 30 83ndash92 1978

Freyer H D Seasonal variation of 15N14N ratiosin atmospheric nitrate species Tellus 43 30ndash44httpsdoiorg103402tellusbv43i115244 2017

Freyer H D Kley D Volz-Thomas A and Kobel K On theinteraction of isotopic exchange processes with photochemicalreactions in atmospheric oxides of nitrogen J Geophys Res98 14791ndash14796 httpsdoiorg10102993JD00874 1993



Fu X Wang S Zhao B Xing J Cheng Z LiuH and Hao J Emission inventory of primary pollu-tants and chemical speciation in 2010 for the YangtzeRiver Delta region China Atmos Environ 70 39ndash50httpsdoiorg101016jatmosenv201212034 2013

Galloway J N Aber J D Erisman J W Seitzinger S PHowarth R W Cowling E B and Cosby B J The nitrogencascade Bioscience 53 341ndash356 httpsdoiorg1016410006-3568(2003)053[0341Tnc]20Co2 2003

Geng L Murray L T Mickley L J Lin P Fu Q Schauer AJ and Alexander B Isotopic evidence of multiple controls onatmospheric oxidants over climate transitions Nature 546 133ndash136 httpsdoiorg101038nature22340 2017

Gobel A R Altieri K E Peters A J Hastings M G and Sig-man D M Insights into anthropogenic nitrogen deposition tothe North Atlantic investigated using the isotopic compositionof aerosol and rainwater nitrate Geophys Res Lett 40 5977ndash5982 httpsdoiorg1010022013GL058167 2013

Gu D Wang Y Smeltzer C and Boersma K F An-thropogenic emissions of NOx over China Reconciling thedifference of inverse modeling results using GOME-2 andOMI measurements J Geophys Res 119 7732ndash7740httpsdoiorg1010022014JD021644 2014

Hastings M G Sigman D M and Lipschultz F Isotopic evi-dence for source changes of nitrate in rain at Bermuda J Geo-phys Res 108 4790 httpsdoiorg1010292003JD0037892003

Heaton T H E 15N14N ratios of NOx from vehicle en-gines and coal-fired power stations Tellus 42 304ndash307httpsdoiorg101034j1600-0889199000007x-i1 1990

Hennigan C J Sullivan A P Collett J L and RobinsonA L Levoglucosan stability in biomass burning particles ex-posed to hydroxyl radicals Geophys Res Lett 37 L09806httpsdoiorg1010292010GL043088 2010

Hoering T The isotopic composition of the ammonia and thenitrate ion in rain Geochim Cosmochim Ac 12 97ndash102httpsdoiorg1010160016-7037(57)90021-2 1957

Itahashi S Yumimoto K Uno I Hayami H Fujita S-I PanY and Wang Y A 15-year record (2001ndash2015) of the ratio ofnitrate to non-sea-salt sulfate in precipitation over East Asia At-mos Chem Phys 18 2835ndash2852 httpsdoiorg105194acp-18-2835-2018 2018

Jaegle L Steinberger L Martin R V and Chance KGlobal partitioning of NOx sources using satellite obser-vations Relative roles of fossil fuel combustion biomassburning and soil emissions Faraday Discuss 130 407ndash423httpsdoiorg101039B502128F 2005

Jedynska A Hoek G Wang M Eeftens M Cyrys J Bee-len R Cirach M De Nazelle A Nystad W MakaremAkhlaghi H Meliefste K Nieuwenhuijsen M de Hoogh KBrunekreef B and Kooter I M Spatial variations and devel-opment of land use regression models of levoglucosan in fourEuropean study areas Atmos Chem Phys Discuss 14 13491ndash13527 httpsdoiorg105194acpd-14-13491-2014 2014

Ji S Cherry C R Zhou W Sawhney R Wu Y CaiS Wang S and Marshall J D Environmental justice as-pects of exposure to PM25 emissions from electric vehi-cle use in China Environ Sci Technol 49 13912ndash13920httpsdoiorg101021acsest5b04927 2015

Jin X Fiore A M Murray L T Valin L C Lamsal L N Dun-can B Folkert Boersma K De Smedt I Abad G G ChanceK and Tonnesen G S Evaluating a space-based indicator ofsurface ozone-NOx -VOC sensitivity over midlatitude source re-gions and application to decadal trends J Geophys Res 122439ndash461 httpsdoiorg1010022017JD026720 2017

Kaiser J C Riemer N and Knopf D A Detailedheterogeneous oxidation of soot surfaces in a particle-resolved aerosol model Atmos Chem Phys 11 4505ndash4520httpsdoiorg105194acp-11-4505-2011 2011

Kendall C Elliott E M and Wankel S D Stable isotopes inecology and environmental science chap 12 2nd Edn Black-well Oxford 2007

Knopf D A Mak J Gross S and Bertram A K Doesatmospheric processing of saturated hydrocarbon surfaces byNO3 lead to volatilization Geophys Res Lett 33 L17816httpsdoiorg1010292006GL026884 2006

Knopf D A Forrester S M and Slade J H Heterogeneous ox-idation kinetics of organic biomass burning aerosol surrogatesby O3 NO2 N2O5 and NO3 Phys Chem Chem Phys 1321050ndash21062 httpsdoiorg101039C1CP22478F 2011

Kojima K Murakami M Yoshimizu C Tayasu INagata T and Furumai H Evaluation of surfacerunoff and road dust as sources of nitrogen using ni-trate isotopic composition Chemosphere 84 1716ndash1722httpsdoiorg101016jchemosphere201104071 2011

Lamsal L N Martin R V Padmanabhan A van Donke-laar A Zhang Q Sioris C E Chance K KurosuT P and Newchurch M J Application of satelliteobservations for timely updates to global anthropogenicNOx emission inventories Geophy Res Lett 38 L05810httpsdoiorg1010292010GL046476 2011

Leighton P Photochemistry of Air Pollution Academic NewYork 1961

Levy H Moxim W J and Kasibhatla P S A globalthree-dimensional time-dependent lightning source of tro-pospheric NOx J Geophys Res 101 22911ndash22922httpsdoiorg10102996JD02341 1996

Li D and Wang X Nitrogen isotopic signature of soil-releasednitric oxide (NO) after fertilizer application Atmos Environ42 4747ndash4754 httpsdoiorg101016jatmosenv2008010422008

Li M Zhang Q Kurokawa J-I Woo J-H He K Lu ZOhara T Song Y Streets D G Carmichael G R ChengY Hong C Huo H Jiang X Kang S Liu F Su Hand Zheng B MIX a mosaic Asian anthropogenic emissioninventory under the international collaboration framework ofthe MICS-Asia and HTAP Atmos Chem Phys 17 935ndash963httpsdoiorg105194acp-17-935-2017 2017

Ling T Y and Chan C K Formation and transformation ofmetastable double salts from the crystallization of mixed ammo-nium nitrate and ammonium sulfate particles Environ Sci Tech-nol 41 8077ndash8083 httpsdoiorg101021es071419t 2007

Liu D Li J Zhang Y Xu Y Liu X Ding P Shen CChen Y Tian C and Zhang G The use of levoglucosanand radiocarbon for source apportionment of PM25 carbona-ceous aerosols at a background site in East China Environ SciTechnol 47 10454ndash10461 httpsdoiorg101021es401250k2013



Liu F Zhang Q Tong D Zheng B Li M Huo Hand He K B High-resolution inventory of technologies ac-tivities and emissions of coal-fired power plants in Chinafrom 1990 to 2010 Atmos Chem Phys 15 13299ndash13317httpsdoiorg105194acp-15-13299-2015 2015

Liu J Li J Zhang Y Liu D Ding P Shen C Shen KHe Q Ding X Wang X Chen D Szidat S and ZhangG Source apportionment using radiocarbon and organic tracersfor PM25 carbonaceous aerosols in Guangzhou South Chinacontrasting local- and regional-scale haze events Environ SciTechnol 48 12002ndash12011 httpsdoiorg101021es503102w2014

Liu X Zhang Y Han W Tang A Shen J Cui Z VitousekP Erisman J W Goulding K Christie P Fangmeier Aand Zhang F Enhanced nitrogen deposition over China Nature494 459ndash463 httpsdoiorg101038nature11917 2013

Lu Z Streets D G de Foy B Lamsal L N Duncan BN and Xing J Emissions of nitrogen oxides from US ur-ban areas estimation from Ozone Monitoring Instrument re-trievals for 2005ndash2014 Atmos Chem Phys 15 10367ndash10383httpsdoiorg105194acp-15-10367-2015 2015

Michalski G Scott Z Kabiling M and Thiemens MH First measurements and modeling of 117O in at-mospheric nitrate Geophys Res Lett 30 1870ndash1872httpsdoiorg1010292003gl017015 2003

Miyazaki K Eskes H Sudo K Boersma K F Bowman Kand Kanaya Y Decadal changes in global surface NOx emis-sions from multi-constituent satellite data assimilation AtmosChem Phys 17 807ndash837 httpsdoiorg105194acp-17-807-2017 2017

Morin S Savarino J Frey M M Yan N Bekki S BottenheimJ W and Martins J M Tracing the origin and fate of NOx inthe Arctic atmosphere using stable isotopes in nitrate Science322 730ndash732 httpsdoiorg101126science1161910 2008

Morino Y Kondo Y Takegawa N Miyazaki Y Kita KKomazaki Y Fukuda M Miyakawa T Moteki N andWorsnop D R Partitioning of HNO3 and particulate nitrateover Tokyo Effect of vertical mixing J Geophys Res 111D15215 httpsdoiorg1010292005JD006887 2006

Park Y M Park K S Kim H Yu S M Noh S KimM S Kim J Y Ahn J Y Lee M D Seok K Sand Kim Y H Characterizing isotopic compositions of TC-C NOminus3 -N and NH+4 -N in PM25 in South Korea Impactof Chinarsquos winter heating Environ Pollut 233 735ndash744httpsdoiorg101016jenvpol201710072 2018

Parnell A C Phillips D L Bearhop S Semmens B X WardE J Moore J W Jackson A L Grey J Kelly D J and In-ger R Bayesian stable isotope mixing models Environmetrics24 387ndash399 httpsdoiorg101002env2221 2013

Phillips D L Inger R Bearhop S Jackson A L Moore JW Parnell A C Semmens B X and Ward E J Best prac-tices for use of stable isotope mixing models in food-web stud-ies Can J Zool 92 823ndash835 httpsdoiorg101139cjz-2014-0127 2014

Price C Penner J and Prather M NOx from lightning 1 Globaldistribution based on lightning physics J Geophys Res 1025929ndash5941 httpsdoiorg10102996JD03504 1997

Reuter M Buchwitz M Hilboll A Richter A SchneisingO Hilker M Heymann J Bovensmann H and Burrows J

P Decreasing emissions of NOx relative to CO2 in East Asiainferred from satellite observations Nat Geosci 7 792ndash795httpsdoiorg101038ngeo2257 2014

Richter A Burrows J P Nuumlszlig H Granier C andNiemeier U Increase in tropospheric nitrogen dioxideover China observed from space Nature 437 129ndash132httpsdoiorg101038nature04092 2005

Savarino J Kaiser J Morin S Sigman D M and ThiemensM H Nitrogen and oxygen isotopic constraints on the origin ofatmospheric nitrate in coastal Antarctica Atmos Chem Phys7 1925ndash1945 httpsdoiorg105194acp-7-1925-2007 2007

Seinfeld J H and Pandis S N Atmospheric chemistry andphysics From air pollution to climate change John Wiley ampSons 2012

Shiraiwa M Poumlschl U and Knopf D A Multiphase chemicalkinetics of NO3 radicals reacting with organic aerosol compo-nents from biomass burning Environ Sci Technol 46 6630ndash6636 httpsdoiorg101021es300677a 2012

Sigman D M Casciotti K L Andreani M Barford CGalanter M and Boumlhlke J K A bacterial method for the nitro-gen isotopic analysis of nitrate in seawater and freshwater AnalChem 73 4145ndash4153 httpsdoiorg101021ac010088e2001

Simoneit B R T Biomass burning ndash a review of organic tracers forsmoke from incomplete combustion Appl Geochem 17 129ndash162 httpsdoiorg101016S0883-2927(01)00061-0 2002

Simoneit B R T Schauer J J Nolte C G Oros D R Elias VO Fraser M P Rogge W F and Cass G R Levoglucosan atracer for cellulose in biomass burning and atmospheric particlesAtmos Environ 33 173ndash182 httpsdoiorg101016S1352-2310(98)00145-9 1999

Smith M L Bertram A K and Martin S T Deli-quescence efflorescence and phase miscibility of mixedparticles of ammonium sulfate and isoprene-derived sec-ondary organic material Atmos Chem Phys 12 9613ndash9628httpsdoiorg105194acp-12-9613-2012 2012

Solomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Climate change 2007The physical science basis contribution of Working Group I tothe Fourth Assessment Report of the Intergovernmental Panel onClimate Change Cambridge University Press New York 2007

Thiemens M H Mass-independent isotope effects in planetary at-mospheres and the early solar system Science 283 341ndash345httpsdoiorg101126science2835400341 1999

Thiemens M H and Heidenreich J E The mass-independentfractionation of oxygen a novel isotope effect and its pos-sible cosmochemical implications Science 219 1073ndash1075httpsdoiorg101126science21945881073 1983

Turekian V C Macko S Ballentine D Swap R J andGarstang M Causes of bulk carbon and nitrogen isotope frac-tionations in the products of vegetation burns laboratory stud-ies Chem Geol 152 181ndash192 httpsdoiorg101016S0009-2541(98)00105-3 1998

Walters W W and Michalski G Theoretical calculation of nitro-gen isotope equilibrium exchange fractionation factors for vari-ous NOy molecules Geochim Cosmochim Ac 164 284ndash297httpsdoiorg101016jgca201505029 2015

Walters W W and Michalski G Theoretical calculation of oxygenequilibrium isotope fractionation factors involving various NOy



molecules OH and H2O and its implications for isotope vari-ations in atmospheric nitrate Geochim Cosmochim Ac 19189ndash101 httpsdoiorg101016jgca201606039 2016

Walters W W Goodwin S R and Michalski G Ni-trogen stable isotope composition (δ15N) of vehicle-emitted NOx Environ Sci Technol 49 2278ndash2285httpsdoiorg101021es505580v 2015

Walters W W Simonini D S and Michalski G Nitro-gen isotope exchange between NO and NO2 and its im-plications for δ15N variations in tropospheric NOx andatmospheric nitrate Geophys Res Lett 43 440ndash448httpsdoiorg1010022015gl066438 2016

Wang H Lu K Chen X Zhu Q Chen Q Guo S Jiang MLi X Shang D Tan Z Wu Y Wu Z Zou Q Zheng YZeng L Zhu T Hu M and Zhang Y High N2O5 concen-trations observed in urban Beijing Implications of a large ni-trate formation pathway Environ Sci Tech Let 4 416ndash420httpsdoiorg101021acsestlett7b00341 2017

Wankel S D Chen Y Kendall C Post A F and Paytan ASources of aerosol nitrate to the Gulf of Aqaba Evidence fromδ15N and δ18O of nitrate and trace metal chemistry Mar Chem120 90ndash99 httpsdoiorg101016jmarchem2009010132010

Wojtal P K Miller D J OrsquoConner M Clark S C and Hast-ings M G Automated high-resolution mobile collection sys-tem for the nitrogen isotopic analysis of NOx Jove-J Vis Exp118 e54962 httpsdoiorg10379154962 2016

Yienger J J and Levy H Empirical model of global soil-biogenic NOx emissions J Geophys Res 100 11447ndash11464httpsdoiorg10102995JD00370 1995

Zhang Q Geng G Wang S Richter A and He KSatellite remote sensing of changes in NOx emissions overChina during 1996ndash2010 Chinese Sci Bull 57 2857ndash2864httpsdoiorg101007s11434-012-5015-4 2012

Zhang R Tie X and Bond D W Impacts of anthro-pogenic and natural NOx sources over the US on tropo-spheric chemistry P Natl Acad Sci USA 100 1505ndash1509httpsdoiorg101073pnas252763799 2003

Zhao B Wang S X Liu H Xu J Y Fu K Klimont ZHao J M He K B Cofala J and Amann M NOx emis-sions in China historical trends and future perspectives At-mos Chem Phys 13 9869ndash9897 httpsdoiorg105194acp-13-9869-2013 2013

Zhao Y Qiu L P Xu R Y Xie F J Zhang Q Yu Y YNielsen C P Qin H X Wang H K Wu X C Li W Q andZhang J Advantages of a city-scale emission inventory for ur-ban air quality research and policy the case of Nanjing a typicalindustrial city in the Yangtze River Delta China Atmos ChemPhys 15 12623ndash12644 httpsdoiorg105194acp-15-12623-2015 2015

Zong Z Wang X Tian C Chen Y Fang Y Zhang F Li CSun J Li J and Zhang G First assessment of NOx sources ata regional background site in North China using isotopic analysislinked with modeling Environ Sci Technol 51 5923ndash5931httpsdoiorg101021acsest6b06316 2017


Abstract

Introduction

Methods

Field sampling

Laboratory analysis

Calculation of N isotope fractionation value (N)

Bayesian isotope mixing model

Results

Sanjiang in northern China

Nanjing in eastern China

Discussion

Sanjiang campaign theoretical calculation and field validation of N isotope fractionation during pNO3- formation

Source apportionment of NOx in an urban setting using a Bayesian isotopic mixing model

Previous 15N--NO3--based estimates on NOx sources

Conclusion and outlook

Data availability

Author contributions

Competing interests

Special issue statement

Acknowledgements

References


CQC module is consistent with that simulated by the WeatherResearch and Forecasting model coupled with Chemistry(WRF-Chem) further confirming the robustness of our esti-mates Our investigations also show that without the consid-eration of the N isotope effect during pNOminus3 formation theobserved δ15NndashpNOminus3 at the study site would erroneouslyimply that NOx is derived almost entirely from coal com-bustion Similarly reanalysis of reported δ15NndashNOminus3 datathroughout China and its neighboring areas suggests thatNOx emissions from coal combustion may be substantivelyoverestimated (by gt 30 ) when the N isotope fractionationduring atmospheric pNOminus3 formation is neglected

1 Introduction

Nitrogen oxides (NOx = NO + NO2) are among the mostimportant molecules in tropospheric chemistry They are in-volved in the formation of secondary aerosols and atmo-spheric oxidants such as ozone (O3) and hydroxyl radicals(OH) which control the self-cleansing capacity of the at-mosphere (Galloway et al 2003 Seinfeld and Pandis 2012Solomon et al 2007) The sources of NOx include both an-thropogenic and natural origins with more than half of theglobal burden (sim 40 Tg N yrminus1) currently attributed to fos-sil fuel burning (224ndash261 Tg N yrminus1) and the rest primar-ily derived from nitrificationdenitrification in soils (includ-ing wetlands 89plusmn 19 Tg N yrminus1) biomass burning (58plusmn18 Tg N yrminus1) lightning (2ndash6 Tg N yrminus1) and oxidation ofN2O in the stratosphere (01ndash06 Tg N yrminus1) (Jaegle et al2005 Richter et al 2005 Lamsal et al 2011 Price et al1997 Yienger and Levy 1995 Miyazaki et al 2017 Dun-can et al 2016 Anenberg et al 2017 Levy et al 1996)The mainultimate sinks for NOx in the troposphere arethe oxidation to nitric acid (HNO3(g)) and the formation ofaerosol-phase particulate nitrate (pNOminus3 ) (Seinfeld and Pan-dis 2012) the partitioning of which may vary on diurnal andseasonal timescales (Morino et al 2006)

Emissions of NOx occur mostly in the form of NO (Se-infeld and Pandis 2012 Leighton 1961) During daytimetransformation from NO to NO2 is rapid (few minutes) andproceeds in a photochemical steady state controlled by theoxidation of NO by O3 to NO2 and the photolysis of NO2back to NO (Leighton 1961)

NO +O3 minusrarr NO2+O2 (R1)NO2+hv minusrarr NO+O (R2)

O+O2Mminusrarr O3 (R3)

where M is any non-reactive species that can take up the en-ergy released to stabilize O NOx oxidation to HNO3 is gov-erned by the following equations During daytime

NO2+OHMminusrarr HNO3 (R4)

During nighttime

NO2+O3 minusrarr NO3+O2 (R5)

NO3+NO2Mminusrarr N2O5 (R6)

N2O5+H2O(surface)aerosolminusrarr 2HNO3 (R7)

HNO3 then reacts with gas-phase NH3 to form ammoniumnitrate (NH4NO3) aerosols If the ambient relative humid-ity (RH) is lower than the efflorescence relative humid-ity (ERH) or crystallization relative humidity (CRH) solid-phase NH4NO3(s) is formed (Smith et al 2012 Ling andChan 2007)

NH4NO3 HNO3 (g)+NH3 (g) (R8a)

If ambient RH exceeds the ERH or CRH HNO3 and NH3dissolve into the aqueous phase (aq) (Smith et al 2012 Lingand Chan 2007)

HNO3 (g)+NH3 (g) NOminus3 (aq)+NH+4 (aq) (R8b)

While global NOx emissions are well constrained individualsource attribution and their local or regional role in particu-late nitrate formation are difficult to assess due to the shortlifetime of NOx (typically less than 24 h) and the high degreeof spatiotemporal heterogeneity with regards to the ratio be-tween gas-phase HNO3 and particulate NOminus3 (pNOminus3 ) (Dun-can et al 2016 Lu et al 2015 Zong et al 2017 Zhang etal 2003) Given the conservation of the nitrogen (N) atombetween NOx sources and sinks the N isotopic compositionof pNOminus3 can be related to the different origins of the emit-ted NOx and thus provides valuable information on the par-titioning of the NOx sources (Morin et al 2008) Such a Nisotope balance approach works best if the N isotopic com-position of various NOx sources display distinct 15N14N ra-

tios (reported as δ15N =(15N14N

)sampleminus(

15N14N)N2

(15N14N)N2times 1000)

The δ15NndashNOx of coal-fired power plant (+10 permil to+25 permil)(Felix et al 2012 2013 Heaton 1990) vehicle (+37 permilto +57 permil) (Heaton 1990 Walters et al 2015 Felix andElliott 2014 Felix et al 2013 Wojtal et al 2016) andbiomass burning (minus7 permil to +12 permil) emissions (Fibiger andHastings 2016) for example is generally higher than thatof lightning (minus05 permil to +14 permil) (Hoering 1957) and bio-genic soil (minus489 permil to minus199 permil) emissions (Li and Wang2008 Felix and Elliott 2014 Felix et al 2013) allowingthe use of isotope mixing models to gain insight on the NOxsource apportionment for gases aerosols and the resultingnitrate deposition (minus15 permil to +15 permil) (Elliott et al 20072009 Zong et al 2017 Savarino et al 2007 Morin etal 2008 Park et al 2018 Altieri et al 2013 Gobel etal 2013) In addition because of mass-independent frac-tionation during its formation (Thiemens 1999 Thiemensand Heidenreich 1983) ozone possesses a strong isotopeanomaly (117Oasymp δ17Ominus 052timesδ18O) which is propagated







2 Methods

21 Field sampling

























)pNOminus3 minusNOx






(1)



) The remainder is









= 1000times


1minus fNO2

)(1minus fNO2

)+(

15αNO2NOtimes fNO2

)] (2)














(4)



represent the O







=23

[1000

(18αNO2NOminus1)(

1minus fNO2

)(1minus fNO2

)+(

18αNO2NOtimes fNO2


13

[(δ18OminusH2O

)+ 1000

(18αOHH2Ominus 1

)]

(5)





56

(δ18OminusN2O5

)+

16

(δ18OminusH2O

) (6)





1000(mαXY minus 1

)=A

T 4 times 1010+B

T 3 times 108+C

T 2 times 106

+D

Ttimes 104 (7)






3 Results








390 plusmn 479 and178452


4 Discussion


3 formation








(meanmax

min plusmn 1σ)



























sources





















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References






2 Methods

21 Field sampling

























)pNOminus3 minusNOx






(1)



) The remainder is









= 1000times


1minus fNO2

)(1minus fNO2

)+(

15αNO2NOtimes fNO2

)] (2)














(4)



represent the O







=23

[1000

(18αNO2NOminus1)(

1minus fNO2

)(1minus fNO2

)+(

18αNO2NOtimes fNO2


13

[(δ18OminusH2O

)+ 1000

(18αOHH2Ominus 1

)]

(5)





56

(δ18OminusN2O5

)+

16

(δ18OminusH2O

) (6)





1000(mαXY minus 1

)=A

T 4 times 1010+B

T 3 times 108+C

T 2 times 106

+D

Ttimes 104 (7)






3 Results








390 plusmn 479 and178452


4 Discussion


3 formation








(meanmax

min plusmn 1σ)



























sources





















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References






















)pNOminus3 minusNOx






(1)



) The remainder is









= 1000times


1minus fNO2

)(1minus fNO2

)+(

15αNO2NOtimes fNO2

)] (2)














(4)



represent the O







=23

[1000

(18αNO2NOminus1)(

1minus fNO2

)(1minus fNO2

)+(

18αNO2NOtimes fNO2


13

[(δ18OminusH2O

)+ 1000

(18αOHH2Ominus 1

)]

(5)





56

(δ18OminusN2O5

)+

16

(δ18OminusH2O

) (6)





1000(mαXY minus 1

)=A

T 4 times 1010+B

T 3 times 108+C

T 2 times 106

+D

Ttimes 104 (7)






3 Results








390 plusmn 479 and178452


4 Discussion


3 formation








(meanmax

min plusmn 1σ)



























sources





















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References


1000(mαXY minus 1

)=A

T 4 times 1010+B

T 3 times 108+C

T 2 times 106

+D

Ttimes 104 (7)






3 Results








390 plusmn 479 and178452


4 Discussion


3 formation








(meanmax

min plusmn 1σ)



























sources





















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References






(meanmax

min plusmn 1σ)



























sources





















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References





















sources





















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References
















References

































































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References









































































Abstract

Introduction

Methods

Field sampling

Laboratory analysis



Results



Discussion





Data availability


Competing interests


Acknowledgements

References

nitrogen isotope fractionation during gas-to-particle ... · ted noxand thus provides valuable...

Documents