Environmental Science and Pollution Research

, Volume 20, Issue 7, pp 4659–4670

Emission characteristics of air pollutants from incense and candle burning in indoor atmospheres

  • A. Manoukian
  • E. Quivet
  • B. Temime-Roussel
  • M. Nicolas
  • F. Maupetit
  • H. Wortham
Research Article

DOI: 10.1007/s11356-012-1394-y

Cite this article as:
Manoukian, A., Quivet, E., Temime-Roussel, B. et al. Environ Sci Pollut Res (2013) 20: 4659. doi:10.1007/s11356-012-1394-y


Volatile organic compounds (VOCs) and particles emitted by incense sticks and candles combustion in an experimental room have been monitored on-line and continuously with a high time resolution using a state-of-the-art high sensitivity-proton transfer reaction-mass spectrometer (HS-PTR-MS) and a condensation particle counter (CPC), respectively. The VOC concentration–time profiles, i.e., an increase up to a maximum concentration immediately after the burning period followed by a decrease which returns to the initial concentration levels, were strongly influenced by the ventilation and surface interactions. The obtained kinetic data set allows establishing a qualitative correlation between the elimination rate constants of VOCs and their physicochemical properties such as vapor pressure and molecular weight. The emission of particles increased dramatically during the combustion, up to 9.1(±0.2) × 104 and 22.0(±0.2) × 104 part cm−3 for incenses and candles, respectively. The performed kinetic measurements highlight the temporal evolution of the exposure level and reveal the importance of ventilation and deposition to remove the particles in a few hours in indoor environments.


Incense Candle Volatile organic compounds Aerosol HS-PTR-MS Indoor air quality 


Because of the life style induced by modern society, people tend to spend most of their time in various kinds of indoor environments such as home, workplace or microenvironments (transport for example) (Klepeis et al. 2001). Many studies (OQAI 2006) show that air in closed environments is frequently more contaminated than outdoors air. Contaminants include a wide range of organic and inorganic substances in the gaseous and particulate phases (Lewis and Gordon 1996). Both acute and chronic exposures to these pollutants could induce adverse health effects such as damage to the nervous system, immune and reproductive diseases, respiratory system dysfunction, developmental problems and cancers (WHO 2005).

A fraction of the indoor pollution can result from the penetration of outdoor pollutants, but most of the indoor environments contain their own sources of pollution (WHO 2005). These sources are commonly classified in the literature according to categories such as building materials, furniture, food preparation, cleaning, heating, combustion and people (He et al. 2004; Afshari et al. 2005; Wallace 2006; Zai et al. 2006; He et al. 2007; See et al. 2007; Mannino and Orecchio 2008). Combustion sources such as candle and incense burning have been identified as sources of volatile organic compounds (VOCs) and ultrafine particles.

Candles and incenses are used for various purposes (e.g., religious or spiritual, aesthetic, and therapeutic reasons) and nowadays also for creating a pleasant household atmosphere. Numerous studies have been carried out to characterize the VOCs and particles emission and to quantify the emission induced by incense (among the latest studies: Hu et al. 2009; Maupetit and Squinazi 2009; Glytsos et al. 2010; Ji et al. 2010) and candle combustion (Maupetit and Squinazi 2009; Pagels et al. 2009; Glytsos et al. 2010; Orecchio 2011; Derudi et al. 2012). Nevertheless, very few of them are dedicated to determine the concentration–time profiles during and after combustion event.

Thus, the aim of this study was to investigate the VOC and particle concentration–time profiles using on-line analysers in order to estimate the emission and elimination rate constants induced by candle and incense stick burning. These results can provide key parameters for future evaluation of the potential health impacts during candle and incense stick burning, such as peak concentration and exposure duration. In addition, a tentative correlation could be made between the elimination rate constants and physicochemical properties (boiling point, melting point, molecular volume, molecular weight, refractive index, vapor pressure, density and solubility in water) of the quantified VOCs under study.

Materials and methods


The incense investigated was in stick form (14 cm) and based with spice fragrances. Its core, made of bamboo stem, was covered by layers of wood powder (m ≈ 0.5 g). The scented candle was hand-poured into a 200-ml container (m ≈ 350 g) and was composed of vegetable oils and beeswax. The wick is in pure cotton. No dye is used for the manufacture of incense and candle under study. To reduce inter-experimental variability the four replicates were carried out using the same kind of incense stick and candle coming from the same supplier batch.

Room experiments

Experiments were carried in the “Mechanised house for Advanced Research on Indoor Air” (MARIA; Scientific and Technical Center of Building Marne-la-Vallée, France). For more details about MARIA house, the readers are referred to Ribéron and O’Kelly (2002).

The experimental room selected for the study has a rectangular shape (2.53 × 5.15 m), with a volume of 32.3 m3. The ceiling was painted concrete and the walls were covered with patches of painted plaster. Incense sticks and candles were put on a support set in the middle of the room at 1 m above floor level. Except for this support, no furniture was present in the room.

Ventilation was provided using a controlled mechanical exhaust system located at the bottom of the door (Fig. 1). This device draws out air from the room and brings in outdoor air through a tiny hole above the window on the opposite wall. The ventilation flow was kept constant at 25.8 m3 h−1. In all experiments, the indoor climatic conditions were measured as follows: air exchange rate (AER) = 0.80 (±0.08) h–1, room temperature (T) = 16.2 (±0.4)°C and relative humidity (RH) = 50 (±6)%. Prior to the experiments, the room was flushed during 12 h with outdoor air in order to reach the atmospheric background level for VOCs and particles.
Fig. 1

Layout of the ventilated room (a 2.53 m; b 2.50 m; c 5.15 m; ① incense stick and candle position; ② narrow opening for outdoor air input; ③ closed door with an opening at the bottom connected to the mechanical extraction system; ④ closed window; ⑤ sampling devices)

Sampling and analyses

Sampling devices (⑤, Fig. 1) were operated in the neighboring corridor of the MARIA experimental room. Three sampling lines were connected to the controlled mechanical exhaust system located at the bottom of the door (③, Fig. 1). Two of them were dedicated to on-line measurements using a high sensitivity-proton transfer reaction-mass spectrometer (HS-PTR-MS) and a condensation particle counter (CPC) while the third is used for off-line measurements: adsorbent trap and DNPH cartridges.

VOCs on-line measurements

To ensure an on-line measurement of VOCs with a high temporal resolution, an HS-PTR-MS (Ionicon Analytik) was used. This analytical technique is described in details elsewhere (de Gouw and Warneke 2007). Briefly, the HS-PTR-MS consists of a discharge ion source that produces H3O+, a drift tube where proton transfer reactions between VOC and H3O+ occur, and a quadrupole mass analyser coupled with a secondary electron multiplier that detects the resulting ions.

The HS-PTR-MS was operated at standard conditions of drift voltage (600 V) and drift pressure (2.2 mbar), resulting in an E/N of 136 Townsend (1 Td = 10−17 V cm2), where E is the electric field (V cm−1) and N is the ambient air number density in the drift tube (cm−3). These source settings made it possible to carry out quantification and ion assignment, i.e., high enough to lessen reagent H3O+ hydration (and thus prevent humidity effect on sensitivity) and low enough to minimize ion product fragmentation.

Measurements were made in the ‘mass scan’ mode, whereby a complete mass spectrum ranging between 20 and 270 amu was acquired at a mass detection rate of 1 s amu−1. According to these experimental conditions, the HS-PTR-MS provides concentration–time profiles with a time resolution of 4 min.

Because of the soft ionisation method, the ions produced corresponded mainly to the pseudo-molecular ion (M + 1 amu). The quantification of VOCs at M + 1 amu was based on calibrations with certified analysed gas standard cylinders (Restek, Praxair; see their composition in Supporting Information) containing different mixtures of hydrocarbons and oxygenated compounds at ppb levels (±10 %). Alternatively, when no reliable gas standard was available, quantification was based on the conversion of ion counts to mixing ratio using the proton transfer rate constants k (cm3 s−1) reported by Zhao and Zhang (2004) and on the experimentally determined m/z specific transmission. According to this procedure, compounds can be determined with an accuracy of ±30 % (Holzinger et al. 2005). Finally, for compounds with no individual proton transfer rate constant value available, the default k value of 2 × 10−9 cm3 s−1 was used to obtain a rough estimate of their concentrations.

VOCs off-line measurements

To validate the HS-PTR-MS measurements and to deconvolute its signal when several VOCs produce the same ion, off-line analyses were simultaneously carried out. VOCs were collected on Tenax TA sorbent tubes (Perkin Elmer), at a flow rate of 90 ml min−1 during 60 min (according to standard NF ISO 16000–6; AFNOR 2005) and analysed by thermal desorption (ATD Clarus 400, Perkin Elmer) and gas chromatography (GC) coupled to a flame ionisation detector (FID) (Varian 3800 GC-FID), for quantification and a mass spectrometer (MS) (Varian Saturn 2000) for identification.

Gaseous carbonyl compounds were collected on DiNitroPhenylHydrazine (DNPH) coated short-body cartridges (Waters), at a flow rate of 900 ml min−1 during 60 min (according to NF ISO 16000–3; AFNOR 2002) and analysed using HPLC-UV (Alliance, Waters), after liquid extraction with 5 ml of acetonitrile.

The limits of quantification for both off-line techniques (ATD-GC-FID and HPLC-UV) are 0.3 μg m−3.

For these two kinds of off-line analysis, five samples were collected during each combustion experiment as follows:
  • Prior to burning, one 1-h sample was collected to measure the indoor VOC background concentrations (S0 in the text).

  • One air sample was collected during the entire burning event corresponding to 0.5 and 1 h for incense sticks and candles, respectively (S1 in the text).

  • After the burning experiment, three successive 1-h samples were collected (S2, S3 and S4, respectively in the text).

Measurement of the number of particles

On-line measurement of particle concentrations was carried out using a CPC (5.403; GRIMM Inc). It was set at a flow rate of 0.3 l min−1 to count submicrometer particles in diameters from 0.004 to 3 μm, which provided the exact number concentration of particles ranging from single events up to 107 part cm−3. A scanning mobility particle sizer (SMPS, CPC combined with a Differential Mobility Analyser column [L-DMA; GRIMM Inc.]) was also used for additional laboratory experiments (n = 4 for both incense and candle experiments) in order to measure the distribution of particles from 11.1 to 1,083.3 nm in diameter. According to the instrument configuration (CPC or SMPS) the measurement frequency was set for analysis every 4 s or 5 min, respectively.

Modelisation of the concentration profiles

The emission rates of VOCs and particles were calculated according to the law of conservation of mass (Eqs. 1a and b, respectively).
$$ \frac{{\mathrm{d}{C_{\mathrm{in}}}}}{{\mathrm{d}t}}=\frac{{{E_{\mathrm{R}}}}}{V}-{k_{\mathrm{obs}}}\times {C_{\mathrm{in}}}+{k_{\mathrm{AER}}}\times {C_{\mathrm{out}}} $$
$$ \frac{{\mathrm{d}{C_{\mathrm{in}}}}}{{\mathrm{d}t}}=\frac{{{E_{\mathrm{R}}}}}{V}-{k_{\mathrm{obs}}}\times {C_{\mathrm{in}}}+P\times {k_{\mathrm{AER}}}\times {C_{\mathrm{out}}} $$
where Cin (μg m−3 or part cm−3) is the indoor VOC or particle concentration, Cout (μg m−3 or part cm−3) is the outdoor VOC or particle concentration, t (h) is the time, ER (μg h−1 or part h−1) is the emission rate of the compound or particle under study, V (m3) is the volume of the experimental room (V = 32.3 m3), kobs (h−1) is the elimination rate constant and kAER (h−1) is the AER (kAER = 0.80 (±0.08) h−1), and P (dimensionless) is the penetration efficiency (fraction of outdoor particles that penetrate the building).
Assuming that the emission rates are constant throughout the combustion event and that the penetration efficiency is close to one for particles, the analytical solution of Eq. 1a and b is obtained according to Eq. 2:
$$ {C_{\mathrm{in}}}=\frac{{{E_{\mathrm{R}}}}}{{{k_{\mathrm{obs}}}V}}+\frac{{{k_{\mathrm{AER}}}{C_{\mathrm{out}}}}}{{{k_{\mathrm{obs}}}}}+\left( {{C_0}-\frac{{{E_{\mathrm{R}}}}}{{{k_{\mathrm{obs}}}V}}-\frac{{{k_{\mathrm{AER}}}{C_{\mathrm{out}}}}}{{{k_{\mathrm{obs}}}}}} \right)\times {{\mathrm{e}}^{{-{k_{\mathrm{obs}}}t}}} $$
where C0 (μg m−3 or part cm−3) is the initial concentration (background levels) of VOC or particle.
When the combustion event was over, the emission rates were set to zero (ER= 0) and pollutant concentrations decreased. As a result, Eq. 2 could be simplified resulting in Eq. 3, which indicates that indoor concentrations decreased according to a first order kinetics:
$$ {{C}_{{{\rm{in}}}}} = \frac{{{{k}_{{{\rm{AER}}}}}{{C}_{{{\rm{out}}}}}}}{{{{k}_{{{\rm{obs}}}}}}} + {{C}_{{\max }}} \times {{{\rm{e}}}^{{ - {{k}_{{{\rm{obs}}}}}t}}} $$
where Cmax (μg m−3 or part cm−3) is the maximum concentration reached immediately after the combustion period.

To determine the elimination rate constant kobs, exponential models were developed based on HS-PTR-MS and CPC data (using OriginPro 8 SR0; OriginLab Corporation). Only values obtained with a correlation coefficient above 99 % were considered.

The elimination rate constant was determined as a sum of the rate constants of the air exchange and all the other removal process including reactive degradation, surface adsorption, phase change and deposition according to Eq. 4.
$$ k_{{{\text{obs}}}} = k_{{{\text{AER}}}} + k_{{{\text{removal}}\;{\text{processes}}}} $$
where kAER (h−1) is the AER constant and kremoval processes (h−1) is the kinetic rate constant including all the other removal processes.

Results and discussion

Volatile organic compounds

Concentrations of the VOCs

According to the HS-PTR-MS measurements, 44 main ions were detected in the mass range 20–270 amu. Due to the absence of chromatographic separation prior to analysis, assignment of ions to VOC appears challenging, as several VOCs and ion fragments may have contributed to the same m/z. This issue was partly resolved performing the aforementioned off-line measurements in addition to the on-line HS-PTR-MS measurements. As a result, the kinetic studies were focused specifically on ions identified and detected by both on-line and off-line techniques. Table 1 summarises the 27 compounds identified by both Tenax and DNPH analysis (AFNOR 2002, 2005) and detected by HS-PTR-MS.
Table 1

Compounds identified by Tenax and DNPH analysis (AFNOR 2002, 2005) and detected by HS-PTR-MS (in bold the pseudo molecular ion)


Ions detected by HS-PTR-MS (uma)



Monoterpenes (C10H16)

81, 137

g, k, n, o, p, t


Acenaphthene (C12H10)


c, m, r, s, u, w, x

Acetaldehyde (C2H4O)


a, b, c, d, e, g, v, w, y

Acetone (C3H6O)

46, 59

d, e, h, i, v, w

Acrylonitrile (C3H3N)



Benzaldehyde (C7H6O)


b, c, d, g, i, m, q, w, y

Benzene (C6H6)


c, d, g, i, j, l, y

Biphenyl (C12H10)


c, m, r, s, u, w

Butadiene (C4H6)


h, k

Butyraldehyde (C4H8O)

46, 53, 55, 73

d, h, i, v

Ethanol (C2H6O)



Ethylbenzene (C8H10)


b, c, d, g, i, m, q, w

Formaldehyde (CH2O)


a, b, c, d, e, f, g, v, w, y

Furfural (C5H4O2)


b, c, m, w

Furfuryl alcohol (C5H6O2)


m, w

Hexaldehyde (C6H12O)

55, 83, 101

h, e, w

Isoprene (C5H8)


h, k, l

Linalool (C10H18O)


c, m, r, s, u, w

5-Methylfurfural (C6H6O2)



Monoterpenes (C10H16)

81, 137

c, g, k, m, n, o, p, q

Naphthalene (C10H8)


g, m, q, r, u, x, y

Pentanal (C5H10O)

69, 87

d, e, h, k, l, w

Pentanone (C5H10O)


d, e, w

Propionaldehyde (C3H6O)


d, e, i, v, w,

Styrene (C8H8)


c, d, g, i, m

Toluene (C7H8)


c, e, g, I, q, y

Xylene (C6H4CH3)2


b, c, d, g, i, m, q, w, y

(a) Lin and Wang (1994), (b) Ho and Yu (2002), (c) Eggert and Hansen (2004), (d) Lee and Wang (2004), (e) Wang et al. (2007), (f) Yang et al. (2007b), (g) Maupetit and Squinazi (2009), (h) Schwarz et al. (2009), (i) Yang et al. (2007a), (j) Navasumrit et al. (2008), (k) Buhr et al. (2002), (l) Löfroth et al. (1991), (m) Tran and Marriott (2007), (n) de Gouw and Warneke (2007), (o) Taipale et al. (2008), (p) Warneke et al. (2003), (q) Tran and Marriott (2008), (r) Yang et al. (2007c), (s) Lung and Hu (2003), (t) Buchbauer et al. (1995), (u) Guo et al. (2004), (v) Lee and Lin (1996), (w) Chuang (2010), (x) Orecchio (2011), (y) Derudi et al. (2012)

Candle and Incense stick combustion

During the candle experiments, HS-PTR-MS detected only signals at m/z 81 and m/z 137 (Table 1). These ions are generally related to monoterpenes (MW = 136) as 137 is the pseudo molecular ion and 81 its main fragmentation product (Tani et al. 2003). This was confirmed by the off-line analysis as limonene and traces of alpha-pinene and beta-pinene (below the limit of quantification) were identified by ATD-GC-MS. Nevertheless the higher experimental ratio m/z 137/81 (1.5) than that obtained in similar HS-PTR-MS conditions (0.8) (Tani et al. 2003) suggests the presence of other compounds contributing to the ion signal at m/z 137.

Several other compounds were detected by off-line analysis (in particular formaldehyde) but at low concentration levels (dozens of ppt) which explains that the HS-PTR-MS did not measure significant temporal variation of their concentrations.

During the incense experiments, 26 compounds were detected by HS-PTR-MS corresponding to 24 pseudo molecular or fragment ions (Table 1). Unfortunately, among the compounds identified in incense smoke in previous works (see references in Table 1), seven ions (m/z 46, 55, 59, 69, 87, 107 and 155) could correspond to several compounds but using the off-line measurement, it was possible to evaluate the contribution of these compounds to their respective m/z:
  • m/z 46: It could correspond both to acetone and butyraldehyde fragments. The contribution of these molecules to the ion m/z 46 was calculated according to the off-line measurement (DNPH cartridge) assuming that the ratio acetone/butyraldehyde was constant during the sampling period.

  • m/z 55: According to the HS-PTR-MS ionisation and fragmentation pattern, the ion m/z 55 can be attributed to the water cluster H3O+,(H2O)2 produced in the drift tube of the HS-PTR-MS but also to the butadiene pseudo molecular ion and to an hexaldehyde and butyraldehyde fragment. Because of the stability of the RH in the experimental room, the signal of m/z 55 does not correspond to the formation of water cluster. In the off-line samples, butadiene was below the ATD-GC-FID limit of quantification (0.3 μg m−3) which demonstrates its low contribution in m/z 55 signal. As a result, m/z 55 could be induced both by the fragmentation of hexaldehyde and butyraldehyde.

  • m/z 59: This m/z could correspond to the pseudo molecular ion of both acetone and propionaldehyde. The HPLC-UV analysis shows a distribution between propionaldehyde (25 %) and acetone (75 %). As a result, the signal must be assumed as a mixture.

  • m/z 69: According to the HS-PTR-MS ionisation and fragmentation procedure, this ion could correspond to the isoprene pseudo molecular ion and to a fragment of pentanal. Nevertheless, the isoprene concentration level measured according to the ATD-GC-FID is low compared to the HS-PTR-MS signal. The fragmentation pattern (Buhr et al. 2002) shows that the signal of m/z 69 could not be attributed only to the pentanal fragment. As a result, two assumptions can be put forward: there is another compound giving a signal at m/z 69 (unfortunately not detected or below the quantification limit of the off-line methods), or ion m/z 87 corresponds to the sum of the pentanal pseudo molecular ion plus another compound.

  • m/z 87: It could correspond both to pentanal and pentanone. In the off-line samples, pentanone was detected at low levels. However its contribution to signal m/z 87 could explain the overestimation of ion m/z 69.

  • m/z 107: In the off-line samples, benzaldehyde was ten times more concentrated than xylene and ethylbenzene. As a result, the ion m/z 107, which can be obtained from benzaldehyde, xylene and ethylbenzene, was mainly due to benzaldehyde and thereafter the corresponding signal it was assumed to be benzaldehyde.

  • m/z 155: In the off-line samples, acenaphthene and biphenyl were detected below the ATD-GC-FID limit of quantification. As a result, the ion m/z 155 was mainly due to linalool and thereafter the corresponding signal supposedly corresponds to linalool.

In Fig. 2a, the intercomparison of five chemicals (acetaldehyde, acetone + propionaldehyde, benzene and toluene) which were chosen as model compounds is presented. This result demonstrates the good agreement between HS-PTR-MS and ATD-GC-FID for VOCs represented here by benzene and toluene, and between HS-PTR-MS and HPLC-UV for carbonyl compounds represented here by acetaldehyde and the sum of acetone and propionaldehyde.
Fig. 2

Concentration–time profile of off-line and HS-PTR-MS measured during and after incense burning (n = 4). Histograms: cartridges analysis (AFNOR 2002, 2005). Dot plot: HS-PTR-MS analysis. a Acetaldehyde, acetone + propionaldehyde, benzene and toluene. b formaldehyde and hexaldehyde

In addition, in Fig. 2b, the disagreement between the on-line and off-line measurements observed for two compounds (formaldehyde and hexaldehyde) can be appreciated. The lower formaldehyde concentration levels and higher hexaldehyde concentration levels were obtained using HS-PTR-MS measurements. The underestimation of formaldehyde was most probably induced by its low proton affinity (170.1 kcal mol−1; Czakó et al. 2009) only slightly higher than that of H2O (166.5 kcal mol−1) (Steinbacher et al. 2004). As a result, a significant back reaction, which efficiency depends strongly on the RH, could have occurred with H2O. This phenomenon has already been observed in a previous work that suggested an underestimation of the formaldehyde concentrations up to a factor of five (Vlasenko et al. 2010). Under our analytical conditions, the underestimation of formaldehyde concentration is slightly higher and reaches a factor of 7.

Regarding the overestimation of hexaldehyde, it is possible that other compounds, not detected by the off-line techniques, may have contributed to the signal at m/z 101 and consequently interfered with the pseudo molecular ion of hexaldehyde. The same overestimation was observed for two other carbonyl compounds (butyraldehyde and pentanal). Attributing the entire signal m/z 73 and 87 to butyraldehyde and pentanal, respectively, induces an overestimation of their concentration indicating that other compounds contribute to these ion signals.

In conclusion, even with a help of complementary off-line methods and comparison with dataset of previous studies (Table 1), the attribution of ion observed in the HS-PTR-MS mass spectrum of a complex mixture such as incense smoke remains a difficult task. In our conditions, this is particularly true for carbonyl compounds for which specific ion assignment could be ambiguous. For the remaining compounds (Table 1), the correlation between off-line and on-line measurements highlighted good results, implying a correct attribution of ions to compounds.

Emission and elimination rate constants

For each ion, minimum and maximum concentration levels corresponding to the extreme values observed during the experiments, the emission (ER) and elimination rate constants (kobs) were determined using the database of the HS-PTR-MS and according to Eq. 2 described in Section Modelisation of the concentration profiles. The calculated results, given in Table 2, are normalized to 0.5 and 5 g of the burning mass for incense (about one incense stick burning during approximately 30 min) and for candle (about 1 h candle burning), respectively.
Table 2

Ions detected by HS-PTR-MS during and after combustion (n = 4), minimum and maximum concentration levels (μg m−3); and estimation of emission rate constant (μg h−1) and elimination rate constant (h−1)

Ions detected by HS-PTR-MS (uma)

Min–max average (μg m−3)

Standard deviation

Emission rate constant (μg h−1) ± SD

Elimination rate constant (h−1) ± SD





352 ± 3

0.95 ± 0.04




513 ± 7

0.89 ± 0.06





1206 ± 17

1.81 ± 0.05




4910 ± 11

1.09 ± 0.01




251 ± 6

1.10 ± 0.10




598 ± 17

0.82 ± 0.09




418 ± 4

0.91 ± 0.04




220 ± 4

0.88 ± 0.08




1730 ± 32

0.95 ± 0.06




2951 ± 28

1.13 ± 0.03




2221 ± 11

0.94 ± 0.02




1078 ± 15

1.43 ± 0.05




937 ± 5

0.86 ± 0.02




608 ± 4

1.39 ± 0.03




651 ± 5

1.06 ± 0.03




997 ± 9

1.50 ± 0.04




770 ± 7

0.90 ± 0.03




1845 ± 17

1.23 ± 0.04




934 ± 8

1.47 ± 0.04




207 ± 7

1.23 ± 0.12




726 ± 7

1.38 ± 0.04




940 ± 49

1.50 ± 0.13




431 ± 6

1.23 ± 0.06




217 ± 4

1.60 ± 0.08




278 ± 9

1.27 ± 0.13




92 ± 4

1.36 ± 0.20

aFor other m/z, the standard k value of 2 × 10−9 cm3 s−1 was used in calculation of estimate concentration

bk value reported by Zhao and Zhang (2004)

cConcentrations are calculated from gas calibration standard cylinder

During candle and incense burning, the identified emitted compounds followed the same temporal profile. The concentrations of VOC reached a maximum at the beginning of the second period (S2), immediately after the burning event (S1). Subsequently, the concentrations decreased to the background level (S0) (Fig. 2).

The calculated emission rate constants (ER) of the investigated gaseous compounds ranged between 92 and 4910 μg h−1 (Table 2). As can be seen in Table 2, for most of the compounds under study, the emission rate constants were in agreement with the values reported in the literature (Eggert and Hansen 2004; Guo et al. 2004; Yang et al. 2007a, b; Li et al. 2009). Unfortunately, in the literature, no data are available for acrylonitrile, furfuryl alcohol and 5-methylfurfural. However, compared to the work carried out by Eggert and Hansen (2004), lower linalool emission rate and higher furfural emission rate were obtained in the present work. This discrepancy could probably be ascribed to the amount of fragrance added in the incense sticks used in this study. Indeed, linalool, a terpene-smelling lily of the valley, and furfural, an aldehyde with the odor of almonds, are found in several essential oils and are added in various proportions in the aroma fragrance. Moreover, furfural may also be formed during the wood burning processes (Eggert and Hansen 2004). As a result, its emission rate during incense combustion depends on the nature of the wood powder surrounding the incense stick.

Regarding the elimination rate constants (kobs) as determined by Eq. 4, they were usually close to the kAER, which is 0.80 (±0.08) h−1. As a result, AER was the main removal process for a majority of the compounds. Indeed, Lung et al. (2003) and Gilbert et al. (2008) showed that, under ventilated conditions, concentrations of the gaseous pollutants decrease to nearly background levels quickly after combustion. However, the kobs values ranged from 0.82 (±0.09) to 1.81 (±0.06) h−1 depending on the compounds under study, which suggests that AER was the main but not the only process of elimination. Other possible removal processes could be exfiltration, reactive degradation and surface sorption described below:
  • The exfiltration is the uncontrolled leakage of indoor air to the outside of the building, through unidentified leaks in the building itself. Akoua et al. (2004) estimated the air leakage rate at these room walls as 1.2 m3 h−1. This corresponds to 4.4 % of the fresh air which does not pass by the inlet air. Therefore, air leakage rate which represents 0.03 h−1 of the room air flow rate has slight influence on the internal flow.

  • The reactivity of VOCs, especially with ozone, in indoor atmosphere has been studied in previous works (Fan et al. 2005; Wang and Morrison 2006). These studies demonstrated that the oxidation of VOCs by ozone leads to the formation of secondary products such as aldehydes, peroxides and condensed phase compounds, as well as OH, HO2 and RO2 radicals that, in turn, could also react with VOCs. However, based on both the known kinetic rate constants of the compounds under study with ozone and OH radical (ERADB, chemical kinetics database) and the estimation of the ozone and OH radical concentrations in MARIA experimental room, it was possible to calculate the elimination rate constants due to VOC reactivity. The ozone and OH radical concentrations were not directly measured during the experiments. Nevertheless, based on the outdoor ozone concentrations and a previous work carried out in the MARIA experimental room (Nicolas 2006), the ozone average concentrations were estimated to range between 1.2 × 1010 and 3.8 × 1010 molecules cm−3 during experiments, that is, 0.5 and 1.5 ppb, respectively. Regarding the OH radical, to the best of the authors’ knowledge, no direct measurements of OH radical concentrations are currently available (Carslaw 2007; Weschler 2011; Gomez Alvarez et al. 2012). The indoor OH radical concentration was estimated as 5 × 104 molecules cm−3 according to Sarwar et al. (2002). According to these assumptions, the estimated half-life time of the more reactive compounds is 1.1 × 103 days (naphthalene) and 5 days (hexaldehyde) with respect ozone and OH radical. As a result, it can be concluded that the oxidation processes are of minor importance in comparison to the effect of AER.

  • Regarding the surface deposition, a number of studies (Kjaer and Tirkkonene 2006; Zhang and Chen 2002; Tichenor 2004; Uhde and Salthammer 2006; Tlili et al. 2010) have demonstrated that an inherent sink effect (adsorption/desorption processes) could take place between VOCs and different kind of materials. The adsorption processes strongly depend on the indoor settings design, physicochemical properties of the VOCs and on the indoor conditions such as temperature, humidity and air velocity.

The dependence of kobs with respect their physicochemical properties (boiling point, melting point, molecular volume, molecular weight, refractive index, vapor pressure, density and solubility in water) was studied. For some of these physicochemical properties, correlations were observed when the compounds were grouped according to the chemical family they belong. Figure 3b–d suggests a qualitative relation between the elimination rate constants of aromatic hydrocarbons and their boiling point, molecular weight or vapor pressure. Data on carbonyl compounds are more scattered (Fig. 3a–c), perhaps because of the difficulty to attribute ions to these compounds (see Section Volatile organic compounds). However, a similar trend is observed both for aromatic and carbonyl compounds.
Fig. 3

kobs versus vapor pressure (a, b) and molecular weight (c, d) of the carbonyl compounds (ac) and aromatic hydrocarbons (bd) identified at room temperature

Influence of the vapor pressure

When the vapor pressure increased, i.e., compounds were more volatile, the ratio adsorption/desorption decreased (Fig. 3a–b). As a result, the adsorption phenomenon became negligible and the elimination rate constant (kobs) was close to the air exchange rate, kAER. The opposite trend was observed with the boiling point (not shown in Fig. 3), but a similar reasoning could be made. Tichenor (2004) (and references therein) and Uhde and Salthammer (2006) made the same conclusion.

Influence of the molecular weight

Although there is no direct relationship between boiling points and the molecular weights, the same trends were observed. Figure 3c–d shows that the high molecular weight favored the high elimination rate (kobs) (i.e., adsorption/desorption ratio increases).


Indoor particle concentration–time profiles for incense and candle are shown in Fig. 4. The concentration values used were those resulting from an average over four replicated experiments normalized for combustion mass of 0.5 g for incense and 5 g for candle. During the combustion stage, the particle number increased significantly from the background level 0.6 (±0.3) × 104 up to 9.1 (±0.2) × 104 part cm−3 after combustion of 0.5 g of incense in 30 min, and from 0.6 (±0.5) × 104 up to 22.0 (±0.2) × 104 part cm−3 after combustion of 5 g of candle in 1 h. Finally, particle concentrations returned in a few hours close to the background level after lighting off the candles and the incenses.
Fig. 4

Concentration-time profile of particle number for incense and candle burning (n = 4) (dashed line combustion start, broken line end of incense burning, dotted line end of candle burning)

The elimination process of particles was potentially influenced not only by AER but also by many other phenomena such as exfiltration, change of phase, coagulation and deposition (Thatcher et al. 2003; Nazaroff 2004; Bhangar et al. 2011). The deposition phenomenon depends on the particle size distribution. Because only the total particle number (without size distribution) was measured during the MARIA campaign, complementary experiments were carried out using an SMPS in a room of the laboratory having the same volume as the MARIA experimental room (see Supporting Information). SMPS scans average over all laboratory experiments show a unimodal and bimodal distributions of the particles for incense sticks and candles, respectively.

For incense (Fig. SI a), no significant variation in the unimodal distribution at 125 (±13) nm was observed between the burning step, the extinction step and the post-combustion step. The absence of evolution between burning and extinction can be induced by the continuity of the combustion process which is constituted by only a smoldering phase. On the other hand, the stability of the particle size distribution during the post-combustion step shows that no important coagulation phenomenon occurs in the room during the experiment time.

For candle (Fig. SI b), during the combustion step, only one mode was observed below 11 nm. On the other hand, when the combustion is stopped, a bimodal distribution is observed with a mode below 11 nm and a new mode centred at 92 (±9) nm. However, the intensity of this second mode is lower than the 11 nm mode. It can be possibly explained by the formation of larger particles during the extinction phase of the candle. Indeed, during this short period, the candle is in the smoldering phase characterized by a more important plume of smoke. Zai et al. (2006) had already observed this phenomenon.

These results are in agreement with the literature data, which clearly show that the diameter of incense particles (Chang et al. 2007; See et al. 2007; Géhin et al. 2008; Ji et al. 2010) is generally larger than that of the candles (Afshari et al. 2005; Zai et al. 2006; Géhin et al. 2008; Pagels et al. 2009; Glytsos et al. 2010).

The emission rate constant for candles is 2.0 (±0.1) × 1013 part h−1, which ranged between 5.3 × 1012 and 2.5 × 1013 part h−1 as reported in previous studies (Zai et al. 2006; Afshari et al. 2005). The emission rate constant for incense is 3.1 (±0.3) × 1012 part h−1, which is close to that estimated by See et al. (2007) for different types of incense sticks (ranging between 5.1 × 1012 and 1.4 × 1013 part h−1).

The time profile of the particles obtained using a CPC made it possible to calculate the elimination rate constant kobs, which is 1.11 (±0.03) and 2.87 (±0.08) h−1 for incense sticks and candles, respectively. Assuming a kAER set at 0.80 (±0.08) h−1, about 70 % and less than 30 % of the particles emitted by incense sticks and candles, respectively, were eliminated by ventilation. Exfiltration and phase change were assessed to be negligible compared to the other removal processes: ventilation and deposition. In addition, particle coagulation was not observed during the laboratory experiments with SMPS since no evolution in the size distribution was observed.

As a result, the elimination processes of particles emitted during combustion of incense and candle were mainly due to ventilation (AER) and deposition phenomena (diffusion, thermal forces, etc.). According to Eq. 5 obtained from Eq. 4, kdeposition was estimated as 0.31 (±0.11) and 2.07 (±0.16) h−1 for incense and candle, respectively.
$$ {k_{\mathrm{obs}}}={k_{\mathrm{AER}}}+{k_{\mathrm{deposition}}} $$

The deposition rate constant depends on particle properties such as size, shape and density, as well as properties of the deposition environment such as surface area and orientation, air velocity (Thatcher et al. 2002). As all these data are not available, it was difficult to use a model to estimate a theoretical value of kdeposition. Nevertheless, several studies have already reported the dependence of kdeposition versus the particle size distribution (Hussein et al. 2006; He et al. 2005 and references therein). The trends obtained in these previous works are in harmony with the present study, which indicate a higher deposition rate constants for 10 nm particles (between 2.6 and 5.4 h−1) than for 120 nm particles (between 0.2 and 1.6 h−1) (He et al. 2005).


This study confirms that the combustion of incense sticks and candles is an important source of a broad range of VOCs and particles indoors. VOC and particle concentrations increase until the stop of the combustion. While the concentrations of compounds emitted by candle under study are low, incense sticks emitted carcinogenic substances (i.e., benzene and formaldehyde) attaining the concentration levels close to the WHO guideline exposure threshold values, i.e., 17 μg m−3 (concentration associated with an excess lifetime risk of 1/10,000) and 100 μg m−3 (30-min average concentration), respectively (WHO 2010). Nevertheless, based on a 95 % removal efficiency and elimination rate constant (Table 2), these concentrations decreased in a few hours after the end of the combustion (1 h39 [m/z 31] ≤ t95% ≤ 3 h39 [m/z 47] for VOCs and 1 h03 [candle] ≤ t95% ≤ 2 h42 [incense] for particles). This event could be ascribed to the ventilation and the adsorption processes for VOCs and ventilation and deposition for the particles. If the exposure levels of the pollutants significantly increase with the occasional use of candles and incense sticks, their impact on indoor air quality remains low because of the rapid elimination processes in a room correctly ventilated.

In this sense, certain precautions ought to be taken in places that are naturally or poorly ventilated such as temples and churches (Wang et al. 2007; Navasumrit et al. 2008; Hu et al. 2009).


The AMBISAFE project is labialized by competitiveness PASS cluster and was funded by a research consortium (Albhades Provence, APF arômes et parfums, Bougie & Senteur, L’Occitane en Provence, TERA Environnement, Terre d’Oc). Audrey Manoukian is grateful to the Région Provence-Alpes-Côte-d’Azur and the Albhades Provence laboratory for the doctoral grant obtained to carry out this study. The authors thank Dr. Sasho Gligorovski for his useful comments and for English correction to the manuscript.

Supplementary material

11356_2012_1394_MOESM1_ESM.doc (136 kb)
ESM 1(DOC 136 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • A. Manoukian
    • 1
  • E. Quivet
    • 1
  • B. Temime-Roussel
    • 1
  • M. Nicolas
    • 2
  • F. Maupetit
    • 2
  • H. Wortham
    • 1
  1. 1.Aix Marseille Université, CNRS, LCE, FRE 3416Marseille Cedex 03France
  2. 2.Centre Scientifique et Technique du Bâtiment (CSTB)Saint Martin d’HèresFrance

Personalised recommendations