ABSTRACT: Hydrogenated soybean oil, referred to as soywax
by candle makers, is a renewable and biodegradable alterna-
tive to paraffin wax in candle manufacturing. Soywax was in-
vestigated for its tendency to produce soot as well as potentially
harmful organic volatiles (acrolein, formaldehyde, and acet-
aldehyde) during combustion. Beeswax and paraffin candles
were used as references. A considerable amount of soot was
produced from the combustion of paraffin candles, but little or
none was observed from soywax candles. Compared to paraffin
candles, soywax candles burned at a significantly slower rate
and required less air. Small amounts of formaldehyde were de-
tected and quantified in the fumes of burning paraffin candles.
However, formaldehyde peaks found in the chromatograms of
soy- and beeswax candles were similar to or slightly higher than
that of the blank. Since soywax candles exhibited burning prop-
erties similar to those of beeswax candles, soywax shows
promise in candle applications.
Paper no. J10259 in JAOCS 79, 803–808 (August 2002).
KEY WORDS: Candle, combustion, hydrogenated oil, organic
volatiles, soot, soybean oil, soywax.
For years, petroleum-based paraffin wax has been used as the
main component of candles. Volatile organic compounds and
small particles, commonly referred to as soot, are either di-
rectly emitted from the source material during combustion or
produced as a result of incomplete combustion of paraffin
wax (1,2). The soiling of interior surfaces of buildings due to
soot deposition was reported by Fine et al. (2). The binding
of soot to delicate materials, such as fine art, is a major con-
cern since it darkens surfaces of valuable artifacts such as
paintings and sculptures.
Soywax is a potential paraffin substitute that is biodegrad-
able, renewable, and environmentally friendly. However, its
burning characteristics are not well known. Lau et al. (1) re-
ported that 27 ng/m3acrolein was produced from the combus-
tion of 540 g stearin (a glycerol-based wax) for 3 h in a 40-m3
room. Soywax is also a glycerol-based material. Acrolein is be-
lieved to be toxic to humans and animals (3), and at 0.13 mg/m3
level, it can irritate the eyes upon exposure for 5 min. At 0.7
mg/m3, it can affect the respiratory tract. The objectives of this
work were to investigate the combustion properties of soywax
and compare them with those of paraffin and beeswax candles.
Materials. Naphthalene (laboratory grade, internal standard)
and formaldehyde solution (37%, external standard) were pur-
chased from Fisher Scientiﬁc (Fair Lawn, NJ). Acrolein (90%)
and acetaldehyde used as external standards were obtained from
Eastman Kodak Co. (Rochester, NY). 2,4,6-Trichlorophenylhy-
drazine (TCPH) (99%) was from Sigma-Aldrich Inc. (Milwau-
kee, WI), and carbon disulﬁde (CS2) was from Matheson Cole-
man & Bell (East Rutherford, NJ). Soft paraffin with a broad
melting range (25–80°C) and hard paraffin with two melting
peaks at 32 (minor) and 49°C (major) were obtained from
Dussek Campbell (Skokie, IL). Soywax (14.5% palmitic acid,
34.1% stearic acid, and 51.4% oleic acid, calculated iodine
value: 46) was purchased from Cargill Co. (Minneapolis, MN),
and beeswax from Strahl & Pitsch, Inc. (West Babylon, NY).
Air (medical grade, USP) used for in-chamber combustion of
waxes was obtained from Praxair Inc. (Danbury, CT).
Candle preparation. Approximately 200 g of wax was used
to prepare each candle. Each one (soywax, beeswax, or paraf-
ﬁn) was melted and poured into a glass container (7.0 cm in di-
ameter) with a wick (made of a cotton strip with a metal disk
connected to one end) held in the center. To avoid cracking,
chipping, and/or deforming of the candles during cooldown,
wax was poured in several steps, allowing the waxes to solidify
between steps. Soywax and beeswax were used without any ad-
ditional manipulation, but a 50:50 ratio of soft and hard paraf-
ﬁns (an appropriate ratio for proper melting; personal commu-
nication with industrial partner) was used for paraffin candles.
Melting and solidification behaviors. A differential scan-
ning calorimeter (DSC) model DSC 6200 (Seiko Instruments
Inc., Chiba, Japan) equipped with a cooling controller using
liquid nitrogen and an Exstar 6000 communication device
(Seiko Instruments Inc.) was used to evaluate the melting and
solidification behaviors of the waxes. AOCS recommended
practice Cj 1-94 (4) with modiﬁcation was used to program the
DSC system. An initial hold for 2 min at 30°C followed by
ramping at 30°C/min to 80°C and holding for 10 min was ap-
plied. Then, a cooling step at 10°C/min to −40°C and 1-min
hold followed by a heating step to 80°C at 10°C/min were used.
Soot determination. For the purpose of collecting soot, can-
dles were placed on a loop standing at a height of 49.5 cm
above the benchtop. Initially, a funnel (15.0 cm in diameter)
that was equipped with a membrane filter and connected to a
vacuum pump was positioned upside-down (Fig. 1A) away
Copyright © 2002 by AOCS Press 803 JAOCS, Vol. 79, no. 8 (2002)
*To whom correspondence should be addressed at 2312 Food Sciences
Bldg., Dept. of Food Science and Human Nutrition, Iowa State University,
Ames, IA 50011. E-mail: firstname.lastname@example.org
Combustion Characteristics of Candles
Made from Hydrogenated Soybean Oil
Karamatollah Rezaei, Tong Wang*, and Lawrence A. Johnson
Department of Food Science and Human Nutrition, and the Center for Crops Utilization Research,
Iowa State University, Ames, Iowa 50011
from the candle. To collect soot, the funnel was moved next to
the burning candle so that the lip of the funnel was 29 mm
below the edge of the candle container. Membrane ﬁlters with
5-µm pore size (SKC Inc., Eighty Four, PA) were used to col-
lect the soot. Since flame disturbance can happen at typical
room conditions, soot collection was performed under both dis-
turbed and undisturbed burning conditions. Disturbance was
deﬁned as the small turbulence occurring on the ﬂame due to
air draft, which was implemented by using the lowest speed of
a model 3733 Lasco fan (Lasco Products, Inc., West Chester,
PA). The fan was located 112 cm away from the candle stand.
Between two successive collections, the candle was allowed to
continue burning but the fan was turned off. Normal room con-
ditions were considered as nondisturbed or undisturbed condi-
tions. Air velocity at the tip of the candle ﬂame during the dis-
turbed condition was 0.71–0.77 m/s as measured by a model
8455-12 air-velocity transducer (TSI, Inc., St. Paul, MN).
A Hunter colorimeter (Hunter Associates Laboratory, Inc.,
Reston, VA) with a 1-in. (2.54 cm) aperture was used to measure
the darkness of the membrane ﬁlters. The range of L value of the
soot-coated ﬁlter is 0 to 100. The instrument was standardized
by using a black tile for 0 and a white tile for 100 values.
Burn rate. To measure the candle burn rate under normal
room conditions, three candles of each type were placed alter-
nately on the benchtop (20 cm apart) and allowed to burn for
380 min. Overall burn rates were obtained by dividing the total
weight loss during candle combustion by the total burn time. A
second set of candles was used to measure dynamic burn rates
wherein burn rates were measured at different intervals through-
out the 5-h burning period. Pool diameters and ﬂame sizes were
measured several times during the combustion periods.
Collection and analysis of acrolein, formaldehyde, and
acetaldehyde. To collect volatiles, burning was performed in a
chamber (Fig. 1B) with an air flow rate of 3.5 L/min, below
which paraffin ﬂame could not remain lit and above which the
TCPH solution, used to collect aldehydes, was bubbled away.
The fumes were allowed to pass through three consecutive im-
pingers containing 30-, 10-, and 10-mL TCPH solutions (1.0
M phosphoric acid saturated with TCPH), respectively. TCPH
precipitates aldehyde and ketone compounds as hydrazones
(5). Collection was continued for 8 h, after which the TCPH
solutions along with the precipitates from the three impingers
were transferred to a separatory funnel. The precipitates re-
maining on the surfaces of the impingers were washed off
(using 18–20 mL CS2as solvent) and added to the separatory
funnel, where all of the hydrazone products were quantita-
tively extracted into the CS2phase, with vigorous shaking, and
separated from the aqueous TCPH solution. As much as 97%
of the total hydrazones is recovered with a one-step
solvent–solvent extraction of carbonyl compounds as 2,4-dini-
trophenylhydrazones into CS2(6). Therefore, a similar extrac-
tion level was assumed when using TCPH in this study, and
only a single extraction into CS2was performed. Also, a blank
extraction was performed using the same volumetric ratio of
TCPH solution to CS2solvent as was used for the sample ex-
traction (i.e., 50:18, respectively). All extracts in the CS2
phase were stored under refrigeration until analysis.
A Hewlett-Packard series II model 5890 gas chromato-
graph equipped with an SPB-5 fused-silica column (30 m ×
0.25 mm ×0.25 µm; Supelco, Bellefonte, PA) was used to an-
alyze the CS2extracts. Both the injection port and detector
(FID) were set at 250°C, and the oven temperature was pro-
grammed as follows: 5 min holding at 50°C, 10°C/min ramp-
ing to 230°C, and 22 min of ﬁnal holding at 230°C.
Naphthalene, which had a retention time of 14.20 min, was
used as an internal standard to normalize the peaks. A primary
solution of internal standard was made by dissolving 30.0 mg
naphthalene in 3.00 mL CS2. Then, 200 µL of this solution
was mixed with 1.00 mL of sample or blank solution, and 2–3
µL of the ﬁnal solution was injected for GC analysis.
Formaldehyde, acetaldehyde, and acrolein were used as
external standards for quantification purposes; these gave
peaks at 20.13 (singlet), 21.17, and 21.70 (doublet) and 22.70
min (singlet), respectively. The doublet peaks in the case of
acetaldehyde hydrazone were due to the production of syn-
and anti-isomers speciﬁcally in CS2as solvent (6). Detection
limits obtained in this study were 1.24, 0.96, and 0.76 ng for
formaldehyde, acetaldehyde, and acrolein, respectively.
RESULTS AND DISCUSSION
Soot production. Preliminary experiments indicated a wide
variation in the level of soot production when burning paraffin
candles, in contrast to soywax and beeswax candles. There-
fore, 10 paraffin candles were allowed to burn on a benchtop.
A wide range of smoking behavior was observed in the early
stages of burning. After about 1 h, all paraffin candles pro-
duced smoke with any air movement. Similar observations of
10 soywax candles did not detect any visible smoke.
Two consecutive 2-min collections (3 min apart) of soot
were obtained from 7 of the 10 paraffin candles above, and
804 K. REZAEI ET AL.
JAOCS, Vol. 79, no. 8 (2002)
FIG. 1. Schematic of the apparatus used for the collection of soot (A)
and aldehydes (B). TCPH: trichlorophenylhydrazine solution.
variations among all observations (i.e., 7 candles ×2 observa-
tions) were analyzed. There was considerable variation in the
amount of soot produced among the seven candles (L values =
65 ± 34). This high variation in soot production was associ-
ated with the nonuniform structure of starting materials (i.e.,
the soft and hard paraffin waxes) as well as possible variations
in the crystallization during the solidiﬁcation of different can-
dle batches. For soywax and beeswax candles, the consecutive
10-min collections indicated that the total soot produced over
each collection period was negligible (hardly visible), for
which colorimetric L-values of 90 or better were obtained.
For paraffin, seven separate collections (2 min per collec-
tion over 62 min of burning) were performed under disturbed
conditions from two of the soot-producing candles identified
above, for which a mean colorimetric L-value as low as 18.7
was obtained. The average L-value was 32.4 ± 14.4 (mean ±
SD). This L-value was far below those obtained for soywax
and beeswax candles under similar conditions (95.0 ± 0.4 and
94.4 ± 2.3, respectively). In a separate experiment, one can-
dle from each type was selected and soot collection from the
disturbed conditions was compared with that of nondisturbed
conditions (Fig. 2). Average colorimetric L-values of 11.6 ±
0.8, 95.0 ± 0.9, and 86.7 ± 0.1 were obtained for two succes-
sive collections of disturbed paraffin (2-min collections), soy-
wax (10-min collections), and beeswax (10-min collections)
candles, respectively (Fig. 2). The 10-min collections for soy-
and beeswax candles were due to the fact that no soot was col-
lected within the 2-min periods applied to these types of can-
dles. For the 2-min soot collections from nondisturbed paraf-
fin candles, an average L-value of 83.6 ± 1.1 was obtained,
indicating a major effect from the disturbance in airﬂow. For
the 10-min collections from nondisturbed soywax and
beeswax candles, average L-values of 95.8 ± 0.0 and 95.7 ±
0.7, respectively, were obtained. During candle combustion
under typical conditions in many places, flame disturbance
by walking individuals is inevitable, and therefore paraffin
candles will produce soot.
The results from the experiments above are for unscented
paraffin candles. A higher amount of soot was reported by
Krause (7) when scented paraffin candles were burned. Like-
wise, limited quantities of soot can possibly be produced with
scented soywax candles depending on source and quantity of
the materials added for such purposes.
Wax consumption rate. For in-chamber burning of candles
(when collecting volatiles) at a 3.5 L/min air flow, burning
rates of 3.21 ± 0.37 and 4.26 ± 0.24 g/h were obtained for soy-
wax and paraffin candles, respectively. The air demand for
burning paraffin candles under controlled air flow was much
higher than that of soywax candles. Paraffin candles were ex-
tinguished if the air ﬂow fell below 3.5 L/min.
Wax consumption rate was also determined when candles
were burned outside the chamber at ambient conditions. Six
soywax, three beeswax, and three paraffin candles were
burned simultaneously for 380 min, and burning rate and pool
diameter from each candle were measured (Table 1). Three of
the soywax candles were trimmed as necessary. Untrimmed
soywax candles burned at somewhat lower rates than paraffin
candles (4.50 ± 0.22 vs. 5.08 ± 0.18 g/h), whereas the
trimmed soywax candles had much lower burn rates (3.89 ±
0.28 g/h). This correlated well with the ﬂame size (Fig. 3) of
soywax candles, which was smaller than that of paraffin can-
dles. Reduced wax consumption rate extends the burning time
considerably for a given candle size. A smaller ﬂame also can
be associated with lower heat production suggesting that for
some indoor applications where heat of combustion is an
issue soywax candles have advantages. Under ambient condi-
tions these observed burning rates were higher than those for
candles burned in a chamber under controlled air ﬂow.
The differences in the chemical compositions of soywax and
paraffin contribute to their different combustion behaviors.
Whereas soywax is a TAG, paraffin is a petroleum-based
COMBUSTION OF HYDROGENATED SOYBEAN OIL CANDLES 805
JAOCS, Vol. 79, no. 8 (2002)
FIG. 2. Top: Colorimetric data as indicators of soot production for dif-
ferent types of candles under different air movement conditions. Results
for paraffin are for 2-min and those for soywax and beeswax are for 10-
min collections. Bottom: Typical images from the soot collected during
the disturbed combustion of the waxes: (A) paraffin (2 min), (B) soywax
(10 min), and (C) beeswax (10 min) candles. The longer times for soy-
wax and beeswax collections were due to the lack of any soot at 2-min
Combustion Properties of Three Different Types of Candles
over a 380-min Burning Period
Burning rateaPool diametera
Candle type (g/h) (cm)
Soywax (trimmed) 3.89 ± 0.28 5.4 ± 0.2
Soywax (untrimmed) 4.50 ± 0.22 6.5 ± 0.4
Beeswax 3.28 ± 0.25 3.3 ± 0.2
Paraffin 5.08 ± 0.18 4.9 ± 0.5
aMean ± SD (n= 3).
mixture of different organic compounds, 78% (w/w) of which
are GC-extractable [i.e., the fraction of the derivatized sample
that is extractable and elutable from the GTC column (2)] con-
sisting of 93% (w/w) alkanes, 6% (w/w) alkanoic acids, and 1%
(w/w) cyclohexylalkanes (2). Compared to paraffin, soywax
contains larger molecules with lower volatility and less mobil-
ity through the wax. This results in reduced ﬂow of the melted
wax through the wick and therefore a lower consumption rate.
The burning rate of beeswax candles is also shown in
Table 1. Among the different candles studied, beeswax can-
dles burned at the lowest rate. Fine et al. (2) reported that 76%
of the total unburned beeswax was GC-extractable organic
compounds consisting of 67% (w/w) long-chain wax esters,
14% (w/w) alkanes, 15% (w/w) alkanoic acids, and 1% (w/w)
Burning soywax candles resulted in a flower- or mush-
room-like formation around the ﬂame (Fig. 3A) as a result of
wick deformation. Such formations, which were not observed
with paraffin candles, could lead to a larger combustion area
and thus higher wax consumption. Therefore, trimming was
added to our protocol for evaluating the burning of soywax
candles. For trimmed candles, the ﬂower-like structures were
broken or cut during the combustion period (380 min) every
45–60 min. The trimmed candles burned at a considerably
lower rate than those not trimmed (Table 1). The reduced sur-
face areas of the wicks with the trimmed candles resulted in
smaller flames and therefore lower consumption rates. In
reality, consumers do not need to perform a continuous trim-
ming for cost concerns. However, to avoid potential ﬁre haz-
ards and to reduce soot production, manufacturers recom-
mend trimming their candle products.
Burning rates under disturbed and nondisturbed conditions
were also examined. The overall burning rate of soywax can-
dles under disturbed conditions for 2 h was 2.62 g/h. This was
considerably lower than the burning rate obtained for soywax
candles burning in normal room conditions (4.50 g/h). Simi-
larly, paraffin candles under disturbed conditions burned at
considerably lower rates than those burned under normal room
conditions (3.85 vs. 5.08 g/h, respectively). For beeswax can-
dles, there was no detectable difference between the burning
rates of the candles burned under disturbed conditions and
those burned under normal room conditions. However, slight
changes in the burn rate were observed after the candles had
burned under normal room conditions for a longer time.
Size of liquid wax (burning) pool. The diameter of the pool
created by each type of candle was measured over the burning
period of 380 min (Table 1). Soywax candles created wider
liquid pools inside the walls of their glass containers than did
paraffin candles (Figs. 3A and 3B, respectively). The liquid
pools of beeswax candles were the smallest (Fig. 3C). Several
parameters can inﬂuence the size of the molten pool. The melt-
ing and solidification profile of the waxes as obtained by a
DSC thermograph (Figs. 3D, 3E, and 3F) plays a major role in
determining the pool size. Melting onset and peak tempera-
tures for soywax were lower than those of paraffin and
beeswax, resulting in a more liquid wax during candle com-
bustion. Another parameter in determining the pool diameter
is the wick diameter. Paraffin and soywax candles with thicker
wicks created larger pools and burned at higher rates. In con-
tainer candles, the larger pool size may be desirable to clean
the container walls as burning progresses, but too much liquid
can drown the wick and extinguish the ﬂame.
Dynamic combustion. The three types of candles were
burned for 5 h in duplicate and the changes in weight were
recorded intermittently (Fig. 4A). Paraffin candles had the
highest combustion rates (Fig. 4B), which increased with time.
806 K. REZAEI ET AL.
JAOCS, Vol. 79, no. 8 (2002)
FIG. 3. Typical differences in liquid wax pool size and flame size among the three types of candles: (A) soywax, (B) paraffin, (C) beeswax. D, E,
and F are the DSC thermograms of these waxes, respectively.
This was consistent with the higher soot production as burning
continued. The combustion rates of soywax candles decreased
within the first ~60 min but then slowly increased as burning
continued. The decrease in combustion rate conformed to the
decrease in soot production. Little or no changes were observed
in the combustion rates of beeswax candles during the 5-h com-
bustion period. The combustion rate of each candle type was
consistent with the results from similar runs performed earlier.
The size of the liquid pool around the ﬂame during the 5-h
combustion period changed at a greater rate during the early
stage of burning (0–40 min, Fig. 4C). The largest mean pool
size was observed in soywax candles. The mean pool size of
beeswax candles was initially similar to that of paraffin can-
dles, but, as burning continued, paraffin preceded that of
beeswax (Fig. 4C). Such behavior was consistent with the
DSC melting and solidiﬁcation behaviors discussed earlier.
The beeswax candles exhibited the most uniform flame
shape, which was relatively small in height and width. After
2 h of burning, flame widths were 4.0, 5.0, and 6.0 mm for
beeswax, soywax, and paraffin candles, respectively. The
flame widths shifted to 6.0, 7.0, and 7.0 mm, respectively,
after another hour of burning, when ﬂame heights of 1.3–1.8,
2.0–2.6, and 3.2 cm were observed for beeswax, soywax, and
paraffin candles, respectively. The flame shapes of the three
different types of candles were quite different with paraffin
candles being the largest and beeswax candles being the
smallest. Untrimmed soywax candles were intermediate with
a bulky flower- or mushroom-like formation in the center of
the ﬂame leading to uneven ﬂames.
Characterization of combustion compounds. A typical chro-
matogram of TCPH-captured combustion products is shown
COMBUSTION OF HYDROGENATED SOYBEAN OIL CANDLES 807
JAOCS, Vol. 79, no. 8 (2002)
FIG. 4. (A) Profile of wax consumption in paraffin, soywax, and
beeswax candles over time. (B) Dynamic combustion rate of paraffin,
soywax, and beeswax candles. (C) Changes in the diameter of the liq-
uid pool in the three different types of candles.
FIG. 5. Typical chromatogram for the trichlorophenylhydrazones of the fumes obtained from
burning paraffin candles.
for paraffin candles in Figure 5. For soywax and beeswax can-
dles, formaldehyde was detected at levels similar to or slightly
higher than that of the blank, but its presence could not be con-
fidently associated with the combustion of these waxes. A
formaldehyde peak was obvious for paraffin candles, with an
average of 1.7 mg formaldehyde per g combusted paraffin.
Based on the minimum formaldehyde recovery of 54% in the
system, as measured by a simulated run for a standard, a maxi-
mum of 3.2 mg formaldehyde is expected to be released for
each gram paraffin consumed. No acrolein was detected in the
combustion products of soywax, beeswax, or paraffin candles
over a 5–8 h period. Lau et al. (1) reported that combustion of
paraffin candles in a nondisturbed manner produced no volatile
organic compounds that were of concern to human health.
Little information is available about health issues that may
arise from the organic compounds emitted from paraffin can-
dles. Lau et al. (1) quantiﬁed some of the materials collected
and found that the levels emitted into a closed environment
were far below their tolerance levels. Fine et al. (2) reported
that many organic compounds were emitted during the com-
bustion of paraffin candles. Their experimental procedure was
not designed to collect lower-M.W. volatile organics, but
many other compounds, such as alkanes and cycloalkanes,
alkanoic acids, alkenes, and polycyclic aromatics, were
found. Polycyclic aromatics are a health concern (8) and
should not be present in a fully reﬁned vegetable oil.
Gratitude is expressed to the Iowa Soybean Promotion Board for ﬁ-
nancial support and to CandleWorks (Cedar Rapids, IA) for provid-
ing candle materials. Journal Paper No. J-19568 of the Iowa Agri-
culture and Home Economics Experiment Station, Ames, Iowa. Pro-
ject No. 0178, and supported by Hatch Act and State of Iowa funds
and a grant provided by the Iowa Soybean Promotion Board.
1. Lau, C., H. Fiedler, O. Hutzinger, K.H. Schwind, and J. Hossein-
pour, Levels of Selected Organic Compounds in Materials for
Candle Production and Human Exposure to Candle Emissions,
Chemosphere 34:1623–1630 (1997).
2. Fine, P.M., G.R. Cass, and B.R.T. Simoneit, Characterization of
Fine Particle Emissions from Burning Church Candles, Environ.
Sci. Technol. 33:2352–2362 (1999).
3. Vermeire, T., Acrolein, in Environmental Health Criteria 127,
World Health Organization, Geneva, 1992, pp. 11–14.
4. Official Methods and Recommended Practices of the American
Oil Chemists’ Society, 4th edn., AOCS Press, Champaign, 1996.
5. Johnson, D.C., and E.G. Hammond, A Sensitive Method for the
Determination of Carbonyl Compounds, J. Am. Oil Chem. Soc.
6. Smith, R.A., and I. Drummond, Trace Determination of Carbonyl
Compounds in Air by Gas Chromatography of Their 2,4-Dinitro-
phenylhydrazones, Analyst 104:875–877 (1979).
7. Krause, J.D., Characterization of Scented Candle Emissions and
Associated Public Health Risks, Ph.D. Dissertation, University
of South Florida, 1999.
8. Spaeth, K.R., Don’t Hold Your Breath: Personal Exposures to
Volatile Organic Compounds and Other Toxins in Indoor Air and
What’s (not) Being Done About It, Prev. Med. 31:631–637
[Received March 4, 2002; accepted May 14, 2002]
808 K. REZAEI ET AL.
JAOCS, Vol. 79, no. 8 (2002)