American Journal of Anal yt ical Chemistry, 2011, 2, 938-943
doi:10.4236/ajac.2011.28109 Published Online December 2011 (http://www.SciRP.org/journal/ajac)
A Greener Way to Screen Toothpaste for
Diethylene Glycol
Yale Fu, Zhigang Hao, Barry Parker, Michael Knapp
1Global Analytical Science Department, Colgate-Palmolive Company, Piscataway, USA
Received October 14, 2011; revised November 21, 2011; accepted December 3, 2011
Abstract
A method developed for the screening of diethylene glycol (DEG) in toothpaste was released by the FDA in
2007. This method could not only quantify the DEG but also confirm if any potential interfering peak is pre-
sent. However, disadvantages of this method such as intermittent shortages of the key reagent acetonitrile
and the shorter than expected column-life issues have prompted a search for alternative solutions. An im-
provement with an alternate “greener” extraction solvent is presented, and the method comparison and vali-
dation are described in this article. The greener extraction solvent, ethanol with limited water, provided a
better efficiency for the toothpaste sampling procedures. The limit of detection (LOD) and limit of quantita-
tion (LOQ) are 0.0025% and 0.0084% in (w/w) unit, respectively. The sample recovery is 101.2%.
Keywords: Greener Way, Diethylene Glycol, DEG, Ethanol, Gas Chromatography-Mass Spectrometry,
Carryover
1. Introduction
Diethylene glycol (DEG, CAS number 111-46-6) is a
clear, colorless, practically odorless, viscous, and hydro-
scopic liquid with a sweet taste. It is miscible in water,
alcohol, and acetonitrile. In addition to its use in a wide
range of industrial products including dehydration of
natural gas, production of polyurethanes and unsaturated
polyester, antifreeze and brake fluids [1,2], DEG can be
formed by polymer degradation in biomaterials [3] and
other sources [4]. A number of DEG detection methods
have been developed over the years. Some methods fo-
cus on raw materials such as glycerin [5], ethylene, and
propylene [6-8]. Other methods are devoted to DEG de-
tection in finished products such as cosmetics [9], oral
care products [10-14], food [15] and pharmaceuticals
[16]. DEG detection methods in biological studies [15],
environmental polution samples [17] and phytoremedia-
tion [18] have also been reported. A common character-
istic of most DEG detection methods is that the sample
matrix must be predictable and the peak identity is not a
critical issue because matrix interference can be avoided
and removed by separation efforts. Method specific ap-
plications have been validated with a known individual
matrix. Difficulties arise when the product matrix is un-
known. As a result, the potential for peak interference
presents a significant challenge to method specificity.
The Food and Drug Administration (FDA) endorsed a
method for DEG detection in toothpaste [19]. This me-
thod provides a resolute ability to confirm the presence
of DEG in a toothpaste matrix, which could not only
quantify the DEG contents but also identify if any inter-
fering peaks are present. The method employs the use of
gas chromatography (GC) coupled to a mass spectros-
copy detector (MS). The GC provides the chroma-
tographic separation of the individual volatile compo-
nents which are then sent into the full scan MS for detec-
tion. The unique ability of the MS detector allows each
individual component to be identified and can distinguish
if any interfering peaks are present. The method has
helped the Colgate-Palmolive Company successfully
screen toothpaste products and provide brand protection
globally. However, several issues have been identified
with the method. Firstly, the solvents (acetonitrile and
water) used in the sampling procedure are not “friendly”
with respect to either the environment or the GC column.
Secondly and more importantly, the issues with DEG
carryover and its retention time shift become more and
more pronounced after numerous samples have been run
on the GC instruments. Finally, the acetonitrile shortage
between 2008 and 2010 significantly increased the
method operating cost to an unacceptable level. As part

Y. FU ET AL.939
of a new solvent strategy, it was decided to search for a
greener alternative to acetonitrile and without sacrificing
performance. Not only was a greener and more cost ef-
fective solvent (ethanol) identified, but also a more col-
umn friendly sample preparation. This new method has
been validated on dentifrice products for DEG and is
currently being used at Colgate-Palmolive Company.
2. Experimental
2.1. Materials and Reagents
Diethylene glycol (99%), anhydrous granule sodium sul-
fate (Na2SO4) and 710 - 1180 microns of glass beads
were purchased from Sigma-Aldrich (St. Louis, MO,
USA). 1,3-propanediol (99%) was from Alfa Aesar
(Ward Hill, MA, USA). 15-mL polypropylene centrifuge
tubes were obtained from VWR (West Chester, PA,
USA). ASTM type 1 water (at point of delivery) was
prepared internally from ELGA PURELAB prima 7/
Purelab Ultra Analytic (Lowell, MA, USA), acetonitrile
(HPLC grade) was purchased from JT Baker (Philips-
burg, NJ, USA) and anhydrous ethyl alcohol (ACS/USP
grade) was purchased from PHARMCO-AAPER (Brook-
field, CT, USA). The anhydrous ethyl ethanol needs to
be kept in seal until use.
2.2. Instrumentation
The Eppendorf centrifuge 5810R was used to centrifuge
the toothpaste sample for 10 min at 5000 g before the
GC-MS analysis. A Genie 2 vortex mixer was used to
assist sample dispersion in a minimum of time.
The gas chromatography system 5890 with 5972 MS
detector plus 7673 autosampler-split/splitless injector or
the gas chromatography system 6890 with 5973 MS se-
lective detector plus 6890 autosampler-split/splitless in-
jector (GC-EI-MS, Agilent Technologies, Santa, Clara,
CA, USA) was used for DEG analysis. Separation was
accomplished using a 30 m long Stabilwax capillary
column, 0.25 mm internal diameter (I.D.) and 0.25 m
film thickness (Restek, Bellefonte, PA, USA). The 1 L
sample was injected with the split mode at ratio of 1:20.
The oven temperature was initially held at 100˚C for 1
minute. Thereafter the temperature was raised at 10˚C
/min until 250˚C and held for 4 minutes. Helium was
used as the carrier gas and delivered at a constant flow
rate at 1 mL/min (the pressure at 8.2 psi and velocity at
37 cm/sec). The injector temperature was set at 250˚C
and the interface temperature was 250˚C. The MS de-
tectors were tuned with the standard spectrum autotune,
and the MS data (total ion chromatogram, TIC) were
acquired with either the full scan mode (m/z of 29 - 400
at a scan rate of 4 scan/sec or selected ion monitoring
(SIM) using the electron ionization (EI) mode for the
fragments at 31, 45 and 75 m/z with an electron energy
of 70 eV. The MS source temperature is 230˚C and quat
temperature is 150˚C. The retention time of DEG is
about 9.9 min, and the solvent delay is 4 minutes.
2.3. Blank, Standard and Internal Standard
Solutions
A blank solution which consists of 10 mL of extraction
solvent (either 50% acetonitrile-water or 98% ethanol-
water) taken through the entire procedure including the
addition of the internal standard was evaluated to make
sure that there was no contamination from the reagents
and containers.
Approximately 1.0 gram of DEG standard was
weighed into a 100 mL volumetric flask and dissolved
with a 50% aqueous acetonitrile or a 98% aqueous etha-
nol solvent, respectively. A series of standard solutions
with a 1:3 dilution from this standard stock solution were
made for method evaluation.
Approxiamtely 0.5 gram of internal standard, 1,3-
propanediol, was weighed into a 100 mL volumetric
flask and dissolved with a 50% aqueous acetonitrile or a
98% aqueous ethanol solvent, respectively. 0.1 mL of
this solution was added into 1 mL of each sample extract
or standard solutions in the autosampler vial prior to
GC-MS analysis. The fixed concentration is 0.05%. The
internal standard with 50% aqueous acetonitrile solvent
was for the original method, and the internal standard
with a 98% aqueous ethanol solvent was for the new
developed method.
2.4. QC and Spiking Sample Preparation
The QC samples were set up at 0.5 mg/mL and 0.1
mg/mL levels. The QC sample at 0.5 mg/mL of DEG
was analyzed at the beginning and the end of the sample
set to provide a basis for quantitative evaluation and to
monitor the amount of drift during the analysis of the set
of samples. QC sample at 0.1 mg/mL of DEG is ana-
lyzed at the center of the sample set for the threshold
level DEG detection.
The standard spiking solution was made from the stan-
dard stock solution with a 1:2 dilution, and 0.2 mL of
this standard spiking solution was directly added to the
representive uncontaminated sample before proceeding
with the sample preparation procedures.
2.5. Sample Preparation
For comparison purposes, both the original and new

Y. FU ET AL. 940
sampling procedures are described here.
2.5.1. Sampling Preparation for the New Developed
Method
Approximately 1.0 gram of toothpaste sample was
weighed into a 15 mL polypropylene centrifuge tube. To
wet the toothpaste materials, 0.2 mL of water was added.
To assist toothpaste sample suspension, 0.5 gram of glass
beads (710 - 1180 micros, Sigma-Aldrich , St. Louis,
MO, USA) were premixed well by using a Genie 2 vor-
tex mixer (Scientific Industrial, Inc., Bohemia, NY,
USA). After vortexing, 9.8 mL of ethanol and 2 grams of
anhydrous granule sodium sulfate was added; after
throrough mixing via vortex, the centrifuge was used for
10 minutes at 5000 g to isolate the supernatant. The su-
pernatant was filtrated by a syringeless filter device with
a 0.45 m PVDF membrance from Whatman (Clifton,
NJ, USA) for GC-MS analysis.
2.5.2. Sampling Preparation for the Original Method
Approximately 1.0 gram of toothpaste was weighed into
a 15-mL polypropylene centrifuge tube from VWR
(West Chester, PA, USA) and 5 mL of water was added.
Mix well to thoroughly disperse the entire sample. A
Genie 2 vortex mixer (Scientific Industrial, Inc., Bohe-
mia, NY, USA) was used to assist this process. After the
toothpaste sample was fully suspended, 5 mL of acetone-
trile in two portions with thorough mixing between each
addition was added. Centrifuge for 10 minutes at 5000 g.
The supernatant was filtrated by a syringeless filter de-
vice with a 0.45 m PVDF membrance from Whatman
(Clifton, NJ, USA) for GC-MS analysis.
3. Results and Discussion
3.1. The Specificity from the Original GC-MS
Method
An important component in toothpaste products is the
flavor, which can be composed of many different volatile
ingredients and produce potential interference to DEG
during the GC separation. For the known toothpaste
samples, all potential interference can be avoided with
GC separation efforts during the method development
because the matrix flavors are known. However, for un-
known toothpaste samples, the potential interference is
not predictable because the matrix flavors are not known.
Two extreme examples are presented in Figure 1 and the
interfering peaks can be located just before (Figure 1(a))
and after (Figure 1(b)) the DEG peak.
MS full scan mode is necessary to confirm if the un-
known samples contain DEG. In the practical application,
single ion monitoring (SIM) mode of MS detection can
(a)
(b)
Figure 1. Two typical GC-MS chromatograms of unknown
toothpaste spiked with DEG analytes. Figure 1(a) (top)
presents an interfering peak just before the DEG peak at
retention time 9.9 min and Figure 1(b) (bottom) presents an
interfering peak just after the DEG peak at retention time
9.8 min. (Note that DEG peak retention time was shifted
after numerous sample analyses).
also be used to screen for DEG in the toothpaste. The
SIM mode can provide better sensitivity, but the peak
identity is not reliable even when the ratio of the indi-
vidual fragment from SIM mode is monitored. Figure
2(a) exhibits a MS spectrum with a full scan mode detec-
tion of the DEG standard, which provided 90% match
quality to the NIST 2002 database. Figure 2(b) displays
a MS spectrum with a SIM mode detection and the indi-
vidual fragment ratios at m/z 31, 45, and 75 are right
when the DEG concentration is above the LOQ level of
full scan detection. When DEG cencetration reached the
LOQ levels in the full scan mode, the ratios at m/z of 31,
45, and 75 started to be twisted, which is presented in
Figure 2(c).

Y. FU ET AL.941
(a)
(b)
(c)
Figure 2. DEG mass spectra obtained from full scan and
single ion monitroing (SIM) modes. 2(a) (top) was from full
scan detection; 2(b) (middle); and 2(c) (bottom) were from
SIM detections.
3.2. The Challenges of the Original GC-MS
Method
Toothpaste materials cannot be dissolved or suspended
well in organic solvents such as polar methanol used
during the sampling procedure. Water is usually a good
media for toothpaste sampling. However, water is a
less-than-ideal solvent from a GC point of view. The
problems associated with water include a large vapor
expansion volume and perceived chemical damage to the
stationary phase [20]. Based on the solvent expansion
calculator software provided by Agilent for GC instru-
ment, the approximate vapor volume of 50% acetone-
trile-water solvent under the flow rate of 1 mL/min (pres-
sure approximately 8.2 psi) can be larger than 1000 L,
which surpasses the volume of all commercially avaiable
GC injection liners. Therefore, the DEG carryover was
observed after two blank injection intervals during sub-
sequent sample or standard injections. Carryover is de-
fined as the appearance of a compound in a blank sample,
especially when the blank sample is injected immediately
after an injection of a sample or standard containing high
concentration of analytes, which can be a result of the
backflash phenomenon [20]. In addition, the boiling
point of DEG is 244˚C - 245˚C which requires a terminal
temperature of 250˚C in the GC oven program. Serious
column bleeding can be observed in Figure 1 due to the
high ratio of water in the injected solvent. DEG retention
time continuously dropped after a larger number of sam-
ple injections. To improve the original FDA method [19],
a new sampling procedure was developed in this study.
The water volume in the revised sampling procedures
has been limited to 200 L (2% of total sampling sol-
vent). Water is required to wet the toothpase matrix, and
many toothpaste samples cannot be suspended without
using the water. To improve the sampling solubility and
suspension, a solvent with a stronger hydrogen-bonding
capability, ethanol, was utilized to replace the acetone-
trile. Ethanol is a cheaper and more environmentally-
friendly solvent compared to acetonitrile. To enhance
toothpaste sample suspension, glass beads (710 - 1180
microns size) and anhydrous granule sodium sulfate were
applied. Anhydrous sodium sulfate provides two critical
functions, it mechanically enhances the toothpaste sus-
pension, and it removes the water from the ethanol phase
[21]. All the tested toothpaste samples showed good
suspension with the assistance of a Genie 2 vortex mixer
(Scientific Industrial, Inc., Bohemia, NY, USA).
3.3. Method Evaluation
To evaluate the solvent replacement, several method
validation parameters have been measured and compared

Y. FU ET AL. 942
between the extraction solvents of 50% aqueous acetone-
trile and 98% aqueous ethanol. A typical chromatogram
of DEG with the improved ethanol extraction procedure
is shown in Figure 3. All the data were obtained from
the newer instrument, the gas chromatography system
6890 with 5973 MS selective detector plus 6890 autos-
mapler-split/splitless injector.
3.3.1. Linearity of Response, Limits of Detection
(LOD) and Quantit ation (LOQ)
The calibration responses were obtained by plotting the
peak area ratio between the DEG standard concentration
range of 0.16% down to 0.01% and 1,3-propanediol at a
fixed concentration of 0.05%. As shown in Table 1,
good linearities were obtained for both 98% aqueous
ethanol and 50% aqueous acetonitrile.
The LODs and LOQs by GC-MS in the full scan mode
are shown in Table 1. The LOD and LOQ with 50%
aqueous acetonitrile extraction are 0.0044% and 0.0146%,
respectively. The better LOD and LOQ with 98% aque-
ous ethanol extraction are 0.0025% and 0.0084%. The
sensitivity of the LOD and LOQ can be significantly im-
proved with the SIM mode detection, but this is not suit-
able for screening unknown toothpaste samples for DEG.
Figure 3. A typical GC-MS chromatogram of DEG with the
improved ethanol extraction proce dur e .
Table 1. Calibration parameters and sample recovery.
Sampling
solvents R2 LOD LOQ
RSD
%
Recovery
%
50% aqueous
ACN 0.9999 0.0044% 0.0146% 12.4% 87.5%
98% aqueous
EtOH 1.0000 0.0025% 0.0084% 6.7% 101.2%
3.3.2. Method Accuracy and Precision
A toothpaste without fluoride or DEG (Grins & Giggles,
by Gerber Product Company, Fremont, MI, USA) was
used as the spiking matrix to perform the analyte recov-
ery for accuracy evaluation. The recovery values are the
results from five injections of spiked standards in matrix
following the entire procedure from sample suspension,
centrifuging and GC-MS analysis.
The precision, as given by relative error (RE%) and
relative standard deviation (RSD%), respectively, was
evaluated (Table 1) by analyzing five replicates of DEG
standard at 0.01%.
3.3.3. Carryo v er
To evaluate the carryover effect, a blank sample was set
next to the upper limit of quantification control sample
(QC sample at 0.5 mg/mL). A significant carryover peak
of DEG was observed with 50% aqueous acetonitrile
extraction procedure but no peak was present after the
extraction solvent was replaced by 98% aqueous ethanol
solvent.
4. Conclusions
When the sampling solvent was switched from 50% ace-
tonitrile-water to 98% ethanol-water, all the validation
data from carryover, linearity, sensitivity, precision to
accuracy exhibited a positive improvement. Ethanol is
not only more cost effective than acetonitrile but also
more friendly (greener) to the environment. Moreover,
column life with this newly developed method is longer
than with the original method.
5. Acknowledgements
Some experimental procedures and experimental condi-
tions are copied from the original FDA GC-MS method,
and we would like to thank its authors, Jonathan Lizau
and Kevin Mulligan.
6. References
[1] R. J. Lewis and S. Hawley, “Condensed Chemical Dic-
tionary,” Van Nostrand Reinhold Company, New York,
1993.
[2] J. M. Marraffa, M. G. Holland, C. M. Stork, C. D. Hoy
and M. J. Hodgman, “Diethylene Glycol: Widely Used
Solvent Present Serious Poisoning Potential,” The Jour-
nal of Emergency Medicine, Vol. 35, No. 4, 2008, pp.
401-406. doi:10.1016/j.jemermed.2007.06.025
[3] A. Ghaffara, P. G. Verschuren, J. A. J. Geenevasen, T.
Handels, J. Berard, B. Plum, A. A. Dias, P. J. Schoen-
makers and Sj. van der Wal, “Fast in vitro Hydrolytic
Degradation of Polyester Urethane Acrylate Biomaterials:

Y. FU ET AL. 943
Structure Elucidation, Separation and Quantification of
Degradation Products,” Journal of Chromatography A,
Vol. 1218, No. 3, 2011, pp. 449-458.
doi:10.1016/j.chroma.2010.11.053
[4] G. Stefania, B. Maura, V. L. Claudia, P. Barbara, M.
Ginevra and R. Francesco, “Biological Effects of Diethyl-
ene Glycol (DEG) and Produced Waters (PWs) Released
from Offshore Activities: A Multi-Biomarker Approach
with the Sea Bass Dicentrarchus labrax,” Environmental
Pollution, Vol. 157, No. 11, 2009, pp. 3166-3173.
doi:10.1016/j.envpol.2009.05.021
[5] C. M. Gryniewicz, J. A. Spencer, M. Hankins and J. F.
Kauffman, “Spectroscopic Methods for Rapid Determina-
tion of Diethylene Glycol in Glycerin,” American Phar-
maceutical Review, Vol. 10, 2007, pp. 1-4.
[6] G. Holloway, R. Maheswaran, A. Leeks, S. Bradby and S.
Wahab, “Screening Method for Ethylene Glycol and Di-
ethylene Glycol in Glycerin-Containing Products,” Jour-
nal of Pharmaceutical and Biomedical Analysis, Vol. 51,
No. 3, 2010, pp. 507-511.
doi:10.1016/j.jpba.2009.08.025
[7] X. Li, S. Arzhantsev, J. F. Kauffman and J. A. Spencer,
“Detection of Diethylene Glycol Adulteration in Propyl-
ene Glycol—Method Validation through a
Multi-Instrument Collaborative Study,” Journal of
Pharmaceutical and Biomedical Analysis, Vol. 54, No. 5,
2011, pp. 1001- 1006. doi:10.1016/j.jpba.2010.11.042
[8] ASTM E611-08 in ASTM, “American Society for Test-
ing and Matericals, Standard Test Methods for Low
Concentrations of Diethlyene Glycol in Ethylene Glycol
by Gas Chromatography,” Annual book of ASTM stan-
dard, Philadelphia.
[9] K. Molever, “Simplfied Assay of Diethylene Glycol and
Ethylene Glycol in Various Raw Materials by Capillary
Gas Chromatography,” Society of Cosmetic Chemists,
Vol. 61, 2010, pp. 225-234.
[10] A. Kenyon, X. Shi, Y. Wang, W. H. Ng, “Simple on-Site
Detection of Diethylene Glycol/Ethylene Glycol Con-
tamination of Glycerin and Glycerin-Based Raw Materi-
als by Thin-Layer Chromatography,” Journal of AOAC
International, Vol. 81, 1998, pp. 44-50.
[11] J. Ding, H. Gu, S. Yang, M. Li, J. Li and H. Chen, “Se-
lective Detection of Diethylene Glycol in Toothpaste
Prod- ucts Using Neutral Desorption Reactive Extractive
Electros- pray Ionization Tandem Mass Spectrometry,”
Analytical Chemistry, Vol. 81, No. 20, 2009, pp. 8632-
8638. doi:10.1021/ac9013594
[12] F. Hernandez, M. Ibanez and J. V. Sancho, “Fast Deter-
mination of Toxic Diethylene Glycol in Toothpaste by
Ultra-Performance Liquid Chromatography-Time of
Flight Mass Spectrometry,” Analytical and Bioanalytical
Chemistry, Vol. 391, No. 3, 2008, pp. 1021-1027.
doi:10.1007/s00216-008-2082-0
[13] M. Lopez-Sanchez, A. Dominguez-Vidal, M. J. Ayora-
Canada and A. Molina-Diaz, “Assessment of Dentifrice
Adulteration with Diethylene Glycol by Means of ATR-
FTIR Spectroscopy and Chemometrics,” Analytica
Chimica Acta, Vol. 620, No. 1-2, 2008, pp. 113-119.
doi:10.1016/j.aca.2008.05.032
[14] X. Xing, X. Shi, H. Zhang, W. Wang and J. Ye, “Deter-
mination of Diethylene Glycol in Toothpaste by Capillary
Electrophoresis with Electrochemical Detection,” Micro-
chimica Acta, Vol. 167, 2009, pp. 297-302.
[15] A. A. Rahim, B. Saad, H. Osman, N. Hashim, S. Yahya
and K. M. Talib, “Simultaneous Determination of Di-
ethylene Glycol, Diethylene Glycol Monoethyl Ether,
Coumarin and Caffeine in Food Items by Gas Chroma-
tography,” Food Chemistry, Vol. 126, No. 3, 2011, pp.
1412-1416. doi:10.1016/j.foodchem.2010.11.121
[16] T. Zhou, H. Zhang and G. Duan, “Simultaneous Determi-
nation of Diethylene Glycol and Propylene Glycol in
Pharmaceutical Products by HPLC after Precolumn Deri-
vatization with P-Toluenesulfonyl Isocyanate,” Journal
of Separation Science, Vol. 30, No. 16, 2007, pp. 2620-
2627. doi:10.1002/jssc.200700097
[17] A. Cappiello, G. Famiglinia, P. Palma, V. Termopolia, H.
Trufelli , R. Di Mentob and M. Mannozzi, “LC-ESI-MS
Determination of Diethylene Glycol Pollution in Sea
Water Samples Collected around Gas Extraction Platform
Plants,” Talanta, Vol. 80, No. 1, 2009, pp. 257-262.
doi:10.1016/j.talanta.2009.06.067
[18] P. Teamkao and P. Thiravetyan. “Phytoremediation of
Ethylene Glycol and Its Derivatives by the Burhead Plant
(Echinodorus cordifolius (L.): Effect of Molecular Size,”
Chemosphere, Vol. 81, No. 9, 2010, pp. 1069-1074.
doi:10.1016/j.chemosphere.2010.09.049
[19] http://www.fda.gov/Food/ScienceResearch/LaboratoryM
ethods/DrugChemicalResiduesMethodology/ucm113209.
htm
[20] E. Kuhn, “Water Injections in GC―How Wet Can You
Get?” LCGC North America, Vol. 20, 2002, pp. 474-478.
[21] F. J. Schenck, P. Callery, P. M. Gannett, J. R. Daft and S.
J. Lehotay, “Comparison of Magnesium Sulfate and So-
dium Sulfate for Removal of Water from Pesticide Ex-
tracts of Foods,” Journal of AOAC International, Vol. 85,
2002, pp. 1177-1180.