Methods for detecting silicones in biological matrixes

Article abstract of DOI:10.1021/ac990157d

Methods for analyzing for silicon and silicone in biological matrixes were developed. A silicone-specific technique involved microwave digestion of samples in acid solution to rapidly break down the biological matrix while hydrolyzing silicones to monomeric species. The resulting monomeric silanol species were then capped with trimethylsilyl groups, extracted into hexamethyldisiloxane, and analyzed by gas chromatography. In serum, positive identification of silicone species with detection limits below 0.5 μg of Si/mL are possible with this technique. The technique is compared with a silicone-specific technique, ²⁹Si NMR, and a non-silicone-specific technique, ICP-AES. ²⁹Si NMR was far less sensitive, with a detection limit of only 64 μg of Si/mL in serum when analyzing for one compound with a single sharp resonance. Inductively coupled plasma-atomic emission spectroscopy (ICP-AES) has potentially lower detection limits, but the technique is not silicone-specific and suffers from species-dependent responses.

Full text of DOI:10.1021/ac990157d

Anal. Chem. 1999, 71, 3054-3060
Me t h o d s fo r De t e c t in g S ilic o n e s in Bio lo g ic a l
Ma t rix e s
J ohn J . Ke nna n,* La urie L. Mc Ca nn Bre e n, Thom a s H. La ne , a nd Ric ha rd B. Ta ylor
Dow Corning Corporation, Midland, Michigan 48686-0994
hemodialysis exhibit significantly elevated serum and urine silicon
Methods for analyzing for silicon and silicone in biological
matrixes were developed. A silicone-specific technique
involved microwave digestion of samples in acid solution
to rapidly break down the biological matrix while hydro-
lyzing silicones to monomeric species. The resulting
monomeric silanol species were then capped with tri-
methylsilyl groups, extracted into hexamethyldisiloxane,
and analyzed by gas chromatography. In serum, positive
identification of silicone species with detection limits
below 0 .5 µg of Si/ mL are possible with this technique.
The technique is compared with a silicone-specific tech-
nique, 2 9 Si NMR, and a non-silicone-specific technique,
ICP -AES. 29Si NMR was far less sensitive, with a detection
limit of only 6 4 µg of Si/ mL in serum when analyzing for
one compound with a single sharp resonance. Inductively
coupled plasma-atomic emission spectroscopy (ICP -AES)
has potentially lower detection limits, but the technique
is not silicone-specific and suffers from species-dependent
responses.
concentrations.3 These are just a few of the factors that make
comparisons based purely on silicon levels difficult.
The ubiquitous presence of silicon also introduces silicon
analysis problems. As shown in Table 1, reported silicon concen-
trations for normal blood and serum vary over a considerable
range.3,4,5,6,7,8,9 At least part of the variability may be due to
contamination from environmental sources of silicon or silicone.
Dust is a source of silicon contamination, and silicone may be
introduced from the surface of laboratory items such as plas-
ticware and syringes. In addition, human contact may introduce
contamination from silicone that is present in many personal care
products. Great care must be taken to avoid sample contamination.
The importance of avoiding sample contamination may be
reflected in the fact that the lowest normal blood-silicon levels
were reported in ref 9, a work in which the authors acquired blood
samples using silicone-free syringes and processed samples under
class 100 clean room conditions to avoid contamination.
Atomic spectroscopic techniques have also been used for
analyzing biological tissues. Typically these analyses involve acid
digestion followed by analysis for total silicon.10,11 These methods
cannot distinguish silicones from inorganic silicates in tissue.
Extraction techniques have been used in an attempt to preferen-
tially extract organosilicon compounds into organic solvents prior
to the analysis.12 However, even this approach cannot preclude
the possibility of extracting complexed silicates.
Because of the increased use of silicones as biomedical
materials, there has been interest in developing methods for
detecting silicone in biological matrixes. Unfortunately, many of
the methods suffer serious limitations. Atomic spectroscopy
techniques such as inductively coupled plasma-atomic emission
spectroscopy (ICP-AES) or atomic absorption (AA), are not
silicone-specific, but rather analyze for silicon as a surrogate for
silicone. With proper controls, this approach is useful, however,
it is important to realize that silicon is naturally present in tissues
and silicon levels can be influenced by factors other than the
presence of silicone. Silicon intake in a typical U. S. diet has been
estimated as 20-50 mg per day, but can range substantially
higher.1 The consequences of high silicon intakes can be dramatic.
Silicon intakes exceeding 1400 mg/ day from ingestion of mag-
nesium trisilicate antacids were linked to the formation of silica
stones.1 Other factors can also affect comparisons based purely
on silicon levels. For example, silicon output in urine varies with
the time of day.2 It has also been found that patients undergoing
Trace silicone analysis is difficult. Many of the potential
problems surrounding analyte losses, contamination, sample
preparation, and sample analysis have been enumerated in
reviews.13-15 Losses may occur as a result of strong adsorption of
silicones at surfaces or because of volatilization of low molecular
(3) Roberts, N. B.; Williams, P. Clin. Chem. (Washington, D.C.) 1990, 36, 1460-
1465.
(4) Sun, L.; Ricci, J. L.; Klein, A.; Alexander, H.; Lattarulo, N.; Blumenthal, N.
C. Abstracts of Papers, Fifth World Biomaterials Congress, Toronto, Canada,
1996; University of Toronto Press: Toronto, Canada, 1996; Vol. 1, p 305.
(5) King, E. J.; Stacy, B. D.; Holt, P. F.; Yates, D. M.; Pickles, D. Analyst
(Cambridge, U.K.) 1 9 5 5 , 80, 441-453.
(6) Bercowy, G. M.; Vo, H.; Rieders, F. J. Anal. Toxicol. 1 9 9 4 , 18, 46-48.
(7) Gitelman, H. J.; Alderman, F. R. J. Anal. At. Spectrom. 1 9 9 0 , 5, 687-689.
(8) Teuber, S. S.; Saunders: R. L.; Halpern, G. M., Brucker, R. F.; Conte, V.;
Goldman, B. D.; Winger, E. E.; Wood, W. G.; Gershwin, M. E. Biol. Trace
Elem. Res. 1 9 9 5 , 48, 121-130.
* Corresponding author: (tel.) 517-496-6409; (fax) 517-496-6824; (e-mail)
john.j.kennan@dowcorning.com
(1) Pennington, J. A. T. Food Addit. Contam. 1 9 9 1 , 8, 97-118.
(2) Kanabrocki, E. L.; Sothern, R. B.; Scheving, L. E.; Vesely, D. L.; Tsai, T. H.;
Shelstad, J.; Cournoyer, C.; Greco, J.; Mermall, H.; Ferlin, H.; Nemchausky,
B. M.; Bushnell, D. L.; Kaplan, E.; Kahn, S.; Augustine, G.; Holmes, E.;
Rumbyrt, J.; Sturtevant, R. P.; Sturtevant, F.; Bremner, F.; Third, J. L. H. C.;
McCormick, J. B.; Dawson, S.; Sackett-Lundeen, L.; Haus, E.; Halberg, F.;
Pauly, J. E.; Olwin, J. H. Chronobiol. Int. 1 9 9 0 , 7, 445-461.
(9) Peters, W.; Smith, S.; Lugowski, S.; McHugh, A.; Baines, C. Ann. Plast. Surg.
1 9 9 5 , 34, 343-347.
(10) Hornung, M.; Krivan, M. A J. Anal. At. Spectrom. 1 9 9 7 , 12, 1123-1130.
(11) Schnur, P. L.; Weinzweig, J.; Moyer, T. P.; Nixon, D.; Harris, J. B.; Petty, P.
M. Plast. Reconstr. Surg. 1 9 9 6 , 98, 798-803.
(12) Barnard, J. J.; Todd, E. L.; Wilson, W. G.; Mielcarek, R.; Rohrich, R. J. Plast.
Reconstr. Surg. 1 9 9 7 , 100, 197-203.
3054 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
10.1021/ac990157d CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/05/1999

ratios in their spectra could not be accurately determined.21 While
short of a retraction, it does raise questions concerning the utility
of 29Si NMR, a technique with intrinsically low sensitivity compared
with those of other techniques. Raman and FT-IR microscopy
have been used to identify silicone in capsular tissue and in
regional lymph nodes of women with implants.22,23,24,25 A limitation
of IR and Raman spectroscopy is that, as applied, they do not give
quantitative information.
Extraction in conjunction with gas chromatographic techniques
was used to follow migration of low molecular weight cyclics (D3
to D7) following injection in mice.26 Silicones were identified by
use of mass spectral or atomic emission detectors. Whereas these
techniques unequivocally identified various cyclics in tissue, the
injected doses of approximately 10 g/ kg of body weight would
be equivalent to injecting a human with half a kilogram of cyclics.
Since medical devices consist of polymeric silicones with only
traces of the volatile low molecular weight cyclics, the extraction
and GC techniques have limited practical utility.
A technique which can specifically detect dimethylsiloxane
oligomers and polymers is known in the silicone industry as the
Aqueous Silanol Functionality Test (ASFT).27 In this procedure,
an aqueous sample containing silicone species is digested by
acidifying to 10-15 wt % HCl and shaking for 1-7 days. During
this period, silicones are hydrolyzed to the monomeric units:
(CH3)2Si(OH)2, CH3Si(OH)3, and Si(OH)4. A small amount, typi-
cally 2 mL, of hexamethyldisiloxane (MM) is then added to the
digested sample, and the sample is shaken an additional 1-7 days.
During this time, all silanol species are capped with trimethylsilyl
groups and extracted into the hexamethyldisiloxane layer. Thus,
(CH3)2Si(OH)2, CH3Si(OH)3, and Si(OH)4 are converted to MDM,
M3T, and M4Q, respectively. The hexamethyldisiloxane layer is
then isolated and analyzed by GC versus appropriate standards.
The technique is capable of detecting sub ppm levels of (CH3)2-
SiO in aqueous samples.
The great sensitivity of the ASFT is attributed to several factors.
Capping silanols with trimethylsilyl groups eliminates the tendency
for the analytes to undergo condensation reactions in the GC. Also,
capping results in a significant mass increase, which greatly
enhances sensitivity when using a GC/ FID detector. For example,
in the case of the D unit, capping increases the mass from 74 to
236. Lastly, depending on the ratio of the original sample to the
hexamethyldisiloxane phase, significant concentration enhance-
ments are possible by extraction of the analytes of interest into a
smaller volume. Another great advantage to the technique is that,
if the hydrolysis can be run under conditions where methyl
Ta b le 1 . Re p o rt e d No rm a l S ilic o n Le ve ls in Blo o d o r
Blo o d Co m p o n e n t s
matrix mean µg of Si/ mL
technique
ref
blood
blood
blood
plasma
plasma
serum
blood
6.24
1.29
0.5a
0.17
0.14
0.13
0.0250
AA
4
5
6
7
3
8
9
digestion/ colorimetric analysis
DC-PES
GF-AAS
DC-PES
ICP-AES
GF-AAS
a
Approximate median.
weight species. Losses may also occur during sample digestion
procedures as a result of the volatility of low molecular weight
silicones or silanes that may be generated during acid digestions.
Instrumental concerns include the materials of construction of a
spectrometer which can cause problems through interaction with
silicon compounds or leaching of silicon compounds as contami-
nates.
The focus of this paper is the detection of silicones. Poly-
(dimethylsiloxane) (PDMS) is by far the most common form of
silicone produced commercially, and it is also the form of silicone
most widely used in medical applications. Therefore, any test for
detecting silicones in biological tissues should be sensitive to the
dimethylsiloxane moiety. In discussing such tests, it is convenient
to use a silicone shorthand notation that is commonly used in
the industry. In this notation, M, D, T, and Q denote (CH3)3SiO1/ 2
,
(CH3)2SiO, CH3SiO3/ 2, and SiO4/ 2 respectively. Thus, MDM, M3T,
and M4Q are short for (CH3)3SiO(CH3)2SiOSi(CH3)3, ((CH3)3-
SiO)3Si(CH3), and ((CH3)3SiO)4Si, respectively. Using this nota-
tion, the common trimethylsilyl-terminated poly(dimethylsiloxane)
would be denoted as MDnM.
Radiolabeling methyl groups of poly(dimethylsiloxane) with
14C provides a very sensitive and specific tag for detecting silicone
in tissues. Silicone labeled with 14C was used to monitor silicone
migration from gels implanted in rats.16 A technique involving
radiolabeled octamethylcyclotetrasiloxane (D4) has been devel-
oped for use in distribution and metabolism investigations.17
Obviously, the technique is only useful if the silicone source has
been labeled with an appropriate tag.
Silicone-specific spectroscopic techniques such as 29Si NMR,
FT-IR, and Raman have been used to detect silicone in biological
tissues. 29Si NMR was used to detect silicone degradation products
in the blood of women with implants.18 However, reported silicon
levels were 4-5 orders of magnitude higher than levels reported
by atomic spectroscopy techniques. As a result, the reported high
levels in blood have been challenged,19,20 and the original authors
have since published an erratum stating that the signal-to-noise
(19) Macdonald, P.; Plavac, N.; Peters, W.; Lugowski, S.; Smith, D. Anal. Chem.
1 9 9 5 , 67, 3799-3801.
(20) Taylor, R. B.; Kennan, J. J. Magn. Res. Med. 1 9 9 6 , 36, 498-501.
(21) Garrido, L.; Pfleiderer, B.; Jenkins, B. G.; Hulka, C. A.; and Kopans, D. B.
Magn. Reson. Med. 1 9 9 8 , 39, 689.
(13) Lugowski, S. J.; Smith, D. C.; Lugowski, J. Z.; Peters, W.; Semple, J. Fresenius’
J. Anal. Chem. 1 9 9 8 , 360, 486-488.
(22) Frank, C. J.; McCreery, R. L.; Redd, D. C. B.; Gansler, T. S. Appl. Spectrosc.
1 9 9 3 , 47, 387-390.
(14) Smith, A. L.; Parker, R. D. In The Analytical Chemistry of Silicones; Smith,
A. L., Ed.; Chemical Analysis 112; Wiley: New York, 1991; pp 71-95.
(15) Cavic-Vlasak, B. A.; Thompson, M.; Smith, D. C. Analyst (Cambridge, U.K.)
1 9 9 6 , 1121, 53R-63R.
(23) Hardt, N. S.; Yu, L. T.; La Torre, G.; Steinbach, B. Mod. Pathology 1 9 9 4 , 7,
669-676.
(24) Kidder, L. H.; Kalasinsky, V. F.; Luke, J. L.; Levin, I. W.; Lewis, E. N. Nat.
Med. (NY) 1 9 9 7 , 3, 235-237.
(16) Schulz, C. O.; Lee, G.; Mathews, J. M. Toxicologist, 1 9 9 3 , 13 (1), 381;
(25) Ali, S. R.; Johnson, F. B.; Luke, J. L.; Kalasinsky, V. F. Cell. Mol. Biol. 1 9 9 8 ,
44, 75-80.
Abstract 1491.
(17) Varaprath, S.; Salyers, K. L.; Plotzke, K. P.; Nanavati, S. Anal. Biochem.
1 9 9 8 , 256 (1), 14-22.
(26) Kala, S. V.; Lykissa, E. D.; Neely, M. W.; Lieberman, M. W. Am. J. Pathol.
1 9 9 8 , 152, 645-649.
(18) Garrido, L.; Pfleiderer, B.; Jenkins, B. G.; Hulka, C. A.; Kopans, D. B. Magn.
(27) Mahone, L. G.; Garner, P. J.; Buch, R. R.; Lane, T. H.; Tatera, J. F.; Smith,
R. C.; Frye, C. L. Environ. Toxicol. Chem. 1 9 8 3 , 2, 307-313.
Reson. Med. 1 9 9 4 , 31, 328-330.
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999 3055

cleavage does not occur, then it is possible to determine the type
of units that were originally present in the polymer, i.e., the
absolute concentrations of D, T, and Q species.
with an electronic reflux ratio controller. Deionized water with a
resistivity of 19.2 MΩ-cm was used.
ICP Analysis. Analyses were carried out using a Fisons
Accuris ICP system 20412 equipped with a low-power, low-flow
torch design. Samples were introduced at a controlled flow rate
using a peristaltic pump. Background-corrected intensities were
determined by subtracting off-peak intensities from the raw peak
intensity at 288.16 nm.
The technique does have a few drawbacks. Detection of other
functional groups on silicon requires that they be stable under
the hydrolysis conditions and that the analyst has appropriate
standards to detect and quantify the capped species. Also, it is
not possible to detect Me3SiOH unless a capping agent other than
the trimethylsilyl group is used. For most analyses, these are not
considered significant problems because the dimethylsiloxane unit
comprises the bulk of commercial silicones. The analyst must
always be wary of the possibility of methyl cleavage. Although
methyl cleavage was not observed in carrying out the standard
ASFT on spiked aqueous samples,27 it is essential to run appropri-
ate recovery experiments to ensure adequate recovery of the
analyte and to ascertain whether methyl cleavage occurs.
Unfortunately, the ASFT does not work as well with biological
samples. Proteins precipitate upon addition of the HCl, and the
solids may interfere with the hydrolysis of entrapped silicones.
In addition, some proteins are extracted into the hexamethyld-
isiloxane layer, leading to formation of gels. The gels can be
dissolved in methanol for analysis, but dilution results in a loss
in sensitivity.
A modified version of the ASFT was published for analysis of
biological tissues.28 In this variation, capsular tissue from a woman
with breast implants was first digested in alcoholic KOH. The
sample digestion was completed in strong acid, followed by
capping with hexamethyldisiloxane. Presumably, the more rigor-
ous digestion technique hydrolyzes proteins to water-soluble
amino acids, thus eliminating both the formation of solids and
the extraction of proteins into the hexamethyldisiloxane layer. The
disadvantage of the technique is the need to digest sequentially
in strong base and acid. Detection limits for this technique were
estimated to be 6 µg ((CH3)2SiO)/ g of tissue; however, recoveries
for the technique were not reported.
29Si NMR. NMR spectra were acquired on a Varian Unity Plus
400 MHz spectrometer equipped with a 16-mm Nalorac probe
capable of 850:1 s/ n ratio in a single scan for 13C on the ASTM
standard, 60% d6-benzene/ 40% p-dioxane. Signal-to-noise determi-
nations were performed on a 0.0312% solution of (CH3)2Si(OH)2
in pig serum. Spectra were acquired with 256 scans in 17 h and
10 min using the following parameters: 240-s pulse delay, 90°
pulse width, D2O lock reference, and 1.5 Hz of line broadening
1
with H decoupling applied during the acquisition period only to
avoid nuclear Overhauser effects.
Synthesis of Dimethylsilanediol. A procedure originally
developed by Compton was used.29 (CH3)2Si(OCH3)2 (26.60 g) and
acetic anhydride (2 µL) were weighed into a three-neck 100-mL
flask outfitted with a stir bar, addition funnel, thermometer, and
short path distillation head. While stirring vigorously, 15.9 g of
deionized water was added to the flask. An immediate exotherm
was noted, with the pot temperature rising to 40 °C. After stirring
15 min, the reaction mixture was vacuum-stripped to remove
methanol. A total of 16.73 g of crude (CH3)2Si(OH)2 was isolated
(82% yield). Crude (CH3)2Si(OH)2 was recrystallized from acetone,
washed with cold pentane, and then dried in a vacuum desiccator
(15 min at 0.01 Torr). The product consisted of white platelets
(mp 94.5-97 °C).
Microwave Digestion. Microwave digestions were carried out
using a CEM model MDS-2100 microwave digestion apparatus
fitted with as many as twelve 100-mL Teflon-lined pressure vessels.
Temperature or pressure was controlled or monitored through
sensors attached to a control vessel. In some early experiments
involving HCl digestions of spiked sera, carryover contamination
was evident. The problem was eliminated by washing the digestion
vessels and centrifuge tubes with detergent. In addition, the
digestion vessels were subjected to a microwave digestion treat-
ment with 35% HNO3 to remove residues. The vessels and
centrifuge tubes were then rinsed with deionized water and
acetone.
This paper describes the detection of silicones in biological
matrixes using silicon- and silicone-specific techniques. The intent
is to alert researchers as to some of the pitfalls and limitations of
techniques such as ICP-AES and 29Si NMR. In addition, a modified
version of the ASFT technique will be presented which allows
for faster sample throughput and routine analysis of biological
samples.
Nitric Acid Digestion of Spiked P ig Serum. Teflon pressure
vessels outfitted with rupture disks were loaded with 0-20 µL of
either 350 cst. PDMS in THF or sodium silicate in water. To each
vessel, 1.5 mL of pig serum and 8 mL of 70% nitric acid were
added. Samples were digested using a five-step pressure program
where the pressure was ramped to 10, 20, 30, 50, and finally 75
PSI with 5-min holds at each pressure. Pressure vessels were
cooled in dry ice, vented, and the solutions then transferred to
30-mL Teflon centrifuge tubes. Transfer was facilitated by rinsing
the pressure vessels with two 3-mL portions of a 2% boric acid
solution. The boric acid served to significantly lower background
intensities in the ICP-AES when analyzing solutions of strong acid.
Samples were analyzed by ICP-AES against sodium silicate
standards in solutions containing 37% HNO3 and 0.66% boric acid.
EXPERIMENTAL SECTION
Reagents. Methyltris(trimethylsiloxy)silane (Gelest), tetrakis-
(trimethylsiloxy)silane (Gelest), dimethyldimethoxysilane (Dow
Corning), methylene chloride, pig serum (Sigma Aldrich), 34%
hydrochloric acid (Fisher Optima Grade), 70% nitric acid (Fisher
Optima Grade), 350 cst. poly(dimethylsiloxane) (Dow Corning 200
fluid), 100 cst. poly(dimethylsiloxane) (Dow Corning 360 fluid),
and pentadecane (Aldrich) were used as received. National
Institute of Standards and Technology (NIST) traceable sodium
silicate standards were obtained from SPEX CertiPrep, Inc.
Hexamethyldisiloxane (Dow Corning OS-10) was treated with
activated charcoal, was filtered, and then a center cut was obtained
from distillation on a glass helices packed 36" column outfitted
(28) Baker, J. L.; LeVier, R. R.; Spielvogel, D. E. Plast. Reconstr. Surg. 1 9 8 2 , 69
(1), 56-60.
(29) Compton, R. A., Dow Corning Corporation, unpublished results, 1964.
3056 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

Ta b le 2 . Re s p o n s e Fa c t o rs fo r An a lys is o f S ila n e s in
S e ru m
compound
response factora
NIST silicate
MeSi(OCH3)3
(CH3)2Si(OH)2
(CH3)3SiOH
1.00
0.894
0.890
0.060
a
Corrected ICP intensity ) response factor × intensity.
Aqueous Silanol Functionality Test. A typical analysis of a
sample spiked with PDMS was carried out as follows. Teflon
pressure vessels outfitted with rupture disks were loaded with
0-50 µL of a silicone solution in CH2Cl2. The CH2Cl2 was
permitted to evaporate, after which the vessel was loaded with
0.25 mL of serum, 14.50 mL of deionized water, and 5.25 mL of
concentrated HCl (34 wt. % HCl). Sealed vessels were subjected
to a microwave program involving a 10-min ramp to 140 °C
followed by a temperature hold for 2 h. Samples were transferred
to 30-mL Teflon centrifuge tubes. To each sample was added 2
mL of hexamethyldisiloxane containing 100 µg/ mL pentadecane.
The samples were tightly capped, shaken for 24 h on a Burrell
Scientific Model 75 wrist shaker, and centrifuged at 3000 rpm for
30 min in a Beckman Model GS-6 centrifuge. The hexamethyl-
disiloxane layer was then analyzed for MDM, M3T, and M4Q by
GC.
Gas Chromatography. Samples were analyzed on a HP 6890
GC outfitted with an autosampler, a 30 m × 0.25 mm ID × 0.1
µm film thickness DB1-HT column, and an FID detector. Injections
of 3 µL were run split (100:1) with a total gas flow of 109 mL/ min
and a constant column flow of 1 mL/ min. Samples were analyzed
versus MDM, M3T, and M4Q standards with pentadecane as an
internal standard.
Fig u re 1 . ICP-AES analysis of sample prepared by microwave
digestion of poly(dimethylsiloxane)-spiked pig serum in HNO3. O -
poly(dimethylsiloxane)-spiked serum, x - silicate-spiked serum.
digestion of PDMS or sodium silicate spiked pig serum in HNO3
was conducted in order to determine the appropriateness of the
technique for sample preparation for total silicon determination.
Samples were analyzed by ICP-AES. Background-corrected in-
tensities were converted to silicon concentrations on the basis of
calibration curves made up from NIST traceable silicon standards.
Figure 1 shows a plot of actual silicon levels based on spiking
with PDMS or sodium silicate versus the concentration as
measured by ICP-AES. As one might expect, serum samples
spiked with sodium silicate give a slope near one, demonstrating
the appropriateness of the standards for that determination. The
nonzero intercept is indicative of a substantial amount of native
silicon that is present in the particular bottle of pig serum
(approximately 6.6 µg/ mL serum). In contrast, the slope of the
line for the PDMS spiked samples is nearly 2.3. This demonstrates
a species-dependent response for the digested PDMS samples,
resulting in an overestimate in the amount of silicon present. This
might be caused by incomplete digestion of the PDMS producing
species which influence sample introduction into the plasma.
Doubling the microwave digestion time did not eliminate the
elevated response for PDMS samples in the ICP-AES. It is likely
that this problem could be overcome by running the digestion
using HNO3/ HF or HNO3/ HCl/ HF as the digesting acid. Unfor-
tunately, this would require the use of HF, a very hazardous acid.
Clearly, many of the techniques in use based on ICP-AES can
detect elemental silicon, but absolute quantitation would not be
possible unless species-dependent responses are eliminated. Even
then, ICP-AES does not reveal anything about the origin of the
silicon species.
NMR does reveal information about the species present in the
sample; however, NMR is not nearly as sensitive as atomic
spectroscopic techniques. This is especially true of 29Si NMR
which has the added disadvantages of long spin-lattice relaxation
times (T1), a negative magnetogyric ratio, and a low natural
abundance of the 29Si isotope. The long T1 relaxation times are
particularly troublesome since obtaining quantitative information
typically requires using pulse delays of five times the longest T1.
Normally, T1 relaxation times can be significantly reduced by the
inclusion of a paramagnetic relaxation agent, but common agents
have low water solubility. In addition, paramagnetic relaxation is
strongly distance-dependent and only effective at close proximity
Designed Experiment. To optimize the microwave digestion
conditions, a 3-factor Box-Behnkin designed experiment was set
up and analyzed using Design-Expert 5.0.7 (Stat-Ease Corp.). Data
were fit to a quadratic model.
RESULTS AND DISCUSSION
Many groups have used ICP-AES to analyze for silicon in blood
or serum. Most of these studies have involved diluting the sample
and then analyzing the solution directly. For the most part, these
procedures should give reasonable estimates of total silicon, but
the analyst should be aware that, if organosilicon compounds are
involved, there is the potential for species-dependent responses
which could significantly impact the accuracy of the result. For
example, serum samples containing spikes of (CH3)3SiOH, (CH3)2-
Si(OH)2, or CH3Si(OCH2CH3)3 (which presumably hydrolyzes to
form CH3Si(OH)3) were diluted 4:1 with dilute HNO3 solution and
then analyzed by ICP-AES. Table 2 shows response factors for
each silane relative to spiking with a NIST silicate standard.
Clearly, the ICP-AES analysis using NIST silicate standards
would give reasonable estimates of the concentration of silicon
arising from dimethylsilanediol or methylsilanetriol. However,
trimethylsilanol gave a response nearly 17 times that of the NIST
standard. Thus, if trimethylsilanol were present, it would result
in a significant overestimate of the silicon concentration.
To eliminate species-dependent response, it is common to
subject samples to an ashing procedure. Wet ashing by microwave
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999 3057

Ta b le 3 . De s ig n e d Ex p e rim e n t t o Op t im ize Mic ro w a ve
Dig e s t io n Co n d it io n s fo r AS FT An a lys is
CH3
HCl
serum temp
(CH3)2SiO
recovery (%) dev.
std. cleavagea std.
(wt. %) (mL)
(°C)
N
(%)
dev.
5
5
5
0.250
0.625
0.625
1.000
0.250
0.250
130
120
140
130
120
130
3
41.8
30.0
36.4
12.4
46.3
75.1
8 8 .2
55.2
51.7
76.9
42.4
20.2
55.5
16.9
13.8
20.8
10.7
7.6
18.3
15.3
1 3 .6
18.9
32.7
18.2
38.1
6.3
0.23
0.13
0.58
0.55
1.34
1.28
1 .1 4
0.9
0.33
0.69
1.74
1.16
1.38
0.76
0.09
0.14
0.86
0.92
0.16
0.20
0 .4 4
0.3
0.02
0.57
0.98
0.22
0.25
0.04
3
3
3
3
3
5
10
10
1 0
10
10
10
15
15
15
15
0 .2 5 0 1 4 0
3
0.625
1.000
1.000
0.250
0.625
0.625
1.000
130 15
Fig u re 2 . 29Si NMR spectrum of a 0.0312% solution of (CH3)2Si-
(OH)2 in pig serum. Spectrum was acquired with 256 scans in 17.1
h. Peak at 3.9 ppm is dimethylsilanediol, s/n ) 34.6. The peak at
-4.2 ppm arises from trace levels of the condensation product,
tetramethyldisiloxanediol.
120
140
130
120
140
130
3
3
3
3
2
3
45.1
15.9
a
Percent of D units undergoing methyl cleavage, estimated from
the concentration of D and T units.
to the nuclei of interest. It might not be effective at all or effective
to varying degrees in a complex biological matrix.
The biological matrix itself may impose limitations on the NMR
experiment. Biological fluids such as serum commonly have
higher dielectric constants than usual organic NMR solvents, like
CDCl3, resulting in poorly tunable probes and reduced pulse power
delivered to the sample. Lower effective pulse power means longer
90-degree pulse widths with concomitant reductions in sensitivity.
Long pulse widths also limit the size of the spectral window that
can be observed while ensuring that all nuclei see the same
effective pulse width which is required for quantitative analysis.
On a VXR400 NMR spectrometer equipped with a 16-mm probe,
typical 90-degree pulse widths of 30 µs for CDCl3 solutions were
increased to 45-50 µs for serum samples. This is too long for
good T1 measurement and severely limits the frequency range
available for quantitative acquisition. Solvent effects can be
minimized by using a lower volume probe, but this also reduces
the sample size and ultimate sensitivity. To balance these factors,
the experiments were moved to a Unity Plus 400 MHz spectrom-
eter that handles higher pulse power yielding useful 25.5-µs 90-
degree pulse widths for 29Si on the 16-mm probe. In a recent paper,
Knight demonstrated the potential advantages of high magnetic
field strengths and polarization transfer pulse sequences in ideal
nonaqueous systems.30 Extrapolation to biological matrixes has
not been demonstrated and may be problematic for large sample
sizes since polarization transfer experiments are highly dependent
on effective pulse power.
The value of T1 for dimethylsilanediol at room temperature in
pig serum was measured as 48.4 ( 0.9 s using fast inversion
recovery.31 As a result of the long relaxation time, the acquisition
time for 256 quantitative scans for determination of the detection
limit required over 17 h. On the basis of the definition of the
detection limit as requiring a signal-to-noise of three, the detection
limit for dimethylsilanediol in pig serum was determined to be
64 µg Si/ mL. One should keep in mind that this is the detection
limit for a discrete water-soluble organosilicon compound with a
single sharp resonance, Figure 2. In actual biological samples,
there would be the potential for numerous species of varying
solubility and resonance frequencies. In addition, if the organo-
silicon compounds were bound to other species in the matrix,
substantial line broadening and a subsequent loss in sensitivity
could occur.
Another technique which can measure silicon levels and reveal
limited information on the form of silicon is the ASFT test.
Unfortunately, the test generally takes a long time to run and may
not work well for biological samples. A solution to both these
problems is to accelerate the test by carrying out the digestion at
a higher temperature in sealed vessels. This serves to rapidly
hydrolyze both the silicone and the proteins that constitute the
sample. This is most easily carried out by acid digestion of the
samples in microwave pressure vessels.
A designed experiment was carried out to determine the
optimum conditions for recovery of dimethylsiloxane from pig
serum spiked with PDMS while minimizing methyl cleavage. For
practical purposes, digestion conditions were sought that could
be accomplished in 2-3 h. This allows the analyst to prepare the
samples and then run the capping portion of the experiment
overnight.
A 3-level Box-Behnken designed experiment was used to
optimize the digestion conditions. The factors chosen for study
were the HCl concentration, the volume of serum, and the
temperature. Serum, water, and HCl were added such that the
total volume added was fixed at approximately 20 mL. The amount
of PDMS spiked into each serum sample was also held constant.
Thus, decreasing the volume of serum also had the effect of
increasing the concentration of PDMS relative to serum, but held
it constant relative to the final matrix. Capping with hexamethyl-
disiloxane was carried out by shaking at room temperature for
24 h. Increasing the capping time seemed to have little impact on
the recovery. The responses analyzed were the recovery of (CH3)2-
SiO and the amount of CH3 cleavage. Methyl cleavage was
determined from the concentration of M3T. The results of the
experiment with this design are shown in Table 3.
A response surface for recovery as a function of sample size
and temperature at a fixed HCl concentration of 10 wt. % is shown
in Figure 3. From this figure, it is immediately apparent that
recovery increases with increasing temperatures and decreasing
serum sample size. Unfortunately, methyl cleavage increases with
(30) Knight, C. T. G.; Kinrade, S. D. Anal. Chem. 1 9 9 9 , 71, 265-267.
(31) Canet, D.; Levy, G. C.; Peat, I. R. J. Magn. Reson. 1 9 7 5 , 18, 199.
3058 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

Ta b le 4 . Re c o ve ry a n d Me t h yl Cle a va g e Ob s e rve d in
AS FT An a lys is Fo llo w in g Dig e s t io n (3 h , 1 4 0 °C, 1 0 %
HCl, 0 .2 5 m L o f S e ru m )
spiked serum
concn.
(µg/ mL)
measured
concn. in serum
(µg/ mL)
% D recovery
(blank
corrected)
% D units
undergoing
Me cleavage
std.
std.
D
n
D
T
Q
ave.
dev. ave. dev.
1866
453
17.7
0.00
3
6
6
6
1634
389.4
18.5
12.9 1.87 87.5
3.89 1.54 85.5
0.27 1.71 92.4
17.6
15.4
2.1
0.8
1.0
1.7
0.3
0.6
1.6
2.07 0.00 1.78
Fig u re 3 . Response surface for recovery as a function of sample
size and temperature during microwave digestion of poly(dimethyl-
siloxane)-spiked pig serum in 10 wt. % HCl.
tion enhancement gained by extraction into the MM layer should
lower the detection limit in serum by as much as a factor of 4.
Moderately increasing the digestion time had little impact on
recovery or methyl cleavage. In the experiment shown in Table
4, 0.25 mL of serum was digested for 3 h in 10%HCl at 140 °C.
Spikes of PDMS were varied between 17.7 and 1866 µg (CH3)2-
SiO/ mL of serum (equivalent to 6.7 to 707 µg Si/ mL of serum).
Over this series, recoveries from individual runs ranged from 65
to 101%. It was noted that the precision was poor with samples
having higher concentrations of silicone spikes, but precision
improved dramatically for the lower concentration samples. This
may be due to the insolubility of poly(dimethylsiloxane) resulting
in a heterogeneous sample. If a means could be found to stir the
sample, the precision might improve dramatically.
Fig u re 4 . Response surface for recovery as a function of sample
size and HCl concentration during microwave digestion of poly-
(dimethylsiloxane)-spiked pig serum at 140 °C.
Although there was considerable variability in the individual
runs, the test is applicable for detection of silicones over a wide
range of silicone concentrations. Also, the observed recovery and
precision at lower silicone concentrations, 92.4 ( 2.1% at 17.7 µg
D/ mL, should be acceptable for most analyses. Thus, if replicate
analyses give irreproducible results, running the analysis on
smaller samples should resolve the problem.
Based on six blank serum analyses, the serum contained 2.07
( 0.38 µg (CH3)2SiO/ mL of serum. Using the convention of
establishing the detection limit from three times the standard
deviation of the blank, the detection limit would be 1.1 µg (CH3)2-
SiO/ mL of serum or 0.43 µg Si/ mL of serum. Detection limits
would have improved if lower background serum blanks could
have been obtained or if larger volumes of serum had been
digested. For critical work near the detection limit, more blank
determinations are highly recommended.
Preliminary experiments have shown that the technique is
applicable to other tissue samples as well. Digestion of PDMS
spiked chicken livers under nonoptimized conditions gave recov-
eries ranging from 51 to 90%.
A few words of caution about the test are appropriate. ASFT
analysis of serum or even deionized water blanks will show traces
of dimethylsiloxane. The origin of the dimethylsiloxane species
has not been established; however, the species may originate as
impurities in the hexamethyldisiloxane or as the result of
environmental contamination. Silicones are widely used as anti-
foam agents, adhesives, lubricants, release agents, etc., and
silicone contaminants may be introduced during sampling or
processing. It is also possible for species present in the sample
to catalyze methyl cleavage from hexamethyldisiloxane during the
capping phase of the analysis, thus generating D or even Q units.
Therefore, to determine if detected silicone is significant, it is
Fig u re 5 . Response surface for methyl cleavage as a function of
sample size and HCl concentration during microwave digestion of
poly(dimethylsiloxane)-spiked pig serum at 140 °C.
increasing temperature as well. Response surfaces for recoveries
and methyl cleavage at 140 °C are shown in Figures 4 and 5. Here
it becomes evident that recovery goes through a maximum as
the HCl concentration is increased. This may be a direct result
of the increased amount of methyl cleavage which is observed at
higher temperatures and HCl concentrations.
Software optimization suggested reasonable recoveries with
minimal methyl cleavage required an HCl concentration of 8.9-
9.5%, a temperature of 137-138 °C, and a sample volume of 0.25
mL. The digestion conditions closest to these optima, as evaluated
in Table 3, were 10 wt. % HCl, 140 °C, and 0.25 mL of serum.
Under these conditions, 88.2% recovery and 1.1% methyl cleavage
were observed. Unfortunately, standard deviations were high, and
further optimization may be warranted.
There are tradeoffs in choosing the digestion conditions. For
example, increasing the serum volume digested from 0.25 to 1.00
mL leads to a decrease in the recovery to 77%, but the concentra-
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999 3059

essential to run a series of matrix-matched blanks or suitable
controls in order to determine the detection limit of the technique
for the samples of interest.
sample to hexamethyldisiloxane, although this may result in
decreased recoveries. It was noted that microwave digestion did
lead to a small amount of methyl cleavage, which precludes the
possibility of absolute quantitation of D, T, and Q units originally
present in the sample. It is possible that methyl cleavage could
be further reduced by changing the digestion conditions. Possible
modifications might include using enzymatic digestion to break
down the matrix prior to the acid digestion, using milder digestion
conditions (at the cost of extending the analysis), or changing
the digesting acid.
When using the test, it is absolutely critical that detection limits
be properly established and that replicates are run to establish
precision. In those cases in which precision is poor, both recovery
and precision can be improved by diluting the sample, although
this comes at the cost of raising the detection limit of the
technique.
CONCLUSIONS
Many researchers analyze for silicon as a surrogate for silicone.
In these cases, it is advisable to keep in mind that silicon and
silicone are commonplace, and contamination can be a significant
issue. Running appropriate matrix-matched blanks is absolutely
essential. Researchers should also take care to determine whether
species-dependent responses may influence their results. This is
particularly true of ICP-AES, in which volatile silanes or incom-
pletely digested silicones may lead to overestimates of the silicon
concentrations. Positive identification of silicone is possible by a
number of spectroscopic techniques; however, most do not yield
quantitative information. An exception is 29Si NMR; but the
technique lacks sensitivity, particularly if the silicone is present
as multiple species.
A sensitive technique for identifying and quantifying silicones
in aqueous or biological samples is to use microwave digestion
to hydrolyze silicone and proteins followed by analysis of the
digested sample using the Aqueous Silanol Functionality Test.
Detection limits below 1 µg (CH3)2SiO/ mL of sample are feasible.
Lower detection limits may be obtained by increasing the ratio of
ACKNOWLEDGMENT
The authors would like to thank Gary Kozerski for invaluable
assistance and helpful discussions on the use of the ICP-AES.
Received for review February 10, 1999. Accepted April 26,
1999.
AC990157D
3060 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999