Investigative Ophthalmology & Visual Science, Vol. 31, No. 10, October 1990
Copyright © Association for Research in Vision and Ophthalmology
Analysis and Fractionation of Silicone and
Fluorosilicone Oils for Intraocular Use
Kimitoshi Nakamura,*t Miguel F. Refojo,*t Donald V. Crobrree,* and Fee-Lai Leong*
Silicone oil (SiO) and fluorosilicone oil (FSiO) are useful in difficult cases of retinal detachment
surgery. Unidentified low-molecular-weight components (LMWC) and residual catalysts in SiO and
FSiO have been implicated in the adverse reactions of the oils in the eye. The authors analyzed
LMWC of SiO and FSiO using a gas chromatograph (GC) equipped with a 6-ft X 2-mm column
packed with 3% SP-2250 and a flame-ionization detector. By commercially available standards and a
homologous series plot, MD3M to MD23M (linear LMWC) and D4 to D30 (cyclic LMWC) were
positively identified in commercial-grade 1000-centistokes (cs) SiO. Commercial-grade 12,500-cs SiO
contained GC-detectable LMWC (up to MD2sM and D30) at higher concentrations than commercialgrade 1000 cs SiO, although the weight percent of acetone-extractable LMWC (including those larger
than MD2gM and D30) was less in the former than in the latter. The GC-detectable LMWC in most
medical-grade SiO were less than those in commercial-grade SiO. Tetramethylammonium siloxanolate (a residual catalyst) and tributylphosphine oxide (a heat-decomposition product of a polymerization catalyst) were tentatively identified in commercial- and medical-grade 12,500-cs SiO, respectively. Commercial-grade 1000- and 10,000-cs FSiO also contained LMWC, including F3 and/or F4
(cyclic LMWC). To eliminate LMWC from the oils, the authors developed a solvent fractionation
method using acetone for SiO and hexane for FSiO. After continuous solvent extraction of SiO for 2
weeks and of FSiO for 3 weeks, all measurable LMWC were eliminated from the oils. Invest Ophthalmol Vis Sci 31:2059-2069, 1990
Silicone oil (SiO, polydimethylsiloxane) has been
used as a vitreous substitute to treat difficult cases of
retinal detachment, such as those complicated with
proliferative vitreoretinopathy, giant retinal tears,
and penetrating ocular trauma.1'2 Fluorosilicone oil
(FSiO, polymethyl-3,3,3-trifluoropropylsiloxane)
also is used for retinal detachment surgery, although
less frequently.3 The National Eye Institute is evaluating the effect of SiO tamponade in a randomized
multicenter surgical trial.4
Both SiO and FSiO are polydispersed synthetic
polymers that consist of mixtures of linear species
with different chain lengths and may contain linear
and cyclic species of lower molecular weights. The
cyclic species are by-products of the polymerization
of the oils (Fig. I).5"7 Some structures of these lowmolecular-weight components (LMWC) of SiO and
FSiO are shown in Figure 2.8
Using gel permeation chromatography, Gabel et
al9 reported that LMWC were present in clinically
used SiO, although individual components were neither separated nor identified. They arbitrarily defined
LMWC of SiO to be linear and cyclic species with a
molecular weight of <2400 daltons.
The presence of LMWC in SiO and FSiO is
thought to be a cause of the toxicity of these oils when
they are applied intraocularly. li9~12 Although not previously detected in intraocularly used SiO, residual
polymerization catalysts in the oil also are thought to
cause adverse ocular reactions.29"11 Our objectives
were to identify LMWC and residual catalysts in SiO
and FSiO and to remove them from the oils to facilitate intraocular use.
From the *Eye Research Institute, and the tDepartment of Ophthalmology, Harvard Medical School, Boston, Massachusetts.
Supported by the National Institute of Health grant EY-00327
(M.F.R.).
Presented in part at the ARVO Annual Meeting, Sarasota, Florida, May, 1989.
Reprint requests: Miguel F. Refojo, DSc, Eye Research Institute,
20 Staniford Street, Boston, MA 02114.
Commercial-grade 1000- and 12,500-centistokes
(cs) SiO (PS-S; Table 1) and 1000- and 10,000-cs
FSiO (PS-F) were obtained from Petrarch Systems
(Bristol, PA). These oils contain neither additives nor
fillers and are not blends.8 Medical-grade 100-, 1000-,
and 12,500-cs SiO (USDC-360) and medical-grade
1000-cs SiO (USDC-Q7), used in The Silicone
Materials and Methods
Materials
2059
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
2060
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Ocrober 1990
catalyst'31
0
monomer* + end-capping
<2)
5
* MDmM< > + D n
(6)
Vol. 31
Table 1. Silicone oil (SiO) and fluorosilicone oil
(FSiO) analyzed by gas chromatography
heat(4)
Key:
(1) Monomer = D 3 , D 4 , or D 5 .
(2) End-capping: MM or MD2M.
(3) Catalyst: potassium hydroxide, potassium siloxanolate,
tetrabutylphosphonium hydroxide, tetrabutylphosphonium siloxanolate, tetramethylammonium siloxanolate, etc.
(4) Heat: 70° to 110°C in the case of tetramethylammonium siloxanolate.
(5) MDmM: mixture of linear species (m 2: 0).
(6) D n : mixture of cyclic species (n ^ 3).
Fig. 1. Polymerization of silicone oil (SiO): high-molecularweight SiO is usually produced by anionic open-ring polymerization of cyclic monomers.5-6 High-molecular-weight fluorosilicone
oil also is polymerized by a similar process.7
Study,4 were obtained from US Dow Corning (Midland, MI). Medical-grade 1000- and 12,500-cs SiO
(JNDC-360) was obtained from Japan Dow Corning
(Yamakita, Kanagawa). Medical-grade 5000-centiLinear LMWC of SiO (MD m M)
Me
Cyclic LMWC of SiO (Dn)
Me
I
i
Ms- Si - O -Si -Me
i
i
Me
Me
Abbreviation*
60421
1000-cs PS-S
70438
12,500-cs PS-S
81145
1000-cs PS-F
90626
10,000-cs PS-F
HH096196
1000-cs
USDC-360
HH036137
12,500-cs
USDC-360
HX1U4
USDC-Q7
HHO381
100-cs USDC360
HH051648
1000-cs
JNDC-360
HH103678
12,500-cs
JNDC-360
16028 17 0105
AM-5000
* Abbreviations of SiO and FSiO are specific usage in this article.
Me
Hexamethylcyclotrisiloxane (D3)
Me
Me-Si-O-Si — O - Si -Me
I
I
I
Me
Commercial grade SiO
Petrarch Systems
1000 cs
Petrarch Systems
12,500-cs
Commercial grade FSiO
Petrarch Systems
1000-cs
Petrarch Systems
10,000-cs
Medical grade SiO
US Dow Corning 360
Medical Fluid 1000
cs
US Dow Corning 360
Medical Fluid
12,500-cs
US Dow Corning Q7
2870 1000 cs
US Dow Corning 360
Medical Fluid 100
cs
Japan Dow Corning
360 Medical Fluid
1000-cs
Japan Dow Corning
360 Medical Fluid
12,500-cs
Adatomed 5000 cps
(5139 cs)
Lot
Me
Hexamethyldisiloxane (MM)
Me
Origin of the oils
/
\
^
Me — Si —'
S
Me
Octamethyltrisiloxane (MDM)
Octamethylcyclotetrasiloxane (D4)
Me
Me
Me
Me
i
l
l
I
M e - Si - O - Si - O - Si - O - Si - Me
I
I
I
I
Me
Me
Me
Me
Decamethylteirasiloxane (MD2M)
Linear LMWC of FSiO (MF m M)
Decamethylcyclopentasiloxane (D5)
Cyclic LMWC of FSiO (Fn)
F
i
F-C- F
CH 3 -Si- O- - S i - O- - S i - CH3
I
I
CH_
CH.
F-C-F
Fig. 2. Molecular structures of linear and cyclic low-molecularweight components of silicone oil (SiO) and fluorosilicone oil
(FSiO) 8 : D and F denote the units -(CH 3 ) 2 SiO- and
-CH2(CF 3 CH 2 CH2)SiO-, respectively. M denotes the unit
(CH3)3SiOi/2-, which is the end-capping of the linear polymers.
Me denotes methyl group.
poises (cps) SiO (AM-5000), which is clinically used
in Europe, was obtained from Adatomed GmbH
(Miinchen, FRG). Origins, lot numbers, and abbreviations of SiO and FSiO are given in Table 1.
Hexamethyldisiloxane (MM), octamethyltrisiloxane (MDM), decamethyltetrasiloxane (MD2M),
hexamethylcyclotrisiloxane (D3), octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane
(D5), a mixture of polymethyl-3,3,3-trifluoropropylcyclosiloxanes (trimethyl-3,3,3-trifluoropropylcyclotrisiloxane (F3) and tetramethyl-3,3,3-trifluoropropylcyclotetrasiloxane (F4), and tetramethylammonium siloxanolate (1.5-2.0% nitrogen as end-capped
polydimethylsiloxane) were obtained from Petrarch
Systems. Pure polydimethylcyclosiloxanes (D6, D9,
and D12), a mixture of polydimethylcyclosiloxanes
(D12, D, 5 , Dig, and D2i) and pure low-molecularweight polydimethylsiloxanes (MD6M, MD7M,
MDgM, MD9M, MD,0M, and MDI2M) were kindly
supplied by Ohio Valley Specialty Chemical (Marietta, OH).
Tributylphosphine oxide was obtained from Aidrich (Milwaukee, WI). Hexane (97% n-hexane; Baker
Analyzed HPLC Reagent), and benzene (Baker Analyzed Reagent) were obtained from Doe & Ingalls
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
No.
10
2061
ANALYSIS AND FRACTIONATION OF SIUCONE OIL / Nokomuro er ol
(Medford, MA). Acetone (Spectranalyzed) and ethanol (HPLC grade, 4.5% methanol and 4.9% isopropanol) were obtained from Fisher (Pittsburgh, PA).
10
120°
160°
200°
240°
280°
8.
Gas-Chromatographic Analysis
A Perkin-Elmer 3920 gas chromatograph (GC)
(Buck, Norwalk, CT) equipped with a flame-ionization detector (FID) and a SP4100 integrator (Spectra-Physics, San Jose, CA) was used. A 6-ft X 2-mm
glass column was packed with 1.7 g of 3% SP-2250 on
80/100 mesh Supelcoport (Supelco, Bellfonte, PA).
Ultrasep-R (Alltech Associates, Deerfield, IL) was
used as GC septa. The temperature program was
80°C for 4 min, then 8°C/min to 280°C, held at
280°C. Temperatures of injector and detector were
290° and 300°C, respectively. The carrier gas was
nitrogen (flow rate, 26 ml/min). It was passed
through a molecular sieve/silica gel trap and an oxygen trap (Buck). Pressure to the flow controller was
maintained at 78 psi. Pressures of hydrogen and air to
the FID were 20 and 50 psi, respectively.
Before injecting the sample into the GC, 2 ml each
of SiO and acetone were placed in a test tube, shaken
by a Vortex-Genie (Scientific Industries, Springfield,
MA) at speed control 10 for 1 min, and centrifuged
for 5 min at room temperature (21° ± 1°C). Then 2
/il of the solvent layer was injected into the GC with a
Hamilton syringe (10 ix\, fixed needle 701N) using the
plug injection technique.13 The LMWC of FSiO were
analyzed by the same procedure as SiO, except hexane was used as the extraction solvent.
We first identified the peaks of 1000-cs PS-S by the
technique of peak coincidence with the aid of LMWC
standards ("spiking")14 using the temperature program described above. In the GC (isothermal condition) of linear and cyclic LMWC of SiO, the linear
correlation between the log retention volume (In VR)
and the number of silicon atoms has been reported.1516 Therefore, the peaks of the LMWC, of
which standards were not obtained, were identified
using a homologous series plot.17 Briefly, the retention volume (VR) of each standard of linear and cyclic
LMWC was determined by GC at 80°, 120°, 160°,
200°, 240°, and 280°C. The In VR of the standards
then was plotted against the number of silicon atoms,
and the slope of the line of the linear and cyclic
LMWC standards was separately calculated by regression analysis at the different temperatures. Acetone extract of 1000-cs PS-S also was measured by
GC under the above isothermal conditions, and the
unknown peaks of 1000-cs PS-S were identified by
the relation to the lines obtained by regression analysis. The peaks in the chromatograms of the other SiO
samples were identified similarly. We also spiked
6.
4.
10
20
30
Number of Silicon Atoms
Fig. 3. Homologous series plot17 of linear and cyclic low-molecular-weight components (LMWC) of silicone oil: log retention volume (In VR) of linear and cyclic LMWC is plotted against the
number of silicon atoms at different temperatures. Retention volume (V R ; ml) = retention time (t R ; min) X flow rate of the carrier
gas (26 ml/min).
1000- and 10,000-cs PS-F using a mixture of F3
and F4.
Because the response of FID is different among
LMWC of SiO,513 the weight response factors (RFi5
Fig. 3) of the LMWC standards (MDM, MD2M,
MD6M, MD9M, MD12M, D4, D5, D6, D9, and D12)
were determined using the temperature program.
Benzene was used as the reference standard, and the
weight response factor (RFstd) of benzene was assigned as l.OO.13
When the base line of the chromatogram was
noisy, the peak areas were determined by cutting out
the photocopied peak and weighing the paper on a
balance (Mettler H54, Caley & Whitmore, Somerville, MA). Before analysis of the oil samples, the
contamination of the injection port or the column
was checked by injecting the pure organic solvents
into the GC. Detector sensitivity was checked frequently by injecting 1.7 mg/ml of cholesterol in decane/toluene (50:50, v/v).
Fractionation of SiO and FSiO by Continuous
Solvent Extraction
Commercial-grade SiO (1000- and 12,500-cs PS-S)
was fractionated with acetone for 2 weeks, and commercial-grade FSiO (1000- and 10,000-cs PS-F) was
fractionated with hexane for 3 weeks using a continuous extraction apparatus (250 ml, liquid/liquid type
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
2062
INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / Ocrober 1990
for use with solvents lighter than water, Kontes,
Vineland, NJ). About ten granules of Chemware TFE
Boiling Stones (Teflon; Fisher) were added to the
boiling flask (500 ml, Pyrex; Corning, Corning, NY).
When the solvent fractionation was finished, the
supernatant layer in the extractor was separated from
the oil layer by decantation and added to the contents
of the boiling flask. Using a rotatory evaporator
(Rinco, Greenville, IL), the solvent in the boiling
flask was evaporated. The solvent dissolved in the
fractionated oils also was evaporated. (In this report,
the fraction of the oils from which LMWC has been
eliminated by the solvent fractionation is referred to
as fractionated oils.)
The fractionated oils were shaken vigorously with
an equal amount of distilled water in a separatory
funnel, then separated from the water. After this procedure was repeated three times, the oils were heated
under vacuum (~0.5 mm Hg) with a stirrer in a SiO
bath at 110°C for 2 hr and at 150°C for an additional
2 hr, mixed with approximately 20 mg/ml of activated charcoal (Nuchar C-190-N; Eastman Kodak,
Rochester, NY), and filtered twice (Whatman filter
paper No. 1, then No. 2; Whatman, Maidstone, England).
Kinematic viscosity of the fractionated oils and the
solvent-extractable LMWC of the oils was determined at 25°C using glass capillary viscometers.5 A
Cannon-Manning Semi-Micro Viscometer (size 400
and 450; Cannon, State College, PA) was used for the
oils with viscosity < 3000 cs and a Cannon-Fenske
Routine Viscometer (size 600; International Research Glassware, Kenilworth, NJ) for the oils with
viscosity ^ 3000 cs. Weight percent (Wt %) of the
fractionated oils and the solvent-extractable LMWC
of the oils to the total oils was determined using a
balance (Mettler PI000, Caley & Whitmore).
Refractive index (nD), density, and surface tension
of 1000-and 12,500-cs PS-Sand 1000-and 10,000-cs
PS-F were determined before and after fractionation.
The nD was measured at 25°C using a Valentine Precision Refractometer Model 350 A (Vista, CA). Density was measured at room temperature using a
Moore-Van Slyke Specific Gravity Bottle (capacity, 2
ml; Fisher) and a balance (Mettler H54). Surface tension was measured at room temperature by the
method previously described.18
To determine the recovery of LMWC from SiO by
our extraction procedure for the GC analysis, 0.01,
0.1, or 1 mg/ml each of linear and cyclic LMWC
standards (D6, MD6M, D9, MD9M, D12, and MD12M)
were mixed together with fractionated 1000-cs PS-S.
Then 2 ml of the oil was extracted with 2 ml of acetone at room temperature, and the concentrations of
the LMWC standards in the acetone were measured.
Vol. 31
The recovery of the LMWC standards from SiO was
calculated according to the following equation:
(Recovery of each LMWC standard) = (concentration of the standard in acetone)/(concentration of the
standard in fractionated 1000-cs PS-S X 100(%).
Elimination of LMWC from SiO and FSiO
by Heating
Because heating of SiO and FSiO is sometimes
used to eliminate LMWC from the oils,1118 we examined the effect of the heating on the elimination of
LMWC. After 10 ml of commercial-grade SiO (1000and 12,500-cs PS-S) and FSiO (1000- and 10,000-cs
PS-F) were placed in separate beakers (capacity, 50
ml), they were heated in an air-circulating oven at
200°C for 24 hr. Then the oils were analyzed by GC
and compared with those before heating. The specimens of SiO and FSiO were weighed using a balance
(Mettler H54) before and after heating, and the
weight loss (%) by heating was calculated. These experiments were done in duplicate.
Analysis of Nitrogen, Phosphorus, and Potassium
Medical-grade SiO (1000- and 12,500-cs USDC360), commercial-grade SiO (1000- and 12,500-cs
PS-S before and after fractionation), and commercial-grade FSiO (1000- and 10,000-cs PS-F before and
after fractionation) were sent to an outside contractor
(Galbraith, Knoxville, TN) for the analyses of nitrogen, phosphorus, and potassium, which are the characteristic components of possible residual catalysts in
the oils (Fig. 1). Nitrogen was measured by the Kjeldahl method. Phosphorus was determined by colorimetry using the molybdenum blue method.1920 In
the positive control, phosphorus also was determined
by inductively coupled plasma atomic emission spectroscopy (ICP-AES) using a Perkin-Elmer ICP 6000
(Norwalk, CT).21 In both analyses of phosphorus, the
oils were decomposed by wet digestion using a mixture of sodium molybdate, nitric acid, sulfuric acid,
and perchloric acid before measurements. Atomic
absorption spectroscopy was used for the determination of potassium. We mixed 10 mg of tetramethylammonium siloxanolate (30-40 ppm of nitrogen)
and 6.80 mg of tributylphosphine oxide (190 ppm
phosphorus) in 5 g of fractionated 1000-cs PS-S separately, and used this for positive controls. Each sample was analyzed once.
Fluorescence Spectroscopic Analysis
Using a fluorescence spectrophotometer (PerkinElmer MPF-4), we measured the fluorescence of
medical-grade SiO (1000- and 12,500-cs USDC-360),
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
ANALYSIS AND FRACTIONATION OF SILICONE OIL / NoKomuro er ol
No. 10
commercial-grade SiO (1000- and 12,500-cs PS-S),
and commercial-grade FSiO (1000- and 10,000-cs
PS-F). These oils were extracted with an equal
amount of ethanol, and the fluorescence in the ethanol was measured.
Results
GC Analysis
By spiking with commercially available standards,
we positively identified linear LMWC (MD6M,
MD7M, MD8M, MD9M, MD10M, and MD12M) and
cyclic LMWC (D4, D5, D6, D9, D12, D15, D18, and D2I)
in 1000-cs PS-S. Because of elution on the solvent
front, the detection limits of D4 and D5 were only 0.1
and 0.01 mg/ml, respectively. The detection limit of
LMWC larger than D5 was 0.001 mg/ml. The MM
and D3 were not separated from the solvent front in
our GC analysis, and MDM and MD2M were not
detected in 1000-cs PS-S. Pyrolysis of LMWC standards was not observed.
Figure 4 shows a homologous series plot of linear
and cyclic LMWC of SiO at different temperatures.
We found a high correlation between the In VR of
LMWC and the number of silicon atoms at all temperature ranges measured (coefficient of determination > 0.983). The VR of linear LMWC was larger
than that of cyclic LMWC of the same number of
silicon atoms. The slope of the line obtained by regression analysis decreased with an increase of tem-
2-
5
10
15
Number of Silicon Atoms
Fig. 4. The weight response factor (RFi) of low-molecular-weight
components (LMWC) of silicone oil (SiO) for flame ionization
detector relative to benzene (RFstd = 1.00): RFi of each LMWC
standard is plotted against the number of silicon atoms. RFi
= Astd/Ai X Wi/Wstd X RFstd, where Astd, Ai, Wstd and Wi are
the peak area of benzene and LMWC of SiO, and the weight of
benzene and LMWC of SiO, respectively.513
2063
perature. This may be explained by the partition coefficient of LMWC between stationary phase and carrier-gas phase, which decreases with an increase of
column temperature because of the increment of the
vapor pressure of LMWC.22
As shown in Figure 3, RFj of LMWC of SiO increased with an increase of molecular weight of
LMWC, indicating the response of FID to LMWC
decreased with an increase of molecular weight of
LMWC. The values of RF; were an average of six
measurements. (Larger aberrations in the values of
RFj for MDM, MD2M, D4, and D5 compared with
those of the other standards probably were caused by
their close proximity to the solvent front.)
The contamination of the injection port was eliminated by injecting organic solvents several times or by
increasing the injection port temperature for several
days. The sensitivity of FID, measured by the response to cholesterol, did not change. However, the
hydrogen flame was occasionally extinguished by a
build-up of combustion products (probably silicon
dioxide) that clogged the detector nozzle.
In the GC of commercial-grade 1000-cs SiO (Fig.
5), main peaks and satellite peaks appeared alternatively. By spiking and by using our homologous series
plot (Fig. 4), we determined the satellite peaks to be
linear LMWC (MD3M to MD23M) and the main
peaks to be cyclic LMWC (D4 to D30). Commercialgrade 12,500-cs SiO (Fig. 6A) contained GC-detectable LMWC (up to MD28M and D30) at higher concentrations than commercial-grade 1000-cs SiO.
(Molecular weights of MD28M and D30 were 2234
and 2220 daltons, respectively.)
At least 17 and 26 LMWC were detected by GC in
commercial-grade 1000-cs FSiO (Fig. 6C) and
10,000-cs FSiO (Fig. 6E), respectively. Using a mixture of F3 and F4, we identified the first peak of these
oils as F3 and/or F4 (Figs. 6C, 6E). (Separation of the
peaks of F3 and F4 could not be achieved with our
system.) The amounts of GC-detectable LMWC in
10,000-cs PS-F (Fig. 6E) were more than those in
1000-cs PS-F (Fig. 6C).
The GC-detectable LMWC in most medical-grade
SiO were less than those in commercial-grade SiO.
Both AM-5000 (Fig. 7B) and 1000-cs USDC-360
(Fig. 8A) contained the least concentrations of
LMWC among all SiO products we analyzed. Particularly in 1000-cs USDC-360, linear LMWC smaller
than MDi2M and cyclic LMWC from D6 to D,o were
less than detection limits. The USDC-Q7 (Fig. 8B),
which is clinically used in The Silicone Study,4 contained relatively high concentrations of LMWC. Although 1000-cs JNDC-360 (Fig. 8C) contained much
larger amounts of LMWC (especially linear species)
than 1000-cs USDC-360 (Fig. 8A), the concentra-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
2064
INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / October 1990
120°
160° 200°
Vol. 31
240° 280
Fig. 5. Gas chromatogram of commercial grade 1000-cs silicone oil: each peak is identified by spiking with standards14 and a homologous
series plot.17 The concentrations* of MD9M and D| 2 in this sample are 0.006 and 0.044 mg/ml, respectively. Concentrations in the acetone
extract.
tions of LMWC in 12,500-cs JNDC-360 (Fig. 9D)
were similar to those in 12,500-cs USDC-360 (Fig.
9A). Except 1000-cs JNDC-360 (Fig. 8C), commercial- and medical-grade SiO contained more cyclic
LMWC than linear LMWC.
Fractionation of SiO and FSiO by Continuous
Solvent Extraction
The efficacy of our solvent fractionation method
can be seen by the before and after chromatograms of
1000- and 12,500-cs PS-S and 1000- and 10,000-cs
PS-F (Figs. 5-7), showing that all measurable peaks of
LMWC were eliminated from the fractionated oils.
After the solvent fractionation of SiO and FSiO, the
contents in the boiling flask separated into two
phases. The top layer was solvent rich, and the bottom layer was rich in LMWC of the oils.
We tried continuous water extraction of fractionated 1000- and 12,500-cs PS-S and 1000- and
280°
Fig. 6. Gas chromatogram of commercial grade 12,500-cs silicone oil (PS-S) and 1000- and 10,000-csfluorosiliconeoil (PS-F)
before and after fractionation. (A) 12,500-cs PS-S before fractionation; (B) 12,500-cs PS-S after fractionation; (C) 1000-cs PS-F before fractionation; (D) 1000-cs PS-F after fractionation; (E)
10,000-cs PS-F before fractionation; (F) 10,000-cs PS-F after fractionation. *F3 and/or F4.
Fig. 7. Comparison of gas chromatogram of fractionated silicone
oil (A) fractionated 1000-cs PS-S with that of clinically used silicone oil; (B) AM-5000; (C) 1000-cs USDC-360. Fractionated
1000-cs PS-S does not have any detectable LMWC, which are still
measurable with flame ionization detection in AM-5000 and
1000-cs USDC-360. These chromatograms are the highest amplification and the drift of the base line is column bleed, which appears
at a column temperature above 240°C.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
No. 10
ANALYSIS AND FR.ACTIONATION OF SILICONE OIL / Nokomuro er ol
2065
The LMWC of FSiO, of which retention time (tR) was
shorter than 15 min (168°), also were almost entirely
eliminated by this heating period (Fig. 11). The
weight loss of 1000- and 12,500-cs PS-S and 1000and 10,000-cs PS-F by the heating was 0.24, 1.17,
0.80, and 0.55%, respectively. Heating of the oils (10
ml) in scintillation vials (capacity, 20 ml) was less
effective in eliminating LMWC from the oils than
that in beakers (capacity, 50 ml), probably because
the surface area of the oils in the former was smaller
than that in the latter.
Fig. 8. Gas chromatogram of medical grade 1000-cs silicone oil
(A) USDC-360; (B) USDC-Q7; (C) JNDC-360; (D) a mixture* of
100-cs USDC-360 and 12,500-cs JNDC-360. (A) 1000-cs
USDC-360 contains much less LMWC than (C) 1000-cs
JNDC-360. (B) USDC-Q7 contains relatively high concentrations
of LMWC. (C) 1000-cs JNDC-360 has a gas chromatogram similar
to that of the mixture of (D) 100-cs USDC-360 and 12,500-cs
JNDC-360. T h e volume ratio of 100-cs USDC-360 and 12,500-cs
JNDC-360 in this mixture is 48:52, and the measured viscosity is
2290 cs.
10,000-cs PS-F using an extraction apparatus (SiO,
1000 ml, liquid/liquid type for use with solvents
heavier than water; FSiO, 250 ml, liquid/liquid type
for use with solvents lighter than water, Kontes). The
FSiO could be extracted with water; however, SiO
was heavily emulsified (water-in-oil emulsion18).
The viscosity of SiO and FSiO considerably increased after fractionation (Table 2), although other
physical properties of the oils, such as nD, density,
and surface tension, were similar before and after
fractionation (Table 3). Because the Wt % of solventextractable LMWC in 1000-cs PS-S and PS-F was
larger than that in 12,500-cs PS-S and 10,000-cs
PS-F, the viscosity of the fractionated oils produced
from the former increased more compared with the
latter (Table 2). (The measured viscosity of 1000- and
12,500-cs PS-S and 1000- and 10,000-cs PS-F before
fractionation was 1341, 12,221, 980, and 11,043 cs,
respectively.)
The recovery of the LMWC standards (D6, MD6M,
D9, MD9M, D12, and MD,2M) from fractionated SiO
by a single extraction with an equal amount of acetone was ~30% in the concentration range from
0.01-1 mg/ml of each standard (Table 4). The value
of recovery was an average of two measurements.
Elimination of LMWC from SiO and FSiO
by Heating
After 1000- and 12,500-cs PS-S were heated in
beakers at 200°C for 24 hr, LMWC of the oils smaller
than D7 became less than detection limits of GC-FID.
Larger LMWC still remained in the oils, although the
peak sizes decreased. (The peaks of 12,500-cs PS-S
decreased more than those of 1000-cs PS-S.) (Fig. 10).
Analysis of Residual Catalysts
High-molecular-weight SiO and FSiO, which are
clinically used, are probably produced by anionic
open ring polymerization of cyclic monomers (Fig.
I).5"7 (Low-molecular-weight oils usually are produced by cationic open ring polymerization of cyclic
monomers, or polycondensation of hydrolyzed dimethyldichlorosilane.5-6) Because catalysts for the
anionic polymerization of the oils are not volatile,2324
they cannot be detected by GC directly.
The presence of a basic catalyst, such as tetramethylammonium siloxanolate, was suspected in 12,500cs PS-S by the following experiments. When the organic solvents in the boiling flask were evaporated
after the solvent fractionation, the residues in the
boiling flask obtained from 1000-cs PS-S and 1000and 10,000-cs PS-F became one layer, which was
LMWC of the oils. Meanwhile, the residue in the
boiling flask obtained from 12,500-cs PS-S separated
into two layers after the evaporation of acetone.
Using an infrared (IR) spectrometer (Perkin-Elmer
1310), the top layer (nD, 1.42; boiling point,
Fig. 9. Gas chromatogram of medical grade 12,5OO-cs silicone oil
(A) USDC-360 without spike; (B) USDC-360 with spike; (C)
USDC-360 purified by activated charcoal; (D) JNDC-360. By
spiking, one of the peaks (*) in (A, B) 12,500-cs USDC-360 are
tentatively identified as tributylphosphine oxide. This peak is not
detected in (C) 12,5OO-cs USDC-360 purified by activated charcoal,
and in (D) 12,500-cs JSDC-360. The size of the peak (*) in the
chromatogram of (A) 12,500-cs USDC-360 corresponds to 0.11
mg/ml of tributylphosphine oxide.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
2066
Vol. 31
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Ocrober 1990
Table 2. Weight percent (Wt %, at 21° ± 1 °C) and viscosity (at 25°C) of fractionated silicone (PS-S) and
fluorosilicone (PS-F) oils and low-molecular-weight components (LMWC)* of the oils
Fractionated oils
Wt %
Oil
69.4
88.1
77.9
87.5
1000-csPS-S
12,500-csPS-S
1000-csPS-F
10,000-csPS-F
±
±
±
±
LMWC*
Viscosity (cs)
13.6
16.6
4.5
3.3
2722
16,395
1556
14,212
Wt%
± 95
± 1435
± 103
± 1446
30.6
11.9
22.1
12.5
Viscosity (cs)
± 7.5
±4.2
±3.4
±3.5
135
149
248
471
± 27
± 40
± 9
± 384
' Solvent-extractable LMWC: this fraction is larger than that of gas-chromatography-detectable LMWC.
152°-154°C) and the bottom layer (nD, 1.40) were
identified as diacetone alcohol (4-hydroxy-4-methyl2-pentanone) and SiO, respectively. The IR spectrum
of the top layer had the characteristic absorption
bands of alcohol (O-H) stretching vibration at 3540
cm"1 and ketone (C=O) stretching vibration at 1705
cm." 125 The remainder of the spectrum was also
identical to diacetone alcohol, except a few bands
possibly caused by impurities.
Diacetone alcohol is formed by base-catalyzed
aldol condensation of acetone.26 When we refluxed
acetone with a trace amount of tetramethylammonium siloxanolate, diacetone alcohol was produced.
When 1000-cs PS-S and 1000- and 12,500-cs
USDC-360 were extracted with acetone, diacetone
alcohol was not formed in the boiling flask. Therefore, we suspect, among these oils, only 12,500-cs
PS-S contained a residual amount of basic catalyst,
which catalyzed the aldol condensation of acetone
during the solvent fractionation.
The GC of 12,500-cs USDC-360 (Fig. 9A) showed
one extra peak between D, i and D12, which had some
tailing. (Thet R ofD M and D !2 was 18.5 and 19.8 min,
Table 3. Refractive index (nD, at 25°C), density
(g/cm3, at 21 ° ± 1 °C), and surface tension
(dyne/cm, at 21° ± 1°C) of commerical grade
silicone (PS-S) and fluorosilicone (PS-F) oils
before and after fractionation
Density
Surface
tension
1.4039
1.4038
0.966
0.967
20.9
21.1
1.4039
1.4039
0.970
0.971
21.2
21.3
1.3809
1.3812
1.267
1.282
24.1
24.1
1.3816
1.3817
1.293
1.293
24.0
24.2
Oil
1000-cs PS-S
Before fractionation
After fractionation
12,500-cs PS-S
Before fractionation
After fractionation
1000-cs PS-F
Before fractionation
After fractionation
10,000-csPS-F
Before fractionation
After fractionation
respectively.) By spiking (Fig. 9B), this peak was tentatively assigned as that of tributylphosphine oxide.
The peak of tributylphosphine oxide standard had a
similar tailing as the extra peak of 12,500-cs
USDC-360, and the tR became shorter with an increase of its amount. (The tR of 0.01, 0.1, and 1
mg/ml of tributylphosphine oxide was 21.3, 18.8, and
17.4 min, respectively.) This phenomenon may be
explained by nonlinear adsorption (Langmuir isotherm) of this compound on the stationary phase.27
By percolation through activated charcoal (Fig. 9C)
or by continuous extraction with water, the extra
peak of 12,500-cs USDC-360 disappeared. (Tributylphosphine oxide is water soluble.) In the GC of
12,500-cs JNDC-360 (Fig. 9D), this extra peak was
not found.
Based on these results, we deduced that tetrabutylphosphonium hydroxide, tetrabutylphosphonium silanolate, or tetrabutylphosphonium siloxanolate was
used as an anionic polymerization catalyst of 12,500cs USDC-360 and that tributylphosphine oxide was
produced by the heat decomposition of the catalyst in
the factory.23'24 (Tributylphosphine oxide also has
been reported to be used for the deactivation of the
other catalysts [ie, potassium hydroxide] in the process of the anionic polymerization of SiO.28)
Table 4. Recovery (%; at 21° ± 1°C) of several
standards of low-molecular-weight components
(LMWC) from fractionated 1000 cs silicone oil
(PS-S) by a single extraction with an equal
amount of acetone
Concentration of the standard in
fractionated PS-S
LMWC
standards
0.01 mg/ml
0.1 mg/ml
I mg/ml
D6
MD 6 M
D9
MD 9 M
Dl2
MD I2 M
Mean
37.4
28.8
32.2
23.2
26.2
22.6
28.4
35.5
25.5
28.2
24.1
25.4
24.2
27.2
35.8
30.3
35.1
24.4
33.0
26.9
30.9
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
ANALYSIS AND FRACTIONATION OF SILICONE OIL / NoUomuro er ol
No. 10
2067
Table 5. Analysis of nitrogen, phosphorus, and
potassium* (ppm) in silicone (USDC-360, PS-S)
and fluorosilicone (PS-F) oils
Fig. 10. Gas chromatograms of commercial grade silicone oil (A)
1000-cs PS-S and (C) 12,000-cs PS-S and the same oils (B) and (D),
respectively, after heating at 200°C for 24 hr.
Analysis of Nitrogen, Phosphorus, and Potassium
As shown in Table 5, 12,500-cs PS-S (in which the
presence of tetramethylammonium siloxanolate was
suspected) contained the largest amount of nitrogen
(31 ppm). After the fractionation of the oil, the
amount of nitrogen was less than 1 ppm. The amount
of phosphorus was the largest (16 ppm) in 12,500-cs
USDC-360 (in which the presence of tributylphosphine oxide was suspected). Potassium in most samples was lower than the detection limit of atomic absorption analysis. Using the Kjeldahl method, we obtained the expected value of nitrogen in the positive
control. The value of phosphorus in the positive control, which was determined by colorimetry, was
slightly lower than we expected; however, a similar
value was obtained by ICP-AES.
Oil
Nitrogen
Phosphorus
Potassium
1000-cs USDC-360
12,500-cs USDC-360
1000-cs PS-S
Before fractionation
After fractionation
Add nitrogen
(30-40 ppm)t
Add phosphorus
(190ppm)t
12,500-cs PS-S
Before fractionation
After fractionation
1000-cs PS-F
Before fractionation
After fractionation
10,000-cs PS-F
Before fractionation
After fractionation
1
11
12
16
<2
7
6
6
8
<2
<2
37
9
<2
13
126(148$)
<2
31
<1
8
9
<2
10
1
7
6
<2
<2
<1
8
8
<2
<1
<2
3
<2
* Nitrogen, phosphorus, and potassium are determined by Kjeldahl
method, molybdenum blue method, and atomic absorption spectroscopy,
respectively.
t Positive controls.
$ Determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES).
Fluorescence Spectroscopic Analysis
Using fluorescence spectroscopy, we found that lot
#44158 of 10,000-cs PS-F wasfluorescent(excitation,
310 nm; emission, 368 nm). This fluorescence was
eliminated from FSiO by continuous extraction with
hexane. However, continuous extraction with water
and percolation through activated charcoal or alumina (Acid and Basic, Brochman Activity I, 80-200
mesh; Fisher) was not effective in eliminating it from
the oil. The cause of thefluorescencein this FSiO was
not identified. Lot #90626 of 10,000-cs PS-F did not
fluoresce. Fluorescence also was not found in 1000and 12,500-cs USDC-360, 1000-and 12,500-cs PS-S,
and 1000-cs PS-F.
Discussion
280°
280"
Fig. 11. Gas chromatograms of commercial grade fluorosilicone
oil (A) 1000-cs PS-F and (C) 10,000-cs PS-F' and the same oils after
heating at 200°C for 24 hours (B) and (D), respectively.
When the samples are introduced into the GC, they
are first volatilized at the injection port, and carried
as vapor through the column by the carrier gas.
Therefore, when a compound is analyzed by GC, the
compound must be volatile and stable at high temperature. The LMWC of SiO and FSiO have been
shown to be volatile and measurable by GC without
pyrolysis.5'29'30 However, high-molecular-weight SiO
and FSiO are not volatile. Hence, when SiO and FSiO
are injected directly into the GC, high-molecularweight oils may accumulate in the injection port and
then undergo pyrolysis, which will interfere with the
analysis of LMWC of the oils. Some organic solvents
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
2068
INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Ocrober 1990
such as acetone, ethanol, dioxane, and dihexyl adipate (partial solvents of SiO) do not dissolve high-molecular-weight SiO, but they do dissolve LMWC of
the oil.813 Hexane, ethanol, and methanol are partial
solvents of FSiO. To inject only LMWC of SiO and
FSiO into the GC, we first extracted LMWC of the
oils with the partial solvents (acetone for SiO and
hexane for FSiO). Then we injected the solvent extracts into the GC.
Only a part of LMWC of the oils were extracted by
a single extraction procedure using the partial solvents. However, the concentrations of LMWC in the
solvents did not change, when the time of vortexmixing was prolonged for 5 min or the test tube was
kept for 1 day after centrifugation. Therefore, we believe that the approximate equilibrium concentrations of LMWC between the oils and the partial solvents occurred during the 1-min shaking time.
We assumed that the amounts of LMWC in the
partial solvents correlate with the amounts in the oils.
To test this assumption, we mixed some of the
LMWC standards with fractionated SiO and extracted with acetone. The recovery of the LMWC
standards from the oil was nearly constant in the
concentration range of 0.01-1 mg/ml of each standard (Table 4). Therefore, we believe it is possible to
compare the relative amounts of LMWC in the oils
by obtaining the peak area of LMWC in the partial
solvents after a single extraction at constant temperature.
After all GC-detectable LMWC (up to MD28M and
D30) were eliminated from SiO by solvent fractionation, a part of the fractionated oil still dissolved in
acetone, indicating that acetone-soluble LMWC contained low-molecular-weight species larger than
MD28M and D30. Although 12,500-cs PS-S (Fig. 6A)
contained more GC-detectable LMWC than 1000-cs
PS-S (Fig. 5), acetone-extractable LMWC were less in
12,500-cs PS-S than in 1000-cs PS-S (Table 2). As
PS-S, 10,000-cs PS-F (Fig. 6E) contained more GCdetectable LMWC than 1000-cs PS-F (Fig. 6C), while
hexane-extractable LMWC were less in the former
than in the latter (Table 2).
Not only commercial-grade SiO and FSiO but also
12,500-cs USDC-360 (Fig. 9A) contained more GCdetectable LMWC than 1000-cs USDC-360 (Fig.
8A). These results appear to indicate that the elimination of GC-detectable LMWC from the high-viscosity
oils (12,500-cs SiO and 10,000-cs FSiO) is more difficult in the factory than from the low-viscosity oils
(1000-cs SiO and FSiO). However, 1000-cs
JNDC-360 (Fig. 8C) contained much higher concentrations of the LMWC (especially linear species) than
12,500-cs JNDC-360 (Fig. 9D). It is known that SiO
of two different viscosities is sometimes blended by
the supplier to reach the required viscosity.8" Be-
Vol. 01
cause the gas chromatogram of 1000-cs JNDC-360
(Fig. 8C) was similar to that of the mixture of 100-cs
USDC-360 and 12,500-cs JNDC-360 (Fig. 8D), it is
our opinion that 1000-cs JNDC-360 may have been
produced by a blending process. (100-cs USDC-360
contained considerable amounts of linear LMWC.)
Elimination of residual catalysts from high-viscosity SiO in the factory also appears to be more difficult
than from low-viscosity oils, because tetramethylammonium siloxanolate and tributylphosphine oxide
were tentatively identified only in 12,500-cs PS-S and
USDC-360, respectively. The elemental analysis
(using the Kjeldahl method) showed that 12,500-cs
PS-S contained more than three times higher concentration of nitrogen than the other oils. This result
supports our speculation that the residual basic catalyst in 12,500-cs PS-S was tetramethylammonium siloxanolate.
By colorimetric determination, a higher concentration of phosphorus was found in 12,500-cs
USDC-360 compared with the other oils. Parel et al1'
and Johnson et al12 previously analyzed the elements
in clinically used SiO by atomic emission spectroscopy (or ICP-AES) and found phosphorus in the oils.
In the first report, the values of phosphorus in the oils
were extremely high (200-2200 ppm).11 However, in
the second report, a much smaller amount of phosphorus was denoted without specified values (< 150
ppm).12 This range of values agrees with our results
(Table 5). If the extra peak between D n and D|2 in
the chromatogram of 12,500-cs USDC-360 (Fig. 9A)
was tributylphosphine oxide, the peak size was consistent with the value of phosphorus which was determined by colorimetry.
Although SiO and FSiO are relatively stable compounds, they are subject to thermal attack at the siloxane bond (Si-O) and oxidative attack at the side
group of the siloxane chain.31 Thermal attack at the
siloxane bond converts the oils into a mixture of cyclic LMWC. This begins to occur around 400°C in a
vacuum.3233 However, oxidation of the side group of
SiO, which produces formaldehyde and formic acid,
begins to occur even below 200°C.31'34 Consequently,
an environment free of oxygen will prevent the oxidative degradation of SiO, when the oil is heated for
purification or sterilization before intraocular use.
Siloxane bonds also are cleaved by alcoholysis. Rearrangement of D3 to D4 in ethanol takes place at
room temperature.35 In the case of FSiO, even highmolecular-weight oil is degraded by ethanol at room
temperature to give linear and cyclic LMWC.35
Therefore, alcohol should not be used for purification
of FSiO.1836
The LMWC and residual catalysts in SiO and FSiO
have been implicated in the ocular toxicity of the
oils.1'2-9"12 Kampik et al37 first indicated that one of
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017
No. 10
ANALYSIS AND FRACTIONATION OF SILICONE OIL / Nokomuro er ol
the causes of the variable frequency of the complications of intraocularly applied SiO among different
research groups may be the varying physical and
chemical properties of the oil they used. Because the
experimental evidence of this assumption has not
been reported as yet, in our opinion, the effect of
LM WC and residual catalysts in SiO and FSiO on the
ocular toxicity is still obscure. Currently, we are
doing in vivo experiments on the tolerance of fractionated SiO and FSiO.
Key words: fluorosilicone oil, gas chromatography, retinal
detachment surgery, silicone oil, solvent fractionation
References
1. Federman JL and Schubert HD: Complications associated
with the use of silicone oil in 150 eyes after retina-vitreous
surgery. Ophthalmology 95:70, 1988.
2. Skorpik C, Menapace R, Gnad HD, and Paroussis P: Silicone
oil implantation in penetrating injuries complicated by PVR:
Results from 1982 to 1986. Retina 9:8, 1989.
3. Tolentino FI, Cajita VN, Chung H, Acosta J, Freeman HM,
and Refojo MF: High-density fluorosilicone oil in vitreous
surgery. In Proliferative Vitreoretinopathy (PVR), Freeman
HM and Tolentino FI, editors. New York, Springer-Verlag,
1989, pp. 177-180.
4. Ryan SJ: The silicone study. In Clinical Trials Supported by
the National Eye Institute, 1988, pp. 35-38. Bethesda, MD,
NIH Publication No. 88-2910.
5. Thrash CR: Silicon compounds, organic, silicone fluids. In
Encyclopedia of Industrial Chemical Analysis, Vol 18, Snell
FD and Ettre LS, editors. New York, Interscience Publishers,
1973, pp. 95-155.
6. Hardman BB and Torkelson A: Silicon compounds (silicones).
In Kirk-Othmer Encyclopedia of Chemical Technology, Vol
20, 3rd ed, Kirk RE, Othmer DF, Grayson M, and Eckroth D,
editors. New York, Wiley, 1982, pp. 922-962.
7. Veith CA and Cohen RE: Kinetic modelling and optimization
of the trifluoropropylmethylsiloxane polymerization. Poly
Mat Sci Eng 57:661, 1987.
8. Anderson R, Arkles B, and Larson GL: Silicon Compounds,
Register and Review. Bristol, PA, Petrarch Systems, 1987, pp.
253-300.
9. Gabel VP, Kampik A, and Burkhardt J: Analysis of intraocularly applied silicone oils of various origins. Graefes Arch Clin
Exp Ophthalmol 225:160, 1987.
10. Zivojnovic R: Silicone Oil in Vitroretinal Surgery. Dordrecht,
The Netherlands, Martinus Nijhoff/Dr W. Junk, 1987, pp.
40-44.
11. Parel JM: Silicone oils: Physicochemical properties. In Retina,
Vol III, Ryan SJ, Glaser BM, and Michels RG, editors. St.
Louis, CV Mosby, 1989, pp. 261-277.
12. Johnson RN, Flynn HW, Parel JM, and Portugal LM: Transient hypopyon with marked anterior chamber fibrin following
pars plana vitrectomy and silicone oil injection. Arch Ophthalmol 107:683, 1989.
13. Smith AL: Analysis of Silicones. Malabar, FL, Robert E.
Krieger Publishing, 1974, pp. 42-45, 217-246.
14. Keulemans AIM: Gas Chromatography, 2nd ed. New York,
Reinhold, 1959, p. 27.
15. Wurst M: Analyse von Organosiliciumverbindungen: III.
Trennung und Bestimmung linearer und cyclischer Polydimethylsiloxane Mittels Gaschromatographie. Collection
Czech Chem Commun 29:1458, 1964.
2069
16. Preisler L: Separation of the homologous series of methylcyclopolysiloxanes, methylhydrocyclopolysiloxanes, methylethylhydrocyclopoly-siloxanes and linear methylpolysiloxanes by
means of gas-liquid chromatography. Z Anal Chem 240:389,
1968.
17. Crabtree DV, Adler AJ, and Handelman GJ: Pentafluorobenzoicanhydride as a derivatizing agent for alcohols and hydroxy
fatty acid methyl esters detected by electron capture in gas
chromatography. J Chromatog 466:251, 1989.
18. Nakamura K, Refojo MR, and Crabtree DV: Factors contributing to emulsification of intraocular silicone and fluorosilicone oils. Invest Ophthalmol Vis Sci 31:647, 1990.
19. Kirk BP and Wilkinson HC: Determination of phosphorus in
coke by the oxygen-flask method. Talanta 19:80, 1972.
20. Ma TS and Rittner RC: Modern Organic Elemental Analysis.
New York, Marcel Dekker, 1979, pp. 247-248.
21. Perkin-Elmer ICP Manual #0993-9609. Norwalk, CT, PerkinElmer, 1981.
22. Littlewood AB: Gas Chromatography: Principles, Techniques,
and Applications. New York, Academic Press, 1962, pp.
55-63.
23. Gilbert AR and Kantor SW: Transient catalysts for the polymerization of organosiloxanes. J Poly Sci 40: 35, 1959.
24. Kirk N: A continuous process for making silicones. Ind Eng
Chem 51: 515, 1959.
25. Bellamy LJ: The Infra-red Spectra of Complex Molecules.
London, Methuen & Co, 1958, pp. 134-136.
26. Fieser LF and Fieser M: Organic Chemistry, 3rd ed. New York,
Reinhold, 1956, pp. 207-210.
27. Golshan-Shirazi S and Guiochon G: Analytical solution for the
ideal model of chromatography in the case of a Langmuir
isotherm. Anal Chem 60: 2364, 1988.
28. Kawazumi K, Maruyama H, and Hashi K: Open ring polymerization of cyclic dimethylpolysiloxane by basic catalysts.
Kogyo Kagaku Zasshi 66:628, 1963.
29. Carmichael JB, Gordon DJ, and Ferguson CE: Analysis of
large linear and cyclic methylsiloxanes and computer calculation of the chromatographic data. J Gas Chromatog 4:347,
1966.
30. Kagan EG, Yuzhelevskii YuA, Klebanskii AL, Larinova ON,
and Kharlamova AV: 3,3,3-Trifluoropropyl(methyl)cyclosiloxanes: I. Isolation and identification of trifluoropropyl(methyl)dimethylcyclosiloxanes. Khim Geterotsikl Soedin
3:819, 1967.
31. Torkelson A: Developments and trends in silicone fluids. In
Applied Polymer Symposia, No. 14, Silicone Technology,
Bruins PF, editor. New York, Interscience Publishers, 1970,
pp. 61-69.
32. Lewis CW: The pyrolysis of dimethylpolysiloxanes. J Poly Sci
33:153, 1958.
33. Novikov SN, Kagan YeG, and Pravednikov AN: Thermal degradation of 3,3,3-trifluoropropyl(methyl)siloxane. Vysokomol
Soyed 8:1015, 1966.
34. Nakao K, Wada N, and Nakajima K: Waterproof of silk fabrics treated with polysiloxanes. Kogyo Kagaku Zasshi 59:361,
1955.
35. Voronkov MG, Mileshkevich VP, and Yuzhelevskii YuA: The
Siloxane Bond, Physical Properties and Chemical Transformations. New York, Consultants Bureau, 1978, pp. 146-157.
36. Miyamoto K, Refojo MF, Tolentino FI, Fournier GA, and
Albert DM: Fluorinated oils as experimental vitreous substitutes. Arch Ophthalmol 104:1053, 1986.
37. Kampik A, Gabel VP, and Spiegel D: Intraokulare Tamponade mit hoch-viskosem Siliconol bei massiver proliferativer
Vitreo-Retinopathie. Klin Monatsbl Augenheilkd 185:368,
1984.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933152/ on 06/17/2017