Polysiloxanes by proton NMR — NorrChemica Analysis Notes, NMR Series Part 22. Navy card with a schematic proton spectrum of a silicone: the vinyl and Si–H signals of the two reactive groups in blue, the tall Si–CH3 signal at 0.07 ppm in gold.

Polysiloxanes by proton NMR: silicone oil in medicines, food and cosmetics, the reactive groups before curing, and the rubber network after it

Analysis Notes · NMR Series · Part 22

Polysiloxanes by proton NMR: silicone oil in medicines, food and cosmetics, the reactive groups before curing, and the rubber network after it

What a silicone is, how a proton spectrum measures its amount, which NMR experiments are used for oils, rings and rubbers, and which questions belong to gas chromatography, to atomic spectrometry or to an instrument for solids

Sources EFSA J., J. Pharm. Sci., PDA J. Pharm. Sci. Technol., Pharm. Ind., Foods, Food Addit. Contam., Front. Pharmacol., Organometallics, Lab Chip, Anal. Chem., Polimery, Polymer, ACS Appl. Mater. Interfaces, Macromolecules, Polymers, Polym. Degrad. Stab., Magn. Reson. Med., Biomaterials, J. Biomed. Mater. Res. A, J. Pharm. Biomed. Anal., Plast. Reconstr. Surg. Glob. Open, Am. J. Roentgenol., Phys. Chem. Chem. Phys., J. Chromatogr. A, Int. J. Cosmet. Sci., J. Soc. Cosmet. Chem., Talanta, BIO Web Conf.; EU and US regulations, an opinion of an EU scientific committee, a German test method, a US government report, two US patents, a trade magazine, an application note Experiments quantitative 1H NMR, 1H DOSY, 29Si–1H HMBC, 1H double-quantum NMR, solid-state 29Si NMR

A syringe, a frying oil, a shampoo, a baking mould and a breast implant all contain silicone. Chemically these materials are one family: chains of alternating silicon and oxygen atoms that carry methyl groups. Polymers with such a chain are called polysiloxanes. This article describes what nuclear magnetic resonance (NMR) measures on them: how much silicone a medicine, a food or a rinse water contains, how many reactive groups a rubber formulation holds before it cures, and how the network is examined after curing. It states the limit of each measurement and names the method that is better beyond it. Part 10 covers the silicon-29 spectrum in detail.

A silicone oil is a mixture of chains of many lengths. Chromatography lacks reference substances for most of them, and atomic spectrometry measures silicon from any source.[1] In the proton spectrum the methyl groups of all chains give one signal near 0 ppm, a region in which the extract of a food or the residue of a rinse water shows almost nothing else.[1, 2] A proton spectrum of the liquid from a prefilled syringe gave the content of silicone oil and of its two degradation products at levels below 10 parts per million, without the hydrolysis, derivatisation and extraction steps of the GC–MS methods.[3] In 1979 a titration reported 3.87 % of vinyl groups in a silicone fluid whose proton spectrum shows none.[4] A meat loaf baked in a silicone mould contained 177 mg/kg of siloxanes, almost three times the overall migration limit.[5] The method has limits. Graphite-furnace atomic absorption detects 7 µg/l of silicone oil in an aqueous medicine,[6] one hundredth of the 0.7 mg/kg detected in an NMR tube.[1] The EU limit of 0.1 % for each of the rings D4, D5 and D6 is tested by gas chromatography,[7, 8] and a cured rubber does not dissolve.

Five terms used throughout (the table of terms before the summary explains twenty more). PDMS (polydimethylsiloxane) is a chain of –Si(CH3)2–O– units; silicone oil is PDMS with trimethylsilyl ends. A siloxane is any compound with Si–O–Si links; D4, D5 and D6 are the rings of four, five and six dimethylsiloxane units. Si–H (hydride) and vinyl (CH2=CH–Si) are the two groups that react with each other when an addition-cured silicone hardens. A cross-link is a bond that joins two chains. qNMR (quantitative NMR) compares a signal area with that of a weighed standard (Part 4). In this article ppm after a chemical shift belongs to the NMR scale; concentrations are given in mg/kg, mg/l or per cent and, where a source does so, in parts per million.

At a glance: the question, the NMR measurement and its limit

Question NMR measurement Limit, or the better method
How much silicone oil has entered an aqueous medicine? 1H spectrum of the formulation: silicone oil, trimethylsilanol and dimethylsilanediol below 10 parts per million[3] Atomic absorption detects 7 µg/l, as total silicon[6]
How much E 900 does a food contain? 1H spectrum of a hexane extract against an internal standard; quantification down to 0.2 mg/kg from 5 g of food[1] Fat must be removed with care: saponification hydrolysed part of the PDMS[1]
How much siloxane passes from a baking mould into food? 1H spectrum of the food extract; detection limit 8.7 mg/kg of oil[9] Single oligomers are separated by HPLC or GC[9]
How much dimethicone does a shampoo leave on hair? 1H spectrum of the evaporated rinse water against an internal standard; recovery 108 %[2] The source states no detection limit; X-ray fluorescence measures silicon on the hair itself[2]
Does a silica for paints carry traces of silicone oil? 1H spectrum of a hexane extract at 500 MHz[10] The powder itself gives broadened lines; infrared spectroscopy is not sensitive enough[10]
How many Si–H and vinyl groups does a prepolymer contain? 1H signals at 4.64 ppm and at 5.75 to 6.40 ppm against a weighed standard[4] On pure vinyl siloxanes the measurement of 1979 gave 92 to 96 % of the true value[4]
How densely is a cured rubber cross-linked? 1H double-quantum NMR on the solid[11, 12] Not a routine solution experiment; swelling and mechanical tests are the classical measures[13]
What do the gel and the shell of an implant contain? 1H NMR and 29Si–1H correlation of dissolved material; solid-state 29Si NMR of the gel[14, 15, 16] The rings D4, D5 and D6 in the gel: 604 parts per million at most in the analyses of national authorities, which cite GC–MS[17]
Which oligomers does a mixture contain? DOSY: the diffusion coefficient falls with the square root of the molecular mass[18] The 1H signals of the oligomers overlap; 29Si resolves them at lower sensitivity[18]
Does a product contain 0.1 % or more of D4, D5 or D6? Not an NMR task GC with flame-ionisation detection after silylation[7, 19]

What a silicone is

A silicone is a polymer whose chain consists of silicon and oxygen atoms in turn. In the commonest silicone every silicon atom carries two methyl groups (CH3); its chemical name is polydimethylsiloxane, PDMS (Figure 1).

H3CSiCH3CH3OSiCH3CH3O OIL · LINEAR CHAIN a liquid trimethylsilyl groups close both ends syringes, antifoam E 900, shampoo CH3SiCH3OSiCH3CH3OSiH3CH3COSiCH3H3CO RING D4: four units, volatile D5 and D6 have five and six units cosmetics, residue in polymers OSiCH3CH3CH2CH2SiCH3OO RUBBER · CROSS-LINKED CHAINS swells, does not dissolve a bridge joins two chains baking moulds, seals, implants

Figure 1. One polymer in three forms. A wavy line marks the continuation of a chain; the blue discs mark the bridge formed in an addition cure. The line in italics names products in which each form occurs. Our drawing.

The Si–O bond is about 1.64 Å (0.164 nm) long, and every second atom of the chain carries no substituent. For both reasons the chain is unusually flexible.[20] The food additive, with average molecular weights of about 6,800 to 30,000, is a clear, viscous liquid.[21]

Three forms of the polymer appear in this article. Oils are linear chains closed at both ends by trimethylsilyl groups. Small rings are volatile liquids. Rubbers, gels and resins consist of branched molecules or of chains joined to one another by cross-links.[20]

A technical silicone is never a single compound. The synthesis gives chains of different lengths, and the product may also contain rings.[20] For two viscosity grades of E 900, 350 and 1,000 centistokes, producers reported number-average molecular weights of about 11,000 and 16,000 to the European Food Safety Authority (EFSA).[21]

One polymer under several names

Name Where the name is used What it denotes
Poly­dimethyl­siloxane, PDMS Chem­istry The chain of –Si(CH3)2–O– units
Silicone oil Pharmacy, engin­eering Liquid PDMS
Dimeticone (di­methicone), sime­ticone Medi­cines, cos­metics PDMS, or a preparation based on it, as an active pharma­ceutical ingredient or a cosmetic ingredient[20]
E 900, dimethyl polysiloxane Food PDMS as an anti­foaming additive[21]
Cyclo­methicone; D4, D5, D6 Cos­metics; chemicals regu­lation Volatile rings[20, 22]
Silicone rubber, elastomer, gel Moulds, seals, implants Cross-linked PDMS[20]

The physical properties of a sample alone do not show whether it is a pharmaceutical ingredient or the food additive.[20]

How a proton spectrum measures an amount

An NMR spectrometer detects the nuclei of hydrogen atoms, the protons. Each kind of proton in a molecule gives a signal at its own position, the chemical shift, which is stated in parts per million (ppm) of the measuring frequency. The area under a signal is proportional to the number of protons behind it (Part 1).

The second property makes NMR a counting method. A known mass of a pure standard is added to the sample, and the two areas are compared. A reference sample of the silicone itself is not required, which matters for a polymer that is a mixture of chains (Part 4).

A model calculation shows the steps (Figure 2). To 5.00 g of a food we add 0.050 mg of the standard bis(trimethylsilyl)benzene, which has 18 equivalent protons in a molar mass of 222.5 g/mol. One unit of PDMS has 6 protons in 74.15 g/mol. If the silicone signal has 1.20 times the area of that of the standard, the tube contains 0.060 mg of PDMS, and the food 12 mg/kg. The numbers are invented for the example; the standard and the amount of food are those of a published method described below.[1]

1 · WEIGH 5.00 g of food + 0.050 mg of standard standard: 18 protons in 222.5 g/mol extract with hexane, evaporate, dissolve in CDCl3 2 · MEASURE 0.3 0.2 0.1 0.0 chemical shift, ppm standard area 1.00 silicone area 1.20 3 · CALCULATE protons of the standard 0.050 mg ÷ 222.5 g/mol × 18 = 4.05 µmol protons of the silicone 4.05 µmol × 1.20 = 4.85 µmol silicone units, 6 protons each 4.85 µmol ÷ 6 = 0.809 µmol mass, 74.15 g/mol per unit 0.809 µmol × 74.15 = 0.060 mg 0.060 mg in 5.00 g = 12 mg/kg

Figure 2. From two signal areas to a content in mg/kg. Our model calculation with invented numbers; the spectrum is drawn, not measured.

The result is reliable under three conditions: the two signals overlap with no others, the nuclei relax fully between two pulses (Part 9), and the standard is weighed accurately.

The silicone signals in the proton spectrum

In deuterated chloroform (CDCl3) the six equivalent protons of the –Si(CH3)2–O– unit give a singlet at 0.07 ppm.[1] The ends of the chains appear beside it. A hydroxyl-terminated chain shows its last unit, HO–Si(CH3)2–O–, at 0.16 ppm and the neighbouring one at 0.11 ppm. With a trimethylsilyl end, (CH3)3Si–O– appears at 0.08 ppm and the adjacent unit at 0.03 ppm. Cyclic dimethylsiloxanes give a sharp singlet at 0.05 ppm.[1] These values come from a review written in Reus, Spain, which takes the four end-group shifts from the work of Helling and co-workers at TU Dresden.[1]

CHAIN UNIT OSiCH3CH3O 0.07 ppm main signal of the polymer TRIMETHYLSILYL END H3CSiCH3CH3OSiCH3CH3O 0.08 ppm, end group 0.03 ppm, adjacent unit SILANOL END HOSiCH3CH3OSiCH3CH3O 0.16 ppm, end unit 0.11 ppm, adjacent unit RING CH3SiCH3OSiCH3CH3OSiH3CH3COSiCH3H3CO 0.05 ppm cyclic dimethylsiloxanes VINYL GROUP ON SILICON H2CCHSiCH3CH3O 5.75 to 6.40 ppm three protons, twelve lines HYDRIDE (Si–H) OSiHCH3O 4.64 ppm proton bonded to silicon WHERE EACH GROUP APPEARS IN THE PROTON SPECTRUM 8 7 6 5 4 3 2 1 0 1H chemical shift in CDCl3, ppm phenyl on silicon 7.59, 7.35 vinyl 6.40 to 5.75 Si–H 4.64 most organic ingredients Si–CH3 0.07

Figure 3. The protons of a silicone and their chemical shifts in CDCl3. Gold and blue discs mark the groups to which the values refer; a wavy line marks the continuation of the chain. Our drawing of published values: methyl groups from the review of 2022,[1] vinyl and Si–H protons from the report of 1979.[4] The strip at the bottom places the groups on one axis; the phenyl positions come from the same report.

Lipid extracts of foods and of other biological materials usually show no signals below 0.5 ppm.[1] The silicone signal is therefore free of overlap with fats and waxes, and its area gives the amount of polymer once a weighed standard is present in the tube.

Two reactive groups lie far from the methyl region. In a spectrum of 1979 the hydrogen of an Si–H unit appears at 4.64 ppm, and the three protons of a vinyl group on silicon form twelve lines between 5.75 and 6.40 ppm. Phenyl groups on silicon, present in some fluids and rubbers, appear at 7.35 and 7.59 ppm.[4]

The position of the methyl signal depends on the solvent. A table of impurity signals compiled in 2010 by Fulmer and seven co-authors in Seattle, Pasadena and at Bar-Ilan University lists silicone grease in eleven deuterated solvents: at 0.07 ppm in chloroform, 0.29 ppm in benzene and −0.06 ppm in dimethyl sulfoxide (Figure 4).[23] A group in Mexico City found the PDMS of wine extracts at 0.285 ppm in benzene-d6.[24] In mixtures of carbon tetrachloride with benzene-d6 linear PDMS appears at 0.18 to 0.21 ppm.[25]

0.3 0.2 0.1 0.0 −0.1 chemical shift of the Si–CH3 signal, ppm TMS benzene-d6 0.29 toluene-d8 0.26 trifluoroethanol-d3 0.16 chlorobenzene-d5 0.14 acetone-d6 0.13 tetrahydrofuran-d8 0.11 methanol-d4 0.10 dichloromethane-d2 0.09 acetonitrile-d3 0.08 chloroform-d 0.07 dimethyl sulfoxide-d6 −0.06

Figure 4. Chemical shift of the methyl signal of silicone grease in eleven deuterated solvents. Our drawing of published values;[23] the table lists no value for heavy water.

Tetramethylsilane (TMS), the usual reference at 0 ppm, is itself a methylsilicon compound: an excess of it covers the signal at 0.03 ppm.[1]

The NMR experiments used for silicones

The studies cited in this article use six kinds of NMR experiment. The first and the fourth are measurements on a dissolved sample with standard equipment; the others require a probe tuned to silicon, long measuring times or hardware for solids.

NMR experiments for silicones

Experi­ment What it gives for a silicone Special require­ment
Quanti­tative 1H NMR (Part 4) The amount of silicone in a product; the content of Si–H, vinyl and phenyl groups[1, 4] A weighed standard, little or no TMS, and a delay of five relax­ation times[1, 25]
29Si NMR (Part 10) The kind of every silicon atom: chain end, chain unit or branch point; the small rings D3 and D4 apart from chains Low sensi­tivity, slow relax­ation and the background signal of the glass tube (Part 10)
29Si–1H corre­lation, HMBC (Part 10) The silicon atom behind each proton signal; it showed silicone oligomers among the lipids of explanted implants[15] A probe tuned to 29Si
DOSY (Part 17) The signals of a mixture sorted by the size of the molecules[18] A probe with field gradients
Solid-state NMR with magic-angle spinning (Part 11) Spectra of gels and of cured rubber[11, 16] A solid-state probe and a rotor
1H double-quantum NMR and relax­ation measure­ments The cross-link density of a rubber and the share of loose chains[11, 12] A time-domain instru­ment at about 20 MHz[13] or a 400 MHz spectrometer[12]

The 29Si experiment is worth its long measuring time where the proton spectrum separates two species poorly. Cyclic dimethylsiloxanes give their methyl signal 0.02 ppm from that of the chains, which is 10 Hz at 500 MHz. In the 29Si spectrum the ring D4 lies 2.14 ppm from the chain units of a silicone oil, 213 Hz on the same instrument, and D3 a further 11 ppm away (Figure 5). Part 10 gives no values for D5 and D6, and we make no statement on them.

PROTON SPECTRUM 0.12 0.10 0.08 0.06 0.04 0.02 0.00 chain 0.07 cyclic siloxanes 0.05 1H chemical shift, ppm 0.02 ppm apart: 10 Hz at 500 MHz SILICON-29 SPECTRUM −8 −12 −16 −20 −24 chain −22.0 D4 −19.86 D3 −9 29Si chemical shift, ppm D4 to chain 2.14 ppm: 213 Hz at 99.3 MHz

Figure 5. Cyclic siloxanes and chains in the proton spectrum, and the rings D3 and D4 beside the chain in the 29Si spectrum. Sticks at published shifts with arbitrary heights: the 1H positions from the review of 2022,[1] the 29Si values as given in Part 10. The differences in hertz are our arithmetic for a 500 MHz spectrometer, on which 29Si resonates at 99.3 MHz.

Silicone oil in a prefilled syringe

The glass barrel of a prefilled syringe is lubricated with silicone oil, so that the plunger moves evenly (Figure 6, A).[26] The amounts are small. A pharmaceutical company in Maryland measured the oil on empty syringes by infrared spectroscopy: units that held 0.25 to 0.8 mg when new reached a stable minimum of 0.15 to 0.26 mg during one to four years of storage with the tip upwards.[26] Part of the oil can pass into the medicine.[3]

The oil matters for protein medicines. In 2005 Jones, Kaufmann and Middaugh at the University of Kansas added 0.5 % of silicone oil to solutions of four model proteins and found significant aggregation in each.[27] In 2009 a group led from a biotechnology company in California studied a monoclonal antibody. The protein adsorbed to the oil, and aggregates formed when the samples were also agitated; the surfactant polysorbate 20 suppressed the loss of antibody completely.[28]

Most earlier methods for the oil in a medicine and its degradation products required hydrolysis, derivatisation and extraction before gas chromatography–mass spectrometry (GC–MS). These steps can lose the two hydrolysis products, trimethylsilanol and dimethylsilanediol, or overestimate them. Both silanols are hydrophilic and remain in the water phase.[3]

A · WHERE THE OIL IS aqueous medicine silicone oil on the glass: 0.25 to 0.8 mg in a new syringe oil droplets and dissolved silanols in the medicine: measured by the proton spectrum below 10 parts per million plunger needle B · WHAT WATER MAKES OF THE OIL H3CSiCH3CH3OSiCH3CH3O silicone oil chain end and one unit of the chain H2O H3CSiCH3CH3OH trimethylsilanol from the chain end + HOSiCH3CH3OH dimethylsilanediol from a chain unit Si–O–Si link cleaved by water

Figure 6. A: a prefilled syringe with the oil layer on the glass and droplets in the medicine. Schematic, not to scale; the amount of oil is the published range for new syringes.[26] B: hydrolysis of silicone oil in an aqueous formulation. Our drawing of the two degradation products named in the source;[3] the stoichiometry is not shown.

In 2019 Joan Malmstrøm of Novo Nordisk in Denmark published a method that records the 1H spectrum of the aqueous formulation taken from the syringe. The spectrum gives the content of silicone oil and of both silanols, also at levels below 10 parts per million, for material intended for clinical use and for samples from stability studies.[3] In syringes with a baked-on coating, dimethylsilanediol was the main silicone compound in the formulation. Its amount increased during storage and with temperature, as expected for hydrolysis of the oil.[3] The method also applies to aqueous samples to which PDMS was added as an antifoam.[3]

Atomic spectrometry is more sensitive. In 2001 Mundry, Surmann and Schurreit, at Schering in Berlin and at the Humboldt University, validated graphite-furnace atomic absorption for silicone oil in eight aqueous drug formulations. The method is linear from 10 to 60 µg/l and has a detection limit of 7 µg/l.[6] An atomic absorption instrument measures silicon; it does not distinguish the oil from a silanol.[1]

The methods answer different questions. Infrared spectroscopy gives all the oil on the glass of an empty syringe.[26] The proton spectrum measures what has entered the liquid and separates the oil from its two hydrolysis products.[3]

E 900 in food

PDMS is the food additive E 900, an antifoaming agent. Annex II to Regulation (EC) No 1333/2008 authorises it in 14 food categories with maximum levels of 10 to 100 mg/kg. Among them are frying oils and fats, jams, confectionery, batters, soups, pineapple juice, flavoured drinks and cider at 10 mg/kg or 10 mg/l, and chewing gum at 100 mg/kg.[21] In the United States the limit is 10 parts per million in the food as consumed, and zero in milk.[29]

EFSA re-evaluated the additive in 2020. Its panel found that more than 99.9 % of an oral dose is excreted unchanged, raised the acceptable daily intake from 1.5 to 17 mg per kg of body weight per day, and saw no safety concern at the reported uses.[21] Two of its recommendations concern analysis: the specification should state the range of molecular weights and set a maximum limit for cyclic siloxanes.[21]

Maximum levels for PDMS as a food additive

Rule and food Maximum level
EU: frying oils and fats, jams, confectionery, batters, soups, pineapple juice, flavoured drinks, cider[21] 10 mg/kg or 10 mg/l
EU: chewing gum[21] 100 mg/kg
USA: food ready for consumption[29] 10 parts per million
USA: milk[29] Zero
USA: dry gelatin dessert mix[29] 110 parts per million, and at most 16 in the dessert as served
USA: salt for cooking[29] 250 parts per million, and at most 10 in the cooked food

EU levels as listed in the EFSA opinion of 2020; US levels from the Code of Federal Regulations as of October 2026.

No official method exists for E 900 in food.[1] The EFSA opinion lists what laboratories have used: flame atomic absorption after solvent extraction for fats, fruit juices and beer, infrared spectroscopy for pineapple and processed vegetables, size-exclusion chromatography, and 1H NMR for wine and edible oils.[21]

A review of 2022 from the Eurecat Centre for Omic Sciences in Reus, written with a commercial analytical laboratory near Barcelona, names the obstacles for these methods. Atomic absorption and plasma spectrometry measure total silicon. Size-exclusion and gas chromatography separate silicon species, but reference substances for cyclic siloxanes of medium and high molecular weight are lacking, and the polydisperse nature of PDMS hinders targeted mass spectrometry.[1]

The method: one standard for extraction and measurement

The Reus group adds bis(trimethylsilyl)benzene (BTMSB) to the food before any work-up. BTMSB is a lipophilic solid, inert to heat, acid and base, and gives a sharp, isolated 1H signal at 0.27 ppm. Standard and PDMS are extracted together with hexane, the solvent is evaporated, and the residue is dissolved in CDCl3.[1] At 500 MHz, the frequency of that work, the 0.20 ppm between the two signals correspond to 100 Hz (our arithmetic; Figure 7).

CH3SiCH3CH3SiCH3CH3H3C 1,4-bis(trimethylsilyl)benzene (BTMSB) internal standard, 18 equivalent protons OSiCH3CH3O chain unit of PDMS 6 equivalent protons in every unit 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 −0.05 1H chemical shift in CDCl3, ppm 0.27 BTMSB 0.07 chain units 0.16 0.11 0.08 0.05 0.03 0.16, 0.11 silanol end, adjacent unit 0.08, 0.03 trimethylsilyl end, adjacent unit 0.05 rings 0.00 TMS, if added (dashed) 0.20 ppm = 100 Hz AT 500 MHz

Figure 7. The internal standard and the analyte between 0.35 and −0.05 ppm. Our drawing: the positions are the published shifts,[1] the heights and widths are schematic. The source names the standard without locants; we draw the 1,4-isomer, which is sold for quantitative NMR.

The validation used four foods with added PDMS: surimi, precooked octopus, canned mussels and strawberry jam. The response was linear up to about 60 mg/kg in the NMR tube. On surimi the extraction recovered 97 to 103 % with a relative standard deviation of 5 %. The limits of detection and quantification were 0.7 and 1.0 mg/kg in the tube. With an extract of 5 g of food the method quantifies down to 0.2 mg/kg.[1]

Fat, saponification and chain ends

Fat is the main interference in the work-up. In 2003 Mojsiewicz-Pieńkowska, Jamrógiewicz and Łukasiak, pharmaceutical chemists in Gdańsk, determined PDMS at 0.06 mg/kg in wine and at 6 mg/kg in edible oils by 1H NMR after extraction and enrichment.[30] The Reus review reports that they removed the triglycerides of the oils with concentrated sodium hydroxide and quantified against hexamethyldisiloxane.[1] When the Reus group saponified a spiked oil by a standard procedure, with ethanolic potassium hydroxide under reflux, part of the PDMS was hydrolysed.[1]

The integral at 0.07 ppm omits chain ends and rings. For PDMS between 0.8 and 30 kDa, the range used in food, the Reus group estimates the resulting error at 3.7 to 0.1 %.[1]

Siloxanes from a baking mould

A silicone baking mould is cross-linked PDMS that still contains oligomers. Between 2009 and 2012 the group of Thomas Simat at TU Dresden measured what passes from the mould into food.[5, 9, 31]

The group characterised 37 moulds by 1H NMR, reversed-phase HPLC and GC. Every mould contained cyclic oligomers with 6 to 50 dimethylsiloxane units; 13 also held linear, partly hydroxyl-terminated chains of 7 to 20. No substances other than siloxanes were detected.[9] The 1H method for the total amount in a food or a simulant has a detection limit of 8.7 mg/kg of oil and a coefficient of variation of 7.8 %.[9]

Cakes took up at most 21 mg/kg, well below the overall migration limit of 60 mg/kg, and the amount fell only slightly from the first to the tenth use; it depended on the fat content of the cake. Migration into the simulants (olive oil, isooctane, 95 % ethanol and a polymer sorbent) differed completely from that into cakes.[9] A meat loaf exceeded the limit in the first and in the third use of a mould, with 177 mg/kg as the highest value; milk-based foods remained below 2.4 mg/kg even at high fat content.[5]

Two factors governed these numbers. The first is tempering, a heat treatment of the mould after its manufacture: moulds that had not received it, with 1.1 % of volatile compounds, released considerably more siloxane than treated ones.[5] The second is use. In a pizza bakery, moulds went through as many as 1,700 baking cycles; the migration fell from between 11 and 18 mg/kg to less than 1 mg/kg, and the volatile content from 0.44 to 0.14 %.[31] No new siloxanes formed in the rubber. The oligomers are side-products of the polymerisation and are present from the beginning.[5, 31]

In 2025 Eckardt, Seidel, Säger and Simat returned to the question with platinum-cured moulds: 1H NMR for the sum of the siloxanes, and coupled liquid and gas chromatography for the single rings D4 to D25. Shortcrust pastry took up 167 mg/kg of siloxanes in the third use of a mould, and chocolate 21.5 mg/kg at no more than 45 °C. From a mould without tempering, roughly 98 % of the ring D5 evaporated into the air during five baking cycles and about 1 % passed into the food. Tempering reduced the migration of the smaller rings; above D20 it had hardly any effect.[32]

In Germany a food-contact silicone is tested by a simpler measure. Pieces of the article are dried for one hour at 100 °C and then heated for four hours at 200 °C; the loss of weight must not exceed 0.5 %.[33] The method carries an expanded uncertainty of 25 % of the result.[33] In 48 silicone materials this loss correlated only weakly with the content of cyclic siloxanes, and moisture taken up from the air disturbed the weighing.[8]

Since the migrants were almost entirely siloxanes, their sum from the 1H spectrum can be compared directly with the overall migration limit (our reasoning).

Dimethicone from a shampoo: the amount left on hair

Dimethicone is added to shampoos to improve the combing of hair. Its effect depends on the amount that stays on the hair after rinsing, a quantity that is hard to measure directly. X-ray fluorescence detects silicon on the fibre but cannot tell a new deposit from silicone already present, and the efficiency of a solvent extraction varies from one hair sample to the next.[2]

In 2014 chemists of an industrial laboratory measured the complementary quantity: the silicone that is rinsed away. They collected the rinse water of a hair tress, about 2 l, evaporated 50 g of it, dissolved the residue in CDCl3 and added p-dinitrobenzene as internal standard. The 1H spectrum at 500 MHz, recorded with a delay of about 20 s, gave the silicone from its signal near 0 ppm. The deposit is the amount applied minus what is found in the water.[2]

For a shampoo with 3 % dimethicone the method recovered 108 % of the silicone, with a relative standard deviation of 1.26 % in three measurements. Among six shampoos with 0 to 5 % dimethicone the deposit grew by about 280 parts per million between 1 and 3 %, by almost 650 between 3 and 4 %, and by only some 110 between 4 and 5 %.[2]

The study appeared in a trade magazine, not in a peer-reviewed journal, and states no detection limit.[2] The older laboratory method for dimethicone in a cream or lotion is infrared spectroscopy after extraction. A paper of 1984 quantified the ingredient against a pharmacopoeial reference material, with a relative standard deviation of 1.78 % and a slight positive bias caused by the other components of the lotion.[34] Part 16 describes other NMR measurements on cosmetics.

Silicone where it is not wanted

Traces in a silica for paints

A German producer of fumed silica described a problem of its customers in a patent granted in 2002. Silica that carries silicone oil at the level of parts per million causes serious adhesion problems in lacquers, paints and adhesives.[10]

Infrared spectroscopy was not sensitive enough, and a 1H spectrum of the powder suspended in a deuterated solvent gave broadened lines that could not be integrated. The patented method removes the solid first. Eight grams of silica are dispersed in 140 g of hexane and centrifuged; the clear liquid is evaporated, and the residue is dissolved in deuterated chloroform. The methylsilicon signal near 0 ppm is integrated at 500 MHz against that of the CHCl3 in the solvent and converted with a calibration from samples of known siloxane content.[10]

Oligomers from a cured laboratory rubber

Microfluidic devices for cell culture are cast from PDMS. In 2009 a group at the University of Wisconsin–Madison detected uncured oligomers by mass spectrometry in the culture medium of such devices, before and after these had been extracted with ethanol, and in the membranes of cells grown in them for 24 hours.[35] Chemists at Harvard had shown in 2003 that solvents which swell PDMS strongly, among them pentane, xylenes and triethylamine, extract such contaminants from the bulk rubber.[36]

The singlet near 0 ppm in a chemist's own spectrum

Synthetic chemists meet silicone as an impurity. Silicone grease from ground-glass joints gives the singlet of Figure 4, and hexamethyldisiloxane lies at the same 0.07 ppm in chloroform (Part 1). A singlet at this position in the spectrum of a product can also come from a silicone septum or from a siliconised syringe (our comment).

Three preparations, one spectrum

The trace methods for food, rinse water and silica differ only in the way the silicone reaches the tube (Figure 8). In each of them the silicone is separated from the bulk of the sample, taken up in deuterated chloroform and measured against a reference signal (our summary).

FOOD 5 g of food + weighed standard extract with hexane evaporate the hexane dissolve in CDCl3 1H spectrum 500 MHz RINSE WATER rinse water of a hair tress, 50 g evaporate the water dissolve in CDCl3 add weighed p-dinitrobenzene 1H spectrum 500 MHz FUMED SILICA 8 g of silica in 140 g of hexane centrifuge, keep the liquid evaporate the hexane dissolve in CDCl3 1H spectrum 500 MHz

Figure 8. Sample preparation in three published methods: E 900 in food,[1] dimethicone in the rinse water of hair,[2] and silicone oil on fumed silica.[10] Our drawing of the published steps.

Published 1H NMR determinations of silicone in products

Product Preparation and standard Reported limit
Aqueous medicine from a prefilled syringe[3] The formu­lation is measured directly; the abstract names no standard Below 10 parts per million
Surimi, octopus, mussels and jam with E 900[1] BTMSB is added, then hexane extraction; residue in CDCl3 LOD 0.7 and LOQ 1.0 mg/kg in the tube; 0.2 mg/kg from 5 g of food
Wine and edible oils[30] Extraction and enrichment; oils saponified with sodium hydroxide; hexa­methyl­disiloxane as standard[1] Determined at 0.06 mg/kg in wine and at 6 mg/kg in edible oil
Cakes, meat loaf and food simulants[5, 9] The tri­glycerides of the extracted fat are the standard[1] LOD 8.7 mg/kg of oil
Pastry and chocolate from platinum-cured moulds[32] 1H NMR for the sum of the siloxanes; single rings by coupled HPLC and GC Not stated in the abstract
Rinse water of hair washed with shampoo[2] The water is evaporated; residue in CDCl3 with p-dinitro­benzene None stated; recovery 108 %
Fumed silica for paints and adhesives[10] Hexane extract evaporated; residue in CDCl3; the CHCl3 signal of the solvent as reference No figure in the passages we read; traces at the level of parts per million

LOD: limit of detection. LOQ: limit of quantification. BTMSB: bis(trimethylsilyl)benzene.

0.001 0.01 0.1 1 10 100 1,000 10,000 mg/kg (aqueous samples: mg/l), logarithmic scale LEGAL AND RECOMMENDED LIMITS Weight loss of a food-contact silicone, 0.5 % 5,000 Each of the rings D4, D5 and D6, 0.1 % 1,000 Overall migration limit, food contact 60 E 900 in most authorised foods, EU 10 FOUND BY 1H NMR Meat loaf from a silicone mould, highest 177 Shortcrust pastry, third use of a mould 167 Cakes from silicone moulds, highest 21 DETECTION AND QUANTIFICATION LIMITS 1H NMR, mould migration: detection 8.7 per kg of oil 1H NMR, syringe formulation: measured below 10 1H NMR, E 900: detection in the tube 0.7 1H NMR, E 900 from 5 g of food 0.2 Atomic absorption, silicone oil in water 0.007

Figure 9. Limits, amounts found and detection limits on one logarithmic scale. Our drawing of published values: weight loss,[33] rings,[22] migration limit and cakes,[9] E 900,[1, 21] meat loaf,[5] pastry,[32] syringe formulation,[3] atomic absorption.[6] The conversions of 0.5 % to 5,000 mg/kg, of 0.1 % to 1,000 mg/kg and of 7 µg/l to 0.007 mg/l are ours.

Si–H and vinyl groups before curing

A liquid silicone becomes a rubber when its chains are joined to one another. In two-part products that cure by addition, one component carries vinyl groups and the curing agent Si–H.[4, 37] The proton spectrum gives both contents in one measurement.

The reaction and its two signals

Liquid silicone rubbers and many release coatings cure by addition: a platinum catalyst adds the Si–H bond of a cross-linker to the vinyl group of another chain, and an ethylene bridge joins the two silicon atoms (Figure 10).[37, 38]

OSiCH3CH3CHCH2 + HSiCH3OO Pt catalyst OSiCH3CH3CH2CH2SiCH3OO vinyl end of a long chain 5.75 to 6.40 ppm Si–H unit of the cross-linker 4.64 ppm ethylene bridge: one cross-link Si–CH2–CH2–Si joins the two chains

Figure 10. The addition cure. Our drawing of the textbook reaction; the chemical shifts are those of the report of 1979, measured in CDCl3.[4]

The ratio of the two groups determines the network. A patent granted in 2026 claims 1 to 5 moles of Si–H per mole of alkenyl groups.[38] In eight commercial liquid silicone rubbers the ratio ranged from slightly above 1.5 to 3.7; the Si–H groups in excess react with one another and lower the molar mass between cross-links.[37] Both groups resonate in regions free of other silicone signals, and each is integrated against a weighed standard.

A titration that counted Si–H as vinyl

In the 1970s the vinyl content of a silicone prepolymer was determined by titration with iodine monochloride (ICl), a reagent that adds to C=C bonds. In March 1979 George Clink, of the development division of the Pantex plant in Texas, compared this titration with 1H NMR on the two parts of two commercial encapsulants (Sylgard 184 and 186) and of a dielectric gel, and on a siloxane fluid rich in Si–H.[4]

Clink dissolved weighed amounts of prepolymer and standard in CDCl3 and recorded the spectra on a Fourier-transform spectrometer. The formyl proton of dimethylformamide at 8.04 ppm was the standard for vinyl and phenyl groups; for Si–H it was p-dioxane.[4]

For Si–H, NMR agreed with the established manometric method, which measures the hydrogen gas evolved from the sample (table below). For vinyl, the two methods agreed only in the parts that contain no Si–H. In the parts with Si–H the titration gave 1.21 to 2.32 % and NMR 0.31 to 0.70 %. The Si–H fluid, whose spectrum shows no olefinic signal, titrated as 3.87 % vinyl (Figure 11).[4]

0 1 2 3 4 vinyl content, % by weight titration with iodine monochloride 1H NMR 0.60 0.51 184 A (31) 0.60 0.51 184 A (36) 0.25 0.27 186 A (47) 0.54 0.44 186 A (64) 0.79 0.63 Gel A (31) 2.32 0.70 184 B (31) 2.14 0.39 184 B (36) 1.91 0.31 186 B (47) 1.93 0.34 186 B (64) 1.21 0.58 Gel B (31) 3.87 none Si–H fluid PARTS WITHOUT Si–H: THE METHODS AGREE MATERIALS WITH Si–H: THE TITRATION READS HIGH

Figure 11. Vinyl content, as CH2=CH–, of eleven silicone prepolymers by two methods. Measured data of 1979, our drawing.[4] A and B are the two parts of each product; the numbers in brackets identify the samples as in the report.

Clink traced the error to the Si–H group, a reducing agent that consumes iodine monochloride. Four pure hydrosilanes without any C=C bond, among them triethylsilane and 2,4,6,8-tetramethylcyclotetrasiloxane, titrated as 2.69 to 3.72 % vinyl. Phenyl groups on silicon raised the iodine number as well.[4]

The NMR value had a bias of its own. On vinyl siloxanes of high purity it reached 92 to 96 % of the true content, and Clink noted that an integral over twelve lines spread across 0.65 ppm is not highly accurate.[4]

Hydrogen bonded to silicon, % by weight, in six prepolymers (1979)

Material Evolved hydrogen, manometric 1H NMR, mean of three
Sylgard 184, part B (31) 0.456 0.45 ± 0.01
Sylgard 184, part B (36) 0.442 0.44 ± 0.01
Sylgard 186, part B (47) 0.088 0.10 ± 0.02
Sylgard 186, part B (64) 0.097 0.11 ± 0.01
Dielectric gel, part B (31) 0.180 0.19 ± 0.03
Si–H fluid 1.575 1.60 ± 0.15

Data of the report.[4] The value after ± is the standard deviation given there. A chain of CH3(H)SiO units alone contains 1.68 % of such hydrogen (our arithmetic), close to the value of the fluid.

In molar terms the first curing agent of the table holds 0.45 mol of Si–H and, from its 0.70 % by weight, 0.026 mol of vinyl per 100 g: seventeen to one (our arithmetic). A titration that responds to both cannot measure the smaller number.

The same two numbers in a patent of 2026

The measurement is routine in industry today. A patent on release coatings for paper and film, granted to a Japanese silicone producer in April 2026, states that the vinyl value and the Si–H content of its raw materials are determined by 1H NMR, and reports both in mol per 100 g.[38]

Its vinyl-terminated polymer with 160 dimethylsiloxane units has a vinyl value of 0.0166 mol/100 g. Its methylhydrogenpolysiloxane, a chain of 100 CH3(H)SiO groups, has an Si–H content of 1.62 mol/100 g.[38] Both figures follow from the formulae: 2 vinyl groups in 12,051 g/mol and 100 Si–H bonds in 6,175 g/mol (our arithmetic).

The first polymer shows the range of intensity that the spectrum must cover. Its 6 vinyl protons stand against 972 in methyl groups, a ratio of 1 to 162 (Figure 12).

H2CCHSiCH3CH3OSiCH3CH3OSiCH3CH3CHCH2 160 2 vinyl groups: 6 protons 324 methyl groups: 972 protons 0 1 2 3 4 5 6 7 1H chemical shift, ppm 972 H Si–CH3, 0.07 ppm 6 H at the same scale VERTICAL SCALE × 100 6.40 5.75 vinyl : methyl = 6 : 972 = 1 : 162 the vinyl signal is 0.6 % of the methyl signal

Figure 12. A vinyl-terminated PDMS with 160 chain units and the relative areas of its 1H signals. Our calculation from the formula in the patent;[38] line shapes and the pattern of the twelve vinyl lines are schematic, and the shifts are those of Figure 3.

Commercial rubbers before curing

In 2012 Delebecq and Ganachaud, in work from a doctoral thesis in Montpellier, analysed eight commercial liquid silicone rubbers in their uncured form. 29Si NMR gave the architecture of the polymers, size-exclusion chromatography in toluene their molar mass, and 1H NMR the amounts of hydride and vinyl groups.[37] The hardness of the products, 20 to 30 Shore A, was adjusted mainly through the filler content and the molar mass between cross-links.[37]

Size-exclusion chromatography of PDMS is run in toluene or with an evaporative light-scattering detector. The polymer has the same refractive index as tetrahydrofuran, the usual eluent, and a refractive-index detector cannot be used with that solvent, as an instrument maker explains in an application note.[39] Part 17 describes how NMR gives a chain length from end groups without a detector response.

Cured silicone: the network

A cured silicone swells in a solvent and does not dissolve. Neither chromatography nor a solution spectrum describes its network; the classical measures are swelling experiments and mechanical measurements.[13]

What the proton signal of a network contains

Proton NMR measures the network through the motion of its chains. In a liquid polymer the segments of a chain turn in all directions, and the magnetic coupling between neighbouring protons averages to zero (our explanation). Cross-links, entanglements and contacts with filler particles restrict this motion, and a residual dipolar coupling remains.[13] In 2012 Chassé, Lang, Sommer and Saalwächter confirmed for PDMS that this coupling is proportional to the reciprocal of the molar mass between cross-links, with systematic errors of its own.[11]

The experiment is a 1H double-quantum measurement, also called multiple-quantum NMR. It records the rise of a signal that exists only between coupled protons (Figure 13). Dangling chain ends and loops are not elastically active, show no coupling and are counted separately.[13] The polymer institute of the Spanish research council in Madrid runs the experiment on a time-domain instrument at 20 MHz with a 90° pulse of 2.2 µs.[13] Such an instrument evaluates the signal in time; it is a different tool from the benchtop spectrometers of Part 18.

filler particle cross-link chain between two cross-links: residual coupling dangling end: no coupling loop: no coupling A · THE NETWORK evolution time double-quantum signal dense network strong coupling loose network weak coupling B · THE MEASURED CURVE

Figure 13. A: the parts of a rubber network. B: the double-quantum signal rises sooner when the residual coupling is stronger. Schematic, not data; the terms follow the Madrid paper.[13]

Four studies on silicone networks

The study of 2012, from the group of Kay Saalwächter in Halle, used two series of randomly cross-linked PDMS with a defined content of dangling chains. The authors compared low-field double-quantum NMR, equilibrium swelling and magic-angle-spinning NMR (Part 11), which gave the conversion of the cross-linking reaction. With the fraction of defects taken into account, the methods agreed well; the remaining deviations of the swelling results were attributed to the Flory–Rehner theory used to evaluate them.[11]

At Lawrence Livermore National Laboratory, Maxwell and co-workers irradiated a peroxide-cured silicone rubber, filled with about 30 % of quartz and silica, with gamma doses of 5 to 250 kGy. They ran the multiple-quantum experiment on a 400 MHz spectrometer with a 5 mm probe. The data required two populations of chains (Figure 14). The residual coupling of the larger one rose from about 700 to 1,100 rad/s over this range and that of the smaller from 3,200 to 3,900 rad/s, while the share of this second population grew from 11 to 25 %. The authors assign the strongly coupled chains to polymer at the filler surface or to regions of higher cross-link density.[12]

0 50 100 150 200 250 0 1,000 2,000 3,000 4,000 gamma dose, kGy residual coupling, rad/s strongly coupled chains bulk of the network A · RESIDUAL COUPLING OF THE TWO POPULATIONS 0 50 100 150 200 250 0 10 20 30 gamma dose, kGy share of strongly coupled chains, % 25 % 11 % B · SHARE OF THE STRONGLY COUPLED POPULATION

Figure 14. A filled silicone rubber after gamma irradiation. A: residual dipolar coupling of the two chain populations. B: share of the strongly coupled population. Measured data of 2005, our drawing. The first and the last value of each series stand in the text of the preprint, which we read as page images; the values between them come from its table, extracted with a retrieval tool.[12]

In a second rubber, filled with 30 % of fumed silica and exposed to doses of up to 250 kGy, the shear modulus rose from 4 to 8.1 MPa; swelling and spin-echo NMR likewise indicated that the cross-link density had about doubled. Multiple-quantum NMR again showed two populations, in a ratio of about 90:10. The larger lies near 1 krad/s; for the smaller the preprint gives 3 in its text and 5 in its abstract, in the same unit. The volatile products pointed to back-biting, the reaction that splits rings from a chain.[40]

Chains at a filler surface are less mobile than the rest. In 2002 Litvinov, of DSM Research in the Netherlands, and two co-authors studied PDMS bound to the surface of silica by relaxation measurements and by 1H and 29Si spectra. The bound chains were loops of four to eight Si–O bonds on average, attached at both ends, and about four chain units next to each point of attachment were immobilised.[41]

A solution spectrometer contributes where part of the material dissolves. The oligomers extracted from a cured article are an ordinary sample, as the work on baking moulds shows, and the Si–H and vinyl groups are measured before curing (our reasoning).

Breast implants: the gel, the shell and what leaves them

The implants discussed here are shells of silicone rubber filled with a gel of the same polymer. NMR has been used for two questions: what the materials consist of, and whether they change in the body.

In 1995 Dorne, Alikacem, Guidoin and Auger recorded high-resolution solid-state 29Si spectra of the gels from unused and explanted breast prostheses. The spectra show the D unit of the chains at −22.26 ppm. Within a detection limit of 0.2 % they show no signal between −90 and −150 ppm, and the authors concluded that the gels contained no silica. The spectra also revealed differences in composition between manufacturers, and no change in the chemical structure of the gel after implantation.[16]

In 2003 a group at the University of Münster examined unused and explanted implants. Solid-state spectra of the shells gave only limited information; high-resolution measurements in solution were more sensitive. With COSY (Part 3) and a 29Si–1H HMBC experiment the authors detected lipids, mainly phospholipids, and silicone oligomers in explanted shells and gels. They found silicone oligomers in the tissue capsule around the implant as well.[15]

A study by the manufacturer of two older implant models, published in 2008, compared explants with unused controls of the same type. Spectra and the transverse relaxation times of the protons showed no hydrolysis or other degradation of the cross-linked silicone after 32 and 13 years in the body. The authors state that their interpretation differs from a report of the Münster laboratory and point to the controls and the sample preparation.[42]

Between 2010 and 2012 the implants of one French manufacturer were withdrawn from the market, because their filler was not medical-grade silicone.[43] At the University of Milan, Beretta and Malacco compared the gel of such explants with that of an unused implant of another make by rheology, infrared spectroscopy, 1H NMR and mass spectrometry. The gel of the withdrawn product lacked a significant part of the cross-linking sites, and it had absorbed cholesterol from the tissue.[44]

After the withdrawal, Formes and Diehl classified the implants of eleven manufacturers by 1H NMR in 2014. The olefinic signal showed whether the vinyl groups are terminal or lie within the chain, and how many remained. With the structure of the shell material, these features sorted the products into four types.[14]

The EU scientific committee SCENIHR collected the chemical analyses of national authorities in its opinion of 2014 (table below). The gel of the withdrawn implants held several hundred parts per million each of D5 and D6, the gels of three other brands between 20 and 132.[17] The committee judged the three rings to be of low toxicity and at most very weak irritants, and expected no health risk from the higher exposure after a rupture. It estimated that 25 to 30 % of the withdrawn implants rupture within ten years, against 2 to 15 % for other makes.[17] The opinion cites analyses by GC–MS and does not mention NMR.[17]

Rings in the gel of breast implants, parts per million

Gel and laboratory D4 D5 D6
Withdrawn implants, Australian authority, median 136 434 474
Withdrawn implants, Swedish authority, highest value 134 457 604
Same manufacturer with medical-grade gel, Swedish authority Not detected 18 30
Two other brands, Swedish authority Not detected 20 22
A third brand, Swedish authority 30 72 132

From Table 4 of the committee opinion.[17] Every value lies below 1,000 parts per million, the 0.1 % of the general limit described in the next section (our comparison).

In three explants attributed to the same manufacturer and analysed in Amsterdam by NMR, near-infrared spectroscopy and GC–MS, each of the three rings was below 100 parts per million.[43]

The proton signal of silicone also serves in the clinic. Magnetic resonance imaging with suppression of the water signal shows the silicone of an implant. In a study of 1994 on 81 removed implants, 18 of them ruptured, a sequence of this kind identified ruptures more often than a three-point Dixon technique.[45]

NMR measurements on cured silicones and implants

Material and source Experiment Result
Model PDMS networks with defined defects, Halle 2012[11] Low-field 1H double-quantum NMR, swelling, magic-angle-spinning NMR Residual coupling proportional to the reciprocal molar mass between cross-links; the methods agree once defects are counted
Filled rubber after 5 to 250 kGy, Livermore 2005[12] 1H multiple-quantum NMR at 400 MHz, swelling Two chain populations; the strongly coupled share grew from 11 to 25 %
Filled commercial rubber after 250 kGy, Livermore 2009[40] 1H multiple-quantum NMR, spin-echo NMR, swelling, mechanical analysis Shear modulus from 4 to 8.1 MPa; two chain populations, about 90:10
Gels of breast prostheses, 1995[16] Solid-state 29Si NMR D units at −22.26 ppm; no silica above 0.2 %
Unused and explanted implants, Münster 2003[15] Solution 1H, 13C and 29Si NMR, COSY, 29Si–1H HMBC; solid-state NMR Lipids in explanted shells and gels; silicone oligomers in the tissue capsule
Implants of eleven manufacturers, 2014[14] 1H NMR Four types, by the position and the remaining amount of vinyl groups and by the shell material

Oligomers and the rings D4, D5 and D6

Where the rings come from

Rings accompany silicones from the start. They can form during the manufacture of the polymer and contaminate the product.[21] They form again when a silicone is heated strongly: Camino, Lomakin and Lazzari at the University of Turin reported in 2001 that PDMS decomposes thermally to cyclic oligomers, through the scission of an Si–O bond in a folded section of the chain (Figure 15).[46]

folded section of the chain an Si–O bond breaks here and the ends rejoin heat shorter chain + SiOSiOSiOSiO ring, drawn here as D4

Figure 15. Thermal formation of a ring from a silicone chain. Schematic after the mechanism named in the abstract of the Turin paper;[46] the source reports cyclic oligomers in general, and D4 is drawn as one example.

Separation by diffusion

A mixture of siloxane oligomers gives 1H signals that overlap; the 29Si spectrum resolves the species at far lower sensitivity. In 2004 Stchedroff, Kenwright, Morris, Nilsson and Harris, at the Universities of Manchester and Durham, addressed this dilemma with DOSY (diffusion-ordered spectroscopy, Part 17). They compared 1H DOSY, 29Si DOSY and a three-dimensional 1H{29Si} experiment on a mixture of cyclic dimethylsiloxanes with 4 to 20 units, and found that the diffusion coefficient falls with the inverse square root of the molecular mass.[18] A ring of 16 units therefore diffuses half as fast as D4 (our arithmetic; Figure 16).

SiOSiOSiOSiO D4 4 units · 296.6 g/mol SiOSiOSiOSiOSiO D5 5 units · 370.8 g/mol SiOSiOSiOSiOSiOSiO D6 6 units · 444.9 g/mol A · THE THREE REGULATED RINGS 4 8 12 16 20 0.4 0.6 0.8 1.0 dimethylsiloxane units in the ring diffusion coefficient relative to D4 16 units: 0.50 D6: 0.82 B · DIFFUSION AGAINST RING SIZE

Figure 16. A: the rings D4, D5 and D6; every line that leaves a silicon atom ends in a methyl group. Our drawing; molar masses are our arithmetic. B: diffusion coefficient of cyclic dimethylsiloxanes relative to D4. Our calculation from the published power law,[18] not measured points.

Diffusion also identifies a silicone signal. A group at the Instituto Politécnico Nacional in Mexico City extracted five wines and measured the extracts by DOSY at 600 MHz. The methylsilicon signal diffused at about 2.1 × 10−10 m2/s, nine times more slowly than TMS, as expected for a polymer. The conference paper of 2023 reports no validation data.[24]

The limit of 0.1 % for each ring

On 13 June 2018 the Member State Committee of the European Chemicals Agency identified D4, D5 and D6 as substances of very high concern on account of their persistence and bioaccumulation. Cosmetic products were the main source of their release.[22]

Entry 70 of Annex XVII to the REACH Regulation, as replaced in 2024, prohibits placing D4, D5 and D6 on the EU market as substances, as constituents of other substances or in mixtures at 0.1 % by weight or more of the respective ring.[22] Mixtures that contain the rings as residues of a silicone polymer have higher limits for named uses, for example 1 % for adhesion, sealing, gluing and casting.[22]

13 June 2018 identified as substances of very high concern 31 January 2020 D4 and D5 in wash-off cosmetic products 6 June 2026 substances, constituents and mixtures in general 6 June 2027 all other cosmetic products 6 June 2031 medical devices and medicinal products 6 June 2034 D5 as a solvent in dry cleaning passed when this article was written (November 2026) still ahead

Figure 17. The rule on D4, D5 and D6 in time: the identification of 2018 and the dates after which the limit of 0.1 % applies. Our drawing from the recitals and the Annex of the amending regulation;[22] industrial uses such as the production of silicone polymers are exempt.

Compliance is tested by gas chromatography with flame-ionisation detection (GC-FID). The analysis has a known source of error: cyclic siloxanes can form during the measurement and give false positives. Brothers and co-authors, from the laboratories of a consortium of manufacturers, silylate the sample to reduce this formation, in silicone emulsions (2016) and, after breaking the emulsion and a liquid–liquid extraction, in personal care products (2017).[7, 19] For emulsions the results differed from those of a reference method by liquid chromatography–mass spectrometry by 0.03 % by weight or less, and four sites agreed within a relative standard deviation of 10 %.[7] The second paper shows false positives obtained with an unsuitable sample preparation.[19]

In 2025 Eckardt and co-workers, at a commercial laboratory in Cologne and at TU Dresden, published a GC-FID method for linear siloxanes of 4 to 24 units and rings of 4 to 25 in silicone elastomers for food and skin contact. They validated it against 1H NMR and thermal-desorption GC–MS.[8] All 48 materials of the study contained rings; linear oligomers with trimethylsilyl ends were not detected. The authors propose the method as an alternative to the weight-loss test.[8]

In that study 1H NMR was an independent reference. The 1H signals of rings and chains lie 0.02 ppm apart,[1] and the spectrum does not distinguish D4, D5 and D6 from one another (our reasoning). The 29Si spectrum separates D4 from the chains (Figure 5); for D5 and D6 we have no values.

NMR beside the other methods

What each method measures on a silicone

Method What it measures Limit
Atomic absorption and plasma spectro­metry Total silicon, at trace level[6] No distinc­tion between the oil, a silanol and other silicon com­pounds[1]
Infrared spectro­scopy The total silicone on a syringe;[26] di­methicone in a lotion after extraction[34] A slight positive bias from the other ingredients;[34] too insen­sitive for traces on silica[10]
Gas chromato­graphy Single volatile siloxanes; the test for D4, D5 and D6[7, 8] Rings can form during the analysis[7, 19]
Size-exclusion chromato­graphy The distri­bution of chain lengths[21, 39] PDMS and tetra­hydro­furan have the same refrac­tive index[39]
Weight loss at 200 °C The sum of volatile com­pounds in a rubber[33] Correlates only weakly with the content of rings[8]
Swelling and mech­anical tests The cross-link density of a rubber[13] The evaluation of swelling rests on a theory with known short­comings[11]
1H NMR All methyl­silicon protons against a weighed standard; Si–H and vinyl groups; the oil apart from its silanols[1, 3, 4] Detection limit of 0.7 mg/kg in the tube;[1] the rings D4, D5 and D6 are not told apart

Practical points for a quantitative silicone spectrum

  • Reference. The solvent should contain no TMS, or very little: its signal lies 0.03 ppm from that of the unit next to a trimethylsilyl end.[1]
  • Standard. BTMSB at 0.27 ppm is clear of all silicone signals and can be added before the extraction.[1]
  • Delay. At 80 MHz the methyl protons of siloxanes have longitudinal relaxation times (T1) between 1.4 and 2.9 s (table below). The Gdańsk group set the delay to 15 s, five times the longest T1, with an acquisition time of 5 s.[25] Part 9 explains the reason.
  • Blank. Greased joints, silicone septa and siliconised syringes must not touch the sample, and a blank passes through the whole preparation (our recommendation).
  • Work-up. Hexane recovered PDMS from food slightly better than chloroform or diethyl ether.[1]
  • Integration. A report should state whether the signals of chain ends and rings are included.

Relaxation times of the methyl protons of siloxanes

Compound T1, s
Hexamethyldisiloxane 2.38 to 2.94
PDMS, 10 cSt (1,200 Da) 2.44
PDMS, 300 cSt (12,000 Da) 1.40 to 1.68
PDMS, 1,000 cSt (30,000 Da) 1.67
Hexamethylcyclotrisiloxane (D3) 2.20
Octamethylcyclotetrasiloxane (D4) 2.06

T1: longitudinal relaxation time, measured at 80 MHz in mixtures of carbon tetrachloride and benzene-d6. Mean values by inversion recovery. The ranges cover two solvent ratios and, for hexamethyldisiloxane, three concentrations and pure benzene-d6.[25] cSt: centistokes, the viscosity grade. T1 depends on the field, so the values are a guide for other spectrometers (our comment).

Terms used in this article

Twenty terms in plain language

Term Meaning
Addition cure Hardening of a silicone by the reaction of Si–H groups with vinyl groups, started by a platinum catalyst
Antifoam A substance that prevents or destroys foam in a liquid during processing
Centistokes, cSt The unit of kinematic viscosity by which silicone oils are graded
Chemical shift The position of an NMR signal, in ppm of the measuring frequency
Derivatisation Chemical conversion of a substance before a measurement, for example to make it volatile for gas chromatography
DOSY An NMR experiment that sorts signals by the speed at which the molecules diffuse; large molecules move slowly
Filler Fine solid particles, such as silica, mixed into a rubber
Hydrolysis Cleavage of a bond by water; in a silicone, of the Si–O–Si link
Internal standard A pure compound added to the sample in a known amount; its signal is the measure for all others
kGy, kilogray The unit of an absorbed dose of radiation
Limit of detection, limit of quantification The smallest amount that a method detects, and the smallest that it measures reliably
Oligomer A short chain or ring of a few to some tens of units
Overall migration limit The largest total amount of substances that a food-contact material may release into food: 60 mg per kg of food in the studies cited here
Prepolymer A liquid polymer that is intended to be cured into a rubber
Relaxation time, T1 The time that nuclei need to return to equilibrium after a pulse; the delay between two pulses must be several times longer
Residual dipolar coupling The part of the magnetic interaction between neighbouring protons that the motion of a chain does not average to zero[13]
Saponification Cleavage of fats by a strong base; it turns them into water-soluble soaps
Silanol A compound with an Si–OH group, formed when water cleaves a siloxane
Simulant A liquid or solid that replaces food in a migration test, for example olive oil or 95 % ethanol
Titration Determination of an amount from the volume of a reagent solution that reacts with it

Summary

  • A silicone is a chain of –Si(CH3)2–O– units: a liquid as an oil, volatile as a small ring, insoluble once cross-linked. A technical product is always a mixture of chain lengths.
  • NMR counts protons. A weighed standard in the tube converts the area of the silicone signal into milligrams, without a reference sample of the polymer.
  • The methyl protons of PDMS give one signal at 0.07 ppm in CDCl3, where extracts of food usually show nothing; in benzene-d6 it lies at 0.29 ppm.
  • A 1H spectrum of the liquid from a prefilled syringe gave silicone oil, trimethylsilanol and dimethylsilanediol below 10 parts per million without derivatisation; atomic absorption detects 7 µg/l, as total silicon.
  • The EU authorises E 900 at 10 mg/kg in most of its food categories, and no official method exists. A hexane extract with an internal standard allowed quantification down to 0.2 mg/kg from 5 g of food.
  • Siloxanes that migrated from silicone moulds reached 21 mg/kg in cakes, 177 mg/kg in a meat loaf and 167 mg/kg in shortcrust pastry, above the overall migration limit of 60 mg/kg. Tempering and repeated use lower the migration.
  • The dimethicone that a shampoo leaves on hair was obtained from the 1H spectrum of the rinse water, with a recovery of 108 %.
  • Si–H at 4.64 ppm and vinyl at 5.75 to 6.40 ppm are counted by 1H NMR. A titration with iodine monochloride counted Si–H as vinyl and reported 3.87 % in a fluid that has none.
  • In a vinyl-terminated chain of 160 units the olefinic protons give 0.6 % of the methyl signal.
  • A cured rubber does not dissolve. Its network is examined by 1H double-quantum NMR; in a filled rubber the share of strongly coupled chains grew from 11 to 25 % between 5 and 250 kGy of radiation.
  • In breast implants NMR identified the type of gel, lipids taken up from the body and silicone oligomers outside the implant. The rings D4, D5 and D6 in the gels, determined by other methods, reached 604 parts per million at most.
  • DOSY separates siloxane oligomers by size: the diffusion coefficient falls with the square root of the molecular mass.
  • Cyclic siloxanes lie 0.02 ppm from the chain signal in the proton spectrum; in the 29Si spectrum the ring D4 lies 2.14 ppm from it.
  • The limit of 0.1 % for each of D4, D5 and D6 is tested by GC-FID; silylation reduces the formation of rings during the analysis.
  • For a quantitative spectrum: no TMS, a delay of 15 s, a blank for silicone grease, and a statement on chain ends.

Silicone analysis by NMR from NorrChemica

Our NMR service measures samples that dissolve, at 400 MHz: a 1H spectrum is Package I, and a quantitative measurement against a calibrated standard is Package V. DOSY is quoted per sample, and spectra at 600 MHz and above are available on request. 29Si is not on the published list, and we do not measure cured rubber. For silicone in a product, or for Si–H and vinyl groups in a prepolymer, please describe the sample, the question and the expected level on the request form; we will reply with what can be measured. A chemist assigns each spectrum, and the signed report includes the native and processed data files. Research use only.

Request NMR analysis → About the NMR service

NMR tubes made in Germany by Deutero and shipped from Finland. See the NMR tube guide →

References

  1. Palacios-Jordan H, Jané-Brunet A, Jané-Brunet E, Puiggròs F, Canela N, Rodríguez MA. Considerations on the analysis of E-900 food additive: an NMR perspective. Foods 2022, 11, 297. DOI
  2. Huang Q, Mei Z-W, Takata K, Yang J. Analyzing deposition from rinse-off hair products. Cosmetics & Toiletries, online 23 April 2014. Article
  3. Malmstrøm J. Quantification of silicone oil and its degradation products in aqueous pharmaceutical formulations by 1H-NMR spectroscopy. J. Pharm. Sci. 2019, 108, 1512–1520. DOI
  4. Clink GL. FT-NMR analysis of silane hydrogen, phenyl, and vinyl content of several siloxane pre-polymer materials. Report MHSMP-79-15, Development Division, Pantex Plant, March 1979. OSTI 6166384
  5. Helling R, Kutschbach K, Simat TJ. Migration behaviour of silicone moulds in contact with different foodstuffs. Food Addit. Contam. Part A 2010, 27, 396–405. DOI
  6. Mundry T, Surmann P, Schurreit T. Trace analysis of silicone oil in aqueous parenteral formulations and glass containers by graphite furnace atomic absorption spectrometry. Pharm. Ind. 2001, issue 3, p. 301. Abstract
  7. Brothers HM Jr, Bovens E, Bruni A, Habitz TM, Hamachi T, Han Y, Ji Z, Kerbleski JJ, Letouche C, Lu YD, Nguyen R, Rivard ML, Qi X, Shoji M, Tanaka K, Tecklenburg RE. A practical gas chromatography flame ionization detection method for the determination of octamethylcyclotetrasiloxane, decamethylcyclopentasiloxane, and dodecamethylcyclohexasiloxane in silicone emulsions. J. Chromatogr. A 2016, 1441, 116–125. DOI
  8. Eckardt M, Seidel A-K, Säger S, Simat TJ. An alternative approach for compliance testing of food and skin contact silicone elastomers by specific determination of volatile cyclic siloxane oligomers via GC-FID. Talanta 2025, 283, 127145. DOI
  9. Helling R, Mieth A, Altmann S, Simat TJ. Determination of the overall migration from silicone baking moulds into simulants and food using 1H-NMR techniques. Food Addit. Contam. Part A 2009, 26, 395–407. DOI
  10. Michael G, Müller A. Method for determining silicone concentrations. US Patent 6,436,715 B1, Degussa-Hüls AG, granted 20 August 2002. Google Patents
  11. Chassé W, Lang M, Sommer J-U, Saalwächter K. Cross-link density estimation of PDMS networks with precise consideration of networks defects. Macromolecules 2012, 45, 899–912. DOI
  12. Maxwell RS, Chinn SC, Solyom D, Cohenour R. Radiation-induced cross-linking in a silica-filled silicone elastomer as investigated by multiple quantum 1H NMR. Macromolecules 2005, 38, 7026–7032. DOI · Preprint
  13. Salamanca FM, Zepeda-Rodríguez Z, Diñeiro L, Escrivá MM, Herrero R, Navarro R, Valentín JL. Introducing “MEW2” software: a tool to analyze MQ-NMR experiments for elastomers. Polymers 2023, 15, 4058. DOI
  14. Formes A, Diehl B. Investigation of the silicone structure in breast implants using 1H NMR. J. Pharm. Biomed. Anal. 2014, 93, 95–101. DOI
  15. Birkefeld AB, Bertermann R, Eckert H, Pfleiderer B. Liquid- and solid-state high-resolution NMR methods for the investigation of aging processes of silicone breast implants. Biomaterials 2003, 24, 35–46. DOI
  16. Dorne L, Alikacem N, Guidoin R, Auger M. High resolution solid-state 29Si NMR spectroscopy of silicone gels used to fill breast prostheses. Magn. Reson. Med. 1995, 34, 548–554. DOI
  17. Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR). Safety of Poly Implant Prothèse (PIP) silicone breast implants (2013 update). European Commission, opinion adopted on 12 May 2014. DOI 10.2772/66097. European Commission
  18. Stchedroff MJ, Kenwright AM, Morris GA, Nilsson M, Harris RK. 2D and 3D DOSY methods for studying mixtures of oligomeric dimethylsiloxanes. Phys. Chem. Chem. Phys. 2004, 6, 3221–3227. DOI
  19. Brothers HM Jr, Boehmer T, Campbell RA, Dorn S, Kerbleski JJ, Lewis S, Mund C, Pero D, Saito K, Wieser M, Zoller W. Determination of cyclic volatile methylsiloxanes in personal care products by gas chromatography. Int. J. Cosmet. Sci. 2017, 39, 580–588. DOI
  20. Mojsiewicz-Pieńkowska K, Jamrógiewicz M, Szymkowska K, Krenczkowska D. Direct human contact with siloxanes (silicones) – safety or risk. Part 1. Characteristics of siloxanes (silicones). Front. Pharmacol. 2016, 7, 132. DOI
  21. EFSA Panel on Food Additives and Flavourings (FAF); Younes M, Aquilina G, Castle L, et al. Re-evaluation of dimethyl polysiloxane (E 900) as a food additive. EFSA J. 2020, 18(5), 6107. DOI
  22. Commission Regulation (EU) 2024/1328 amending Annex XVII to Regulation (EC) No 1907/2006 (REACH) as regards octamethylcyclotetrasiloxane (D4), decamethylcyclopentasiloxane (D5) and dodecamethylcyclohexasiloxane (D6). Official Journal of the European Union, L series, 2024. EUR-Lex
  23. Fulmer GR, Miller AJM, Sherden NH, Gottlieb HE, Nudelman A, Stoltz BM, Bercaw JE, Goldberg KI. NMR chemical shifts of trace impurities: common laboratory solvents, organics, and gases in deuterated solvents relevant to the organometallic chemist. Organometallics 2010, 29, 2176–2179. DOI
  24. Herbert-Pucheta JE, Hernández-Rangel ÁO, Vargas-Díaz ME, Hernández Sánchez K, Zepeda-Vallejo LG, Jiménez-García M. Silicone-specific identification of trace polydimethylsiloxanes in wines with 2D-diffusion-ordered nuclear magnetic resonance spectroscopy (DOSY-NMR). BIO Web Conf. 2023, 68, 02004. DOI
  25. Jamrógiewicz Z, Mojsiewicz-Pieńkowska K, Jachowska D, Łukasiak J. Study on the dependence of longitudinal relaxation time in 1H NMR method on the structure of polydimethylsiloxanes and optimization of spectra registration parameters (in Polish). Polimery 2005, 50, 737–741. DOI
  26. Bee JS, Frey VV, Javed U, Chung J, Corcoran ML, Roussel PS, Krause SO, Cash PW, Bishop SM, Dimitrova MN. Characterization of the initial level and migration of silicone oil lubricant in empty prefilled syringes for biologics using infrared spectroscopy. PDA J. Pharm. Sci. Technol. 2014, 68, 494–503. DOI
  27. Jones LS, Kaufmann A, Middaugh CR. Silicone oil induced aggregation of proteins. J. Pharm. Sci. 2005, 94, 918–927. DOI
  28. Thirumangalathu R, Krishnan S, Ricci MS, Brems DN, Randolph TW, Carpenter JF. Silicone oil- and agitation-induced aggregation of a monoclonal antibody in aqueous solution. J. Pharm. Sci. 2009, 98, 3167–3181. DOI
  29. US Code of Federal Regulations, Title 21, § 173.340, Defoaming agents. Electronic Code of Federal Regulations, text as of 19 October 2026. eCFR
  30. Mojsiewicz-Pieńkowska K, Jamrógiewicz Z, Łukasiak J. Determination of polydimethylsiloxanes by 1H-NMR in wine and edible oils. Food Addit. Contam. 2003, 20, 438–444. DOI
  31. Helling R, Seifried P, Fritzsche D, Simat TJ. Characterisation and migration properties of silicone materials during typical long-term commercial and household use applications: a combined case study. Food Addit. Contam. Part A 2012, 29, 1489–1500. DOI
  32. Eckardt M, Seidel AK, Säger S, Simat TJ. Migration of cyclic dimethyl siloxanes from silicone elastomers into food – determination of D4 to D25 via on-line coupled HPLC-GC-FID. Food Addit. Contam. A 2025, 42, 1764–1782. DOI
  33. Determination of volatile compounds in silicone consumer products. Method published by the German Federal Institute for Risk Assessment (BfR), version of March 2022. PDF
  34. Sabo M, Gross J, Rosenberg IE. Quantitation of dimethicone in lotions using Fourier transform infrared spectral subtraction. J. Soc. Cosmet. Chem. 1984, 35, 273–281. Journal archive
  35. Regehr KJ, Domenech M, Koepsel JT, Carver KC, Ellison-Zelski SJ, Murphy WL, Schuler LA, Alarid ET, Beebe DJ. Biological implications of polydimethylsiloxane-based microfluidic cell culture. Lab Chip 2009, 9, 2132–2139. DOI
  36. Lee JN, Park C, Whitesides GM. Solvent compatibility of poly(dimethylsiloxane)-based microfluidic devices. Anal. Chem. 2003, 75, 6544–6554. DOI
  37. Delebecq E, Ganachaud F. Looking over liquid silicone rubbers: (1) network topology vs chemical formulations. ACS Appl. Mater. Interfaces 2012, 4, 3340–3352. DOI
  38. Ihara T, Kobayashi A, Kimura T. Organopolysiloxane composition for use in release paper or release film. US Patent 12,612,494 B2, Shin-Etsu Chemical Co., Ltd., granted 28 April 2026. Google Patents
  39. Agilent Technologies. Polysiloxane analysis on Agilent PLgel 5 µm MIXED-D using GPC/SEC with RI and ELS detection. Application note 5990-7897EN, 2018. PDF
  40. Maxwell RS, Chinn SC, Alviso CT, Harvey CA, Giuliani JR, Wilson TS, Cohenour R. Quantification of radiation induced crosslinking in a commercial, toughened silicone rubber, TR-55, by 1H MQ-NMR. Polym. Degrad. Stab. 2009, 94, 456–464. DOI · Preprint
  41. Litvinov VM, Barthel H, Weis J. Structure of a PDMS layer grafted onto a silica surface studied by means of DSC and solid-state NMR. Macromolecules 2002, 35, 4356–4364. DOI
  42. Taylor RB, Eldred DE, Kim G, Curtis JM, Brandon HJ, Klykken PC. Assessment of silicone gel breast implant biodurability by NMR and EDS techniques. J. Biomed. Mater. Res. A 2008, 85, 684–691. DOI
  43. Bachour Y, Heinze Z, van Selms G, Ritt M, Niessen F, Keizers P. Poly Implant Prothèse silicone breast explants: chemical analysis of silicone gel and implant shell. Plast. Reconstr. Surg. Glob. Open 2019, 7, e2093. DOI
  44. Beretta G, Malacco M. Chemical and physicochemical properties of the high cohesive silicone gel from Poly Implant Prothèse (PIP) breast prostheses after explantation: a preliminary, comparative analytical investigation. J. Pharm. Biomed. Anal. 2013, 78–79, 75–82. DOI
  45. Gorczyca DP, Schneider E, DeBruhl ND, Foo TK, Ahn CY, Sayre JW, Shaw WW, Bassett LW. Silicone breast implant rupture: comparison between three-point Dixon and fast spin-echo MR imaging. Am. J. Roentgenol. 1994, 162, 305–310. DOI
  46. Camino G, Lomakin SM, Lazzari M. Polydimethylsiloxane thermal degradation. Part 1. Kinetic aspects. Polymer 2001, 42, 2395–2402. DOI

About this article. The sources were read between 18 and 20 November 2026. Read as page images in the browser: source 4, pages 1 to 5 of its text with all four tables and both spectra (the last two pages were not read); the row for silicone grease in Table 1 of source 23; pages 12 and 13 of the preprint of source 12. Read as text in the browser: of source 1 the passages on the NMR signal, the standards, the validation and the other methods, without its figures; of source 13 the abstract, the passage on the residual coupling and the experimental conditions (its equations are images and were not read); of source 22 recital 2 and the Annex, paragraphs 1 to 3 in full and paragraphs 4 to 6 in part; of source 38 the passage on the measurements, the first two raw materials and claim 1; of source 10 the background, the summary and the first example; of source 21 the passages on the specification, the cyclic siloxanes, the methods of analysis, the authorised uses with most of Table 2, the absorption and the conclusions, in the full text held by Europe PMC; of source 20 Table 1 and three passages in the same archive; of source 2 the method and the results, without its tables and equations, which are images; of source 34 the synopsis and the introduction; the abstract of source 3. Read as abstracts in the records of Europe PMC, complete or lacking only their last words: 5, 7, 8, 9, 14, 15, 16, 19, 27, 28, 31, 32, 35, 36, 42, 43 and 44; of 26, 30, 37 and 45 the first part of the abstract, and of 37 the remainder as an extract on the page of the publisher. Read as complete abstracts elsewhere: 11, 12, 18 and 41 in the records of OpenAlex, 40 in the record of the US Office of Scientific and Technical Information, 46 on two index pages that reproduce it. Read in full through a retrieval tool that returns extracts and answers to questions about a document: sources 25 and 24, the preprints of sources 12 and 40, the text of source 33 and the entry of source 29; of source 17 about four fifths of the opinion itself, with its Table 4 extracted twice with the same values. Source 24 contains inconsistent numbers, and we quote only its solvent, its shift and two diffusion coefficients; the preprint of source 40 gives two values for one coupling, and we report both. Read through the same tool as an abstract: 6. The affiliations of the authors of sources 23 and 18 come from the records of OpenAlex. Of the application note 39 we read the first page as an extract in a search result. What we state about the work of the Gdańsk group in source 30 beyond its abstract comes from sources 1 and 21, and the affiliation of that group from source 25. The place of the thesis behind source 37 comes from a library record. The 29Si shifts of D3, D4 and silicone oil and the 29Si frequency of 99.3 MHz are those of Part 10 of this series. The bibliographic data of the journal articles were compared with Crossref or with the records of Europe PMC,. Our own reasoning and arithmetic, marked in the text: the model calculation of Figure 2 with its invented amounts and area ratio; the conversion of 0.20 ppm to 100 Hz and the differences of 10 Hz and 213 Hz in Figure 5; the summary of the three preparations in Figure 8; the comparison of 7 µg/l with 0.7 mg/kg; the comparison of the siloxane sum with the migration limit; the molar amounts of Si–H and vinyl in the curing agent of 1979 and the 1.68 % of hydrogen in a CH3(H)SiO chain; the formula masses and the proton ratio of the patent polymers; the explanation of the averaged coupling in a liquid; the percentages of Figure 14B, which are one minus the published fraction; the statements on extracts of cured articles and on the three rings in the 1H spectrum; the diffusion ratios and the molar masses of Figure 16; the unit conversions of Figure 9; the comparison of the ring contents of implant gels with the limit of 0.1 %; the possible origins of a silicone singlet and the recommendation of a blank. The definitions in the table of terms are ours. NorrChemica sells NMR tubes and an NMR service; none of the cited authors or companies took part in this article.

NMR Series

  1. Part 1 — Reading 1H and 13C NMR
  2. Part 2 — Reading 19F, 31P and 11B NMR
  3. Part 3 — Reading 2D NMR: COSY, HSQC and HMBC
  4. Part 4 — Quantitative NMR: how a purity figure is made
  5. Part 5 — 15N NMR: which nitrogen is alkylated, protonated or coordinated?
  6. Part 6 — NMR tubes by experiment: which tube for which measurement
  7. Part 7 — Quadrupolar nuclei in NMR: broad lines and missing carbons
  8. Part 8 — Two sets of NMR signals: rotamers, diastereomers or an impurity
  9. Part 9 — NMR yield and isomer ratios: relaxation and the delay between pulses
  10. Part 10 — Silicon NMR: the glass signal, the negative NOE and M, D, T, Q units
  11. Part 11 — Solid-state NMR: magic-angle spinning, cross-polarisation and what a CP-MAS spectrum counts
  12. Part 12 — Enantiomeric excess by NMR: when NMR is the best method, the chiral auxiliaries and the detection limit for the minor enantiomer
  13. Part 13 — Predicted NMR shifts: how accurate ChemDraw, MestReNova and other predictors are
  14. Part 14 — Phosphorus NMR in biology: phospholipids, phosphorylated proteins, oligonucleotide drugs and phosphates in food
  15. Part 15 — Multiplets in 1H NMR: coupling trees, second-order patterns and two signals from one CH2
  16. Part 16 — NMR for cosmetics: adulterated essential oils, free formaldehyde, aloe vera, rancid lipsticks and mineral oil in lip products
  17. Part 17 — Polymers by NMR: composition, sequence, chain ends and branches, and what a spectrum of many chains can and cannot measure
  18. Part 18 — Benchtop NMR: what a permanent-magnet spectrometer does well, and when a sample needs high field
  19. Part 19 — NOESY, ROESY and TOCSY: regioisomers, E and Z isomers, anomers, and the signals of one compound in a mixture
  20. Part 20 — Rapid-injection NMR: starting a reaction inside the spectrometer, how fast the spectra can follow it, and the intermediates it has revealed
  21. Part 21 — Fluorine NMR for PFAS: total organofluorine, what targeted LC-MS/MS does not detect, and the detection limit
  22. Part 22 — Polysiloxanes by proton NMR: silicone oil in medicines, food and cosmetics, the reactive groups before curing, and the rubber network after it (this article)

© 2026 NorrChemica™ · Content may not be reproduced without written permission

Back to blog

Leave a comment

Please note, comments need to be approved before they are published.