nanoparticle synthesis and growth in a continuous plasma reactor from organosilicon precursors

Full Paper

Nanoparticle Synthesis and Growth in aContinuous Plasma Reactor fromOrganosilicon Precursors

Christian Roth, Gina Oberbossel, Elizabeth Buitrago, Roman Heuberger,Philipp Rudolf von Rohr*

Silica-like nanoparticles are produced from four different organosilicon monomers HMDSO,TMDSO, TEOS and TMOS in a continuous non-equilibrium plasma reactor. The nanoparticlesynthesis is studied as a function of the process pressure, plasma power, gas velocity,and gascomposition (Ar:O2:monomer). The morphology, massproduction,and chemical composition of the plasmaformed particlesare investigated. An adapted particlegrowth model for a continuous plasma reactor is intro-duced which explains the influence of the differentprocess parameters on particle evolution. Themorphology of the produced amorphous particles issimilar to fumed silica, with primary particles in the sizerange of 10nm building hard-agglomerates of severalhundred nanometers during the synthesis.

1. Introduction

The plasma surface modification of particulate substrates

hasmany potential applications in industry, since it allows

combining powder bulk properties with highly specialized

surfaces. One such application is the flowability improve-

ment of cohesive powders.[1] In this process nanoparticles

are continuously formed in a non-equilibriumplasma from

C. Roth, G. Oberbossel, Prof. Dr. P. Rudolf von RohrETH Zurich, Institute of Process Engineering, Sonneggstrasse 3,8092 Zurich, SwitzerlandE-mail: [email protected]. BuitragoEPFL Lausanne, Nanoelectronic Devices, NANOLAB, 1015 Lausanne,SwitzerlandDr. R. HeubergerRMS Foundation, Bischmattstrasse 12, 2544 Bettlach, Switzerland

Plasma Process. Polym. 2012, 9, 119–134

� 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlin

organosilicon monomers and directly attached to the

surface of substrate particles. The attached nanoparticles

increase the surface roughness of the substrate particles

reducing the attractive interparticle vanderWaals forces in

between them.[2] This in turn reduces the friction between

the substrate particles improving the flowability such that

an originally sticky and very cohesive powder can be

classified as easy-flowing after the plasma surface mod-

ification.[2]

A well suited reactor concept for the low-temperature

plasma treatment of fine-grained particles is the plasma

downstream reactor (PDR) which is used for the surface

modification of polymers,[3] cohesive drug like sub-

stances[2] and other temperature sensitive compounds.

This reactor concept enables the quasi-continuous and

homogeneous treatment of fine-grained substrates within

a very short and narrow residence time in the plasma of

approximately 0.1 s.

elibrary.com DOI: 10.1002/ppap.201100180 119

120

C. Roth, G. Oberbossel, E. Buitrago, R. Heuberger, P. Rudolf von Rohr

The macroscopic behavior of a surface modified powder

depends on the amount, shape, distribution, and chemistry

of the deposited structures. Therefore, the emphasis of this

work is to study the nucleation and growth of silica-like

nanoparticles in a PDR systemand to correlate the chemical

properties of themonomers and PDR process parameters to

the emerging nanoparticles’ chemistry, morphology and

mass production rate.

In this investigation four organosilicon monomers

hexamethyldisiloxane (HMDSO), tetramethyldisiloxane

(TMDSO), tetraethyl orthosilicate (TEOS), and tetramethyl

orthosilicate (TMOS) are compared. The monomer vapor

was always mixed with oxygen and argon. Oxygen is

required to yield particles with low carbon content while

argon is used to guarantee a stable discharge and good

monomer fragmentation.

We examined for each monomer the influence that the

process pressure, gas composition, residence time and

plasma power has on the particle production rate,

morphology and chemistry of the particles. The nanopar-

ticleswere collected onafilter downstreamthedischarge to

ascertain the production rate. Morphological information

about the nanoparticles was gained by transmission

electron microscopy (TEM). Fourier transform infrared

spectroscopy (FTIR) and X-ray photoelectron spectroscopy

(XPS) were used to investigate the chemistry and bond

structure of the produced nanoparticles.

A multiplicity of plasma processes have been investi-

gated for the controlled synthesis of nanopowders in the

last years.[4] Large scale nanoparticle production is nor-

mally performed in thermal plasmas,[5] not suited for

processes containing temperature sensitive materials. In

order to simultaneously produce and deposit nanoparticles

onto temperature sensitive substrate materials, the nanos-

tructures need to be produced in non-equilibrium plasma

reactors to prevent the substrate particles from melting.

A review about the formation and behavior of powder in

low-pressure plasmas was recently published by Wata-

nabe.[6] Commonly investigated discharge systems are

capacitively coupled parallel plate reactors for thin film

deposition. In such systems the evolution of particles is

oftendescribedasdust formationandlargeeffortsaremade

to avoid the onset of particle growth.

The standard growth model established for such kind of

reactors consists of three main phases. For a proper

differentiation between aggregates and agglomerates we

use the proposed definition of Nichols et al.[7] The

nucleation of particles starts with the development of

primary clusters also known as protoparticles. These

protoparticles tend to be electrically neutral but can also

acquire positive or negative charges since the statistic

fluctuation of charges are large.[6,8] As soon as these

nanometer-sized primary particles reach a critical number

density, aggregation occurs leading to particles in the range


� 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

of tens of nanometers.[9] Further aggregation and agglom-

eration is often inhibited by the negative charge of these

aggregates,which results fromthecurrentbalanceofheavy

ions and mobile electrons impinging on the nanoparticles

surface. The particle growth in the third phase is then

governed by accretion of monomers and its radicals.

This three step process is well described for silane

discharges where the three phases are sometimes also

called ‘‘start of rapid growth,’’ ‘‘rapid growth,’’ and ‘‘growth

saturation’’.[8,10,11] In general particles are most readily

formedat elevatedprocesspressures and inelectronegative

plasmas[12] like the investigated organosilicon-O2-Ar dis-

charge. This is due to the increased number of collisions at

higher pressures, reduced losses of negative species based

on ambipolar diffusion and the mostly anionic or neutral

particle formation pathways.[13]

Up to our best knowledge a detailed investigation of the

controlled nanoparticle formation in a continuous reactor

type such as the PDR with different organosilicon mono-

mers is not available yet. The PDR differs significantly from

a standard plasma reactor for thin film deposition in terms

of residence time distribution, operating pressure, and

power density. Thus, also an adaptedmodel for the particle

growth is necessary. Since the properties of the nanopar-

ticles produced in the PDR can be tuned by the process

parameters and the choice of monomer, the general

findings of this study do not only deepen the knowledge

about the plasma assisted surface modification of fine-

grained substrate powders but may also serve other

applications connected to the production of silica-like

nanoparticles.

2. Experimental Section

2.1. Plasma Downstream Reactor

The original plasma downstream reactor is designed for the quasi

continuous surface modification of powders and described else-

where.[2] Sinceno substrate powder is fed in this study, a simplified

process scheme for the pure nanoparticle production is shown in

Figure 1. The plasma chamber (1) consists of a 1.5m long double

wall glass reactor with an inner diameter of 40mm. Thus, the

effective reactor volume accounts to only 0.5–1.5 l depending on

the actual expansion of the discharge in the reactor. The gap

between inner andouter glass tube is flushedwithdeionizedwater

(2) of 20 8C to ensure a constant reactor temperature.

The discharge is driven by an inductively coupled plasma (ICP)

source which operates at a radio frequency (RF) of 13.56MHz. The

RF-generator (3) is connected over amatching network (4)with the

water cooled copper coil (5) on the outside of the cooling jacket. The

flow rates of oxygen (purity>99.999%), argon (purity> 99.999%)

and the organosiliconmonomers (see Table 1) are adjusted by flow

controllers. The liquid monomers are fed through a controlled

evaporation mixing device (6) and stored under a 2 bar argon

atmosphere (7) to prevent monomer degradation.

DOI: 10.1002/ppap.201100180

Figure 1. Process scheme of the plasma downstream reactoradapted for nanoparticle production, FIC: flow indicator control-ler, PIC: pressure indicator controller.

Nanoparticle Synthesis and Growth in a Continuous Plasma Reactor

Below the plasma zone the produced nanoparticles are

separated from the gas stream by a downcomer (8), cyclone (9)

and filter unit (10), and collected in the solid collection vessels (11).

Since the produced silica structures are very small, all powder was

Table 1. Organosilicon monomers.

Name Abbreviation

Hexamethyldisiloxane HMDSO

Tetramethyldisiloxane TMDSO

Tetraethyl orthosilicate TEOS

Tetramethyl orthosilicate TMOS



collected from the polyester filter (Type: TFV 478/170 A, effective

diameter:0.385m,efficiencygrade99.94%forparticlesbetween0.2

and 2mm, Tecnofil, Switzerland). A constant pressure in the reactor

ismaintainedduring theprocess by a butterfly control valve (12) in

frontof the twostage roots and rotaryvanevacuumpump (13). The

standardprocess conditionswhicharenormallyused tomodify the

surfaces of substrate powder materials are given in Table 2 and

were also taken as starting point for this study.

2.2. Organosilicon Monomers

The four investigated organosilicon precursors can be classified in

two groups; the disiloxanes HMDSO and TMDSO and the

orthosilicates TEOS and TMOS. The chemical structures and purity

declarations are summarized in Table 1. The disiloxanes consist of

two silicon groups which are bound to a central oxygen atom and

saturated by either three methyl groups (HMDSO) or two methyl

groupsanda singlehydrogenatom(TMDSO). In contrast to this, the

orthosilicates consist of only one central silicon atom which is

surroundedbyfouroxygenatoms, saturatedeachbyeitheranethyl

(TEOS) or methyl group (TMOS).

2.3. Filter Weight Measurements

Theproduction rateofnanoparticles isdeterminedbyweighingthe

polyester filter before and after each experimental run. Depending

on the process conditions (degree of filter clogging) particles were

produced during 150–1800 s to obtain a maximal amount of

particles and to minimize the weighing error. The scale (EG420-

3NM, Kern, Germany) has a linearity of�0.003g and the filter was

packed into an aluminum envelope before the measurement to

avoid electrostatic disturbances. The main measurement error

originatesnot from theweighing itself but from theunknownwall

depositions in the reactor tube.Asmall fractionof theparticlesmay

Chemical formula Supplier Purity [%]

Aldrich >98

Fluka >97

Aldrich >98

Fluka >97

www.plasma-polymers.org 121

Table 2. Standard process conditions and its variation.

Parameter Standard conditions

Plasma power 300W

System pressure 200 Pa

Monomer flow rate 50 sccm

Oxygen flow rate 500 sccm

Argon flow rate 950 sccm

122


stick to the glass wall and can therefore not be evaluated. To keep

this error as small as possible, the reactor tube was intensively

cleaned before each experiment with a rotating steel brush and

pressurized air to ensure a smooth and powder-free surface.

Thenanoparticleproductionat standardparameter settings (see

Table 2) was measured three times for each monomer and the

measurementwas found to be quite reproducible; the error bars of

these points correspond to twice the standard deviation. All other

data points represent single measurements. Due to the unknown

wall depositions the overall uncertainty of themethod is assumed

to lie in a 20% error frame.

2.4. Transmission Electron Microscopy

The structure and morphology of the collected nanoparticles were

investigated with transmission electron microscopy. For this, we

collected the nanoparticles directly on a TEM grid (S160, Plano

GmbH, Germany), which was mounted on a stainless steel holder

and placed concentric in the tube between the cyclone and the

filter chamber. The discharge ran always 60 s during the

morphology analysis experiments to attain an optimal particle

concentrationon theTEMgrid. Themicroscope (Philips, CM12)was

operated at an acceleration voltage of 100kV and micrographs

were recordedwith a GATAN 794MultiScan CCD camera. Nominal

magnificationsof8 000,28 000,45 000,and100000wereemployed

to image the agglomerated structures as well as the primary

particles.

2.5. Fourier Transform Infrared Spectroscopy

The produced silica nanoparticles were analyzed with a FTIR

spectrometer (Perkin-Elmer, Spectrum BX II, USA) in combination

with an attenuated total reflection (ATR) diamond probe (Pike

Technologies, MIRacle, USA). For each spectrum 16 scans in the

range of 600–3600 cm�1 at 2 cm�1 resolution were collected and

averaged. For background correction the spectrum against air was

measured and subtracted. We analyzed each sample three times

and the error bars of all data points represent twice the standard

deviation of the average.

Even though FTIR is no quantitative method, the spectra of the

produced particles were further evaluated to illustrate the

influence of the process parameters on the bond structures.

According to the law of Lambert–Beer, the absorbance A is

proportional to the molar extinction coefficient e, the effective

optical path length l, and the specific concentration c of the excited



bond structure.

A ¼ "�l�c (1)

In ATR operation, l is proportional to the penetration depth

which is in turnproportional to thewavelength l. Thus, the spectrawere divided by the wavelength to obtain a signal, which is

independent of the penetration depth and proportional to the

product of e and c.

A

l� "�c (2)

In order to study the concentration ratio c1/c2 between two

specific bond features (1 and 2) as a function of the process

parameters, the baseline corrected absorbance spectra were

integrated over the respective peak widths (ll� lh) of interest asoutlined in Equation (3).

c1c2

� "2"1

�

Zlh;1

ll;1

A1

l1dl

Zlh;2

ll;2

A2

l2dl

(3)

n though the ratio e /e is unknown, it can be assumed to be
Eve 2 1
constant as long as the same features are compared for different

samples.

2.6. X-ray Photoelectron Spectroscopy

The surface chemistry of selected powders was investigated with

X-ray photoelectron spectroscopy (XPS) using a Kratos Axis Nova

spectrometer (Kratos Analytical, Manchester, UK). A between 0.5

and 1mm thick layer of the powder was fixed with double sided

carbon tape on the sample holder. The samples were illuminated

withmonochromaticAlKa irradiation, runat 225W(15kV, 15mA).

An aperture is included in the lens system and allowed to

analyze an area of 700� 300mm2 at a take off angle of 908. Thephotoelectrons were analyzed with a hemispherical analyzer,

operated in the fixed analyzer-transmission mode with a pass

energy of 40 eV for the detailed spectra and 80 eV for the survey

spectra (fullwidthathalf-maximum(fwhm)forAg3d5/2¼ 0.55and

0.9 eV, respectively).

A charging of the powder was over-compensated with slow

electrons from the neutralizer. The spectrometer was calibrated

according to ISO 15472:2010 with an accuracy better than

�0.05 eV. The residual pressure in the spectrometer was below

1�10�6 Pa for the analysis.

The spectrawereprocessedusing theCasaXPSsoftware (V2.3.14,

Casa Software Ltd, UK). Peak shifting was corrected by referencing

aliphatic carbon to 285.0 eV.[14] The peak fitting was performed

after subtraction of an iterated Shirley background.[15] The

quantitative composition was calculated by correcting the peak

areasby the transmission functionandthesensitivity factorsgiven

by Kratos assuming a homogeneous compound. Every analyzed

powder was measured twice and the presented atomic factions

DOI: 10.1002/ppap.201100180

Figure 2. Dependency of nanoparticle properties on process andintermediate parameters.


represent the arithmetic mean of the two measurements. The

errors bars of all data points correspond to twice the standard

deviation.

2.7. Parameter Variation

In an ideal case process parameters can be varied independently of

each other to study their influence. In practice this is almost never

possible and the variation of one process parameter implicates a

manifold of changes in the reaction conditions. In Figure 2 an

overviewofvariable inputparametersandconnected intermediate

parameters of the present study is shown.

The process pressure for example has a major influence on

properties of the produced nanoparticles. If the pressure is simply

adjusted by the butterfly valve (12) in Figure 1 by keeping the total

gas flow rate and its composition constant, the pressure and the

mean gas velocity are varied at the same time and it is not possible

to assign the found tendency to one or the other parameter. If on

the other hand the pressure is varied by adjusting also the total

gas flow rate, either the gas composition has to be changed (for

instance by adjusting only the argon flow rate) or, if all flow rates

are adjustedproportionally, the effect of variedmonomerflow rate

needs to be considered (different energy per monomer molecule).

Wewere well aware of this problem and thus, performed seven

different experimental series of parameter variations to identify

the most important process parameters and its combinations and

to understand their influence on the produced nanoparticles. The

series are consecutively labeled from A to G and described in the

following. The explicit settings for the filter weightmeasurements

arepresentedinTable3. For thechemicalanalysis someparameters

were analyzed with higher resolution than in Table 3 while TEM

images are presented for selected parameter values only to

illustrate its influence best.

In series Awevaried only the plasmapower.With rising plasma

power more energy is available for excitation, ionization and

dissociation processes. Thus, not only themonomer fragmentation

increases at higher plasma powers but the discharge expands also

more through the reactor such that the effective plasma residence



time becomes slightly longer. Thus, in this series in fact a

combination of increased energy input and prolonged plasma

residence time can be studied.

In the second series B we investigated the influence of the

oxygen tomonomer ratio. By keeping themonomer flow rate such

as the total gas flow rate constant, the mean gas velocity is not

influenced and therefore the plasma residence time is similar

within this series. Since the changes in oxygen flow rate had to be

compensated by the argon flow rate to maintain a constant mean

gas velocity, the varied argon to oxygen ratio could have a side-

effect in this series too.

Series C was performed to study the influence of the mean gas

velocity.We varied only the argon flow rate and kept all other flow

rates suchas thepressureconstant. Since themeangasvelocityand

the gas compositionwere varied at the same time, the influence of

the mean gas velocity cannot be studied independently. Never-

thelessweassume that the residence timehas a larger influence on

the particle evolution than the altered dilution of the gas mixture

with argon. In series D again the mean gas velocity was varied

while keeping the gas concentration and the pressure constant

and adjusting all flow rates proportionally. Here the change in

monomer flow rate adulterates the insulated study of the mean

gas velocity most.

The influence of the pressure on the nanoparticle formation is

also studied in two different series. First in series E, we varied the

pressure by keeping the mean gas velocity, the oxygen and

monomer flow rate constant and adjusting the total gas flow rate

with the argon flow rate only. The connected change in gas

composition acts as a secondary influence here. In series F on the

other hand we kept all flow rates constant and adjusted the

pressure with the butterfly control valve in front of the vacuum

pumps. Hence, in this series the mean gas velocity changed

inversely proportional to the set pressure.

Finally in series G we varied themonomer flow rate by keeping

the total gas flow rate constant (by a slight adjustment of the argon

flow rate). As a consequence the oxygen tomonomer ratio, such as

the overall gas composition changed accordingly.

3. Results and Discussion

The nanoparticles produced in the PDR exhibit different

morphology, primary particle size, and chemical composi-

tion depending on the precursor monomer and applied

process conditions. A qualitative model describing the

nanoparticle growth in the PDR and monomer specific

differences are introduced first and in the subsequent

sections the influence of the varied process parameters is

shown.

3.1. Qualitative Particle Growth Model

The investigated PDR is a continuous reactor and differs

strongly from a standard parallel plate plasma chamber.

Thus, the available particle growth models cannot be

applied in a straight forward manner. In the classic model,


Table 3. Experimental series to study the influence of the process parameters (standard parameter settings in italics).

Series Plasma

power

[W]

Argon

flow rate

[sccm]

Oxygen

flow rate

[sccm]

Monomer

flow rate

[sccm]

Total gas

flow rate

[sccm]

Mean gas

velocity

[m/s]

Process

pressure

[Pa]

(A) Power variation 200 950 500 50 1 500 10.8 200

250 950 500 50 1 500 10.8 200

300 950 500 50 1 500 10.8 200

350 950 500 50 1 500 10.8 200

(B) Oxygen to monomer

ratio variation

300 1 200 250 50 1 500 10.8 200

300 950 500 50 1 500 10.8 200

300 700 750 50 1 500 10.8 200

300 450 1 000 50 1 500 10.8 200

300 200 1 250 50 1 500 10.8 200

(C) Total gas flow rate

variation with constant O2

and monomer flow rate

300 200 500 50 750 5.4 200

300 650 500 50 1 200 8.7 200

300 950 500 50 1 500 10.8 200

300 1 250 500 50 1 800 13.0 200

300 1 700 500 50 2 250 16.2 200

(D) Total gas flow rate

variation with constant

gas composition

300 475 250 25 750 5.4 200

300 760 400 40 1 200 8.7 200

300 950 500 50 1 500 10.8 200

300 1 140 600 60 1 800 13.0 200

300 1 425 750 75 2 250 16.2 200

(E) Pressure variation

with constant mean

gas velocity

300 200 500 50 750 10.8 100

300 950 500 50 1 500 10.8 200

300 1 700 500 50 2 250 10.8 300

300 2 450 500 50 3 000 10.8 400

(F) Pressure variation

with constant gas

flow rates

300 950 500 50 1 500 14.4 150

300 950 500 50 1 500 10.8 200

300 950 500 50 1 500 7.2 300

300 950 500 50 1 500 5.4 400

(G) Monomer flow rate

variation with constant

mean gas velocity

300 975 500 25 1 500 10.8 200

300 950 500 50 1 500 10.8 200

300 925 500 75 1 500 10.8 200

300 900 500 100 1 500 10.8 200

124


fluid dynamics are neglected and the number density and

size of the particles in a confined area are studied as

function of time. In the PDR, the evaporated monomers,

mixed with the process gases oxygen and argon, stream

through the reactor tubewithameanvelocityof10.8m/sat

standard conditions (see Table 2). The plasma expands no

more than 0.5m up- and downwards from the ICP source,

which results in a very short residence time of the gas

molecules in the discharge of approximately 0.1 s. During

this short period the major phases of particle growth take



place. A sketch of our adapted qualitative model for

continuous plasma chambers is shown in Figure 3.

The monomer vapor mixed with the gases oxygen and

argon is feed to the reactor tube from the top. Entering the

discharge the number of charge carriers in a distinct fluid

element increases rapidly and consecutive ionization and

dissociation processes lead to an increased number of

electrons, ions, and radicals. The number of charge carriers

rises until equilibrium between production and losses is

reached. In this initial phase the monomer molecules are

DOI: 10.1002/ppap.201100180

Figure 3. Qualitative model for the nanoparticle synthesis in acontinuous plasma reactor such as the PDR.


ionized and broken up into fragments. They react with

oxygen radicals and ions or other monomer fragments

forming stable chemical compounds and larger structures,

known as clusters. This first particles growth phase is

identical with the classic models developed for large

vessels.

If a critical number density of clusters is reached, the

second phase (aggregation and fast growth) is initiated too

and the clusters grow to primary particles of several

nanometers in diameter. Due to the geometry of the reactor

and the arrangement of the ICP coil an axial ion density

profile develops, with a maximal charge carrier density in

the center below the coil. As the ion density increases

towards the center of the reactor, most probably also the

growth of clusters and aggregation to larger structures is

intensified.We assume that depending on the local plasma

parameters the aggregates reach a critical size from which

on further aggregation is inhibited by the accumulated

negative charge on the particle surface. As a consequence

also agglomeration of primary particles to large structures

is prevented as long as the surface charge is high enough.



Reaching the lower end of the discharge the surface

charge declines and the primary particles are able to build

hard-agglomerates in a third growth phase.

The still available monomer fragments and ions in the

afterglow or the lower end of the discharge contribute then

also to the growth by accretion. In this fourth growth phase

some additional atomic layers can be deposited on the

agglomerated structures.

Since no further energy is available for fragmentation or

dissociation processes in the afterglow, the particle growth

is quenched as soon as all radicals and ions are saturated.

We assume that the ‘‘growth saturation’’ phase, known

from the classic models, is mostly inhibited due to the very

short residence times. Therefore, the particles keep their

initial structure from the agglomeration phase and do not

grow much further to smooth spherical particles with

diameters in the micrometer range.

In typical RF plasma chambers at low pressures and gas

flow rates the onset of particle formation can take several

seconds or evenminutes. In contrast to this, at least the two

first particle formation steps (cluster formation and fast

growth) take place in about 0.1 s in the presented process.

Due to thehigher pressure andpower density, collisions are

more frequent and the formation of protoparticles such as

aggregates is supposed to be considerably accelerated.

3.2. Nanoparticles Produced at Standard Conditions

In Figure 4 TEMmicrographs of particles producedwith the

four different monomers at standard conditions (see

Table 2) are shown. Themicrographs show large structures

consisting of strongly agglomerated primary particles. The

primary particles are amorphous, non-spherical and in the

size range of about 10nm, while the agglomerate size

reaches 500nm and more. The structural differences

between the monomers are only small and hardly

quantifiable at standard conditions.

Similar particle structures were also found in ref.[16]

where SiO is evaporated in a 200 Pa argon atmosphere to

form nearly spherical primary particles between 8 and

24nm in size. The found particles in Figure 4 look also

similar to fumedsilica inearly stagesof laminarflamesor in

premixed diffusion flames where the reaction time is

limited and no sintering spherical particles occurs.[17]

In a discharge system like the PDR a multiplicity of

reactions is possible and always stochastically distributed,

whereas the probability for a reaction is determined

by the respective cross sections, available energies and

the required reaction enthalpy. For HMDSO for instance

the dissociation of a methyl group is the most probable

fragmentation of the monomer molecule.[18] The dissocia-

tion energy between the silicon and oxygen atom is higher

than the required energy to split off an ethyl or methyl

group.[19] ThereforeSi–Obondsare relatively stable, leading


Figure 5. FTIR spectra of silica particles processed from the fourdifferent monomers at standard conditions.

Figure 4. TEM images of silica particles processed from the fourdifferent monomers at standard conditions, nominal magnifi-cation: left images 28000, right images 100000.

126


to the preferred formation of silicon oxide (SiOx) struc-

tures.[20] In argon and oxygen diluted organosilicon

plasmas the hydrocarbon tails undergo oxidation steps

forming, e.g., formaldehyde (CH2O), formic acid (CH2O2),

carbon monoxide (CO), carbon dioxide (CO2), or as a

byproduct water (H2O).[21] These gaseous species leave

the process through the vacuum pump.

The absorbance spectra obtained by FTIR from the

particles produced at standard conditions with the four



monomers are illustrated in Figure 5. The wave number of

the predominant Si�O�Si asymmetric stretching mode

depends on the oxygen concentration and is generally

found between 940 cm�1 for O doped amorphous silicon

and1 075 cm�1 for stoichiometric SiO2.[22]Within thisstudy

the band was found between 1 025 and 1055 cm�1 which

indicates a slight sub-stoichiometric nature (SiOx, x< 2) of

the nanoparticles.

The symmetric bending mode of associated methyl

groups Si(CH3)x between 1270 and 1280 cm�1 [23] was only

present for particles derived fromHMDSO and TMDSO. The

amount of thesemethyl groups are ameasure of the carbon

content of the synthesized structures as shown by

Spillmann et al. in an earlier study.[24] The broad band

between 3 000 and 3 600 cm�1 corresponds to O�H

stretching vibrations of silanol in the form of associated

surface hydroxyl groups and silanol species trapped in the

SiOx lattice during nanoparticle growth.[25]

The monomers TEOS and TMOS contain no Si�CH3

groups and the synthesis of such groups in the plasma is

improbableunder the chosenprocess conditions. Therefore,

no absorbance in the respective wave number region was

found. HMDSOwith a methyl groupmore per silicon atom

compared to TMDSO showed also higher absorbance of

Si�CH3 groups in the therewith produced particles. On the

other hand particles derived from the monomer TMOS

featured at standard conditions the highest absorbance

originating fromOH-groups,while thesignal in theHMDSO

derivedparticleswas rather low. The carbon tooxygen ratio

in the monomer molecule is 6 for HMDSO and only 1 for

TMOS.Thus,moreoxygen isavailable in theTMOSmolecule

to form silanol groups instead of oxidizing the hydrocarbon

tails. This stands also in good correlationwith the proposed

reaction pathway for TMOS in ref.[23]where amethyl group

is spilt from themonomer and the remaining oxygen atom

DOI: 10.1002/ppap.201100180

Figure 6. Atomic fraction determined by XPS of particles pro-cessed from the four different monomers at standard conditions.


is saturated with a hydrogen radical to form a silanol

group.

ðCðC

Plasma

� 2012

H3 � O�Þ4 Si ! ðCH3 � O�Þ3 Si� Oþ CH3

H3 � O�Þ3 Si� OþH ! ðCH3 � O�Þ3 Si� OH(4)

Figure 7. Normalized peak area ratio between methyl groupbending and Si�O�Si stretching for particles processed fromHMDSO and TMDSO for varying plasma power, studied inseries A.

A prominent C�H absorption (stretching 2 900–

2 980 cm�1) as found in ref.,[26] where spherical particles

areplasmapolymerized fromTEOS ina tubular reactorwith

high residence time, could not be detected in any acquired

spectra, indicating a comparably low hydrocarbon content

in the orthosilicate derived powders.

An XPS analysis of all powders produced at standard

conditions is shown in Figure 6. The particles derived from

HMDSO reveal the highest carbon content on the surface,

while the particles originating from TMOS consist of

less carbon, but show a higher oxygen fraction. Carbon

contaminations are found on nearly all surfaces which

were exposed to the atmosphere. Thus, the absolute carbon

content of the particle surfaces represents both, methyl

groups from the plasma synthesis and carbon adsorbed

from the atmosphere. The atomic fractions shown in

Figure 6 could therefore overestimate the actual carbon

content of the nanoparticles. Nevertheless the trends

support the findings from the FTIR measurements. TMDSO

havingonemethyl group less per siliconatomthanHMDSO

resulted also in particles with a significantly reduced

carbon content. Similarly the orthosilicate TEOS with

8 carbon atoms resulted in particles with more carbon

than those originating from TMOS with only 4 carbon

atoms in the monomer molecule.

Since the synthesis of pure SiO2 is preferred, the silicon

bonds are of main interest and can be assessed by

evaluating of the Si2p binding energy. The Si2p peak was

found between 103.6 and 104.0 eV for all four monomers,

which suggest a highly oxidized configuration of the

silicon.[27]

Process. Polym. 2012, 9, 119–134

WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Already the FTIR spectra and XPS analysis at standard

conditions lead to the conclusion that the chemistry of the

monomer is well represented in the deposited nanoparti-

cles. Since the monomer is not fragmented completely,

chemical features of the precursor molecule can still be

found in the produced particles.

3.3. Influence of the Plasma Power

The influence of the plasma power (series A) on the particle

chemistry of the disiloxanes is shown in Figure 7,where the

peak area ratio between the associated methyl group and

the Si�O�Si feature is depicted. The orthosilicate derived

particles showed no methyl group features within the

investigatedparameter range. Thepeakareawasevaluated

between 1250 and 1290 cm�1 for the Si(CH3)x feature and

between 965 and 1 240 cm�1 for the Si�O�Si feature,

respectively. The particles derived from HMDSO incorpo-

rated generally more methyl groups compared to the

particlesoriginating fromtheTMDSOmonomer,which is in

good correlationwith themonomer chemistry. TheHMDSO

molecule contains 50% more methyl groups than the

TMDSO molecule and since a full fragmentation of the

monomer is very unlikely in the investigated system a

higher carbon content of the particles originating from

HMDSO can be expected. With higher plasma power the

ion density and electron temperature increases and the

monomer fragmentation is enhanced. The carbon tails

are oxidized to gaseous compounds as CO2 which leave the

chamber with the off gas. Thus, the relative amount of

methyl groups decreased and led to more stoichiometric

silica.

The nanoparticle production as a function of the applied

plasma power is depicted in Figure 8. In general a higher


Figure 8. Nanoparticle production rate as a function of theapplied plasma power, varied in series A.

128


mass of particleswas collected per timeunit for disiloxanes

comparedtoorthosilicates. This correlateswellwithstudies

where the deposition rate of different organosilicon

monomers on flat substrates is compared. HMDSO yields

generally in higher deposition rates than TEOS.[28]

Figure 9. TEM images of silica particles processed from HMDSO and TEOS at 200 and300W plasma power, respectively (series A), nominal magnification: 45 000.

We assume that the different number

of silicon atoms in the monomers;

orthosilicates contain two silicon atoms

while orthosilicates incorporate only

one central silicon atom per molecule;

are responsible for the higher mass

production rate with disiloxane precur-

sors. Hence, at the same monomer flow

rate of 50 sccm the doubled amount of

Si atoms was available for the formation

of SiOx-like particles.

The influence of the plasma power on

the production rate was rather small in

the investigated power range and

slightly fewer particles were collected

for higher plasmapowers. Andersonet al.

found the same tendency in a very

similar continuous non-equilibrium

plasma reactor for the synthesis of

Si3N4 particles in the pressure range of

39–130 Pa and specific power inputs

between approximately 0.1 and 0.2W/

cm3.[29] They assumed that electron

collision reactions break up again the

evolving particles at high specific power

inputs and thus, reduce the yield. In

comparison to ref.[29] the power density

in the investigated PDRwaswith around

3W/cm3 much higher but SiO2-like

structures are very stable.



Recently also Hundt et al. discussed the influence of the

discharge power on the particle formation in an electro-

positive argon-acetylene discharge.[30] They found smaller

growth rates at increased discharge powers, where they

also measured a higher electron temperature and density

and thus, attributed this finding to an increased negative

charge of the growing particles (higher repulsive forces).

Negative ions may partially compensate this effect in the

investigated electronegative discharges. But since the

particle charge depends mainly on the electron tempera-

ture, an increased negative particle charge at higher

discharge powers could still be expected.

Over all we suppose that the small increase in particle

mass at lower discharge power could originate either from

additional carbon groups incorporated in the growing

particles, a lower repulsive particle charge, the reduced

discharge length in the reactor (leading to a shorter plasma

residence time) or a combination of these factors. The

influence of the mean residence time is discussed below in

the corresponding sub-section.

The size of the produced particles is also a function of the

applied plasma power, as visible in the corresponding TEM

micrographs inFigure9exemplarily shown forHMDSOand

TEOS. The primary particles were larger at low plasma

DOI: 10.1002/ppap.201100180

Figure 11. Normalized peak area ratio between methyl groupbending and Si�O�Si stretching for particles processed fromHMDSO and TMDSO for various O2 to monomer ratios, studied inseries B.


power. Thisfindingmightagainbecorrelated to theparticle

charging in combinationwith the proposedmodel depicted

in Figure 3. At high specific power inputs the growing

particles could acquire a higher specific charge, which in

turn lead to higher repulsive forces between equally

charged particles. The increased electrostatic forces would

reduce the collisional particle growth in the aggregation

phase and thus, lead to smaller primary particles at higher

power inputs. The same tendency in particle size and a

similar explanation was also reported by Knipping et al.

where silicon nanoparticles were formed from silane in a

microwave discharge.[31] Since more energy is available at

higher plasma power the cluster formation is also assumed

to take place faster which is another possible explanation

for the difference in primary particle size. The more

monomer was used for the initial growth of clusters, the

fewer molecules were available later on for the growth by

accretion. In fact the primary particle diameter distribution

seems narrower at 300W, where many chains of equally

sized primary particles are observable. In the structures

found at 200W on the other hand the primary particles are

hard to identify since growth by accretion of ions and

radicals after the agglomeration may smoothed edges and

led to structures of various sizes and forms.

3.4. Influence of the Oxygen to Monomer Ratio

The variation of the oxygen tomonomer ratio (series B) and

its impact on the nanoparticle production rate is shown in

Figure 10. While the nanoparticle production rate from the

orthosilicates TEOS and TMOS was rather independent on

the oxygen admixture, a higher particle mass resulted

for the disiloxanes HMDSO and TMDSO with increasing

oxygen contents. Orthosilicates with its central silicon

atom, bound to four oxygen atoms, are less affected by the

Figure 10. Nanoparticle production rate as a function of theoxygen to monomer ratio, varied in series B.



changing oversaturation with O2. Disiloxanes on the other

hand require oxygen radicals to formSiOx-like particles and

thus, the production rate ismore affected byaltered oxygen

content in the process gas. Higher oxygen content in the

process gas seems to improve themonomer fragmentation

and thus, the particle growth.

The relative methyl group content of the particles

produced from disiloxanes as a function of the oxygen to

monomer ratio is illustrated in Figure 11. With rising

oxygen content in the process gas, the methyl group

content decreased asymptotically. This stands in good

correlationwith studies investigating growing films onflat

substrates, where a high influence of the oxygen concen-

tration on the resulting carbon content is stated.[32] The

higher abundance of oxygen in the plasma leads to a higher

monomer fragmentation.[27] The methyl groups are oxi-

dized to volatile hydrocarbons and the arising nanoparticles

contain less incorporated methyl groups the more oxygen

radicals are present. Even though the carbon content of the

orthosilicate derived particles cannot be assessed by FTIR it

is assumed that the carbon content decreases as well with

rising oxygen content in the process gas.

As already found in series A, the particles produced from

HMDSO includedmoremethyl groups than the correspond-

ing particles produced from TMDSO, independent of

the oxygen to monomer ratio. Again this stands in good

correlation with the chemical composition of the two

disiloxanes.

In order to validate the FTIRmeasurements, we analyzed

the influence of the oxygen concentration on the chemical

composition of the resulting particles also by XPS as shown

in Figure 12 for the monomer HMDSO. With rising oxygen

to HMDSO ratio in the process gas the oxygen content in

the growing particles increased and at the same time the


Figure 12. Atomic fraction determined by XPS of particles pro-cessed from HMDSO for various O2 to monomer ratios, studied inseries B.

Figure 13. Conversion to nanoparticles as a function of the meangas velocity, studied in series C and D.

130


carbon content decreased. At the same time the averaged

binding energy of the Si2p peak shifts from103.45 eV for an

oxygen to HMDSO ratio of 5 to 103.6 and 103.75 eV for the

ratios of 10 and 20, respectively. This indicates as well a

highly oxidized configuration of the silicon for higher

oxygen content in the process gas, as previously shown for

flat substrates.[27]

Hence, the tendencies found by FTIR are well supported

by the XPS analysis, as a reduced amount of methyl groups

correlates with a reduced carbon content on the particle

surface. Both, the FTIR and XPS measurements lead

therefore to the conclusion that the formation of highly

stoichiometric SiO2 particles is favored by high oxygen to

monomer ratio.

3.5. Influence of the Mean Gas Velocity

Themean gas velocity and thus also the residence timewas

varied in both series C and D. Since we adjusted in series

D also the monomer flow rate, the conversion to

nanoparticles is used as ordinate in Figure 13 instead of

the absolute nanoparticle production rate to display the

influence of the mean gas velocity on the nanoparticle

formation. Assuming a full conversion from monomer to

stoichiometric SiO2, a maximal yield of 0.32 g/h for

disiloxanes and 0.16 g/h for orthosilicates would result

per monomer flow rate of 1 sccm. While the absolute

nanoparticle production ratewashigher for the disiloxanes

the conversion was in general higher for orthosilicates as

cognizable in Figure 13. This means that at the samemolar

flow rate a higher fraction of the monomer is converted to

particles and clusters for TEOS and TMOS compared to

HMDSO and TMDSO.

Interestingly the conversion became higher for both

series C and D, the shorter the residence time was. Hence,



the faster the different growth processes depicted in

Figure 3 take place, the higher is the fraction of the

monomer which is converted to particles. The rising

conversion to particles with shorter residence times could

be explained by the competing processes of film growth at

the reactor wall and particle growth in the plasma bulk.

With increasing residence time (and lower mean gas

velocity, respectively), the interactionwith the reactorwall

becomes more important and a higher fraction of the

monomer could be used to build up a silica like film on the

reactor wall.

Even though the monomer flow rate was doubled or

halved respectively in series D compared to series C at

the maximal and minimal mean gas velocity (see Table 3),

the conversion differed by only a few percentages. Hence,

the available energy per monomer molecule seems

not to be a limiting factor for the particle production in

the chosen parameter range and the process is rather

monomer deficient regarding powder formation.

Since the particle morphology was very similar for all

investigated residence times, we assume that the reaction

time has only aminor effect on the particle evolution, such

that the same particle growth processes just happen faster

or slower depending on the mean gas velocity but yield in

particles of the same size and structure.

3.6. Influence of the Monomer Flow Rate

The influence of the monomer flow rate, varied in series D

and G, is shown in Figure 14 again in the form of the

fractional conversion. Even though always the same

amount of energy was available, a higher conversion was

measured at increased monomer flow rates, such that the

process in the applied parameter range can be classified as

monomer deficient, concerning the nanoparticle produc-

tion. Higher monomer concentrations lead to an increased

DOI: 10.1002/ppap.201100180

Figure 14. Conversion to nanoparticles as a function of the mono-mer flow rate, studied in series D and G.


number of monomer radicals and ions. Hence, a higher

particle formation rate could be expected. In seriesG,where

we kept the residence time constant, the conversion was

even slightly rising for the monomer TEOS while it was

nearly unaffected by the monomer flow rate for the other

three monomers.

Figure 15. TEM images of silica particles processed from HMDSO and100 sccm monomer flow rate, respectively (series G), nominal magn



In seriesD,wherewekept thegas concentrationconstant

and adjusted all flow rates proportionally the conversion

was even higher at elevated monomer flow rates. This

can be correlated to the reduced residence time at high

monomer flow rates in this case. As already shown in

Figure13, a short residence timeor ahighmeangasvelocity

respectively favors the nanoparticle formation in the

investigated reactor type, since wall interactions are

reduced. Hence, we assume that the trend towards higher

conversion at elevated monomer flow rates in series D is

mainly due to the increased mean gas velocity.

An increased monomer flow rate had a similar effect on

the nanoparticle morphology as a reduction of the plasma

power, shownbefore in Figure 9.As exemplarily depicted in

Figure 15 for the monomers HMDSO and TEOS produced in

series G, the particles became much larger compared to

standard conditions if themonomer flow ratewas doubled.

Even though the reactor volumes were larger and the

residence times much longer in comparative studies the

same tendency towards larger particles at increased

monomer flow rates are reported for TEOS or for silane-

O2-Ar-discharges.[33,34] The effect of monomer concentra-

tionon the size of synthesizedparticles is also studied in the

field of flame spray pyrolysis. In this case the equilibrium

TEOS at 50 andification: 45 000.

chemistry governs and particle charging

plays a minor role but still, larger

particles are formed at higher monomer

concentrations as for instance shown in

ref.[35] where spherical silica particles

were produced from TEOS.

We assume that the changing particle

morphology in our case is a result of the

specific energy per monomer molecule

which is lower if the monomer flow rate

is increased. Thus, theparticle chargewas

reduced, leading to larger aggregates in

the fast growth phase and enhanced

agglomeration. We further assume that

due to the high availability of monomer

radicals and ions at high monomer flow

rates also the accretion phase was

intensified, leading to large structures

where the primary particles are only

hardly cognizable.

3.7. Influence of the Pressure

The impact of the process pressure on the

particle formationwas studied in the two

experimental series E and F, once with a

constant mean residence time and once

with constant gas flow rates, respec-

tively. In Figure 16 the production rate

for both series is shown. It is clearly


Figure 16. Nanoparticle production as a function of the processpressure, studied in series E and F.

Figure 17. Normalized peak area ratio between methyl groupbending and Si�O�Si stretching for particles processed fromHMDSO and TMDSO for varying process pressure in series F.

132


perceivable that the production rate was higher for

increasing pressure for both series. Form the value at

200 Pa on, which is the same for both series, the production

rate increased steeper in series E, where the residence time

was kept constant by adjusting the argon flow rate. Since

always the same amount of gas (1 500 sccm) was fed to the

reactor in series F, an increased pressure led also to a

prolonged residence time of the monomer molecules and

growing particles in the reactor. Thus, series F shows how

the production rate evolved by increasing the pressure and

simultaneously decreasing the mean gas velocity. In series

C and D in Figure 13 the negative effect of a prolonged

residence time on the particle formation is clearly shown.

Hence, the curve propagation of series Fmight be explained

as a superposition of both effects of increased pressure and

residence time.

The particle growth is a stochastic process and depends

strongly on the collision frequency. At elevated pressure

monomer fragments collide more frequently to form

clusters. Similarly aggregates grow faster based on the

numerous collisions with clusters or monomer fragments.

We therefore assume that the increased mass production

rate at elevated pressure is based upon the increased

collision frequency.

In the investigated parameter range the particle size and

morphology is rather insensitive to pressure changes, such

that no significant differences were found on the TEM

micrographs for pressures between 100 and 400 Pa, even

though in literature (e.g.,[33] for TEOS derived particles)

larger particles are reported for increasing pressure.

The influence of the process pressure on the chemistry of

the disiloxane derived particles is shown in Figure 17 for

series F by comparing the FTIR peak area ratios. In general

less methyl groups were found for increasing pressure.

At 400 Pa and low mean gas flow rate of 5.4m � s�1 no



associated methyl groups could be detected any more by

FTIR analysis for the monomer TMDSO.

With rising pressure, more collisions of high energetic

electrons, neutrals and ions occurred. The higher collision

frequency increased not only the nucleation rate of

particles but had an influence on the chemistry of the

evolving particles as well. The beginning thermalization

of the plasma under such conditions[36] led to a higher

monomer fragmentation and more methyl groups were

transformed to volatile hydrocarbons. This resulted in less

methyl groups in the evolving nanoparticles.

High process pressures are therefore not only suited to

form relatively large amounts of nanostructures but yield

as well in more stoichiometric silica compared to lower

pressure settings.

4. Conclusion

We produced nanoparticles with a silica-like chemical

composition and primary particle size in the range of

approximately 10nm in a plasma downstream reactor.

Asmonomers the two disiloxanes HMDSO and TMDSO and

the two orthosilicates TEOS and TMOS were investigated.

The production rate was determined using filter weight

measurements, the morphology of the nanoparticles was

characterized using TEM and the composition and bond

structure was analyzed with XPS and FTIR, respectively.

In order to explain the found tendencies, we propose

a qualitative model to describe the different phases of

particle growth in a continuous plasma reactor such as the

PDR. Entering the discharge the monomer molecules build

radicals to shape first clusters also known as protoparticles.

These clusters aggregate and grow to larger structures of

several nanometers in diameter until further growth is

DOI: 10.1002/ppap.201100180


inhibited by the negative charge on the particles surface.

As soon as the particles reach the end of the discharge,

the surface charge drops and the primary particles

agglomerate to larger structures up to several hundred

nanometers. The remaining radicals and ions in the

afterglow are still able to attach on the surface of the

agglomerated structures and lead to a fourth growth

phase by accretion.

The primary particle size became smaller for low

monomer flow rates and high powers what could be

correlated to the increased particle charging under

the respective process conditions. On the other hand the

particle morphology was rather unaffected by changes of

the residence time, oxygen tomonomer ratio, and pressure

which is in good correlation with the proposed model

of a charge dominated particle growth.

A high nanoparticle production rate was favored by

high process pressures, high monomer flow rates, and low

residence times. The disiloxanes, with two silicon atoms in

the monomer, yielded in a higher particle mass per molar

flowrate than theorthosilicates,withonlyone siliconatom

in the monomer molecule. Nevertheless the fractional

conversion from monomer to silica like structures was

higher for orthosilicates.

The chemistry of the produced particles is mainly

dependent on the composition and chemical structure of

the monomer and the reaction pathways to the silica-like

nanoparticles. In general the chemistry of the monomer

was well represented in the derived particles since no

complete fragmentation of the monomer took place under

the investigated discharge conditions. As a consequence,

orthosilicates resulted in particles with higher silanol

content while particles derived from disiloxanes contained

still some methyl groups. The content of methyl groups in

the nanoparticles produced from disiloxanes decreased

with rising oxygen to monomer ratio, plasma power, and

process pressure.

The investigated plasma process yielded in similar

nanoparticle structures as obtained in flame synthesis

but at low temperature. This is important for the main

application of the plasma downstream reactor, where the

nanostructures are simultaneously produced and attached

to temperature sensitive substrate particles such as drugs,

polymers or other fine chemicals. With the knowledge

gained in this study it may be possible in future to tailor

the powder properties by means of plasma surface

modification even better, since the influence of the process

parameters on the nanoparticle formation is at least

partially understood.

Acknowledgements: The authors gratefully acknowledge supportof the Electron Microscopy Centre EMEZ of the Swiss FederalInstitute of Technology ETHZ, the company Tecnofil for the supply



of the filter material and financial support by the Claude &Giuliana Foundation (Switzerland).

Received: October 7, 2011; Revised: November 25, 2011; Accepted:November 29, 2011; DOI: 10.1002/ppap.201100180

Keywords: continuous reactor; organosilicon precursors; parti-cles; silicon oxide; synthesis

[1] A. Spillmann, A. Sonnenfeld, P. Rudolf von Rohr, PlasmaProcess. Polym. 2007, 4, S16.

[2] C. Roth, Z. Kunsch, A. Sonnenfeld, P. Rudolf von Rohr, Surf.Coat. Technol. 2011, 205, S597.

[3] C. Arpagaus, A. Sonnenfeld, P. Rudolf von Rohr, Chem. Eng.Technol. 2005, 28, 87.

[4] D. Vollath, J. Nanoparticle Res. 2008, 10, 39.[5] R. M. Young, E. Pfender, Plasma Chem. Plasma Process. 1985, 5,

1.[6] Y. Watanabe, J. Phys. D, Appl. Phys. 2006, 39, R329.[7] G. Nichols, S. Byard, M. J. Bloxham, J. Botterill, N. J. Dawson, A.

Dennis, V. Diart, N. C. North, J. D. Sherwood, J. Pharm. Sci.2002, 91, 2103.

[8] M. Shiratani, H. Kawasaki, T. Fukuzawa, T. Yoshioka, Y. Ueda,S. Singh, Y. Watanabe, J. Appl. Phys. 1996, 79, 104.

[9] C. Hollenstein, Plasma Phys. Control. Fusion 2000, 42, R93.[10] A. Bouchoule, L. Boufendi, Plasma Sources Sci. Technol. 1993, 2,

204.[11] L. Boufendi, A. Bouchoule, Plasma Sources Sci. Technol. 1994, 3,

262.[12] A. Garscadden, B. N. Ganguly, P. D. Haaland, J. Williams,

Plasma Sources Sci. Technol. 1994, 3, 239.[13] C. Hollenstein, W. Schwarzenbach, A. A. Howling, C. Cour-

teille, J. L. Dorier, L. Sansonnens, J. Vacuum Sci. Technol. AVacuum Surf. Films 1996, 14, 535.

[14] G. Beamson, D. Briggs, ‘‘High Resolution XPS of Organic Poly-mers’’, Wiley, Chichester 1992.

[15] D. A. Shirley, Phys. Rev. B 1972, 5, 4709.[16] H. Hofmeister, P. Kodderitzsch, J. Dutta, J. Non-Crystal. Solids

1998, 232, 182.[17] A. Camenzind, H. Schulz, A. Teleki, G. Beaucage, T. Narayanan,

S. E. Pratsinis, Eur. J. Inorganic Chem. 2008, 2008, 911.[18] D. Theirich, C. Soll, F. Leu, J. Engemann, Vacuum 2003, 71, 349.[19] M. Hahnel, V. Bruser, H. Kersten, Plasma Process. Polym. 2007,

4, 629.[20] A. Sonnenfeld, T. M. Tun, L. Zajickova, K. V. Kozlov, H. E.

Wagner, J. F. Behnke, R. Hippler, Plasmas Polym. 2001, 6, 237.[21] C. Hollenstein, C. Deschenaux, D. Magni, F. Grangeon, A.

Affolter, A. A. Howling, P. Fayet, ‘‘On the Powder Formationin Industrial Reactive RF Plasmas’’, Elsevier Science B.V.,Amsterdam 2000.

[22] P. G. Pai, S. S. Chao, Y. Takagi, G. Lucovsky, J. Vacuum Sci.Technol. A Vacuum Surf. Films 1986, 4, 689.

[23] Y. Inoue, O. Takai, Plasma Sources Sci. Technol. 1996, 5, 339.[24] A. Spillmann, A. Sonnenfeld, P. Rudolf von Rohr,Appl. Surf. Sci.

2008, 255, 1911.[25] S. M. Han, E. S. Aydil, J. Vacuum Sci. Technol. A Vacuum Surf.

Films 1996, 14, 2062.[26] T. Ihara, S. Kawamura, M. Kiboku, Y. Iriyama, Prog. Org. Coat.

1997, 31, 133.


134


[27] L. Korner, A. Sonnenfeld, R. Heuberger, J. H.Waller, Y. Leterrier,J. A. E. Manson, P. Rudolf von Rohr, J. Phys. D, Appl. Phys. 2010,43, 115301.

[28] K. Aumaille, C. Vallee, A. Granier, A. Goullet, F. Gaboriau,G. Turban, Thin Solid Films 2000, 359, 188.

[29] H. Anderson, T. T. Kodas, D. M. Smith, Am. Ceram. Soc. Bull.1989, 68, 996.

[30] M. Hundt, P. Sadler, I. Levchenko, M. Wolter, H. Kersten,K. Ostrikov, J. Appl. Phys. 2011, 109, 123305-1.

[31] J. Knipping, H. Wiggers, B. Rellinghaus, P. Roth, D. Konjhodzic,C. Meier, J. Nanosci. Nanotechnol. 2004, 4, 1039.



[32] D. Magni, C. Deschenaux, C. Hollenstein, A. Creatore, P. Fayet,J. Phys. D, Appl. Phys. 2001, 34, 87.

[33] T. Fujimoto, K. Okuyama,M. Shimada, Y. Fujishige, M. Adachi,I. Matsui, J. Appl. Phys. 2000, 88, 3047.

[34] C. Hollenstein, A. A. Howling, C. Courteille, D. Magni,S. M. Scholz, G. M. W. Kroesen, N. Simons, W. de Zeeuw,W. Schwarzenbach, J. Phys. D, Appl. Phys. 1998, 31, 74.

[35] H. D. Jang, H. Chang, Y. Suh, K. Okuyama, Curr. Appl. Phys.2006, 6, e110.

[36] C. Roth, S. Bornholdt, V. Zuber, A. Sonnenfeld, H. Kersten,P. Rudolf von Rohr, J. Phys. D, Appl. Phys. 2011, 44, 095201.

DOI: 10.1002/ppap.201100180

nanoparticle synthesis and growth in a continuous plasma reactor from organosilicon precursors

Documents