-
IOP PUBLISHING PLASMA SOURCES SCIENCE AND TECHNOLOGY
Plasma Sources Sci. Technol. 17 (2008) 035024 (11pp)
doi:10.1088/0963-0252/17/3/035024
Repetitively pulsed atmospheric pressuredischarge treatment of
rough polymersurfaces: I. Humid air dischargesAnanth N Bhoj1,3 and
Mark J Kushner2,4
1 Department of Chemical and Biomolecular Engineering,
University of Illinois, Urbana, IL 61801, USA2 Department of
Electrical and Computer Engineering, Iowa State University, Ames,
IA 50011, USA
E-mail: [email protected] and [email protected]
Received 8 March 2008, in final form 29 May 2008Published 31
July 2008Online at stacks.iop.org/PSST/17/035024
AbstractPlasmas generated at atmospheric pressure are used to
functionalize the surfaces of polymersby creating new
surface-resident chemical groups. The polymers used in textiles
andbiomedical applications often have non-planar surfaces whose
functionalization requirespenetration of plasma generated species
into sometimes complex surface features. In thisregard, the
atmospheric pressure plasma treatment of a rough polypropylene
surface wascomputationally investigated using a two-dimensional
plasma hydrodynamics modelintegrated with a surface kinetics model.
Repetitively pulsed discharges produced in adielectric
barrier–corona configuration in humid air were considered to affix
O. Macroscopicnon-uniformities in treatment result from the spatial
variations in radical densities whichdepend on the polarity of the
discharge. Microscopic non-uniformities arise due to the
higherreactivity of plasma produced species, such as OH radicals,
which are consumed before theycan diffuse deeper into surface
features. The consequences of applied voltage magnitude
andpolarity, and the relative humidity on discharge dynamics and
radical generation leading tosurface functionalization, are
discussed.
1. Introduction
Plasma functionalization of polymers entails the modificationof
the topmost surface layers with chemically different atomsor
functional groups to modify surface properties. Forexample,
atmospheric pressure plasmas in air [1–3] arecommonly used to
functionalize the surfaces of hydrocarbonpolymeric materials by
affixing O-containing groups. Giventhe hydrocarbon backbone
(represented by RH) some ofthese groups are alcohols (R–OH), peroxy
(R–OO•), carbonyl(R–C=O) and acid (O=R–OH) [4, 5]. Higher O/C
ratios inthe chemical composition of the topmost layers increase
thesurface energy and so improve properties such as
adhesion,wettability or reactivity.
There is interest in using plasmas to modify thesurface
properties of natural and synthetic textiles as being
3 Present address: Novellus Systems, Inc., 3011 N. 1st St, San
Jose,CA 95134, USA.4 Author to whom any correspondence should be
addressed.
more environmentally friendly than wet processing [6, 7].Plasma
treatment of textile fibers improves shrink
resistance,hydrophilicity and color fastness [8,9]. Textile fabrics
usuallyconsist of fibers in a warp and weft arrangement,
creatingsurface features of a few to tens of micrometers [10].
Since theentire fabric surface is not completely exposed to the
discharge,the ability of plasma produced reactive species to
penetrate intosurface features is important to the quality and
uniformity oftreatment [11–13].
Biomedical surfaces are also often non-planar and rough,and
functionalized by plasmas to enhance (or prevent)
desiredinteractions with living cells [14]. For example,
plasmatreated implants and grafts have been used for many
years.Polymer substrates are also modified using plasmas for
cellmicro-patterning where selectively activated regions of
thesurface are affixed with functional groups which promotecell
adhesion [15]. Tissue engineering routinely involves
thefunctionalization of topologically rough scaffold-like
surfacesto promote cell adhesion and proliferation [16, 17].
0963-0252/08/035024+11$30.00 1 © 2008 IOP Publishing Ltd Printed
in the UK
http://dx.doi.org/10.1088/0963-0252/17/3/035024mailto:[email protected]:[email protected]://stacks.iop.org/PSST/17/035024
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
Common to both of these classes of applications is thedesire
that the plasma process uniformly treats the surfaceindependently
of the local topography which requires thatreactive plasma produced
species penetrate into this roughness.Gaining a deeper
understanding of the interaction betweengas phase transport
processes and surface reaction pathwaysleading to functionalization
will enable optimization of therelative abundance of functional
groups on the surface andeven the creation of tailored functional
group gradients [18].
In this paper, and in part II [19], we discuss the resultsfrom a
multiscale computational investigation of the plasmatreatment of
rough and porous polypropylene (PP) surfaces ina repetitively
pulsed atmospheric pressure corona dischargeusing an integrated
two-dimensional plasma hydrodynamics–surface kinetics model.
Discharges in air can efficientlyfunctionalize PP surfaces with
O-containing groups due tothe abundant generation of reactive
oxygen containing gasphase species such as O, OH and O3, and the
fact that O2can affix to radical sites. Although charged particles
do notplay a large role in directly functionalizing surfaces, we
foundthat the charged particle dynamics indirectly influence
theuniformity and the magnitude of treatment. For example,
thepolarity of the voltage pulse has an effect on the extent
ofplasma penetration into rough features. With positive
coronadischarges, the electron density near small features (a few
tomany micrometers) is low due to the formation of a
cathodefall-like region, thereby reducing the amount of electron
impactdissociation in the immediate vicinity of the feature.
Radicalstake longer to penetrate into features by diffusion
whereasrecombination of positive ions increases the supply
radicalsat the surface.
We found that variations in the relative humidity affectthe
microscopic uniformity of surface functionalization. Inlow humidity
discharges, gas phase OH radicals are generatedclose to the surface
as products of H abstraction by O atoms,in densities comparable to
that produced by electron impactof H2O in higher humidity
discharges. At higher appliedvoltages, the energy deposition per
pulse increases, producingmore O atoms. This results in a higher
overall degree offunctionalization but at the cost of increased
macroscopic non-uniformity.
The model used in the investigation is described insection 2 and
the results from these simulations are discussedin section 3. Our
concluding remarks are in section 4.
2. Description of the model and reactionmechanisms
The two-dimensional modeling platform, non-PDPSIM, hasclosely
coupled modules addressing plasma dynamics and gasphase kinetics.
These modules are the same as describedin [20]. Briefly, the plasma
dynamics module is used to solvePoisson’s equation for the electric
potential simultaneouslywith the multi-fluid charged and neutral
species conservationequations on an unstructured mesh. Poisson’s
equation and theconservation equations for charged species are
simultaneouslysolved. The conservation equations for electron
temperature,neutral species densities and flow properties are
sequentially
integrated in a time-splicing manner following the updatesof
Poisson’s equation and charged particle densities. Theseupdates are
then followed by integration of the continuityequations for
surface-resident species. A radiation transportmodule is used to
track photons emitted from designatedexcited species. These photons
photoionize the gas and interactwith the polymer surface.
The model is capable of simulating the repetitively
pulseddischarges (a few to tens of kilohertz). In these
discharges,the duration of the pulse is typically short (10s ns at
most)compared with the time between pulsing (usually �100s
µs).During the discharge pulse, all of the equations
addressingelectrostatics, and charged and neutral particle
transport aresolved. After the termination of the pulse and charged
particledensities have diminished to negligible values, the
interpulseperiod (IP) begins. During the IP, only the
conservationequations for neutral transport, and gas and surface
reactionkinetics are integrated. After the IP, the discharge is
reinitiated.This procedure is continued until quasi-steady state
conditionsare achieved. At this point, and with prior knowledge
thatfurther evolution of the surface coverages will not
significantlyaffect the plasma properties, the time-varying fluxes
of allplasma species at all locations on the surfaces are
recordedduring an additional discharge pulse and IP. The
recordedfluxes are then interpolated as a function of time
whileexecuting the surface kinetics module (SKM) for
additionalpulses and IPs.
The SKM is the same as that described in [20]. TheSKM consists
of a modified surface site-balance algorithmimplemented at the
nodes of specified materials that borderthe plasma. Briefly, the
initial conditions for the SKM are thesurface coverages of species
found on the virgin polymer.The SKM accepts fluxes of gas phase
species and photons,and implements a user-specified surface
reaction mechanismconsisting primarily of gas phase species
reacting on thesurface sites. The mechanism also allows for
photochemistryand reactions between surface sites. Outputs from the
SKMinclude the fractional coverage of surface-resident species
andreturning fractions for gas phase reactants and products.
The reaction mechanism used for humid air gas phasechemistry is
essentially the same as that described in [21]. Gasspecies include
N2, N2(A 3�), N∗∗2 , N
+2 , N
+, N+4 , N, N(2D),
O2, O2(a 1�), O2(b 1�), O+2, O−2 , O
−, O(1D), O(1S), O+,O, O3, H2O, H2O+, H2, H, OH and HO2. N∗∗2
nominallydenotes N2(b 1�, b′ 1�) but is used as a lumped
stateincluding transitions higher than N2(A 3�). In order
topropagate the positive corona streamer, photoionization
wasincluded with N∗∗2 being the emitting species with a lifetimeof
5 ns, producing photoionization of O2 with a cross sectionof 10−19
cm−2. Electron impact reactions with N2, O2 andH2O including
vibrational excitation, electronic excitation,attachment and
ionization are included in the calculation ofthe electron energy
distribution for generating rate coefficientseven though the
vibrational states are not explicitly includedin the model.
The most important gas phase reactions of relevance tosurface
treatment consist of neutral chemistry, initiated by
2
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
electron impact reactions of O2 and H2O producing O andOH
radicals,
e + O2 → O + O + e, (1)e + H2O → H + OH + e. (2)
At atmospheric pressure, three-body reactions dominatethe
radical kinetics, where O reacts with O2 to form O3 and Hreacts
with O2 to form HO2,
O + O2 + M → O3 + M (k = 6.9 × 10−34T −1.25g cm6 s−1),(3)
H + O2 + M → HO2 + M (k = 6.1 × 10−32T −1.6g cm6 s−1),(4)
where k is the rate coefficient at room temperature unless
thereis specific temperature dependence noted. There are few
majorgas phase consumption pathways for O3 at room temperatureand
so it tends to accumulate from pulse to pulse. O3 doesreact at slow
rates with H to form OH, with O to form O2 andwith OH to form
HO2,
H + O3 → OH + O2 (k = 1.4 × 10−10e−470/Tg cm3 s−1),(5)
O + O3 → O2 + O2 (k = 8 × 10−12e−2060/Tg cm3 s−1),(6)
OH + O3 → HO2 + O2 (k = 1.9 × 10−12e−1000/Tg cm3 s−1).(7)
O3 does undergo other dissociative processes, forexample,
reacting with O2 to form O2 and O. This process, aswell as others,
are included in the reaction mechanism howeverthey are important
only at higher gas temperatures than ofinterest here.
HO2 reacts slowly with O, H and OH forming radicalsor stable
products. OH and O react mutually at slow rates toform H,
HO2 + O → OH + O2 (k = 2.9 × 10−11e−200/Tg cm3 s−1),(8)
HO2 + OH → H2O + O2 (k = 8 × 10−11 cm3 s−1), (9)HO2 + H → H2O +
O (k = 2.4 × 10−12 cm3 s−1), (10)O + OH → H + O2 (k = 2.3 ×
10−11e110/Tg cm3 s−1).
(11)
The surface reaction mechanism is essentially the sameas that
discussed in [20]. Surface reactions are initiated byH abstraction
from the hydrocarbon backbone (RH) by gasphase radicals,
principally O and OH, creating active surfacealkyl sites (R•). The
probability of a reaction may vary with thelocal H bonding at the
site of the C atom due to bond polarityeffects, and that variation
is accounted for in the mechanism.As a result, secondary and
primary C atom sites were lessreactive than tertiary and secondary
sites by a factor of 5–10,respectively. The initiating steps
are
RH + O → R• + OH p = 10−3, (12)
RH + OH → R• + H2O p = 0.1, (13)
where p is the probability of reaction for the tertiary sites.
R•
sites rapidly react with O and O3 to form alkoxy (R–O•)
groups,
R• + O → R–O• p = 0.1, (14)R• + O3 → R–O• + O2 p = 1.0. (15)
In contrast, the reaction of R• sites with O2 to form
peroxy(R–OO•) groups has a lower probability, but due to the
largerdensities of O2 are fairly rapid,
R• + O2 → R–OO• p = 10−3. (16)Reactions between surface sites
also occur (with the rate
coefficients specified in units of cm2 s−1). For
example,abstraction of H atoms from surrounding sites by R–O•
groupsresults in alcohol (R–OH) groups with a rate coefficient
of10−16–10−15 cm2 s−1,
R–O• + RH → R–OH + R•, (17)which leaves an alkyl radical to
sustain further reactions.Alternatively the C–C bond at the R–O•
site cleaves, leading tochain scission and the formation of
carbonyl (HR=O) groups,and the creation of a new R• radical site at
rates of 103 s−1,
R–O• → R=O + R•. (18)The reaction of carbonyl groups with O and
OH createscarbonyl radicals (•R=O), which, upon further reaction
withOH, results in an acid group (HO–R=O),
HR=O + O → •R=O + OH, (19)HR=O + OH → •R=O + H2O, (20)
•R=O + OH → HO − R=O. (21)Continued chain scission similar to
that occurring with alkoxygroups leads to the formation of CO2, a
terminal step in thereaction chain,
•R = O + O → RH + CO2. (22)Other termination steps include the
recombination of alkylradicals on the backbone with H and OH
radicals,
R• + H → RH p = 0.2, (23)R• + OH → R–OH p = 0.2. (24)
The contribution of ions to the surface reactions inatmospheric
pressure discharges is less clear. The highcollisionality of
atmospheric pressure discharges results inions having near thermal
energy close to the surface. It isfeasible for ions arriving at the
surface to contribute to surfacechemical reactions in a manner
similar to their neutral radicalcounterparts. For instance, thermal
O+ and O+2 can reacton the surface in a manner similar to O and O2
since theyare typically neutralized prior to striking the surface.
Non-reactive ions may still break chemical bonds due to the
potentialenergy they bring or due to secondary electron
emission.There are no specific data available in the literature
for
3
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
the secondary electron emission coefficient (γ ) on
dielectricsurfaces due to ion bombardment at atmospheric
pressurethough estimates give γ = 0.01 [22]. In the context
ofsurface chemical reactions, ion bombardment accompanied
bysecondary electron emission most likely leads to the formationof
alkyl sites (R•). A sensitivity analysis was performed byvarying γ
from 10−3–1.0. It was determined that for thetransient discharges
investigated in this work, the contributionsof ion-initiated
reactions to surface functionalization are notsignificant due to
the short pulse durations.
Photons emitted from the plasma are thought to play amore
important role in surface kinetics [23–25]. UV photonsemitted with
energies greater than 4–5 eV can cleave C–Cbonds forming alkyl
sites. Skurat, et al [26] proposed that UVphotochemistry on
hydrocarbon polymers proceeds via threemajor pathways, the dominant
one being the elimination of H2forming adjacent alkyl radical sites
(–·RR·–). These sites havea higher probability of undergoing
scission to form R• sitesthan recombining [23],
RH + hν → •RR• + H2, (25)•RR• → R=R, (26)
•RR• → R• + R•. (27)The other two pathways result in the
formation of R•
radicals by breaking either the C–H bond releasing
atomichydrogen or the C–C bond scission in the PP backbone,
RH + hν → R• + H, (28)
RH + hν → R• + R•. (29)Skurat et al [26] estimated the relative
rates of equation (29) areabout 4–5 times faster than those of
equation (28). In this work,the total quantum yield of photons
leading to the formation ofalkyl (R•) sites was taken to be 0.2
[27].
Surface radicals (R•) on adjacent backbones can react witheach
other to crosslink [28–31]. The crosslinking reactionsincluded in
the mechanism include
R• + R• → R × R p = 10−15 cm2 s−1, (30)
2(•RR•) → R × R p = 10−15 cm2 s−1, (31)where × indicates a
crosslinked site.
3. Functionalization of rough surfaces in humid
airdischarges
The humid air, atmospheric pressure discharge functionaliza-tion
of PP with rough surface features of a few micrometerswas
computationally investigated. The device considered isa
corona–dielectric barrier similar to industrial
web-treatmentdevices [32], and is shown schematically in figure 1
as imple-mented in the model. The geometry is symmetric across
thecenterline. The upper metal electrode is housed in a
dielectricwith an exposed tip and is typically powered at a few to
10 kHz.The gap between the tip of the upper electrode and the
lowergrounded metal electrode is 2 mm. The rough PP surface is
Figure 1. Schematic of the corona treatment device. (a)
Typicalweb-roller arrangement for polymer treatment. (b)
Modelrepresentation of the discharge device with magnification of
therough polymer surface having micrometer-sized strand-like
features.
on the lower grounded electrode and so it effectively operatesas
a dielectric-barrier discharge. Roughness on the polymersurface is
resolved having strand-like features of a few mi-crometers to
resemble textiles or scaffold-like surfaces for celladhesion. Note
that the fine surface structure is included onlywithin 600 µm of
the centerline to reduce the computationalburden. In the following
discussion, macroscopic uniformityrefers to scale lengths of the
discharge, up to a few millimeters.Microscopic uniformity refers to
scale lengths of the surfacefeatures, many to tens of micrometers.
These uniformities arebeing examined for a stationary surface (that
is, a non-movingweb) without forced gas flow and so reflect the
fundamentalprocesses. As discussed in [20], macroscopic
non-uniformitiesmay be averaged out by motion of the web and forced
gas flow.
The unstructured mesh is locally refined close to thepowered
electrode and near the surface features with aresolution of about 1
µm so that reactor- and surface-scaleprocesses can be
simultaneously resolved. Regions distantfrom the discharge zone
have node spacings of 100 µm. Thetotal number of nodes is 21 296
with 9643 of these being in theplasma. Results from simulations
with a smaller number ofsurface features and increased resolution
did not significantly
4
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
Figure 2. Plasma properties after the first breakdown pulse in
anegative humid air discharge. (a) Densities of electrons,
positiveions and oxygen atoms. (b) Densities of electrons, O+2 ,
negativesurface charge and O atoms in the vicinity of the surface.
Thecontour labels are fractions of the maximum density noted in
eachfigure.
differ. The base case conditions are atmospheric pressureand
N2/O2/H2O = 79/20/1 (relative humidity of 30%). Thedischarge
operates at 10 kHz by biasing the powered electrodewith unipolar 15
kV pulses of either negative or positivepolarity.
The densities of electrons, all positive ions and O atoms atthe
time the plasma closes the gap after a propagation time of1.6 ns
are shown in figure 2 for a −15 kV pulse. The avalanche,terminated
by charging of the PP, produces a plasma densityof 1013 cm−3 in the
center of the gap and nearly 1014 cm−3 atthe tip of the electrode.
The major radicals for PP processingproduced by electron impact
dissociation are O and OH withdensities of a few times 1013 and
1012 cm−3 in the vicinity ofthe PP surface.
Densities of electrons, O+2, O, OH and surface charges atthe end
of the discharge pulse near the surface are also shownin figure 2.
The electrons and ions are able to penetrate intothe surface
features only to a limited extent. The size of the
Figure 3. Plasma properties midway during the first
interpulseperiod (duration 100 µs) of the negative discharge. (a)
Density of O,OH and O3 in bulk plasma. (b) Density of O, OH, O3 and
HO2 inthe vicinity of the surface. The contour labels are fractions
of themaximum density noted in each figure.
features are commensurate with the Debye length (λD ≈ 1–10 µm).
However, at times during and after the pulse, theion density inside
the features is low enough that it is notmandatory that charged
particle densities be quasi-neutral. Themore mobile electrons in
the leading edge of the avalanchenegatively charge the surface of
the features with local volumedensities of up to 100 µC cm−3. This
surface charging, whichoccurs dominantly at sites with large view
angles to the plasma,prevents electrons from deeply penetrating
into the features.What penetration of electrons there is continues
to produceO atoms by electron impact dissociation, though at the
end ofthe discharge pulse, radicals are largely absent from deep
insidethe features. Positive ions are accelerated into the features
bythe negative surface charge and fill the features with
densitiesof 1011 cm−3.
The densities of O, O3 and OH at 50 µs, halfway throughthe IP,
are shown in figure 3(a). At this time, the peak densityof O atoms
in the discharge region has been reduced by more
5
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
than an order of magnitude largely by the formation of
O3,producing an ozone density of nearly 1015 cm−3. Since themain
pathway for gas phase production of OH is electronimpact of H2O,
the density of OH steadily decreases duringthe IP in the bulk.
The densities of O, OH, O3 and HO2 near the surfacefeatures near
the end of the IP of 100 µs are shown infigure 3(b). At this late
time, diffusion has enabled thedensities of species that slowly
react with the surface to beessentially uniform over the scale of
the roughness. Neutralradicals continue to penetrate into surface
features by diffusion.O atoms, further depleted by the formation of
O3 and reactionswith the surface, have densities of 5 × 1011 cm−3
inside thefeatures. O3 and HO2, species which tend to accumulate
due totheir low reactivity, have densities of 8×1014 and 9×1013
cm−3inside the features. Earlier in the IP, the surface is a net
sourceof OH as a result of H abstraction by O atoms. By the end of
theIP, these sources decrease with the depletion of the O atoms.As
a result, the surface becomes a net sink for OH given itsown
reactivity with the surface, and the density of OH insidefeatures
decreases.
The densities of electrons, positive ions and O atomsduring
breakdown for a positive voltage pulse are shownin figure 4(a). As
is characteristic of positive coronastreamers compared with
negative streamers, the plasmacolumn is more confined because the
positive corona is morecritically dependent on space charge and
photoionization forpropagation. However, the charged particle
densities in thebulk plasma are commensurate between the positive
and thenegative discharges. With electrons drifting upwards,
whenthe avalanche strikes the surface charging is positive and
thedischarge spreads out forming a sheath like, cathode fall
regionat the surface. The source of electrons in this region
isdominantly by secondary emission by ions from the surface.The
bulk of the ions are driven into the features by electricfield
drift producing densities of 1011–1012 cm−3 and surfacecharge of up
to a positive 1 µC cm−3. Since few electrons enterinto the features
(
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
Figure 5. Plasma properties at the end of the first interpulse
period(duration 100 µs) of the positive discharge. (a) Densities of
O, OHand O3 in bulk plasma. (b) Densities of O, OH, O3 and HO2 in
thevicinity of the surface. The contour labels are fractions of
themaximum density noted in each figure.
positive discharge provides for more uniform fluxes on
themacroscopic scale.
The densities surface groups on the PP in the negativedischarge
during a single IP period [alkyl, (R•)] and afterseveral IPs
[alkoxy (R–O•) and peroxy (R–OO•)] are shownin figure 7. Alkyl
radicals are rapidly generated during andimmediately after a
discharge pulse by abstraction of H atomsfrom the PP backbone by O
and OH created within a fewmean-free paths of the surface. The
alkyl coverage reachesa maximum fractional value of 2 × 10−5 (2 ×
1010 cm−2)a few microseconds into the IP. For the remainder of
theIP, alkyl sites are consumed by O, O3 and O2 from the gasphase
producing alkoxy (R–O•) and peroxy (R–OO•) groups.This reduces the
density of alkyl sites to
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
(a)
(b)
(c)
Figure 7. Coverage of functional groups with position along
therough surface in repetitively pulsed negative discharges at
differenttimes. (a) alkyl radicals (R•), (b) alkoxy (R–O•) and (c)
peroxy(R–OO•).
The analogous time evolutions of alkyl (R•), alkoxy(R–O•) and
peroxy (R–OO•) coverages in the positivedischarge are shown in
figure 8. As with the negative discharge,the alkyl sites are
regenerated during and immediately aftereach pulse, but to
densities in excess of 2×1011 cm−2 due to thelarger fluxes of OH.
These sites are consumed by passivationdominantly by O3 and O2 in
the interpulse period. In contrast tothe negative discharge, there
is greater uniformity on the scaleof a few millimeters. This
results from the spreading of thepositive discharge on the PP,
which reduces spatial gradientsin the fluxes of O and O3. However,
slower radical penetration
(a)
(b)
(c)
Figure 8. Coverage of functional groups with position along
therough surface in repetitively pulsed positive discharges at
differenttimes. (a) alkyl radicals (R•), (b) alkoxy (R–O•) and (c)
peroxy(R–OO•).
into features results in increased microscopic non-uniformityof
alkoxy sites.
The coverage of alcohol (R–OH) and carbonyl (R–C=O)groups after
1 s of treatment is shown in figure 9 for positive andnegative
discharges. There are qualitative differences in bothsmall-scale
and large-scale coverages. Although the negativecorona has higher
coverages of carbonyl groups on axis, themore local production of O
and OH in the positive coronadischarges at larger radii increases
the coverage of all groupsat larger radii. Microscopic variations
in both carbonyl andalcohol groups are larger in the positive
discharge, similar tothat of the alkoxy coverage. The larger fluxes
of more reactive
8
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
Figure 9. Coverage of selected groups along the rough surface
after1 s treatment in repetitively pulsed negative and positive
discharges(a) carbonyl (R–C=O) and (b) alcohol (R–OH).
OH provides for higher and more uniform peak coverage
butdifferentiates more between sites having larger and smallerview
angles to the plasma. As such, the choice of polarity ofthe
discharge does give some degree of control of compositionand
uniformity. Positive corona discharges tend to have moremacroscopic
uniformity due to the spreading of the dischargeacross the surface.
At the same time, positive dischargeshave less microscale
uniformity due to the limited ability forelectrons to penetrate
deeper into the microstructure wherethey may produce radicals.
Through parametrization of the model and sensitivitystudies of
reaction probabilities, it was found that the mostlikely role of
low energy ions—abstracting H atoms from PPto create alkyl radical
sites—was not significant in atmosphericpressure corona discharges.
The reaction probability for an ionM+ reacting with a PP site to
abstract an H atom,
M+ + RH → R• + M–H, (32)was varied between 10−3 and 1.0. The
resulting change insurface functionalization over a treatment time
of 1 s wasnegligible. This was to be expected since the ions have
amuch shorter lifetime compared with neutral radicals and
theirdensities do not accumulate pulse to pulse. As a result,
theirtime-averaged fluxes to the surface are generally orders
ofmagnitude smaller than O, OH and O3. Although low energyions
might also produce chain scission and crosslinking, the
Figure 10. Photon properties and photon initiated products for
anegative discharge. (a) Photon fluxes at the end of the
dischargepulse and (b) density of crosslinked sites after 1 s of
treatment. Thesharp microscopic variations arise due to line of
sight photontransport being shadowed by surface features.
low fluxes of ions (and lack of accumulation on a
pulse-to-pulsebasis) reduce the importance of these processes.
Crosslinking is a terminal process for consumptionof surface
radical sites by creating non-reactive sites.Crosslinking can be a
significant process when the flux ofreactive gas species is low,
such as in rare gas discharges. Indischarges where the flux of
passivating species is large, thelikelihood for radical sites to
react with and be passivated bygas phase species prior to
crosslinking is large. UV photonsfrom the plasma contribute to
crosslinking by adding anotherpathway for chain scission
reactions.
For example, the UV photon flux along the surface at theend of
breakdown for a −15 kV discharge pulse is shown infigure 10(a).
These fluxes generally track the spatial gradientsin plasma
density. The line of sight transport of photons resultsin shadowed
surface features receiving a lower photon flux thansites that are
directly exposed to the discharge. The density ofcrosslinked sites
resulting from the UV illumination after 1 sof processing is shown
in figure 10(b). Crosslinking competeswith functionalizing and
passivating gas phase reactions. Atthose sites where fluxes of
reactive gas phase species islarge, surface radicals are more
likely to be passivated thancrosslinked. For the negative
discharge, these fluxes arehighest on axis and so the density of
crosslinked sites is loweron axis. For the same reasons, the
densities of crosslinked sites
9
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
Figure 11. Surface properties after 1 s treatment in
repetitivelypulsed discharges. (a) Surface coverage of peroxy
(R–OO•) for−12.5, −15 and −17.5 kV. (b) Surface coverage of alcohol
(R–OH)groups while varying the relative humidity.
are larger inside features on axis than outside the features.
Thisresults from the fluxes of passivating species being lower
insidethe features but there being sufficient penetration of the
plasmainto features that produces photons that photon produced
chainscission reactions occur. Off axis, the penetration of
plasmainto features is less and the shadowing of photon fluxes from
thehigh plasma density regions is more severe. In those
locations,there is more photon-induced crosslinking at the top of
featuresthan inside features.
The consequences of the magnitude of the negative appliedvoltage
on the coverage of peroxy groups (R–OO•) after 1 sof treatment are
shown in figure 11(a) With an increase inapplied voltage, the
densities of radicals generated in thedischarge increase and so
surface coverages increase. Themacroscopic uniformity decreases
with increasing voltage dueto a narrowing of the discharge (more
rapid gap closure)producing larger gradients in radical production.
These higherfluxes on axis begin to saturate the surface sites at
about 0.7coverage. Microscopic variations in coverage do not
changesignificantly as the local spatial dynamics of reactive
speciesfluxes are not sensitive functions voltages.
Relative humidity or the fraction of H2O, [f (H2O)] inair
ultimately determines the densities of OH generated in
thedischarge. Larger f (H2O) generally produce larger fluxes ofOH.
These radicals contribute both to more rapid H abstractionand to
the slower OH addition to alkyl (R•) sites producingalcohol (R–OH)
groups. The effect of f (H2O) on the coverage
of alcohol groups is shown in figure 11(b). At low f (H2O),tens
of ppm, the large fluxes of O atoms that abstract H atomslocally
produce OH densities that are comparable to the plasmaproduced flux
of OH in discharges having high f (H2O) of afew per cent. As a
result, increasing f (H2O) has a limitedeffect on functionalization
and production of alcohol groupson the macroscopic scale. Since
plasma produced OH fluxescome from outside of the features and OH
radicals have largerreaction probabilities with the surface, there
is little penetrationof plasma produced OH into features. The OH
fluxes insidefeatures dominantly result from H abstraction by O
atoms. Asa result, the alcohol coverage outside features is more
sensitiveto f (H2O) than inside features.
4. Concluding remarks
The functionalization of rough PP surfaces with featuresof a few
micrometers using repetitively pulsed atmosphericpressure
discharges in humid air was computationallyinvestigated. Electrons
and ions produced during the pulsepenetrate into the rough surface
features to a limited extent.The penetration of charged species
depends on the dischargepolarity. In negative discharges, there is
limited penetration ofelectrons into surface features which locally
produce reactivespecies by electron impact. In positive discharges
a sheath-like region near the surface prevents electrons from
penetratinginto the surface features and so there is less local
productionof radicals. Longer lived radicals are able to penetrate
into allsurface features over timescales of hundreds of
microseconds.The more reactive radicals such as OH are depleted
nearexposed surface features which reduces fluxes penetratinginside
the features. In atmospheric pressure discharges,the contributions
of ions to surface chemistry are limited.Although the contribution
of UV photons is small, somecrosslinking does occur at locations
where reactive fluxes ofpassivating radicals are low.
The dynamics of the discharge and radical generationhave
significant effects on the degree and uniformity
offunctionalization. The macroscopic distribution of
reactivespecies in the positive discharge is more uniform due
tospreading of the discharge along the surface, and this leadsto
higher macroscopic uniformity of treatment. However poorplasma
penetration into features produces more microscopicnon-uniformities
than in the negative discharge. Increasinghumidity had a limited
effect on alcohol coverage due to theformation of OH at the surface
by H abstraction by O atoms.
Acknowledgments
This work was supported by the National Science
Foundation(CTS-0520368). The authors thank Dr Mark Strobel for
hisadvice and guidance.
References
[1] Liston E M, Martinu L and Wertheimer M R 1993 J. Adhes.Sci.
Technol. 7 1091
[2] Sun C, Zhang D and Wadsworth L C 1999 Adv. Polym.Technol. 18
171
10
http://dx.doi.org/10.1163/156856193X00600http://dx.doi.org/10.1002/(SICI)1098-2329(199922)18:23.0.CO;2-8
-
Plasma Sources Sci. Technol. 17 (2008) 035024 A N Bhoj and M J
Kushner
[3] Strobel M, Lyons C S and Mittal K L (ed) 1994 PlasmaSurface
Modification of Polymers (Zeist, The Netherlands:VSP Press)
[4] Lynch J B, Spence P D, Baker D E and Postlethwaite T A
1999J. Appl. Polym. Sci. 71 319
[5] O’Hare L-A, Leadley S and Parbhoo B 2002 Surf.
InterfaceAnal. 33 335
[6] Borcia G, Anderson C A and Brown N M D 2004 Appl. Surf.Sci.
221 203
[7] Carneiro N, Souto A P, Silva E, Marimba A, Tena B, Ferreira
Hand Magalhaes V 2001 Coloration Technol. 117 298
[8] Hoecker H 2002 Pure Appl. Chem. 74 423[9] Kan C W 2007 J.
Adhes. Sci. Technol. 21 911
[10] Poll H U, Schladitz U and Schreiter S 2001 Surf.
Coat.Technol. 142–144 489
[11] De Geyter N, Morent R and Leys C 2006 Plasma Sources
Sci.Technol. 15 78
[12] Vohrer U, Muller M and Oehr C 1998 Surf. Coat. Technol.98
1128
[13] Tsai P P, Roth J R and Chen W W 2005 Textile Res. J. 75
819[14] Castner D G and Ratner B D 2002 Surf. Sci. 500 28[15] Pu F
R, Williams R L, Markkula T K and Hunt J A 2002
Biomaterials 23 2411[16] Jones I, Currie L and Martin R 2002 Br.
J. Plast. Surg. 55 185[17] Chim H, Ong J L, Schantz J-T, Hutmacher
D W and
Agrawal C M 2002 J. Biomed. Mater. Res. A 65A 327[18] Lee J H,
Kim H W, Pak P K and Lee H B 1994 J. Polym. Sci. A
32 1569
[19] Bhoj A N and Kushner M J 2008 Plasma Sources Sci.
Technol.17 035025
[20] Bhoj A N and Kushner M J 2007 J. Phys. D: Appl. Phys.40
6953
[21] Dorai R and Kushner M J 2003 J. Phys. D: Appl. Phys.36
666
[22] Tochikubo F, Chiba T and Watanabe T 1999 Japan. J.
Appl.Phys. 38 5244
[23] Wertheimer M R, Fozza A C and Hollander A 1999
Nucl.Instrum. Methods Phys. Res. B 151 65
[24] Wilken R, Hollander A and Behnisch J 1999 Surf.
Coat.Technol. 116–119 991
[25] Truica-Marasescu F, Jedrzejowski P and Wertheimer M R2004
Plasma Proc. Polym. 1 153
[26] Skurat V E and Dorofeev Y I 1994 Angew. Makromol. Chem.216
205
[27] Kuvaldina E V, Rybkin V V, Titov V A, Shikova T G andShutov
D A 2004 High Energy Chem. 38 411
[28] Poncin-Epaillard F, Vallon S and Drevillon B 1997
Macromol.Chem. Phys. 198 2439
[29] Poncin-Epaillard F, Brosse J C and Falher T
1997Macromolecules 30 4415
[30] Hong J, Truica-Marasescu F, Martinu L and Wertheimer M
R2002 Plasmas Polym. 7 245
[31] Hansen R H and Schonhorn H 1966 J. Polym. Sci. Lett. B4
203
[32] Strobel M, Jones V, Lyons C S, Ulsh M, Kushner M J, Dorai
Rand Branch M C 2003 Plasmas Polym. 8 61
11
http://dx.doi.org/10.1002/(SICI)1097-4628(19990110)71:23.0.CO;2-Thttp://dx.doi.org/10.1002/sia.1217http://dx.doi.org/10.1016/S0169-4332(03)00879-1http://dx.doi.org/10.1111/j.1478-4408.2001.tb00079.xhttp://dx.doi.org/10.1351/pac200274030423http://dx.doi.org/10.1163/156856107781393884http://dx.doi.org/10.1016/S0257-8972(01)01055-6http://dx.doi.org/10.1088/0963-0252/15/1/012http://dx.doi.org/10.1016/S0257-8972(97)00549-5http://dx.doi.org/10.1177/0040517505057526http://dx.doi.org/10.1016/S0039-6028(01)01587-4http://dx.doi.org/10.1016/S0142-9612(01)00377-5http://dx.doi.org/10.1054/bjps.2002.3800http://dx.doi.org/10.1002/pola.1994.080320818http://dx.doi.org/10.1088/0022-3727/40/22/016http://dx.doi.org/10.1088/0022-3727/36/6/309http://dx.doi.org/10.1143/JJAP.38.5244http://dx.doi.org/10.1016/S0168-583X(99)00073-7http://dx.doi.org/10.1016/S0257-8972(99)00282-0http://dx.doi.org/10.1002/ppap.200400020http://dx.doi.org/10.1002/apmc.1994.052160114http://dx.doi.org/10.1023/B:HIEC.0000048241.43342.f4http://dx.doi.org/10.1002/macp.1997.021980807http://dx.doi.org/10.1021/ma961585dhttp://dx.doi.org/10.1023/A:1019938424698http://dx.doi.org/10.1002/pol.1966.110040309http://dx.doi.org/10.1023/A:1022817909276
1. Introduction2. Description of the model and reaction
mechanisms3. Functionalization of rough surfaces in humid air
discharges4. Concluding remarks Acknowledgments References