fabrication of rf circuit structures on a pcb material ... · fabrication of rf circuit structures...

116

Transactions of The Japan Institute of Electronics Packaging Vol. 2, No. 1, 2009

Fabrication of RF Circuit Structures on a PCB Material Using Inkjet

Printing-Electroless Plating and the Substrate Preparation for the

SameA. Sridhar*, M. A. Perik**, J. Reiding**, D. J. van Dijk* and R. Akkerman*

*Production Technology Group, University of Twente, P.O. Box 217, 7500 AE, Enschede, the Netherlands

**Saxion Hogeschool, 7500 KB, Enschede, the Netherlands

(Received August 12, 2009; accepted November 2, 2009)

Abstract

This paper describes the optimisation of the surface characteristics of a high-frequency substrate material widely used

in the PCB (printed circuit board) industry by means of CF4/O2 plasma etching, in order to make it suitable for the fab-

rication of RF (radio frequency) circuit structures by a combination of inkjet printing and electroless plating. A statistical

DoE (design of experiments) based on a CCRD (central composite rotatable design) was used to systematically vary the

plasma etching parameters and explore the characteristics of the etching process. This experimental design yielded 31

substrates, all of which were assessed in terms of surface energy, surface roughness and adhesion. Out of these sub-

strates, 5 were identified as having the most favourable surface characteristics. Finally, RF circuit structures in the form

of S-band filters were fabricated on these substrates using an inkjet printing-electroless plating combination, and the RF

performance of these structures was characterised and compared.

Keywords: Printed Circuit Board (PCB), Plasma Etching, Inkjet Printing, Stripline Filter, Design of Experiments

(DoE), Surface Roughness, Surface Energy, Adhesion

1. IntroductionInkjet printing is an additive fabrication method, seen as

an important enabler to realise (RF-) electronic circuit

structures on dielectric substrates. In recent years, consid-

erable progress has been made in fabricating electronic

circuits/circuit components using this method.[1–4]

Development of inks with low sintering temperatures,

such as nanoparticle-based silver ink, has resulted in wide-

spread research activity in this field.

Reliability and robustness are crucial factors that deter-

mine the extent to which a process or a product is suc-

cessful. Lately, important results concerning the reliability

of inkjet-printed circuit structures have been published.

Caglar et al.[5] and Sridhar et al.[6] discussed the mechan-

ical reliability of inkjet-printed silver structures in terms of

tensile adhesive performance, determined using the pull-

off test method. Caglar et al.[5] also performed a DMA

(dynamic mechanical analysis) to assess the effect of

dynamic mechanical stresses on inkjet printed NPS (nano-

particle silver) structures. Joo and Baldwin[7] developed a

new adhesion test method called the MBST (modified

button shear test) to estimate the interfacial fracture

energy of NPS films. Other publications, such as the one

by Kaydanova et al.,[8] have dealt with the adhesion per-

formance of inkjet-printed structures using qualitative test

methods like the scotch-tape test. However, the effects of

surface roughness and surface energy of the substrate

material on the mechanical reliability of inkjet-printed

structures has not been given sufficient attention so far,

though there are some exceptions including the work of

Park et al.[9]

Prior experience has shown that mechanical interlock-

ing of inkjet printed structures to the substrate, due to the

latter’s surface roughness, plays a dominant role in deter-

mining the adhesion. This is especially true in the case of

polymer substrates, as their chemically inert nature gener-

ally does not favour strong chemical bonding. Scotch-tape

tests done on inkjet-printed structures on commercially

available PCB laminates have indicated that the adhesive

strength was poor and unsuitable for further processing

117

steps. Figure 1 depicts the Scotch-tape used for one such

test, in which an inkjet printed silver track was almost com-

pletely removed from the substrate. A survey of the litera-

ture indicates that plasma etching offers a way to modify

the substrate surface characteristics, thereby promoting

adhesion.

This research was carried out with two principal objec-

tives: (1) to determine the influence of the surface cha-

racteristics of the substrate material on the adhesion of

structures inkjet-printed on them, so as to find the optimal

surface characteristics to realise better adhesion without

compromising the accuracy of the inkjet-printed struc-

tures; and (2) to fabricate S-band filter structures on the

optimised substrates using a combination of inkjet printing

and electroless plating. This fabrication method involves

inkjet printing of silver seed tracks and subsequently

copper plating them using an electroless plating process.

The growth of copper on the silver seed track resulting

from the plating process imparts the necessary thickness

to the RF structures. The mechanism behind the copper

growth has been explained by Kao and Chou.[10] The

fabrication of RF structures using the abovementioned

method was demonstrated by Sridhar, et al.,[11] with the

example of an S-band filter and an RF transmission line. In

order to determine the influence of the surface character-

istics of the substrate material on adhesion, a CF4/O2

plasma etching process was used to impart varying

degrees of roughness and surface energy to the substrate

material. The plasma process parameters were systemati-

cally varied using an experimental design based on a second-

order CCRD. The substrates with optimal surface charac-

teristics were selected based on surface energy calcula-

tions, surface roughness measurements, and adhesion

tests using the scotch-tape test method.

2. Materials and Methods2.1. Materials

RO4000 series high-frequency laminate (Rogers Corpo-

ration, USA) was used as the substrate material. It is a

glass-reinforced hydrocarbon/ceramic thermoset laminate

with a Tg (glass transition temperature) greater than 280°C.

A TePla 3067-E (Technics Plasma GmbH, Germany), an

industrial-scale barrel-type plasma-etching machine, was

used for the CF4/O2 plasma etching. TEC-IJ-040, an

organic silver-complex compound (InkTec Company

Limited, Korea), containing less than 77% Ag-complex by

weight, was used to print the silver seed tracks. A Jetlab-4,

a commercially available piezoelectric drop-on-demand

inkjet printer (MicroFab Technologies Inc., USA), was

used for the inkjet printing trials. A nozzle with a diameter

of 80 μm was used for printing. Electroless copper plating

was done using an Envision-2130 electroless copper system

(Enthone Inc., USA). The contact angle measurements

were done using an OCA (Dataphysics Instruments

GmbH, Germany), an optical contact-angle measuring

instrument. A DEKTAK surface profiler (Veeco Instru-

ments Inc., USA) was used for surface roughness mea-

surements. The SEM (scanning electron microscope)

images in this paper were captured using a JSM-6400

(JEOL Limited, Japan).

2.2. Experimental designThe experimental design techniques commonly used for

process analysis are full factorial, fractional factorial and

CCRD. CCRD gives sufficient information to describe the

majority of steady-state process responses.[12] It requires

much fewer runs when compared to the full factorial

design and gives a clearer picture about interactions

between the process variables than a fractional factorial

method.

The CCRD was chosen in such a way that it contains ‘2n’

factorial treatment designs, ‘2n’ axial or star points, and

sufficient replications at the centre of the design. Here, ‘n’

represents the number of process variables under study.

Initial plasma-etching trials showed that four factors,

namely power (P), time of exposure of the substrate to the

plasma (t), flow rate of O2 (f_O2), and flow rate of CF4

(f_CF4), are the most relevant parameters that need to be

studied. The operating pressure, which is generally con-

sidered important in plasma etching, could not be pre-set

in the available equipment. As a result, the CCRD con-

sisted of 16 factorial treatment designs, with 8 star points

and 7 centre points; thus, 31 experiments in total. In com-

parison, a full factorial design for the same process would

Fig. 1 Scotch tape after testing on a silver track printed onan untreated RO4000 laminate.

Sridhar et al.: Fabrication of RF Circuit Structures on a PCB Material (2/9)

118


have resulted in 34 = 81 experiments. Star points and rep-

licates were added to the design to estimate the curvature

and error of the model.[13] The response model for 4 vari-

ables can be expressed as[14]:

(1)

Here, Y represents the yield; β0 is a constant intercept; βi,

βii and βij represent linear, quadratic and interaction coef-

ficients, respectively; and xi represents the coded indepen-

dent variables. The DoE model was used to understand

the influence of the process parameters on the outcome of

the plasma-etching process. The experimental design and

the coded parameter levels are listed in Table 1; the corre-

sponding parameter names and actual values are listed in

Table 2. The magnitude of the process parameters are

much higher than those found in the literature. The reason

for this is that the plasma-etching equipment used for this

study is an industrial-scale machine with a large chamber

that necessitated these parameter values for effective etch-

ing. In Table 1, runs 1–16 represent the 2-level full factorial

model, runs 17–24 represent the star points, and runs 25–

31 represent the centre points of the model.

3. ExperimentsAs per the experimental design, 31 substrates, each

measuring 100 mm × 100 mm, were cut and then plasma

etched. In this etching process, the specimen, i.e. sub-

strate, is immersed in plasma containing gases that react

with it. At relatively high process pressures of more than

0.2 mbar, the mechanism for etching is predominantly

chemical; these chemical reactions are promoted by

radicals in O2 and CF4.[15] For the chosen flow rates, the

operating pressure was above 0.2 mbar for all the experi-

mental runs. The process temperature is a complex func-

tion of power input and heat-transfer phenomena and was

not controlled during the experimental runs. The substrate

temperature during all the runs was well below the glass

transition temperature of the substrate material; hence, it

was not expected to play a major role in the outcome of the

etching process.

After etching, the contact angle of water on these

substrates was measured for the purpose of calculating the

surface energy of the substrates. Neumann’s equation of

state, shown as Eqn. (2), relates surface energy with

contact angle. These calculations were done using a soft-

ware program coupled to the contact-angle measuring

system.

, (2)

where ‘θ ’ is the contact angle, ‘σS’ is the surface energy of

Y x x x xi i ii i ij i jj iiii

= + + += +===∑∑∑∑β β β β0

2

1

4

1

3

1

4

1

4

Table 1 CCRD for four coded process variables.

Runs x1 x2 x3 x4

1 –1 –1 –1 –1

2 1 –1 –1 –1

3 –1 1 –1 –1

4 1 1 –1 –1

5 –1 –1 1 –1

6 1 –1 1 –1

7 –1 1 1 –1

8 1 1 1 –1

9 –1 –1 –1 1

10 1 –1 –1 1

11 –1 1 –1 1

12 1 1 –1 1

13 –1 –1 1 1

14 1 –1 1 1

15 –1 1 1 1

16 1 1 1 1

17 –2 0 0 0

18 2 0 0 0

19 0 –2 0 0

20 0 2 0 0

21 0 0 –2 0

22 0 0 2 0

23 0 0 0 –2

24 0 0 0 2

25 0 0 0 0

26 0 0 0 0

27 0 0 0 0

28 0 0 0 0

29 0 0 0 0

30 0 0 0 0

31 0 0 0 0

Table 2 Actual values of the process variables.

CodeP (x1)watt

t (x2)min

f_O2(x3)

ml/min

f_CF4(x4)

ml/min

–2 2500 10 0 0

–1 2900 20 500 50

0 3300 30 1000 100

1 3700 40 1500 150

2 4100 50 2000 200

Cos eS

L

L Sθσσ

β σ σ= −− −( )2 12

119

the substrate, ‘σL’ is the surface tension of the liquid, and

‘β ’ is a constant with a value of 1.247E-04.

The next step was the surface roughness measurement,

using a DEKTAK surface profiler. Subsequent to surface

characterisation, rectangular test patterns (30 mm × 10

mm) were inkjet-printed on the substrates. The thickness

of the pattern was highly dependant on the surface rough-

ness and surface energy of the individual substrates, and

was difficult to characterise due to the pronounced rough-

ness of certain substrates. Measurements on selected sub-

strates after sintering of the test patterns indicated that the

thicknesses were in the order of 1 μm. Scotch tape tests

were done on these patterns to qualitatively rank the adhe-

sive strength of the substrates under study. The next step

was to study the spreading of the ink on the substrates in

order to identify the optimal substrate surface energy

value(s). For this purpose, a micropipette was used to

deposit ink droplets with a constant volume of 20 μ l. The

reason behind the deposition of such a large droplet, the

diameter of which was nearly 20 times that of a droplet

from the inkjet printer, was to make the ink cover a larger

area on the substrate, to keep the influence of local rough-

ness peaks and valleys minimal. Prior experience showed

that the roughness peaks and valleys on a roughened sub-

strate’s surface can be several times the thickness of an

inkjet printed droplet after spreading on the substrate.

Moreover, a local cluster of roughness peaks or valleys

will radically modify the spreading behaviour of such a

droplet.

4. Selection of Optimal SubstratesThe measured contact-angle values and the correspond-

ing surface-energy values are depicted in Fig. 2. In Figure

3, the measured surface roughness values in terms of Ra

are plotted. In both these graphs, specimen number 0 indi-

cates the untreated RO4000 series substrate. The numbers

of the substrates correspond to the runs as listed in Table 1.

For a given material, a substrate with high surface

energy provides better adhesive strength than one with

lower surface energy, by decreasing the contact angle of

the deposited liquid, thereby increasing the interfacial

area. Mechanical interlocking resulting from surface

roughness, which is a dominant factor in metal-polymer

adhesion, is also enhanced by a larger interfacial area. On

the other hand, if the surface is too rough, it is not possible

to inkjet print with accuracy. The edge as well as the cross-

sectional accuracy of printed structures is very important

for (RF–) electronic applications, especially in the fre-

quency range (S-band) dealt with in this research. For this

reason, we attempted to select substrates that were suffi-

ciently rough, but not too rough, with sufficiently high sur-

face energy, so that the printed droplets did not bead up

and have high contact angles.

It is interesting to note that three substrates with the

highest surface roughness (numbers 7, 8 and 22) also had

surface energy values among the highest measured. Fig-

ure 4 shows a scatter plot of surface roughness (Ra) versus

Fig. 2 Contact angle and surface energy values of theplasma treated substrates.

Fig. 3 Surface roughness values of the plasma treated sub-strates.

Fig. 4 Surface roughness versus surface energy (S.E.) andcontact angle (C.A.).


120


contact angle and surface energy. As seen clearly in this

graph, low surface roughness values yield high contact

angles and high surface roughness values result in low

contact angles. Similarly, the surface energy generally

increases with surface roughness. However, the relation-

ship between these physical quantities is too complex to

be derived based on these observations alone. Even

though an increase in surface roughness of the substrate

results in a higher contact area between the substrate and

the ink, resulting in lower contact angles, there are devia-

tions as seen in Fig. 4, since the substrate material con-

tains more than one phase. This complexity due to the

composition of the substrate material is also mentioned in

the ‘Results and Discussion’ section.

The results from the ranking-based adhesion tests and

droplet-spreading studies showed that substrates 4, 5, 15,

16, and substrates 25 to 31, exhibited the desired results

and met the requirements mentioned above. Since sub-

strates 25 to 31 were subjected to plasma treatment with

the same parameter set (as they represent the centre points

of the DoE model), only one, number 25, was chosen from

them. Thus, 5 substrates were chosen for fabricating the S-

band filter. As seen from Figs. 2 and 3, the surface energy

values of the chosen substrates are on the higher side

(> 45 mNm–1), and their Ra values are in the order of 1 μm,

which, incidentally, corresponds to the layer thickness of

the inkjet printed test structures. 5 substrates were chosen

instead of only 1 for the following reasons:

(1) It was desired to determine a range of surface rough-

nesses and surface energy values that are suitable and

not one particular value.

(2) Fabricating the S-band filter on a number of substrates

and comparing their performance will give an idea

about the repeatability of the fabrication method under

study.

Figure 5(a) depicts the RO4000 laminate before plasma

etching and Fig. 5(b) depicts the same material after

plasma etching; in this case, the parameter set used was

from the centre point of the DoE model. It can be clearly

seen from these SEM images that the etched substrate is

roughened and the filler material (silica) present in the

substrate is exposed. The surface energy of this substrate

was also higher, due to the change in functional groups on

the surface. The increased roughness and surface energy

resulted in better adhesion of the deposited ink. The sub-

strate shown in this image was one of those selected for

the fabrication of the S-band filter.

5. S-band FilterThe design of the S-band band-pass filter is shown in

Fig. 6. The reason for choosing to make a filter is that it is

relatively straightforward to compare performance charac-

teristics like return loss (S11), insertion loss (S21) and

passband.

The next step was the fabrication of the designed filter.

The selected substrates were cleaned to remove any impu-

rities present on their surface and dried in a convection

oven at 100°C for 30 minutes. After drying, seed tracks for

the filters were inkjet printed, with the substrate and the

ink at room temperature. A sintering step followed, in

which the printed substrates were heated at 150°C for 30

minutes, as specified by the ink supplier. Copper was elec-

troless plated on these seed tracks to impart the desired

thickness to the filter structure. The thickness of plated

copper was approximately 2.5 μm, which is sufficient for

the S-band due to the skin effect. More details concerning

the filter design, skin effect and plating process used in

(a)

(b)

Fig. 5 Micrograph of RO4000 laminate (a) before plasmaetching and (b) after plasma etching.

Fig. 6 S-band filter design; dimensions in mm.

121

this research have been dealt with in a separate journal

paper.[11] Figure 7 shows one of the fabricated filters. All

the filters were subjected to RF measurements using a net-

work analyser.

6. Results and Discussion6.1. Plasma etching and DoE model

The reproducibility of plasma etching, exemplified by

responses at the centre points of the design (25 to 31), can

be ascertained from Figs. 2 and 3. It can be seen that sur-

face roughness values are more reproducible than surface

energy values. However, the tolerance of the contact-angle

measuring device may also account for surface energy

variations. The experimental responses of the DoE were

analysed using MINITAB15 statistical software. The model

equations for the two variables under study, namely sur-

face roughness and surface energy, were derived based on

Eqn. (1).

Surface roughness model:

(3)

Surface energy model:

(4)

In these models, x1, x2, x3 and x4 represent the process

variables listed in Table 2. Substituting the coded values of

the process variables gives the yield in terms of μm for

Eqn. (3) and mNm–1 for Eqn. (4). It is clear from Eqn. (3)

that the O2 flow rate has the strongest effect on surface

roughness, as it has the highest β coefficient. The positive

sign of this coefficient indicates that the surface roughness

will increase with increasing O2 flow rate. The CF4 flow

rate also has a strong effect, albeit a negative one. As the

CF4 flow rate increases, the surface roughness values tend

to be lower, even at high O2 flow rates. This phenomenon

can be explained by fact that a high concentration of fluo-

rine radicals results in passivation via the formation of a

fluorinated surface layer, which inhibits attack, i.e. etching,

by oxygen.[16] The analysis also revealed that the effect of

power on surface roughness is not significant for the range

of parameter values studied.

As far as the surface energy is concerned, the O2 flow

rate again has the strongest effect, as seen from Eqn. (4).

In this case also, the CF4 flow rate has a strong negative

effect due to its contribution towards the formation of the

surface passivation layer. The interaction effect of the flow

rates of these gases are, understandably, very important as

well. The influence of power and time are insignificant.

The ANOVA (analysis of variance) showed that there is

a significant lack-of-fit in the models and they contain a few

observations with large residuals. These deviations can be

explained by the following influences:

(1) The presence of different phases in the substrate

(thermoset, silica and glass fibre) makes the etching

process highly complicated. SEM images revealed

that after plasma etching, a number of substrates had

their polymer top layer completely etched away,

revealing silica and, in some cases, the glass fibres

present underneath. Consider the surface roughness

model represented by Eqn. (3): the measured surface

roughness, i.e. the observed responses, might repre-

sent the roughness of the polymer top layer for a par-

ticular measurement, whereas for another measure-

ment, it might represent the roughness of the exposed

silica or glass fibre, or a combination of these materi-

als. The effect of the etching parameters is difficult to

predict in this case, exemplified by the very low pre-

dicted R-squared values, which indicate how well the

model will predict future data. For both models, this

value is less than 30%. The R-squared values for the

current dataset are about 85% for both models. The

presence of outliers in the model supports this expla-

nation.

(2) The influence of the operating pressure inside the

plasma equipment could be pronounced. Since it

could not be pre-set and hence was left out of the DoE

model, its influence could not be quantified.

6.2. RF characterisationNetwork analysis of the fabricated filters showed that

return and insertion losses of the 5 tested filters show

some variation, as depicted in Figures 8 and 9. This can be

Fig. 7 A fabricated S-band filter.

Y x x x x x

x

= + + + − +

+ +

0 643 0 099 0 168 0 32 0 19 0 034

0 044 01 2 3 4 1

2

22

. . . . . .

. .1131 0 011 0 042 0 0750 101 0 138 0

32

42

1 2 1 3

1 4 2 3

x x x x x xx x x x

+ + +− + −

. . .. . .. .16 0 2352 4 3 4x x x x−

Y x x x x x

x x

= − − + − −

− + −

50 32 1 43 1 35 7 6 5 05 0 27

1 7 0 731 2 3 4 1

2

22

32

. . . . . .

. . 00 77 0 65 0 911 21 0 81 1 13 3 59

42

1 2 1 3

1 4 2 3 2 4

. . .. . . .

x x x x xx x x x x x

− +− + − + xx x3 4


122


attributed to variations in the cross-sections of the seed

tracks and differences in the dielectric properties of the

substrate. The cross-sectional variation is due to differ-

ences in the surface characteristics of these substrates.

For a given set of parameters of the inkjet-printing process,

if the surface characteristics like surface roughness and

surface energy are different for different substrates, then

the spreading behaviour of the ink droplet also varies,

resulting in differing dimensions of the printed structures.

Variation in the dielectric properties between substrates

Fig. 8 Return loss (S11) of the tested filters.

Fig. 9 Insertion loss (S21) of the tested filters.

123

could have been caused by the different parameter sets

used for the plasma etching of these substrates, resulting

in differences in the extent of etching. It can be seen from

Fig. 8 that the filter fabricated on substrate 15 has the low-

est return loss in the passband. However, the insertion

losses of all the filters in the passband, as seen in Figure

9, agree well with each other and are close to 0 dB, indi-

cating very low losses during transmission through the fil-

ters. The passbands of the filters are in fair agreement with

each other.

The network analyses revealed that it is indeed possible

to fabricate RF structures using the process combination

of inkjet printing and electroless plating; in addition, the

repeatability of this fabrication method was also verified.

7. ConclusionsThe goals stated at the beginning of this paper were

accomplished. Substrates with optimal surface characteris-

tics were identified based on DoE, and S-band filters were

fabricated on them using a combination of inkjet printing

and electroless plating. The RF measurements proved the

validity of the chosen fabrication method and brought the

importance of consistency in substrate surface characteris-

tics to the fore. DoE analysis showed that the model devel-

oped based on the experimental results does not account

for all the outcomes. The reasons for this are two-fold and

have been explained. Future work will involve refining the

DoE model by including plasma-process pressure as a

parameter.

AcknowledgementsThis research falls under the auspices of the PACMAN

project, promoted by SENTER, the Netherlands. The

authors would like to gratefully acknowledge the contribu-

tions of Mr. J. Mannak and Dr. R. Legtenberg, both of

Thales Nederland B.V. The contribution of their project

partner, ASTRON (Netherlands Institute of Radio Astron-

omy), is acknowledged as well.

References

[1] D. Redinger, S. Molesa, S. Yin, R. Farschi and V.

Subramanian, “An ink-jet-deposited passive compo-

nent process for RFID”, IEEE Transactions on Elec-

tron Devices, Vol. 51, No. 12, p. 1978, 2004.

[2] Y. Yoshioka, P. D. Calvert and G. E. Jabbour, “Simple

modification of sheet resistivity of conducting poly-

meric anodes via combinatorial ink-jet printing tech-

niques”, Macromolecular Rapid Communications,

Vol. 26, No. 4, p. 238, 2005.

[3] Y. Liu and T. Cui, “Polymer-based rectifying diodes

on a glass substrate fabricated by ink-jet printing”,

Macromolecular Rapid Communications, Vol. 26,

No. 4, p. 289, 2005.

[4] T. H. J. van Osch, J. Perelaer, A. W. M. de Laat and

U. S. Schubert, “Inkjet printing of narrow conductive

tracks on untreated polymeric substrates”, Adv-

anced Materials, Vol. 20, No. 2, p. 343, 2008.

[5] U. Caglar, K. Kaija and P. Mansikkamaki, “Analysis

of mechanical performance of silver inkjet-printed

structures”, 2nd IEEE International Nanoelectronics

Conference, INEC 2008, p. 851.

[6] A. Sridhar, D. J. van Dijk and R. Akkerman, “Charac-

terisation of the interface between inkjet printed

components and substrate materials”, 2nd DPI Inkjet

Workshop, Eindhoven, the Netherlands, 2007.

[7] S. Joo and D.F. Baldwin, “Performance of silver

nano particles as an electronics packaging intercon-

nects material”, Electronic Components and Tech-

nology Conference 2007, p. 219.

[8] T. Kaydanova, A. Miedaner, J. D. Perkins, C. Curtis,

J. L. Alleman and D. S. Ginley, “Direct-write inkjet

printing for fabrication of barium strontium titanate-

based tunable circuits”, Thin Solid Films, Vol. 515,

No. 7–8, p. 3820, 2007.

[9] S.-C. Park, S.-H. Cho, H.-C. Jung, J.-W. Joung and Y.-

B. Park, “A study on improvement of interfacial

adhesion energy of inkjet-printed Ag thin film on

polyimide by CF4 plasma treatment”, Korean

Journal of Materials Research, Vol. 17, No. 4, p. 215,

2007.

[10] C.-Y. Kao and K.-S. Chou, “Electroless copper plat-

ing onto printed lines of nanosized silver seeds”,

Electrochemical and Solid-State Letters, Vol. 10,

No. 3, p. D32, 2007.

[11] A. Sridhar, J. Reiding, H. Adelaar, F. Achterhoek, D.

J. van Dijk and R. Akkerman, “Inkjet-printing- and

electroless-plating - based fabrication of RF circuit

structures on high-frequency substrates”, Journal of

Micromechanics and Microengineering, Vol. 19,

doi: 10.1088/0960-1317/19/8/085020, 2009.

[12] D. P. Obeng, S. Morrell and T. J. Napier-Munn,

“Application of central composite rotatable design to

modelling the effect of some operating variables on

the performance of the three-product cyclone”,

International Journal of Mineral Processing, Vol.

76, p. 181, 2005.


124


[13] K. Acikalin, F. Karaca and E. Bolat, “Central com-

posite rotatable design for liquefaction of pine

barks”, Fuel Processing Technology, Vol. 87, p. 17,

2005.

[14] B. Manohar and S. Divakar, “Application of central

composite rotatable design to lipase catalysed syn-

thesis of m-cresyl acetate”, World Journal of Micro-

biology & Biotechnology, Vol. 18, p. 745, 2002.

[15] D. Hegemann, H. Brunner and C. Oehr, “Plasma

treatment of polymers for surface and adhesion

improvement”, Nuclear Instruments and Methods in

Physics Research B, Vol. 208, p. 281, 2003.

[16] S. Sapieha, A. M. Wrobel and M. R. Werthheimer,

“Plasma-assisted etching of paper”, Plasma Che-

mistry and Plasma Processing, Vol. 8, No. 3, p. 331,

1988.

fabrication of rf circuit structures on a pcb material ... · fabrication of rf circuit structures...

Documents