Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion...

8
12 Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012 1. Introduction To realize copper wiring on resin materials with high frequency signal propagation, low electrical power con- sumption and high-density packaging, flat copper wiring on a smooth resin material with a low dielectric constant and a low dielectric loss is essential. Cycloolefin polymer (COP) resin is a promising material which satisfies these requirements, where its relative dielectric constant and dielectric loss tangent are around 2.35 and an order of 10 –4 for a wide frequency region within 70 GHz, respectively.[1] Furthermore, COP has excellent low moisture absorption characteristics. Thus, process technologies for realizing copper wiring on the smooth COP surface have been stud- ied.[2–5] Our group proposed adhesive copper seed-layer formation processes on the smooth-surface COP using magnetron sputtering as an alternative to the electroless deposition which usually requires an intentional rough- ness-induced process to obtain practical adhesion between the resin and the metal. Rough surfaces degrade the sig- nal propagation characteristics of wiring, especially as sig- nal frequency increases, because the traveling length of the electrical current, and hence conductor loss, increases due to the decrease of skin depth which defines the level of current flow from the metal surface. At the same time, electroless deposition involves a heavy environmental load. The adhesive seed layer can be obtained by introduc- ing consecutive processes of plasma nitridation of the COP surface, thin copper nitride film deposition, and copper deposition by magnetron sputtering before the electroplat- ing.[3, 4] Notably, because we have developed the novel magnetron sputtering equipment called rotation magnet sputtering which can realize high target utilization,[6] pro- duction costs can be reduced compared to the use of con- ventional magnetron sputtering equipment by realizing high productivity. In this paper, characteristics of the pro- posed adhesive COP-metal stacked layer are investigated [Technical Paper] Adhesion Characteristics of Magnetron-Sputter-Deposited Copper on Smooth Cycloolefin for Realizing Wiring with High-Frequency Signal Propagation Tetsuya Goto*, Takatoshi Matsuo**, Masamichi Iwaki*, Kazuki Soeda*, Ryosuke Hiratsuka*, Shigetoshi Sugawa*, and Tadahiro Ohmi* *New Industry Creation Hatchery Center, Tohoku University, Sendai 980-8579, Japan **R&D Center, ZEON Corporation, Kawasaki 210-9507, Japan (Received July 23, 2012; accepted September 25, 2012) Abstract Copper wiring formation on a resin material with a low dielectric constant, a low dielectric loss and a smooth surface is indispensable to realize high-frequency signal propagation with fine wiring patterns. Cycloolefin polymer (COP) resin is a promising material to meet these requirements. We propose adhesive copper seed layer formation on the COP by magnetron sputtering as an alternative to the electroless deposition which usually requires an intentional roughness- induced process to obtain practical adhesion between the resin and the metal. The proposed process steps include plasma nitridation of the COP surface, thin CuN film deposition and Cu film deposition before the electroplating. Excellent adhe- sion strength between the COP and the metal, greater than 1 kN/m, can be obtained while maintaining a smooth surface, which is attributed to the strong chemical bond generated between the nitrided COP surface and the CuN film. The coplanar transmission line was fabricated using the proposed process steps with semi-additive processes, and we found that the introduction of relatively high-resistive CuN film does not cause degradation of the propagation characteristics. Keywords: Cycloolefin Polymer, Magnetron Sputtering, Seed layer, Copper Wiring, Coplanar Line, Propagation Loss

Transcript of Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion...

Page 1: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

12

Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012

1. IntroductionTo realize copper wiring on resin materials with high

frequency signal propagation, low electrical power con-

sumption and high-density packaging, flat copper wiring

on a smooth resin material with a low dielectric constant

and a low dielectric loss is essential. Cycloolefin polymer

(COP) resin is a promising material which satisfies these

requirements, where its relative dielectric constant and

dielectric loss tangent are around 2.35 and an order of 10–4

for a wide frequency region within 70 GHz, respectively.[1]

Furthermore, COP has excellent low moisture absorption

characteristics. Thus, process technologies for realizing

copper wiring on the smooth COP surface have been stud-

ied.[2–5] Our group proposed adhesive copper seed-layer

formation processes on the smooth-surface COP using

magnetron sputtering as an alternative to the electroless

deposition which usually requires an intentional rough-

ness-induced process to obtain practical adhesion between

the resin and the metal. Rough surfaces degrade the sig-

nal propagation characteristics of wiring, especially as sig-

nal frequency increases, because the traveling length of

the electrical current, and hence conductor loss, increases

due to the decrease of skin depth which defines the level

of current flow from the metal surface. At the same time,

electroless deposition involves a heavy environmental

load. The adhesive seed layer can be obtained by introduc-

ing consecutive processes of plasma nitridation of the COP

surface, thin copper nitride film deposition, and copper

deposition by magnetron sputtering before the electroplat-

ing.[3, 4] Notably, because we have developed the novel

magnetron sputtering equipment called rotation magnet

sputtering which can realize high target utilization,[6] pro-

duction costs can be reduced compared to the use of con-

ventional magnetron sputtering equipment by realizing

high productivity. In this paper, characteristics of the pro-

posed adhesive COP-metal stacked layer are investigated

[Technical Paper]

Adhesion Characteristics of Magnetron-Sputter-Deposited Copper

on Smooth Cycloolefin for Realizing Wiring with High-Frequency

Signal PropagationTetsuya Goto*, Takatoshi Matsuo**, Masamichi Iwaki*, Kazuki Soeda*, Ryosuke Hiratsuka*, Shigetoshi Sugawa*,

and Tadahiro Ohmi*

*New Industry Creation Hatchery Center, Tohoku University, Sendai 980-8579, Japan

**R&D Center, ZEON Corporation, Kawasaki 210-9507, Japan

(Received July 23, 2012; accepted September 25, 2012)

Abstract

Copper wiring formation on a resin material with a low dielectric constant, a low dielectric loss and a smooth surface is

indispensable to realize high-frequency signal propagation with fine wiring patterns. Cycloolefin polymer (COP) resin is

a promising material to meet these requirements. We propose adhesive copper seed layer formation on the COP by

magnetron sputtering as an alternative to the electroless deposition which usually requires an intentional roughness-

induced process to obtain practical adhesion between the resin and the metal. The proposed process steps include plasma

nitridation of the COP surface, thin CuN film deposition and Cu film deposition before the electroplating. Excellent adhe-

sion strength between the COP and the metal, greater than 1 kN/m, can be obtained while maintaining a smooth surface,

which is attributed to the strong chemical bond generated between the nitrided COP surface and the CuN film. The

coplanar transmission line was fabricated using the proposed process steps with semi-additive processes, and we found

that the introduction of relatively high-resistive CuN film does not cause degradation of the propagation characteristics.

Keywords: Cycloolefin Polymer, Magnetron Sputtering, Seed layer, Copper Wiring, Coplanar Line, Propagation

Loss

Page 2: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

13

Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (2/8)

in detail to gain further understanding of the mechanism

of generating excellent adhesion. Also, the high frequency

propagation characteristics of the coplanar transmission

line fabricated using the proposed process steps are pre-

sented.

2. Experimental SetupFigures 1(a) and 1(b) show the cross-sectional structure

of the newly developed rotation magnet sputtering and the

bird’s eye view of its magnet system, respectively.[6] The

magnet system includes helical magnets with alternative

polarities, and multiple moving plasma loops are excited at

the target surface by rotating the helical magnets using a

motor. This results in full erosion of the target surface, and

easily realizes a high target utilization, larger than 60%,

much better than the 30% realized in typical conventional

fixed rectangular magnet systems. In the experiment, a

rectangular 6-mm-thick copper target (350 mm × 112 mm)

was used. The width of the deposition region, which is

defined by a slit formed by a electrically grounded plate as

shown in Fig. 1(a), was 60 mm. A 0.1-mm-thick COP sub-

strate with a typical size of 50 mm × 100 mm was set on a

movable grounded stage below the target, and the film

deposition as well as the plasma treatment, such as plasma

nitridation, were performed by scanning the movable

stage through this deposition region. The moving speed of

the stage was set at 8 mm/s. Figure 2 shows the proposed

seed layer formation steps before the Cu electroplating.

The plasma nitridation was carried out by Ar/N2 plasma

with various N2 additive ratios [ratios of N2 flow rate to the

total flow rate (400 sccm in the experiment)]. The movable

stage was scanned two times for the plasma nitridation,

which corresponds to a plasma irradiation time at an arbi-

trary position of the COP substrate of 15 seconds. Hereaf-

ter, process time will be described in this way. Working

pressure in the chamber was 52 Pa. A 13.56 MHz-rf power

was applied to the Cu target with a power density of 0.25

W/cm2. Since this power density was small and working

pressure was relatively high compared to the typical pres-

sure for film depositions, the target bias voltage generated

was small. Thus, the sputtering of the Cu target and result-

ing film deposition on the COP surface were negligibly

small, at least in the short plasma treatment time of 15 sec-

onds. After the plasma treatments, a CuN film with a thick-

ness of 35 nm was deposited on the COP surface by reac-

tive sputtering for the condition of Ar/N2 = 490/210 sccm

with a total pressure of 0.67 Pa. The deposition time was 15

seconds. Then, Cu with a thickness of 280 nm was depos-

ited for an Ar flow rate of 585 sccm with a working pres-

sure of 0.67 Pa. The deposition time was 90 seconds. In

these depositions of CuN and Cu, dc power was applied to

the target with a power density of 1.26 W/cm2. After the

seed layer deposition, the samples were unloaded from the

magnetron-sputtering chamber, and Cu with a thickness of

22 μm was deposited by electroplating.

Adhesion strength, the force required per width (5 mm)

to peel the Cu film at 30 mm/min, was measured using a

peel testing machine. Scanning electron microscopy

(SEM), energy dispersive X-ray spectroscopy (EDX), X-ray

photoelectron spectroscopy (XPS), atomic force micros-

copy (AFM), and a contact angle measurement machine

were used to characterize the samples.

The coplanar transmission line was fabricated on the

COP surface using the widely used semi-additive method.

The seed layer was formed using the proposed process

steps. Then, a dry-film resist was laminated on the seed

layer surface, and that was followed by photo-lithography Fig. 1 (a) A cross sectional structure of the newly developed rotation magnet sputtering and (b) its magnet system.

Fig. 2 The proposed seed layer formation steps before the Cu electroplating.

Page 3: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

14

Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012

patterning of the dry-film resist. Then, Cu electroplating

was carried out, and that was followed by wet etching of

the dry-film resist. Finally, the exposed seed layer was

eliminated by slight wet etching to isolate the wiring pat-

tern. The high-frequency propagation characteristics were

measured using the network analyzer (E8364B Agilent

Technologies) and the G-S-G probe (I50-A-GSG-150 Cas-

cade Microtech).

3. Results and Discussion3.1 Adhesion characteristics

Figure 3 shows the adhesion strength between the COP

resin and the metal as a function of the N2 additive ratio in

the plasma nitridation processes. The samples were fabri-

cated by the process steps shown in Fig. 2. The N2 additive

ratio was changed from 0% to 60%, while the total gas flow

rate was kept at 400 sccm. Note that the N2 additive ratio of

0% means that the COP surface was irradiated by pure Ar

plasma, so there is no nitridation of the COP surface. In

addition, a case without the introduction of CuN film depo-

sition was carried out in the pure Ar-plasma treated case

(i.e., the Cu seed layer was deposited after pure Ar plasma

treatment), and plotted in Fig. 3, showing very poor adhe-

sion strength around 0.1 kN/m. On the other hand, when

the CuN film is introduced, adhesion strength improves

drastically to about 0.8 kN/m even in the pure-Ar plasma

treated case. Furthermore, by introducing the Ar/N2

plasma treatment, a strong adhesion of 1 kN/m or more

can be obtained for the N2 additive ratio of at least 15% or

more, as shown in Fig. 3. The results show that the intro-

duction of both plasma nitridation and the CuN film is

indispensable to obtain excellent adhesion. Figures 4(a)–

4(e) show the surface morphology of the COP resin sur-

face for the 1 μm × 1 μm area evaluated by AFM. The

result of the initial COP surface is shown in Fig. 4(a), and

Figs. 4(b)–4(e) show the cases of plasma treatments with

the N2 additive ratios of (b) 0% (pure Ar plasma treatment),

(c) 15%, (d) 30% and (e) 60%, respectively. In the plasma

treated case, the AFM measurements were carried out

after the removal of the Cu film (and of the CuN film)

formed on the COP by wet etching using hydrated ferric

oxide. Surface roughness, Ra, and the peak-to-valley

length, P-V, obtained from the measured data are also indi-

cated in the figure. Surface roughness, Ra, is defined by

the following surface integral over the measurement field:

RS

z x y Z dxdya AVE= ( ) − ∫∫1, , (1)

Fig. 3 Adhesion strength between the COP resin and the metal as a function of N2 additive ratio in the plasma nitrida-tion processes.

Fig. 4 The surface morphology of the COP resin surface for the area of 1 μm × 1 μm evaluated by AFM. The result of the initial COP surface is shown in (a), and (b)–(e) show the cases of plasma treatments with the N2 additive ratios of (b) 0% (pure Ar plasma treatment), (c) 15%, (d) 30% and (e) 60%, respectively.

Page 4: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

15

Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8)

where S is the area of the measurement field, z(x, y) is the

height at position (x,y) and ZAVE is the averaged height in

the measurement field. It is found that both Ra and the P-V

can be maintained at very low levels, around 0.5 nm and

less than 10 nm, respectively, even after the plasma treat-

ments and film depositions, demonstrating that the plasma

treatment does not significantly roughen the COP surface.

Figures 5(a)–5(d) show SEM images and EDX mappings

of surfaces of the stripped metal (Cu or CuN/Cu) films.

Measurements were carried out for the metal-COP-inter-

face side of the metal films stripped from the COP resin, as

illustrated in the bottom of the figure. Figure 5(a) shows

the case of the N2 additive ratio of 0% (pure Ar plasma

treatment) without introducing CuN film, while Figs. 5(b)–

5(d) show the cases of the N2 additive ratios of (b) 0%, (c)

30%, (d) 60% with CuN film, respectively. The adhesion

strengths of these samples were measured at (a) 0.1

kN/m, (b) 0.8 kN/m, (c) 1.2 kN/m and (d) 1.3 kN/m, as

shown in Fig. 5. The atomic compositions at the surface of

the stripped metal films were investigated by EDX mea-

surements for the regions defined by dotted lines in each

of the SEM images. The resulting EDX mappings of C and

Cu are shown at the bottoms of the SEM images, where

the colors are brighter as the amounts of C and Cu are

higher. Note that even when the CuN film was present, the

N component was under the detection limit of EDX

because of the thinness of the CuN film. Clear differences

of the surface structures of the stripped metal films and

the relative amounts of C on them were observed for the

different cases as follows. In the poor adhesion case (0.1

kN/m), where only the Ar plasma treatment was carried

out and the CuN film was not introduced [Fig. 5(a)], the

surface of the stripped Cu film is very smooth. However,

when the CuN film is introduced, adhesion strength

improves drastically to 0.8 kN/m even with the Ar plasma

treatment [Fig. 5(b)], and it seems from Fig. 5(b) that the

surface roughness of the stripped metal and relative

amount of C increases slightly. When both the Ar/N2

plasma treatment with an N2 additive ratio of 30% and the

CuN film are introduced, the adhesion strength improves

to 1.2 kN/m [Fig. 5(c)]. In this case, island shaped clus-

ters are observed on the surface in the SEM image, and

the corresponding EDX mapping indicates that these clus-

ters consist of C. Thus, it is expected that the clusters con-

sist of COP resin. When the N2 additive ratio is further

increased to 60%, where the maximum adhesion strength

of 1.3 kN/m is obtained [Fig. 5(d)], the amount of C is the

maximum of all the cases, and a very rough surface with a

dimple-like structure is observed. It is expected from

these results that the entire surface is covered with the

COP resin in the case shown in Fig. 5(d). This also sug-

gests that the peeling takes place at the inside of the COP

resin for the entire region, because the adhesion strength

is better than the mechanical strength of the COP. On the

other hand, for the poor adhesion case shown in Fig. 5(a),

the amount of C on the film is relatively small and a very

smooth surface can be seen in the SEM image, indicating

that the peeling takes place between the COP resin and

the metal layer due to poor adhesion. It is noted that the

surface morphologies of the clusters seen in Fig. 5(c) and

the entire surface seen in Fig. 5(d) are clearly different,

where the surface in the latter case is very rough with the

dimple-like structure described above. Although the rea-

son is not clear at present, the dimple-like structure might

result from the generation of strong tension inside the

COP resin during the peel test due to the strong adhesion.

As for the Cu component in the EDX measurement, there

Fig. 5 SEM images and EDX mappings of surfaces of the stripped metal (Cu or CuN/Cu) films. Measurements were carried out for the metal-COP-interface-side of the metal films stripped from the COP resin, as illustrated in the bottom of the figure. (a) shows the case of the N2 additive ratio of 0% (pure Ar plasma treatment) without introducing CuN film, while (b)–(d) show the N2 additive ratios of (b) 0%, (c) 30%, (d) 60% with introducing CuN film, respectively.

Page 5: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

16

Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012

are no significant differences between the various cases,

indicating that the thickness of the attached C layer is suf-

ficiently thinner than the measurement depth of EDX,

which is a few μm.

3.2 Measurements of plasma-treated COP surfaceTo investigate chemical changes in the COP surface

induced by the plasma treatment, XPS was carried out for

the COP surface after the plasma treatment; i.e., the

plasma-treated COP films without Cu and CuN film deposi-

tions were applied to the XPS measurements. Figure 6(a)

shows the atomic composition ratios of the COP surface as

a function of the N2 additive ratio in the 15-second plasma

treatment, which is the same condition as that in the sam-

ple fabrication for the adhesion measurements. The initial

atomic composition ratios before the plasma treatment are

also shown. The take-off angle of the secondary electron

was set at 45° in the XPS measurement. Note that a small

amount of Cu with a composition ratio less than 1% (near

the detection limit) is observed in the plasma-treated case

(which is not indicated in the graph). N is observed in the

Ar/N2 plasma treatment case, suggesting the nitridation of

the COP surface. In the case of the pure Ar plasma treat-

ment, the O component increases to about 10% from the

initial state of about 5%. It is believed that such an increase

of the O component is mainly induced by oxidation of dan-

gling bonds at the COP surface generated by the plasma

irradiation, which occurs with air exposure after the sam-

ple is unloaded (thus, this phenomenon is not related to

the in-situ processes where the seed layer is deposited

after the plasma treatment without air exposure). When

the N2 additive ratio increases, however, the O component

decreases as shown in Fig. 6(a). The results suggest that

the nitridation of the COP surface can effectively prevent

the COP surface from being oxidized even in air after

unloading the sample. Figure 6(b) shows the C1s photo-

electron spectra from the XPS measurements for the initial

sample, the pure Ar-plasma treated sample, and the Ar/N2

(30%)-plasma treated sample. A hump on the high-energy

side (around 290 eV) of the tail of the main spectral peak

can be clearly seen only in the Ar/N2–plasma treated case.

It has been confirmed that similar tails were also found for

the other N2 additive ratios, 15% and 60%. The results sug-

gest that these tail components result from the nitridation

of the COP surface. These XPS data, shown in Fig. 6, were

obtained for the surface of the COP. We also measured

depth profiles of the XPS data with the widely used in-situ

Ar ion beam etching technique. Figure 7 shows the depth

profiles of the photoelectron spectra of C1s, N1s and O1s

for the Ar/N2 (30%)-plasma treated sample. The spectra for

each atomic orbital of the topmost layer of the resin are

shown at the top of the figure, and the spectra of the next

three layers inside from the topmost layer are shown in

order in Fig. 7. The spectra of the inside layers were

obtained by cyclic Ar ion beam etchings and the XPS mea-

surements. The Ar ion beam flux density was set to corre-

spond to an etching depth of 1 nm in the case of SiO2 for

one cycle, which is a rather small etching rate. It is found

from Fig. 7 that the hump structure at the C1s photoelec-

tron spectrum and the spectral peak of the N1s photoelec-

tron spectrum are only observed in the topmost layer, and

these features are not observed for the second or deeper

layers after the Ar ion beam etching. These results sug-

gest that the thickness of the nitrided layer of the COP is

very thin with an order of nm.

We also checked the wettability of the COP surface after

the plasma treatment. It is generally difficult to discuss a

relationship between wettability and adhesion, especially if

the adhesion layer is formed by dry processes such as

Fig. 6 (a) Atomic composition ratios of the COP surface as a function of the N2 additive ratio in the plasma treatment for 15 seconds. (b) C1s photoelectron spectra in the XPS measure-ments in the cases of the initial sample, pure Ar-plasma treated sample and the Ar/N2 (30%)-plasma treated sample.

Fig. 7 The depth profiles of photoelectron spectra of C1s, N1s and O1s for the Ar/N2 (30%)-plasma treated sample.

Page 6: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

17

Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (6/8)

sputtering depositions. However, since surface chemical

changes can be quickly and easily checked, the wettability

measurements were carried out. Figures 8(a) and 8(b)

show the water contact angle of the COP surface as a func-

tion of (a) the N2 additive ratio in the 15-second plasma

treatment and (b) the nitridation time for the N2 additive

ratio of 30%, respectively. The contact angle of the COP

surface without plasma treatment was 99°, as denoted in

Fig. 8(a) and Fig. 8(b) (the value at 0 seconds). Since the

COP resin consists of nonpolar hydrocarbons, the wettabil-

ity of the COP resin is low, and thus, the contact angle of

the initial state is high (99°). It is found from Fig. 8(a) that

wettability increases and thus the contact angle decreases

with the plasma treatment. The contact angle in the pure

Ar-plasma treated case decreases to about 80° from the

initial angle of 99°, and further reduction to about 40° is

observed in the plasma-nitrided case. It is empirically

found that the inclination of the N2-additive-ratio depen-

dence of the reduction of the contact angle is qualitatively

similar to that of the increase of the adhesion strength in

the case of introducing the CuN film shown in Fig. 3. Since

the surface roughness can be kept low even after the

plasma treatment, as shown in Fig. 4, the change in the

contact angle is induced by a change in the chemical struc-

ture of the COP surface, not by an increase of surface

roughness. Because both nitridation and oxidation

increase polarity at the COP surface, the water contact

angle will decrease with such chemical structure changes.

Since the XPS results indicate the post oxidation of the

COP surface in the Ar-plasma treated case, it is considered

that the reduction of the contact angle observed in the

pure Ar-plasma case is induced by the post oxidation, and

that the further reduction of the contact angle obtained in

the Ar/N2-plasma treated case is induced by plasma nitrid-

ation at the COP surface. As for the nitridation time depen-

dence, Figure 8(b) shows that the contact angle decreases

steeply to 40° for the nitridation time of around 15 sec-

onds, and increases slightly and asymptotically approaches

about 55°. Although the reason of this asymptotic value is

not clear at present, the results suggest that the nitridation

time of 15 seconds is sufficient to obtain an effective

nitrided COP surface.

3.3 Coplanar transmission lineThe proposed metallization processes described above

were applied to the fabrication of a coplanar transmission

line in order to evaluate the electrical performance of wir-

ing. Figure 9 shows a SEM image of the pad region of the

fabricated coplanar transmission as well as its schematic

cross-sectional structure. The width of the signal line and

the spacing between the signal line and the ground were

200 μm and 20 μm, respectively, which corresponds

approximately to the characteristic impedance of 50 Ω.

The length of the coplanar line was 2 cm, and the Cu thick-

ness was 12 μm. Widely used semi-additive processes

were successfully applied to the fabrication of the coplanar

line with CuN and Cu film deposition by magnetron sput-

tering as the seed layer as shown in Fig. 9. The existence

of the CuN film increases the total resistivity of the wiring

and thus might degrade its electrical properties. There-

fore, it is important to evaluate the effect of the introduc-

tion of CuN. It has been already confirmed that CuN resis-

tivity increases with an increase of the N2 additive ratio in

the rf-dc coupled sputtering.[3] In this study, the CuN film

was deposited by dc sputtering to reduce the heat flux on

the COP substrate. We have also confirmed that the behav-

ior of the resistivity in the dc-sputtering case is similar to

Fig. 8 The water contact angle of the COP surface as a func-tion of (a) N2 additive ratio in the plasma treatment for 15 sec-onds and (b) nitridation time in the case of the N2 additive ratio of 30%, respectively.

Fig. 9 A SEM image of pad region of the fabricated coplanar transmission as well as its schematic cross-section structure.

Page 7: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

18

Transactions of The Japan Institute of Electronics Packaging Vol. 5, No. 1, 2012

that in the rf-dc coupled sputtering case, and that the resis-

tivity is about 18 μΩcm in this study, which is about 10

times higher than the pure Cu resistivity of 1.8 μΩcm. To

overestimate the effect of introducing the high-resistive

CuN film, the thickness of the CuN film was set at 0.14 μm

which was 4 times thicker than that in the case of the

aforementioned adhesion measurements. Figure 10 shows

the frequency dependence of the propagation loss (S21

parameter) per wiring length of the fabricated coplanar

line in the cases with and without the introduction of CuN

film. Note that since the COP surface was nitrided before

seed layer formation in both cases, the metalizing adhe-

sion even in the case without the CuN film was adequate to

at least measure the electrical properties by making the

G-S-G probe contact the pad of the coplanar line. It is found

that a signal propagation of –0.3 dB/cm at around 10 GHz

can be achieved in both cases, and also, no significant dif-

ference between the two cases is observed for the wide

frequency region within 50 GHz. It is well known that

when the frequency increases, the electrical current flows

mainly at the surface of the conductor, between the outer

surface and a level called the skin depth defined by δ ωµσ= 2 / , where ω is the angular frequency of the cur-

rent, μ is the absolute magnetic permeability of the con-

ductor, and σ is the conductivity of the conductor.[7] The

skin depth decreases as the frequency increases, and is

about 0.3 μm at 40 GHz (if a Cu conductivity value of 58 × 106

S/m is used), which is comparable to the thickness of the

CuN film. In the case of the coplanar line, however, the

electrical current mainly flows at the sidewall of the wiring

including the 12 μm-thick electroplated Cu layer. This

might be a main reason for no degradation of the propaga-

tion properties even when introducing the relatively-high

resistive CuN film.

4. ConclusionTo realize Cu wiring with high-speed signal propagation,

low electrical power consumption, and high density pack-

aging, adhesive copper seed layer formation processes

using magnetron sputtering technology was developed for

metallization on a smooth COP resin with good high-fre-

quency electrical properties such as low dielectric loss and

a low dielectric constant. By introducing the proposed in-

situ processes including plasma nitridation, thin CuN film

deposition, and Cu film deposition before the electroplat-

ing, excellent adhesion capability can be obtained. It is

found that the smooth surface of the COP can be main-

tained even when introducing the proposed plasma pro-

cesses, and thus, the excellent adhesion is attributed to the

strong chemical bond between the nitrided COP surface

and the CuN film. The XPS results and the water contact

angle measurements suggest the nitridation of the COP

surface. The proposed seed layer formation processes do

not use electroless plating, which requires an intentional

roughness-inducing process to obtain practical adhesion

between the resin and the metal, and entails a heavy envi-

ronmental load as well. A coplanar transmission line was

fabricated using the proposed seed layer formation pro-

cesses with the semi-additive processes, and we found that

the introduction of relatively-high resistive CuN film does

not cause degradation of the signal propagation character-

istics for the wide frequency region within 50 GHz. The

proposed technology will greatly contribute to next-gener-

ation fine wiring on resin materials with high-frequency

signal propagation.

References[1] S. Kaneko, Y. Kobayashi, and Z. Ma, “Heikougataen-

bankyousinki no sokuteigenkai ni kansuru kentou,”

IEICE Technical Report MW2010-79(2010-09), 2010

(in Japanese).

[2] H. Imai, M. Sugimura, M. Kawasaki, A. Teramoto, S.

Sugawa, and T. Ohmi, “High-Frequency Propagation

on Printed Circuit Board Using a Material With a Low

Dielectric Constant, a Low Dielectric Loss, and a Flat

Surface,” IEEE Trans. Components and Packaging

Technol., Vol. 32, pp. 415–423, 2009.

[3] T. Ohmi, T. Goto, H. Imai, M. Sugimura, and O.

Kawashima, “Proposal of Very High Performance and

High Density Printed Wiring Board and Its Very High

Productivity Manufacturing Processes,” IEEE Trans.

Components Packaging and Manufacturing Technol,

Vol. 1, Issue 4, pp. 486–494, 2011.

Fig. 10 A frequency dependence of propagation loss (S21 parameter) per wiring length of the fabricated coplanar line in the cases with and without the introduction of CuN film.

Page 8: Adhesion Characteristics of Magnetron-Sputter-Deposited ... · 15 Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (4/8) where S is the area of the measurement

19

Goto et al.: Adhesion Characteristics of Magnetron-Sputter-Deposited Copper (8/8)

[4] T. Goto, O. Kawashima, and T. Ohmi, “Adhesive Cop-

per Seed Layer Formation as an Alternative to Elec-

troless Deposition for Printed Wiring Board Fabrica-

tions by Rotation Magnet Sputtering,” International

Conference on Electronics Packaging 2010, TC3-1,

Sapporo, May 2010.

[5] K. Baba, Y. Nishimura, M. Watanabe, and H. Honma,

“Formation of Fine Circuit Patterns on Cyclo Olefin

Polymer Film,” Transaction of The Japan Institute of

Electronics Packaging, Vol. 3, No. 1, pp. 73–77, 2010.

[6] T. Goto, T. Matsuoka, and T. Ohmi, “Rotation magnet

sputtering: Damage-free novel magnetron sputtering

using rotating helical magnet with very high target

utilization,” J. Vac. Sci. Technol. A, Vol. 27, No. 4, pp.

653–659, 2009.

[7] R.E. Collin, Foundations for Microwave Engineering,

2nd Edition, McGraw-Hill, New York, 1992.

Tetsuya Goto was born in Chiba, Japan, in 1972. He received the B. S., M. S., and Ph.D. degrees in physics from Tsukuba University, Ibaraki, Japan in 1995, 1997, and 2000, respectively. He researched physics of high-temperature plasmas in Tsukuba University. In 2000, he moved to Tohoku University

where he is currently an Associate Professor in the New Industry Creation Hatchery Center, Tohoku University. He is currently engaged in advanced semiconductor process technologies includ-ing sputtering film depositions, plasma-enhanced chemical vapor depositions, and plasma etchings for semiconductor devices and packages, and flat panel displays.

Takatoshi Matsuo was born in Osaka, Japan, on December 4, 1981. He received the B. S. and M. S. degrees in material engi-neering from Tokyo Institute of Technology, Tokyo, Japan in 2004 and 2006.He joined Zeon Corporation, Tokyo, Japan, in 2006. In 2006–2012, he has been engaged in the research and development of new materials used in semiconductors.

Masamichi Iwaki was born in Hokkaido, Japan, in 1982. He received the B. S. and M. S., degrees in engineering from Hiroshima University, Higashihiroshima, Japan in 2006, and 2008, respectively. He joined NIPPON VALQUA INDUSTRIES, LTD., Tokyo, Japan, in 2008. In 2010-2012, he has been seconded to the New Industry Creation Hatchery Center, Tohoku University as a researcher. He was engaged in surface modification by plasma treatment. He is currently engaged in the development of sealing materials for semiconductor manufacturing equipment.He is a member of the Japan Society for Analytical Chemistry.

Kazuki Soeda was born in Hyogo, Japan, on June 9, 1980. He received the B. S. and M. S. degrees in engineering from Univer-sity of Shiga prefecture, Shiga, Japan, in 2003 and 2005, respec-tively. He joined Nichias Corporation, Shizuoka, Japan, in 2005. He is currently engaged in the development of inorganic fiber insulation technology.

Ryosuke Hiratsuka was born in Saga, Japan, on August 10, 1984. He received the B. S. and M. S. degrees in engineering from Kyushu Institute of Technology, Fukuoka, Japan, in 2008 and 2010, respectively. He joined Nichias Corporation, Shizuoka, Japan, in 2010. He is currently engaged in the study of surface modification of fluorine resin.

Shigetoshi Sugawa received his M. S. degree in 1982 in Physics from the Tokyo Institute of Technology and his Ph.D. degree in 1996 in Electrical Engineering from Tohoku University. During 1982–1999, he worked in Canon Inc., where he researched high S/N ratio solid-state imaging devices

and other electronic devices. In 1999, he moved to Tohoku Uni-versity and he is currently a professor at the Graduate School of Engineering, Tohoku University. He is currently engaged in researches on CMOS image sensors, high-performance ULSIs and advanced displays.

Tadahiro Ohmi was born in Tokyo, Japan, in 1939. He received the B. S., the M. S., and Ph.D. degrees in electrical engineering from Tokyo Institute of Technology, in 1961, 1963, and 1966, respectively. In 1972, he moved to Tohoku University. He was a Full Professor with Department of Electronics, Faculty of

Engineering in 1985, with Department of Electronic Engineering, Graduate School of Engineering in 1997, and with New Industry Creation Hatchery Center, Tohoku University in 1998. Dr. Ohmi has been a Senior Research Fellow with the New Industry Cre-ation Hatchery Center, Tohoku University since 2011. From 1972, He is engaged in researches on high-performance ULSI, high-speed flat-panel display, and advanced semiconductor process technologies based on ultraclean technology concept. His research activities are summarized as 1,400 original papers and 1,700 patent applications.