Instead of being wrapped around roller 1, the nickel mold was aligned and then stacked over the green ceramic substrate, as shown in Fig。 7, and a supporting plate made of Flame Retardant 4 (FR4, a typical material used for making a printed circuit board) was adopted to provide a hard surface for roller embossing in order to minimize the influence of roller 2。 Comparing with the approach of wrapping a mold directly on a roller, the method illustrated in Fig。 7 has advantages of: (1) ease in mold attachment;

(2) ease in mold replacement; (2) reduction in limitation of mold thickness and lateral  dimension。

2。3 Green ceramic substrates

In this study, micro roller embossing started from laminating green ceramic tapes to a stacked  substrate。   Commercially

Fig。 6 Three dimensional measurement of a  square  inductor  in Group B on the nickel mold。 The line-width and pitch were 50 and 100 lm, respectively, with a pattern height being about 38    lm

Fig。 7 A scheme of micro roller embossing。 The substrate is sandwiched with the mold above it and a rigid supporting plate beneath it

available low temperature co-firable ceramic green tapes (HL2000® from Heraeus, Germany) were used in this study, which has a near-zero shrinkage ratio in its x–y plane but  a

high shrinkage ratio in its thickness direction。 Four layers of such green tapes (150 mm 9 150 mm) were aligned and sealed in a vacuum package for squeezing out air bubbles, and then laminated by means of an isostatic laminator。 The lamination pressure, temperature and  holding  time were 10 MPa, 70°C and 10 min, respectively。 A green substrate of 0。5 mm thick was obtained after lamination。

3 Experimental results

3。1 Micro roller embossing process

Micro roller embossing was performed according to pro- cedures described below:

Fig。 9 Effect of pressure versus depth of embossed channels (75 lm channel-width and 300 lm pitch) within unit LC。 Process temperature and feeding speed were 75°C and 1。6 mm/s,   respectively

1。 Process parameters, i。e。 pressure, temperature and feeding speed, were pre-determined  via  a controller of roller embosser。

2。 The roller embosser was warmed up for about 10 minutes to stabilize the process pressure and temperature。

3。 The nickel mold, a green ceramic substrate and a FR4 supporting plate were sandwich structured and aligned with a guider for sample  feeding。

4。 Roller 1 was then lowered down for process execution。

While the rotation of roller 1 drove the sample to move through the roller-pair under pre-determined process parameters (pressure, temperature and feeding speed), micro patterns were then replicated on the green substrate with reversed profiles。 The ratio of polymeric additives versus ceramic powders in HL2000 green tapes was    only

16。5 wt%; this low ratio of polymeric additives would slow down material flow during roller embossing even at raised temperature。 Nevertheless, microstructure formation on large-area green ceramic substrate was successfully dem- onstrated using roller embossing。 Some process issues will be investigated in details in following  sections。

Fig。 8 Variations of embossed depth of embossed channels (75 lm channel-width and 300 lm pitch) in all six units at 70°C, 14 bars  and

1。6 mm/s

Fig。 10 Effect of feeding speed versus depth of embossed channels (75 lm channel-width and 300 lm pitch) within pattern UC。 Process temperature and pressure were 65°C and 10 bars,   respectively