A new cost-effective setup for silicon bulk micromachining is presentedwhich makes use of a polymeric protective coating, ProTEK B2 coating,instead of a conventional hardmask. Different concentrations of KOH andbath conditions (pure, with surfactant, with stirrer, with both surfactant andstirrer) have been considered. ProTEK B2 coating exhibits good adhesionto Si substrates, no degradation, etching rates and surface roughnesscomparable to literature data, and etching times greater than 180 minwithout damaging front side microstructures. Microcantilevers have alsobeen fabricated using two different process flows in order to demonstrate thesuitability of such a protective coating in microelectromechanical system(MEMS) technology.
Bulk micromachining often requires the deposition of ahardmask layer, such as silicon nitride or silicon oxide, on thesample front side to prevent undesired etching of the alreadypatterned structures [10, 11]. However, hardmask depositiontechniques such as low-pressure chemical vapour deposition(LPCVD) or plasma-enhanced chemical vapour deposition(PECVD) both exhibit some problems. LPCVD requires hightemperatures, which can induce unwanted mechanical stressin the layer [12], while PECVD silicon nitride and/or oxideexhibit a fair etch rate in KOH.
Alternatives to using a hardmask can be the use of aTeflon cell with O-rings or a polymeric protective coating.The first approach requires the development of a morecomplex experimental setup to mask the sample, while sealingand clamping of the cell can seriously compromise the effectiveness of this method. Furthermore, any possiblefracture of a membrane can directly compromise the integrityof the whole wafer surface. In contrast, the use of a polymericprotective coating, directly deposited on a Si surface, has beenrecently proposed as a very efficient and cost-effective solutionto the problem [13, 14]. This configuration does not introduceany mechanical stress and prevents the possible etching ofthe coated front side surface when a fracture of a membraneoccurs.
curs.The aim of this work is to investigate the compatibilityof a polymeric protective coating, ProTEK B2 coating,under different bath conditions during a KOH Si bulkmicromachining. For such purpose, three main parameterswere identified: the Si etch rate, selectivity of the Si etch ratewith respect to the SiO2 etch rate and the roughness of theetched surface. These three values have been characterizedwith varying KOH concentration (20%, 30%, 40% by weight)under four different bath conditions: pure, with sodiumdihexyl sulfosuccinate (SDSS) surfactant, with mechanicalagitation (by stirrer) and with both surfactant and stirreragitation.
Before the ProTEK B2 coating is deposited, a specificprimer (ProTEK Primer which produces a monolayer) mustbe spin-applied onto the Si surface in order to improve theadhesion of the film. Then a soft bake at 130◦–150 ◦Con a hot plate is required. The protective coating is easilyspin-applied onto samples at 1000 rpm for 90 s; then a hardbake is performed in three steps: 150 ◦C for 2 min, 170 ◦Cfor 2 min, 225 ◦C for 1 min, respectively. These steps arerequired to reach the optimal stability and adhesion of thepolymer, thus avoiding an unwanted lift-off during the KOHetch. With these conditions the average thickness of thecoating is 12 µm according to the receipt provided by thesupplier (Brewer Science). After wet etching, sample cleaningwas performed as follows: the protection layer is first madesoluble by applying a specific remover solution for about 20–40 min; then, for a total removal, an additional ‘piranha’ etchin a 3:1 solution of H2SO4 and H2O2, respectively, must beperformed for 10 min.
Fig1
Silicon etch-rate characterization is reported in figure 3.The results show a nonlinear decrease of the etch rate withincreasing KOH weight concentration (20%wt, 30%wt and40%wt), as reported in [10, 15]. For the different testedbath conditions, a minimal effect on the Si etch rate wasachieved by means of surfactant addition, while the stirreragitation played a more important role. This behaviourcould be due to the better minimization of the so-called‘mask bubble’ phenomenon [16] due to stirrer addition withrespect to surfactant addition. The combination of stirrerand surfactant does not seem to induce any noticeable effect,inducing roughly the same etch rate of the test with agitationonly. Since all the measurements are in fair agreementwith the wide dispersion of data reported in the literature,it is possible to argue that the ProTEKR B2 coating has no remarkable effect on the Si etch rate under the tested KOHbath conditions.
