• No results found

4. Device Fabrication

4.2. Detailed Process Workflow

4.2.2. Electrode Definition

The following paragraphs describe the fabrication of the top electrodes, including the small tips on the BLS structures.

BLS tips preparation layer. First, a Ti(5.0)/Au(120.0) layer is sputter-deposited uniformly over the substrate as indicated in Figure 4-3-D. The only function of this

layer is to enable the manufacturing of the small BLS tips at the end of the regular top electrodes. The Ti(5.0) layer provides improved adhesion of the Au layer onto the underlying SiO2. Although thicker adhesion layers such as Ti(10.0) and Ti(30.0) are more commonly employed as they provide even better bonding of the Au layer to the SiO2 passivation layer, the thickness of the Ti layer is purposely maintained at a minimum here. This is due to the lower ion-mill etch rate of Ti compared to that of Au and the etch time limit imposed by the FOX12 resist process in the definition of the small tips (see further).

Top electrode definition. Rather than instantly moving on to the definition of the small tips with e-beam lithography, first, the top electrodes are defined in a second Au(120.0)/Ti(5.0) layer that is sputter-deposited on top of the uniform Ti(5.0)/Au(120.0) layer already in place. The top electrodes are defined using conventional optical lithography and lift-off as illustrated in Figure 4-3-E. Note that optical alignment of the top electrodes is performed through the uniform Ti(5.0)/Au(120.0) layer. This poses no problem since the markers incorporated into the spin valve layer provide sufficient contrast for optical alignment purposes, even after they have been covered by the uniform Ti(5.0)/Au(120.0) layer. The purpose of the top Ti(5.0) layer is to provide an buffer ion-milling as to maintain maximum electrode thickness, which will be considered when the final ion-mill step is discussed.

E-beam marker clearing. At this point, the process has delivered a substrate carrying spin valve microstructures covered with an insulating SiO2 layer which are reached with conducting material through side and point contacts. Everything is covered with a uniform Ti(5.0)/Au(120.0) layer and the top electrodes have been defined in the Au(120.0)/Ti(5.0) layer, except for the small tips of the BLS structures. These tips will be defined using a FOX12 e-beam lithography step and ion-mill etching into the uniform Ti(5.0)/Au(120.0) layer, whose purpose now becomes clear. Because the BLS tips are only slightly oversized with respect to the point contacts they have to cover, exact location of these tips above the point contacts that were previously etched into the SiO2 layer is a prerequisite for obtaining a functional device, imposing strict alignment requirements on the e-beam definition of these tips. Note that the e-beam markers required for high-precision automated alignment by the e-beam system have been covered by a uniform Ti(5.0)/Au(120.0) layer. Of course, a 50 nm thick SiO2

layer is also present, but this layer is transparent to the e-beam alignment process. In order to provide sufficient contrast for alignment, the clearing of a set of e-beam markers which were covered by the uniform Ti(5.0)/Au(120.0) layer is required, which is depicted in Figure 4-3-F. This is performed through an additional optical lithography step that defines squares on top of the e-beam markers. The Au in these square areas is then removed using a chemical Au wet etching process using a 11.29 g

NaI, 64 g I2 and 800 ml H2O solution. The etch time is determined experimentally so that sufficient contrast is generated between the e-beam markers and the remaining gold. The marker clearance step is illustrated in detail in Figure 4-7.

Figure 4-7: The e-beam markers that were covered by the uniform Ti(5.0)/Au(120.0) layer have to be cleared to enable e-beam alignment and definition of the small BLS tips. The clearing is performed by defining squares on top of the e-beam marker crosses with conventional optical lithography, followed by chemical etching of the Au until sufficient contrast is obtained.

BLS tips definition. After the e-beam markers have been cleared, the process continues with the definition of the small BLS tips for the devices with optical access.

The 400 nm wide tips require precise definition and accurate alignment over the point contacts underneath and are therefore defined by an e-beam process using FOX12 flowable oxide negative resist. Upon exposure with energetic electrons, the FOX12 resist hardens into SiO2 that acts as a hard mask (illustrated by the black area in Figure 4-3-G) for the subsequent ion-milling of the tip pattern into the underlying uniform Ti(5.0)/Au(120.0) layer. The thickness and etch rate of the cured FOX12 determine the ion-mill time available before complete removal of the hard mask. The hard mask should provide sufficient ion mill time to remove the uniform layer in between structures, while at the same time maximum thickness of the tips should be maintained. The required ion-mill time for a spin-coated 80±5 nm FOX12 layer can be calculated based on an ion-mill rate calibration for cured FOX12. This rate was determined to be 15±2 nm/min, based on an ion-mill rate of 35 nm/min for Au (see Section 4.3). Based on this mill rate, total removal of the FOX12 layer results for a mill time between 4,4 and 6,5 minutes. The corresponding maximum pure Au layer thickness that can be milled in this time lies between 154 and 227 nm. This thickness determines the maximum height of the tips at the location where they make contact with the point contact. However, due to the insertion of a Ti adhesion layer (whose mill rate was determined to be much lower than that of Au at approximately 3 nm/min) between the SiO2 passivation and Au layer, the thickness of the uniform Au

layer for the combination with a 5.0 nm Ti layer is reduced to 120 nm. While a thicker Ti adhesion layer improves the adhesion of the Au to the SiO2 passivation layer, the maximum thickness limit on the Au part of the uniform layer would be reduced considerably because of the associated reduction in remaining time available for milling the Au part of the tips when a considerable amount of Ti is present. To maintain maximum thickness of the Au layer, the Ti adhesion layer is kept minimal at 5.0 nm. The insertion of a 5.0 nm Ti adhesion layer was observed to visibly enhance adhesion properties, preventing the Au to be easily scratched off when electrical connections to the devices were made through measurement probes, while it only marginally affected the thickness limit on the Au part of the layer (and thus, the total layer thickness).

The time required to mill entirely through the uniform layer was estimated based on the obtained etch rates for Ti, Au and cured FOX12 and verified experimentally by testing the (absence of) electrical conductivity between separate structures. The tips are milled in three steps of two minutes, while helium cooling is applied at the back of the substrate at all times. This is depicted in Figure 4-3-H.

Final device. The resulting device structure after the final ion-mill step is presented in Figure 4-3-I. The top electrode beginning at the left shows some overlap with the small BLS electrode tip that contacts the point contact. Note that the e-beam markers have disappeared during the last ion-mill step. The careful reader may wonder whether the steps for defining the BLS tips and the main part of the top electrodes may be reversed. While this is in principle possible, the procedure adapted above does not suffer from residual cured FOX12 that may impede the electrical contact in the overlapping area between the small tips and the rest of the electrodes.