A series of splits using MWA only is performed and summarized in Table 3-1.
Splits vary in power levels, one or two step MWA, and setup 1 versus setup 2. M1 is the Ni silicide formed by only one-step 360W MWA in setup 1 for 300sec. M2 and M3 are formed by two step MWA. The first step of M2 and M3 are as the same as M1, but different in the second step annealing conditions. The second step annwaling of M2 and M3 are 600W in setup 2 for 300sec and 1300W in setup 2 for 300sec, respectively. M4 is the Ni silicide formed by only one-step 360W MWA in setup 2
25
for 300sec. M4 and M5 are formed by two step MWA. The first step of MM5 and M6 are as the as M4, but different in the second step annealig conditions. The second step annealing of M5 and M6 are 600W in setup 2 for 300sec and 1300W in setup 2 for 300sec, respectively. The plot of the sheet resistance versus temperature for each condition is shown in Fig. 3-1. Sheet resistance compared to M4 (180 ohm/sq.), M1 (450 ohm/sq.) is much higher due to the relatively lower temperature in setup 1 than in setup 2. For M2 and M3 with the same first annealing condition to M1, stronger power (higher temperature) applied as second step annealing results in lower sheet resistance, which are 180 (ohm/sq.) and 100 (ohm/sq.) respectively. For M4 and M5 with the same first annelaing condition to M4, stronger power (higher temperature) applied as second step annealing results in lower sheet resistance, which are 100 (ohm/sq.) and 20 (ohm/sq.) respectively. So far, extremely low sheet resistance has achieved in condition M6.
The plot of the sheet resistance versus silicide thickness for each condition is shown in Fig. 3-2. Higher sheet resistance usually results in thin silicide thickness.
Silicide thickness compared to M4 (9nm), M1 thus reached ultrathin silicide thickness (3.2nm) with its high sheet resistance. For M2 and M3 with the same first annealing condition to M1, the sheet resistance is dramatically declined with very little increment of thickness, which are 4.7 nm and 6.5 nm from 3.2nm respectively. For M4 and M5 with the same first annealing condition to M4, the same conccept is confirmed as well. The sheet resistance is dramatically declined with very little increment of thickness, which are 10 nm and 10.5nm from 9 nm respectively.
Now we concentrate the concern into phase issue. The conventional tool for phase examinaiton is by XRD, however, the thickness formed by MWA is too thin to detect. Therefore, selective area diffraction pattern (SADP) is applied to confirm the
26
phase of ni silicde. SADP is a crystallographic experimental skill that could be performed inside the transmission electron microscope (TEM), the image on the screen of the TEM will be a series of spots each spot corresponding to a satisfied diffraction condition of the sample's crystal structure. As shown in Table 3-2, smaller spots surrounded with red halo-shaped stand for the diffraction pattern of NiSi. The large and bright spots stand for the diffraction pattern of Si. The result of SADP indicates that amorphous phase of Ni silicide in M1 due to no ring pattern was discovered. For M2 and M4-M6, the phase of Ni silicide is indicated to be NiSi die to the existence of the ring pattern, and they could be further confirmed by EDS analysis as shown in Fig. 3-3. However, specific condition will result in the formation of NiSi2, and its diffraction pattern right next to Si spot is specified in M3, the phase could also be confirmed by EDS as shown in Fig. 3-3. Fig. 3-4 will help to elaborate this phenomenon. The amorphous NiSi phase of M1 is transformed to NiSi2 after second step MWA with formation of NiSi2 pyramids. M4 is in NiSi phase after the first step MWA in setup 2 at higher temperature than M1 (as Fig. 2-3 indicated), and M6 is also in NiSi phase after 2nd MWA with low sheet resistance. The second step MWA drives Ni to redistribute in the silicide instead of penetrating into the underlying Si because the NiSi phase is formed in the first step. The thermal budget of the second step MWA is sufficient to transform the original silicide into NiSi films without increasing the silicide thickness.
3.3 The blanket Ni silicide film formed by RTA and MWA
In this section, a series of splits comparing first step RTA and second step (RTA/MWA) annealing are summarized in Table 3-3. C1 is the Ni silicide formed by only one-step RTA in 180oC for 15sec. The first step of C2 and C3 are as the same as
27
C1, but different in the second step annealing conditions. The second step annwaling of C2 and C3 are in RTA 600oC for 15sec and MWA 2000W in setup 2 for 300sec, respectively. C4 is the Ni silicide formed by only one-step RTA in 260oC for 15sec.
The first step of C5 and C6 are as the same as C4, but different in the second step annealing conditions. The second step annwaling of C5 and C6 are in RTA 450oC for 15sec and MWA 1300W in setup 2 for 300sec, respectively. The plot of the sheet resistance versus temperature for each condition is shown in Fig. 3-5. Sheet resistance compared to C4 (78 ohm/sq.), C1 (400 ohm/sq.) is much higher due to the lower temperature in first-step of RTA. For sheet resistance in C2 (83 ohm/sq.) is alomost identical to C3 (79.3 ohm/sq.), which suggests the same sheet resistance for 600oC RTA to achieve simply requires 420oC for MWA. For sheet resistance in C5 (8 ohm/sq.) is a little higher than C6 (6.37 ohm/sq.) even annealing temperature for C6 is lower than C5, which reveals that ultalow sheet resistance for MWA to achieve merely requies 360oC. These results indicates that MWA meet the current requirements and tendency of low-temperatur e in downscaled device fabrication.
The plot of the sheet resistance versus silicide thickness for each condition is shown in Fig. 3-6. Higher sheet resistance usually results in thin silicide thickness, consistent as the data in seciton 3.2. Silicide thickness compared to C4 (18.09nm), C1 thus reached extremely thin silicide thickness (5.42nm) due to its lower first annealing temperature in RTA. Besides lower second step annealing temperature, C3(6.84 nm) is even thinner than C2 (7.03 nm) for their silicide thickness. For C5 and C6 with the same first annealing condition to C4, silicide thickness for C6 is thinner than C5 with their sheet resistance both declined from 78 (ohm/sq) to less than 10 (ohm/sq.). However as shown in Table 3-4, specific anneal condiiton such as C2 and C3 will result in the formation NiSi2 which will lead to great leakage in ultra shallow
28 junction (USJ).
Therefore here we concentrate the concern into phase issue, again. For silicide thickness is less than 10nm in the case of C1-C3, SADP and EDS are applied to analysis the phase as shown in Table 3-5. The composition for Ni:Si is almost 2:1 for C2 and C3 which indicates that the phase of Ni silicide is NiSi2, and is consistent to the SADP pattern. For silicide thickness is not less than 10 nm, XRD is able to examine the phase for each condition. As shown in Fig. 3-7, the phase of C4, C5 and C6 are Ni2Si, NiSi and NiSi including corresponded orientation. Besides, the phase of M6 is also able to be detected by XRD due to its complete transformation into NiSi as shown in Fig. 3-7.