2.1 Device Char acter ization Method
A four-terminal nMOSFET was used with a gate width of 100µm and a gate length of 0.25µm. The device has a gate oxide thickness about 50Å . Maximum Ib stress around Vg=0.5Vd was performed in DTMOS-like mode (Vb>0) and in the standard mode (Vb=0). Drain current in the triode region was measured to monitor drain current degradation. Temperature and drain bias dependence of the Auger enhanced degradation was also characterized. Hot carrier luminescence measurement was performed with single photon counting system that allowed for spectral analysis in the range 1.2eV-2.9eV [27].
Photons emitted from a MOSFET are detected by a photon counting camera through the optical microscope. The photon numbers at each wavelength are counted individually using band pass filters. The measured data are then corrected for the energy dependence of the filter transmittance. During the luminescence measurement, drain current was monitored to check the possible presence of aging, which was found to be negligible.
2.2 Auger Recombination Assisted Hot Electr on Ener gy Gain Pr ocess
It has been reported that Auger recombination can enhance hot electron tail and cause more serious degradation in MOSFETs [11], [28], [29]. The process for Auger recombination in various device operation conditions is illustrated in Fig.2.1. In Fig.2.1 (a), a small part of holes created by impact ionization may flow to the region near the source (where electron concentration is high) [30] and provide for Auger recombination.
In Fig.2.1 (b), a positive substrate bias is applied and the channel hole concentration and thus Auger recombination rate are increased due to substrate hole injection [29]. In ultra-thin oxide nMOSFETs, a positive gate bias can cause valence-band electron tunneling to
the gate and leave holes behind in the channel (Fig.2.1 (c)). In the following, we will first investigate the significance of positive substrate bias injected holes to device reliability.
Fig.2.2 illustrates the electron energy gain process in conventional hot carrier stress and in DTMOS hot carrier stress, respectively. In conventional hot carrier stress, the electron energy gain mechanism is field heating near the drain junction as shown in Fig.2.2 (a). In DTMOS hot carrier stress, i.e., with a positive substrate bias applied, holes are injected from the positively biased substrate to the channel as shown in Fig.2.2 (b).
The injected holes can provide for recombination with electrons in the channel and give excess energy to other channel electrons as shown in Fig.2.3 (a). This is the Auger recombination process and also a major energy gain process in DTMOS operation. The energetic electrons arising from the Auger process are then accelerated by a channel electric field, thus resulting in a larger hot electron tail than in the standard MOSFET’s operation condition as shown in Fig.2.3 (b).
2.3 Hot Electr on Light Emission Measur ement
As we know, the hot electron luminescence and the light emission in nMOSFET's can reflect the electron energy distribution [31], [32]. Fig.2.4 (a) and Fig.2.4 (b) are the micrographs of the hot electron light emission from nMOSFET's with Lg=0.25µm and Wg=100µm. Bias conditions are: Vds=2.9V, Vgs=1.5V, Vbs=0V and Vds=2.9V, Vgs=1.5V, Vbs=0.7V, respectively. The bandwidth of the band pass filter is 800Å . Square block regions are aluminum pad with 100µm by 100µm for electrical contacts. The total exposure time is 100 seconds. Note that the light intensity is relatively stronger as the positive substrate bias is applied. Fig.2.5 shows the light intensity and the substrate current as a function of the gate voltage with different substrate biases. It should be mentioned that the substrate current arising from impact ionization near the drain junction strongly depends on channel field and so does the hot electron light emission. Therefore, the hot electron light emission correlates well to the substrate current in the conventional
hot carrier stress. However, the substrate current in the DTMOS hot carrier stress condition is more complex due to an additional current path from the substrate to the channel. Therefore, the light emission provides a reliable monitor for hot electron energy in MOSFET's, which can be used as an alternative monitor to the substrate current.
The hot electron light emission spectrum is measured to analyze the hot electron distribution, as illustrated in Fig.2.6 In this figure, the y-axis represents the normalized light intensity and the x-axis represents the photon energy. The light intensity is normalized to the drain current to compensate for the different carrier flux in the channel.
As the substrate bias increases from 0V to 0.5V, the hot electron actually decreases due to a smaller electric field. As the substrate bias continues to increase to 0.8V, the hot electron tail is significantly enhanced by an order of magnitude.
2.4 Hot Electr on Gate Cur r ent
Another evidence is the hot electron gate injection current. Since hot electron gate inction current, Ig, is a sensitive measure of the high-energy tail of the hot carrier distribution [33]. Fig.2.7 demonstrates the normalized hot electron gate injection current as a function of gate bias with different substrate biases. The drain bias is fixed at 3.5V and the substrate bias is increased from 0V to 0.8V. Again, the gate injection current first decreases and then increases with the substrate bias. In Fig.2.8, we measure the hot electron gate current at different drain biases. From the above two figures, the hot electron gate current depends on both drain bias and substrate bias. The drain bias determines the field heating and the substrate bias determines the Auger effect. The dependence on both the drain bias and the substrate bias confirms that the electron energy gain process in DTMOS operation consists of Auger recombination and field acceleration.
2.5 Two-Dimensional Device Simulation Results
Figure 2.9 shows the two-dimensional device simulation results of the Auger recombination rate and electric field in the channel. In this figure, the x-axis is the distance from the source junction. The open symbol represents the result with a substrate bias of 0.5V and the full symbol represents a substrate bias of 0.8V. Note that the electric field slightly decreases as the substrate voltage increases from 0.5V to 0.8V due to body effect, whereas the Auger recombination rate increases by several orders of magnitude due to the exponential dependence of hole injection. Notably, hole injection is restricted to the low field region near the source. Therefore, the energetic electrons arising from the Auger recombination are then accelerated by a lateral electric field, thus leading to a large hot electron tail, and so do the hot electron light emission and the gate injection current.