5.1 Transfer Characteristics

The $I_{\mathrm{d}}$- $V_{\mathrm{g}}$ curves for Device Generation A and Device Generation B are shown in Fig. 5.1 and Fig. 5.2 on a semi-logarithmic scale. The curves of the PCD devices in forward and reverse modes (drain and source interchanged) and of the corresponding uniformly doped device are given for the linear ( $V_{\mathrm{d}}$ = 0.1 V) and saturated ( $V_{\mathrm{d}}$ =  $V_{\mathrm{dd}}$) cases.

The Gaussian parameters of the PCD devices are taken from Table 4.3 with peak doping 2.18$\cdot$10$^{18}$ cm$^{-3}$ and bulk background doping 3.04$\cdot$10$^{16}$ cm$^{-3}$ for Device Generation A and peak doping 5.73$\cdot$10$^{18}$ cm$^{-3}$ and bulk background doping 1.21$\cdot$10$^{17}$ cm$^{-3}$ for Device Generation B, respectively. The uniformly doped devices have bulk doping 5.47$\cdot$10$^{17}$cm$^{-3}$ for Device Generation A and 2.39$\cdot$10$^{18}$ cm$^{-3}$ for Device Generation B, respectively.

Table 5.1 lists the transfer characteristics of the PCD device in forward and reverse mode and of the uniformly doped device extracted from Fig. 5.1 and Fig. 5.2. Generally, the PCD device has a steeper subthreshold slope than the uniformly doped device. This can be considered as one reason for the higher drive current of the PCD device. The DIBL of the PCD device is higher due to the reduced length of the effective channel area (the doping peak), but stays within an acceptable range. Anyway, it has been shown that there exists a fundamental tradeoff between current drive performance and short-channel effects in deep-submicron MOS devices [59].

Figure 5.1: The transfer curves of the PCD device and the uniformly doped device in the linear (top) and saturated (bottom) regions, Device Generation A.

Figure 5.2: The transfer curves of the PCD device and the uniformly doped device in the linear (top) and saturated (bottom) regions, Device Generation B.

Table 5.1: Transfer characteristics of the PCD device.

		Generation A			Generation B
		uniform	PCD	PCD rev.	uniform	PCD	PCD rev.
$I_{\mathrm{off}}$	(pA)	1	1	188	1	1	3.2
$I_{\mathrm{on}}$	($\mu$A)	259	369	377	131	214	229
S	(mV/dec)	83	74	79	81	71	71
DIBL	(mV/V)	7	20	140	12	39	81

Due to the fact that the PCD device in forward mode and the uniformly doped device were designed to have the same off-state current of 1 pA in the saturated case ( $V_{\mathrm{d}}$ =  $V_{\mathrm{dd}}$), the drain current at $V_{\mathrm{g}}$ = 0 V differs in the linear case ( $V_{\mathrm{d}}$ = 0.1 V). Actually, this variation results from the different DIBL values.

Moreover, the 0.25 $\mu$m PCD device happens to have about the same threshold voltage (defined at $I_{\mathrm{d}}$ = 100 nA) as its uniformly doped counterpart because the shifts of the linear transfer curves due to the DIBL difference and due to the different subthreshold swings fairly compensate at this specific drain current. This is not the case for the 0.1 $\mu$m device,therefore the threshold voltage of the PCD device is slightly lower compared to the uniformly doped device.

If the PCD device is operated in reverse mode, the transfer curves for the linear case stay fairly the same, but for the saturated case the DIBL effect is drastically increased. Therefore, the off-state current becomes worse by more than two decades for the 0.25 $\mu$m device (Device Generation A). The 0.1 $\mu$m device (Device Generation B) suffers less off-state current deterioration in reverse mode (only about half a decade) due to the smaller supply voltage. This strong DIBL effect is the result of the much higher voltage at the right-hand side (intrinsic drain side) of the doping peak during reverse mode operation.

The difference in the device characteristics of the PCD device and the uniformly doped device and of the PCD device in forward and reverse modes turned out to be bigger for Device Generation A because of the higher ratio between the gate length and the doping peak length. Thus, the stronger asymmetry of the device structure causes a stronger asymmetry in the electrical behavior of the device.