Kin P. Cheung

On the mechanism of plasma enhanced dielectric deposition charging damage

Photoconduction is shown to be the mechanism for plasma charging damage during plasma enhanced dielectric deposition. Details of the conduction process, including polarity effects, are explained. The recently measured oxide photoconductivity is shown to be in agreement with expectation. The main cause of severe charging damage is the low level of photoconduction coupled with high processing temperature.

Plasma charging damage of ultra-thin gate-oxide-the measurement dilemma

We examined the problem of detecting plasma charging damage in deep submicron technology where the gate-oxide is ultra-thin. From the viewpoint of damage impacting gate-oxide reliability, we show that the current available method is incapable to provide sufficient sensitivity.

Field dependent critical trap density for thin gate oxide breakdown

We have found that the total trapped negative charge in a thin gate-oxide at the point of breakdown is a strong function of the stress field. This observation is in direct contrast with previous reports in the literature. The field dependent behavior of total trapped charge leads to the conclusion that the critical trap density for breakdown is also field dependent. We use field dependent hopping conduction to explain why the critical trap density for breakdown in the percolation model should be field dependent.

On the use of Fowler-Nordheim stress to reveal plasma-charging damage

It has been well establlished that to properly evaluate plasma-charging damage, one needs to bring out the latent defects hidden by various temperature cycles. A method of doing so is lo anneal the wafer to detrap all charges and then use Fowler-Nordheim (FN) stress to repopulate discharged traps. By measuring the post FN stress transistor threshold voltage (Vt) and transconductance (Gm) !Shifts, plasma damage is revealed by comparing the damaged devices to a control device or by detecting antenna ratio dependency. Another way to use FN stress to measwe plasma-charging damage is to monitor the voltage required for maintaining a constant current injection through the gate oxide. From the voltage curve, one measures a quantity called Initial Electron Trapping Rate (IETR) or Slope (IETS). IETR=IETS/init. voltage, both are directly proportional to the electron trap density in the oxide under test. Electron trap density in turn is an indicator of plasma damage.

Charging damage in thin gate-oxides-better or worse?

The question of whether or not thinner gate oxide is less susceptible to plasma charging damage depends on a number of factors. One important factor is the definition of damage itself. The measurement method is linked to the definition of damage. When the charging current is low and charging voltage is high, thinner oxides are indeed far less prone to damage. When the charging current is high and charging voltage is low, as in most new plasma systems, thinner oxides are more susceptible to damage. In a very crude way, one may conclude that older plasma systems tend to belong to the low current, high charging voltage class, while modern plasma equipment tends to belong to the high current, low charging voltage class. In this sense, it is the concomitant change to high density plasma processing with advanced technology where thinner gate oxides are used that make plasma charging damage continue to be a major problem.

A model of the stress time dependence of SILC

A number of groups have reported that the stress-induced leakage current (SILC) follows a power law dependence on the stress time. In this study, we observed that the power-law behaviour is only an approximation of the fast rising part of a more complex behaviour. SILC rises during the initial stress stage and saturates after a long stress time. Based on the trap-assisted tunneling (TAT) model, we show that the stress time dependence of SILC is better described as the depletion of multi-precursors of traps. Although the new model involves many fitting parameters, we show that the fitting results are consistent with the physical interpretation of these parameters. To further support the physical interpretation, we examined these parameters with annealing experiments.

Fast hot-carrier aging method of charging damage measurement

To monitor plasma-charging damage, it is well known that one must not overlook the latent defects that are created by the damage but are not readily observable by most commonly used measurement methods. These latent defects are either passivated defects due to temperature cycling after the defects are created, or simply due to charge detrapping that render the defects invisible to many measurements. The common way to reveal these latent defects is to look for accelerated degradation of device parameters under electrical stress. A common stressing method is uniform Fowler-Nordheim tunneling through the gate-oxide. A less often used method is channel hot-carrier stress (Hook et al., Proc. P2ID96, p. 164, 1996). In this paper, we have demonstrated the suitability of using very short hot-carrier stress time to monitor plasma charging damage without the use of a very high acceleration voltage. When the stress is long enough to extract the lifetime from a large number of devices, the method allows us to find the intrinsic (damage free) lifetime of a particular transistor design. This ability to determine how much the devices are degraded quantitatively without the help of a reference device is unique among plasma-charging damage measurement methods.

Relationship between plasma damage, SILC and gate-oxide reliability

The debate on whether plasma-charging damage will become less of a problem or not when the gate-oxide is 30 /spl Aring/ or thinner has been going on for some time. This is, of course, a very important question. We showed, in a previous publication (Cheung et al., 1998), that charging damage continues to be a serious problem for ultra-thin gate-oxides when a high-density plasma is used. In this paper, we show that even when low-density plasma is used, charging damage is also a serious problem for ultra-thin gate-oxides by looking at the relationship between plasma damage, stress-induced leakage current (SILC) and gate oxide reliability.

Improved CVD aluminum deposition using in-situ sputtered nucleation layers

Superior-quality chemical vapor deposited (CVD) Al films have been achieved using tri-isobutylaluminum (TIBA) on in-situ sputtered nucleation layers. The nucleation layer enhances the growth of CVD Al, resulting in smooth, high-quality films suitable for VLSI application. The use of in-situ sputtered seed layers allows the deposition of CVD Al on SiO/sub 2/ without having to expose the wafers to TiCl/sub 4/, which may leave Cl residue and cause corrosion. In addition, the CVD films deposited on a TiN seed layer demonstrate smooth morphology and high conductivity and show no presence of the pinholes or interfacial voids which rendered earlier CVD Al films unusable for VLSI. The TIBA process is done at low temperature of 250 degrees C, and TiN is an accepted barrier metal. The combined TiN/CVD Al process promises a low-cost, high-quality manufacturable process for VLSI metallization.<<ETX>>

Is surface potential measurement (SPM) a useful charging damage measurement method

Recently, there has been strong interest in using the surface potential measurement (SPM) method to monitor plasma charging damage. This method is also called contact potential measurement (CPM). The appeal of this method is obvious, in that it is fast, inexpensive and simple to use. More than one commercial system is available for this measurement. Given the importance of plasma-charging damage in advanced VLSI manufacturing, the use of this method is spreading quickly. However, the data presented in this paper shows that this method does not always work. The SPM method produces a voltage map that does not always correlate with damage. Since a highly nonuniform or high value SPM map does not imply damage, nor does a uniform and low value map imply no damage, it cannot be used as a damage monitor directly. Until it is understood how and where the residual charges are created, the relation between SPM and plasma damage cannot be established. Although we provide a possible explanation for the residual charge distribution for a specific case, it cannot be generalized to all situations.