Production Monitoring of Charge-Induced &nDamage from Strippers
Laura Peters, Senior Editor -- Semiconductor International, 9/1/1998
M icrowave strippers are known for their low damage potential. Even so, at 0.25 µm geometries, production monitoring of cumulative charge damage induced by multiple photoresist stripping processes and their impact on yield becomes important. The impact of charge-induced damage is expected to increase as gate oxides are scaled below 40-60 Å in thickness and as more aggressive, high-density plasma ashing techniques are employed to remove polymer residues.
CHARge Monitor (CHARM) wafers, capable of being reprogrammed and reused, allow process engineers to tune ashers, etchers or ion implanters so that no portion of the wafer is subjected to charging levels greater than the oxide damage threshold. "The key issue is monitoring how much current is drawn from the plasma and what effect this cumulative charge will have on a particular device," said Shams Tabrez, product manager for resist removal systems at Lam Research Corp. (Fremont, Calif.).
The effects of charge-induced damage, resulting in potentials in the 10-30 V regime, can range from a shift in critical transistor parameters, such as threshold voltage, to catastrophic breakdown of the gate. CHARM wafers monitor wafer-surface substrate potentials, ultraviolet emission and current density versus voltage characteristics of the charging source. If the voltage resulting from plasma-induced charging exceeds the voltage-causing Fowler-Nordheim tunneling through the gate oxide, and if sufficient current is drawn from the plasma, device damage is likely.
 |
| Fig. 1. After exposure to ashing, positive voltage sensor response on the CHARM map shows uniform voltages around 2.0 V, the minimum detection limit. |
The first methodology involves the use of antenna structures, similar to commonly used antenna capacitors, except the charge sensing element is not a gate oxide but rather an EEPROM transistor with its gate connected to an antenna charge collection electrode (CCE). Charge build-up on the CCE develops a voltage across the CCE-substrate capacitor, changing the threshold voltage of the EEPROM transistor. Based on threshold voltage shift, voltage build-up on the CCE during processing is determined. Antenna structures with poly/gate oxide areas of 214,000:1, 821,000:1 and 9,600,000:1 were used, and antenna test responses were measured in as gate oxide leakage after plasma exposure. Results showed 100% yield, damage-free gates measuring <1 nA leakage, even with a 1,000,000:1 antenna ratio.
The CCE lies directly above the EEPROM transistor, shielding it from the effects of UV exposure. CHARM wafers with unshielded or partially shielded EEPROMs measure UV dose. If the CCE is connected through a family of known resistor values to the substrate for current measurement, a complete I-V characteristic curve of the plasma source can be mapped using a standard parametric tester.
Figure 1 shows the response of a positive voltage sensor after exposure
in the microwave asher. Resulting voltages are uniform at about 2.0 V,
the minimum detection limit for the CHARM wafer, indicating no
measurable charging. The current sensors also show no response,
establishing an upper limit of <3.5 µA/cm2 on the
current drawn from the discharge, a low level that is consistent with
data from the antenna tests. In production, CHARM testing time is
approximately 20 min. 
