Mobile Ionic Contamination (MIC)

Custom Search

Mobile Ionic Contamination - Page 2 of 2

In general, mobile ionic contaminants come from: 1) the environment; 2) humans; 3) processing chemicals such as etchants; 4) processing equipment such as furnaces; and 5) even from assembly materials such as lead frames and adhesives if the device's protective surface layers are inadequate or defective.

The most common sources of Na⁺ contamination during wafer fabrication, however, include: 1) gate or contact metallization; 2) oxidation and annealing furnaces and gases; 3) diffusion furnaces and gases; 4) photoresist bake; 5) incomplete resist stripping; and 6) contaminated chemicals used in wafer cleaning. It is therefore necessary to minimize the introduction of Na⁺ ions from these wafer fab sources in order to reduce the risk of failures due to mobile ionic contamination.

Mobile ionic contamination pose a serious reliability risk that needs immediate attention. Failures can occur after electrical testing, or even after affected devices have been operating in in the field for quite a while. Fortunately, lots affected by mobile ionic contamination are easy to identify.

These contaminated lots will degrade or fail after being subjected to burn-in, since the high temperature and electrical bias of the said stress test will accelerate mobile ionic charging at the Si-SiO2 interface, causing V_T shifts and high leakage currents. These burn-in-induced failures are recoverable by an unbiased bake, which tends to scatter the mobile ions and relocate most of them back to the gate-SiO2 interface. Thus, a tell-tale sign that a device is suffering from mobile ionic contamination is if it's failing after burn-in, and then becoming good again after bake.

The failure analysis (FA) process for suspected MIC-induced failures is likewise not complicated. Once a lot has been verified to exhibit failure after burn-in which recover after bake, the worst failures are taken for use as FA samples. Bench testing and curve tracing should confirm that these samples exhibit failure modes that are associated with mobile ionic charging, e.g., V_T shifts or high leakage currents.

Photoemission microscopy may also show line emissions (not point emissions) around the gate area of the affected MOS components. Affected areas may then be subjected to EDX analysis for identification not only of the mobile ions present, but possibly their source as well. A commonly encountered EDX spectrum for MIC cases will show peaks of one or more of the following elements: Na, Cl, K, P, Ca, and S. Human spittle is a potential source if this spectrum is revealed, while the same spectrum without the S peak may point to human perspiration.

Ensuring a clean wafer fab process alone is not enough to prevent mobile ionic contamination, since mobile ions from external sources after wafer fabrication can easily seep into devices. The solution to this problem is to protect the device from these external contaminants by depositing protective layers over the die surface.

For instance, a phosphosilicate glass (PSG) layer can act as a getter or Na⁺ ions, making it a practical choice for interlevel dielectric between the gate and the metal level. Silicon nitride is often used as the final surface passivating layer of the die, since this material is not only mechanically resistant, but impervious to Na⁺ as well.

A wide range of values for the activation energy of mobile ionic contamination failures have been observed, but 1 eV is typically used.

<Back to Page 1>

HOME