Mobile Ionic Contamination (MIC)

Mobile Ionic Contamination

Mobile Ionic Contamination (MIC) refers to the presence of ionic contaminants in the active circuitries of semiconductor devices, the most common of which are alkali ions such as Na⁺, K⁺, and Cl^-. It is usually observed in gate oxide layers of MOS transistors. These contaminant ions are free to move about, hence the phrase 'mobile ionic contamination.' This mobility is accelerated by temperature and electric field.

The mobile ions often enter the gate oxide through the interface between the gate (usually metal or polysilicon) and the gate oxide (usually SiO₂). Some of the ions then drift to the Si-SiO₂ interface under the influence of electric fields created by voltages applied to the gate. Given the high mobility of these ions in SiO₂, they can drift under field assistance even at room temperature.

The presence of these ionic contaminants at the gate-oxide and oxide-semiconductor interfaces and in the oxide itself results in a mobile ionic charge, Qm, which can cause long-term changes in the threshold voltage, V_T, of the transistor. The V_T shift aggravates as more charges accumulate at the Si/SiO₂ interface.

According to S. Wolf and R. N. Tauber, a Qm value in the low 10¹⁰/cm²range will cause a shift of only a few tenths of a volt for a MOS device with a gate oxide of 1000 angstroms. However, a Qm value in the 10¹⁰/cm² range can result in V_T shifts of several volts. Thus, reducing Qm density should be a key ingredient of any program designed to eliminate MIC failures.

A high Qm value can also promote the formation of conducting channels that increase leakage currents. Gain reduction due to mobile ionic contamination has likewise been observed. Bipolar devices can also experience beta degradation due to the presence of mobile ionic contaminants, mainly because these can change carrier concentrations.

Figure 1. The mobile ionic contaminants present in the gate oxide can

accumulate as an ionic charge that affects the channel of a MOS transistor.

Among the common contaminants, Na⁺ exhibits the greatest mobility due to its small atomic radius. It is also usually the first mobile ionic contaminant to suspect if MIC is being dealt with because Na⁺ is widely distributed, being present in air and in human byproducts such as perspiration and saliva.

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.

HOME