Environmental Vibration Criteria

These generic criterion curves have been developed on the basis of data from individual systems and measurements made in facilities both before and after vibration problems had been solved. Moreover, they have been used extensively by leading vibration consultants for the semiconductor manufacturing industry for almost 20 years, and have been extended and refined as the industry has moved to narrower line widths.

The curves take into account that equipment used for the most exacting tasks (such as manufacturing semiconductors with smaller device geometries) is stiffer and better-isolated. It is, however, important to note that these criteria are for guidance only. For example, like any useful and general rule of thumb, the criterion curves are reasonably conservative for some specific cases, especially equipment with well-designed built-in vibration control systems. It should also be noted that these criterion curves do not replace high-resolution narrowband spectrum analysis for diagnostic studies. If a comprehensive study is required, both methods should be used.

The criteria are specified as a set of “1/3-octave-band velocity spectra” that define guidelines for allowable vibration levels for various activities and equipment. One of the major benefits of these criterion curves is that vibration is expressed in terms of root-mean-square velocity instead of units of displacement or acceleration. Various studies have shown that while individual equipment may show unique displacement responses to different frequencies, these points often lie on a curve of constant velocity. Moreover, the threshold constant velocity that affects the performance of equipment within each class tends to be rather uniform. The International Standards Organization (ISO) also uses a velocity-based standard for human exposure to vibration, which is incorporated into the criterion curve chart.

Floor vibration in both manufacturing and research environments is typically dominated by random, “broadband” energy, as opposed to pure tone energy consisting of discrete frequencies. Although both broadband and pure tone vibrations will excite system resonances, the degree of their excitation will be different. Because random vibration closely approximates the measured floor vibrations observed in laboratories and manufacturing facilities, a specification based on random vibration more accurately reflects real world applications.

Evaluation of sites and the design goal of many new facilities confirm that “1/3-octave” vibration criteria accurately reflect typical environmental vibration spectra. This means that the bandwidth of the random vibration — the “window” of frequencies which is taken into account — is 23% of the band’s center frequency. A proportional bandwidth is used instead of fixed bandwidth based on a conservative estimate of the system’s ability to damp resonances.

Much of the current design activity for fab construction is setting goals in the VC-D to VC-E (250 µin./sec to 125 µin./sec). This range is believed to be appropriate to production of line widths of 0.3 µm. Older facilities will increasingly need supplemental vibration control systems to achieve the required level of vibration control. Next generation systems seeking line widths of 0.18 µm and below will almost certainly need active vibration isolation to complement even the most thorough fab construction.