Considerations regarding theappropriate timing for advancedtechnology facility vibration surveysHal Amick, Ph.D., P.E., Michael Gendreau, INCE.Bd.Cert, Colin Gordon & Associates, San Bruno, CA, USA

Purposes of facilityvibration surveysIt is quite common to carry out avibration survey of a site prior toconstruction of a facility. This may bejust the first of several vibrationevaluations carried out over the life of afab. There are a variety of reasons forthese subsequent surveys, some ofwhich are discussed below.

Site Survey – The purpose of the pre-construction site survey is to aid in thesite selection or acceptance effort.Thereis a limit to the amount of attenuationof poor site vibration conditions thatcan be achieved at or inside a building.Thus, depending on the circumstances,it may be necessary to verify thatvibration due to external sources (rail

and automobile traffic, industr ialfacilities, etc.) will be adequately lowfor the proposed function of the siteand future buildings and processes.

Fab Construction Process – At leasttwo occasions during the constructionphase provide opportunities to evaluatedistinct characteristics of the facilitywith a minimum of interference fromextraneous conditions. Surveys done atthese two times would serve quitedifferent purposes. Those surveys arethe primary focus of this paper. Theearlier of the two is what we often calla “Structural Evaluation;” the later onewe commonly call a “Final Evaluation”or “As Built Evaluation.” Both may besomewhat extensive in scope.

Tool Site Evaluation – This type of

evaluation is generally rather limited inscope. Its intent is to evaluate theplanned locations for one or moreparticular tools to document that thoselocations meet the tools’ specificinstallation requirements.

Diagnostics – As a fab matures,problems may occur with ageingmechanical systems or specific tools,which may suggest that vibrations areexceeding a particular tool’s tolerances.The vibration measurements are moreof a diagnostic aid for the vibrationconsultant as a source of excessivevibration is sought.

Documentation of “Current” State –Some facility owners wish to updatetheir records periodically with regard to the environmental status of aproduction area as a facility ages. Thismay include vibration and/or acousticssurveys as well as particulate surveys.

The timing of the last three surveytypes is somewhat obvious.Circumstances or owner policy willdictate when they must be carried out.The timing of the surveys carried outduring construction is less obvious, andpoor timing can lead to misleadingresults, or complications withremediation of any inadequacies theyidentify.

The intent of a structural evaluationis to validate specific structuralparameters associated with thestructural dynamics design, such as thestiffness and resonance frequencies offloors, performance of structuralisolation breaks (SIBs), and structuralattenuation or amplification betweenpoints in the fab (such as between thesubfab and fab). If inadequacies2 arefound, they are much more easilycorrected during construction thanafter the cleanroom and mechanicalsystems are in place and operating.

The intent of a final evaluation is todocument the performance of thevibration-critical portions of a fab atthe time that the facility is “delivered”

FT25 – 01/ 4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION 1

CLEANROOM

ABSTRACT

Vibration surveys are carried out for a variety of reasons over the lifetime of afacility for semiconductor fabrication (fab) or other advanced technology process.1

Evaluations may take place as part of the site selection effort, during construction,or during the productive life of the facility. Generally, there should be a reason for avibration survey-a question or set of questions that need to be answered. Thatreason will likely suggest an appropriate time at which a survey should be carriedout. However, it is not unusual for a survey to be scheduled at an inappropriatetime, which may lead to useless data, or data that lead to an incorrect conclusion.

The question or questions providing the reason for the survey may not even beanswered. For example, suppose the object of the survey is to assess whether thevibration design goals of a facility have been met. In essence, the survey is intendedto predict the quality of the vibration environment in a fab after it becomesoperational. When operational, the vibration environment is some combination ofambient site vibrations, vibrations generated by the facility’s mechanical systems(fans and air handling systems, pumps and liquid circulation systems, etc.), andthose generated by personnel activities. Will the survey objective be met by a sitesurvey carried out during the construction phase, in the middle of the night (toavoid construction activities), prior to turning on the mechanical systems? Mostlikely not. Ambient site vibrations will likely be at their lowest during the night, andthere will be no contribution from the ubiquitous mechanical systems.

The intent of this article is to identify some of the reasons for vibration surveysand discuss the appropriate timing for those surveys. This will necessarily involvesome discussion of the IEST/ISO operational states of a fab as well as some aspectof measurement protocols. The issue of facility “maturation” will be touched upon,but a detailed presentation of that phenomenon is beyond our present scope.Measured data from multiple surveys taken over periods of many months (orseveral years) will be used to illustrate the discussions.

1.This paper specifically addresses vibration surveys, but most of the issues we raise apply equally well to other environmental “contaminants,” including EMI, acoustic noise,particulates, and airborne molecular contaminants.

2.These inadequacies could include design flaws, construction defects, misinterpretation of design, etc.

to the owner. Inadequacies may befound at this time as well, but thoseinadequacies are (hopefully) limited toeasily correctible items such asimproperly aligned vibration isolatorsfor mechanical systems.

Vibration criteriaVibration criter ia fall into twocategories-tool-specific and generic-and their usage depends upon thenature of the evaluation. In general, afacility will include more than one tool,so we try to avoid designing a facility tomeet the requirements of a specifictool. On the other hand, if a specifictool site is being evaluated-whether forinstallation validation or diagnosticpurposes-the manufacturer’s criteria aremost appropriate.

Designing a facility for a specific toolmight severely limit the future flexibilityof that facility3. In addition, differentmanufacturers state their criteria indifferent ways, which may lead toinconsistencies if they are used together.For example, one manufacturer maystate their criterion in peak-to-peakdisplacement, while another might givetheirs in rms velocity, and yet another inpeak acceleration.

It was for these reasons that genericvibration criteria were developed.Twopopular families of criteria are based onthe group developed two decades ago byUngar and Gordon (1983) while theywere employed at Bolt Beranek andNewman (BBN). Slight modificationshave occurred over the years, and theyare now promulgated by IEST (1993,2005), where they are denoted by theacronym “VC” (VC-A through VC-G),and ASHRAE (1999), where they aredenoted by a “Class” designation (ClassesA through F)4.The VC curves are shownin Figure 15.The issue of generic criteria(along with the related measurementmethodologies) is explored at greaterlength in Amick (1997).

The generic vibration criter ia arebased in part on the environmentalvibration specifications provided by toolmanufacturers. The latter environmentsare expected by the manufacturer to bepresent at the time the tool is installed.It is appropriate to apply them to tool-specific measurements, which aregenerally made at any time in anoperational fab. However, if the intent is

to demonstrate that a fab as a wholeconforms to its design goals, the surveyand the cr iter ia should represent asmuch as possible the total environmentencompassed by the design (and thusthe responsibility of the design team).Ideally, the survey should be made withall of the mechanical systems providedby the design and construction teams inan operational state, before systems areinstalled that are not the responsibilityof the design and construction teams.The appropriate criterion is that usedfor the design of the facility ispresumably a generic criterion.

Measurement statisticsMeasured vibration data are commonlyacquired as analog time history signalsproduced by acceleration or velocity

transducers such as accelerometers orseismometers, respectively. From a dataanalysis viewpoint, Amick (1997)divided time history signals into twobroad categories, after Bendat andPiersol (1986), each with twosubcategories, as follows:

1. Deterministic data signals: (a) steady-state signals; (b) transient signals.

2. Random data signals: (a) stationarysignals; (b) nonstationary signals.

Ambient site vibrations are usually amix of steady-state (sinusoidal, generallyfrom mechanical equipment) andstationary (random, from a variety ofsources). Thus, a site or buildingvibration survey involves measurementprotocols appropriate for theseenvironments, generally involving

W W W.FA B T E C H.O R G

FT25 – 01/ 4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION2

CLEANROOM

Figure 1. Generic vibration criteria, from IEST(2005).

3. Suppose a facility was designed for a particular stepper with moderate vibration requirements.A few years into operation, the owner may see the need to convert to a morecurrent model or to a different technology, such as a scanner.The manufacturer’s criterion for the later tool might be more stringent than those of the earlier one. Genericcriteria-though conservative-are intended to avoid this problem.

4.The reader is cautioned that the letter designations of ASHRAE (2003) are not the same as those used in IEST (1993, 2005). For example, the ASHRAE Class B criterioncorresponds to IEST VC-D (both are 6.3 mm/s).

5.The VC curves are being updated in the version to be issued in the 2005 edition of IEST RP-12.The newer version is shown in Figure 1.

energy-average signal processing.However, there are often transientvibrations, such as those associated withvehicle pass-by, and these may requirethe use of maximum-hold processing orcentile analysis [Amick and Bui (1991)].

Temporal Statistics – It is customaryto character ize the vibrationenvironment at some location by meansof the energy average (linear average)spectrum, or a centile spectrumapproximating the linear average, suchas A10

6. If one is documenting atransient event, such as a truck or trainpass-by, it may be more appropriate touse a spectrum form that captures orapproximates the upper bound of thespectrum occurring during the event,such as the maximum-hold spectrum(often denoted Amax). In some caseswith centile processing, the A1 spectrumwill be used, so as to eliminate theeffect of extremely short-durationtransients that may not be trulyrepresentative of the event.

If the vibrations at a measurementlocation are stationary (or a reasonableapproximation of stationary), then arelatively short sample time (e.g., 1 to 2min) is acceptable to capture the linearaverage, but a longer sample time (e.g.,10 to 30 min) may be desirable ifcentile processing is being used. If Amaxis being obtained using eithermaximum-hold or centile processing,it is important to note that Amaxincreases as longer sample times areused, even if the environment isstationary.

Spatial Statistics – If one ischaracterizing a single location, such asthat of a proposed tool, then it is logicalto use a single measurement location,though some will measure severalpoints around the footprint of the tool.However, if one is attempting tocharacter ize a large area, such as anentire cleanroom, a single location isprobably not representative of thewhole. One can either present all of themeasured spectra, or use some format inwhich all the spectra are condensedinto a single spectrum or set of spectra.

A practice has evolved in which theambient vibrations of a fab or otherlarge area are character ized by aspectrum representing the mean plusone standard deviation spectrum,Amean+sig, of a collection of spectra

obtained at a statistically significantnumber of locations randomlydistributed throughout the cleanroomor other area of interest.The Amean andAsig spectra are defined for a collectionof spectra Xi as:

log(Amean) = Average(log(Xi)) (1)

log(Asig) = StdDev(log(Xi)) (2)

The statistics are calculated for eachfrequency. The two spectra arecombined in log space to obtainAmean+sig, defined as: 7

log(Amean+sig) = log(Amean) + log(Asig) (3)

When presenting data in a report,we will typically summarize the data in a plot that shows four spectra:Amin, Amean, Amean+sig, and Amax. TheAmean+sig spectrum is usually shown in abolder line. The reader can then easilysee the range of the data, the statisticallymeaningful representations, and thespectrum used to characterize the spaceas a whole, Amean+sig. If one wants tocompare the performance of the entirespace at two different times, one mayshow the Amean+sig spectra from eachset of measurements.

Operational states of a fabVibrations may come from a number ofsources. Some may be from distantactivities such as vehicle or rail traffic,or even from minor seismic activities.No matter how remote, we will findthese vibrations, and they may be called“cultural” vibrations.As a site transitionsfrom remote to developed, the level ofvibrations generally increases. Thepresence of a building will alter thevibrations in a number of ways,including the contributions from itsmechanical systems. It is important tocarefully consider the point in time atwhich a survey is made, and the natureof the associated cultural vibrations.

Several key milestones in the life of afacility may be defined. It may beappropriate to carry out measurementprograms at each one for one reason oranother, but there might be importantreasons why specific benchmarkmeasurements should be done atparticular milestones. These milestonesmay be defined in terms of pre-

construction, during construction, andpost-construction. We find industrystandardization of some of themilestones, but not of others. Theprocess of commissioning orcertification forms a boundary of sortsbetween construction and post-construction.We will consider it part ofthe former, as one of its purposes is todocument the performance of thedeliverable-the completed facility.

Pre-Construction Milestones – Wemake a distinction between two sitestates. We use the term “greenfield” todenote a site at which there are noexisting facilities, and “developed” todenote the site of a future building on alarge piece of property with one ormore existing facilities (e.g., a“campus”). The distinction is betweenwhether there are other operatingfacilities at the time of the survey.Thus,when considering a campus (orindustrial park) that will have multiplefabs, the very first set of measurementscould be considered “greenfield” andmeasurements subsequent to startup ofthe first fab would be considered“developed.” 8

Construction Milestones – The mostimportant vibration evaluation duringconstruction is the one that takes placeas the completed facility is “delivered”to the owner. We call it the “final”evaluation. It demonstrates that thefacility performs as intended, at thetime the owner takes possession.

The “final” evaluation is intended todocument the “as built” state, asconventionally accepted. ISO Standard14644 defines this state as “allconstruction is complete, all servicesconnected and functioning but with noproduction equipment, materials orpersonnel present.” In this state, toolshave not yet been installed, butequipment associated with the facilityitself (i.e., fans, pumps, chillers, etc.) arerunning normally and there is flow inthe ducting and piping. It should beacknowledged that this is an idealizedcondition, and very rarely occurs, sincemost owners start installing productionequipment before the facility is fullycertified.Also, it is not possible to run allelectrical and mechanical equipment atnormal operating conditions at this time,as the tools are not in place to provide a“load”. However, a reasonably good

FT25 – 01/ 4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION 3

CLEANROOM

6.When using centile spectra, the designation Ln denotes the spectrum that is exceeded n per cent of the time.This is discussed at greater length in Amick and Bui (1991).7. Carrying out statistical operations on levels expressed in decibels produces the same results as operations in log space.The latter is simply a more general format for definition.8.The presence of commercial, industrial, or transportation facilities on adjacent property may contribute to the site vibration environment to some degree, but they are not

considered as part of the distinction between “greenfield” and “developed”. It is important to note, however, that future significant changes in this “external” environment (i.e.,the addition of a rail system) can have a significant impact on the future use of a site, and so these are also considered, to the degree possible, in a pre-construction site survey.

approximation of this state can usually be(and indeed must be) achieved. 9

Post-Construction Measurements –There are several reasons to carry out avibration survey at some time after the“as built” state not longer exists. ISOStandard 14644 defines two states inwhich measurements of various sortsare commonly carried out in a fullyfunctional facility: “at rest” and“operational.” The “at rest” state isdefined as one in which tool“installation is complete withequipment installed and operating in amanner agreed upon by the customer

and supplier, but with no personnelpresent.” The “operational” state is onein which the “installation is functioningin the specified manner, with thespecified number of personnel presentand working in the manner agreedupon.” These two states are also quiteidealized, and the only significantdifferences between the two are thepresence of personnel and the state ofproductivity.There may be a significantdifference in the particle count oracoustic noise in these two states, butthe only significant manner in whichvibrations are affected are with regard

to walker-generated vibrations andmaterial handling (such as automaticvehicles or other rolling loads).

It is not uncommon to find asignificant difference in the vibrationenvironments in the “as built” and the“at rest” or “operational” states.We haveobserved a variation that can occur overtime as a facility ages. New tools areinstalled, or old ones modified, andoften this will add to the vibrationenvironment. We call this effect“maturation” and use the term “mature”to describe an older facility in whichthe vibration environment has evolved.

W W W.FA B T E C H.O R G

FT25 – 01/ 4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION4

CLEANROOM

Figure 2. Narrowband vibration Lmean+sig spectra measured in a cleanroom at “as built” and “operational” states, 20 months apart,illustrating effect of maturation. (22 measurement locations, vertical direction, 0.375 Hz bandwidth, Hanning window) [From Gendreau andAmick (2004)].

Figure 3. Ratio of increase in vibrations from “as built” to “operational” state using data shown in Figure 2. Shaded zone shows variation of±6 decibels (halving to doubling) [from Gendreau and Amick (2004)].

9.An important distinction must be made with regard to EMI surveys. EMI is highly dependent on load (i.e., electrical current), and a proper as-built EMI survey willrequire the use of artificial loads.

Fab maturation processA detailed discussion of the fabmaturation process is beyond the scopeof this paper, but a condenseddiscussion is appropriate. A lengthydiscussion was presented by Gendreauand Amick (2004).Several factorscontr ibute to the vibrationenvironment in a fab or othercompleted technology building:

• Greenfield Site ambient vibrations– These are due to such varied sourcesas distant traffic or other transportationsources, small microseismic events, etc.,and exist at the site with or withoutthe presence of the building [Gordon(1987)]. These are sometimes called“cultural” vibrations.

• Developed Site ambient vibrations– These are due to the added presenceof other buildings on the site(including their mechanical systems),as well as the added cultural vibrationsassociated with on-site traffic, etc.

• “Building Effect” – This results froma soil-structure interaction, whereby

the presence of the building’s massand stiffness suppresses some vibrationcomponents and may amplify others.A detailed discussion of this effect isbeyond the scope of this paper, but isaddressed by Amick, et al. (2004).

• Building mechanical systems –Vibrations are generated by themechanical systems that are requiredin any technical building, includingpumps, air handling systems, air flow,fluid flow in piping, etc. Most of thesefactors are controlled to some extentusing vibration isolation methods, butthe effects are only reduced, noteliminated. In addition, the effectsmay be reduced only temporarily ifthe isolation methods and devices arenot properly maintained.

• Owner-installed mechanical systems– Many items of mechanicalequipment, particularly vacuumpumps, must be installed withproduction or research equipment.Typically these are installed after thebuilding has had its “as built”evaluation, and are the responsibility of

the owner/operator, rather than thedesign/construction team.

• Building occupants and personnelactivities – Personnel must walkabout the building. Material must bemoved. Goods must be delivered. Allof these activities generate vibrations.The effects of these sources may beaddressed to some extent, but they areonly reduced, not eliminated.

It should be clear that when oneconsiders all these factors and how theymay change over time, a facilityvibration environment cannot be staticover any extended period of time,especially as a facility progresses fromgreenfield site to operational facility. Infact, developed site ambient vibrationsand those from owner-installedmechanical generally increase over timeas an area is built up and new equipmentis installed. This is the primary part ofthe process we call maturation.

Maturation of one facility isillustrated in Figure 2, which shows twoconstant-bandwidth spectra, each ofwhich is the Amean+sig resulting from measurements at 22 locations in a representative semiconductorproduction cleanroom. The same 22locations were used in both surveys, butthe surveys were significantly separatedin time. The first was a formal10 “asbuilt” survey, and the second was an “atrest” survey taken 20 months later.

The sharp peaks in Figure 2correspond to vibrations generated byrotating mechanical equipmentoperating at single frequencies. Broaderpeaks-though still sharp-are formedwhen a number of single-frequencypeaks are at nearby frequencies.We seeconsiderable increases at frequenciesbetween 55 and 60 Hz, and between 28and 30 Hz. Other minor increases occurat other discrete frequencies, as well.

The increases are shown more clearlyin Figure 3, which presents the changein terms of the ratio of the “at rest”spectrum divided by the “as built”spectrum.The area between the dashedlines represents changes of ±6 decibels(from halving to doubling). Thoughsome of the increase may includeperformance degradation of isolators onbuilding mechanical systems (which canoccur over time), we have found thatthe vast majority of maturation-relatedincreases are due to new equipmentinstalled as part of production systems(such as dry pumps).

FT25 – 01/4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION 5

CLEANROOM

Figure 4. Data taken at four times prior to startup of a typical fab on a campus.

10.We are calling this a formal “as built” survey because the owner elected to start installing production equipment only after the “as built” documentation surveys werecompleted.Thus, we could be assured that there was no contribution from owner-supplied equipment.This formality is rarely followed now.

Measurements from arepresentative facilityWe are in the process of gatheringvibration data measured at several fabsover a long period of time.The data forone such facility are shown in Figure 4.This facility is one of several on acampus, and was just about to go intoproduction at the time of the mostrecent survey. In addition to the designcriterion VC-D, the figures show datafrom four studies:

• The greenfield site survey, prior toany of the fabs on the campus beingconstructed. There was littledevelopment in the area nearby.

• The site survey for the fab underdiscussion.The survey was carried outprior to starting construction, butafter several other nearby fabs on thecampus had been in operation forsome time. This was a “developedsite” survey. Measurements were madeat several locations within thefootprint of the fab, and the spectrumshown is Amean+sig.

• Ambient vibrations measured duringthe structural evaluation. Normally,the intent of this evaluation is themeasurement of stiffnesses, resonancefrequencies, and transmissionproperties of the facility, and all ofthese measurements benefit frombeing carr ied out at the quietestpossible time. The structure itself wascompleted and some of the buildingshell was in place, but none of themechanical equipment was operating.The measurements were made in themiddle of the night, without anyother construction activities.

• Ambient vibrations measured on thestructural floor of the cleanroom at a number of randomly selectedlocations distr ibuted throughout the cleanroom, represented as an Amean+sig spectrum. Thesemeasurements were made inapproximately the “as-built” state.Thecleanroom and the mechanicalinfrastructure were fully operationaland ready for certification.A few toolshad been installed but were notoperational, and there was minimalpersonnel activity. However, this is the“quietest” state achievable whilemaintaining clean class.

Table 1 summarizes the maxima fromthe four spectra. The predominantvibrations prior to construction are inthe 12.5 Hz band, the predominant sitefrequency. After construction, thepredominant vibrations are in the 50 Hzband, which contains the floor’s verticalfundamental resonance frequency.

These data are shown in sequentialorder in Figure 5. Several things areimportant to note:

1.The presence of other fabs on thecampus introduces a significantamount of vibration energy to the site.

There is an increase by a factor of 5.5 times (15 dB) associated withdevelopment at the campus. All of thefabs have had about the same level ofeffort given to vibration control.

2.The fab building itself tends tosuppress some of the vibration energywhen there are no vibration sourceswithin the building.We call this the“building effect.” In this case, there wasa reduction by a factor or 0.4 (8 dB)11.

3.The increase from site ambient priorto construction to the as-built state isby a factor of 2.9, about 3 dB.There isa much greater difference fromstructural to as-built, 7.6 times (18 dB),due to operation of the well-isolatedbuilding mechanical equipment.

It is clear that data taken prior to the introduction and operation ofmechanical equipment is notrepresentative of the environmentpresented to fab tools when they areinstalled.

RecommendationsWhen considering the timing of avibration survey, one should start byasking one important question:“Why doI want this survey?” The answer may

W W W.FA B T E C H.O R G

FT25 – 01/4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION6

CLEANROOM

Figure 5. Comparison of maxima.

11. In some situations, there is a diurnal variation in vibrations. In many others, particularly at locations with around-the-clock production and constant load on mechanicalsystems, there may not be.The engineer must exercise some judgment as to whether this is relevant in a given study.

State Amplitude, µm/s Frequency, Hz

Greenfield Site 0.47 12.5

Developed Site Ambient Prior to Construction 2.6 12.5

Ambient at Structural Evaluation 1.0 50

“As-Built” State 7.6 50

TABLE 1: SUMMARY OF VIBRATION MAXIMA (IN ONE-THIRD OCTAVE BANDS)FROM FOUR SURVEYS AT A FAB AND ON THE SITE PRIOR TO CONSTRUCTION

dictate when you have your surveycarr ied out. Following are somepossible answers to the question, andthe measurement times suggested foreach.We have carried out surveys basedupon all of these questions.

Do I want to buy this piece of property?

“Greenfield” conditions; during times atwhich the site is most energetic (rushhour traffic, fastest traffic, rail passages,etc.). If your facility will be used forresearch, your scientists may want toknow how it performs in the middle ofthe night. This is relevant only if theycan work during off hours.

Is this location on my campusacceptable “as is“, or do I need to takecorrective action, perhaps with respectto a neighboring building or trafficlayout?

“Developed site“ conditions; duringconditions when the most mechanicalsystems are running in existing facilitiesand environmental vibrations are at amaximum (rush hour traffic, fastesttraffic, rail passages, etc.).

Does the structure meet its engineering designparameters, such as stiffness, resonancefrequencies, propagation properties, etc.?

During construction phase, after mainload-bearing structural elements are inplace (foundation, columns, shear walls,floors, etc.) Formwork must have beenremoved, and concrete cured to designstrength (at least 28 days).Measurements must be made withoutimpact from construction, and are oftendone at night. Ambient vibration maybe documented to validatemeasurement quality, but thesemeasurements do not predict the finalvibration environment to which thetools will be subject.

What is the effect of this particular buildingfoundation and structure at this site andwith this construction type?

This is rarely needed, but if it is, it canbe done any time after the foundationand a significant portion of thestructure are complete. The test shouldbe scheduled at a time that inter ioractivities are at a minimum, andexterior activities may be controlled bythe evaluation team.

Does the facility meet its performance goals?Is it ready to deliver to the owner?

Ideally, the “as built” state, meaningmechanical systems are operational butproduction or research equipment isnot. Survey should be carr ied outduring normal working hours, ifpossible. Avoid taking data during

disruptive construction or toolinstallation activities.

A survey carried out prior to startupof the facility’s mechanical systems willnot provide the documentationrequired by this question.

What is the current vibration environment ofmy facility?

Presumably, you want to document thenormal working environment of yourfacility. Carry out the survey undereither at rest or operational conditions,but document which one. The surveyshould be (1) at a time representativeof normal working conditions, or (2)at the most energetic time of day (for“worst case” condition). A spatialstatistical evaluation of multiple pointsmight be the most useful, especially if comparing with similar data fromthe past.

What is the current vibration environment ata tool location?

Carry out the measurement undereither at rest or operational conditions,but document which one. Themeasurement protocol should beconsistent with the tool manufacturer’srequirements.

What is the quietest my facility can be?

This isn’t usually done for fabs, but isoften of interest for research facilities,where scientists may wish to work atthe “best” time. This survey mayrequire several measurements todocument the diurnal var iation. Acontinuous 24-h survey with centileprocessing may be useful. This isdiscussed in Amick and Bui (1993).

A survey carried out in the middleof the night during the constructionphase will likely not give a correctanswer, as it will not include the effectsof mechanical equipment. If themechanical systems will be operatingat the quietest part of a normal day,they must be operating dur ing thevibration survey. One may also wish toknow the contr ibution due toequipment operable by the user, incase temporarily powering these offmay provide an improvement in thevibration environment.

ConclusionA vibration survey (or anyenvironmental survey, for that matter)can be costly, and it should be moneywell spent. Most surveys have anobjective-a question to be answered.The scheduling of the survey should be

driven by what one wants to know. Ifthe objective is to assess a site’ssuitability, and it is to be used duringthe day, then evaluating it in the middleof the night will not characterize howit will perform during the day. If thefacility is to be used at night, that isanother matter. If it is to be used bothday and night, it might be important tohave data for both time periods.

Likewise, if you are characterizingyour fab, schedule the survey so that itsresults characterize the operational stateyou really want characterized. An as-built survey cannot be carried out priorto the mechanical equipment beinginstalled and balanced.An as-built surveywill not necessarily represent how afacility will perform after a year of use(and modification). A survey when thefacility is in full production probably willnot document whether the design/buildteam met its contractual obligation todeliver to you a facility with particularvibration criteria.

All this sounds terribly obvious, butseveral times a year we are asked toevaluate a facility at the “wrong” time.In one typical case we were asked tomeasure vibrations for a 24-h periodduring construction (after the structurewas completed) and predict thevibrations in the cleanroom. Since thesite was “contaminated” by constructionduring the day and very quiet at night,it was unreasonable to use any part ofthis survey to character ize the timewhen the cleanroom’s recirculation airhandlers, ducting, and chilled waterpiping would in the future beoperational.The money would not havebeen well spent due to potentiallymisleading conclusions.

The particularly troubling casementioned above came to us as adiagnostic problem: “Why isn’t our brandnew fab meeting tool specs?”The consultantused during design had carried out amidnight survey during the middle ofthe construction phase in order tocharacterize how the fab would performduring operation.The owner was lulledinto a sense of security by the fact that atmidnight at this semi-rural site, thevibrations were an order of magnitudebelow the facility requirements. Nobodyconsidered the effects of the mechanicalsystems. It turned out the design andconstruction both had some flaws, mostof which could have been identified andmitigated if the survey had been delayeduntil the “as-built” stage and owneracceptance based upon the fab’sperformance at that time. A significant

FT25 - 01/4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION 7

CLEANROOM

amount of production was delayed bythe need to correct the problems so thetool requirements could be met.

These minor catastrophes needn’thappen to you. Use common sense toschedule surveys when they willprovide the information you need.Also,be sure to communicate to yourvibration consultant the reason why thesurvey is being done-it may influencethe measurement protocol.

Amick, H. (1997). “On GenericVibration Criter ia for AdvancedTechnology Facilities: with a Tutorialon Vibration Data Representation,” J.Inst. Env. Sci., September/October1997, v. XL, no. 5, pp. 35-44

Amick, H., and Bui, S. K. (1991). “AReview of Several Methods forProcessing Vibration Data,” Proceedingsof SPIE Conference on VibrationControl and Metrology 1619, pp. 253-264, San Jose, CA (November 1991)

Amick, H., Gendreau, M., and Bayat,A.(1999). “Dynamic Characteristics ofStructures Extracted From In-situTesting,” Proceedings of SPIEConference on Opto-mechanicalEngineering and Vibration Control, July20, 1999, Denver, Colorado

Amick, H., Xu, T. and Gendreau, M.(2004). “The Role of Buildings andSlabs-on-Grade in the Suppression ofLow-Amplitude Ambient GroundVibrations,” Proc. 11th Intl. Conf. onSoil Dyn. & Earthquake Eng. (11thICSDEE) & the 3rd Intl. Conf. onEarthquake Geotech. Eng. (3rd ICEGE),7-9 January, 2004, Berkeley, CA.

ASHRAE (2003). ASHRAEHandbook: Applications, Chapter 46,“Sound and Vibration Control,” 47 pp.

Bendat, J. S., and Piersol, A. G. (1986).Random Data: Analysis andMeasurement Procedures, 2nd ed., JohnWiley & Sons, Inc. (1986).

Gendreau, M. and H. Amick,“‘Maturation’ of the Vibration andNoise Environments in SemiconductorProduction Facilities,” Proc. ESTECH2004, 50th Annual Technical Meeting,Institute of Environmental Sciences andTechnology (IEST), Las Vegas, Nevada,April 28, 2004

Gordon, C. G. (1987). “A Study ofLow-Frequency Ground Vibration inWidely Differing Geographic Areas,”Proceedings of Noise-Con 87, StateCollege, Pennsylvania (June 1987), pp.233-238.

Gordon, C. G. (1991). “Generic criteriafor vibration-sensitive equipment,”

Proceedings of SPIE Conference onVibration Control and Metrology, pp.71-85, San Jose, CA (November 1991).

Institute of Environmental Sciences(1993), “Considerations in CleanRoom Design,” IES-RP-CC012.1

Institute of Environmental Sciences(2005), “Considerations in CleanRoom Design,” IES-RP-CC012.2

ISO 14644-4:2001, “Cleanrooms andassociated controlled environments-Part 4: Design, construction and start-up,” 51 pp.

Ungar, E. E., and Gordon, C. G. (1983).“Vibration Challenges in Micro-electronics Manufacturing,” Shock andVibration Bulletin, 53(I):51-58 (May1983).

W W W.FA B T E C H.O R G

FT25 - 01/4

SEMICONDUCTOR FABTECH – 25 TH EDIT ION8

CLEANROOM

Hal Amick is Vice President for Technology at

Colin Gordon & Associates. His education was in

civil engineering and structural dynamics, and he

has been involved since 1985 with design and

assessment of vibration-sensitive facilities for

technology. He is a registered civil engineer and

leads CGA’s consulting activities pertaining to

nanotechnology. A list of his publications may be

found at www.colingordon.com.

Michael Gendreau is President of Colin Gordon &

Associates. He holds a degree in physics, and is

Board Certified by the Institute of Noise Control

Engineering. He has been involved with the full

range of vibration and noise control consulting

services for fabs and research facilities. He is CGA’s

senior project manager, and is responsible for

projects around the world. A list of his publications

may be found at www.colingordon.com.

Hal Amick

Vice President, Technology

Colin Gordon & Associates

883 Sneath Lane, Suite 150

San Bruno

CA 94066

USA

Tel: +1 (650) 358-9577

E-mail: hal.amick@colingordon.com

Michael Gendreau

President

Colin Gordon & Associates

883 Sneath Lane, Suite 150

San Bruno

CA, 94066

USA

Tel: +1 (650) 358-9577

E-mail: michael.gendreau@colingordon.com

ABOUT THE AUTHORS ENQUIRIES

REFERENCES