The Effect of Humidity On Static Electricity
The Effect of Humidity On Static Electricity
The Effect of Humidity On Static Electricity
DE-13-031
Fayu Wan is a post doc and Michael Hillstrom and Carlton Stayer are bachelor students in the Department of Electrical & Computer Engi-
neering, Missouri University of Science and Technology, Rolla, MO. David Swenson is the president of Affinity Static Control Consulting,
L.L.C Round Rock, TX. David Pommerenke is a professor in the Department of Electrical& Computer Engineering, Missouri University of
Science and Technology, Rolla, MO.
2013 ASHRAE. THIS PREPRINT MAY NOT BE DISTRIBUTED IN PAPER OR DIGITAL FORM IN WHOLE OR IN PART. IT IS FOR DISCUSSION PURPOSES ONLY
AT THE 2013 ASHRAE ANNUAL CONFERENCE. The archival version of this paper along with comments and author responses will be published in ASHRAE
Transactions, Volume 119, Part 2. ASHRAE must receive written questions or comments regarding this paper by July 15, 2013, for them to be included in Transactions.
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 1 ASHRAE temperature & humidity guidelines for data centers and telecom facilities (Hydeman 2010, Hydeman and
Swenson 2010).
2 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 2 Personnel ESD Events per shift based on Simonic’s data (Mardiguan2009)(left). Estimation based on power-law
distribution (based on Simonic’s data). A total of 10,000 events were used in order to illustrate the power-law having
all positive values (right).
DE-13-031 3
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
ronmental conditions change and different types of ESD from the skin is less severe due to the skin’s resistance) is
events are introduced, but, in general, a fairly consistent distri- approaching a surface very slowly, the spark will occur at a
bution curve is expected. given distance. This discharge can be calculated using
In controlled environments, discharge events tended to Paschen’s law (Pommerenke 1995). Now we need to remind
have a higher probability of occurring at lower voltage levels. ourselves that the spark is an electron avalanche process which
Therefore expectations for uncontrolled (such as a non- needs to be initiated by a starting electron. If the person is
conductive carpet and tennis shoes) environment ESD event approaching faster, then the initiation of the spark can be
rates were used to determine the worst case percentage for delayed due to a lack of initial electrons and the metal part will
high voltage ESD events. approach closer than the value predicted by Paschen’s law.
Considering uncontrolled environments and accounting Once the spark is initiated it will develop under much higher
only for 1 kV or greater ESD events, discharges of at least 8 kV field strengths as the voltage remains nearly the same, but the
are expected to account for around two percent of discharges distance of the spark gap has been reduced. Such a spark will
whereas discharges of at least 4 kV should account for around develop much faster (hundreds of picoseconds) and have a
seven percent of discharges. Discharges with voltages greater higher peak value.
than 12 kV are highly unlikely to occur, but this does not mean The process above indicates the importance of sources for
that they do not pose a risk to sensitive equipment. In order to initial electrons: If many initial electrons are available a spark
determine the risk posed by ESD events at different voltage will develop over a longer distance, have a slower rise time and
levels, the susceptibility of equipment at a certain voltage level a lower peak value. Of all factors that provide initial electrons,
must be considered alongside the probability of an ESD event humidity is the dominating one, as moisture in the air will form
occurring at a voltage level. water layers on the surface and electrons easily get attached,
A major consideration in the design of any environment but also detached due to the highly polar water molecules.
where electronic equipment is being installed, operating or Other factors are: Electrodes with sharp corners allow easier
will be serviced is the ability to control charge generation on electron emissions; and surface contamination which usually
personnel. The level of static charge that can develop on a will also enhance the ability to create initial electrons
human body is determined by the person’s resistance to (Pommerenke 1993).
ground, the charging rate and the dissipation rate. If the charg- Two types of experiments are conducted: In charge accu-
ing rate and dissipation rate are nearly the same, the resultant mulation experiments the voltage levels reached by people
body voltage relative to ground will be maintained near “0 operating in lower and higher humidity environments is
volts”. While conventional flooring and footwear will gener- measured as a function of flooring, footwear, grounding and
ally allow charge generation that results in body voltage above environmental conditions. These experiments provide a
“0” volts, high charging levels are unlikely in any reasonable measurement of maximum voltage levels, their distribution
office area where some consideration has been given to the functions and dependencies. The second set of experiments
flooring materials. For office and data center environments, investigates the discharge process for given voltages, as
maintaining personnel below 2,500 volts potential will humidity will also influence the spark development. Here
prevent perceptible electrostatic shocks. While this level is not potentially damaging discharge currents and electromagnetic
appropriate or acceptable for areas where unprotected ESD fields are measured for discharges of humans, but also for
susceptible parts are handled or equipment is being repaired, sparks that can occur if a charged cable is inserted into a
maintaining personnel at or below 2,500 volts will ensure that connector.
operational electronic equipment that meets ESD immunity
standards are not affected. Once this study is completed, the core data analysis will
determine:
Electronic equipment such as it is used in data centers is
tested for its ESD robustness in accordance with the IEC Did the observed voltage levels surpass the minimal guar-
61000-4-2 ESD test standard (IEC 2008). During this testing anteed robustness of the telecom equipment at the lower end
the equipment is subjected to controlled ESD discharges at set of the recommended humidity window as defined by
levels of charge voltage (minimal: 4 kV for discharges to ASHRAE? Or under the lowest humidity level that could be
metallic surfaces and 8 kV for discharges to non conducting achieved in the test chamber?
surfaces) and it is observed if damage or upsets occur. Highly Using the voltage distribution function from the test
reliable data center equipment is tested to higher levels such as data, and referenced to published distribution functions one
8 kV/15 kV and more by most manufacturers. Just using the can extrapolate the likelihood of having charge voltages of
minimal test levels a guaranteed minimal robustness can be 4/8 kV. If further the number of discharge relevant actions
assumed. At 4 kV the equipment will be subjected to about in a data center per day is estimated one can estimate the like-
17A of current having a rise time of about 850 ps. lihood of having critical ESD events. By observing the
A third basic property of ESD that needs to be considered change of the probability distribution function with humidity
is the influence of the humidity on discharge development. If one can estimate the effect of reduced humidity on the like-
a charged person holding a rounded metal part (a discharge lihood of a critical ESD event.
4 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
The second part of the analysis evaluates the severity of a STM97.2 procedure in shown in right of Figure 3. During
discharge. It is not only given by the charge voltage, but more the walking test, the person repeats the walking pattern a
directly by the discharge current and its rise time. At lower minimum of 10 times while holding an electrode that is
humidity the severity as defined by current and rise time will connected by wire to the voltage monitor as shown in left
increase on average. The measured set of electrical current and of Figure 3. A data acquisition system records the walking
electromagnetic field data can then be compared to qualifica- voltage and associated software computes the statistics of
tion levels at which typical telecommunication equipment is the voltage recorded for the test iterations. A typical walk-
tested at prior to usage in data centers. ing voltage pattern explains the difference between walking
and standing voltage is shown in Figure 4.
HUMAN CHARGING EXPERIMENTS: The pattern is very repeatable for each walking step
The human charging test has been run so far at 60% cycle. The analysis software calculates the mean of the peak
RH, 35% RH and nominal 15% RH at a temperature of voltage and mean of the standing voltage for each test set
73.4°F±5.4°F(23°C ± 3°C). Additional testing at 59°F iteration. The peak voltage value is calculated as the 90
(15°C) and 15% RH has been done on the flooring samples percentile of the 3 sigma maximum voltage for the
using a subset of the footwear. The testing also includes STM97.2 test method. The peak values are compared in the
measurement of resistance to groundable points from the following tables for a representative sample of flooring
surface of the flooring test plates using an industry recog- materials and footwear examples.
nized contact electrode and the measurement of resistance In the example shown in Figure 5, five (5) different
to groundable points of the floor test plates from personnel types of footwear were used in the walking voltage test on
wearing the various types of footwear1. The main dynamic two types of high pressure laminate (HPL) flooring
test procedure measures the accumulation of voltage on a commonly used in raised flooring applications (designated
person while walking in a defined pattern on the floor test as “F” and “N” from the supplier). The HPL flooring
plates according to industry standard test method ANSI/ samples were rated as standard HPL, not as static protec-
ESD STM97.22 .The walking pattern established in the tive. The samples had slightly different resistance values
1. 2.
ESD Association Standard Test Method – For the Protection of ESD Association Standard Test Method – For the Protection of
Electrostatic Discharge Susceptible Items – Floor Materials and Electrostatic Discharge Susceptible Items – Floor Materials and
Footwear – Resistance Measurement in Combination with a Footwear – Voltage Measurement in Combination with a Person,
Person, ibid ibid
Figure 3 Walking Voltage Test Set-up (left), walking pattern (right) according to ANSI/ESD STM97.2.
DE-13-031 5
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 4 Typical walking voltage pattern explain difference between walking and standing voltage.
Figure 5 Two types of High Pressure laminate flooring. Two types of normal shoes, two types of static control footwear and
a pair of conventional heel grounding straps.
which accounts for the difference in walking voltage, espe- allow over the nominally accepted level of 2,500 volts for
cially when using the footwear or shoe grounding devices an office or IT environment at the lower humidity level.
with dissipative properties. The resistance point-to-point as Personnel wearing these types of shoes in that environment
measured by ANSI/ESD S7.1-Flooring for the Type F might be able to feel a static shock if they were to touch a
material (24 inches x 24 inches) was 3.1 x 1010 ohms and grounded metal object. However, these voltage levels are
for the Type N the resistance was 2.0 x 1010 ohms. still below the threshold at which operating system damage
The flooring samples were adhered to an aluminum or upset might occur if the most sensitive location is
ground plane supported on a hardboard backing. The two touched by a person, even if the person is holding a metal
humidity conditions shown represent the extremes of the object in his or her hand.
testing accomplished to date: 60% RH at 73.4°F (23°C) and As shown in Figure 6, a standard floor tile allows the
15% RH at 59°F (15°C). In these examples, the normal generation of a modest amount of electrostatic voltage with
shoes represented by the running shoe and the deck shoe most shoes tested. None of the voltage levels exceeded the
6 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
normally accepted office or IT environment voltage level of date shows that the risk is low for generation of voltage in
2,500 volts. excess of system level ESD immunity requirements for IT
A standard asphalt tile with a dissipative wax commonly equipment even at very low relative humidity.
used in commercial applications (grocery stores, offices, some
The following table (Table 1) summarizes the Standing
manufacturing) is shown in Figure 7.
and Walking voltage levels for flooring that represents the
The dissipative wax is affected by removal of moisture range tested to date. The standing voltage is determined at the
from the air as would be expected. However, the overall volt-
point in which the test subject returns to the starting position
age level for all shoe types is still very low.
in the walking test and pauses for about 2 seconds. The voltage
The walking voltage results on a conductive vinyl tile on the test subject has a chance to decay to some lower level.
commonly used in manufacturing of electronics is shown in
The software in the data analysis system used in this study
Figure 8.
calculates the standing voltage as well as the walking peak
Many additional floors have been evaluated with the vari- voltage for each test.
ety of footwear to observe the walking voltage levels. The
above examples are shown as they represent the full range of The walking voltage as reported represents the maximum
results. In only one instance was a level above 4 kV recorded voltage on the test subject while in motion. When the person
and only a few above 3 kV. While not every possible shoe type stops, the voltage begins to decay as a function of the environ-
has been evaluated in this study, the information gathered to ment but most importantly as a function of the electrical resis-
Figure 6 Standard asphalt based floor tile and various footwear at three humidity ranges 73.4°F (23°C).
Figure 7 Dissipative wax on standard asphalt tile using various footwear at three humidity conditions at 73.4°F (23°C).
DE-13-031 7
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 8 Dissipative Vinyl Tile and various footwear walking voltage at three humidity conditions at 73.4°F (23°C).
Table 1. Summary of Floor and Footwear Walking and Standing Voltage Values
HPL Type N
Dissipative Shoe 8 7 87 11
HPL Type F
8 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
tance through the body to ground. As shown in Table 1 and The Triboelectric interaction of the Standard tile to the
reported in Figure 5, the peak voltages occurred on the HPL insulative footwear allows less charging than the HPL flooring.
type flooring with a Deck Shoe and a Running Shoe. These are Please note that footwear with dissipative properties charges
commonly worn types of footwear and would appear much less on the HPL type flooring since the contact resistance
frequently in data center type environments. It should be noted is lower overall. The Deck Shoe and Running Shoe charge just
that even with these shoes at low environmental moisture as easily on the Conductive Vinyl tile as on the Standard tile.
content in the air, the Standing voltage is well below any prob- This evidence strongly suggests the use of at least moderately
conductive flooring systems in controlled areas to reduce the
lem level. It would be possible however for a person wearing
overall level of electrostatic charge accumulation, regardless of
these types of footwear on a high electrical resistance floor to
the footwear or environmental moisture. Flooring has to be
generate above 4 kV on a regular basis. Care would be needed installed anyway so the cost associated with a conductive floor
to discharge the body voltage to ground prior to handling or verses an insulative floor is minor compared to continuing oper-
touching any sensitive piece of equipment or exposed suscep- ational costs to sustain moisture levels. Existing data centers
tible parts during service activity, however equipment in its may see value in the use of commercially available wax and
standard operating condition would not be endangered. The coatings that provide reduced charging levels.
Standard asphalt based tile allowed lower charging in this set
While it may prove impossible to control the footwear
of experiments than the HPL even though the electrical resis-
worn by personnel who enter or work in data centers with
tance of the Standard tile is slightly higher as shown in certainty, it would be a good idea for facility owners and manag-
Figure 9. Electrical resistance is used to categorize flooring ers to have an awareness that footwear can lead to issues in the
types. The Red Line in Figure 9 at the 1 x 109 ohms level is daily operation of the data center. Just about any conventional
maximum suggested resistance to ground for a floor in an elec- polymer based sole material may lead to high charge levels,
trostatic protected area. Comparing Table 1 voltage to some more so than others – regardless of humidity. A conduc-
Figure 9 shows some interesting information about how tive floor will help to mitigate electrostatic charging even from
charging relates to resistance to ground. the worst possible pair of shoes.
Figure 9 Electrical resistance of flooring samples using industry standard electrode (S7.1), heel ground strap and ESD
control rated shoes (Low Dissipative Shoe).
DE-13-031 9
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
CABLE CHARGING EXPERIMENTS: the electrode toward the current target, a discharge occurs. The
Cables can charge by dragging, unrolling or by people wall of a shielded cabinet was used as a ground plane with
carrying a device, such as a laptop that later connects via a current and field sensor is placed on it while the scope is placed
cable to a computer rack. An apparatus was built to unwind inside, as shown in Figure 12. The measurement setup needs
CAT 6 data cable at a controlled rate across various floor to capture the discharge current, transient fields, the arc length,
samples. More data collection is needed in this area but the and approach speed of human metal ESD (Chun-
results to date are shown for preliminary information. druet.al.2004).
The apparatus is shown in the left side of Figure 10.
Human metal discharge measurement result
The CAT 6 cable is unwound from a spool to another
spool that is motor driven at 50 rpm. The cable is allowed to Typical waveform. There are two groups of parameters
drag with tension across a sample of flooring. The voltage is of ESD:
measured on the wires at one end of the cable. The voltage Group I: Voltage, Peak Current. Group II: Rise time,
level increases with time as shown in right of Figure 10. In this Current derivative, transient field magnitude, and transient
example, the floor sample is HPL type N. At approximately 32 field time derivative. The first group is more likely to cause
seconds of unwind, a step in voltage occurs caused by a damage, while the second group is more likely to cause upset
momentary increase in drag (friction) of the cable across the events.
floor sample. An additional step in voltage occurred at about The arc length is influenced by the approach speed, time
44 seconds. At about 48 seconds the motor was shut off and the lag and charge voltage. Fast approach speeds lead, in average,
small positive spike was recorded. Natural decay of the static to shorter arc lengths. A longer time lag tends to lead to shorter
voltage occurs from the 48 second mark to about 65 seconds arc lengths for the same approach velocity. While the arc
when the connector end was grounded causing the cable to length is the most important parameter that determines the
discharge approximately 800 volts in this example. Similar resistive phase of the arc for a given charge voltage (Pomme-
results were obtained for other floor samples. Voltage levels renke 1993), one also needs to know the approach speed for
were lower for cable that was unwound across conductive two reasons: (1)The approach speed in conjunction with the
vinyl and conductive rubber based floor samples. time lag will determine the arc length. (2) The range of normal
Additional work is needed in evaluating cable charging approach speeds during human operation is known to be
and discharging to fully characterize risks as related to data 3.28 ft/s (1m/s) so it is useful to test if the approach speeds
centers. This preliminary work is reported here to show that selected during testing fall into the normal speed range.
some techniques are being developed although refinement of The typical waveform of discharge current, transient field
the procedures is necessary. are shown below in Figure 13. Here we selected discharges
from the same voltage but different arc lengths to illustrate the
HUMAN METAL DISCHARGE effect of arc length on the discharge current:
The test setup of human metal discharge is shown in In human metal discharge measurement, the arc length is
Figure 11, the person holds a metallic discharge electrode in a very important factor that decides the repeatability of the
his hand. The person is charged to a high voltage via a test. Even if the arc length cannot be controlled it is important
450 M current limiting resistor. Then, as the person moves to know that the discharge current will repeat very well for the
Figure 10 .CAT 6 Cable Unwind Test Set up (left), CAT 6 unwind across HPL Sample “N” (right).
10 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
same charge voltage, the same discharging object having the discharge current of the short arc length reaches 40 A, while
same arc length. For a given voltage, long arc lengths induce it only reaches 10 A for the long arc length case. The other
a low ESD risk while short arc lengths cause a high ESD risk. parameters such as, current derivative, electric field and
Three different arc length cases have been chosen for plotting magnetic field all indicate the same tendency.
to illustrate the effect, long -42.47 mil (1.07 mm), mid-28.98 ESD vs. arc length. Human metal discharge measure-
mil (0.73 mm), and short- 15.08 mil (0.38mm). The top left ments are done in seven environmental conditions around the
figure shows discharge current in time domain. The peak envelope of the ASHRAE recommend range. For each envi-
Figure 11 Human metal discharge setup showing the current measurement and the electric and the magnetic field sensors. The
system has a bandwidth > 3 GHz for the current and > 2 GHz for the transient electromagnetic fields
Figure 12 Photo of human metal discharge setup inside the cabinet, and one of the authors with an electrode was charged to
a high voltage and discharge to the ESD target.
DE-13-031 11
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 13 ESD parameters as a function of arc length at 5 kV with long- 42.47 mil (1.07 mm), mid-28.98 mil (0.73 mm) and
short-15.08 mil (0.38 mm) arc length; Top left is current; Top right is current derivative; Bottom left is E field; Bottom
right is H field.
ronmental condition, the high voltage supply is set to5 kV and direct effect on the arc development, but it influences the arc
10 kV separately; For each voltage level, 20 discharges are length. For a given arc length the peak current is not a function
performed. For each discharge, the current, transient field, arc of humidity. The current and rise time correlate well to arc
length and approach speed are recorded. Before the test, the length. A long arc length induces a smaller peak current and
electrodes are slightly polished with 1000 grit paper and a slower rise time. The ratio of the current and arc length is the
cleaned with alcohol. The test process can ensure that the same in all environmental conditions. This indicates that the
surface influence on arc length, caused by changes from prior humidity has no primary effect on the current, however,
discharges is eliminated. certainly humidity has an indirect effect by influencing the
average arc length via the time lag for approaching electrodes.
The ESD parameters vs. arc length are shown in
Figure 14. The top left is the current (5 kV), top right is current In IEC testing, the two parameters that indicate the sever-
(10 kV) the bottom left is the rise time (5 kV), and the bottom ity of ESD are rise time and current. The standard sets values
right is the rise time (10 kV). The same color is used for same of 850 ps and 3.75 A/kV. If the rise time is less than 850 ps or
relative humidity. With a slow approach speed, Paschen’s law the current is larger than 3.75 A/kV, then it may cause prob-
is satisfied in all charge voltage settings and all environmental lems to equipment in a data center. For a given voltage, we
conditions. This just verifies the correctness of the measure- want to determine how short does the arc length have to be to
ment setup. For faster approach speeds the combined effect of surpass the limit of 850 ps rise time and/or 3.75 A/kV current.
approach speed and time lag lead to discharges over a shorter For 5 kV, the current exceeds the threshold at about 23.6 mil
distance. Here it is important to notice that humidity has no (0.6 mm) arc length, while at 27.5 mil (0.7 mm) the rise time
12 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 14 ESD parameters as a function of arc length for human metal ESDs. Top left is current-5kV. Top right is current-10kV.
Bottom left is rise time-5kV,Bottom right is rise time -10kV.
is less than the 850 ps threshold. For 10 kV, the arc length in Figure 14. As indicated previously, for 5 kV charge voltage,
threshold for current is 62.9 mil (1.6mm), while it is 66.9 mil when the arc length is less than 23.6 mil (0.6mm), the
(1.7 mm) for rise time. discharge current violates at least one parameter in the IEC
Arc length vs. approach speed. There is no deterministic settings 3.75 A/kV limit. In high humidity, only when the
relationship between speed of approach and arc length, approach speed is higher than 6.56 ft/s (2 m/s) which is outside
however, on average, the arc length will be reduced once the of the normal human approach speed(0.0328 ft/s to 3.28ft/s
speed of approach surpasses a threshold value. This threshold (0.01m/s to 1m/s)), does relative arc length decreases to 0.6. In
value is a function of humidity, surface conditions and voltage. low humidity (outside the recommended envelope), the arc
This process had been observed prior by (Pommerenke,1993) length decreases to 0.6 when the approach speed is less than
and (Frei,1999]. The left side of Figure 15 is for the high rela- 0.328 ft/s (0.1m/s).The data shown below have been taken for
tive humidity (70% RH), and the right side is the low relative clean, rounded surfaces. More realistic sharp cornered elec-
humidity (20% RH) at 73.4°F (23°C. As the high humidity trodes would enhance the electron emission, thus, reduce the
keeps the time lag short, the influence of approach speed is time lag and lead to higher threshold speeds even in dry condi-
small. The threshold of approach speed is also considerably tions. Based on this argument the data shown below will over-
large. In low humidity, the time lag is longer, and the influence estimate the potential effect in data centers.
of approach speed is stronger. The threshold of approach Relative humidity or absolute humidity. Relative
speed is small. They are all in accordance with results shown humidity or absolute humidity, which is more important to
DE-13-031 13
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 15 Relative arc length as a function of approach speed in 23°C left-70% RH, right-20% RH)(Frei,1999).
indicate the risk of ESD? The question is relevant as the rela- For each charge voltage, 20 single discharges with slow,
tive humidity will vary strongly within a data center: It will be medium and fast approach speeds are done.
much lower at an air outlet of an air cooled rack. If absolute
With a 5 kV charge voltage, the current as a function of
humidity would be the better measure to influence ESD risk a
simpler assessment of the data center situation would be possi- arc length is shown in Figure 16. The two cases on the left side
ble. The study will answer this question, as of now, only a qual- of Figure 16 are with the same absolute humidity. The ratio of
itative analysis can be given. current and arc length is the same so we don’t see an influence
by the humidity level. With the same approach speed, the
For the discharge phase we know that the arc length discharge current range of 89.6°F (32°C), 40% RH is much
depends on the time lag. For a given approach speed, a longer larger than that of 80.06°F (26.7°C), 55% RH. The two envi-
time lag induces a shorter arc length, and a higher discharge ronmental conditions on the right side of Figure 16 are with
current. Once the charge voltage reaches the breakdown volt-
same relative humidity. With the same approach speed, though
age, one electron can initiate the breakdown. The electrons
the absolute humidity is different (1.12 g/ ft3 (2.5 g/m3)vs.
that initiate the breakdown can be detached from water mole-
2.74 g/ft3 (6.78g/m3)),the discharge current range are similar.
cules on the cathode surface. The more electrons that are avail-
able, the more likely is the initiation of a breakdown and the They both indicate that the relative humidity is more mean-
shorter the time lag which causes a longer arc length, and ingful to represent the severity of ESD. In a data center, the
therefore in general reduces the severity of the ESD. As we absolute humidity is controlled very well, but the local temper-
explained in the previous part of the paper, the initial electron ature might be higher which is caused by the operational elec-
is very important for the discharge phase. If many initial elec- tronic equipment. The high local temperature causes the low
trons are available a spark will develop over a longer distance, relative humidity.
causing a slower rise time and a lower peak value. Of all
factors that provide initial electrons, humidity is the dominat- CABLE DISCHARGE EVENT
ing one, as moisture in the air will form water layers on the
surface and electrons easily get attached and detached. Water The cable CAT5e was spooled and placed on the 2 inch
layers are more likely in high relative humidity, Thus, the rela- thick foam for insulation. A high voltage supply charges at the
tive humidity is assumed more important than the absolute Al-foil that is wrapped around the cable. This emulates a situ-
humidity. ation in which the cable has been dragged along a floor and
To evaluate the effects of relative humidity and absolute carries charge, created by tribocharging, on the surface of the
humidity, two groups of measurement are done. In the first cable jacket. The high voltage polarizes the wires inside the
group, the relative humidity is kept the same (20% RH), but the LAN cable which leads to an ESD event if the gap, formed by
temperatures are varied -59°F to 89.6°F (15°C to 32°C). In the inserting the cable connector into a connector, is small enough.
second group, the absolute humidity is kept the same 5.8 g/ With a 10 kV charge voltage, the polarized voltage on the wire
ft3(13.4g/m3), but the temperatures and relative humidity vary is about 5 kV which can be estimated from the arc length at a
(80.06°F (26.7°C), 55% RH, and 89.6°F (32°C), 40% RH). very slow speed. A typical waveform is shown in Figure 17.
14 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Figure 16 Current as a function of arc length with 5kV charge voltage, left (same absolute humidity), right (relative humidity).
The peak current is about 11 A, and the rise time is 0.6 ns. The frequency and the longer the cable, the lower the ringing
ringing frequency of the current is about 4 MHz. The ringing frequency. Further investigations will differentiate between
frequency is dependent on the length of the cable being the common mode and the differential mode discharge. In
discharged, i.e., the shorter the cable, the higher the ringing common mode all wires of a LAN cable are charged against
DE-13-031 15
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
ground, and the differential mode discharge that results from humidity. Compared with absolute humidity, relative humid-
sequenced contacting of the pins inside a connector. Due to the ity is more important.
circuit used in LAN ports differential mode discharges pose a The impact on the charging process in reduced humidity
greater risk. conditions has been investigated. Flooring and footwear mate-
rials commonly used have been evaluated. The difference in
DISCUSSION charge accumulation on a person at different humidity and
moisture content levels has been investigated. Charging levels
If a human or an object is charged to the same voltage and beyond the immunity testing levels required for IT equipment
discharged repeatedly, a large variation of current waveforms were exceeded in only one data point out of thousands
will be observed. The variations in the current are a result of collected, and that was at the most extreme condition. This
differences in arc lengths. The arc length is determined by the paper presented the motivation and the setup of this study and
approach speed, charge voltage and time lag. With the same showed typical results which allow the authors to give prelim-
charge voltage and the same approach speed, the time lag is the inary conclusions. Presenting a quantified risk analysis of
most important factor. A longer time lag induces a shorter arc reduced humidity in data centers is the goal for the next paper.
length. Time lag is determined by the availability of seed elec-
trons. These can be provided by field emission, by surface REFERENCES
roughness enhanced field emissions and by detachment from
ANSI/ESD S20.20 -2007. ESD Association Standard –For
water molecules. The surface condition is kept roughly the
the development of an Electrostatic Discharge Control
same for all the cases. With the same charge voltage; the
Program for –Protection of Electrical and Electronic
humidity is the most important factor. Humid air has a strong
Part, Assemblies and Equipment (Excluding Electrically
effect in reducing the time lag, most likely due to electron
Initiated Explosive Device).
detachment from the water molecules on the cathode surface.
As the relative humidity indicates the moisture on the cathode Chundru, R., D.Pommerenke, K.Wang,T.Van Doren, F.
surface, it is the more important component when compared P.Centola, and J. Huang.2004. Characterization of
with absolute humidity. The measurement results with same human Metal ESD reference discharge event and corre-
relative humidity but different absolute humidity support this lation of generator parameters to failure levels-part I:
conclusion. reference event., IEEE Transactions on EMC, 46(4):
498-504.
Charge accumulation by human activity has been shown Frei, S., M. Senghaas, R. Jobava, and W. Kalkner.1999. The
to increase as the moisture content of the air decreases. influence of speed of approach and humidity on the
However, as has been discussed in industry standards, adding intensity of ESD. 13th Int. Zurich Symp. EMC, Zurich,
moisture to the air does not change the charging process to any Switzerland:105–110.
great extent, only the accumulation of charge is affected. The Horvath, A., and A. Shehabi.2008.Improving the Energy
human body can store charge for slightly longer periods of Performance of Data Centers. University of California
time when the moisture content of the air is low (Swenson and Energy Institute.
Kinnear 2009). The relaxation of the charge on the human
Hydeman, M. 2010. Implications of Current Thermal Guide-
body can be observed, for example in Figure 4. When the
lines for Data Center Energy Use. ASHRAE Journal.
walking motion is stopped, the charge dissipates (relaxes) at a
Hydeman, M., and D. Swenson. 2010. Humidity Controls for
rate controlled by the resistance of the discharging circuit – in
Data Centers. Are they Necessary? ASHRAE Journal.
this case the person – footwear and floor resistance. The floor
IEC 61000-4-2, 2008. Testing and Measurement Tech-
and footwear resistance is affected by moisture for some floor-
niques—Electrostatic Discharge Immunity Test, IEC
ing materials and also most of the footwear soles. While not a
Standard 61000-4-2 Ed. 2.0.
perfect predictor of charge generation, the resistance to ground
Mardiguian, M. 2009. Electrostatic Discharge. New York:
of a person will give a good indication of the ability of the
Hoboken, NJ: Wiley-IEEE Press: 15-19.
person to hold a static charge and provide an estimate of how
long the charge will reside (decay time). The installation of a Jayong K., C. Qing, K.Wang; J. Maas, T. Takahashi, A. Mar-
conductive floor in any environment where electrostatic twick, and D. Pommerenke.2008. Correlation between
charge control is desired is a sound investment. The use of EUT failure levels and ESD generator parameters. IEEE
conductive or dissipative coatings or wax would be of benefit Trans. EMC. 50(4):794 - 801.
in existing facilities to reduce charge generation regardless of Pommerenke, D. 1993. On the influence of the speed of
the humidity level. approach, humidity and arc length on ESD breakdown.
ESD forum, Grainau: 103-111.
Pommerenke, D. 1995.ESD: Transient field, arc simulation
CONCLUSION
and rise time limit.J. Electrostatics, 36(1): 31–54.
A low humidity supports an increase in ESD severity. The Renninger, R.G. 1992.Optimized Statistical Method for Sys-
effect of humidity in ESD severity is a function of charge volt- tem-Level ESD Tests.IEEE International Symposium on
age; the higher the charge voltage, the weaker the effect of Electromagnetic Compatibility: 474-484.
16 DE-13-031
PREPRINT ONLY. Authors may request permission to reprint or post on their personal
or company website once the final version of the article has been published.
Simonic, R. 1981.Personnel ESD Statistics.IEEE/EMC tions of economizer use in California data centers. Pro-
Symposium, Boulder, CO. ceedings of the 2008 ACEEE summer study on energy
Swenson, D. and J. Kinnear. 2009. The role of relative efficiency in buildings, Asilomar, CA.
humidity and dew point on electrostatic charge genera- US Department of Energy. 2007. Annual Energy Outlook
tion and electrostatic discharge (ESD).The Green Grid - 2007, with projections to 2030.Energy Information
White paper. Administration, Washington, DC 20585.
Shehabi A., S.Ganguly,K.Traber,H.Price,A.Horvath,
WW.Nazaroff,and AJ. Gadgil. 2008. Energy implica-
DE-13-031 17