TSI AEROTRAK® PORTABLE

PARTICLE COUNTER
MODEL 9110

APPLICATION NOTE CC-107

Introduction
This purpose of this document is to detail the advanced, state of the art features TSI has
incorporated in the design of the AeroTrak 9110 0.100 µm 1 CFM particle counter.
Measuring 0.100 µm airborne particles in clean spaces using optical scattering techniques is
challenging. The amount of light scattered by 0.100 µm particles is very small. Ensuring that this
signal is clear of any electronic noise is a key design requirement. 0.100 µm particles must be
detected and counted efficiently; they must not be lost in electronic noise.

End users need an optical particle counter to have excellent size resolution in order that
the sizing and counting of particles is consistent year after year. Calibration sizing errors
due to poor size resolution are unacceptable.
Semiconductor cleanroom air is very clean and commonly very low levels of airborne particles
are counted. When measuring the particulate cleanliness of the air it is unacceptable to end
users that the measurement be confounded by false particle counts. The difference between
good and poor performing instruments is their ability to differentiate between false counts and
true particle count events.
Understanding the root cause of airborne particle contamination in a manufacturing
environment is important to end users. Optical particle counters with a wide dynamic size
range can help to provide important information to support this work.
With all good designs it is not one single feature, it is the combination of many advanced
features that delivers excellence. The result is a reliable, easy to use instrument that is
lightweight and delivers an excellent measurement in semiconductor applications.
What follows are details of innovative, state of the art techniques developed by TSI that build on
25 years of 0.100 µm particle detection technology in cleanroom applications.
State of the Art- Active Cavity Laser
Basic Optical Particle Counter technology has been around since the 1960s, and continues to
evolve.
The AeroTrak 9110 portable particle counter deploys 633nm Helium-Neon or HeNe laser
active cavity laser sensing technology. HeNe lasers from the 1990s are still in operation today,
so they have a proven long life.
Just because laser diode technology is newer it does not mean that it is better. Laser diode
based “0.100 µm” products are difficult and expensive to maintain, real laser diode lifetimes are
unknown and unproven.
Typically only 1% of the laser power available in a HeNe laser ever makes it to the outside
world. Active cavity simply means we are releasing the laser power that is locked up between
the mirrors of a HeNe laser into the sensing area of the particle counter optics.
Active cavity HeNe laser based optics technology is over 25 years old. The optics technology
deployed by TSI in the Model 9110 is based on this foundation. However, TSI has taken all the
lessons learned over these years and completely redesigned the AeroTrak 9110 particle
counter’s optics, detectors and sensing electronics from the ground up.
The result is that airborne particles will be illuminated by laser cavity powers in excess of 40
Watts at a low wavelength of 633nm. The advantage of this combination is that it delivers at
least 50X more scattered light than the same particle illuminated by a laser diode. This is the
starting point for an excellent signal to noise ratio. It is not the whole story though, the side
effect of high laser cavity power is a high background optical noise caused by molecules of air
scattering light. How we overcome this steady state background optical noise is covered later in
this document, for the moment we will continue to focus on the optics.
Figure 2 shows how historically two sets of Mangin mirrors were required to collect scattered
light from the particle. Shown are two detectors which were needed in the past to facilitate a
large dynamic size range and noise cancellation.

Sample Airflow

Particle inlet nozzle

HeNe Laser

Mangin Mirrors
Fig 1. HeNe Active Cavity Optical Bench- Top
View

Detector Diode #2
Mangin
Mirrors #2

Detector Diode #1 Mangin

Mirrors #1

Page 2 of 9
Fig 2. HeNe Active Cavity Optical Bench- Side View, (air flow is towards the reader)

Page 3 of 9
In the past detector diode #2 handled sizing of ≥0.1 µm particles and detector diode #1 sizing of
≥0.2 µm particles. Twin detector systems as shown in figure 2 have been used to inhibit false
counts from high energy particles. Typically, but not always exclusively, a high energy particle
would be detected by only one of the detectors, not both, enabling the high energy or dark
particle count to be rejected, more on this later.
The challenge with twin detector systems is that they are costly, and are complex to
manufacture and maintain. Aligning both detectors and crisply focusing the part of the laser
beam through which the particles travel onto both detectors simultaneously is difficult and
time consuming.
Based on these lessons learned, TSI has redesigned the HeNe optical bench of the AeroTrak
9110 particle counter from the ground up. We have reduced the complexity and the number of
components needed, see figure 3. There is only one set of Mangin mirrors and only one detector
in the latest design. The Mangin mirrors collect scattered light over 80° and deliver excellent
image quality to the detector.

Inlet ~10 mm wide x

1 mm high

Only 1 set of
Only 1 Mangin
Detector Mirrors
Diode
Fig 3. AeroTrak 9110 HeNe Active Cavity Optical Bench- Side
View

The resulting improvement to particle image quality and optimal laser cavity power improves
our signal to noise ratio and size resolution, enabling the Model 9110 to meet the requirements
of the ISO 21501-4 calibration standard.
In recent years 0.100 µm instruments have claimed to be able to count and size 0.100 µm
particles. The reality was that in fact the particles being used to calibrate these instruments
were not exactly 0.100 µm in size, in some cases they were calibrated using 0.107 µm particles.
At first glance this may not seem to be a big problem; we are talking about only 7nm here.
However, examining the Rayleigh scattering curve in figure 4 below, it is clear that a 0.107 µm
particle scatters 50% more light than a true 0.100 µm particle. This means that a 0.100 µm
particle counter that only measures down to 0.107 µm will count ~13% less particles than a
particle counter that actually measures down to 0.100 µm.
In order to meet ISO 21501-4 optical particle counter calibration standard the instrument must
have a counting efficiency of 50% ± 20% at the first particle size threshold, 0.100 µm in this
case. It was clear that a significant improvement was needed in optical power and signal to
noise ratio to deliver the extra signal and performance needed to meet the ISO 21501-4
requirement. The good news is that the AeroTrak 9110 particle counter easily meets and
exceeds this requirement, typically achieving a first size channel counting efficiency of 60%.
See figure 5 for an oscilloscope screen shot showing a 0.100 µm calibration particle being
detected.

Page 4 of 9
 Fig 4. Rayleigh Scattering Curve or 0.100 and 0.107 µm Particles

Fig 5. A typical 0.100 µm Particle Signal Pulse Taken from a Single Detector
Element

An additional advantage of the excellent image quality onto the detector is excellent size
resolution. The minimum requirement for size resolution per ISO 21501-4 is 15%, we regularly
achieve ≤5% resolution @ 0.2 µm, easily exceeding the ISO 21501-4 requirement. Excellent size
resolution means reduced size calibration errors when determining position of a peaking
calibration particle during instrument calibration, leading to consistent year on year
instrument calibration. Additionally, as a result of the excellent sizing resolution of the
AeroTrak 9110, we are able to count and report particle size at 8 individual size thresholds
across our measurement range. End users will now have more information about airborne
particulate contamination in their environment.

Page 5 of 9
The concentration limit of the AeroTrak 9110 particle counter per ISO 210501-4 is
100,000 c/ft³ or 3,500,000 c/m³ at a 10% coincidence level. This is the number of
particles
sampled by the instrument at which there is a 10% probability that there will be two particles
present in the laser beam at the same time. A significant factor that influences this specification
is the physical size of the laser beam. All optical particle counters assume that there is only ever
one particle in the beam at any given moment in time. If two particles are present in the laser
beam an optical particle counter cannot tell, it simply assumes that there is one particle and
sizes the particle count event as one large particle, see figure 6.

Fig 6 Particle coincidence, two particles in the laser beam

A concentration limit of 100,000 c/ft³ or 3,500,000 c/m³means that the AeroTrak 9110 particle
counter easily meets the requirements per ISO 14644-1 ISO Class 6, see figure 7. For
environments that contain large numbers of airborne particles, such as ISO 7-9, ISO 14644-1
guides end users towards more suitable instruments with a minimum sensitivity of 0.3-0.5 µm.

Fig 7. ISO 14644-1 1999 Airborne Particulate Cleanliness Classification

Table

The most common cause of technical problems when using high sensitivity optical particle
counters is optics contamination. Running a 0.100 µm optical particle counter in an
environment where there are high levels of airborne particles will increase the risk of optics
contamination. This is why TSI has ensured that the advanced HeNe optical bench can be easily
cleaned on site if needed. There is no need to send the AeroTrak 9110 particle counter off site
for cleaning.
In the next section we will detail other technical advances that have enabled TSI to use a single
detector optical system in the AeroTrak 9110 particle counter.

Page 6 of 9
State of the Art – Advanced Detector Diode
In the late 1980s detector diodes were developed that facilitated optical noise reduction. Up
until then the part of the laser beam through which the particle travels was imaged onto a
single detector diode. This image is rectangular in shape because the inlet nozzle of 0.1 µm
1CFM instrument is also rectangular in shape, ~ 1mm high by about ~10mm wide, see figure 3.
The technical challenge was that the detector diode was viewing a very large area of a high
cavity power laser beam. This meant that it was detecting a large amount of steady state
scattered light, another name for this scattered light is optical noise. The result is that the small
amount of light scattered by a 0.100 µm particle is masked or hidden by this background optical
noise.
To overcome this problem detector diodes were developed that had many smaller elements.
These are called Array detector diodes. Because each element is small it is only exposed to a
small area of the laser beam thereby reducing background optical noise. The number of
elements within the detector has advanced over the years. See Fig 8 below to see this
evolution.

Inlet Nozzle
Laser
Beam
Mangin Mirrors

Array Detector Diode

10 mm

1988: 12 Element Detector Diode

2003: 14 Element Detector Diode

TSI 2009: 16 Element Detector Diode

Fig 8. Detector Diode Evolution

Page 7 of 9
The advantage of the TSI AeroTrak particle counter is that it utilizes a 16 element array
detector diode, instead of having 1 detector we now have 16 very small detectors. This means
we are getting a ~16X reduction in background optical noise resulting in very low detector
electronic noise. The whole area of the laser beam through which the sample air flow passes is
imaged onto the 16 element detector diode, see figure 9, and is another factor that leads to an
excellent signal to noise ratio.

Sample air flow region of laser beam

Laser Beam – air flow is towards the viewer

Mangin 0.100 µm particle

Mirrors scattered light is
imaged onto a
detector element

Sample air flow region of laser beam

is imaged onto multi-element detector

Fig 9. Laser Beam is sharply focused onto the Multi Element Detector

In order realize this optical noise reduction benefit each of the individual detector elements has
to be processed individually. In the next section we are going to examine the important role of
signal processing when counting 0.100 µm particles in air.

State of the Art – Signal Processing

We can process each array detector element individually to help us reduce electronic noise, and
facilitate a wide dynamic size detection range. Signal processing technology and techniques
have evolved over the last 25 years. It is this foundation of lessons learned, and what works
best that have inspired us to develop and enhance signal processing methods to deliver an
excellent measurement performance.
Interestingly, many patents have been generated over the last 25 years to facilitate optical and
electronic noise reduction and thereby improve the signal to noise ratio, many of these are
currently owned by TSI. However our new detector design and signal processing perform so
well we do not even need to use this older intellectual property in the AeroTrak 9110 particle
counter.

Page 8 of 9
Electronic Noise Cancellation
High levels of electronic noise will severely impact a signal to noise ratio. As discussed earlier,
electronic noise can be caused by optical noise inside the measurement optics, although this is
not the only cause. Other components of this noise are inherent in the detector diode itself and
the associated electronics. The optical signals are extremely small and the electronic
amplifiers are needing to work hard to amplify them, and will be contributing towards
electronic noise.

Fig 10. Electronic Noise Cancellation by Cross Feeding

To reduce electronic noise TSI takes individual element signals and Cross Feeds them with
other elements, see figure 10. Cross Feed is a term used to describe the practice of combining
two individual element signals to significantly reduce the electronic noise component of each
signal. The technique of Cross Feeding individual elements is not new, what is new is that way
we are doing this. We have taken a tried and tested foundational technique and improved it.
Because a very large amount of light is scattered from larger particles we can also choose not to
use all the detector elements when sizing large particles. This means we do not saturate the
detector electronics with huge amounts of scattered light signal when sizing large particles. This
delivers an excellent sizing performance of 100:1 dynamic range from 0.100 µm through to 10
µm.

Analog to Digital Conversion

The analog signals from each individual detector are then converted to digital signals to allow
us to perform a more complex analysis. This analysis will determine whether or not the particle
is a true particle count or a false count due to high energy particles such as cosmic rays.

False or Dark Particles

The main sources of false or dark particle counts are typically Cosmic Rays. These are high
energy particles that originate from outside the Solar System. When they pass through a solid
state detector such as a photo-diode detector they will generate a pulse that could be counted
as a real particle. The frequency of Cosmic Rays is not constant; they are influenced by Earth’s
magnetic field. The incidence of Cosmic Rays can vary depending on your location on the planet
surface. The glass within the array detector itself can also be a source of dark particle counts
due to radioactive decay,
Different approaches have been used in the past to prevent 0.100 µm 1 CFM optical particle
counters from counting dark or false counts. One approach has been to use 2 physically
separate detectors, see figure 2. This is based on the principle argument that the same Cosmic
Ray is unlikely to pass through two physically separate detectors at the same time. Simply, if
one of the detectors sees a pulse and the other does not then the “particle” is rejected and not

Page 9 of 9
counted.

Page 10 of
9
Another approach used utilizes statistical filtering algorithms. This is an algorithm in the
particle counter firmware that artificially subtracts particle counts over a period of time. It
makes assumptions that dark counts occur at a typical frequency. The side effects of statistical
filtering algorithms are more significant in very clean environments as they may eliminate
significant percentage of actual particle counts, giving end users a false sense of security.
Because TSI has now digitized the signals within the AeroTrak 9110 particle counter we can
perform some detailed analysis on the particle signals using a Field Programmable Gate Array
or FPGA. These dark particles are high energy particles and so their signal characteristics are
different from an airborne particle signals. We analyze the signal for many characteristics, rise
time, transit time and shape, to determine if the signal is a result of airborne particulate
contamination, see figure 11. TSI does not use statistical algorithms to reject dark particles and
we do not use a second separate detector. We analyze each individual signal.

Accepted and Counted Rejected and Not Counted

True Particle Signal Cosmic Ray or High Energy Particle

Fig 11. Dark Particle Rejection using Field Programmable Gate Array FPGA

Conclusion
The TSI AeroTrak 9110 Particle Counter meets and significantly exceeds the ISO 21501-4
particle counter calibration standard. TSI has built on technological advances over the last 25
years and has used them as a solid foundation to develop new and innovative ideas. The result is
a 21st century 0.100 µm particle optical counter that is easy to use and maintain, and delivers an
accurate and reliable measurement each and every day.

TSI Incorporated – Visit our website www.tsi.com for more information.

USA Tel: +1 800 874 2811 India Tel: +91 80 41132470

UK Tel: +44 149 4 459200 China Tel: +86 10 8219 7688
France Tel: +33 4 91 11 87 64 Singapore Tel: +65 6595 6388
Germany Tel: +49 241 523030

CC-107 Rev. A (A4) ©2013 TSI Incorporated Printed in U.S.A.