Session Paper No' Emerging Applications for Tunable Diode Laser, the Promise and the Problems

1
Session / Paper No. Emerging Applications for
Tunable Diode Laser, the Promise and the Problems

• Trevor Knittel, Don Wyatt, Alan Cowie, Jie Zhu
• Analytical Specialties Inc. 910 Gemini, Houston,
  TX, 77058 
•  Linh D. Le, J.D. Tate
• The Dow Chemical Company, Freeport, TX.

2
TDL BACKGROUND

• TDL has its origins in Research, Atmospheric
  measurements and Aerospace applications. Early
  limitations were due to the lack of available
  lasers that were low cost and reliable.
• Emissions Open Path markets were opened up when
  the telecomm market developed low cost, long life
  laser sources. These applications tended to be
  for applications that had no attractive
  alternatives (NH3, HCl, HF, etc.). TDL analyzers
  have been installed in these applications for gt9
  years.
• Combustion applications were the next step for
  TDL analyzers. Due to the ability to measure at
  high temperatures (up to 1500C) with a path
  average (vs. point) , and no sensor contact to
  the process, TDL was applied to many of the most
  severe service applications (high dust,
  corrosive, etc.).
• PROCESS applications have not seen widespread TDL
  implementation (when compared to other analytical
  techniques such as paramagnetic, zirconia,
  infrared, GC, etc.
• This is primarily due to requirements not seen in
  research, emissions or combustion applications
• Measurement with simultaneous changes of
  background gases, temperature and pressure
• Requirement to seal the process for corrosive and
  lethal service
• Basic instrument features required such as
  auto-validation, user interface, valve control
• PROCESS APPLICATIONS ALSO REQUIRE MUCH MORE
  ATTENTION TO APPLICATION DETAILS

3
TDL OPERATION

• TDL is an optical measurement. The measured gas
  absorbs the laser light at a specific wavelength.
  The amount of light absorbed is a function of
  gas concentration, pressure, temperature and
  optical path length.
• Method of operation
• The system electronics control the current fed to
  the laser (1)
• The laser (2) wavelength changes with current
• The modulated laser light is collimated to a
  circular beam (2), and sent through process
  windows (3) and the process gas
• The light transmitted through the process gas
  strikes a detector (4)
• This signal will show an absorption (loss of
  light) at the wavelengths the gas absorbs

1
3
2
4
Signal at Detector
Current ramp to laser
4
POSSIBILITIES

• Accuracy improvements
• Speed of analysis
• Representative measurement
• Measurement under extremely aggressive process
  conditions
• Installation cost reduction
• Operational cost reduction
• Positive indication of operation

5
DIFFICULTIES

• Window fouling
• Particulate loading
• Sensitivity drift issues at high temperatures
  or short path lengths
• Temperature/pressure compensation
• Calibration / validation

6
ACCURACY IMPROVEMENTS

• Accuracy improvements generally fall into three
  categories
• Interference Free Measurement. Due to high
  resolution of measurement, spectral overlap can
  usually be avoided. If the measurement method is
  immune to foreign gas broadening a truly
  interference free measurement is possible.
• Speed of Analysis. With no sample transport and
  conditioning delays and a fast analysis cycle
  (1-5seconds) process changes can be seen that are
  not visible to slower analysis methods
  (continuous or batch)
• Representative Measurement. With no requirement
  to remove a point sample (probe) the entire
  process is measured eliminating distribution
  errors. Also with the possibility of measurement
  at process conditions the sample is not altered
  prior to analysis (i.e. sample cooling and loss
  of moisture heavy components)

7
INTERFERENCE REJECTION

• Some interference typically considered are,
  spectral (photometers/spectrometers),
  paramagnetic background gases (paramagnetic
  oxygen), and hydrocarbons for zirconia oxygen.
• TDL in addition to spectral interference must
  consider foreign gas (collisional) broadening.
• Spectral interference is generally easy to
  overcome due to the ability to leverage high
  resolution (narrow laser linewidth) and fine
  rotational structure of target gas.
• Foreign gas broadening can be minimized with
  fitting or eliminated with peak area integration
  (unfitted)

Interference Free CO2 Peaks
8
EXAMPLE OF INTERFERENCE REJECTION IN FIELD

• Aggressive waste destruction system with TDL
  (green trend) installed downstream of 2-zirconia
  analyzers (red-blue trend)
• All analyzers dropped during change in
  incinerator load. One zirconia dropped below 0
  (not possible) other dropped to 1 oxygen.
  Correlation with CO data from CEMS suggests that
  both zirconia analyzers dropped artificially low
  (possible due to unburned hydrocarbons)

9
SPEED

• Depending on process dynamics extractive
  analyzers may miss process events due to sample
  transport and conditioning lag times
• Short duration excursions (process or fault
  driven) may be missed or smoothed due to
  measurement delays
• TDL can offer lt5s analysis times with measurement
  through the process
• Diagnostics can prove measurement validity if
  operations questions data

10
SPEED II

• Both systems exposed to 4.91ppm H2O for one hour,
  then returned to 0 ppm N2
• Maximum P2O5 response after 1 hour exposure
  2.6ppm

8ppm short term exposure missed by slower
response analyzer
11
REPRESENTATIVE MEASUREMENT

• Representative measurement can be achieved by
  measuring in the process or at process conditions
  (temperature, pressure, etc.)
• Two main advantages are for
• Processes were there is a spatial distribution of
  the analyte (i.e. large scale combustion)
• Processes with high levels of water that need to
  be removed for extractive analysis (net vs. wet
  measurement)

12
DIAGNOSTICS

• IF the analyzer has the ability to automatically
  capture and store spectra diagnostic ability is
  greatly improved.
• In this example the TDL readings were questioned
  since the measurement speed showed reading
  spikes that were not expected.
• HISTORICAL spectra was already present in the
  analyzer allowing analysis and confidence that
  the readings were correct.

13
TRANSMISSION LOSSES (WINDOW FOULING, PARTICULATE)

• Since both resonant (peak) and non-resonant
  (baseline) information is collected during each
  scan, loss of power has no significant effect on
  the measurement until gt95 of laser power is
  lost. Pathlength can be reduced with insertion
  tubes to overcome laser power losses.
• Window fouling presents more difficulty since
  coatings can affect the optical surface and beam
  quality. Generally this can be overcome with
  protective purge in front of the process window
  (measurement is now non-contacting). Window
  purge is possible for both in-situ and close
  coupled extractive installations.

14
AGGRESSIVE SERVICE

• Dealing with a non-conditioned sample requires
  careful attention to the process interface
  design.
• For high pressures and aggressive/lethal samples
  the process seal must be independent of alignment
  mechanisms.
• Window and seal materials must be carefully
  selected for process conditions (i.e. sapphire)
• Window purge will protect optics from excessive
  temperature and coating/fouling

15
EFFECT OF OPTICAL PATHLENGTH, TEMPERATURE AND
PRESSURE

• The fundamental performance limitation of TDL
  systems is typically etalon noise/drift
  (constructive interference from laser
  reflections).
• Etalon measurement interference is difficult to
  eliminate but can be minimize with proper optical
  design and signal processing.
• Etalon noise/drift is typically an absolute error
  (independent of measurement level). As such the
  measurement equivalent is affected by
• Optical pathlength. Linear relationship, as
  pathlength increases the etalon error
  contribution is decreased.
• Process gas temperature. Most lines will
  decrease with temperature, for example one
  candidate oxygen line will decrease by 50 when
  increasing temperature from 80F to 550F. For
  high temperature applications it is sometimes
  possible to select lines that increase (or
  decrease less) with temperature
• Process gas pressure. With peak area integration
  pressure has a linear effect. Higher pressure
  process conditions will proportionally reduce
  etalon noise/drift.
• For reference most suppliers supply performance
  data at a typical process condition (25C, 1m
  path, 1barA). It is important to evaluate the
  effects of process conditions on the measurement
  performance (noise, drift, accuracy)

16
PRESSURE/TEMPERATURE COMPENSATION

• Peak height and area are pressure and temperature
  dependant
• Pressure effect on peak area is linear, pressure
  effect on peak height is non linear
• Temperature effect is non-linear for peak height
  or area
• Pressure and temperature can be compensated for
  real time by feeding 4-20mA signal to analyzer
• Pressure and temperature tests should be
  performed on the analyzer under controlled
  (offline) conditions to validate compensation
  curves.
• Pressure can be simple to generate and test, and
  is easy to compensate (esp. since area is linear
  compensation)
• Temperature is more difficult when targeting
  measurements gt1000C. Specific test equipment is
  needed.
• As with any analytical measurement when
  developing calibrations and compensation curves
  accurate tests and standards are critical
  especially when operating under uncontrolled
  conditions

High Temperature gt1200C Test Furnace
17
INSTALLATION OPTIONS

• CROSS STACK/PIPE (IN-SITU)
• Measurement across the process
• Path integrated measurement

• BYPASS LEG
• Large diameter pipe run
• Process flow through measurement leg or-
• Process slipstream through measurement leg
• Allows isolation from process
• FLOW CELL
• Pull or push sample through flow cell

18
VALIDATION

• Full off-line calibration/check using flow cell
• Zero and span
• Block in place calibration/check for bypass and
  extractive applications
• Zero and span

• On line dynamic spiking using serial bump cell
• Manual or Auto initiate
• Allows adding successive spikes of gas being
  measured
• Validates response
• Validates response level
• No functional test of zero
• Positive indication of operation for complete
  analysis system