EPA-AA-SDSB-80-19
                          Technical Report
         An Investigation of Photoacoustic Spectroscopy as  a
        Technique for Measuring Diesel Particulate  Emissions
                                 by


                          Daniel P. Reiser


                          September, 1980
                               NOTICE

Technical Reports do not necessarily  represent  final  EPA  decisions
or positions.  They  are intended to present technical analysis  of
issues using data  which are currently  available.   The purpose  in
the  release of  such reports is  to facilitate the  exchange  of
technical information and to inform  the public  of  technical  devel-
opments which may form the basis  for a final EPA decision,  position
or regulatory action.

               Standards Development and Support Branch
                 Emission Control Technology Division
             Office of Mobile Source Air Pollution  Control
                  Office of Air,  Noise and Radiation
                 U.S. Environmental Protection  Agency

-------
I.   Introduction and Background

     EPA  is  in  the process  of proposing  a particulate emission
standard for  heavy-duty  diesel  vehicles.   With this  mandated
requirement to  regulate  particulate  emissions, EPA must propose a
test procedure  for the measurement  of particulate emissions from
diesel engines.  A draft  recommended  practice for  measuring partic-
ulate  emissions  from heavy-duty  diesel  engines  has  already been
developed and described in  a  previous report.^/   In  the  draft
recommended practice,  the  necessary additions  and changes  to the
current  Federal  Test  Procedure  (FTP) were discussed  and generally
involved  1)  the  use of a  dilution tunnel  coupled with a constant
mass sampler and 2) the use of  filter  media  to collect the partic-
ulate  matter over both  the  cold  and hot  start  portions  of the
test.  The latter  of these two  requirements  (i.e.,  filter measure-
ments) is of special interest  here.

     The method of measuring  emitted  particulate matter with filter
media is believed to yield  accurate results.   Filter-based measure-
ments  are  currently required for  the  certification of  light-duty
diesel vehicles ,2j  and  have  also  been accepted  as  the  method  of
measurement in EPA's current  program to determine  heavy-duty diesel
emissions over the  transient  cycle._3/  One major  drawback in using
filter media  to  measure  particulate  emissions is that  the time
needed to  obtain results,  or data processing  time,  is very slow,
at least one hour  after the end of the test.^/  In addition, there
are  also the disadvantages of  being unable  to analyze  mass par-
ticulate  emissions  instantaneously as a  function of  the  mode  of
engine operation,  and  of being unable to  rapidly and effectively
monitor  the  effects  of  minor  engine  adjustment  with  the  engine
operating.  Also,  filter  measurements  are  subject to a number of
errors during  the  filter  stabilization period, such as changes in
humidity, faulty handling procedures, etc.

     The technique  of photoacoustic spectroscopy may  be an attrac-
tive  alternative to  filter  measurements  in that  it would  bring
about  instantaneous  measurement  and a short data processing time
and  thus help eliminate some  of  the problems mentioned  above.
Photoacoustic spectroscopy  involves the use  of a laser beam to heat
particles in  a cell.   If  a  rapidly modulated laser beam is used,
the  particles  heat and  cool  quickly,  which  in turn heat and cool
the surrounding air resulting  in pressure waves which  are picked up
by a spectrophone,  which is a sensitive microphone with  associated
electronics.  The  response  time  is  between  0.5  and  1.0 seconds4/_5_/
while the data processing time is  less  than  one minute.6/

     Photoacoustic  spectroscopy  is currently  being used by Ford,
General  Motors,  and possibly  others for  monitoring  light-duty
diesel  particulate emissions.   This  report  will  investigate
these  present  applications.    This  report  will  also discuss the
feasibility of this technique as an alternative to filter measure-
ments  for  certification  testing of light-duty  diesel vehicles and
heavy-duty diesel engines.

-------
II.  Present  Operation of Photoacoustic Spectroscopy

     The photoacoustical  effect can be used  to  measure  the light
absorption  of airborne particulate matter  in  a suitable  cell.
In theory,  the spectrophone  response should be proportional to the
amount of light  absorbed  which is in  turn assumed  to  be propor-
tional  to the mass concentration of  particles in the  exhaust.

     Measurement  of particulate  emissions  from diesel  exhaust
has  been  attempted with  photoacoustic spectroscopy .J5/7/8/   The
spectrophone measures  only  the  portion of particulate  which
is elemental  carbon,  as  elemental  carbon is the  only component
of the particulate which absorbs light to a high enough degree .J7/
This  presents a problem  with diesel  particulate because  it is
composed  of both  elemental  carbon  and  organic compounds  that
are  attached to the carbon.   If the  elemental carbon  to  total
mass  ratio  is known, then  the airborne  particulate concentra-
tions  can be  determined.   However,  the fraction of  attached
organic material ranges from  10 to 70 percent  of  the  total mass
particulate,^/ depending  on the type  of  engine  and  the operating
conditions.   Thus,  the  elemental carbon to  total  mass  ratio
must  be approximated so  that spectrophone  correlations  can be
made.

     Ford  and General  Motors appear  to be doing  the most research
on the  use  of photoacoustic  spectroscopy for  measuring  diesel
particulate   emissions.    At  present,  photoacoustic  spectroscopy
is used by  both Ford and GM  for  an  "instant"  readout  of partic-
ulate  emissions  so that  the  process   of  modifying  engine design
as a  function  of particulate  emissions can be  done  more rapidly
and  effectively  than that possible  with  filter  measurement.
To better understand  the  present  role of photoacoustic spectro-
scopy,  the  Ford and GM  set-up and their most  recent  studies on
monitoring of  diesel  particulate emissions  will be examined in
detail below.

     A.   Ford 5/7/i/

     Ford  uses an argon  (Ar+)  laser of  1.5 watts  with a wavelength
of 514.5 nanometers  (nm) as  the source  of  light.   The laser beam is
split  into  two beams.   One  beam serves as a  reference  and is
detected by  a photodiode (a  device which converts  a light beam into
a  usable electrical signal).   The second  beam  is directed into a
spectrophone  cavity which  intensifies  the sound  waves produced by
the  heated  particulate  in  the  cavity.   The  second beam is then
detected  by another  photodiode.   The output  signals  from the
microphone are  detected by  amplifiers  and sent  to an appropriate
instrument  (ratiometer)  to obtain the normalized spectrophone
signal.  Response time  is  approximately one second, while the data
processing time  is  less than one minute.

     Ford  estimates the cost  of their  set-up to be about $10,000
for the laser,  and  $2,500  for  the remaining equipment.10/

-------
     Ford's  latest  documented work  of photoacoustic  spectroscopy
measurements involves a study of diesel particulate  from a  1979 2.3
liter Opel  diesel which  could be run with two different rear axle
ratios (3.89 or 2.70) and with or without  exhaust  gas  recirculation
(EGR).J/   In this experiment, the vehicle  exhaust  is  diluted and
mixed in a dilution tube that is  about 30  feet  long  and 1.5  feet  in
diameter,llj with  samples being  taken for  both  spectrophone and
conventional filter  analysis.  The  results  of several  tests per-
formed  on this Opel diesel  show that the  average  spectrophone
response  (in MV)  is  proportional  to the  particulate  mass  loading
(within reasonable accuracy)  only under conditions  where the ratio
of  elemental  carbon  to organic compound is  fairly constant.

     For example, one  set of Ford's  data  shows that  when  the Opel
diesel  is  run  without  EGR,   the  correlation between  spectrophone
response  and particulate mass loading  is  accurate  within ^_5 per-
cent. YLJ   The organic  fraction in this  case is  believed to be
constant at about 55 percent  of total particulate  mass.J_3_/   Another
set of Ford data on the Opel  diesel,  this  time  with  EGR, also shows
good correlation between spectrophone response  and particulate mass
loading, again within an accuracy of about +5  percent.   Again, the
organic fraction is believed  to be constant, this time about 35-40
percent of total particulate.  However, this  latter  correlation has
a slope (change in spectrophone  response per change in particulate
mass  concentration) that  is about 20 percent  higher than the
slope  for the data  without  EGR.J_3_/  Although good correlation
exists  for  each case,  a combined plot of  these  two sets of data
yields a poor correlation between spectrophone  response and  partic-
ulate mass   loading  with an  accuracy of  roughly ^50  percent.13/
Thus, the  Ford  data  indicate that the correlation  of spectrophone
response  and particulate mass  loading is  poor  for   the  combined
results of  the Opel diesel with  and  without  EGR.  Also, these test
data  show that  the  elemental carbon  to organic  ratio 1) must  be
known for an accurate correlation between  spectrophone response and
particulate  mass  loading to  be  developed,  and  2) may vary  greatly
from  one engine (or vehicle)  design or operation to another.

     In addition to the problem of determining  the elemental carbon
to  organic  ratio,  another   potential  limitation with the photo-
acoustic spectroscopy system, as  discussed  by Ford,  is that  optical
absorption  per  particle mass may not  be  independent  of  particle
size.   For  carbon particles  it  has  been shown that the absorption
of  light  is constant per unit  of mass only  for  particles with
diameters much  smaller than  the incident wavelength  of light  (known
as  Rayleigh particles) .J_4_/    It  is   true that  a  laser beam of ex-
tremely  long wavelength could  bypass these small  particles com-
pletely,  with  no  absorption taking place.    However,  this would
occur  only  at  particulate   concentrations  much  lower than those
expected  to  be measured  from light-duty diesels,  even with  partic-
ulate emission  control  devices.   In the Ford  set-up,  the  particle
diameter may not  be  sufficiently less than the argon laser beam's
wavelength  of  514.5  nm  (0.5 micrometer)  to  avoid  this  problem.
This  problem is  also compounded by  the  fact that  the particle

-------
size  distribution  may change under  various vehicle operating
conditions.

     Ford has tried laser beams of  longer  wavelength  in  an  effort
to solve this problem  involving particle size.  For an argon  laser,
a  wavelength  of 514.5  nm  is  the longest  possible  wavelength.ll/
Lasers other  than  argon with longer wavelengths  may  be  used,  but
such  lasers  have  difficulty producing the  power  output  necessary
to  obtain  sufficient  spectrophone  response (approximately  1
watt).  For  example,  Ford tried using  a  krypton  ion laser because
of its longer wavelength,  but the  power of this laser was insuffi-
cient. _!_!_/  Ford also investigated  the  use of a carbon dioxide
laser, which happens to be a laser with a wavelength in the  infra-
red region  that is powerful  enough for spectrophone response.
However, Ford found that the response  to  the  carbon dioxide laser
had to be corrected  for gas  phase absorption by both  carbon dioxide
and water,  the  concentrations  of  which both vary considerably  in
diesel exhaust.   Also,  absorption of the  carbon dioxide  laser  may
occur  for some  organic  compounds.    Thus,  from Ford's  published
work, systems with  lasers of longer wavelength appear to have  as
many of their own problems as the  argon laser.   (See discussion on
GM set-up below.)

     Thus, while Ford uses  the photoacoustic  spectroscopy  method
as  an effective means  to  detect  the  relative effects  of  engine
design  on particulate emissions when  the organic  fraction  is
constant, they have  not demonstrated  at  the moment that this
technique can be  used  to measure absolute levels  of particulate
emissions with  the  accuracy necessary for  certification  work;  an
accuracy which  is currently  achievable with the  filter technique.

     B.   General Motors (GM) 8/15/

     According  to  GM's  latest published report,  photoacoustic
measurements are made using a dual  cell  method  in series.   Both
particulate  and exhaust  gases  enter this apparatus  at  the  second
cell.  The  particulate  is filtered  after  leaving  the  second cell,
so  that the  first  cell contains gases  only.  The  particulate
absorption  signal  can then be determined differentially.  For
diesel  particulate  measurements a  carbon  dioxide  laser of 10.6
micrometer wave length  is  directed   into  the  first  cell  and then
into  the  second cell,  operating at  a power  of  3.4 watts.  (GM  has
used  an  argon  laser  of 514.5 nm wavelength   for acetylene  smoke
measurements. )_1_5_/   With this high  modulation frequency, background
noise is neglible.   In the second cell, the infrared  absorption  due
to  particulate  plus gases (C02, H20, etc.) is measured,  while  in
the  first cell,  the  absorption of  gaseous  species  only  are mea-
sured.   The particulate absorption   is computed by  electronically
subtracting the response of  the two  photoacoustic  cells.   The
spectrophones are  placed in  each cell midway between the open ends
of  the  cells, to  provide  good coupling of  both the laser and  the
microphone to the resonant  mode.   The response time  is  about  0.5
seconds.

-------
     GM estimates  the  cost  of the whole unit  to  be  about $35,000.
16/   This breaks down  to  $12,000  for the  C02  laser,  $4,000  -
$5,000 for the optical bench, $10,000 - $12,000 for the electronics
and amplifier, and about  $4,000 for each of  the  two photoacoustic
cells.

     GM has used  this  system to measure  diesel exhaust particulate
concentrations from a 5.7  liter diesel  engine.   This engine  was
operated at various  steady-state conditions and apparently was  run
without any modifications during the test program.   Results  indi-
cate that for this  diesel engine only, the  photoacoustic  signal is
proportional  to  the diesel  particulate mass  concentration,  within
an  overall  accuracy of   +^.5 percent.   GM  believes  that  by  their
measuring of  particulate  emissions in a cell  sufficiently removed
from the combustion source,  the optical effects due to variation in
the elemental carbon  to organic  compound ratio have  been  minimized
for this  test.    However,  they  admit  that  data are  too  scattered
(as shown on a plot of particle mass concentration vs. optoacoustic
signal) to  appreciably determine that the optoacoustic signal  was
not affected by variation in particle composition.

     GM concludes  from this  study  that the  photoacoustic  effect is
a  convenient  and  sensitive  method  for  measuring mass emissions.
However,  GM  states  that  there are problems involved  with  the
photoacoustic  spectroscopy   method  which may  lead   to  inaccurate
results.   First,   as  mentioned  above,  the chemical  composition of
the particulate may vary and  affect  optical properties for a  given
mass  concentration.    Second,  changes in particle size  may  still
affect  optoacoustical calibration in  the visible light region
somewhat because  the  wavelength is not  sufficiently larger  than
particle size.    This  may be  solved by using an infrared  source
(such as  a carbon  dioxide  laser)  of sufficiently  long wavelength
which  should  give  measurements  that are  independent of particle
size variations,  if the  particles are spheres.   However, infrared
light response can be  affected  by  the  shape of a particle if  it is
not  spherical,  which occurs  frequently with diesel  particulate
emissions,  where  long chains of small  particles tend  to form.17/

     In conclusion,  GM uses photoacoustic spectroscopy to observe
the effects  of engine operating conditions on  diesel particulate
emissions.   GM  does not presently  have  data   which  compare  photo-
acoustic spectroscopy  to filter  measurements  over  the Federal Test
Procedure.   At the  moment they  do  not  intend  to use this  technique
as an alternative to conventional filter  measurements for  measuring
exact amounts of  particulate emissions,  such  as  would be required
in the Federal Test Procedure.

III. Future Applications of Photoacoustic Spectroscopy

     Photoacoustic spectroscopy has been  proven to be effective for
relating  changes   in  particulate  emissions  to  changes   in  engine
design  and operating conditions  of  light-duty  diesel  vehicles,
particularly  if  the  organic fraction of  the particulate  is  con-

-------
stant.  This effectiveness  is  due to quick response (about 0.5 to
1.0  seconds) and data processing  times  (less than  1  minute) in-
herent in the system which allow it to show variations of emissions
with  driving  conditions  such as acceleration,  steady  highway
speeds, and  idles.  However, the  photoacoustic spectroscopy system
can not be  depended  on for exact measurement of particulate  emis-
sions, as the overall  accuracy  of  the  correlation between  spectro-
phone  response  and  particulate mass  loading is estimated  to be
about _+15 percent  for the engine  tested  with no modifications by
GM,  and  about  ^_50 percent for  the Opel  diesel engine tested  with
modifications (EGR) by Ford.

     Because of  this  limited accuracy, photoacoustic spectroscopy
should not   be  considered for  certification application  at   this
time.  Also, photoacoustic spectroscopy has only been studied  using
a  few light-duty diesel  vehicles  by Ford  and  General  Motors.
Given  the  problems  already seen with this  technique  these few
studies would  not be  adequate to  show  that the accuracy of the
photoacoustical technique would not be affected to an even greater
degree by different  engines  or vehicles.    In other words, before
photoacoustic spectroscopy  could  be  considered  for certification
purposes  the technique must  correlate with the filter measurement
technique for  all  types  of  light-duty diesel vehicles  and heavy-
duty  diesel engines.   Until  data  show  otherwise, photoacoustic
spectroscopy should  not  be allowed  as an alternate certification
technique for measuring particulate emissions.

Conclusion

     Photoacoustic spectroscopy  is currently used  to  monitor the
effects of  engine  design  and  operating  conditions  on particulate
emissions.   Accurate measurements  of the  correlation  between
spectrophone response  and particulate mass  loading  have not  been
obtained  for particulate  emissions because  the  absorption of the
laser  beam used  in this  technique  seems to be  affected  by 1)
particle  size and shape and 2)  particle composition.  Thus, photo-
acoustic  spectroscopy should be rejected  at  the  moment  as a  tech-
nique  for  measuring  particulate   emissions  during  certification.
Progress  in  photoacoustic  spectroscopy should be monitored in the
future as  improvements may  justify it as an alternative  to the
filter measurement test procedure.

-------
                            References

JY   Danielson, Eugene, "Draft Recommended  Practice for Measurement
     of  Gaseous  and Particulate Emissions  from Heavy-Duty Diesel
     Engines Under  Transient  Conditions,"  SDSB,  ECTD,  EPA, April,
     1979.

2J   "Standard for  Emission  of Particulate Regulation  for Diesel-
     Fueled Light-Duty Vehicles and  Light-Duty Trucks," FR Vol. 45,
     No. 45, March 5,  1980,  pp. 14496-14525.

_3_/   Southwest  Research  Institute  Diesel  Baseline Emissions
     Summary, EPA, June 1,  1980.

47   "Optoacoustics  Measure  Diesel  Particulates,"  Automotive
     Industries,  September,  1979, p.  31.

5/   Japar,  S. M., D.  K.  Killinger,  and  J. Moore,  "The  Use of
     Photoacoustic  Spectroscopy   to  Characterize  and  Monitor
     Soot  in Combustion Processes,"  presented at Symposium of
     Lasers  in Combustion Chemistry,  ACS Meeting, Washington,
     D. C., September,  1979.

6J   Truex,  T.J.  and  J. E.  Anderson,  "Mass Monitoring of Carbon-
     aceous Aerosals with a Spectrophone,"  Atmospheric Environment,
     Vol. 13, pp. 507-509,  September 26,  1978.

Tj   Japar,  S. M.  and Ann Guneo Szkarlat,  "Measurement  of Diesel
     Vehicle Exhaust Particulate Using Photoacoustic Spectroscopy,"
     Engineering  and Research Staff,  Ford Motor  Co.

8/   Faxvog, Fred R. and David M. Roessler, "Optoacoustic Measure-
     ments  o'f Diesel  Particulate   Emissions,"   Journal of Applied
     Physics,  Vol. 50,  No.   12, December,  1979,  pp. 7880-7882.

_9_/   Japar, S.M. , and  O.K.  Killinger, "Photoacoustic and Absorption
     Spectrum  of  Airborne  Carbon Particulate Using  a  Tunable Dye
     Laser," Chemical  Physics Letters,  Vol. 66, No.  1,  Sept. 15,
     1979, pp. 207-209.

10/  Telephone  conversation   with  Steven  Japar,  Engineering  and
     Research Staff, Ford Motor Co.,  July 3, 1980.

ll/  Telephone  conversation   with  Steven  Japar,  Engineering  and
     Research Staff, Ford Motor Co.,  August  18,  1980.

12/  Unpublished   data   from  Ford Motor  Company,  Engineering  and
     Research Staff.

-------
13/  Telephone  conversation  with  Steven Japar,  Engineering  and
     Research Staff, Ford Motor Co., August 21, 1980.

14/  Faxvog, Fred  R.  and David M.  Roessler,  "Carbon Aerosol  Visi-
     bility vs.  Particle Size Distribution,"  Applied  Optics,  Vol.
     17, No. 18, August  15, 1978, pp. 2612-2616.

15/  Faxvog, F.R.  and  D.M. Roessler, "Optoacoustic  Measurement  of
     Optical Absorption in  Acetylene Smoke,"  Optical Society  of
     America, Vol. 69, No. 12, Dec. 1979, pp. 1699-1704.

16/  Telephone  conversation with Dr. Fred Faxvog,  Physics  Depart-
     ment,  General Motor Research Laboratories,  July 3,  1980.

_1_7_/  Lipkea, William L., John H.  Johnson,  and Carl T. Vick,  "The
     Physical Chemical  Character  of Diesel  Particulate Emissions -
     Measurement  Techniques  and  Fundamental  Considerations,"  SAE
     780108, 1978.

-------