EPA-AA-SDSB-85-4
Technical Report
Durability of Oxygen Sensors
By
Lisa Snapp
March 1985
NOTICE
Technical Reports do not necessarily represent final SPA
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
developments 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 Sources
Office of Air and Radiation
U. S. Environmental Protection Agency
-------
I. Introduction and Summary of Comments
As emission requirements continue to tighten, new emission
control technology is constantly being developed. This
technology must result in equipment which is not only initially
effective, but also durable. One important aspect of
durability is the amount of maintenance required. Despite auto
manufacturers' maintenance schedules for warranty requirements,
many consumers do not maintain their cars at recommended
mileage intervals. This is especially true if no problem is
apparent to the driver who fails to do the maintenance. The
longer the affected equipment lasts without maintenance, then,
the better the emission control will be in use.
The oxygen sensor is such a piece of equipment. Some
manufacturers suggest replacement at 30,000 (30K) miles, while
most suggest 50K for model year 1984 vehicles, [i] if the life
of the sensors could be extended to higher mileage, emissions
could be significantly reduced. In light of this, oxygen
sensors were included in the recent EPA proposal to extend
maintenance intervals to lOOK for emission-related
components.[2]
In response to EPA' s proposal, most manufacturers of
gasoline-fueled engines submitted comments which opposed the
extended interval. These manufacturers argued that the
maintenance interval should not be extended beyond the 30-50K
range. Their position was based on the claim that no data
existed to support the feasibility of lOOK durability for
oxygen sensors. Manufacturers did not present factual
information indicating a lack of high-mileage durability for
oxygen sensors. Rather, they indicated that it was unknown
whether or not a lOOK interval was feasible. It is the purpose
of this report to discuss the technological feasibility of such
an extension for the oxygen sensor and evaluate any data
available on high-mileage performance of oxygen sensors.
II. Background
Oxygen sensors have been widely used since model year 1981
in conjunction with three-way catalysts (TWC). The sensors
give feed-back control , of oxygen partial pressure in the
exhaust to an electronic control module (ECM), in order to keep
the. air/fuel (A/F) ratio in a narrow range near stoichiometry.
Figure 1 shows this range, which gives maximum conversion
efficiency of HC, CO and NOx emissions, as well as high fuel
economy.[3] Two types of sensors have so far been developed:
a galvanic zirconia sensor and a resistive titania sensor.[4]
The zirconia sensor has virtually the entire market, but this
is probably due to its timelier development rather than any
inherent technical advantage. Some manufacturers are looking
into titania sensors for 1986 or later model years.
-------
Figure 1
PRINCIPLE OF OPERATION 3-WAY CATALYST
CO
100
80
60
CONVERSION
EFFICIENCY
40
20
13:1
14:1 15:1
AIR-FUEL RATIO
16:1
-------
-3-
The popular zirconia sensor is shaped like a thimble of
zirconia ceramic, coated with a noble metal. The open end of
the thimble points out of the exhaust flow, as can be seen in
Figure 2. [3] The inner, reference surface works like an anode,
ionizing the oxygen of the atmosphere. The outer surface acts
as a cathode, combining carbon monoxide from the exhaust with
oxygen ions. The relative rates of the two reactions can be
combined in the Nernst equation to determine the voltage
produced:
E = e" in -Po
VThere :
E = the potential difference,
k = the Boltzmann constant,
T = temperature in degrees kelvin,
e = the electronic charge,
PC>2(atm) = the atmospheric oxygen partial pressure, and
PO2(exh) = the exhaust oxygen partial pressure. [3]
A lean A/F ratio, with low carbon monoxide and relatively
high oxygen partial pressure in the exhaust, gives a low
voltage reading. A rich A/F ratio, conversely, gives a high
voltage reading. The voltage change indicating a rich/lean
change occurs as a step near stoichiometry , as in Figure 3. [3]
This voltage information is fed to the ECM, which alters the
air/fuel ratio in order to maintain a value near stoichiometry.
Thus, a switching action from rich to lean and back constantly
occurs . [ 3]
The same switching action takes place with the titania
sensor, but it is a semiconductor rather than an el'ectrolyt ic
material. Oxygen vacancies in the crystal lattice form with
rich A/F ratios, and are replaced when the mixture is lean.
During rich operation/ electrons are donated into the
conduction band, and resistance is low compared to that in the
lean state, which has few donor electrons. A voltage divider
network utilizes these extreme resistances, as well as the
ideal, which is midway between the logarithmic lean and rich
resistance values. Voltage output is then similar to that of
the zirconia sensor: low for lean A/F ratios and high for
rich. The same switching response maintaining A/F ratio near
stoichiometry occurs, controlled by the ECM. [4]
-------
.-4-
Figure 2
SCHEMATIC - ZIRCONIA SENSOR
ENGINE
EXHAUST
STREAM
OUTPUT
VOLTAGE, V
02 AT HIGH PRESSURE
02 + *»£—-20 »
\
POROUS Pt aECTRODE
(ANODE)
STABILIZED Zr02
aECTROLYTE
(CERAMIC TUBD
POROUS Pt aECTRODE
(CATHODE)
02 AT LOW PRESSURE
NERNST EQUATION: V - (RTAF) IM (?<% AiR/p02
-------
Figure 3
ZIRCONIA EXHAUST GAS OXYGEN SENSOR (AC)
CHARACTERISTIC CURVE
700
VOLTAGE
MV 500
400
300
200
100
13:1
LEAN
14:1 15:1
AIR-FUEL RATIO
U1
16:1
-------
-5-
Upon failure of the oxygen sensor, the ECM defaults to a
preselected condition. In most cases, the default is to full
rich operation. This strategy ensures driveability of the
car. However, the driver cannot tell that anything is
malfunctioning; meanwhile, emissions increase greatly. Warning
lights are installed in some models to warn the driver when the
oxygen sensor has failed or when it is time for its scheduled
maintenance. However, some drivers will not heed the lights,
and will thus allow high emissions to continue. Some models
default to stoichiometry, which is a preferred default from the
emissions standpoint. This cannot, however, keep emissions as
low as can the feedback loop with the sensor. It is also
possible to set the default to lean, but as this will give the
car only limp home capability, manufacturers disfavor it.
Therefore, in order to keep HC, CO, and NOx emissions at a
minimum, it is imperative that the oxygen sensor remain
operative. Early sensors were predicted to be good at least to
15K miles. As will be explained below, current testing shows
that sensors are good for at least 50K miles and perhaps far
beyond that point. However, the new sensors that are beginning
to reach the market, particularly heated and/or titania
sensors, are being developed for reasons other than greater
durability. For instance, the new heated sensors allow
closed-loop operation within a few seconds of engine start-up;
non-heated require open-loop until the engine temperature
climbs high enough for quick sensor response. Titania sensors
seem to have quicker response times than the currently used
zirconia. Vendors indicate that they also may be substantially
cheaper, although auto manufacturers express doubts.
Development in terms of extended durability could easily take a
back seat to such improved performance and cost issues.
Satisfactory durability from the point of view of the auto
manufacturers and their warranty requirements has already been
reached. Whether this lifetime can be extended has not
previously been of particular interest. It is possible that
this durability can be extended to as much as 100K miles.
Such an extension, of durability of oxygen sensors is
important for manufacturers from a competitive point of view.
The deteriorating sensor with the default to rich condition
found on most models results in a significant loss of fuel
economy. On the one hand, this may encourage owners to seek
vehicle maintenance, thus effecting a replacement of the faulty
sensor. On the other hand, it may decrease owner satisfaction
with the vehicle, due either to the expense and trouble of
replacing the sensor or the decreased fuel economy for those
who do not seek maintenance. Thus, extended durability should
also be preferable in order to maintain consumer satisfaction.
-------
-7-
III. Sensor Failures
The causes of sensor failure have been investigated by
manufacturers and vendors. One extensive study was done by the
Robert Bosch GmbH, which investigated control efficiency of
sensors under a variety of field operating conditions.[5]
According to.this and other sources, there are several causes
of sensor failure: extremely high or low temperatures, large
temperature gradients, leaded fuel, physical damage, and ECM
malfunctions. Occurrence of these problems can be reduced, to
some extent, with proper control and maintenance of the sensor
and engine.
Erosion of the coating apparently takes place at very high
operating temperatures (900°C at the sensor tip) over long
periods of time. The outer electrode decomposes, and the
protective coating flakes off. On the other hand, deposits of
oil accumulate at low temperatures (500°C) and are then glazed
on at intermediate temperatures (650°C). These deposits
consist of phosphorous, calcium, zinc and lead; They can be
reduced by controlling oil type and consumption.
Contamination of the sensor by leaded fuel is also a
problem: lead contamination at low temperatures, followed by
high temperature operation, creates lead deposits which cannot
be removed. Lead contamination at high temperatures can be
cleaned up with the subsequent use of lead-free fuel.
Therefore, low operating temperatures should be avoided in
order to reduce both oil deposits and lead contamination.
In order to allow higher temperature operation, high
temperature ceramics have been developed. The temperature
range to which the sensor is exposed can also be controlled by
choice of placement in the exhaust manifold. The use of high
temperature ceramics then allows placement closer to the
engine, avoiding the lower temperatures which decrease
lifetimes due to deposits. These ceramics can also better
withstand cracking due to extreme temperature gradients.
Cracking of oxygen sensors due to external forces is
difficult to prevent, as such forces are unpredictable.
Occurrences of this type of physical damage are infrequent,
however. Also infrequent, but serious, is malfunctioning of
the ECM so as to damage the sensor. This malfunction involves
a current overload to the sensor. The output voltage of the
sensor under rich conditions then sinks to a low level. The
voltage step is reduced to below the threshold level read by
the ECM as a switch and differences between rich and lean
operation are thus not registered. Of course, the ECM can also
malfunction on its own, but this is not considered an oxygen
sensor failure.
-------
-3-
IV. Warranty Practices
Current warranty practices for Light duty vehicles include
minimum maintenance intervals of 30K and 50K miles. The 30K
requirement appears to be very much on the safe side,
especially in light of comments by Toyota: [63 "We are nor
opposed to the extension of the minimum maintenance interval to
50,000 miles. We have already adopted maintenance-free type
02 sensor for useful life for our 1981 models." Toyota
subsequently did increase their interval to 50K.
The percent of light duty vehicle families equipped with
TWC which are certified with 50K maintenance schedules has been
steadily increasing. For model year 1980, over 90 percent had
30K schedules; the remainder had 50K. For model year 1984, 67
percent of the light duty engine families with TWC have 50K
maintenance schedules. This includes all of the domestic
manufacturers, as well as some foreign automakers. It appears
that there are no particular problems with a 50K warranty on
oxygen sensors. This conclusion is confirmed by the test data
discussed below.
V. Vendor/Manufacturer Data
Oxygen sensor vendors and auto manufacturers have done a
substantial amount of testing on sensors. This testing has
typically addressed aging characteristics only up to 50K in
order to meet warranty requirements. Up to this mileage, the
tests point to acceptable sensor performance, particularly in
the most recent designs.
Early results did not promise much in the way of
durability, as evidenced in a 1977 SAE paper from Bosch.[7] "A
sensor lifetime of more than 25,000 km can be predicted from
the results of the sensor tests . . .;" about 15,000 miles were
the extent of the reliable lifetimes at that date. As use of
the sensor has grown, however, assurance of its lifetime has
also grown. Test vehicles have had a chance to accumulate
mileage, and numerous dynamometer tests have been conducted,
all indicating longer lifetimes. Sensor design itself has
developed, also adding to expected durability. Current
projections run to 50K and beyond.
The main thrust of durability testing has been on the
dynamometer. This is due to its ease and controllability in
comparison to in-use vehicle mileage accumulation. Early
(1979) testing by Bendix[8] shows a gradual deterioration of
sensor characteristics, particularly internal resistance, over
dynamometer testing up to about 40K miles. This degradation,
however, does not appear to be serious. Rich voltage stays
-------
well above the threshold of 600mV, and response times do not
drastically increase or decrease. More recent publications,
particularly a customer product information booklet by
-------
-10-
excellent durability when the output voltage switching
threshold was chosen to be 350 mV: "... sensor output in this
range (is) extremely stable over the life of the- sensor."
Other in-use vehicle testing was done by C?4.ri3] Four vehicles
were run on a 50K mile certification durability schedule. Two
exceeded emission standards at 10K. Various components were
replaced, including oxygen sensors, despite -the fact that they
were still functioning. The other two cars had maintained
allowable emission levels to 50K. All of these cars had system
maintenance performed as necessary, since they were development
vehicles. Conclusions on the entire emission control system
are therefore difficult to draw. The two cars with original
oxygen sensors, however, show acceptable performance for 50K,
and at that point are still well within compliance. Road
testing, therefore, verifies the results of dynamometer
testing: oxygen sensors generally have lifetimes of at least
50K miles. In fact, there was no indication that any of these
in-use sensors were even approaching failure at this level of
mileage accumulation.
For various reasons, durability data beyond 50K is scarce
from these sources. Manufacturers have not had warranties
beyond 50K miles; therefore, they have no information about the
replacement of sensors after this point. Because of the
manufacturers' lack of interest in durability beyond 50K miles,
vendors have not emphasized high mileage lab testing. With the
extension of useful life for LDT1 s, manufacturers have begun
looking at durability beyond 50K miles. These programs have
been started too recently, however, to have yet yielded any
results.
VI. EPA Data
The EPA, however, has some in-use emissions data available
which provides information on oxygen sensor failure rates above
50K. Testing involves model year 1981, 82 and 83 cars in EPA
emission factors testing programs. These vehicles are
recruited from their owners and emission tested. The sample
used to generate this data includes 393 cars equipped with
oxygen sensors. The vehicle sample encompasses a variety of
domestic and foreign manufacturers and a range of .sizes and
accumulated mileages.
In this testing program, gross emitters receive a system
performance check in order to identify reasons for excess
emissions. Engine components with suspected problems are
individually tested; tune-up, repair or replacement is
performed as necessary. The test is then rerun. With this
sequence of events, it is likely that all grossly failed
components are noticed and reported. However, gross emitters
-------
-11-
often have a variety of low performance components which
contribute to the high emission level. Such generally poor
performance is due to aging, misfueling, lack of maintenance,
hostile environmental conditions, and/or abuse; these can
affect the entire vehicle, not just one component. Hence, it
is difficult to determine the emissions impact caused by the
failure of any one component, such as the oxygen sensor.
It is also important to note that oxygen sensors are not
strictly pass/fail devices. Some of the most common failure
modes result in a gradual deterioration of performance.
Sensors with such deterioration may still perform well enough
to allow the car to pass emissions testing, while acting well
below peak performance, and may not provoke a default response
from the ECM. Such sensors would not be reported in test
results so long as the vehicle passes.
The sensors which are reported in the following results
were determined to be bad by a "fail" reading on an oxygen
sensor tester. This tester can check the sensor operational
characteristics for idle and off-idle modes, and registers a
"fail" when the voltage level has dropped below the threshold
which the ECM reads as a rich/lean switch. This tester is used
on gross emitters in cases where the sensor is suspected,
either due to engine operational characteristics or to a
generally malmaintained engine. This requires some judgment
and may not result in a perfect reporting of poor sensors, but
probably reports all grossly failed sensors due to their
crucial role in emission levels. These results, then, report
grossly failed sensors in vehicles with high emission levels;
these vehicles may also have other engine problems.
Results of this testing show that most in-use oxygen
sensors continue functioning to very high mileages. This
conclusion is based upon the assumption that the sensors in
place at the time of testing are the originals; that is, that
the owners haven't had the sensors replaced prior to testing.
Indications from various EPA recall and surveillance programs
and owner surveys are that this assumption is largely correct.
When owners were questioned, most said that they ignored
maintenance intervals for oxygen sensors, even when such
intervals were indicated by flags or lights; in most cases for
vehicles with such indicators, the flags or lights were reset
without replacement of the sensors, despite the fact that
replacement may have been necessary for the warranty. Owners
cited the high cost of replacement and no noticeable loss in
driveability or fuel economy as reasons for not following
maintenance schedules. For those vehicles which also had
warning lights indicating malfunctions rather than just
maintenance intervals, most owners again responded that
-------
-12-
maintenance was not performed, particularly if the light went
off again before the owner could take any action. Based on
these owner responses, EPA has good reason to believe that
oxygen sensors are generally not replaced, and therefore that
those in the testing program which generated the following data
are most likely original sensors.
As can be seen in Table 1, failures accumulate with
mileage at a slow rate, reaching an overall value of 3.3
percent at the highest mileages, up to 100K. This value
represents the total failure rate for all cars in the sample.
When this data is divided into pre- and post-SDK ranges,
failure rates are 1.4 percent and 9 percent, respectively. So,
even high mileage sensors are performing satisfactorily for
over 90 percent of the vehicles.
The 9 percent failure rate is fairly constant throughout
the post 50K range, although there is some indication that it
may have increased in the 80K-100K mileage group. However, the
sample size in that group (8 vehicles) is too small for any
final conclusion to be drawn. The failure rate for those
vehicles over 70K is 9 percent.
For those cars with failed sensors, emissions far exceeded
standards. The HC and CO standards were consistently exceeded:
the HC by all 13 and the CO by 12 of the vehicles with failed
sensors, at an average emission rate of three times the
standard. NOx standards were exceeded by 5 of the 13 vehicles.
When vehicles received restorative maintenance, HC and CO
emissions dropped considerably. A sample of nine gross
emitters received maintenance which included the replacement of
failed oxygen sensors. The HC levels dropped from four times
the standard to less than twice the standard, while CO levels
fell from greater than seven to one and a half times the
standard. NOx emissions also showed a slight decrease, from
one and a half times the standard to just slightly over it.
Because of the limited sample size, the values may not be
representative, but the trend is clearly toward a much more
acceptable level of emissions.
However, it must be noted that these reductions are not
due entirely to oxygen sensor replacement. All nine vehicles
received other maintenance, including the replacement of air
filters. Several also had carburetors replaced or timing
adjusted. These and various other repairs all contributed to
the reduction in emissions. This concurs with the idea that an
oxygen sensor failure and the corresponding elevation in
emission levels usually occurs not as a cataclysmic event but
as a gradual deterioration. A deterioration also occurs in
-------
-13-
Table 1
Total Oxygen Sensor Failures
Total Total Total Total
Mileage Cars Failures Failure Rate
Up to 10K 24 1 4.2*
20K 139 1 0.7
30K 2L3 1 0.5
40K 250 2 0.8
50K 293 4 1.4
60K 327 8 2.4
70K 360 10 2.8
80K 385 11 2.9
LOOK 393 13 3.3
Unreliable due to small sample
-------
-14-
other engine components, which together affect emission
reduction efficiency. It is this in-use deterioration that
should be slowed with improved component design in order to
achieve extended lifetimes and reduce emissions.
VII. Conclusions
Most manufacturers (67%) currently specify maintenance for
oxygen sensors at 50K and durability to this mileage is well
supported by EPA data, manufacturer comments, and technical
literature. According to in-use test data, the majority of
oxygen sensors currently in use perform satisfactorily to even
higher mileages, in the range of 80K. This is despite the fact
that sensor manufacturers have not yet expended many resources
to extend durability beyond current maintenance practices. The
data available at the present time are inconclusive about
sensor performance beyond 80K, but seem to indicate an
increasing risk of failure. Therefore, it is appropriate to
revise the 100K allowable maintenance interval originally
proposed for oxygen sensors downward to 80K.
-------
1. "Public Hearing to Consider Adoption of Regulations for
Malfunction and Diagnostic Systems and Administration Extending
the Maintenance Interval for Oxygen Sensors for 1988 Model Year
Gasoline Powered Vehicles," California Air Resources Board,
report scheduled for consideration April 25, 1985.
2. "Control of Air Pollution From New Motor Vehicles and Mew
Motor Vehicle Engines: Gaseous Emission Regulations for 1987
and Later Model Year Light-Duty Vehicles, Light-Duty Trucks,
and Heavy-Duty Engines; Particulate Emission Regulations for
1987 and Later Model Year Heavy-Duty Diesel Engines; Proposed
Rule" (49 FR 40286, 10/15/84).
3. Patterson, D. J. and N. A. Henein, "Economy and Emission
Control in Combustion Engines," Engineering Summer Conferences,
University of Michigan College of Engineering, June 1984.
4. Pfeifer, J. L. , T. A. Libsch, and H. P. Wertheimer,
"Heated Thick-Film Titania Exhaust Gas Oxygen Sensors," SAE
Paper No. 840142, International Congress and Exposition,
Detroit Michigan, February 27 - March 2, 1984.
5. Gruber, H. U. and H. M. Wiedenmann, "Three Years Field
Experience with the Lambda-Sensor in Automotive Control
Systems, " SAE Paper No. 800017, Congress and Exposition, Cobo
Hall, Detroit, February 25 - 29, 1980.
6. "Toyota's Comments on the Proposed Changes to Adjustable
Parameter and Allowable Maintenance Requirement," California
Air Resources Board, August 24, 1982.
7. Eckehardt Hamann, Hansjorg Manger, and Leo Steinke,
"Lambda-Sensor with Y203~Stabilized Zr02~Ceramic for
Application in Automotive Emission Control," SAE Paper No.
770401, International Automotive Engineering Congress and
Exposition, Cobo Hall, Detroit, Michigan, February 28 - March
4, 1977.
8. Young, C. T.__ and J.D. Bode, "Characteristics of
Type Oxygen Sensors for Automotive Applications," SAE Paper No.
790143, SAE Automotive Engineering Congress and Exposition,
Detroit, Michigan, February 26 -March 2, 1979.
9. .Exhaust Oxygen Sensor Customer Product Information, AC
Spark Plug and GM Research Laboratories, Revised 6/83.
10. Howarth, D. C. and A.L. Micheli, "A Simple Titania Thick
Film Exhaust Gas Oxygen Sensor," SAE Paper No. 840140,
International Congress and Exposition, Detroit, Michigan,
February 27 - March 2, 1984.
-------
Li. Esper, M. J., E.M. Logothetis, and J.C. Chu, "Titania
Exhaust Gas Sensor for Automotive Applications," SAE Paper No.
790140, SAE International Engineering Congress and Exposition,
Detroit, Michigan, February 26 -March 2, 1979.
12. Reddy, J.N., "Application of Automotive Sensors to Engine
Control," .SAE Paper No. 780291, Congress and Exposition, Cobo
Hall, Detroit, MI, February 1978.
13. Zemke, B. E. and J.J. Gumbleton, "General Motors Progress
Towards the Federal Research Objective Emission Levels," SAE
Paper No. 800398, Congress and Exposition, Cobo Hall, Detroit,
Michigan, February 25 - 29, 1980.
------- |