Measuring optical filters

Application Note

Author Introduction
Don Anderson and Michelle Bandpass filters are used to isolate a narrow region of the optical spectrum.
Archard The filter‟s operation is based on interferometric principles and hence is
Agilent Technologies, Inc. angularly sensitive. Most commercial bandpass filters have full width, half
Mulgrave, Victoria 3170, max (FWHM) bandwidths of approximately 10 nm. Figure 1 illustrates some
Australia of the characteristics of interest when studying bandpass filters
Note: Applications Development Language programs, suitable for use when
measuring filters, are available free of charge from your local Agilent office.

Figure 1. The FWHM is measured as the band width at half the maximum peak height

Measuring the optical characteristics of bandpass filters requires special

techniques. In this study, techniques to negate the angular and polarization
sensitivity of the filters were examined. The blocking capabilities of
bandpass filters were measured and various instrument corrections were
performed on the data.
Experimental spectral band width to be studied. Figure 2 illustrates
the effects of changing the spectral band width.
Instrumentation
 Cary 5 UV-Vis-NIR spectrophotometer
 Solid Sample holders
 2 mm beam masks
 Glan Taylor polarizer
 Depolarizer
Sample types
Figure 2. The spectral band width should be set to less than 1/10th of the
Two different types of bandpass filters were examined. expected natural bandwidth of the filter. Setting a wider SBW will result in
One was a very narrow filter, nominally 0.3 nm centered peak suppression and broadening

at 630 nm, the other, a comparatively broad filter with a Signal Averaging Time
FWHM of 50 nm centered at 260 nm. Both the angular
sensitivity and the effects of various instrument The Signal Averaging Time (SAT) is the time that the
parameters were studied. instrument spends at each wavelength averaging the
signal. This affects both the photometric noise
A sharp cut filter was also used, to study the effects of associated with the measurement as well as the time it
polarization. A „sharp cut‟ filter has low transmission in takes to collect the data. As the filters were being
one region and a very sharp change to high scanned over a very short wavelength range, the scan
transmission in another. They are often called time was not an issue. The SAT was set to 0.1 second
„cut-on‟/‟cut-off‟ filters or „hot‟/‟cold‟ filters. to provide acceptable photometric noise levels.
Optimum instrument parameters Figures 3a and 3b illustrate the effect of changing the
Signal Averaging Time.
Spectral Band Width
As is the usual practice with UV-VIS
spectrophotometry, the Spectral Band Width (SBW)
was set to less than one 10th of the expected
bandwidth. This ensures the best resolution of the data.
When measuring the narrow filter the SBW was set to
0.03 nm (0.04 measured). A wider SBW of 2 nm was
suitable for the 50 nm bandpass filter.
Figure 3a. The Signal Averaging Time should be selected to produce
The Cary 5 has an NIR „Fixed SBW‟ mode that
acceptable noise levels. This scan was performed with 0.1 s SAT
facilitates the measurement of NIR filters. The normal
method of operation in the NIR is to keep the „Energy
level‟ (the voltage on the detector) constant while
varying the spectral band width. This allows the
complete dynamic range of the instrument to be used.
The „Fixed SBW‟ mode of operation reverses this and
keeps the spectral band width constant while varying
the Energy level. This allowed the effects of varying the
Figure 3b. The same sample, with the Signal Averaging Time set to 0.033 s

2
Data interval The Cary 5 in the standard configuration has beam
It is recommended that the data interval be set to angles that are within ±3.2° horizontally and ±7.6°
ensure that at least three data points are collected vertically. Reducing the slit height limits the vertical
across each spectral band width. With the SBW set to angles to ±5.0.
0.03 nm, the Data interval was set to 0.01 nm. This
For the critical applications of very narrow band filters,
provided 200 data points in the 631 nm to 629 nm scan.
(FWHM of less than 1 nm) suitable masking can be
The interval was increased to 1 nm when the 50 nm
introduced to limit the solid angle further. Inserting a
filter was measured. The effect is shown in Figure 4.
5 mm aperture at the vertical grating image (50 mm
after the sample compartment centre) limited the
vertical angles to ±2.8°. Placing a 2 mm aperture at the
vertical slit image (50 mm prior to the sample
compartment centre) and at the vertical grating image
(50 mm after the sample compartment center) limited
the horizontal and vertical beam angles to ±0.6°. In this
case suitable Rear Beam Attenuation was used to
preserve the dynamic range of the instrument. The
Figure [Link] Data interval should be set to less than 1/3 of the SBW. attenuation value was chosen so that the transmittance
Setting too large an interval will result in poor resolution.
did not exceed 100 %T when a baseline was collected.
Angle of incidence In this case about 1 Absorbance (optical density)
attenuation was adequate. Care was also taken in the
The centre wavelength of a bandpass filter changes
mounting of the filter to ensure that it was not tilted.
with the angle of the incident radiation. As the angle
increases, the peak moves to a shorter wavelength. The first set (Figures 5 to 9) of overlaid traces below
This relationship is given by; show the effect of various configurations on a very
narrow (nominally 0.3 nm) filter. Figure 5 is from the
= 0 (neff
2- sin2 )/neff
instrument in the standard configuration, and shows
Where: some asymmetry. Figure 6 is in the reduced slit height
configuration, note the improved symmetry. Figure 7 is
= Peak wavelength when ≠0
for reduced slit height and a 5 mm aperture placed at
neff = effective refractive index. This is typically in the the vertical grating image. Figure 8 is with 2 mm
range 1.3 to 3 (refer to manufacturer‟s details) apertures at the vertical slit and vertical grating images.
Figure 9 shows all the scans overlaid for comparison
= Angle of incidence
purposes.
=Peak wavelength when ≠0
0
As the solid angle is reduced, the measured bandwidth
In any optical system the beam must have a solid angle of the filter decreases (refer to Table 1), the peak
for there to be any flux transmitted. The result of this is wavelength gradually moves to a longer wavelength,
that the measured profile will be broadened. This and the peak transmission increases. The increase in
broadening becomes more significant with narrow peak transmission is due to the better overlap of
filters; particularly when the spread in peak contributions from different angles.
wavelengths due to the finite solid angle becomes
comparable with the filter bandwidth.

3
Figure 5. The 0.3 nm Bandpass filter measured with the standard Figure 9. The 0.3 nm bandpass filter - all configurations
configuration of the Cary 5 Table 1. The measured bandwidth for each configuration studied

Configuration Bandwidth (nm)

Standard 0.462
Reduced height slit 0.397
Reduced height slit and 5 mm mask 0.361
2 mm masks 0.360

The second set (figures 10 to 14) of traces are for the

same configurations on the relatively broad (50nm)
filter. Clearly this filter is not as sensitive to the solid
Figure 6. The 0.3 nm Bandpass filter measured with reduced slit height
angle effects. There was no change in the measured
FWHM of 46.93 nm.

Figure 7. The 0.3 nm bandpass filter measured with reduced slit height and 5
mm apertures
Figure 10. The 50 nm Bandpass filter measured with the standard
configuration of the Cary 5

Figure 8. The 0.3 nm bandpass filter measured with 2 mm apertures

Figure 11. The 50 nm Bandpass filter measured with reduced slit height

4
 plot, the scan should be plotted on a logarithmic scale -
Absorbance (Optical Density) is the norm.
The instrument stray light level directly effects the
measurement of out-of-band leakage. The amount of
stray light present determines the maximum
absorbance that can be measured. Many applications
require blocking levels in the range of 10-6 T or
6 absorbance units.
Figure 12. The 50 nm bandpass filter measured with 5 mm apertures The two scans shown in figures 15 and 16 are of the
two bandpass filters. The narrow bandwidth filter has
blocking of approximately 5 Abs (0.001 % T). The
broader filter had poorer blocking capabilities. As both
filters were to be used in applications where the
detector is a photomultiplier, the upper wavelength was
set to 800 nm. The filters could have been scanned up
to 3300 nm if they were to be used in photodiode/NIR
detector applications.
Figure 13. The 50 nm bandpass filter measured with 2 mm apertures

Figure [Link] blocking capability of the 0.3 nm bandpass filter

Figure 14. The 50 nm bandpass filter - all configurations

Blocking
The „Blocking‟ capability of a filter is a measure of the
transmission outside the bandpass region (out-of-band
leakage). This should be very low, so that the only light
that the filter transmits is that of the required
wavelengths.
In general a second scan will be performed over a wide Figure [Link] blocking capability of the 50 nm bandpass filter

wavelength range to assess the blocking capabilities.

Often the transmission values will be very small. To
highlight small values and compress peak
transmittance values so that all points remain on the

5
Polarization these The Cary instruments typically have a zero
transmission level of approximately 0.000005 T.
The optical characteristics of many filters are affected
by the plane polarization of the incident light. Figure 18 The American Society for Testing and Materials
illustrates the effects that plane polarized light can (ASTM) have published1 a method for the correction of
have on filter measurements. “Solar absorptance, reflectance and transmittance of
materials using integrating spheres” which can also be
The „step‟ at the detector change (800 nm) is due to the
applied to transmission measurements performed
change in polarization of the incident light as the
without an integrating sphere. The correction is in the
grating changes from the UV-Vis to the NIR. To remove
form:
this effect a depolarizer was placed before the filter.
The depolarizer changed the plane polarized light T(λ) = (Sλ - Zλ)/(100λ - Zλ)
exiting the monochromator into randomly polarized Where:
light, before it reached the sample. Figure 17 shows the
data collected with the depolarizer in place. Sλ = the signal recorded with the sample in place
Zλ = the zero line reading, with the sample beam
blocked
100λ = the 100% line reading with no sample
To perform these corrections, the following
measurements were performed over the wavelength
range of interest:
1. A scan with nothing in the sample beam
2. A scan with the sample beam blocked with a piece
Figure 17. Measuring a sharp-cut filter with a depolarizer prior to the filter of black painted sheet metal
eliminates any change in %T at the grating/detector change point
3. A scan with the filter in place
Each scan was performed without baseline correction
as the ASTM correction is inherently a baseline
correction. In a batch measurement situation the first
two scans would be only performed at the start of the
measurement session.
The correction was then performed using an ADL
program written for the purpose. Figure 19 shows the
difference between the uncorrected and the corrected
Figure 18. Measuring a filter which is sensitive to the plane polarization of
the incident light results in a different response when the grating is changed
data. Most of the error was introduced by the zero line,
(800 nm) which caused the uncorrected data to have higher
Data corrections transmission than actual.

When measuring very low transmission levels it is This correction can be easily performed after each scan
important to consider the intrinsic electronic and optical and the corrected data stored for later use.
errors associated with the measurement and correct for

6
Figure 19. ASTM method E903 was used to correct the data for the 100%
and zero line errors

Discussion
The measurement of optical filters requires
consideration of the optical effects of the incident light.
This study found that several techniques can be
employed to improve the accuracy of such
measurements, including;
 Beam masking to reduce the solid angle
 Selection of optimum instrument parameters
 Instrument considerations when determining the
blocking capabilities of bandpass filters
 The use of depolarizers
 The correction of data for instrumental errors
The use of a high performance spectrophotometer with
flexible parameter settings and excellent optical
performance was also highlighted.
References
1. ASTM Standards on Color and Appearance
measurement, American Society for Testing and
Materials, Philadelphia, 1987
.

7
[Link]/chem
© Agilent Technologies, Inc., 1992, 2011
Published March, 2011
Publication Number SI-A-1119