Everything You Always Wanted to Know
About Narrowband Filters, But Were
Afraid to Ask
Don Goldman
Astrodon Imaging
Advanced Imaging Conference
October, 2007
Steve Mandel (what it looks like)
Re-decorating Steve’s and Sandy
Barnes’ SRO Observatory (we
use it a lot!)
If that doesn’t work, drastic action
may be needed. Keith Quatrocchi,
M.D. helping Mike Mayda.
Narrowband Filter (NB)Topics
•
•
•
•
Benefits
Terminology
Definition
Bandwidth
– Contrast, light pollution, moon, angular velocity, NII/H-α
• Manufacturing variability
– FWHM myth
• Spectral (angular) blue shift
– Faster optics
– One filter works for all?
• Practical implications
Benefits of NB Imaging
Enhance contrast
Use in light polluted
areas
Emphasize different
structures H-a, OIII, SII..
Extend Imaging Time
when the moon is up
Contrast Improvement
Crescent Nebula, Wolf-Rayet
Bubble in Cygnus, NGC 6888
Wide Red
(Orange)
Narrow Red
H-α 9 nm
H-α 6 nm
3x5min exposures, RCOS 12.5” Ritchey-Crétien, SBIG
STL11000XM CCD, Bisque Paramount ME
H-α 4 nm
Enhanced Structure at Hi-Res
Crescent Nebula in Cygnus, NGC 6888
Conventional RRGB
(courtesy Rob Gendler)
Red = H-α
Blue=Green=OIII
(Don Goldman)
Narrowband Terminology
• Wavelength
• Transmission Plot
• OD, or Blocking Plot
• S/N or IB, Integrated Blocking
Filters and Wavelength
Photometric Red
Imaging Red
OIII
Photometric Red Imaging Red
(UVBRI)
(RGB)
H-alpha
H-alpha
SII
H-alpha
Narrowband Terminology
Photometric Red
Imaging Red
OIII
H-alpha
SII
H-a Filter Transmission Plot
100
Out-of-Band
In-Band
Out-of-Band
% Transmittance (I/Io)
90
80
70
Peak Transmittance, Tpk
60
50
40
CWL, λ0 Center
Wavelength
30
20
FWHM = Full Width at
Half Maximum
Transmittance
(aka BW = Bandwidth)
10
0
640
645
650
655
Wavelength, nm
660
665
670
Out-of-Band Blocking
• O.D. = Optical Density = Absorbance
• OD = -log(Transmittance) = -log (%T/100)
• Larger OD means less out-of band light leakage
OD
Transmittance
%T
0.3
0.50
50
1.0
0.10
10
2.0
0.01
1
3.0
0.001
0.1
4.0
0.0001
0.01
5.0
0.00001
0.001
• An OD 4 filter blocks 99.99% of the light outside the bandpass
• Must use OD scan, not %T scan
Blocking Scan
H-a Filter Blocking
8
7
O.D
6
5
4
3
QE KAF16803E
2
1
0
350
450
550
650
750
850
950
1050
Wavelength, nm
• Blocking covers the sensitivity range of the CCD
• Light leakage outside of the bandpass will be detected
• Comparing significant light coming through the narrow
bandpass to little light coming in over a huge wavelength
range outside the bandpass
• This filter could be specified as > OD4 (good)
• Quantified by S/N or IB (Integrated Blocking)
What is a Narrowband Filter?
• Use photometric filters as a
guide
Bessell Approximation to UVBRI
120
U
%Transmittance
100
B
V
R
I
80
• Johnson-Cousins (Bessell)
UVBRI as broadband
60
40
20
0
200
400
600
800
1000
Wavelength, nm
• Stromgren* uvby as
intermediate filters
Stromgren Passbands
%Transmittance
120
100
u
v
b
y
80
60
40
20
0
300
350
400
450
500
550
600
Wavelength, nm
* Originally designed for fast surface gravity and
metallicity studies of stars
What is a Narrowband Filter
Filter Bandwidth Terminology
http://spiff.rit.edu/classes/phys440/lectures/filters/filters.html (M. Richmond)
120
Kitt Peak H-α #1468 FWHM 8.5 nm
NARROWBAND
%Transmittance
100
80
Stromgren b FWHM ~20 nm
INTERMEDIATE BAND
60
Bessell V FWHM 85 nm
BROADBAND
40
20
0
550
50 nm600
100 nm
650
700
750
800
850
Wavelength, nm
• Melles Griot Glossary
–
–
–
–
http://www.mellesgriot.com/glossary/w
ordlist/glossarydetails.asp?wID=100
Broadband FWHM >60 nm
Intermediate FWHM 20-60 nm
Narrowband FWHM <20 nm
• Michael Bessell
–
–
–
–
http://www.mso.anu.edu.au/~bessell/araapaper
.pdf (Annu. Rev. Astron. Astrophys, 2005, 43)
Broadband FWHM < ~100 nm
Intermediate FWHM ~7 – 40 nm
Narrowband FWHM < ~7 nm
Proposed Narrowband Definition
•
•
•
•
Broadband > 60 nm (FWHM)
Intermediate Band 11 – 60 nm
Narrowband ≤ 10 nm
Ultra-narrowband < 4 nm (split H-a, NII)
How NB Filters Are Made
%T
BandPass Layer
Substrate
%T
Blocking Layer
Result
%T
Close Up of Narrowband Filter
H
Ta2O5
tantalum pentoxide
L
SiO2
silicon dioxide
H
Ta2O5
L
SiO2
H
Ta2O5
Micrograph ref. ThermoCorion
0.2 microns
0.0002 mm
Human hair 50 mic.
H = High refractive index layer (e.g. 2.3)
L = Low refractive index layer (e.g. 1.45)
Ne = effective index of layer = sqrt ( H * L) = 1.8 (affects blue shift)
Are Astronomy NB Filters Special?
Off-the-Shelf
Company
Part
CWL
FWHM
OD
Tpk (%)
25mm
50mm
Thermo Corion
S10-656-
656 ±2
11 ±2
4
50
$106
$256
Edmund Optics
M43-138
656 ±2
10 ±2
3
≥50
$79.50
$153.70
Melles Griot
03FIR06
656.3 +2/-0
10 ±2
4
45
$111
-
Astronomy NB CCD Filters
• Smaller FWHM - typically 4.5 – 7 nm
• Higher Tpk - 70-95%
• OD - often not specified (important?)
• Need better than ±2 nm CWL for 6 nm FWHM
• Let’s see what effect ±1 nm CWL has
Manufacturing - ±1 nm CWL 6 nm H-α
6nm FWHM H-a Filter Transmission Plot +/- 1 nm CWL
100
% Transmittance (I/Io)
90
80
+1
-1 nm
nm
70
60
50
• No practical change in %T
40
30
20
10
0
648
H-α
H-α
H-α
656.3
656.3
nm
656.3nm
nm
650
652
654
656
658
660
Wavelength, nm
662
664
666
668
Manufacturing ±1 nm CWL 3 nm OIII
3 nm FWHM OIII Filters Transmission Plot +/-1 nm
100
90
-1 nm
%Transmittance
80
70
60
94%T
to
72%T
+1 nm
50
40
30
20
OIII OIII
OIII
500.7
500.7 500.7
nm nm
nm
10
0
490
492
494
496
498
500
502
504
506
508
Wavelength, nm
• No change for -1 nm shift (~94%T)
• 94% to 72% for +1 nm shift at OIII
• If Tpk at λ0 is 80%, +1 nm shift →
<62%T at OIII
510
Tpk and FWHM Myth
% Transmittance (I/Io)
6 and 3 nm FWHM H-a Filter
110
100
90
80
70
60
50
40
30
20
10
0
645
• Idealized case – same Tpk,
different FWHM
Goal
• Can compare effect of just
FWHM on S/N
647
649
651
653
655
657
659
661
663
665
667
669
• Very difficult to CONSISTENTLY
keep Tpk large with narrower
filters
Wavelength, nm
• Lower graph is typical
• Causes confusion (two
parameters are changing)
6 and 3 nm FWHM H-a Filter
100
% Transmittance (I/Io)
90
S
80
70
60
• Smaller FWHM lowers N in S/N
(moon, light poll)
• Smaller T lowers S in S/N
50
N
40
30
20
10
0
645
647
649
651
653
655
657
659
Wavelength, nm
661
663
665
667
669
• ~Cancels the S/N
“perceived” advantage of
narrower filter
OIII FWHM Image Comparison
6 nm, Tpk = 90%
3 nm, Tpk = 93%
• Apogee U16M (KAF16803) at
-25C (gain = 1.2)
• Takahashi FSQ106N at f/5
• 3 x 10 min mean combined
with darks and flats
• City backyard (18.3
mag/sqarcsec, no moon)
• S = (Sig-Bkg)·g
• N = SQRT(S · g + BKG STD2)
• Compare regions with no
stars
Sig = 325 ADU
Bkg = 257 ADU
Bkg STD = 16.84
S = 81.6
N = 19.1
S/N = 4.27
Sig = 187 ADU
Bkg = 118 ADU
Bkg STD = 13.04
S = 82.8
N = 15.9
S/N = 5.21
S/N improvement of
22% with similar Tpk
but half the FWHM
OIII Test Implications
• Halving the FWHM halved the bkg ADU
• Signals are comparable due to similar Tpk
• OIII is more sensitive to moonlight and
light pollution
• May mix narrower OIII filter with wider H-α
and SII filters
• Need Tpk to be similar to achieve
advantage
H-α - FWHM vs. Background
H-a Background
• 12.5” RCOS RC
• Apogee U16M at -20 °C
30
y = 3.6393x + 5.2
Mean ADU
25
• 3 x 10 min unbinned exp.
R 2 = 0.9944
• Darks, flats, bias
calibrated
20
15
• Mean combined
10
• Taken in same region w/o
stars in CCDStack
5
0
2
3
4
5
6
7
• Tpk ~ 92-95%T
FWHM
• Regular relationship
• Lower FWHM leads to lower background
• Need to keep high Tpk for good S/N
Moonlight & Light Pollution
Photometric Red
Imaging Red
OIII
H-alpha
SII
5750 Blackbody
1.20
Incandescent,
Outdoor Flood
Lamp
Normalized Units
_
OIII
1.00
H-α
SII
Sodium Street Lamp
0.80
Fluorescent, Hg
0.60
H-α
0.40
OIII
0.20
0.00
400
500
600
Wavelength, nm
700
800
SII
Blue Shift With Faster Optics
• Simple tilts in collimated light will cause λ0 to
shift toward shorter (blue) wavelengths
• λΘ = λ0·√ (ne2 – Sin2 Θ) / ne
Θ
• ne effective index of the coating (~1.8)
• λΘ is the shifted wavelength
• Faster optics may shift the passband off of the
emission line
Angular Measurement 6 nm OIII
Θ
501nm Filter, 6nm BW, from 0° to 10° AOI
A-6nm-10° %T
Θ
f/3
100
A-6nm-9° %T
0°
90
10°
A-6nm-8° %T
f/3.5
80
f/4
70
A-6nm-7° %T
60
%T
_
A-6nm-6° %T
f/5
50
A-6nm-5° %T
40
OIII Emission
Line
30
A-6nm-4° %T
20
A-6nm-3° %T
10
A-6nm-2° %T
0
490
492
494
496
498
500
502
504
506
508
510
Wavelength (nm)
• Not much effect at 6 nm FWHM
• Because emission line at low wavelength side
• If at high side, will see lower T at OIII
A-6nm-1° %T
f/7
f/10
Angular Measurement 3 nm OIII
Θ
OIII Filter 3 nm FWHM (from 0° to 8°)
Θ
100
Decrease
80
1°
70
2°
60
%T
0°
8°
90
0°
50
40
OIII Emission
Line
30
3°
f/10
4°
f/7
5°
6°
f/5
20
7°
10
8°
0
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
Wavelength (nm )
f/4
f/3.5
Compare Theory & Experiment
OIII Spectral Shift at 500.7 nm 3 nm FWHM
f/10 f/9
f/7
f/5
f/4
f/3
100
%Transmittance
90
80
70
60
Meas.
50
40
Theory
30
20
10
0
0
1
2
3
4
5
6
7
8
9
10
Degrees Off Norm al
•
•
•
•
Measurements show less shift than theory
3 nm filter is reasonable to f/3.5
Not significant effect for >4 nm FWHM
Not the whole story
– Telescope optics
3 nm FWHM OIII on f/5 530 mm Refractor
%T at 500.7 nm 3 nm FWHM OIII Filter
f/10 f/9
f/7
f/5
f/4
•
•
f/3
100
%Transmittance
80
60
f/5 5.7º
40
•
20
0
0
1
2
3
4
5
6
7
8
9
10
Degrees Off Normal
•
Distributionn
Half-Angle Distribution f/5 Refractor
100 mm Aperture 88mm Field
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
•
•
•
0
2
4
6
Angle Off Norm al
Ray trace results courtesy of Mike Jones, Optical
Designer, Lockheed Martin
8
10
Peak distribution at f/5 ~ 5.7°
Many other angles striking
the CCD
Multiply the frequency of the
angle (lower graph) * %T at
500.7 nm in the upper graph
Effective %T for 3 nm FWHM
OIII is 92%
Not much loss from 94%
Slower telescopes, including
RCs, SCTs will not be
affected.
Wider NB filters will not be
affected
3 nm FWHM OIII on 10” f/3.5 mm Astrograph
%T at 500.7 nm 3 nm FWHM OIII Filter
f/10 f/9
f/7
f/5
f/4
f/3
100
%Transmittance
80
60
40
f/5 5.7º
20
0
0
1
2
3
4
5
6
7
8
9
10
Degrees Off Normal
Half-Angle Distribution
10" f/3.5 Astrograph 20% Central Obstruction
50x50mm Detector Area
1.2
1
Distribution
0.8
0.6
0.4
0.2
0
0
2
4
6
8
10
Angle Off Norm al
Ray trace results courtesy of Mike Jones, Optical
Designer, Lockheed Martin
12
14
• 20% obstruction (area)
• 50 mm square detector area
• Peak distribution at f/3.5 ~ 8°
• Many other angles striking
the CCD
• Multiply the frequency of the
angle (lower graph) * %T at
500.7 nm in the upper graph
• Effective %T for 3 nm FWHM
OIII is 80%
• Still useable at f/3.5?
• Slower telescopes, including
RCs, SCTs will not be
affected.
• Wider NB filters will not be
affected
H-a Comparison on E-180 at f/3
5nm FWHM >92%T
3nm FWHM > 92%T
SBIG STL11000XM, -25C, 1 10 min unbinned exposure, calibrated (dark, flat, bias), equalize screen stretch in MaximDL, same S/N
3 nm FWHM OIII Spatial Sensitivity
f/5 100mm System, 88 mm Field Angle Histograms in Each “Strip”
Effective %T
90.1%
88.0%
91.7% 92.7% 93.1%
KAI11000
KAF16803, 9000
40
30
20
mm
10
0
Ray trace results courtesy of Mike Jones, Optical
Designer, Lockheed Martin
FWHM, H-α and Nitrogen
6 and 3 nm FWHM H-a Filter with NII Lines
110
H-α 656.3 nm
% Transmittance (I/Io)
100
NII 658.4 nm
90
80
70
60
50
40
NII 653.8 nm
30
20
10
0
645
647
649
651
653
655
657
659
661
663
665
667
669
Wavelength, nm
• NII doublet occurs close to H-α on either side
• NII can be 20% to >100% of H-α in PN and some galaxies
• Most H-α filters will include H-α AND NII
• Need ≤3 nm FWHM to exclude most NII in H-α
• Imagers want narrower filter to exclude light pollution & moonlight
• Imagers want highest signal, even if it comes from NII + H-α
• 3 nm H-α may not have S/N of 4.5 – 7 nm H-α on some objects (e.g.
planetary nebula, WR bubbles, etc.) due to minimizing NII signal
M76 H-a S/N Comparison
3 nm, 94% Tpk
6 nm, 92% Tpk
• Apogee U16M
at -20C
• Gain 1.2
• RCOS 12.5” RC
• 5 x 10 min 1x1
• Dark, Flat, Bias
calibrated in
CCDStack
• S = (Sig-Bkg)*g
• N = SQRT (S *
g + BKG
STD^2)
• Compare
regions with no
stars
Sig = 208 ADU
Bkg = 16.1 ADU
Bkg STD = 1.68
S = 230
N = 15.3
S/N = 15.1
Sig = 257 ADU
Bkg = 30.4 ADU
Bkg STD = 1.61
S = 271
N = 16.6
S/N = 16.4
Unlike OIII, narrower
H-α may not improve
S/N perhaps due to
loss of NII
FWHM & Radial Velocities
6 and 3 nm FWHM H-a Filter with NII Lines
110
H-α 656.3 nm
% Transmittance (I/Io)
100
90
80
70
60
50
40
-400 km/s
30
+600 km/s
20
10
0
645
647
649
651
653
655
657
659
661
663
665
667
669
Wavelength, nm
•
AAO/UKST H-α survey of the galactic plane
–
Q.A. Parker and J. Bland-Hawthorn, “Technical Aspects of the New AAO/UKST Hα Interference Filter”,
Publ. Astron. Soc. Aust. 1998, 15, 33-7
• -400km/s to +600 km/s
• -0.9 nm to +1.3 nm
• 655.4 nm to 657.6 nm
• No significant effect for most H-α filters
Red Continuum Filter
• Subtract stars from H-a, SII data
Conclusions
• Narrowband Filters
–
–
–
–
–
–
–
FWHM ≤ 10 nm
Tuned to specific emission lines
Increase contrast by eliminating continuum
Show different structures (OIII, SII, NII, etc.)
Extend imaging time with moonlight
Allow us to image in light pollution
Astronomical NB filters are special
Conclusions (2)
Narrower Filters
–
–
–
–
–
–
–
–
Want lowest FWHM with highest Tpk
More difficult to make CONSISTENTLY
More expensive
Cannot achieve S/N advantage unless high Tpk is
maintained.
Wavelength blue shift is not as significant as
previously thought
Off-the-shelf appropriate for ≥ f/3.5
Specially shifted filter for <f/3.5?
May not achieve large advantage for H-α due to loss
of NII in PN and galaxies
Thank You!!
Clear Skies
How Do the Pros Do It – AAO/UKST
•
•
•
•
•
•
•
•
•
•
•
•
“Technical Aspects of the New AAO/UKST Hα Interference Filter”, Pub.
Astron. Soc. Aust, 1998, V15, pp 33-37
Consortium for Hα survey of the Galactic plane
Cover radial velocities of galaxies of -400 to +600 km/s (shifts of -0.9 nm
and +1.2 nm from 656.3 nm)
Placed in the f/2.48 convergent beam
Coated 305 mm Hα diameter (~10”) on a 356 x 356 mm plate (12”)
7 nm FWHM
CWL 659 nm ±2.5 nm to account for blue shift at f/2.48
Blocking 4 OD from 200 to 699 nm
Tpk >85%
¼ wave transmitted wavefront
0.004 nm/C blue shift with temperature.
Narrow enough for use under a bright moon