smps/cpc comparisons beachon 2012
TRANSCRIPT
10/14/2012
1
Update on SMPS/CPC comparisons at BEACHON‐RoMBAS 2012
1. Deployed 5 different SMPSs and 3 stand‐alone CPCs at RoMBAS• SMPS: TSI‐Kr, TSI‐Xray, Custom, BMI, Grimm• Often only 1 SMPS at a field site, used to “determine” CE• Opportunity to see how well that may work out• Basic check: compare total number concentrations
2. Reminder of where we stood at the end of campaign3. Identification of some of the problems4. Moral of the story
Brett Palm, Jose‐Luis JimenezUniversity of Colorado‐Boulder
AMS Users Meeting 13‐Oct‐2012
1
As shown at 2011 AMS Users Mtg
2
Number Concentration
Factor of x5 between instrumentsNone of SMPS agree with CPC!
10/14/2012
2
As shown at 2011 AMS Users Mtg
Three different CPC models match taken to be “true” int. number density
Compared to CPCs:SMPS #4 and #5 are too highSMPS #1 matches except for afternoonsSMPS #2 and #3 are too low
3
SMPS#1: CPCs have a lower size cutoffNucleation during afternoons
Small particle losses in SMPS explain afternoon deviations
10/14/2012
3
SMPS#1: CPCs have a lower size cutoffNucleation during afternoons
Small particle losses in SMPS explain afternoon deviations
Investigating SMPS #2: dN/dlogDm
scans compared to SMPS #1• Three different (and stable in time) size distributions on 8/12/11• Residence time in SMPS #2 inlet was much longer than in SMPS #1 inlet large losses of
small particles in inlet
6
12x103
8
4
0
dN/d
logD
m
2 3 4 5 6100
2 3 4 5 6
Dm
8/12/2011 15:16
15x103
10
5
0
dN/d
logD
m
2 3 4 5 6100
2 3 4 5 6
Dm
8/12/2011 21:56
2500
2000
1500
1000
500
0
dN/d
logD
m
2 3 4 5 6100
2 3 4 5 6
Dm
8/12/2011 00:00
SMPS #1 SMPS #2
10/14/2012
4
Empirical Correction Function for SMPS #2 dN/dlogDm
1.0
0.8
0.6
0.4
0.2
0.0
% T
rans
mitt
ance
3 4 5 6 7 8 9100
2 3 4 5 6 7
Dm7
SMPS #2: After size‐dependent correction
8
12x103
8
4
0
dN/d
logD
m
2 3 4 5 6100
2 3 4 5 6
Dm
8/12/2011 15:16
15x103
10
5
0
dN/d
logD
m
2 3 4 5 6100
2 3 4 5 6
Dm
8/12/2011 21:56
2500
2000
1500
1000
500
0
dN/d
logD
m
2 3 4 5 6100
2 3 4 5 6
Dm
8/12/2011 00:00
SMPS #1 SMPS #2
10/14/2012
5
Integrated Number after SMPS #2 size‐dependent correction
9
Integrated Volume after SMPS #2 size‐dependent correction
10
10/14/2012
6
As shown at 2011 AMS Users Mtg
Three different CPC models match taken to be “true” int. number density
Compared to CPCs:SMPS #4 and #5 are too highSMPS #1 matches except for afternoonsSMPS #2 and #3 are too low
11
Direct Comparisons of SMPS #1, #3‐5 w/ CPC
12
DMA set to a one size
SMPS #1, #3, #4, #5
CPCAmmonium sulfate from atomizer
• Compare CPC counts to integrated SMPS counts• Compare DMA‐selected particle size to measured SMPS size
10/14/2012
7
1500
1000
500
0
cm-3
4:00 PM8/28/2011
4:15 PM 4:30 PM
Integrated SMPS Number Concentration (1nm - 1000nm only) CPCcounts
25nm
50nm
125nm
280nm
SMPS #1 vs. CPC
13
• Decent agreement in size and integrated number
SMPS #4 vs. CPC• Selected vs. measured size matches, but integrated number density does not
14
100nm 200nm 50nm 20nm
10/14/2012
8
As shown at 2011 AMS Users Mtg
Three different CPC models match taken to be “true” int. number density
Compared to CPCs:SMPS #4 and #5 are too highSMPS #1 matches except for afternoonsSMPS #2 and #3 are too low
15
SMPS #5 vs. CPC• Neither size nor integrated number density match• Known problem: a TSI X‐Ray neutralizer was used, which produces a more symmetrical
equilibrium charge distribution (see next slide). This means the software inversion is wrong, resulting in ~15‐40% higher particle number density measured
• Unknown problem: why do selected and measured particle sizes not match?
16
16x103
14
12
10
8
6
4
2
0
Num
ber
Den
sity
(cm
-3)
6:00 PM8/23/2011
6:10 PM 6:20 PM
SMPS #5 CPC counts
20nm selected by CU DMA(<14.6nm meas. by LANL)
50nm selected(37.2nm meas.)
280nm selected(216.7nm meas.)
100nm selected(73.7nm meas.)
10/14/2012
9
Slides from a TSI Webinar: X‐Ray vs Krypton Source
17
18
Slides from a TSI Webinar: X‐Ray vs Krypton Source
10/14/2012
10
SMPS #3 vs. CPCStill working on this comparison, but there was definitely a software or user error in converting dN/dLogDp to integrated number. More to come…
19
What’s going on?• Nobody gets funded to bring an SMPS to a field campaign, or gets
excited about its data (e.g. good luck recruiting a good PhD student to do SMPS measurements!)
• In a field study, people tend to take SMPS out of the box, set it up, and not pay much attention to it as they work on the more scientifically interesting instrument they really care about‐ Unfortunately SMPSs can have many problems and still produce reasonably‐
looking concentrations and size distributions. Most users won’t even notice!‐ I’d estimate the probability of a problem is ~90% for unattended SMPS
• AMS users need to be the exception because of the need of high‐quality SMPS measurements for comparison
Moral of the Story
20
SMPS #4: Inversion routine error
SMPS #5: known X‐Ray source
problem + unknown size issue
SMPS #3: software / user error?
SMPS #2: Transmission loss + CPC problem
• Always use 2+ independent measurements to compare to AMS (especially if trying to determine CE).
• Need to ALWAYS compare number and calibrate size on SMPS• Even “routine” instruments are wrong more often than not.
SMPS #1: Not seeing nucleation particles
10/14/2012
11
BACKUP SLIDES
21
Summary• 5 SMPSs from 5 experienced groups at same location– 1 working OK (not seeing nucleation particles, as expected)
– 1 losing particles < 200 nm due to both long inlet and CPC issues
– 1 likely has error in inversion software
– 1 has too many particles due to TSI X‐Ray charging efficiency, also unknown sizing problem
– 1 unclear, but maybe sofware / user error
• Moral of the story: you always want TWO independent measurements to compare to AMS
22
10/14/2012
12
Integrated Volume after SMPS #2 size‐dependent correction (zoom)
23
Number Comparison #1 vs #2
24
10/14/2012
13
Volume Comparison #1 vs #2
25
SMPS #2 now matches CPCs and SMPS #1
26
What about SMPS #3, #4, #5?
10/14/2012
14
27
Slides from a TSI Webinar: X‐Ray vs Krypton Source