Rapid Measurements of Aerosol Size Distributions Using a Fast Integrated Mobility Spectrometer (FIMS)

Transcription

1 Rapid Measurements of Aerosol Size Distributions Using a Fast Integrated Mobility Spectrometer (FIMS) Jason Olfert, Brookhaven National Laboratory Jian Wang, Brookhaven National Laboratory

2 Measurement of Aerosol Size Distributions from Aircraft For aircraft-based atmospheric studies an instrument must have: high sensitivity and good counting statistics for low particle concentrations fast response time for good spatial resolution. 100 m/s 500 m Time in cloud = 5 s

3 Fast Size Distribution Measurements Scanning Mobility Particle Sizer (SMPS) - The industry standard, but ~1 min is needed for each scan. Electrical Aerosol Spectrometers (Cambustion s DMS & TSI s EEPS) - Response time much less than 1 s, but have low sensitivity. Electrical Low Pressure Impactor (ELPI) - Fast inertial-based instrument, but has limited size resolution and low sensitivity. Optical Particle Counter (OPC) - Fast measurements but range limited to dp>100 nm, and uncertainties due to particle shape and refractive index.

4 Operating Principle of the Fast Integrated Sheath Flow x Mobility Spectrometer (FIMS) Neutralizer x Butanol-saturated Sheath Flow Aerosol Flow Condenser Separator Aerosol Flow Electrostatic Field 25 o C 10 o C Laser Separator Condenser High Voltage Camera Laser x CCD Images Geometry High Speed CCD Camera

5 FIMS Classification Examples 110 nm 45 nm from DMA from DMA Polydisperse Ground Side High Voltage Side smaller particles 11 mm

6 From Particle Location to Size Distribution Ground Side x High Voltage Side Binning Particle Counts x-channels x Zp dp Inversion dn/dlog(dp) 11 mm dp

7 Comparing FIMS to SMPS Steady-state ambient aerosol size distribution dn/dlog(d p ) (#/cm 3 ) 2.5 x SMPS FIMS D p (nm)

8 Measurements from a field study FIMS 1 s data SMPS 60 s data

9 Dynamic Characteristics of the FIMS

10 Sources of transient error - Residence time in separator and condenser Particle detection time must be corrected for differences in detection time. 8 Time (s) Bin 10.. Bin 3 Bin 2 Separator x Flow Profile High Voltage Bin 1 2 Condenser x * /a

11 Sources of transient error - Smearing in the FIMS inlet Mixing of the aerosol in the inlet causes particles sampled at one instant to enter the separator at different times. Particles at center of tube will travel faster than particles near tube wall Flow mixing in neutralizer Distributing the flow causes time delays Aerosol Sample line Neutralizer Separator entrance Critical Orifice 0.27 L/min 5.45 L/min 5.05 L/min

12 Determining the time constant Step-response of the instrument Time constant can be found by measuring the response to a step-change in the input. Diffusion Dryer Produce monodisperse aerosol Neutralizer (a) Step between classifying voltage and max. voltage Make up Air Make-up air to make flow turbulent Atomizer DMA ndma Filter FIMS or CPC 3025A Excess flow

13 Step-response of FIMS and CPC Normalized aerosol concentration (N out /N 0 ) CPC 3025A (τ=0.14 ± 0.02 s) FIMS without Neutralizer (τ=0.76 ± 0.10 s) FIMS with Neutralizer (τ = 1.69 ± 0.10 s) CPC 3010 (τ = 0.83 s; Buzorius (2001)) N o ut N 0 = e (t t 0)/τ Time (s)

14 Determining the time constant Frequency-response of the instrument Time constant can also be determined by measuring the response to sinusoidal input Diffusion Dryer Produce aerosol distribution (b) Sinusoidally dilute the aerosol Neutralizer Dilution system Blower Filter Atomizer DMA Flowmeter FIMS or CPC 3025A Make up Air Filter Excess flow

15 Attenuation of the FIMS measurement

16 Frequency-response curve of the FIMS

17 De-smearing the data de-convolve the time series of particle counts in each size bin before inverting the data. C i (t) = 1 t Concentration measured in each size bin, i, at time, t (known) t 0 R i (t )Θ(t t )dt Concentration entering FIMS in each size bin, i, at previous time, t (unknown) N out N in = e (t t 0)/τ First-order system Response function, Fraction of particles that enter at previous time, t, that are detected at time t. (known from model)

18 De-smearing the FIMS data With de-smearing the attenuation and delay is eliminated but the random error almost doubles: σn went from 7.9% to 15% σgmd went from 2% to 3.8% Normalized concentration Normalized concentration (a) Standard Inversion (no de smearing) FIMS data FIMS fit Input concentration σ N = 86 cm 3 (7.9 %) σ GMD = 0.88 nm (2 %) σ GSD = (1.3 %) σ SKW = (4.8 %) σ N = 162 cm 3 (15 %) σ GMD = 1.6 nm (3.8 %) (c) τ D =1.0 fd=0.2 Hz σ GSD = (2.1 %) σ SKW = (7.7 %) fd=0.2 Hz Time (s)

19 De-smeared frequency-response curve 1.5 Concentration ratio (N out /N in ) τ=0 s τ=1.02 s Standard inversion De smeared inversion; τ D = 1.0 s Frequency, f d (Hz)

20 Conclusions The FIMS is a highly sensitive aerosol sizing instrument that is useful in many applications, esp. aircraft-based studies. Steady-state measurements agree well with SMPS. Error in transient measurements can occur due to smearing in the aerosol inlet. Data can be de-smeared to correct for inlet mixing, but random error will increase.

21 Acknowledgments Department of Energy, Atmospheric Science Program Goldhaber Distinguished Fellowship, Brookhaven National Laboratory Any Questions?