Particulate Emissions Control by Advanced Filtration Systems for GDI Engines

(ANL/Corning/Hyundai CRADA)June 11, 2015

DOE Annual Merit Review & Peer Evaluation Meeting

PI: Hee Je SeongCo-investigator: Seungmok ChoiArgonne National Laboratory

DOE Project Managers: Ken Howden & Gurpreet SinghOffice of Vehicle Technologies

Project ID: ACE024

This presentation does not contain any proprietary or confidential information

Overview

Budget Total project funding

‒ DOE share: $1.5M‒ Contractor share: $1.5M

Funding for FY15− DOE: $500K− Project partners: $500K

in-kind + fund-in $75K (Hyundai)

Barriers B. Lack of cost-effective emission

control C. Lack of modeling for

combustion and emission control E. Durability

Partners Corning and Hyundai Motor University of Illinois at Urbana-

Champaign University of Illinois at Chicago

Timeline Start: Oct. 2011

‒ Contract signed: Sept. 2012 End: Sept. 2015 80% finished

2

Relevance and Objectives PM emissions from GDI engines are mandated to be reduced

‒ Current GDI engine-out emissions cannot meet future PM regulations (mass & number)

‒ Cold start and transient modes are recognized to produce high PM emissions

‒ New test procedures reflecting real life drives require robust PM reduction technologies

GPFs have been developed to meet stringent PM regulations‒ High PN efficiency & low pressure drop‒ Developing directions: “Add-on” GPF and “All-in-one” GPF

Corning, 2012 DEER Conference

“Add-on” GPF

“All-in-one” GPF

3

Ash impact could be more appreciable with GDI engines than with diesel engines‒ Ash impact on DPF performance has been well characterized by MIT

researchers. However, few reports on GPF performance with ash loading were disclosed.

‒ Backpressure increase and TWC functionality are of great interest.‒ Increased ash fraction was observed to enhance soot oxidation (ANL,

2014 AMR). Objectives

‒ Further validate ash enhancing effects on soot oxidation.‒ Understand filter performance in terms of pressure drop and filtration

efficiency, related with filter geometry and pore structure.‒ Evaluate impact of ash loading on TWC-coated GPFs, based on

backpressure increase, filtration efficiency and particle penetration with filter regeneration.

Relevance and Objectives (Cont’d)

4

Project Milestone (FY14-15)Quarter,

YearMilestone Description Status

Q3, 2014 Detailed morphology data of particulates from a stock GDI engine with variation of injection parameters Complete

Q4, 2014 Analytical data to optimize the filter design MIP & X-ray microtomography,Ongoing

Q1, 2015 Complete development of GPF test protocol, installation of test system, and preliminary test Complete

Q2, 2015 Complete tests for 4 bare filters Complete

Q3,2015 Complete tests for 3 catalyzed filters and an aged filter Ongoing

5

Overall Approach – soot & filter characterization

Soot oxidation experiments and bench-scale filter tests (2”(D)x6”(L))2.4L GDI Engine

TGA

DSC

Characterization of GDI sootand filter substrates (APS, CNM, UIUC)

Gravimetric sampling

In-line filter

PM properties• Operation-specific PM emissions source (#

& mass)• Morphology & physicochemical properties• Oxidation reactivity

Filter geometry• Mercury intrusion porosimetry (MIP)

X-ray microtomography: 2D & 3D• Impact on ΔP change and filtration

performance• Coating effects on filter geometry

Evaluation of filter performance• Understanding soot oxidation mechanism• Aging impact on filter performance using

accelerated ash loading • Physical & chemical effects with ash

loading - ΔP change, filtration efficiency, TWC function

6

GPF Test Approach – targeting feasible GPF testing Corning provided advanced cordierite-based filters

Selected sample % porosity Medium diameter (d50, µm)

AC 200/12KEX 200/8HP 300/8

HP 200/12

50.5657.2264.6065.76

20.7411.6720.3820.93

Hyundai provided TWC-coating services on bare filters through OEM- In-wall coating used in current TWC-coating technology

Selected Samples Catalyst Coating Loading (g/L) PGM loading (g/ft3)

AC 200/12KEM 300/8

HCW 200/12

w/ PGM, OSC, PGM&OSCw/ TWCw/ TWC

5025, 50 and 100

50

0.50.50.5

Argonne evaluated different types of filters for fair comparison at real engine operating conditions using a bench-scale flow reactor under proposed test protocol.

7

Summary of Technical Achievements in FY15

Enhancement in soot oxidation was further validated by TGA experiments.- Different engine operating conditions and simulated tight & loose contact- Examinations of soot oxidation for three major inorganic additives (Ca, P &

Zn) formulated by fuel doping Pore structures of filters were examined by X-ray tomography and mercury

intrusion porosimetry (MIP) to understand catalyst coating effects.- Medium porosity filter (AC 200/12) vs high porosity filters (HP 200/12 &

300/8)- Catalyst loading: 25 g/L (1X), 50 g/L (2X) and 100 g/L (4X) TWC-coated GPF performance was evaluated in the 2.4L GDI engine using

the newly installed bench-scale reactor.

8

Ash effect on soot oxidation is further validated

Injection timing varies

TGA: isothermal 600ºC, 8% O2

0

20

40

60

80

100

0 50 100 150 200

Printex U (PU)LD Diesel_Ash 0.36%GDI_1250-25_Low SootGDI_1250-25_High SootGDI_1500-50_Low SootGDI_1500-50_High SootGDI_3000-50_Low SootGDI_3000-50_High Soot

Car

bon

soot

mas

s (%

)

Time (minutes)

Ash fraction and soot mass were inversely correlated under the same condition sets Low-mass soot was always found to be more reactive due to increased ash fraction Ash, taken from filters of 100,000 mile run vehicle, enhanced soot oxidation reactivity

at simulated tight contact and loose contact conditions

0

20

40

60

80

100

0 50 100So

ot m

ass (

%)

Time (minutes)

PU onlyPU pressed on fresh PGM/filterPU pressed on aged TWC/filterPU + PGM powder PU + ash powder from aged filter

9

Based on major additive components, Ca-, P- & Zn-P-specific engine oils were formulatedICP analysis: Proc-Rev 1158-3.9

0

500

1000

1500

2000

2500

3000

3500

Non-detergent(SAE30)

Gasoline(5W20)Conventional

Gasoline(5W20)Synthetic

Gasoline(5W20)Longer life

Diesel(15W40)

Conc

entr

atio

n (p

pm)

Na

Ca

Mg

B

Zn

P

Mo

4,659 ppm 3,805 ppm 5,212 ppm 6,780 ppm646 ppm

Ca source: Calcium dodecyl-benzene sulfonate

P source: Triethyl phosphite

Zn & P: Zinc dithiophosphate(ZDDP)

Dosage in fuel (ppm) Ca Zn P Na TotalGasoline Only 0.0

1% Non-detergent oil 0.4 3.3 2.7 6.4 1% Conventional oil 21.2 11.0 9.6 4.6 46.5 Calcium Sulfonate

in 1% non-detergent oil 4 – 24 4 – 24

Phosphitein 1% non-detergent oil 18 – 55 18 – 55

Zinc Dialkyl Dithiophosphate (ZDDP)in 1% non-detergent oil 8 – 206 8 – 191 16 – 397

10

Soot oxidation enhanced with Ca, while deteriorated with P

TGA: isothermal 600ºC, 8% O2

0

20

40

60

80

100

0 100 200 300 400 500

Non-detergent oil 1%Conv. oil Ash 3.7%Conv. oil Ash 23.4%Ca Ash 3.5%Ca Ash 5.8%P Ash 3.2%P Ash 5.6%ZDDP Ash 8.5%ZDDP Ash 16.7%

Car

bon

soot

mas

s (%

)

Time (minutes)

Impact on soot oxidationby 1.0 wt.% of ash fraction in soot

Ca +14.5%

P -15.4%

ZDDP (Zn+P) -13.7%

Conv. oil +11.1%

Baseline: non-detergent oil 1%

Ca-derived ash significantly improvedsoot oxidation reactivity, while P-derivedash impaired reactivity

Impact of Zn-derived ash seems to be minor Enhanced soot oxidation by ash present is

because Ca is a dominant component in ash

11

Testing set-up that enables aging test and realistic GPF conditions has been newly built

Engine oil tank

Air heater

GPF

Previous system Current system

Optical GPF setup (half cut 2”(D)x6”(L))- Visualization- Low exhaust T with long line- Sealing problem with quartz windowDilution setup and emissions measurements

- Cumbersome handling for inlet & outletNo lube-oil injection system

-

In-line GPF setup (2”(D)x6”(L))- No visualization - Hot exhaust T as actual test- No sealing problem

Dilution setup and emissions measurements- Quick access: 3-way valve for fast measurements

Lube-oil injection system- Aging test enabled

12

GPF test protocol was prepared to evaluate GPF performance for the bench-scale reactor Require test protocol that evaluate filter performance with ash loading Better understand GPF regeneration condition Need clean filter condition with no soot contained for fair comparison

ΔP test with air

Soot/ash loading 1250 rpm & 25% loadQm = 13.8 kg/hT=350°C

Soot oxidation at increased T

# & mass emissions, ΔP change

2000 rpm & 65% loadQm = 23 kg/hT=600°C

Fuel-cut operation 2000rpm or 30002000rpmQm = 23 kg/h

Complete soot oxidation with air Air heaterT=550°CQm = 15 kg/h

Repeat

# & mass emissions, ΔP change

# & mass emissions, ΔP change

13

High porosity filter had relatively minor changes in pore structures with catalyst coating

0.00

0.20

0.40

0.60

0.80

1.00

Pore

size

dis

trib

utio

n ((d

50-d

10)/

d50)

AC200/12-Bare50.56%

HP200/12-Bare65.76%

HP300/8-Bare64.60%

50.56

47.92

65.76 64.17 64.60 60.82 60.39 56.91

20.74

9.17

20.93 18.67

20.38 18.83

18.82

12.05

0

5

10

15

20

25

010203040506070

Med

ian

pore

dia

met

er (n

m)

Poro

sity

(%)

Porosity

MPD

Mercury intrusion porosimetryWide

Narrow

High porosity filters have relatively big pores uniformly spread in filters Changes in total porosity were minor with catalyst coating, regardless of filter type. However,

MPD decreased significantly with catalyst coating for medium porosity filter. With catalyst coating, PSD became wider for medium porosity filter than for high porosity filters. High porosity filter lost porosity benefits with high catalyst coating (4X).

2D images of X-ray microtomography from APS

14

Low & medium catalyst loadings had minor impacts on ΔP, mass & # filtration efficiencies for HP filter

0

50

100

150

200

250

0

20

40

60

80

100

△Pc

lean

(kPa

/(m

3 /se

c))

△P

soot

(kPa

/(m

3 /se

c))

dPsoot @ 0.05 g/L dPsoot @ 0.10 g/L dPsoot @ 0.15 g/L Clean filter dP

40

60

80

100

120

0 0.03 0.06 0.09 0.12 0.15Mas

s filt

ratio

n ef

ficie

ncy

(%)

Soot Loading (g/L)

AC200/8-Bare

HP300/8-Bare

HP300/8-1X

HP300/8-2X

HP200/12-2X

50

60

70

80

90

100

10 100

Num

ber f

iltra

tion

effic

ienc

y (%

)

Mobility diameter (nm)

HP300-8-BareHP300-8-1XHP300-8-2X

Initial ηmass (%)

Soot loading (g/L) @ ηmass 100%

AC200/8-Bare 70.0 0.040

HP300/8-Bare 65.0 0.140

HP300/8-1X 79.5 0.120

HP300/8-2X 76.6 0.175

HP200/12-2X 87.4 0.110

High porosity filters had low ΔP increase and slow increase in filtration efficiency up to the max point with soot loading.

High porosity filter (HP300/8) has comparable ΔP increase, mass & # filtration efficiencies with low (1X) and medium (2X) catalyst loadings.

For high porosity filter, lowest # filtration efficiency was obtained for 40 – 60 nm particles, which are smaller than what others observed (100 – 150 nm).

- Greenfield gap showing the most particle penetrating range tends to slightly shift to smaller size range with catalyst loading.

Soot mass: 10 mg/m3

15

70

75

80

85

90

95

100

0.0E+00

2.0E+06

4.0E+06

6.0E+06

8.0E+06

1.0E+07

1.2E+07

10 100

Filtr

atio

n ef

ficie

ncy

(%)

dN/d

logD

p (#

/cc)

Mobility diameter (nm)

Engine out

Filter out - W/O Ash

Filter out - Ash 2.0 g/L

FE - W/O Ash

FE - Ash 2.0 g/L

With pore plugging by ash loading, depth filtration duration significantly decreased, improving filter efficiencies

Ash loading- Engine oil: 5W20 conventional (p.10)- Oil consumption rate: 2% in fuel - Ash loading of 2 g/L in 3 hrs Ash loading changed ΔP patterns as

observed from previous DPF studies- Short depth filtration (0.02 vs 0.2 g/L

of soot) - Lower ΔP slope (inverse point:

0.18g/L) Ash loading improved filtration

efficiency- Initial mass 𝜂𝜂𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 (98 vs 87.4%)- Initial # 𝜂𝜂𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 (98.3 vs 81%)

Total # FE (%)

W/O Ash 81.06

Ash 2.0 g/L 98.28

80

85

90

95

100

0

50

100

150

200

250

300

0 0.1 0.2 0.3 0.4 0.5

Filtr

atio

n ef

ficie

ncy

(%)

Pres

sure

dro

p (k

Pa/(

m3 /

sec)

)

Soot loading (g/L)

△P - W/O Ash

△P - Ash 2.0 g/L

FEmass - W/O Ash

FEmass - Ash 2.0 g/L

At the start of soot loading

Soot mass: 10 mg/m3

Filter: HP200/12-2X

16

Ash induced continuous regeneration and suppressed particle penetration with pore plugging

Continuous regeneration was observed at an initial stage with ash loading.

- After soot burning at the initial stage, soot loading resumed

- No or very slow soot oxidation with no ash even at 600°C due to the extremely low O2 availability

Particle emissions were relatively high during continuous regeneration with no ash loading.

Right after fuel cut, particle penetration increased with pores opened, resulting from soot oxidation, when there was no ash loaded. With ash loading of 2 g/L, however, particle penetration was minor after fuel cut.

W/O Ash

Ash 2.0 g/L

Fuel cut operationsContinuous operation

17

Responses to FY14 Reviewer Comments Ash effects were noteworthy, but they need more supportive data.

‒ As examined in this work, enhancement in soot oxidation with ash present is due to Ca additive that is the most abundant in engine oil.

‒ Continuous regeneration occurred at 600°C with ash loading without fuel cut. In comparison, soot oxidation is quite slow even at such a high T with very low O2availability when ash loading is negligible.

There need studies on fundamentals of PM formation in GDI engines. ‒ Based on this work as well as others, increased PM emissions from GDI engines

are directly related with delayed fuel vaporization and resultant fuel-air mixing problems. Although deeper investigation is out of scope, several operating strategies such as increased injection pressure and fast warm-up are shown to decrease soot emissions.

Integration with kinetics expertise is required‒ After more investigations on filter type, catalyst loading and filter position, kinetics

research is recommended for the future. Collaborations with research partners are not clear.

‒ Collaborations with other partners were not clearly defined (for instance, Tokyo Tech and workshop hosting). Clear roles of research partners are listed in the following page.

18

CollaborationsCollaborating Partners Corning Incorporated

– Provided 8 different filter substrates including most advanced GPF filter substrates

– Had several technical meetings for future directions– Performed MIP of uncatalyzed and catalyzed filters

Hyundai Motor Company– Provided a 2.4 production GDI engine and open ECU for full control– Gave technical advice of GPF research direction and provided catalyst coating

services for GPF substrates based on current state-of-the-art coating technologyOther Internal and Outside Partners University of Illinois at Urbana-Champaign

– Performed XPS analysis University of Illinois at Chicago

– Xiao Fu (Ph.D. student) helped to analyze X-ray tomography as a guest graduate User Facilities at Argonne (Advanced Photon Source & Center for

Nanoscale Materials)– X-ray microtomography, TEM, Raman, FTIR and SEM

19

Remaining challenges and barriers Accelerated ash loading needs more investigation for qualification.- Rarely examined in GDI engines- Will be compared with aged filters from field tests Long-term TWC functionality in TWC-coated GPF has not much

known with ash loading.- Ash loading challenges maintaining TWC performance as well as

backpressure increase over time - Development of ash-durable catalyst and filter systems may be required Development of cost-effective and durable GPF system is

complicated.- “Add-on” GPF vs “All-in-one” GPF - “All-in-one” GPF has cost benefit over “Add-on” GPF, but the former could

be more vulnerable toward catalyst poisoning and backpressure increase

20

Future Work (to the end of FY15)

Mechanisms of enhanced soot oxidation in the presence of Ca additive

Long-term aging tests with catalyzed filters– High-porosity catalyzed filters (HP 300/8 & 200/12)– Targeted ash loading: 30g/L– Aging effects: pressure drop, filter regeneration and TWC performance

Advancing X-ray micro-tomography and SEM analysis– 3-D image analyses of uncatalyzed, catalyzed and aged filters– Post-Mortem analysis: ash loading in filters

21

Summary Enhancement in soot oxidation by ash is due mainly to Ca additive that is a

major component in engine oil additives. The newly-built in-line GPF system with an engine oil injection system enabled

“TWC-coated GPF” tests with ash loading, in which TWC performance, pressure drop and soot oxidation performance are measured.

High porosity filter had relatively minor changes in pore structure with catalyst coating, resulted in minor impact on ΔP, filtration efficiencies, in comparison with medium porosity filter.

- However, high catalyst loading (100g/L) sacrificed high porosity benefits.- Greenfield gap was observed to be in small particle range of 40 - 60 nm in

mobility diameter. Ash loading had significant impacts on ΔP and filtration efficiency.- Shorter depth filtration duration and increased mass & number filtration

efficiencies.- Continuous regeneration occurred spontaneously at high T with ash loading.- Ash loading helped suppress particle penetration after soot oxidation. While continuous regeneration was interfered even at 600°C with extremely

low O2, fuel cut conditions induced continuous regeneration.

22

Technical Back-up

23

Developing micropore measurement capability using the X-ray microtomography facility at APS-Argonne

AC 200/12

HP 300/8

KEX 200/8

HP 200/12

APS facility provides high-fidelity microporestructures with a spatial resolution of 0.65 µm/voxel

3-D images reconstructed from ~2000 2-D images Will examine effects of catalyst coating and ash

loading on detailed micropore structures

AC 200/12 – 3D image

Open pores

Dead pores

50%

57%

66%

Porosity

65%

24

Despite similar crystalline structure, GDI soot shows different oxidation patterns, than Printex U

100

120

140

160

180

200

Printex-UGDI_cold idle

GDI_1250-25%-330o

GDI_1500-50%-190o

GDI_1500-50%-330o

Diesel_1400_50% load

FWHM

D1 (c

m-1 )

Raman Spectroscopic Analysis

90% oxidized GDI sootat 500°C

90% oxidized diesel sootat 500°C

90% oxidized Printex-Uat 500°C

With oxidation, GDI soot experienced internally burned-out process like Printex-U (PU) However, oxidation enhancement with partial oxidation is significant with PU,

whereas it is minor with GDI soot need more investigation

0

0.2

0.4

0.6

0.8

1

800 1000 1200 1400 1600 1800 2000

0%33%67%90%

Raman shift (cm-1)

Norm

alize

d in

tens

ity Increasing crystalline degree

0

20

40

60

80

100

0 50 100 150 200

Printex U

LD Diesel_Ash 0.36%GDI_Ash 0.10%GDI_Ash 0.58%GDI_Ash 1.35%GDI_Ash 3.36%GDI_Ash 17.27%

m/m

0 [%]

Time [min.]

Increase of ash fraction

From 2014 AMR

0

20

40

60

80

100

0 50 100 150 200

Original PU30% oxidized PU60% oxidized PUOriginal GDI soot30% oxidized GDI soot60% oxidized GDI soot

Rem

aini

ng m

ass

(%)

Time (min)

Trends with partial oxidation

0

0.01

0.02

0.03

0.04

0.05

0.06

500 1000 1500 2000 2500 3000 3500 4000

0% oxidized33% oxidized67% oxidized90% oxidized

Abso

rbanc

e (A.U

.)

Wavenumber (cm-1)

Aromatic carbon

Surface oxygencomplexes (SOCs)

Acetylenic C-H& O-H stretching

TGA Isothermal 600°C, 8% O2

25

Many of sub-23-nm particles seem to be soot

Most sub-15-nm particles are amorphous carbon

0.E+00

1.E+07

2.E+07

3.E+07

4.E+07

5.E+07

6.E+07

7.E+07

8.E+07

10 100 1000

dN/d

logD

p [#

/cm

3 ]

Mobility diameter [nm]

1500-50-Default

1500-50-Retarded

1500-50-Advanced

Non-discernable

Background- Amorphous carbon

Concentric pattern- indication of soot

Particles still remained, despite high energy beam under high vac.

Original image After 600KXAt 600kX

Technical Accomplishments

Images processed

26

Lube oil additives increase the number of sub 30 nm particles and decrease number of large particles

104

105

106

107

108

10 100 1000

Gasoline Only

Non-detergent oil 1%

Conv. oil 1% (46 ppm)

dN/d

logD

p (#/c

m3 )

Dp (nm)

104

105

106

107

108

10 100 1000

Non-detergent oil 1%

P 18.2 ppm

P 54.7 ppm

dN/d

logD

p (#/c

m3 )

Dp (nm)

104

105

106

107

108

109

10 100 1000

Non-detergent oil 1%

ZDDP 15.6 ppm

ZDDP 62.4 ppm

ZDDP 397.8 ppm

dN/d

logD

p (#/c

m3 )

Dp (nm)

012345678

1.E+06

1.E+07

1.E+08

PM (m

g/m3)PN

(#/c

m3 )

PNPM

1.E+04

1.E+05

1.E+06

1.E+07

1.E+08

PN (#

/cm

3 )

< 30> 100

Conventional oil Phosphite ZDDP

27

Engine oil may emit as liquid drops or less-ordered carbon without complete oxidation – Engine condition-dependent

From P additive

Some contain metal compounds

104

105

106

107

108

109

10 100 1000

Non-detergent oil 1%

ZDDP 15.6 ppm

ZDDP 62.4 ppm

ZDDP 397.8 ppm

dN/d

logD

p (#/c

m3 )

Dp (nm)

Amorphous carbon

Numerous liquid drops

0

1

2

3

4

5

6

7

8

0 0.5 1 1.5 2 2.5 3 3.5 4

1250 rpm - 25% load1500 rpm - 50% load3000 rpm - 50% load

VOF

(%)

Ash content (%)

CaseXPS, atomic % From TGA

(mass %)Ca P Zn S

Ca additive

1.23 0 0 0 5.8

P additive 0 2.83 0 0 5.6

ZDDP 0 2.88 0.08 0 16.7

28