Exploration of Melt Spinning as a Route to Large Volume Production of Skutterudite Thermoelectric Materials - James Salvador David Brown and ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Exploration of Melt Spinning as a Route to
Large Volume Production of Skutterudite
Thermoelectric Materials.
James Salvador
David Brown and
David Miller
Hsin Wang and
Andrew A. Wereszczak
james.salvador@gm.com
Acknowledgements
U.S. Department of Energy Grant # DE-FC26-04NT 42278
John Fairbanks (DOE), Carl Maronde (NETL)
U.S. Department of Energy, Assistant Secretary for Energy Efficiency and Renewable
Energy, Office of Vehicle Technologies, as part of the Propulsion Materials Program,
under contract DE-AC05-00OR22725 with UT-Battelle, LLC
GM R&D Thermoelectrics Team:
Researchers:
Gregory P. Meisner Collaborators/Subcontractors:
Mike Reynolds Marlow Industries: Jeff Sharp, Jim Bierschenk, Josh
Postdocs: Moczygemba, and Alan Thompson
Xun Shi, Jung Cho, Zuxin Ye Oak Ridge National Laboratory: Hsin Wang, Andy Wereszczak
Engineering Operations:
University of Nevada, Las Vegas: Changfeng Chen, Yi Zhang
Kevin Rober
John Manole Future Tech: Francis Stabler
Gov. Contracts: Heat Technology, Inc
Ed Gundlach,
Amanda Demitrish Emcon (Faurecia)
Rick Leach Shanghai Institute of Ceramics: Lidong Chen
Management:
University of Michigan: Ctirad Uher
Jan Herbst
Mark Verbrugge University of South Florida: George Nolas
GMPT Integration & Testing: Brookhaven National Laboratory: Qiang Li
Jennifer Staneck
Joshua Cowgill Michigan State University: Don Morelli
Stuart Smith General Electric Global Research: Todd Anderson, Peter DeBock
Melt Spinning: Back Ground The use of melt spinning in metallurgy dates back to the early 1960’s for the production of amorphous metallic materials. The technique was developed into a continuous feed process in the mid 1970’s Over the years the technique has been developed to produce more and more complex intermetallic phases such as the rare earth magnets Nd2Fe14B (NIB) Magnequench was formed from the development of NIB at GM R&D in the mid to late 80’s
Melt Spinning and Thermoelectric Materials
Some of the first reports of the application of melt spinning to thermoelectric
materials came around 2007
H. Li and X.F. Tang at Wuhan University China reported the MS
production of Yb-filled skutterudites. The focus of these efforts was to
make nanostructured thermoelectric materials with improved properties.
Subsequently Q. Li at Brookhaven National Lab and T. Tritt from Clemson
have reported the use of MS for the production of both n-type (Clemson) and p-
type (BNL) skutterudite materials. BNL has advanced to other materials systems
such as HMS. Other systems prepared by this method include clathrates and
half-Heuslers.
There has been many recent reports on improved thermoelectric properties of
MS Bi-Te based materials, and the improvement is attributed to nanostructuring
imparted by the MS process.
Skutterudite-Based Thermoelectric Materials We concluded from our last program that the traditional method of preparing skutterudite materials was too time consuming and energy intensive to be economically viable. This is due to the peritectic decomposition of the CoSb3 below its melting point: these materials cannot be melted and cast, but require lengthy heat treatment. Skutterudite materials can also be made by mechanical alloying
Traditional preparation of n-type filled-skutterudites
Co and Sb shot pre- 2 weeks Annealed samples are
melted by induction at annealing at ground into powder for
1400oC in a ratio of 1:3 in 750oC hot pressing or spark
BN. Followed by adding
plasma sintering
Ba, Yb and Sb to the
desired composition and
re-melting at 1200oC for
The resulting ingot is not
5 min.
the desired skutterudite
phase since skutterudites Resulting billets are >98%
decompose above fully dense pure phase
840oC; products are: After annealing the skutterudite
CoSb2 + Sb+ YbSb2 product is
BayYbxCo4Sb12
For traditional method 2 weeks preparation time used
Proposed melt-spin preparation of n-type filled-skutterudites
Co and Sb shot pre- Fine-grain structure of the Ribbons are ground into
Resulting ingot melt-spun
melted by induction at rapidly cooled ribbon powder for spark plasma
using specified
1400oC in a ratio of 1:3 in product facilitates solid- sintering
temperature, ejection
BN. Followed by adding state reaction, eliminating
pressure, and wheel speed
Ba, Yb and Sb to the need for lengthy
desired composition and annealing
re-melting at 1200oC for
5 min.
Ribbon product: Resulting billets are >98%
not the desired fully dense pure phase
skutterudite phase skutterudite
For melt spinning method preparation time is reduced to 4-5 hours
Microstructure of as-spun materials
10µm
1 µm
1µm 100 nm
Contact Surface Free Surface
Direct Sintering of As-Spun Ribbons 400
Meas. data:p-type LCprocessed annealed
600C/Data 1
Kieftite,Co Sb3,01-073-9522
300
Intensity (cps)
200
100
0
10 20 30 40 50
Kieftite, Co Sb3, 01-073-9522
Integrated Intensity (cps deg)
200
100
0
10 20 30 40 50
2-theta (deg)
1500 Meas. data:p-type MQ processed SPS2 co
nsolidated slic center/Data 1
Meas. data:p-type MQ processed SPS2 co
nsolidated transport bar rough/Data 1
Meas. data:p-type MQ processed SPS2 co
X-ray micro-diffraction of nsolidated transport barsmooth/Data 1
Praseodymium iron tetraantimonide,Pr0.87
1000
transport disk and bars
Fe4 ( Sb4 )3,01-075-8585
Antimony Nickel,Sb2 Ni5,00-034-0947
Intensity (cps)
500
0
10 20 30 40
Praseodymium iron tetraantimonide, Pr0.87 Fe4 ( Sb4 )3, 01-075-8585
Antimony Nickel, Sb2 Ni5, 00-034-0947
Integrated Intensity (cps deg)
1000
800
600
400
This method is scalable. 200
0
10 20 30 40
2-theta (deg)
We have obtained pure phase billets on the 5 g (1/2” diameter) to the 80+ g scale
(3.0 cm diameter/1.5 cm thick).
Direct Conversion to Pure Skutterudite Phase Can
be Accomplished in Minutes Instead of Days.
150 150
Meas. data:LC processed P-type 5 days 70 Meas. data:LC processed P-type 5 days 70
0/Data 1 0/Data 1
Seinajokite, nickeloan, syn,( Fe.75 Ni.25 ) Seinajokite, nickeloan, syn,( Fe.75 Ni.25 )
Sb2,01-086-2263 Sb2,01-086-2263
100 Iron Praseodymium,Fe7 Pr,00-019-0618 100 Iron Praseodymium,Fe7 Pr,00-019-0618
Antimony,Sb,01-072-4134 Antimony,Sb,01-072-4134
Nisbite, syn,Ni Sb2,01-082-0395 Nisbite, syn,Ni Sb2,01-082-0395
Antimony Nickel Praseodymium,Sb3 Ni Pr, Antimony Nickel Praseodymium,Sb3 Ni Pr,
00-058-0264 00-058-0264
Iron,Fe,00-050-1275 Iron,Fe,00-050-1275
Intensity (cps)
Intensity (cps)
50 50
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Seinajokite, nickeloan, syn, ( Fe.75 Ni.25 ) Sb2, 01-086-2263 Seinajokite, nickeloan, syn, ( Fe.75 Ni.25 ) Sb2, 01-086-2263
Iron Praseodymium, Fe7 Pr, 00-019-0618 Iron Praseodymium, Fe7 Pr, 00-019-0618
Antimony, Sb, 01-072-4134 Antimony, Sb, 01-072-4134
Nisbite, syn, Ni Sb2, 01-082-0395 Nisbite, syn, Ni Sb2, 01-082-0395
Antimony Nickel Praseodymium, Sb3 Ni Pr, 00-058-0264 Antimony Nickel Praseodymium, Sb3 Ni Pr, 00-058-0264
Iron, Fe, 00-050-1275 Iron, Fe, 00-050-1275
Integrated Intensity (cps deg)
Integrated Intensity (cps deg)
100 100
80 80
60 60
40 40
20 20
0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
2-theta (deg) 2-theta (deg)
SPS for 20 min.
Annealed for 1 week at 650oC
This was found to be true for
both n- and p-type skutterudite 400
materials and did not matter if Meas. data:p-type MQprocessed annealed
600C/Data 1
the materials were rapidly
Kieftite, syn,Co Sb3,01-074-4343
300
quenched (higher wheel speed)
or slow quenched
Intensity (cps)
200
100
N-type materials are more stable
at higher temperatures 0
0 10 20 30 40 50 60
Kieftite, syn, Co Sb3, 01-074-4343
Integrated Intensity (cps deg)
P-type tended to decompose in 200
the SPS if sintering temperature 100
exceeded 600 oC 0
0 10 20 30
2-theta (deg)
40 50 60
Transport Property Evaluation P-Type Materials Thermal and electrical transport parameters (300 K) Sample ID κ(W/m-K) S(µV/K) ρ(mΩ-cm) PF(µW/cm2-K) FQ-SPS 2.0 115 0.79 16.3 FQ-An-SPS 2.7 121 0.80 18.4 SQ-SPS 2.5 100 0.77 12.6 SQ-An-SPS 2.7 109 0.72 16.5 FQ = Fast Quench SQ Slow Quench An-SPS= annealed and then SPS
High Temperature Transport Properties and Composition
SQ FQ
LT QD PPMS
Pr0.176+0.004Nd0.50+0.01Fe3.41+0.02Ni0.588+0.007Sb12.09+0.02
ZTav. ~ 0.65
(100g batch)Transport Property Evaluation N-type Materials Sample ID κ(W/m-K) S(µV/K) ρ(mΩ-cm) n (cm-3) µΗ (cm2/V-s) PF(µW/cm2-K) VFQ-SPS 3.5 -114 0.39 4.5x1020 36.6 34 FQ-SPS 3.5 -127 0.46 3.5x1020 34.5 35 SQ-SPS 3.9 -122 0.41 3.8x1020 39.3 35 All samples had the same nominal composition Yb0.13Ba0.10Co4Sb12
Composition and Microstructure of N-type Materials SQ FQ VFQ Yb0.11Ba0.085Co4.00Sb12.06 Yb0.09Ba0.070Co4.00Sb12.02 Yb0.11Ba0.067Co4.00Sb12.09 Grain size for VFQ and FQ samples are comparable, and much finer than the SQ samples. All samples irrespective of quench rate can be sintered into single phase billets by heat to 650 0C and holding for 10 to 15 min. Despite differences in the carrier concentration for the VFQ sample the filling fraction is not significantly higher than the other samples as assessed by EPMA .
Transport Property Evaluation N-type Materials
A small and likely statistically insignificant improvement
in ZT with a faster quench rate:
(ZT= 0.97 for SQ vs. ZT =1.05 for FQ and VFQ at 740 K)
Sample transport properties were repeatable over the
heating and cooling cycle.
Tests are underway to look at cycling reproducibility.Mechanical Property Evaluation of Melt-Spun Materials Tensile fracture bars of developmental materials prepared by high volume manufacturing methods were tested to see of their fracture strength is improved compared to materials prepared by the standard melt-quench anneal method.
Does Finer Microstructure Lead To Stronger Materials?
Small Size Range - Allowable Stress vs. Flaw Size Large Size Range - Allowable Stress vs. Flaw Size
800 800
700 700
600 KIc = 1 600 KIc = 1
KIc = 2 KIc = 2
500 500
400 400
300 300
200 200
100 100
0 0
0 20 40 60 80 100 0 200 400 600 800 1000
Flaw Size, c (µm) Flaw Size, c (µm)
Volume type flawsMechanical Property Evaluation N-type Materials
Traditional Melt / Quench / Anneal 80 g Billets
Melt spun N-type Skutterudites 80 g Billet
The data presented here are uncensored. Meaning failure is initiated by surface, edge and volume
flaws. The test bars (95 in total) for the melt spun materials were cut into prismatic bars that
were 2 x 2 10 mm. 20 bars were tested at room temp and 15 bars at subsequent temperatures
Two distinct surfaces were present one was a matte finish the other a cut surface, both were
tested at room temperature and the matte surface was found to be weaker and so subsequent
measurements were performed on the matte surface to get a conservative estimate of the fracture
strength.
CTE of n- type material ~ 10.0 PPM to 14 .0PPM
Young’s Modulus (RUS) = 139 GPa
Poisson’s Ratio (RUS) = 0.20
These values are comparable to those of materials prepared in the standard way.Other Opportunities for Streamlining Production and Improving Properties Net-shape sintering of TE legs to reduce materials loss and increase processing through-put. Advantage of potentially eliminating edge and surface flaws that can lead to failure. Net shape sintering allows for the formation of TE legs with more exotic shapes for high packing densities Can eliminate sharp corners for TE legs which FEA shows to be concentrated centers of stress when under large temperature gradients.
Concluding Remarks/ Future Work Melt spinning combined with Spark Plasma Sintering provides a potential route to the mass production of Skutterudite based thermoelectric materials. This route seems to obviate the long term annealing processes. Consolidation of melt spun ribbons is scalable producing phase pure billets from the 5 g to ~80 g scale. There is no reason to believe that this is the limit to the size of billets that can be made. Fracture tests performed on FQ n-type materials find that them to be of comparable characteristic strength to materials made by the conventional method, with a weak temperature dependence. Thermoelectric properties of the n-type materials compare well with those published in the literature. Power factors are comparable to other Yb/Ba-filled materials produced using conventional methods. Thermal conductivity is about 25 to 30% higher for the MS/SPS case resulting in ZT’s that are ~1 to 1.1 at 740 K. The p-type materials, due to the very low thermal conductivity approach a ZT of 1 in the same temperature range. Power factors are comparable to those reported to materials made by traditional methods. Future work to include pursuing compositions and processing conditions which lower the thermal conductivity of the n-type skutterudites to levels reported for materials made by the traditional methods. Continue to optimize p-type compositions to push the onset of intrinsic conduction to higher temperatures while maintaining the very low thermal conductivity values. Measure the elastic properties CTE’s and fracture strength of p-type materials and examine scaled-up sintering operations.
You can also read
