BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260421T090513Z
LOCATION:Bldg. 6 - Room 002
DTSTART;TZID=Europe/Stockholm:20260630T141500
DTEND;TZID=Europe/Stockholm:20260630T144500
UID:submissions.pasc-conference.org_PASC26_sess141_msa111@linklings.com
SUMMARY:Data-Efficient Multiscale Learning for Catalytic Property Predicti
 on on Amorphous Surfaces
DESCRIPTION:Xuewei Zhang, Mattia Turchi, and Ivan Lunati (Empa)\n\nAmorpho
 us silica (a-SiO2) plays a key role in catalysis and gas adsorption. Under
 coordinated surface defects enhance reactivity and gas adsorption and can 
 serve as anchoring sites for transition metals. However, the intrinsic str
 uctural disorder of a-SiO2 poses significant challenges for conventional a
 tomistic analysis and systematic structure-property mapping. Here, we pres
 ent a multiscale machine learning framework that integrates classical mole
 cular dynamics (MD) with density functional theory (DFT) to predict cataly
 tic properties across statistically representative a-SiO2 surfaces. Local 
 atomic environments (LAEs), extracted from MD-generated structures, are ch
 aracterized using the MACE structural descriptors. After dimensionality re
 duction, clustering is applied to distinguish different non-bridging oxyge
 n (NBO) defects from non-defective oxygen sites without misclassification.
  Catalytic properties computed via DFT for representative NBO clusters are
  then used to train a surrogate Gaussian process regression model that map
 s descriptors of the LAEs onto the property of interest, such as the bindi
 ng energies of transitional metals. The model is iteratively refined throu
 gh active learning, improving predictive accuracy while minimizing the num
 ber of expensive DFT evaluations, enabling efficient and scalable explorat
 ion of catalytic activity of disordered surfaces.\n\nDomain: Chemistry and
  Materials, Computational Methods and Applied Mathematics\n\nSession Chair
 s: Mattia Turchi (Empa) and Ivan Lunati (Empa)\n\n
END:VEVENT
END:VCALENDAR