Atomic Data Program at NIST:

databases, codes, uncertainties

Yuri Ralchenko Atomic Spectroscopy Group

National Institute of Standards and Technology

Gaithersburg MD, USA

July 28, 2016

IAEA, Vienna, Austria

Team

• Alexander Kramida

• Joseph Reader

• Karen Olsen

• Gillian Nave

• Joseph Tan

• Guest researchers

2

3

Standard Reference Data Act

Physical Reference Data Program

http://pml.nist.gov/data

Atomic data: what and why?..

• Atomic Structure

▫ Energy levels (states)

• Radiation

▫ Wavelengths

▫ Transition probabilities

• Non-radiative processes

▫ Autoionization probabilities

• Collisional processes

▫ e-A cross sections & rates

▫ A-A cross sections and rates

▫ photon-A cross sections

▫ Scattering amplitudes

• Fusion & plasma physics

• Astrophysics

• Atmospheric science

• Quantum information

• Military

• Industry

▫ Lithography

▫ Lighting

▫ Energy

▫ Health

▫ Chemistry

▫ …

5

History

• Atomic reference data compilations ▫ Charlotte Moore and others

▫ Long-term effort (no other place)

• First version of ASD: 1995 ▫ Provided online access but lacked consistency

▫ Very simplistic approach to data management

• New version of ASD: completed in late 2004

• Plasma kinetics databases added ▫ Online code FLYCHK (2006)

• Atomic Bibliographic Databases

6

Decades-old program with stable funding

Present status

• Distributed system of atomic and plasma kinetics reference data based on a modern RDBMS

• User-friendly interface with various data selection/output options and graphical capabilities

• Options to generate data online

• Numerical and bibliographic data are interconnected

• ASD is integrated into the Virtual Atomic and Molecular Data Center (VAMDC) system and International Virtual Observatory Alliance (IVOA)

7

Atomic Spectra Database v.5.3

Lines

Levels

109,000

250,100

• The only source of critically valuated atomic spectroscopic data in the world.

• The most accessed database in PML with ~2000 queries/day.

• It is regularly updated (Oct 2015) and expanded.

• It is extensively used by researchers, applied scientists, and educators in various fields of science.

8

academic, 76%

industry, 7%

gov, 1%

astro, 6% military, 10%

ASD users

energy, 13%

instruments, 12%

health, 10%

materials, 9%

semicond, 8% aerospace, 7%

defense, 7%

electronics, 7%

chemical, 6%

lasers, 4%

auto, 3%

lighting, 3%

bio, 3%

environ., 3% nuclear, 2% food, 2%

optics, 1% plasma tech, 1%

ASD user industries

NIST ASD content

ASD Citations: 2010-2014

2010 2011 2012 2013 2014

179 210 276

361 382

11

Detailed content of ASD 5.3

Levels Lines

W

Z

Ion charge

Xe Cs Ba

Mo

Fe

Th

Where does the data come from?

• There are ~30,000 published papers with measured or calculated spectral data

• Each year, ~500 more are published

• Each paper is a fragmentary study

• ASD Team evaluates each study and builds self-consistent data sets

• Bibliographical databases are essential

13

Bibliographic databases

• Principal developer: A. Kramida

• Updated regularly (~2 weeks)

• Automatic retrieval

• Atomic Energy Levels and Spectra ▫ 19,221 references, 1802-2016

• Atomic Transition Probabilities ▫ 9,069 references, 1914-2016

• Atomic Lines Broadening and Shifts ▫ 6,862 references, 1889-2016

• Annually submitted to IAEA

• Search options ▫ Elements/ions

▫ Isoelectronic sequence

▫ Word/patterm

▫ Publication years

▫ Publication source

▫ Method type

▫ Keywords

▫ General category

▫ Specific subjects of interest

http://www.nist.gov/pml/data/asbib/index.cfm

15

Critical compilation workflow

Strategy for estimating uncertainties

• Investigate internal uncertainties

• by varying model parameters and comparing results

• by comparing results (e.g., in length and velocity forms – exact principles)

• by extending the model and looking at convergence trends

• Investigate external uncertainties of the method by comparing with results of other methods

• Investigate possible contributions of neglected effects

Evaluation of transition probabilities

• Checking critical factors

• Matching calculated TP with experimental energy levels

• Matching different calculations with each other

• Selecting best TP values

• Checking for regularities

ASOS12, São Paulo, Brazil, July 2016

Critical factors in TP evaluation Theory:

• Configuration interaction

• Near coincidences of energy levels

• Cancellation effects

• Relativistic corrections

• Convergence of results and of length and velocity forms

Experiment (emission spectroscopy):

• Validity of the plasma model

• Self-absorption effects

• Spectral calibration of intensities

Experiment (lifetime measurements):

• Selective excitation, cascades

• Collisional effects and radiation trapping

• Absence of line blending

• Polarization effects and quantum beats

ASD: more than just numbers

• Visualization options: Grotrian diagrams

• Saha/Local Thermodynamic Equilibrium online-generated spectra

• Line identification spectra

19

Plasma databases

• Benchmark data from Non-LTE Code Comparison Workshops ▫ Ionization distributions ▫ Power losses ▫ Synthetic spectra

• Online collisional-radiative code FLYCHK (developed by HKC) ▫ >840 users from all over the world ▫ Calculated ionization balance and spectra from arbitrary plasmas

20

Collisional-Radiative Models

• Solve rate equations to determine atomic state populations and all relevant parameters

▫ Ionization balance

Mean ion charge 𝑍

Central moments

▫ Spectral emission

▫ Power losses

• CRMs can be very different!

tNtA

dt

tNd ˆˆˆ

What data?..

• Energy levels (different nature, e.g., m-sublevels, levels, terms, configurations, superconfigurations)

• Radiative rates (Einstein coefficients or oscillator strengths)

• Autoionization rates

• Collisional cross sections or rate coefficients

▫ Electron-impact (de)excitation and ionization

▫ Photoexcitation, photoionization, photorecombination

▫ Three-body recombination

▫ Dielectronic capture or dielectronic recombination

▫ Heavy-particle collisions

▫ …

Non-LTE Code Comparison Workshops

• Goal: to benchmark CR models against ideal cases (practically no “clean” experiments)

• Models may differ in various parameters, e.g., atomic structure, number of states, nature of states, quality of atomic data, etc.

• 15-20 codes, 20-25 participants

• Last: NLTE-9, Paris, 2015

24

Data exchange

• SLAP

• XSAMS (next talk, Christian Hill)

• NLTE Code Comparison workshops

▫ ASCII files, relatively free format, GB’s of data

Ionization distributions

Radiative power losses

Ion and level populations

Rates of physical processes

Spectra

…

25

Monte Carlo analysis

• Generate a (pseudo-)random number between 0 and 1

• Using Marsaglia polar method, generate a normal distribution

• Randomly multiply every rate by the generated number(s)

• To preserve physics, direct and reverse rates (e.g. electron-impact ionization and three-body recombination) are multiplied by the same number

• Ionization distribution is calculated for steady-state approximation

σ2 vs mean ion charge

27