1. A first search

You want the papers that simulate Hall thrusters with particle-in-cell codes. One command asks every database mosaic can reach:

$ mosaic search "hall thruster particle-in-cell" --max 5

 #     Title                             Authors       Year   DOI                 Source         OA     PDF       Cited
────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
 1     Plasma Diagnostic and             Jose          2004   10.48550/arXiv.ph   arXiv          yes    ✓             –
       Performance of a Permanent        Leonardo             ysics/0410170
       Magnet Hall Thruster              Ferreira et
                                         al.
 2     Reduced-order particle-in-cell    Maryam Reza   2023   10.1088/1361-6595   arXiv          yes    ✓             –
       simulations of a high-power       et al.               /acdea3
       magnetically shielde…
 3     Effects of the neutral dynamics   Farbod        2023   10.1063/5.0143292   arXiv          yes    ✓             –
       model on the particle-in-cell     Faraji et
       simulations of a …                al.
 4     Dimensionality reduction of       Maryam Reza   2022   10.48550/arXiv.22   arXiv          yes    ✓             –
       Poisson's equation with           et al.               06.13892
       application to particle-in-…
 5     Enhancing one-dimensional         Farbod        2022   10.1063/5.0090853   arXiv          yes    ✓             –
       particle-in-cell simulations to   Faraji et
       self-consistently res…            al.
 6     PIC-DSMC Simulation of a Hall     Sang Hun      2023   10.3390/aerospace   DOAJ           yes    ✓             –
       Thruster Plume with Charge        Kang                 10010044
       Exchange Effects Using…
 7     Effects of Magnetic Field         Qi Liu et     2023   10.3390/aerospace   DOAJ           yes    ✓             –
       Gradient on the Performance of    al.                  10110942
       a Magnetically Shielde…
  ⋮  (101 lines not shown)
 41    Modeling of surface-dominated     Taccogna F    2008   10.1063/1.2802565   PubMed         no     –             –
       plasmas: from electric thruster   et al.
       to negative ion s…
41 result(s)

• --max 5 asks each source for five results (the default is ten). A dozen sources answered, so the table has far more rows than five, after mosaic merged the papers that more than one source returned.
• The sources run in parallel, up to eight at a time, behind a spinner. A search takes 10–30 seconds, most of it spent waiting for the slowest database. Rows come in the order the sources answered.
• When two sources return the same paper, it appears once, with the richer of the two records. Papers are the same when their DOI matches, or their arXiv ID, or (with neither) their title.
• OA says whether the paper is open access, PDF whether mosaic knows a link to its PDF. Cited appears when a source reported citation counts.

Which sources answered

With no configuration, mosaic queries the sources that need no key: arXiv, Semantic Scholar, DOAJ, Europe PMC, Cochrane, OpenAlex, BASE, Zenodo, Crossref, DBLP, HAL, PubMed, PubMed Central and bioRxiv/medRxiv. The others (ScienceDirect, Scopus, Springer Nature, IEEE Xplore, CORE, NASA ADS) join when you give them a free API key or a browser session; Sources lists what each one needs.

A source that fails does not stop the search: the others still answer, and the failure stays hidden. Put --verbose before the command to see why a source is missing:

$ mosaic --verbose search "hall thruster particle-in-cell" --max 5 2>&1 | grep Warning
Warning: DBLP: answered with a web page instead of JSON (a bot check); try again later
Warning: Semantic Scholar: rate limited (HTTP 429); try again later

Without an API key, Semantic Scholar shares a public quota and often answers HTTP 429, "too many requests"; mosaic retries short waits before it gives up. DBLP sometimes answers scripts with a "making sure you're not a bot" page instead of results.

The query

Each source receives the query in its own syntax. For most of them it is a full-text search of title, abstract and keywords, so this broad query also finds papers that only mention a Hall thruster in passing. The next chapter narrows it.

What you learned

mosaic search asks every enabled source at once, merges duplicates and prints one table; --max counts per source; --verbose shows the sources that failed. Reference: search, Sources.

Next: 2. Narrowing the results.