[3dem] MicroED on small molecule

[Takanori Nakane]

Hi,

Looking at your crystals, I feel the crystal density is OK.
Certainly several smaller crystals are attached at the bottom
of the main plate. However, they are much smaller than the main
plate and their orientation is different. Thus, I guess the dirty
spot shape is caused by the property of the main, plate-shaped
crystal itself. Indeed, it has wrinkles on it.

Regarding data processing:

DIALS can index multiple-lattice and (optionally) remove overlapped
spots but cannot deconvolute overlapped intensities.
I myself process the dominant lattice alone.
Because I collect 100 - 500 crystals and work at a very high
multiplicity (usually 50 - 300), I can select better crystals
with less overlaps. Even if some reflections are overlapped,
their intensities are averaged out and/or rejected during
scaling and merging.

I agree that ideally multiple lattices should be avoided.
But whether this is practically possible and worthwhile
depends on the goal of the experiment and the availability
of the scope time.

In our case, the goal of most experiments is the understanding
of packing and conformations. We don't pursue very high
accuracy in molecular geometry. Moreover, other sources
of systematic errors (multiple scattering, lens hysterisis,
inaccurate camera length) probably have bigger influences
on the final model accuracy.

Another issue is that we have only one weekend beam time in three to
four weeks. In one session, we can load up to 9 grids of
various compounds (bottom three slots of a cartridge tend to
get frosted) in one go, only once. Thus, we cannot try-and-error
various pre-treatments, optimizing solvents and sonication duration.
In my experience, even when a chemist says a solvent (say, hexane)
is a poor solvent and does not dissolve a compound, the crystal
often melts a bit on the surface and diffraction becomes worse.
Sonication and grinding also damages crystals often.
This depends on the crystal; some crystals are
hard and very robust. We need to test and optimize various parameters.
But we cannot afford that, unless absolutely necessary.

TFS scopes with an autoloader are great for automatic data collection
over night or over weekend but not convenient for successive grid
exchanges, as all grids have to be pulled out simultaneously.
JEOL scopes are better in this regard, as they allow swapping
only a subset of the loaded grids.

Best regards,

Takanori Nakane

________________________________________
From: 3dem <3dem-bounces at ncmir.ucsd.edu> on behalf of Yu, Xiaodi [JRDUS] 
via 3dem <3dem at ncmir.ucsd.edu>
Sent: 16 January 2026 23:39
To: Tim Gruene; Yu, Xiaodi [JRDUS] via 3dem
Subject: Re: [3dem] MicroED on small molecule

Thanks everyone — these are very helpful suggestions: recrystallization, 
resuspending the sample in solution (with or without sonication), 
sliding glass plates, and more. We’ll try each approach. I hope this 
thread helps not just our research but others facing the same trouble. 
I’ve uploaded a couple of images showing the issue; we’re using a 
Glacios without SA.

Best,

Xiaodi

________________________________________
From: Tim Gruene
Sent: Friday, January 16, 2026 4:32 AM
To: Yu, Xiaodi [JRDUS] via 3dem
Cc: Yu, Xiaodi [JRDUS]
Subject: [EXTERNAL]  Re: [3dem] MicroED on small molecule

On Thu, 15 Jan 2026 18:37:05 +0000
"Yu, Xiaodi \[JRDUS\] via 3dem" <3dem at ncmir.ucsd.edu> wrote:

> Hi everyone,
>
> I’d like to consult MicroED experts about a small‑molecule sample:
> the powder contains diffracting microcrystals, but they stick
> together and make data collection difficult. Does anyone have
> recommended approaches or tips to prevent clumping and improve data
> collection?
>
> Thanks in advance,
> Xiaodi
>

Dear Xiaodi,

my experience is slightly different from Takanori's. We put a lot of
effort in sample preparation, so as to get as good data as possible
rather than leaving it to the software to deconvolute multiple
lattices. No matter how good the software (even talking SAINT/TWINABS
;-), a dataset from multiple lattices is hardly ever as good as a single
crystal dataset.

Our goal is a sufficiently dilute crystal density on the grid so to
have isolated particles. Like Takanori, match the beam diameter to the
crystal size. For organic and metalorganic samples, this can be 2-3um
in diameter. Smaller diameter is a greater challenge for how good you
can centre the crystal and the stability of the rotation axis.

My personal favourite is a suspension, because it is reproducible. The
liquid must match two requirements:
- it evaporates
- it does not modify the crystal structure

In suspension, we either vortex the sample, or place it in a sonication
bath. This can be up to several minutes, depending on how strongly they
stick. The lab record for a zeolite was about 1-2 hourse.

n-hexane is a good candidate to start, but also isopropanol, MeOH,
EtOH. You can check water, too.

Sometimes particles stick together while you watch the solvent
evaporate. To reduce their time of socialization, you can either dip
the grid vertically to the suspension surface.

Or you add a drop to the grid and suck off the supernatant from
underneath: lacey carbon makes a great sieve for crystals.

By creating a dilution series, we typically get to the ideal density
with 3-5 trials.

Good luck!

Tim

P.S.: in Europe, there is a great number of installation of electron
diffractometers (more than a dozen at least), and the Bruker mailing
list or the CCP4bb may also be great resources for crystallographic
questions. Crystallography is not an imaging techniques, while
TEM is.

--
--
Tim Gruene
Head of the Core Facility Crystal Structure Analysis
Faculty of Chemistry
University of Vienna

Phone: +43-1-4277-70202

https://urldefense.com/v3/__https://ccsa.univie.ac.at__;!!Mih3wA!HD3edpvrNnrwqJcjr0poHrj-8C7Vz9d26HDyYgX2ofMcuF0eliP_uLCj50AyQWhK91-uVk1ThTy0qqs2$

GPG Key ID = A46BEE1A