---
title: "Ferritin crystals form in acoustically levitated droplets; lower MW PEG preserves crystalline order."
sdDatePublished: "2026-09-11T05:09:00Z"
source: "https://arxiv.org/pdf/2609.11226"
topics:
  - name: "science and technology"
    identifier: "medtop:13000000"
locations:
  - "Lund"
  - "Stockholm"
  - "Germany"
---


Ferritin crystals form in acoustically levitated droplets; lower MW PEG preserves crystalline order.

Kinetics of ferritin crystal formation and melting in acoustically levitated droplets

Kinetics of ferritin crystal formation and melting
in acoustically levitated droplets
Anita Girelli,∗,† Iason Andronis,† Aigerim Karina,† Sampad Bag,† Lennart
Bergstr¨om,‡ Tom´as S. Plivelic,¶ Felix Roosen-Runge,§ and Fivos Perakis∗,†
†Department of Physics, AlbaNova University Center, Stockholm University, 10691
Stockholm, Sweden
‡ Department of Chemistry, Stockholm University, 10691 Stockholm, Sweden
¶ MAX IV Laboratory Lund University Sweden
§ Division of Physical Chemistry, Lund University, Naturvetarv¨agen 22, 22100 Lund,
Sweden
E-mail: anita.girelli@fysik.su.se; f.perakis@fysik.su.se
1
arXiv:2609.11226v1 [cond-mat.soft] 10 Sep 2026

Abstract
Understanding protein crystallization pathways is essential for controlling crys-
tallization in structural biology, materials science, and pharmaceutical applications.
Classical nucleation theory does not fully capture crystallization processes for several
proteins, including ferritin. Here, we combine acoustic levitation with small- and wide-
angle X-ray scattering (SAXS and WAXS) to monitor ferritin crystallization in evapo-
rating aqueous polyethylene glycol (PEG) solutions. Acoustic levitation rapidly drives
the droplets through a broad range of protein and polymer concentrations, enabling
time-resolved measurements of crystallization during evaporation. The scattering data
show that ferritin crystals form during evaporation and subsequently lose their crys-
talline order upon further dehydration. Varying the PEG molecular weight switches
between distinct crystallization pathways: one dominated by attractive protein–protein
interactions and another dominated by repulsive interactions and excluded-volume ef-
fects. Furthermore, we find that lower-molecular-weight PEG (1000 g/mol) suppresses
the dehydration-induced loss of crystalline order observed for higher-molecular-weight
PEG (6000 g/mol), providing a simple strategy for improving protein crystal stability.
TOC Graphic
2

Introduction
Protein crystallization underpins structural biology and numerous industrial separation pro-
cesses,1 yet predicting and controlling crystallization pathways remains a major challenge.2
Traditional models have largely relied on classical nucleation theory.3 Classical nucleation
theory describes crystallization as a stochastic process in which individual monomers cluster
until a critical nucleus forms, after which crystal growth proceeds. Although this frame-
work has provided useful information, it has proven insufficient to explain the full range of
crystallization phenomena observed in proteins.3,4 Increasing evidence points to alternative
non-classical pathways that involve metastable intermediate phases, liquid–liquid phase sep-
aration, or amorphous precursors.4–6 Such processes can strongly influence nucleation rates,
polymorph selection, and ultimately the quality of the resulting crystals.
Ferritin, an iron-storage protein with a highly symmetric cage-like architecture, has
emerged as a particularly intriguing case: Recent studies have demonstrated that ferritin un-
dergoes a non-classical crystallization pathway when crystallized in the presence of CdCl2.7,8
Instead of following a two-step nucleation model, ferritin molecules first assemble into amor-
phous aggregates,8 which then undergo structural rearrangements until a crystalline order
is reached.
Ferritin has been reported to crystallize not only in the presence of CdCl2, but also in
the presence of polyethylene glycol (PEG).9 This polymer is widely used as a crystallizing
agent,10,11 as it induces liquid-liquid phase separation12 via depletion interactions13 which
can enhance crystallization,14 and is associated with a two-step crystallization mechanism
in which the crystals form at the expense of liquid-like aggregates.15
To probe ferritin crystallization in situ, acoustic levitation provides a powerful container-
less method for studying nucleation processes at high supersaturation.16–18 This approach
eliminates heterogeneous nucleation from container walls and enables the use of minimal
protein volumes while accessing a wide range of concentrations during the evaporation pro-
cess.
3

In this work, we used acoustically levitated droplets to monitor the kinetic pathways of
ferritin aggregation, crystal nucleation, and crystalline ordering under controlled conditions,
varying PEG concentration and molecular weight (PEG1000, with 1000 g/mol, and PEG6000
with 6000 g/mol). By combining Small-Angle X-ray Scattering (SAXS), Wide-Angle X-ray
Scattering (WAXS), and optical microscopy, we captured the evolution from dispersed pro-
teins through concentrated intermediate states to crystalline order, and mapped the interplay
between crowding, protein–protein interactions, and dehydration. This approach enables the
estimation of a state diagram of ferritin crystallization and provides direct insight into the
mechanisms of crystallization.
Results and Discussion
Droplets of protein solution with radius ≈1mm (with volume ≈10µL) were injected into an
ultrasonic acoustic levitator and simultaneously probed with X-ray scattering and optical mi-
croscopy. The volume of the droplets was estimated from time-resolved optical images, while
the molecular structure of the protein and water were monitored through X-ray scattering.
Figure 1a shows the volume of a droplet of ferritin and polyethylene glycol (PEG) in saline
solution (initial concentrations of cprotein
i
= 50 mg/mL Ferritin, cPEG
i
= 3 w/v % PEG6000
and 150 mM NaCl) as a function of evaporation time t (Fig. 1a). Representative microscopy
images from which the droplet volume was estimated are shown in Fig. 1e-g. During the
collection of the microscopy images, the X-ray signal was collected by two detectors to cover
both SAXS and WAXS ranges.
Representative SAXS and WAXS curves are shown in Fig. 1b and Fig. 1c respectively.
The colors from blue to yellow indicate different evaporation times at which they were
collected, with t = 0 s being the time at which the droplet is injected into the levitator. The
corresponding estimated volumes are shown in Fig. 1a as circles with corresponding colors.
In Fig. 1a, the estimated volume decreases linearly with time in the first 600 s, and after a
4

Figure 1: Time evolution of an acoustically levitated ferritin droplet during evaporation.
The initial solution contained 50 mg/mL ferritin, 3% (w/v) PEG6000, and 150 mM NaCl.
(a) Droplet volume as a function of evaporation time t. b) SAXS intensity profiles as a
function of momentum transfer q at different evaporation times t. The curves are shifted
for clarity. c) Corresponding WAXS intensity profiles. d) Protein structure factor obtained
by dividing the SAXS intensity by the ferritin form factor. e-g) Optical microscopy images
corresponding to the labeled time points in panel (a). Colors indicate evaporation time and
are shared across panels (a–d).
turning point at 1100 s, it becomes almost constant.
At early times (Fig. 1b, dark violet curve), the SAXS profile is close to that of ferritin
single proteins in solution. At intermediate evaporation times, the SAXS intensity increases
at low q, while at t = 700 s protein Bragg peaks appear at q ≈0.6 nm−1, q ≈0.7 nm−1,
q ≈1.0 nm−1, as well as additional smaller peaks at higher q. Interestingly, from t ≈1200 s
the height of the Bragg peaks decreases, until t ≈1400 s when they are absent. In the
WAXS signal (Fig. 1c), in the earlier times the main contribution is due to two water peaks
at q ≈20 nm−1 and q ≈28 nm−1. With increasing time t, the water peaks decrease in height
5

as the water evaporates, which is also evident by the volume changes with time t in Fig. 1a.
At later times, two additional sharp peaks appear at q ≈13.7 nm−1 and q ≈16.5 nm−1,
which correspond to Bragg peaks from PEG semi-crystals.19 At higher q-values, the curve
presents three peaks at q ≈25, 28 and 32 nm−1. We attribute these peaks to ferrihydrite
(6LFh) contained in the protein core together with magnetite and hematite.20–22
The experimental protein structure factor S(q), shown for representative conditions in
Fig. 1d and for the different sample compositions in the SI, Fig. S7, was estimated by dividing
the scattering intensity I(q) by the form factor P(q). P(q) was extracted by measuring a
dilute protein solution (with a concentration of 10 mg/mL) in a capillary; see Supplementary
Information section ”Ferritin Form Factor” for more information. The signal was normalized
to ensure that the oscillations of the structure factor at large q were around 1, i.e. S(q →
∞) = 1.
To understand the kinetics of ferritin crystallization, we need to monitor the different
components of the solution: water, proteins in solution, proteins in crystals, and PEG. The
average ferritin concentration cprotein in the droplet can be estimated from the changes in
droplet volume with time. Fig. 2a on the left-hand side y-axis shows the time dependence
of the normalized volume V (t)/V (t = 0) for samples at different initial concentrations of
PEG (cPEG
i
= 0, 3, 5, 7 w/v % ).
The right-hand side y-axis shows the corresponding
average protein concentration in the droplet. To compare similar evaporation stages (that
is, comparable protein concentration), time t was normalized to t∗, which is the turning point
where volume changes are minimal. Before t∗, the volume changes follow a D2-law,23,24 which
states that the square of diameter D changes linearly with time. This master curve implies
that, at the same normalized time, the droplets have the same protein concentration. After
t∗, the volume decreases linearly with time with a shallow slope, implying that the volume
(and therefore the protein concentration) is almost constant. The final protein concentration
depends on cPEG
i
, with larger values of cprotein for lower cPEG
i
.
To monitor the presence of crystals, we estimate the amplitude of the protein crystals
6

as S(q = 0.6 nm−1), i.e. the structure factor at the first protein Bragg peak (Fig. 1b). The
amplitude increases at earlier normalized times for a higher initial PEG concentration cPEG
i
(Fig. 2b). In contrast, the amplitude rapidly decreases at t/t∗≥0.9 (indicated by the black
dashed line) for all cPEG
i
> 0. This indicates a loss of crystalline order once the protein
concentration reaches approximately cprotein ≈400-500 mg/mL.
The behavior of PEG in the solution is analyzed in Fig. 2c. The WAXS intensity at
qpeak = 13.7 nm−1 provides information on the crystalline ordering of PEG that develops
at low hydration.
In semicrystalline PEG, polymer chains partially fold into crystalline
lamellae separated by amorphous regions.19 The molecular ordering within the crystalline
regions gives rise to a Bragg peak at qpeak. To separate the crystalline PEG contribution
from the underlying water scattering, its amplitude was estimated as I(qpeak) −I(qoff−peak)
with qoff−peak = 14 nm−1. Fig. 2c shows that this amplitude is constant around 0 until time
t/t∗≈1. Afterwards, it increases linearly. This indicates that PEG molecules do not form
semi-crystals up to the time when the volume is almost constant, which is after the protein
crystals have melted. These observations suggest that PEG semi-crystallization is unlikely
to be responsible for the loss of crystalline order.
To understand the mechanism of formation of ferritin crystals, protein interactions were
estimated as the structure factor at q = 0.1 nm−1. Further details of the SAXS analysis
and representative fits are provided in the Supplementary Information (Fig. S8). At this q,
the length-scale probed is around 60 nm, i.e. corresponding to the size of several proteins
(2Rprotein ≈12 nm). Attractive protein–protein interactions result in S(q = 0.1 nm−1) > 1,
while repulsive interactions in S(q = 0.1 nm−1) < 1. In Fig. 2d, S(q = 0.1 nm−1) increases
linearly from 1 for all cPEG
i
> 0. It reaches its maximum with a value