2.1. Preparation of DNA–Functionalized Nanoparticles

DNA–AuNPs were prepared according to a previously reported method ^29,30. Single-stranded DNA with a thiol group at the 5\('\) end (Nihon Gene Research Laboratories Inc., Japan) was added to a dispersion of spherical gold nanoparticles with a diameter of 20 nm (Nanopartz).

DNA was attached to the nanoparticle surface using a salt-aging process in which the NaCl concentration was gradually increased. Excess DNA was removed through centrifugation, and the particles were redispersed in phosphate buffer. Four different DNA strands were used in this study. Two thiolated-DNA strands, #1 and #2, were attached to the AuNP surfaces to prepare two types of DNA–AuNPs. Two linker-DNA strands, #3 and #4, were designed to hybridize with the particle-bound DNA strands, thereby bridging the two types of DNA–AuNPs and inducing superlattice formation. Details on the DNA strands can be found in our previous study ²⁹.

2.2. Droplet Generation and Crystallization

DNA–AuNP superlattices were formed in crystallization droplets dispersed in oil, following our previously reported droplet-based approach ³¹. The droplets were generated using a microfluidic device fabricated by standard soft lithography. A mold was prepared by patterning SU-8 photoresist on a substrate, and polydimethylsiloxane was cast on the mold to replicate the microchannel structures. The molded PDMS was then bonded to a glass substrate. The solution and oil phases were introduced into the device using a pressure-driven flow control system (Flow EZ, Fluigent, LU-FEZ-7000).

The DNA–AuNP and linker-DNA solutions were mixed in the microchannel at a flow rate ratio of \(1:1\). The mixed solution was subsequently sheared using an oil phase consisting of isopropyl palmitate containing 7 wt% ABIL EM180 (Evonik Operations GmbH), resulting in droplet formation. The flow rates of the crystallization solutions were maintained constant, whereas the oil flow rate was varied under two different conditions for each salt concentration. The flow rates of the two crystallization solutions were each set to 1 μL/min. The oil flow rates were set to 1 and 2 μL/min for NaCl concentrations of 0.5, 1.0, and 2.0 M and to 1 and 4 μL/min for the 0.25-M condition to obtain droplets with comparable size ranges.

The condition with the highest oil flow rate was defined as the small-droplet (S) condition, whereas the condition with the lowest oil flow rate was defined as the large-droplet (L) condition. The final concentrations in the droplets were 0.005 pmol/μL for each DNA–AuNP strand (#1 and #2) and 2 pmol/μL for each linker-DNA strand (#3 and #4). The NaCl concentrations in the crystallization solutions were set to 0.25, 0.5, 1.0, and 2.0 M.

The obtained droplets were cooled from 65°C to 25°C at a rate of 0.1°C per 10 min using a temperature-controlled chamber (IS401, Yamato), resulting in the formation of DNA–AuNP superlattices.

2.3. SAXS and SEM Characterization

Structural characterization of the DNA–AuNP crystals were performed using SAXS. SAXS measurements was performed using the structural biology beamline BL40B2 at SPring-8. The X-ray source was a bending magnet; the camera length was 4 m; and the X-ray wavelength was 1.1 Å. Scattering patterns were recorded using a two-dimensional detector (Pilatus3 S2M). For SAXS measurements, the crystallized droplet samples were loaded into a quartz capillary (OD 1.0 mm, Hilgenberg), and scattering patterns were obtained through X-ray irradiation.

For SEM observation, the crystals were prepared using a silver nitrate treatment according to a previously reported method ³². This treatment stabilizes the DNA-bridged structure and preserves crystal morphology under vacuum conditions. Specifically, the crystals collected from the oil droplets were transferred to a phosphate buffer with the same NaCl concentration as that of the crystallization condition. Silver nitrate was then added, and the reaction was allowed to proceed at 25°C. After the reaction, 0.5 μL of the sample was dropped onto a Si substrate, dried, and observed using a scanning electron microscope (S-4800, Hitachi High Technologies) at an accelerating voltage of 5 kV and a beam current of 5 μA.

Images of the droplets were obtained using an optical microscope (IX83, Olympus). The images were analyzed to determine the droplet diameters and evaluate the sizes of the crystals formed in each droplet.

2.4. Ultraviolet–Visible (UV–VIS) and Dynamic Light Scattering (DLS) Measurements

UV–VIS absorbance measurements were performed to evaluate the temperature-dependent behavior of the DNA–AuNP solution. A spectrophotometer (V-730BIO, JASCO) equipped with a water-cooled Peltier cell holder (ETCS-761, JASCO) was used for the measurements. The sample was sealed in a capillary tube (CAP-100Q, JASCO). The absorbance at 258 nm was monitored while cooling the sample from 65°C to 25°C at a rate of 0.1°C/min.

DLS measurements were performed to evaluate the diffusion behavior of the DNA–AuNP solution. A zeta potential and particle size analyzer (ELSZ-2000ZS, Otsuka Electronics) was used for the measurements. The sample was sealed in a micro cell and measured while cooling from 65°C to 25°C at a rate of 1°C/min, with measurements taken at intervals of 1°C–5°C.

3.1. Crystallization Behavior in Microdroplets

DNA–AuNP crystallization in droplets was first examined using optical microscopy. Fig. 1(a) shows a representative optical micrograph of droplets after crystallization (0.5 M (L)). Crystals were observed as high-contrast objects inside individual droplets, and in most droplets, a single crystal was formed. This observation suggests that crystallization proceeded independently in each droplet.

The size distribution of droplets obtained under each NaCl concentration and flow condition is shown in Fig. 1(b). The generated droplets had diameters ranging from approximately 18–27 μm. The average droplet diameter was approximately 25 μm for the L condition and approximately 21 μm for the S condition. The coefficient of variation (\(\mathrm{CV} = \textrm{standard deviation}~/~\textrm{mean}\)) ranged from 4.5% to 9.9%.

The size distribution of crystals formed within the droplets is shown in Fig. 1(c). Crystal sizes were determined based on optical micrographs after crystallization. Most crystals exhibited an approximately hexagonal projected shape, consistent with a projection of the rhombic dodecahedral equilibrium morphology expected for a bcc crystal. Therefore, the projected shape of each crystal was approximated as a regular hexagon, and the crystal size was defined as the length of its long diagonal. The crystal size distribution was relatively narrow for each condition, with the coefficients of variation ranging from 4.6% to 11.3%.

Comparison of crystal sizes across different NaCl concentrations should be interpreted with caution, because NaCl concentration affects not only the final crystal size but also the lattice constant and crystalline order, as shown in the following sections. Therefore, crystal sizes measured under different NaCl concentrations are not directly comparable based on their external dimensions alone. To evaluate the effect of droplet size independently of these salt-dependent structural differences, the L and S conditions were compared primarily within the same NaCl concentration, whereby differences in droplet size were accompanied by corresponding differences in crystal size. For NaCl concentrations other than 2.0 M, the average crystal size in the large-droplet samples was consistently larger than that in the small-droplet samples by approximately 0.3–1.0 μm. In contrast, at 2.0 M, the average crystal size in the small-droplet samples was approximately 0.3 μm larger than that in the large-droplet samples, which may reflect the variation in droplet size. Overall, within each NaCl concentration, the relative differences in crystal size generally followed those in droplet diameter, suggesting that the available droplet volume influenced the final crystal size.

These results indicate that droplet-based crystallization enables the formation of crystals with relatively low size dispersity, while within the same NaCl concentration, the final crystal size can be partially tuned by varying the droplet size. This controllability is advantageous for preparing crystals with dimensions suited to the intended intracellular delivery method.

3.2. Effect of Salt Concentration on Crystallinity

To evaluate the crystal structure formed in droplets, SAXS measurements were performed for samples prepared under different NaCl concentrations. Fig. 2(a) compares the SAXS profiles obtained for small-droplet samples. All samples exhibited diffraction peaks consistent with a bcc lattice, indicating that the DNA–AuNP crystals formed in the droplets adopted the same bcc structure over the investigated NaCl concentration range. However, systematic changes in the diffraction profiles were observed with increasing NaCl concentration. In particular, the number of observable higher-order peaks gradually decreased. In addition, the first diffraction peak broadened and shifted toward higher \(q\)-values at higher NaCl concentrations. These changes indicate a reduction in crystalline order and a decrease in the lattice constant as the NaCl concentration increases.

To quantitatively evaluate the peak broadening observed in the diffraction profiles, lattice disorder was assessed using the g-factor based on Hosemann’s paracrystalline theory ³³. The g-factor is a dimensionless parameter that represents the magnitude of relative lattice distortions; an ideal crystal with no positional or lattice spacing disorder has \(g = 0\), and the value increases with increasing lattice disorder. In this study, the analysis was performed for the first diffraction peak of the structure factor obtained from SAXS. The peak profile was fitted with a pseudo-Voigt function while accounting for crystallite size broadening. From the fitting results, the distortion-related peak width \(\beta_\mathit{dist}\) was extracted, and the g-factor was calculated using the corresponding relation ²⁹. The resulting g-factor values for each NaCl concentration are shown in Fig. 2(b). Increasing g-factor with increasing NaCl concentration indicates greater lattice disorder in the DNA–AuNP crystals formed at higher ionic strengths. Higher salt concentration stabilizes DNA duplex formation and alters interparticle interactions ^15,17,18, which may promote particle attachment during crystal growth. If particle incorporation outpaces structural rearrangement, kinetically trapped configurations may remain within the growing crystal ^34,35. The resulting positional and lattice spacing disorder would broaden the diffraction peaks and reduce their intensities, with the effect on higher-order reflections being more pronounced ³³.

3.3. Morphology of DNA–AuNP Crystals

To further examine the influence of NaCl concentration on crystal morphology, SEM observations were performed for crystals obtained at NaCl concentrations of 0.5 and 2.0 M (Fig. 3). At the lower NaCl concentration (i.e., 0.5 M), many crystals exhibited well-defined polyhedral shapes consistent with rhombic dodecahedra (Fig. 3(a)). The nanoparticles on the crystal surfaces appeared to be relatively well ordered, and step-like features were frequently observed on the crystal facets (Fig. 3(b)). In contrast, at the higher NaCl concentration (i.e., 2.0 M), many crystals exhibited rounded shapes without clearly defined crystal facets (Fig. 3(c)). In addition, the nanoparticle arrangement on the crystal surfaces appeared less ordered, and regions with locally disordered nanoparticle arrangements were occasionally observed (Fig. 3(d)).

These observations demonstrate that crystal morphology and surface organization vary with NaCl concentration, suggesting differences in the crystal growth pathway. At 0.5 M NaCl, the step-like surface features suggest a step-mediated growth mechanism ³⁶. Under this condition, anisotropic growth, in which different crystallographic faces grew at different rates, may have led to the formation of well-defined faceted crystals ³⁷. In contrast, the rounded morphology and disordered surface observed at 2.0 M NaCl may reflect a condition in which particle incorporation proceeded more rapidly relative to the structural rearrangement required for ordering, thereby favoring kinetically trapped disorder ^34,35. Together with the SAXS results in Section 3.2, these observations indicate that the loss of well-defined external faceting at high NaCl concentration was accompanied by increased disorder within the nanoparticle lattice. These correlated changes suggest that the ionic environment influenced both crystal morphology and internal structural order likely through changes in the crystal growth pathway.

3.4. Relationship Between Ionic Environment and Crystallization Behavior

To investigate the effect of NaCl concentration on the self-assembly behavior of DNA–NP in solution, temperature-dependent UV–VIS absorbance measurements and DLS measurements of the diffusion coefficient were performed.

Figure 4(a) shows the temperature dependence of the absorbance at 258 nm during cooling for the 0.5 and 2.0 M NaCl conditions. UV–VIS absorbance was measured during cooling from 65°C to 25°C. Because the absorbance was unstable during the initial stage of cooling, only data obtained in the range of 60°C–25°C after temperature stabilization were used for normalization and analysis. For both samples, the absorbance changed abruptly at a characteristic temperature during cooling. This transition is attributed to a change in the association state of the DNA–AuNPs induced by DNA hybridization. Notably, the transition occurred at a higher temperature for the 2.0 M NaCl sample compared with that for the 0.5 M sample. This shift is consistent with the greater stability of the DNA duplexes at higher salt concentrations ¹⁵.

The diffusion coefficients of the nanoparticle solutions during cooling were also measured using DLS, and the results are shown in Fig. 4(b). Under comparable temperature conditions, the diffusion coefficient for the 2.0 M NaCl solution was consistently smaller than that for the 0.5 M solution. This difference may reflect changes in the effective interparticle interactions that depend on salt concentration ^17,18; nevertheless, the influence of changes in solution viscosity cannot be ruled out.

Overall, the UV–VIS and DLS measurements suggest that the association and dispersion states of the DNA–AuNPs depend on the NaCl concentration. UV–VIS and DLS probe different aspects of the same DNA-mediated assembly process: UV–VIS absorbance at 258 nm mainly reflects the average changes in DNA hybridization and base stacking, whereas DLS reflects changes in the hydrodynamic size of the particle assemblies. Therefore, the relative order of these changes provides information on how DNA binding and particle assembly proceed during cooling. In the 2.0-M NaCl condition, the absorbance change appeared at a higher temperature compared with that of the diffusion coefficient change, suggesting that DNA-mediated binding had already progressed appreciably before the formation of particle assemblies large enough to be detected by DLS. Under such conditions, particle assembly may proceed after DNA-mediated interparticle binding has become relatively stable, which may influence subsequent particle rearrangement. In contrast, in the 0.5-M NaCl condition, the diffusion coefficient change appeared at a higher temperature compared with that of the UV–VIS change. This behavior suggests that partial DNA-mediated binding may first produce small clusters that are detectable by DLS, followed by further DNA hybridization and growth over a broader temperature range. Such a gradual assembly process is consistent with the possibility that particle rearrangement occurs during growth. Although these measurements do not directly resolve nucleation and growth within individual droplets, they indicate that the solution state of the DNA–AuNPs varies with NaCl concentration. These differences may contribute to the salt-dependent crystallinity and morphology of the resulting crystals.

For future application of the obtained crystals to intracellular sensing, appropriately designing both the delivery method and the post-crystallization solution conditions is essential. Although the crystals obtained in this study, with sizes of approximately 1.5–2.5 μm, are relatively large for efficient endocytic uptake by non-phagocytic cells ^38,39, direct physical delivery methods may enable their introduction into cells ⁴⁰. Furthermore, although the NaCl concentrations used for crystallization (0.25–2.0 M) are higher than the physiological ionic strength, post-crystallization chemical stabilization of the DNA-linked structures followed by gradual exchange into a lower-ionic-strength solution may enable transfer of the crystals into a more biologically compatible environment, while preserving their structure ^32,41. Therefore, the control of the crystal size and crystallinity through droplet size and salt concentration demonstrated in this study may provide a basis for biological applications combining direct physical delivery with post-crystallization stabilization and solution exchange.