Introduction

The recent and widespread interest in the potential environmental health and safety risks of nanomaterials (nanoEHS) has included substantial work on silver nanoparticles (AgNPs) (Tolaymat et al. 2010; Fabrega et al. 2011; Meng et al. 2010; Sharma et al. 2009; Wijnhoven et al. 2009). AgNPs are of interest primarily due to their increased use in consumer products (Woodrow Wilson International Center for Scholars 2009; U.S. EPA 2012), purportedly to exploit their antimicrobial properties. Other examples include medical products (Chen and Schluesener 2008) and food packaging (Gottesman et al. 2010). The predicted growth of AgNP use in consumer products contributed to the decision by the Organization for Economic Cooperation and Development (OECD) Working Party on Manufactured Nanomaterials (WPMN) to include AgNPs as one of fourteen nanomaterials in their international testing program (OECD 2008). One of the mechanisms of antimicrobial action is believed to arise from the surface oxidation of the metallic AgNPs and subsequent dissolution into silver cations (Ag+) (Gorham et al. 2012; Kennedy et al. 2010; Liu and Hurt 2010a; Liu et al. 2010b), which act as the active biocide (Bradford et al. 2009; Kim et al. 2009). Dissolution leads to a mass loss from the AgNPs, with subsequent changes to the size distribution (Elzey and Grassian 2010; Zook et al. 2011a), increased colloidal instability and aggregation effects, (Li et al. 2011; Lok et al. 2007; Romer et al. 2011) and potentially cycling into new “incidental” AgNPs via reduction of released silver ions (Akaighe et al. 2011; Glover et al. 2011; Maurer et al. 2012; Yin et al. 2012). Changes in size distribution, agglomeration state, and dissolved silver fraction could potentially have significance when interpreting environmental persistence or biological mechanisms of toxicity (Hussain and Schlager 2009; Levard et al. 2012; Rahman et al. 2009; Xiu et al. 2012), making long-term studies on fractions from a single suspension challenging. As a result of these many potential and likely transformations, and based on available evidence, we conclude that long-term stability of solution-phase AgNPs is an intractable issue. However, creating suspensions on-demand from AgNP powders has also proven challenging, as some frequently used commercial AgNP powders yield highly polydisperse particle distributions that are inadequate for hypothesis testing of size-based phenomena.

The instability of AgNPs obviously presents unique technical challenges for the development of AgNP reference materials, which must exhibit long-shelf life (i.e., at least 5 years) without significant changes to composition or physical properties. To increase shelf life in the pharmaceutical industry, lyophilization (or freeze drying) is frequently employed to preserve compounds, drugs, vaccines, and other biological materials (Endres et al. 2012; Lee et al. 2009; Zhang et al. 2008). Lyophilization also avoids introducing heat-related instabilities (such as AgNP dissolution) through traditional solvent-removal techniques such as rotary evaporation. Recently, lyophilization of nanoparticles has been studied systematically to explore the role that formulation and process conditions play in the lyophilization and reconstitution of nanoparticles (Abdelwahed et al. 2006; Beirowski et al. 2011a, b, 2012; Chen and Wang 2007). Beirowski et al. have repeatedly come to the conclusion that the formulation of the nanoparticle suspension to be lyophilized is the most critical parameter to control (not the lyophilization processing conditions) when the objective is to recover the original size distribution. However, Beirowski et al. focused only on two ball-milled crystalline active pharmaceutical ingredients, leaving a critical knowledge gap for metal nanoparticles such as AgNPs.

Therefore, this study demonstrates the capacity to stabilize AgNPs for on-demand use over extended periods of time and across different laboratories. Lyophilization of AgNP suspensions, in the presence of appropriate lyophilizing agents, mitigates the risks from solution-based changes that occur on long time scales (months to years), and subsequent addition of water yields on-demand singly dispersed AgNPs. The selection of polyvinylpyrrolidone (PVP) as a combination of lyophilizing and dispersing agent is based on its inclusion in the OECD WPMN testing program on AgNPs, and on the substantial published nanoEHS literature involving PVP-capped AgNPs (Huynh and Chen 2011; Kent and Vikesland 2011; Levard et al. 2011; Lowry et al. 2012; Song et al. 2011; Stamplecoskie and Scaiano 2010; Tejamaya et al. 2012). The vacuum stoppering mechanism of the freeze drier enables backfilling with an inert gas before sealing the lyophilized material, thereby reducing further oxidative reactions. This study aims to fill the critical knowledge gap of successfully formulating freeze–dried AgNPs for the future development of reference material prototypes. Specifically, this study aims to achieve on-demand singly dispersed nanoparticle suspensions exhibiting a reproducible size distribution that reconstitutes the original (pre-lyophilization) distribution with minimal involvement by the user. Enabling the end user to just add water to the PVP–AgNP lyophilized solid (or “cake”) immediately before use eliminates the variations introduced by aging, sonication (Taurozzi et al. 2012, 2011), and user-induced variability or error. The approach of freeze drying and on-demand reconstitution by just adding water offers a promising basis for the development of AgNP reference materials; toward that end, the National Institute of Standards and Technology (NIST) is currently producing RM8016, nominally 10 nm AgNPs, and RM8017, nominally 75 nm AgNPs, which will be freeze dried to extend the shelf life of the particles over solution-based storage conditions. In addition, applications beyond reference materials could be enabled, such as providing greater flexibility for nanocomposite synthesis, where optical property control is structure dependent (Calandra et al. 2010).

Materials and methods

Preparation of PVP–AgNP lyophilized cakes

Nominally 20 nm diameter citrate-capped AgNPs were synthesized, purified, and concentrated to 1 mg mL−1 Ag as described elsewhere (Chinnapongse et al. 2011; MacCuspie 2011). A single batch was divided into several fractions, and then the desired amount of PVP (PVP10, Sigma-Aldrich, St. Louis, MO, USA)Footnote 1 was added to the suspension and stirred 24 h while covered with parafilm to reduce evaporation. Within 48 h, the resulting solutions were dispensed in 5.00 mL aliquots into 10 mL serum bottles for lyophilization. Samples were frozen at −20 °C for 24 h, then loaded into the vacuum stoppering adapter (Part #7593000 Labconco, Kansas City, MO, USA) and attached to a 6 L freeze drier (FreeZone 6, LabConco, Kansas City, MO, USA). One bottle had a metal thermocouple frozen into the initial aliquot to monitor the internal temperature of the solid during sublimation. Lyophilization was considered complete when the temperature of the cakes rose above the temperature of the shelf, indicating that heat was no longer being transferred from the shelf to the cake to provide the energy required to sublime the ice. Lyophilization was complete within 24 h.

For nominally 10 and 75 nm AgNPs with 40 and 10 mg mL−1 PVP, respectively, suspensions containing 1 mg mL−1Ag were purchased from nanoComposix, Inc. (San Diego, CA, USA), stored in the dark at 4 °C, and used as received within 3 weeks of receipt. Resulting solutions were dispensed in 2.00 mL aliquots into 10 mL serum bottles for lyophilization. The samples were frozen for 3 h in the stoppering tray drier (Model 7948020, LabConco, Kansas City, MO, USA) with a shelf temperature of −40 °C, then lyophilized for 30 h. See Fig. SI1 in the Supplementary Information (SI) for an example record of the shelf temperature, cake sample temperature, and system pressure over the course of one lyophilization run. The stoppering tray drier was backfilled with argon gas to approximately 400 Pa before the serum bottles were stoppered inside the freeze drier. Bottles were crimp sealed and stored in the dark at room temperature until reconstitution.

Instrumentation

Dynamic light scattering (DLS) was performed using a Malvern Zetasizer Nano (Westborough, MA, USA), and using disposable semi-micro UV-transparent plastic cuvettes (BrandTech, Inc., Essex, CT). A general protocol was followed to operate the DLS instrument and perform measurements (Hackley and Clogston 2007). Samples were diluted 1:100 with biological grade deionized (DI) water before measurement. DLS measurements were analyzed with both the refractive index and viscosity values of water and the refractive index and viscosity values of a PVP solution, however, at these Ag and PVP concentrations, <1 % difference in reported z-average size was found. All reported z-average diameter values are the mean of five consecutive measurements, made under repeatability conditions, with uncertainty of one standard deviation about the mean. Note, this uncertainty informs on the precision of the measurement, and is not reflective of the width of the size distribution. DLS intensity-based size distributions were obtained with Zetasizer Software v6.20 (Malvern, Westborough, MA, USA) using the general purpose analysis model.

Immediately upon completion of DLS measurements, cuvettes were transferred to a Lambda 750 Spectrophotometer (Perkin-Elmer, Waltham, MA, USA) for ultraviolet–visible (UV–vis) absorbance spectroscopy.

Atomic force microscopy (AFM) was performed on a Dimension 3100 (Bruker AXS, Santa Barbara, CA, USA), with the average height (z-displacement) of no fewer than 100 particles registered, and uncertainty of one standard deviation of the mean. Note this uncertainty is impacted by the width of the size distribution. Care was taken to minimize sample preparation-induced image artifacts, including drying-induced agglomeration (Bonevich and Haller 2010; Grobelny et al. 2009) and incidental AgNP formation (Glover et al. 2011). In brief, a 20 μL drop of suspension was placed on a 5 × 5 mm 3-aminopropyldimethylethoxysilane-functionalied Si chip for about 10 min, then rinsed with DI water for 10 s and dried using short blasts of compressed air. All AFM images were collected within 3 days of particle deposition onto the Si chip.

Ultra small angle X-ray scattering (USAXS) was performed at ChemMatCARS, sector 15-ID at the Advanced Photon Source (APS), Argonne National Laboratory, Argonne, IL, USA. The instrument is described in detail elsewhere (Allen et al. 2008; Cookson et al. 2006; Ilavsky et al. 2009b). Experiments were performed at ambient temperature (approximately 25 °C). For USAXS scans, a 0.4 × 0.4 mm monochromatic (energy = 10.5 keV, wavelength λ = 0.118 nm) incident X-ray beam was used. The scattering data were corrected for parasitic background scattering and for attenuation using data from DI water. The USAXS data, which are inherently slit-smeared perpendicular to the scanning direction, were absolute calibrated with respect to the incident beam intensity according to the first-principle methods (Ilavsky et al. 2009b), and desmeared using the Lake algorithm (Lake 1967). For measurements, two experimental setups were employed. For nominally 20 nm AgNPs, approximately 0.3 mL of suspension was enclosed within a 1-mm thick liquid cell with polyamide film windows. For nominally 10 and 75 nm AgNPs, suspensions were pumped into a nominally 1.5-mm diameter quartz capillary. To obtain quantitative nanoparticle volume fraction size distributions, the calibrated 1-D USAXS intensity data, I(Q) versus Q was analyzed (where scattering vector Q has a magnitude of Q = (4π/λ)sinΘ, where λ is the X-ray wavelength and Θ is half of the scattering angle). The data were analyzed using an entropy maximization algorithm known as MaxEnt (Potton et al. 1988). The USAXS data were reduced and analyzed (including implementation of MaxEnt) within the Irena 2 package of data evaluation and modeling macros for Igor Pro (Wavemetrics, Lake Oswego, OR, USA) (Ilavsky and Jemian 2009a).

Reconstitution of lyophilized cakes

Reconstitution of lyophilized PVP–AgNP cakes involved addition of 2.00 mL of DI water, gentle shaking of the bottle by hand for approximately 1 min, and allowing the suspension to stand for 1 h under ambient laboratory light and temperature conditions before measurements. A video of this “just add water” procedure can be found in the SI.

Note all measurements reported on “stock” suspensions (i.e., suspensions that were not lyophilized) were typically collected within 48 h of the commencement of lyophilization, and always within 7 days, to compare the before- and after-lyophilization attributes of the size distributions most effectively.

Statistics

All statistical comparisons were performed in MATLAB (Natick, MA, USA) using the one-way ANOVA test with a t test to compare the individual means.

The term residual will be defined in this study as the sum of the absolute values of the differences between each data point in the size distribution, normalized to a maximum possible value of 1, yielding a unitless parameter. Residuals will be applied as a numerical assignment to estimate the difference between DLS intensity-based size distributions. The residual parameter is intended to approximate the degree of overlap between two size distributions, to provide a numerical representation of the shifts that are easy to identify visually when two size distribution plots are overlaid.

Results and discussion

Formulation optimization

When suspensions of electrostatically stabilized metal colloids (such as AgNPs) are frozen, the nanoparticles irreversibly aggregate. The freezing process forces the nanoparticles to have such a small interparticle spacing that the nanoparticles either become trapped together by attractive forces or sinter together into irregularly shaped aggregates (Albert et al. 2009). However, if sufficient repulsive force or physical barrier between NPs is present, such as through sterically bulky coatings and/or suitable lyophilizing agents, the NPs should be able to return to a singly dispersed suspension under appropriate conditions.

PVP was selected as a lyophilizing agent for reasons previously discussed. To determine the minimum concentration of PVP required to successfully recover singly dispersed AgNPs after one lyophilization/resuspension cycle for nominally 20 nm citrate-capped AgNPs at 1.0 mg mL−1 total Ag, a series of formulations were examined before and after one cycle.

Visual inspection of the cakes (Fig. 1) also reveals observable differences in the lyophilized product as a function of formulation. For the smallest PVP concentrations (1 and 10 mg mL−1, Fig. 1a, b), only a trace of fine black dust was observed in the bottle after lyophilization. When an optimal PVP concentration was added to the formulation (50 mg mL−1, Fig. 1c), a reddish-brown cake color was observed. When excess (100 and 200 mg mL−1, Fig. 1d, e) PVP was added; a yellow–brown cake color was observed.

Fig. 1
figure 1

Photographs of various lyophilized AgNP–PVP cakes. Formulations for the 20 nm AgNPs included a 1.0 mg mL−1 PVP, b 10 mg mL−1 PVP, c 50 mg mL−1 PVP, d 100 mg mL−1 PVP, and e 200 mg mL−1 PVP

Full size image

A suite of tools, including DLS, UV–Vis, and USAXS, was employed, which were previously validated to be synergistic for characterizing AgNP aggregation phenomena (MacCuspie et al. 2011a). DLS is an excellent tool for detecting early onset of aggregation, i.e., small numbers of aggregates (Baalousha and Lead 2012; MacCuspie et al. 2011b). Figure 2 presents the DLS mean diameters for pre- and post-lyophilization suspensions (also referred to as “stock” and “reconstituted”, respectively) as a function of PVP concentration for nominally 20 nm AgNPs. Insufficient PVP (≤10 mg mL−1) leads to significant aggregation in the reconstituted suspensions, as indicated by the increased DLS diameter. However, excess amounts of PVP (≥100 mg mL−1) result in pre-lyophilization suspensions that have a significantly increased DLS diameter, as compared to the diameter of the citrate AgNPs before the addition of the PVP. This diameter increase is likely due to an increased adsorption of PVP onto the surface of the AgNPs, creating perhaps multilayers of adsorbed PVP. Interestingly, after the lyophilization process, all the resuspended AgNPs for ≥50 mg mL−1 PVP reveal a similar DLS diameter which is consistent with the thickness of the PVP coating on the AgNPs in the pre-lyophilization 50 mg mL−1 PVP solution. Therefore, the excess PVP likely still provides good protection against aggregation during the lyophilization process, despite the increase in hydrodynamic size for the pre-lyophilization suspensions. This pre-lyophilization size increase could arise from either a large corona of PVP adsorbed onto the AgNPs (as mentioned above), or formation of weakly bound agglomerates that break apart during the lyophilization-reconstitution process.

Fig. 2
figure 2

DLS z-average (cumulants) diameter before and after one lyophilization/reconstitution cycle for the nominally 20 nm AgNPs with various PVP concentrations reported by mg mL−1 PVP. Data are average values from five consecutive measurements under repeatability conditions with uncertainty of one standard deviation. Asterisks indicate significant difference between pre-lyophilization (stock) and post-lyophilization (reconstituted) samples (p < 0.05)

Full size image

DLS intensity-based size distributions provide additional insight about the recovery of the original size distribution after lyophilization. A significant increase in the size distribution occurs for the 10 mg mL−1 PVP formulation (Fig. 3a). For the 50 mg mL−1 PVP formulation (Fig. 3b), only very small shifts in the size distribution were observed, with slight increases toward larger sizes in the reconstituted samples. For the 100 mg mL−1 PVP formulation (Fig. 3c), a noticeable decrease in the size distribution was observed. The residuals were 0.68, 0.16, and 0.86 for the 100, 50, and 10 mg mL−1 PVP formulations, respectively. All three of these size distribution changes are also consistent with the average diameter results. While the 10 mg mL−1 PVP formulation does indeed provide a smaller initial and final DLS diameter compared to the 50 mg mL−1 PVP formulation, the 50 mg mL−1 PVP formulation likely has a significantly thicker shell of PVP coating the metallic AgNP core compared to the lesser PVP concentration formulations, and it is likely that the shell remains intact through the freeze drying process. Thus, it is clear that of the broad range of PVP concentrations studied by DLS, the 50 mg mL−1 PVP formulation provides the optimal recovery of the pre-lyophilization size distribution of the nominally 20 nm AgNPs.

Fig. 3
figure 3

DLS intensity-based size distributions for nominally 20 nm AgNPs with a 10 mg mL−1 PVP, b 50 mg mL−1 PVP, and c 100 mg mL−1 PVP formulations for the pre-lyophlization (stock, red circles) and post-lyophilization (reconstituted, blue triangles) suspensions. Lines are to guide the eye. (Color figure online)

Full size image

USAXS provides a method to simultaneously measure the in situ size distribution and volume fraction (i.e., silver concentration) (Allen et al. 2008; Zhang et al. 2009). This enables efficient characterization for changes to the aggregation state (MacCuspie et al. 2011a), or monitoring for potential reactions (Koerner et al. 2012). When the USAXS is properly calibrated, measurement of the absolute volume fraction provides comparisons to both DLS data and UV–Vis data. USAXS is also only sensitive to the size of the silver metal “core” of the PVP-capped AgNPs, as the X-ray scattering cross section of the PVP corona or “shell” is negligible compared to that of silver; this provides an excellent complement to the DLS hydrodynamic size, which includes the core and attached corona.

In Fig. 4, USAXS-derived size distributions post-reconstitution for a series of formulations with nominally 20 nm AgNPs ranging from 20 to 100 mg mL−1 PVP are shown to determine the minimum amount of PVP required to avoid aggregation during lyophilization. This range was selected based upon the range identified by the DLS results. For comparison, the size distribution for the original citrate-capped AgNP suspension is also shown. The 20 mg mL−1 PVP formulation had incomplete recovery of the silver and a noticeable change in the shape of the size distribution. The formulations that provided the best recovery of silver and least change to the size distribution were the 40, 80, and 100 mg mL−1 PVP formulations. The USAXS data reveals that post-lyophilization suspensions are singly dispersed when sufficient amounts of PVP are present in the formulation. Higher levels of PVP lead to increases in the corona thickness (and therefore the hydrodynamic size), but provide the silver cores protection from aggregation. When considered together, the DLS and USAXS data reveal that, for nominally 20 nm AgNPs, 40 mg mL−1 PVP is the minimum formulation for a successful recovery of singly dispersed AgNPs.

Fig. 4
figure 4

USAXS derived MaxEnt volume-based size distribution functions, post-reconstitution for nominally 20 nm AgNPs for various formulations of 20–100 mg mL−1 PVP, and for the original citrate stock suspension. Note that the diameter (x-axis) is on a log scale

Full size image

Size-based observations

Three sizes of AgNP formulations were also examined in detail (Fig. 5).

Fig. 5
figure 5

DLS intensity-based size distribution histograms (a–c), absorbance spectra (d–f), and USAXS volume-based size distribution functions (g–i) of pre-lyophilization (stock, red circles) and post-lyophilization (reconstituted, blue triangles). a, d, g nominally 10 nm AgNPs with 40 mg mL−1 PVP, b, e, h nominally 20 nm AgNPs with b, e 50 mg mL−1 PVP or h 40 mg mL−1 PVP, and c, f, i nominally 75 nm AgNPs with 10 mg mL−1 PVP. DLS data are the mean of five consecutive measurements with uncertainty of one standard deviation; absorbance data are the mean of three separate sample preparations with uncertainty of one standard deviation. Uncertainties, if not visible, are smaller than the symbols; lines are to guide the eye. (Color figure online)

Full size image

Nominally 10 and 75 nm (core) AgNPs were selected, as they match the sizes chosen for the forthcoming NIST reference materials RM8016 and RM8017, as well as nominally 20 nm AgNPs synthesized in our laboratory by a commonly reported method. For the nominally 10 nm AgNPs with 40 mg mL−1 PVP (Fig. 5a, d, g), the mean DLS diameter remained statistically unchanged, with values of 21.0 ± 0.5 nm for the pre-lyophilization and 20.8 ± 1.1 nm for the reconstituted post-lyophilization samples. However, slight increase to the DLS intensity-based size distribution (Fig. 5a) was observed for both the mode diameter and the width of the size distribution. It is important to note that the forthcoming NIST reference materials will provide reference values for the mean diameter by a variety of measurement techniques. It is also important to keep in mind that the repeatability and reproducibility in the mean size measurement will determine the uncertainty assigned to the reference value. Reference materials must be uniform with respect to the assigned measurand within the stated uncertainty across all units produced. Heterogeneity testing is an integral part of reference material production, as it ensures that the units comply with the stated reference values such that each unit can be utilized for quality control applications by the customer. In the present case, a broader size distribution is tolerable, since only the mean value is of consequence.

A larger DLS diameter is consistent with AFM observations (Fig. 6a, b). The height image reveals an approximately 24 nm average AgNP height for the nominally 10 nm AgNPs, however, the phase image provides an interesting insight into why there is a size difference observed between the AFM imaging and both the USAXS values (Fig. 5g) and the TEM value reported by the vendor. A very thick shell of compliant (softer) material surrounds a core of rigid (harder) material, as indicated by the dark ring surrounding the bright circular core for each of the PVP-capped AgNPs imaged by AFM. This dark ring is not observed on citrate-capped AgNPs (Fig. SI2). It is therefore worth considering that the DLS and AFM techniques are measuring the diameter of the PVP AgNPs including this thick corona of PVP adsorbed onto the AgNP surface, while techniques such as TEM and USAXS are sensitive to the metal core only. A detailed discussion of this point can be found elsewhere (MacCuspie et al. 2011b).

Fig. 6
figure 6

AFM height (a, c) and phase (b, d) images for: a, b nominally 10 nm AgNPs with 40 mg mL−1 PVP and c, d nominally 75 nm AgNPs with 10 mg mL−1 PVP. Scale bars are a, b 100 nm and c, d 50 nm; white shades in color scale represent a 20 nm and c 100 nm

Full size image

Absorbance spectroscopy at the surface plasmon resonance (SPR) wavelength was employed to determine if significant amounts of suspended aggregates are present. Red-shifted SPR absorbance can be easily observed from aggregates of metallic nanoparticles (Tsai et al. 2011; Zook et al. 2011b, c). The absorbance measurements of the nominally 10 nm AgNPs (Fig. 5d) suggest the presence of aggregation, with a small tail and shoulder peak at red-shifted wavelengths from the primary singly dispersed AgNP absorbance peak.

Comparing the pre- and post-lyophilization suspensions of nominally 20 nm AgNPs (Fig. 5b, e, h), only trace levels of aggregation were detectable by either DLS or absorbance measurements. As Fig. 5b is a reproduction of Fig. 3b for ease of comparison, this data have already been discussed. The absorbance spectra for the reconstituted nominally 20 nm AgNPs (Fig. 5e) reveal the appearance of a slight tail at red-shifted wavelengths. The tail is not as pronounced as the 10 nm AgNPs’ shoulder peak, and thus indicates only a very slight degree of aggregation. The overall shape of the two absorbance spectra for the nominally 20 nm AgNPs is also much more similar for the main peak, compared to the 10 nm AgNPs. The USAXS size distribution of nominally 20 nm AgNPs with a 40 mg mL−1 PVP formulation (Fig. 5h) reveals that the slightly polydisperse size distribution is approximately recovered after one lyophilization cycle.

For the nominally 75 nm AgNPs with a 10 mg mL−1 PVP formulation (Fig. 5c, f, i), a lower PVP amount was selected due to the decrease in number concentration of larger nanoparticles when the silver mass concentration remains constant. The DLS size distributions (Fig. 5f) for the pre- and post-lyophilization suspensions were identical within the uncertainty of the measurements. While the DLS diameter was larger than the nominal silver core diameter, the AFM data (Fig. 6c, d) also revealed a thick PVP corona over the silver core. In addition, Fig. 5f reveals no significant changes to the shape of the averaged absorbance spectra before and after lyophilization of nominally 75 nm AgNPs for three replicate samples, indicating recovery of the singly dispersed AgNP suspensions with no detectable aggregation. The shape of the USAXS size distribution (Fig. 5i) is quite similar to the DLS size distribution (Fig. 5c). While the USAXS data do suggest the presence of low concentrations of small scattering objects in the reconstituted samples, unfortunately pre-lyophilization samples were not analyzed due to synchrotron beamline time constraints; additionally, the presence of small AgNPs resulting from dissolution, or as produced by measurements in electron beams has been reported by others (Bogle et al. 2006; Kent and Vikesland 2011; Kim et al. 2008). Nevertheless, the data clearly indicate the absence of detectable aggregation in the reconstituted nominally 75 nm AgNPs.

To compare the degree of aggregation by numerical differences in DLS size distributions, the residuals were again examined. For the nominally 10, 20, and 75 nm AgNPs with PVP concentrations of 40, 50, and 10 mg mL−1, respectively, the residual values were 0.19, 0.16, and 0.03, respectively; recall that values of 0.68 and 0.86 were observed for nonoptimized 20 nm AgNP samples with 100 and 10 mg mL−1 PVP, respectively.

It is also important to note that even though the lyophilized cakes are potentially a stable way to provide long-term storage of AgNPs, after reconstitution of the freeze–dried material, the resulting aqueous suspensions are still subject to short-term changes. For example, Fig. 7 shows one of the 1:100 dilutions of the reconstituted 75 nm AgNPs observed for 7 days by DLS. The z-average diameter trends downward, likely due to the dilution effect causing dissociation of PVP into solution from the thick PVP shell initially coating the AgNPs. A similar downward trend was observed in a replicate sample, and was observed to continue decreasing for 18 days of observation to a value of 99.4 ± 1.0 nm, highlighting the need to use freshly reconstituted materials to ensure consistency among experiments.

Fig. 7
figure 7

DLS diameter for a reconstituted suspension diluted 1:100 in DI water and observed over 7 days. Diamonds represent data from individual measurements, the thick line represents the average of N = 97 DLS diameter measurements collected in the first 24 h after dilution, and the thin lines represent the boundaries of two standard deviations about the mean

Full size image

Conclusion

This study demonstrates a proof of concept for a “just add water” approach to generate on-demand suspensions of singly dispersed AgNPs from a lyophilized cake of PVP and AgNPs. A formulation for nominally 75 nm AgNPs is presented that reproduces the pre-lyophilization size distribution within the uncertainty of the measurements used. Formulations for nominally 20 and 10 nm AgNPs are presented that yield consistent size distributions from batch to batch upon reconstitution, with trace levels of aggregation in the post-lyophilization suspensions. In all cases, the lyophilized cakes were easily reconstituted by the addition of DI water, gentle shaking by hand, and allowing the suspensions to equilibrate on the bench top for 1 h. In addition, these formulations serve as prototypes for AgNP reference materials currently under development at NIST (reference materials RM8016 and RM8017). The “just add water” approach for resuspension after lyophilization and packaging under inert gas could potentially enable new classes of moderately stable nanoparticles to serve as candidates for future reference material development or for general use as test materials to assess the environmental and biological risks associated with nanoscale particles.