Facile benchtop reactor design using dendrimer-templating technology for the fabrication of polyethyleneimine-coated CuO nanoparticles on the gram scale

As we move forward toward an economy that includes increasing numbers of nanoparticle (NP)-enabled or nanoparticle-containing consumer products, manufacturers will require functional nanoparticles that are prepared from earth-abundant and inexpensive metals.^1–6 One particular area of interest is the fabrication of nanoparticles that act as viable replacements for the rare and expensive noble metals (e.g., Ag, Au, and Pt) currently used in many commercial chemical processes.² For these reasons, copper (Cu) and copper oxide (CuO) nanoparticles have received much attention because of the abundance of available copper and the importance of Cu to modern technology due to its high electrical conductivity, high melting point, low electrochemical migration behavior, and wealth of low-cost methods for production.⁷ CuO nanoparticles, in particular, are becoming widely used because of their unique chemical and physical properties observed at the nanometer scale.^8–21 

Copper oxide nanoparticles with a size range between 1 and 100 nm have been reported to show superior catalytic activity and selectivity when compared to that of common copper oxide powder.⁸ More specifically, CuO nanoparticles exhibit unique properties such as improvement in chemical activity, thermal resistance, catalysis, quantum size effects, and macroscopic quantum tunneling effects in magnetic and optical absorption which widens its potential applications to solar energy transfer, sensors, storage devices, and superconductors.¹¹ Many available methods have been reported in the literature for the production of CuO nanoparticles such as surfactants as precursors while using temperature to control the shape and size of desired CuO nanoparticles,¹¹ hydrothermal microwave methods for the synthesis of CuO flower-nanostructures after thermal treatment at 393 K for 1 h,¹² brown algae extract where the need for chemical reagents and surfactants are eliminated,¹³ solgel techniques,¹⁴ and aqueous precipitation methods,¹⁵ just to name a few. Despite the numerous methods cited here for the synthesis of CuO nanoparticles, in order for CuO nanoparticles to become a useful industrial commodity, new synthesis methods capable of producing CuO NPs in large (gram) scale quantities must be developed. This challenge has promoted the development of methods that are capable of producing alternative copper-based NPs with complex structures or systems based on copper oxides with scale-up capability; however, most still suffer from an inability to control size, shape, and agglomeration during the scale-up growth process. This work demonstrates how a continuous flow reactor combined with a chemical reduction method achieves a gram-scale synthesis of copper oxide nanoparticles.

Previous studies have shown that dendrimer-mediated synthesis processes are able to generate small quantities of Cu nanoparticles (≥1 nm in diameter) with precisely controlled sizes and shapes, and these methods are able to mitigate aggregation during the Cu nanoparticle growth.^2–3,22 Polyamidoamine (PAMAM) dendrimers are star-shaped, hyperbranched macromolecules with nanometer-scale dimensions that are commonly used in the small-scale Cu NP syntheses, despite being expensive. Specifically, 2.5 g of PAMAM dendrimer with OH terminal end groups from Sigma-Aldrich is 100 times more expensive than 2.5 g of its polyethyleneimine (PEI)-branched counterpart (i.e., $248 vs $1.82, respectively). Furthermore, due to the interdigitation of the dendrimer branching units, the scalability of the dendrimer synthesis process is limited.¹ By using dendrimer-mediated synthetic routes, exploiting inexpensive linear amine/imine polymers, it is possible to take advantage of the metal ions forming complexes to the amines/imines to produce metallic nanoparticles while avoiding interdigitation.^3,23,24 This significantly simplifies the system requirements for the formation and control over nanoparticle size, reduces the associated production cost, and allows for high-throughput synthesis of metal nanoparticles (i.e., scalability).

Similar to the PAMAM dendrimer, PEI has many primary amino groups (see Scheme 1) that can be used as chelating sites for the formation of various metal or metal oxide nanoparticles, and due to the similar branching structural arrangement of the PEI, it can also be used to stabilize the Cu NPs as they form.^25–28 Furthermore, similar to the PAMAM dendrimer method, controlling the metal-to-polymer ratio introduced into a flow reactor and the rate of flow during the chemical mixing process allows for the optimization of the size, shape, and concentration of the nanoparticles forming in the reactor.²⁹ 

Herein, we report on recent work demonstrating the construction of a flow reactor for the synthesis of PEI-coated nanoparticles and comparing the resulting NPs to those synthesized through traditional batch synthesis with the PAMAM dendrimer. PEI was chosen as the capping agent because it is a low-cost alternative to the PAMAM dendrimer with known affinity for copper.³⁰ In the flow reactor, metal salt solutions are premixed with PEI and subsequently reduced to metal zero via chemical reduction while under steady flow conditions. However, immediate oxidation is observed due to exposure to ambient conditions, producing Cu NPs with a native oxide layer. We provide a comparative characterization of small-scale synthesized PAMAM- and large-scale synthesized PEI-capped copper nanoparticles using x-ray diffraction (XRD), transmission electron microscopy (TEM), and ultraviolet-visible (UV-vis) spectroscopy to demonstrate the great potential of this flow reactor.

Branched polyethylenimine, H(NHCH₂CH₂)NH₂, average M_w ∼ 25 000 by light scattering, average M_n ∼ 10 000 by gel permeation chromatography, and generation four (G4) PAMAM dendrimer were purchased from Sigma-Aldrich. Copper(II) nitrate hemi(pentahydrate), Cu(NO₃)₂⋅ 2.5 H₂O was used as the metal source. De-ionized water was used as the solvent, and sodium borohydride (NaBH₄) was used as the reducing agent. All chemicals were of analytical reagent grade and purchased from Sigma-Aldrich Chemical Company. Stock solutions of PEI at 0.5 and 1 wt. % and micromolar concentration of dendrimer were prepared. (Note that PEI was weighed and dissolved in 500 ml of de-ionized water, and the solution was heated to 50 °C for 30 min.) A 0.2 M solution of the copper salt was prepared by dissolving 5.815 g of the as-received copper salt into 125 ml of de-ionized water and used as the Cu metal source.

Using UV-vis spectroscopy, absorbance spectra were collected for the polyethylenimine metal salt solution and for the complexation of polyethylenimine-copper solution [PEI-(Cu²⁺)_X] before chemical reduction. All absorbance measurements were carried out using a Hewlett-Packard HP 8453 UV-visible spectrophotometer equipped with a 1.0 cm optical path length quartz crystal cuvette. The wavelength range of analysis was 250–800 nm.

X-ray photoelectron spectra (20 sweeps, 59.7 eV pass energy, 0.125 eV/step) were collected on a PHI 5600 XPS system equipped with an Mg Kα x-ray (1253.6 eV) source (Physical Electronics, Chanhassen, MN). Powdered black/blue Cu NPs were dusted onto double sided carbon tape, and initial photoelectron spectra were collected before and after sputtering with an ion gun at 10 × 10⁻³ Pa Ar⁺ with at a constant current of +0.7 μA. XPS spectra were measured after 1, 5, 15, 30, and 60 min of sputtering (100 sweeps, 59.7 eV pass energy, 0.125 eV/step). 99.99% copper foil (Strem Chemicals), Ar⁺ sputtered for 15 min, was utilized as a standard for metallic copper. Similarly, a copper(I) oxide sample sputtered for 1 min to remove Cu(II) oxide that formed at the surface was used as a copper(I) oxide reference.

XRD measurements were used to determine the crystallographic structure of the polyethylenimine-coated Cu NPs. X-ray diffraction analyses were carried out using a Rigaku D/MAX 2200 x-ray diffractometer with a diffracted beam graphite monochromator using Cu Kα radiation. The analysis was performed from 0° to 80° of 2θ angle at a rate of 5°/min and a sample width of 0.02. The data were collected, and peaks were analyzed using PDF database of Joint Committee on Powder Diffraction Standards.

TEM was used to image the Cu NPs produced by the batch process, dendrimer-mediated Cu NPs, and the new method of nanoparticle production, fluidic flow PEI-synthesized Cu NPs. Before imaging, Cu NPs were dispersed in Milli-Q water in a one-dram vial (15 mm width × 45 mm length) and diluted with Milli-Q water until the color of the resulting dispersion was not apparent looking through the vial but was apparent looking along the length of the vial. Note that the dendrimer-mediated Cu NP solution exhibited a brown/yellow hue and PEI-synthesized Cu NP solution exhibited a blue hue. A 6 μl drop of each suspension was deposited on 200 mesh copper TEM grids with Formvar and carbon supports (Ted Pella Inc., Redding, CA). The grids were allowed to dry in ambient conditions with a protective cover to prevent contamination by dust. Images were collected using a Tecnai T12 transmission electron microscope with an operating voltage of 120 kV at various magnifications. Images were processed in imagej. The line tool was used to determine the diameter of the Cu NPs formed by the dendrimer and PEI processing methods; more than 600 individual nanoparticles were sized in each condition.

A standard chelation and chemical reduction procedure was used.^3,21 In short, the prepared dendrimer and copper salt solutions were mixed with a 55 mol equivalent of Cu(NO₃)₂⋅ 2.5H₂O to dendrimer with continuous stirring. Subsequently, a complex solution of dendrimer and metal ions Den-(Cu²⁺)_X where X = 55 mole ratio was formed. After 15 min of continuous stirring, the addition of 10-fold excess of freshly prepared aqueous reducing agent “NaBH₄” was added in a drop-wise manner to the complex solution to produce zero-valent Cu nanoparticles via chemical reduction. All nanoparticle growth experiments were carried out under an N₂ atmosphere to prevent oxidation as the Cu nanoparticles formed.

Figures 2(a) and 2(b) show the flow reactor diagram and design, detailing the materials used to construct the multifunctional nanoreactor and flow process involved in the nanoparticle synthesis. As shown in Fig. 2(a), the heart of the nanoreactor flow system is the use of a peristaltic pump capable of mixing several solutions via a t-mixer setup during the nanoparticle formation process. The peristaltic pump is a Thomas 3386 Mini Variable Speed Tubing Pump with 0.4–85 ml/min flow rate control range. Polyvinyl tubing with a 6.35 mm internal diameter and a t-mixer with an internal diameter of 4.32 mm were used to transport and mix both the PEI-(Cu²⁺)_X solution and reducing agent in concert and, subsequently, the solution was collected in the flow reactor cell. A close view of the flow process is shown in Fig. 3. The flow rate was controlled to manipulate the residence time of the premixed PEI-(Cu²⁺)_X solution and reducing agent flowing through the reaction tubing. Particle formation residence times of 3–10 min from the point of mixing to the point of entering the reactor cell resulted in CuO nanoparticles. [Note that since the PEI-(Cu²⁺)_X solutions are premixed, in this flow reactor, the required residence time necessary in other reactors to produce nanoparticles is significantly reduced.] Furthermore, although this system is equipped with photoreduction capability, this report focuses only on the chemical reduction reaction.

Polyethyleninime (concentration 0.5 or 1 wt. %) was purchased and used as the stabilizer and capping agents and as the ion-pair exchange medium. Micromolar concentrations of the copper salt solutions were prepared and mixed with the 0.5 or 1.0 wt. % PEI solution to promote the formation of a complex of polyethylenimine and metal ions [PEI-(Cu²⁺)_X, where X = 0.5 or 1 wt. %]. Under predetermined pumping conditions, freshly prepared 1.0M NaBH₄ reducing agent and the complexed PEI-(Cu²⁺)_X solutions were added to the reaction cell at a constant rate of flow via a t-mixer that connected tubes 1 and 2, see Fig. 3. Based on our setup, a flow rate of 23.0 ml/min was selected for the growth of the copper oxide nanoparticles in this reactor. Furthermore, to enhance the lifetime of the stable PEI-coated Cu NPs, the NPs were freeze-dried and stored in test tubes at temperatures below 25 °C to minimize aggregation caused by the NP/NP interactions that initiate nanoparticle aggregation in batch reactor processes.

Mass measurements were conducted to determine the total amount of nanoparticles produced at the end of the flow reactor runtime. Based on these mass measurements made after freeze-drying of the resulting PEI-encapsulated copper oxide nanoparticles, gram-scale synthesis of CuO nanoparticles can be achieved through continuous flow mixing of the complex solution and reducing agent at micromolar and nanomolar concentrations or by increasing the concentration to millimolar to reduce the amount of solution needed and reaction runtime. Since PEI is used to stabilize the Cu nanoparticles as they are synthesized, higher concentrated precursor solutions (ca. 0.20 mM) can be used to produce gram-scale Cu nanoparticle without the occurrence of NP aggregation due to the mitigation of metal-to-metal attraction. That is, increasing the concentration of the precursor solution does not lead to aggregation of the formed copper oxide nanoparticles. In the lower concentration cases, 1000.00 ml of PEI-(Cu²⁺) at 0.5 or 1.0 wt. % is mixed with equal volumes of reducing agent and flowed at a rate of 23 ml/min with an in-residence runtime of 50 min. Similar conditions are used at higher concentrations (>μM); however, in this case, only 125 ml of the PEI-(Cu²⁺) solution is mixed with an approximate total in-residence time of 6 min.

Although many advances have been made in the batch synthesis processing of metallic Cu and Cu-based nanoparticles, the inability to control agglomeration and size distribution during the scale-up process is a common drawback that remains unsolved.^31,32 We demonstrate that by using a flow reactor combined with the chemical reduction process and PEI as the chelating and stabilizing agent, nanoparticles with well-defined sizes and shapes can be achieved at the gram scale without a significant aggregation.³³ In this report, we introduce a flow reactor design [Figs. 1(a) and 1(b)] capable of high-throughput production of CuO nanoparticles with spherical morphology. The flow reactor was constructed from commercially available components that can be purchased at any local hardware store or laboratory supplier. Fundamental to its mode of operation, a peristaltic pump is used to drive fluids through millimeter-diameter tubing at flow rates that range between 0.4 and 85 mL/min, see Fig. 2. A flow rate of 20 ml/min was used to produce PEI-capped Cu nanoparticles. It should be noted that the flow rate used in this reactor will likely depend on the type of metal salt solution and reaction type chosen. Murphy et al. demonstrated the construction and operation of a similar millifluidic reactor capable of facilitating the aqueous gram-scale synthesis of a variety of functionalized gold nanoparticles, including the synthesis of gold nanospheres and nanorods with tightly controlled core diameters and aspect ratios.²⁹ Using that millifluidic reactor, Au NPs were synthesized through the mixing of a growth solution (i.e., HAuCl₄, ligand, and ancillary reagents) and a reaction initiator solution that contained either Au NP seeds or reducing agent. In contrast, in the reactor we report here, the growth solution used contains only the polymer ligand and the metal salt. Subsequently, NaBH₄ (reducing agent) can be used to produce Cu-based NPs.

Our goal was to exploit the ion-pair exchange chemistry that serves as the basis for PAMAM dendrimer methods^21–23 and expand this chemistry to less expansive polymers, thereby facilitating scale-up in a cost-effective way. Rather than using the expensive PAMAM dendrimer, we used the secondary amine groups present in PEI to create the ligand-to-metal-charge transfer exchange to drive chemical reduction of the metal salt to form nanoparticles. That is, in this reaction, coordination chemistry through ligand substitution is used to form bonds around the aqua-copper ions to form a Cu–NH bond.³⁴ It is likely that this approach to nanoparticle synthesis could be easily adapted to large-scale production of other monometallic or bimetallic nanoparticles (e.g., Ni, Co, Pt, CuNi, etc.). When a premixed aqueous solution of PEI and aqueous solution of copper salt was introduced into the flow reactor with various residence times (the amount of time the solution spends flowing in the tubing after it meets at the t-mixer), copper oxide NPs are formed. Upon exiting the flow reactor, the PEI-coated particles were freeze-dried to promote long-term stability until characterization or use.

It has been well documented in the literature that native copper NPs show an UV-visible extinction peak around 500–600 nm, and copper oxide nanoparticles show two extinction peaks: one at 250–350 nm and another at 600–800 nm associated with Cu-O charge transfer transitions and d–d transitions of dispersed CuO, respectively.^34–37 Figure 3 shows UV-vis spectra pertaining to the process by which the Cu nanoparticles are formed. In this figure, a spectrum of the copper salt (solid line) and of the reduced copper nanoparticles (dash line) were analyzed to determine if Cu or CuO nanoparticles are being formed. As can be observed, under the chelation and chemical reduction method, once the PEI-(Cu²⁺)_X has been reduced, copper oxide nanoparticles are formed. The initial absorbance of the copper salt (ca. 808 nm) shifts dramatically after chelation with the amines of the PEI and subsequent reduction via sodium borohydride. That is, the extinction spectrum shows a broad, low intensity peak at approximately 612 nm and another peak at 275 nm that is indicative of the formation of Cu_xO (i.e., Cu₂O and CuO) NPs (Fig. 3).^38–40 

Figures 4(a) and 4(b) show photographs of the PEI-coated CuO NPs immediately after freeze-drying [Fig. 4(a)] and after exposure to ambient conditions for one month [Fig. 4(b)]. Apparently, macroscale physical changes of the freshly prepared PEI-coated CuO NP powder are observed when compared to its 1-month aged counterpart. As can be observed, obvious color and texture changes are evident in the aged sample. To determine the physical and chemical state of the aged PEI-coated CuO NPs, further investigations were conducted using XPS and XRD.

Figure 5 shows the Cu(2p) and Cu(L₃M₄₅M₄₅) Auger transitions observed during XPS analysis of different Cu-containing samples. XPS analysis of the as-synthesized CuNPs revealed the presence of significant concentrations of boron (11%), carbon (35%), nitrogen, oxygen (38%), and sodium (3%), with only a small (1%) Cu concentration. The boron and sodium can be ascribed to the use of NaBH₄ in the NP synthesis while the carbon and nitrogen presumably originate from the PEI polymer adsorbed onto the copper NP surface. A detailed scan of the Cu (2p) region (lower left hand spectrum in Fig. 5) shows the presence of Cu(II) species, as evidenced by the presence of shake-up peaks at ≈9 eV higher than the principal Cu 2p^1/2 and 2p^3/2 peaks.⁴¹ Moreover, a comparison of the Cu(2p) and Cu(L₃M₄₅M₄₅) Auger transitions with those observed from a sputter-cleaned Cu(0) metal foil and a Cu(I) oxide reference (uppermost two sets of spectra in Fig. 5) indicates that the surface of the Cu NPs is composed exclusively of Cu(II)O. Upon sputtering the Cu NPs, the signals associated with the organic and inorganic species (C, N, O, and B) greatly decreased and the intensity of the elemental Cu peaks increased. Figure 5 shows that argon sputtering produced significant changes in the Cu(2p) and Cu (L₃M₄₅M₄₅) Auger line shapes. Unfortunately, Cu(II)O is known to be extremely susceptible to ion beam induced reduction, compromising the ability of Ar⁺ sputtering to provide unambiguous diagnostic information on the depth dependent composition of copper oxides.^42,43 However, since the escape of the photoelectrons being analyzed in XPS (2–3 nm) is comparable to the size of the Cu NPs, the spectra indicate that the majority of the copper atoms are present as Cu(II)O.

XRD and TEM analyses were performed to determine the crystal structure, size, and shape of the PEI-coated copper oxide NPs produced using the flow reactor. Figure 6 shows the typical XRD pattern for the PEI-coated copper oxide NPs along with simulated XRD spectra for several relevant possible products; these simulated patterns were made using crystaldiffract software based on structural parameters from the Crystallography Open Database. It is well documented in the literature that XRD peaks observed at diffraction angles (2θ) of 43.6°, 50.8°, and 74.4° correspond to (111), (200), and (220) reflections of elemental Cu(0) in a face-centered cubic structure. These peaks are not observed in our XRD spectrum. However, the simulations using crystaldiffract confirm the UV-vis and XPS data showing that copper oxide is formed after the reduction of the PEI-Cu²⁺ complex with sodium borohydride. The presence of Cu₂O is confirmed by the observance of XRD peaks at 36.4°, 42.6°, 61.76°, and 73.4° and corresponds to (111), (200), (220), and (311) diffraction planes, respectively, and the presence of CuO is confirmed by the observance of XRD peaks at 32.5°, 35.5°, 38.7°, 48.7°, 63.4°, and 66.2° corresponding to (110), (002), (111), (202), (113), and (311) diffraction planes, respectively.³⁹ As can be observed in Fig. 6, the simulated spectra for Cu₂O and CuO show XRD peak positions that match well with the expected XRD peak positions in our experimental XRD spectrum. That is, the simulated XRD patterns for both copper oxides correspond well with the pattern shown for the experimental data presented in Fig. 6. Furthermore, this supports the XPS observations that the majority of the copper nanoparticles is present as CuO when grown in our system and freeze-dried. The peak at around 26.6° can be attributed to the silicon substrate used to hold the sample during the XRD analysis. Thus, we conclude that Cu NPs grown in our system are primarily in the oxidized (Cu₂O and CuO, hereafter, CuO) form. XRD and XPS measurements of PEI-coated copper oxide NPs that were stored for 1 month in aqueous suspension show similar results for copper oxide formation.

A representative TEM micrograph of the CuO nanoparticles produced using traditional dendrimer-encapsulation templating with G4 PAMAM is shown in Fig. 7(a), and its corresponding particle size histogram is shown in Fig. 7(b). The size of Cu NPs reported here exceed the typical average size (ca. 1.8 nm) that has been reported in the literature when Cu²⁺ ions are chelated using an OH-terminated, G4 dendrimer and chemically reduced using excess NaBH₄.³ A G4 dendrimer has an internal diameter around 4.5 nm which limits the overall average size of the Cu NP that forms within the protected cavity. Hence, observance of Cu nanoparticles with average sizes that exceed the diameter reported for an OH-terminated dendrimer is inconceivable based on the physical approach used to explain the Cu nanoparticle formation process reported by Crooks and co-workers.³ Nanoparticles formed with a generation 4 dendrimer is often reported with the amine (NH₂)-terminated dendrimer structure where the pH has been controlled (pH > 3.5) to initiate growth on the periphery via protonation of the external amine groups. In this case, instead of formation of dendrimer-encapsulated Cu (intradendrimer) nanoparticles, nanoparticles bind at the terminal amines of the dendrimer structure and are stabilized by the dendrimer-to-dendrimer interactions (interdendrimer). This explanation is ruled out in the present study, since OH-terminated PAMAM dendrimers are used to conduct this study. One other viable solution is that a competing reaction took place, yielding two sets of nanoparticle size regimes. Crooks et al. reported that when using the OH-terminated dendrimers, Cu²⁺ ions are present both inside the dendrimer and as hydrated ions in solution when excess sodium borohydride is used as the reducing agent.³ After reduction, it is reported that these excess Cu²⁺ ions form dark precipitates with 9 ± 4 nm in average diameter. It can be speculated that, in this case, it appears that the larger sized nanoparticles are formed due to the excess Cu²⁺ ions available in the solution, skewing the overall size distribution deduced from the TEM images. However, further analysis of the particle size distribution in both the dendrimer and PEI cases, single Gaussian fits with better than 95% confidence were obtained. All Gaussian fit analyses were performed using data obtained from the TEM particle size analysis.

Representative TEM micrographs of the PEI-coated copper oxide nanoparticles synthesized using the flow reactor are shown in Fig. 7(c). Clearly, copper oxide NPs produced under the flow reactor conditions exhibit good uniformity in their shape and size. The prominent shape observed is spherical with an average diameter of 5 ± 1 nm (N = 813 nanoparticles). The average diameter of the nanoparticles is supported by the Debye–Scherrer equation; based on the FWHM of the (111) XRD reflection, the average particle diameter is estimated to be 5 nm, in close agreement with the diameter determined by TEM. It should be noted that although the Cu NPs are prepared and surface protected by the PEI-coating during the synthesis process, exposure to ambient conditions results in immediate oxidation of the PEI-coated Cu.⁴⁴ Hence, based on XRD and TEM analyses, we can conclude that PEI-stabilization within the flow reactor facilitates production of single-phase (monoclinic structure only) Cu_xO NPs with spherical geometries and diameters less than 5 nm during the reaction process. Further verification of particle size using dynamic light scattering was attempted with various solvents and analysis methods but did not reveal reliable sizing data about the primary particle size due to particle agglomeration.

Because one goal of this work is to demonstrate that the chemical reduction method using the PEI-mediated synthesis approach can be used to scale up copper oxide NP production to the gram scale without sacrificing NP quality, it is critical to compare the Cu_xO particles produced using the reactor to those produced by traditional dendrimer-based synthesis (Table I). More specifically, in this section, we provide a comparison of the Cu nanoparticles produced by both methods to demonstrate the effectiveness of the flow system in producing similar well-defined, spherical shaped CuO NPs with scaled up quantities.

Figures 7(a) and 7(c) reveal plan view TEM images of Cu and Cu-based NPs produced by both chemical reduction methods, dendrimer-mediated and flow reactor controlled syntheses, respectively. Both chemical reduction methods produce nanoparticles with small average diameters [Figs. 7(b) and 7(d)] and spherical geometries (Table I). However, in the case of traditional synthesis with the PAMAM dendrimer, Cu ions are coordinated with the internal tertiary amine groups and are subsequently reduced as zero-valent Cu (see Fig. 8) in the internal cavities of the dendrimer, providing control over the particle size as aggregation is prevented by dendrimer encapsulation.³ That is, the metal-to-metal affinity that leads to nanoparticle aggregation has been mitigated by the physical entrapment of the nanoparticles inside the cavity structure of the dendrimer once chemically reduced. Since the dendrimer controls nanoparticle aggregation using cavity entrapment, direct control of the particle size is achieved by controlling the ratio of dendrimer-to-Cu salt and the chemical characteristics of the dendrimer template. In addition, as confirmed in our earlier literature reports on the oxidation of dendrimer-encapsulated Ni(0) NPs, the dendrimer only protects the encapsulated NP from surface oxidation for less than 24 h.²¹ 

Unlike the dendrimer, copper oxide nanoparticles are formed when using PEI as the chelation source under the flow reactor process. However, the flow reactor using PEI overcomes some weaknesses of the one-pot method of mixing; the complexed PEI-(Cu²⁺)_X solution is allowed to react while coming in contact with the reducing agent in a dynamic environment. This allows for direct control over the amount of reducing agent being exposed to the complex solution and, thus, provides some level of control over the rate of nucleation of the CuO NPs and PEI-coating process. Similar to the dendrimer method, with the introduction of the PEI modification, the copper oxide NPs produced in this flow reactor exhibit stability against aggregation into larger oxide particles. Furthermore, the combination of the coating process and introduced flow mixing of the reducing agent allows for scalability of the as-synthesized CuO nanoparticle to gram amounts without any significant aggregation.

We have demonstrated that by adapting the nanoparticle synthesis mechanism reported using the PAMAM dendrimer, a benchtop flow reactor can be used to scale up the production of CuO nanoparticles to the gram level. This reactor is made from all commercially available components that can be found in most local hardware stores or scientific laboratories. As has been reported in this paper, this reactor allows for gram-scale production of small CuO nanoparticles. From a comparative standpoint, although the PEI structure provides a similar chelation method to that of the PAMAM dendrimer, it is clear that in this case a mixture of copper oxide NPs is formed, whereas in the case of the dendrimer, Cu(0) with a native oxide layer is formed.³ Furthermore, this low-cost instrument is easy to assemble and maintain and no seeding solution is needed to initiate the growth of Cu nanoparticles since this system relies on the chelating phenomenon reported for the PAMAM dendrimer, which limits the amount of expensive solutions and time required for preparation.

Note that although this report only addresses the issue of scaling up CuO nanoparticles, the PAMAM dendrimer nanoparticle growth mechanism has been reported to produce a wide range of different metallic nanoparticles with well-controlled shape and sizes.³ Hence, it can be postulated that our reactor design would likely also represent a versatile, portable device capable of scaling up a wide range of metallic nanoparticle with well-defined shapes and sizes.

This work was supported by the National Science Foundation (NSF) under the Center for Sustainable Nanotechnology (No. CHE-1503408). N.V.H.-S. acknowledges support through the NSF Graduate Research Fellowship under Grant No. 00039202. The authors thank the Surface Characterization Facility at the Johns Hopkins University Department of Materials Science and Engineering for use of their facilities. The authors gratefully acknowledge that parts of this work were carried out in the Tuskegee Center for Advance Materials, Tuskegee University, and the Characterization Facility, University of Minnesota, which receives partial support from NSF through the CREST and MRSEC programs, respectively.