Capacitively coupled nonthermal plasma synthesis of aluminum nanocrystals for enhanced yield and size control

Uniform-size, non-native oxide-passivated metallic aluminum nanoparticles (Al NPs) have desirable properties for fuel applications, battery components, plasmonics, and hydrogen catalysis. Nonthermal plasma-assisted synthesis of Al NPs was previously achieved with an inductively coupled plasma (ICP) reactor, but the low production rate and limited tunability of particle size were key barriers to the applications of this material. This work focuses on the application of capacitively coupled plasma (CCP) to achieve improved control over Al NP size and a ten-fold increase in yield. In contrast with many other materials, where NP size is controlled via the gas residence time in the reactor, the Al NP size appeared to depend on the power input to the CCP system. The results indicate that the CCP reactor assembly, with a hydrogen-rich argon/hydrogen plasma, was able to produce Al NPs with diameters that were tunable between 8 and 21 nm at a rate up ∼ 100 mg h−1. X-ray diffraction indicates that a hydrogen-rich environment results in crystalline metal Al particles. The improved synthesis control of the CCP system compared to the ICP system is interpreted in terms of the CCP’s lower plasma density, as determined by double Langmuir probe measurements, leading to reduced NP heating in the CCP that is more amenable to NP nucleation and growth.


Introduction
Nanoparticle (NP) synthesis of metals has been the subject of scientific research for over a century [1]. In many cases, the Nanotechnology Nanotechnology 34 (2023) 395601 (12pp) https://doi.org/10.1088/1361-6528/ace193 Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. focus of metal NP research is on the unique size-dependent characteristics that make these materials interesting for catalysis, optics, and biomedical applications [2][3][4]. Noble metal NPs, such as gold, are readily synthesized in the liquid phase with precise size characteristics [5,6]. The more recent efforts in the synthesis of NPs of Earth-abundant metals constitute a continuously developing field. Among these materials, aluminum (Al) shows desirable properties at the nanoscale for electrode materials in Al cell batteries [7], for the production and storage of hydrogen (H 2 ) gas [8][9][10][11], and in fuelenhancement applications [4]. Al NPs are also promising for ultraviolet plasmonics [12][13][14][15].
Because of Al's high oxidation sensitivity and tendency to agglomerate [16], the production of non-oxide-passivated Al NPs has been challenging to achieve with top-down nanopowder production methods, such as wire explosion and laser ablation [17][18][19]. The native oxide shell, formed due to atmospheric exposure, presents an undesirable barrier to energy and mass transfer for many applications. Additionally, achieving precise repeatable control over particle size and monodispersity is a challenge.
Synthesis of Al NPs has been reported using liquid-phase (wet synthesis) techniques including ligands to control the growth of the particles [20]. The use of catalysts, such as titanium isopropoxide, has been identified as a unique colloidal synthetic technique for Al nanocrystals, but this is associated with the drawback of low purity NP synthesis. However, liquid-phase techniques have recently been shown to address challenges associated with the synthesis process, such that controlled synthesis is possible for shape tuning and size tuning over a narrow range [21].
Solid-phase or high-temperature techniques, such as laser ablation and wire explosion methods, are also commonly used to create Al NPs. Broad size distributions of Al particles have been reported in wire explosion techniques. A large particle size range may be unfavorable for the characterization of sizedependent material properties [22,23].
Pulsed laser ablation has been used to form colloids of Al NPs in both organic and inorganic solvents in the absence of chemical catalysts. It has been reported that the size of the particles was dependent on the solvent used for this technique and that the size distributions were broad (not monodisperse) [24,25]. These techniques offered size tuning through the use of different solvents. Still, they had similar drawbacks to wet synthesis techniques since they form particles suspended in a solution that constrains the possible applications of the Al NPs.
Among the gas-phase synthesis processes for NPs, nonthermal plasmas are known for excellent control of particle size and relatively narrow size distributions [26]. Previous work using an inductively coupled plasma (ICP) successfully demonstrated the synthesis of Al NPs from an aluminum trichloride precursor (AlCl 3 ) [27]. However, this work, using a plasma reactor of essentially identical dimensions as the one in the present study, achieved uncharacteristically little control over particle size and a relatively poor precursor utilization of <1%, leading to a small production yield of ∼10 mg h −1 [27].
The current study hypothesizes that the high plasma density associated with ICP plasmas prevents appropriate control of the Al NP synthesis and that improved control can be gained by using a lower plasma density capacitively coupled plasma (CCP). NPs immersed in plasmas are heated to temperatures exceeding the gas temperature by surface reactions and interactions among active species, with the rates of these reactions expected to increase with increasing plasma density [28][29][30][31]. It is plausible that Al NPs cannot form in the bulk of the ICP plasma due to too intense heating, but only in the downstream afterglow where the plasma density decreases rapidly. This leads to poor control of the synthesis process.
In this study, we use a lower plasma density CCP, where the plasma density can be carefully controlled via the applied radio frequency (RF) power. We investigate how plasma power and H 2 can aid the reduction of an Al precursor and be used to control the Al NP size and material characteristics of the product.

Capacitively coupled plasma reactor and conditions
Crystalline Al NPs were synthesized in a single-stage nonthermal plasma reactor, similar to that which was previously discussed in detail by Beaudette et al [27]. In this work, however, a CCP was utilized to test the hypothesis described above. A schematic of the experimental setup is shown in figure 1. Precursor gases consisting of AlCl 3 vapor diluted with argon (Ar) and H 2 were flowed through a length of two 25 mm outer diameter, 21.8 mm inner diameter, 254 mm long quartz tubes, which were coupled with Swagelok Ultra-Torr fittings. A plasma was generated by applying a 13.56 MHz RF (AE RFPP RF-5S Model from Advanced Energy) through a custom-made impedance matching network to two externally mounted copper ring electrodes, which were spaced approximately 15 mm apart.
Anhydrous AlCl 3 (2.5 g, ACS reagent, 99.999%, Sigma-Aldrich) was added to a sealed stainless-steel vessel attached to the Ar gas line. It was heated using an externally wrapped Fisher Scientific fiberglass heat tape connected to a Watlow EZ-Zone thermal controller at a constant temperature of 110°C, as indicated by a Type K thermocouple, to sublimate the precursor material (AlCl 3 ). The sublimation chamber was heated for 10 min after reaching 110°C before the reaction. Ar flowed upstream at a constant flow rate of 75 standard cubic centimeters per minute (sccm) through the AlCl 3 sublimation chamber, serving as the diluting carrier gas. The estimated average flow rate of AlCl 3 was 113 mg h −1 based on the difference between the starting mass of the precursor and the remaining mass after 22 min of operation at a carrier gas (Ar) flow rate of 75 sccm. The flow rate of H 2 varied between 0 and 150 sccm. The supplied power varied between 40 and 80 W. Reactions were carried out at pressures between 3 and 5 Torr, depending on the flow rate of H 2 .
The collection of the particles occurred downstream of the plasma region on a glass optical microscope slide by impaction. A 0.5 mm slit nozzle was used to facilitate the collection of Al NPs. During the removal and transfer of the particles from the deposition stage, an airlock assembly of two manually operated valves kept the samples in an oxygenfree environment to minimize air and moisture exposure.
Basic plasma parameters, electron temperature (T e ), and positive ion density (n i ) were measured using a prototype double Langmuir probe. The cylindrical axis of the probe was aligned with the centerline of the reactor tube. Tungsten tips of 10 mm in length and 0.075 mm in diameter were used. The I-V characteristics were measured by a source meter (Tektronix 2450, Tektronix, USA) connected to the double probe. More details can be found in [32].

Aluminum nanoparticle heating
A simple energy balance was used to estimate the temperature of Al NPs in the plasma for ICP and CCP conditions. The equilibrium temperature of a NP was calculated as where G is an NP heat gain term and L is a heat loss term.
The gain term describes the heat released by electron-ion recombination on the particle surface. For very small NPs with a diameter of around 10 nm, the orbital-motion-limited theory [33] is well suited to calculate the collision frequency between NPs and charged species. Therefore, the generation term is calculated as [34]: where n i is the ion density of the plasma, S represents the surface area of the Al NPs, T i is the ion temperature, and m i is the mass of an ion. Here, E ion is the ionization energy of Ar,  The loss term L is calculated as [34,35]: The first term describes the heat conduction loss to the background gas, and the second term describes the heat loss due to evaporation. The radiation cooling term is found to be negligible for an NP [35][36][37]. For the conduction term, n v is the background gas density, T g is the gas temperature, m g is the mass of the background gas molecules, and T p is the particle temperature. For the evaporation term, Dh v is the enthalpy of vaporization per Al atom. Here, is the number density of evaporated Al atoms, where p v is the vapor pressure of the Al atoms at the gas/surface interface given by the Antoine equation, with temperature-dependent parameters from NIST (National Institute of Standards and Technology) [35].
Al is the mean thermal speed of Al vapor, where m Al is the mass of an Al atom.

Characterization methods
X-ray diffraction (XRD) measurements were taken using a Bruker D8 Discover diffractometer equipped with a cobalt (Co) Kα source (wavelength (λ) = 1.79 Å). The XRD patterns were converted to copper (Cu) Kα wavelengths (λ = 1.54 Å) for data analysis, which was performed using the Material Data Incorporated Jade 8.0 software package. The size of the crystallites was determined from the XRD peak broadening using the Scherrer equation analysis with a shape factor of K = 0.89 [38].
X-ray photoelectron spectroscopy (XPS) was performed using a PHI VersaProbe III XPS and UPS system with a monochromatic Al Kα anode x-ray source (photon energy = 1486.6 eV) and a hemispherical analyzer. The binding energy of C 1 s at 284.6 eV was used as a reference. XPS survey scans were taken with a pass energy of 280 eV. A 55 eV bandpass energy was used to collect high-resolution scans. Peaks were fitted and atomic percentages were calculated using PHI's Multipak software v9.0.
Transmission electron microscopy (TEM) was performed on a Thermo Scientific Talos F200x electron microscope at 200 kV accelerating voltage. The samples were directly deposited onto 3 mm diameter lacey carbon-supported TEM (Cu) grids from Pacific Grid-Tech. Particle dimensions were manually measured using the open-domain scientific image analysis software ImageJ for the sizing of approximately 200 particles. Geometric mean sizes (diameters) were reported together with geometric standard deviations.
The samples were briefly exposed to the air during sample transfer in XPS and TEM measurements. XRD measurements were performed in ambient air. The samples were otherwise maintained in a dry nitrogen glovebox (an oxygenfree environment) after synthesis.

Results and discussion
In this section, we first report and discuss the experimental results on the synthesis of Al NPs in the CCP reactor and compare them to previous results from those produced by an ICP reactor. We then use plasma diagnostics and the NP heating model to interpret the differences between the CCP and ICP systems.

Effect of plasma power
We explored the impact of input plasma power on the crystallinity, particle size distribution, and production yield. In general, we found that the CCP system was capable of producing Al NPs at RF powers that were about three times lower than previously required by the ICP system [26]. Figure 2 shows the XRD patterns of Al NPs at plasma powers varying from 40 to 80 W. The H 2 flow rate was kept at 150 sccm, and the total pressure of the reactor was maintained at approximately 5 Torr for this reaction set. The XRD patterns revealed two peaks at 38.4°and 44.7°, which correspond to the (111) and (200) lattice planes, respectively, of standard face-centered cubic (fcc) Al (JCPDS 03-065-2869). Figure 3 exhibits a comparison between crystallite sizes, determined using the Scherrer equation [38] for each respective plasma power based on the (111) peaks in the diffraction patterns, and geometric mean particle diameters, manually calculated using TEM images of the samples. The gas flow rates for this sample set were kept constant at conditions that were most suitable for this experimental system. In general, there was good agreement between the observed geometric mean particle diameter from TEM and the estimated crystallite size from XRD. As shown in figure 3, crystallite sizes increased from 8 nm to 18.5 nm with increasing plasma power. Geometric mean diameters also follow a similar trend, increasing from 8.2 to 20.8 nm with increasing plasma power. This trend suggests that there is enhanced particle growth with power, leading to larger particles. This observation is consistent with a prior study, where metal NP diameters were likewise shown to be influenced by input plasma power [39]. It is noteworthy that particle size for most other NP materials produced by nonthermal CCP systems, such as silicon, scales with the gas residence time in the reactor, and not with plasma power [26,28]. This suggests that the growth mechanism of metal NPs may be different from other NP materials, which should be investigated in the future. Figure 4 shows the TEM images, selective area electron diffraction (SAED) patterns, and particle size distributions with geometric mean diameters and geometric standard deviations at 40, 60, and 80 W plasma power, respectively.
Note that the scale for the high-magnification TEM images (inset) at each power varies for comparison.
As shown in figure 4(a), Al NPs form chain-like agglomerates consisting of seemingly spherical particles at all three different plasma powers of 40, 60, and 80 W.
We observed a geometric mean of 8.2 nm with a geometric standard deviation of 1.3 for the Al nanocrystals produced at 40 W. In comparison, we observed geometric means of 11.9 nm and 20.8 nm with standard deviations of 1.2 and 1.4 for 60 and 80 W, respectively as shown in figure 4(b). This confirms that plasma power variation in the current CCP reaction conditions enables size tunability of Al nanocrystals. Geometric standard deviation is indicative of the size dispersity of the Al NPs, which suggests higher powers above 60 W could lead to a slightly polydisperse Al particle distribution. Geometric standard deviations in the previous ICP plasma study [27] varied from 1.3 to 1.7, which suggests that the CCP system produces more monodisperse Al NPs than the ICP system.  -065-2869). Additionally, the SAED patterns intensified with the increase in plasma power. This observation is consistent with prior work performed in dusty plasmas, where plasma power was noted to result in NP heating when the particles were immersed in the ionized gas, leading to crystallization [29]. The authors comment that a precise understanding of the interaction between NPs and the nonthermal plasma is not fully understood; particle heating can still be shown to affect the crystallization of particles. Moreover, we hypothesize that particle heating may be a contributing factor for the increased agglomeration of crystallites, which was observed at higher powers (80 W) and causing larger particle sizes.
The NP production yield was assessed by weighing the deposited material after 3 min of continuous deposition in an oxygen-free environment. Figure 5 exhibits the impact of input power on the particle production yield, with a constant H 2 and Ar flow rate. The production yield increases significantly from 40 to 50 W at a rate of 45 to 96 mg h −1 . The yield stayed relatively constant at 90 mg h −1 at 60 W but decreased again above 60 W. The yield of 96 mg h −1 at 50 W plasma power is approximately a ten-fold increase compared to the previous ICP study [27].
The yield measurements in figure 5, show that there is a decrease between 60 and 80 W. The reduced yield at 70 and 80 W that we indicated here can be explained by observable increased losses to the interior reactor walls in the form of both Al-Cl clusters and Al metal film formation, as well as a decreased tendency to form Al NPs in the plasma region due to particle overheating, as discussed in section 3.3. Conversely, yields at 50 and 60 W indicated that loss pathways, such as particle overheating, were less prevalent.
The operation time for the reactor was recorded and the precursor (AlCl 3 ) conversion rate was estimated since these factors were important for improving the yield and optimizing experimental conditions. The average consumption rate of AlCl 3 was determined to be 113 mg h −1 , calculated by its initial weight (2.5 g) loaded in the precursor chamber divided by the average reactor runtime (22 min) when the power was varied and the Ar and H 2 flow rates were constant at 75 and 150 sccm, respectively. The estimated conversion rate, in terms of moles of Al (collected as NPs) to moles of AlCl 3 fed into the reactor, was ∼7%, assuming the flow of AlCl 3 did not change when the power was varied. This estimated conversion rate accomplished with the CCP reactor was more than a seven-fold increase compared to the previous ICP study [26], where the conversion yield was <1%. Overall, this work has clearly demonstrated the success of the improved mass yield, size control, monodispersity of NPs, and precursor conversion rate using H 2 -rich plasma in a CCP system.

Effect of hydrogen (H 2 )
We further investigated the impact of the H 2 flow rate on the resultant crystallinity, particle size distribution, and production yield of Al nanocrystals, while keeping a constant Ar flow rate (75 sccm) and a plasma power (60 W). The Ar flow rate was kept constant to ensure a constant precursor (AlCl 3 ) feed rate since Ar acted as the carrier gas. The plasma power was kept constant at 60 W since it was observed that this  yielded a desirable NP size distribution and yield rate. It was shown previously that H 2 is a key component that directly dictates the precursor conversion efficiency in the synthesis of crystalline NPs when using a chlorinated precursor, such as in the case of SiCl 4 in a nonthermal plasma [40]. In contrast, previous work with an ICP system reported that Al nanocrystals can be synthesized with AlCl 3 , with or without the presence of H 2 . However, higher NP crystallinity with H 2 was found. Here, H 2 was identified as a scavenger for chlorine (Cl) to enable particle nucleation since HCl could be formed as a result of a H 2 and Cl reaction. However, its actual contribution to particle nucleation and crystallization is still poorly understood. Figure 6 demonstrates the XRD patterns of four samples produced using different H 2 flow rates. Without H 2 (0 sccm), the XRD pattern was featureless, suggesting that the NPs were amorphous without a H 2 flow. We observed peaks corresponding with fcc Al in the samples produced with 50, 100, and 150 sccm of H 2 . Additionally, by increasing the H 2 flow rate to 100 sccm and above, the crystallinity of the sample was significantly enhanced, as evident from the more pronounced (111) and (200) peaks. Therefore, our findings suggest that H 2 is vital for the synthesis of crystalline Al NPs in the currently studied CCP conditions. We calculated crystallite sizes from the (111) peak in the XRD patterns for 100 and 150 sccm of H 2 , respectively, using the Scherrer equation. The (111) peak was used since it has the highest relative intensity in the XRD pattern. The estimated crystallite sizes are 15, 13, and 12 nm for 50, 100, and 150 sccm of H 2 , respectively, indicating that the H 2 flow rate has only a minor influence on crystallite sizes. To confirm the above observation, the particle size and morphology of Al NPs produced under similar conditions were also evaluated by TEM. Figure 7 shows representative TEM images, size distribution histograms, and SAED patterns of Al NPs synthesized with 0, 50, and 100 sccm H 2 flow rates. Figure 4 includes the image and size distribution histogram for 60 W plasma power and a 150 sccm flow rate of H 2 .
These TEM images exhibit spherical Al NPs agglomerated in clusters for all the discussed above. However, a clear shell structure with an approximate thickness of 2-6 nm was observed in Al NPs in the absence of H 2 . We assume that surface oxidation is prevalent in these conditions, which leads to the formation of a thick Al 2 O 3 shell after exposure to the atmosphere during XRD characterization. The impact of the Al 2 O 3 shell will be further discussed in the SAED analysis.
The geometric mean diameters for 50, 100, and 150 sccm are 15.9, 14.7, and 11.9 nm with geometric standard deviations of 1.3 and 1.2, respectively, and are indicative that there is little appreciable difference in particle size (<5 nm) with increasing H 2 flow rate. However, at 0 sccm of H 2 , Al NPs tend to be larger with a geometric mean of 20.1 nm and a geometric standard deviation of 1.2. Due to the arrangement of the reactor, the flow rate of H 2 gas impacts the total flow rate through the plasma region. Increasing H 2 results in a decrease in the residence time for particles within the CCP. Decreasing the residence time of the particles results in the size changes observed in figure 7, where increased H 2 flow resulted in a slight reduction in the average particle size.
The sampling limitations associated with the size distributions (N = 199-205) led to the appearance of a slightly broadened size distribution as the flow rate of H 2 was increased. Despite the variation between geometric standard deviations, the distribution histograms presented in figure 7(b) are indicative of a generally narrow distribution in all cases. Geometric standard deviations of between 1.2 and 1.3 in all three reaction conditions imply narrow size distributions of Al NPs.
In this particle size range, it was predicted that Al NPs would exhibit pyrophoric combustive behavior [41]. Pyrophoricity of the sample deposited at 60 W, 75 sccm Ar, and 150 sccm H 2 was observed. An image is presented in figure  S1 showing the combustion.
The SAED patterns in figure 7(c) confirm that Al NPs produced under varied H 2 flow rates, including without H 2 , are crystalline and can be assigned to the (111), (200), (220), and (311) diffraction planes for the fcc Al crystallographic structure. Interestingly, this is different from what we observed from XRD, where a featureless diffraction pattern was obtained for the sample produced without H 2 . We postulate that the thick amorphous alumina shell together with poor crystallinity (figure 7(a)) may contribute to the featureless XRD pattern. While this suggests that crystalline Al NPs can be synthesized with or without the addition of H 2 , the addition of H 2 clearly improves the crystallinity of Al NPs. Note that the inset images in figure 7(a) are intended to provide selected higher magnification examples of single particles to demonstrate the general NP morphology and size.
We assessed the yield at various H 2 flow rates in figure 8, similar to the method described in section 3.1. The range was chosen based on experimental observations that suggested the occurrence of detectable crystalline Al NP at approximately 100 sccm H 2 ( figure 6). The production yield varied between 72 and 90 mg h −1 . The fluctuations observed in the graph are indicative of the error associated with the measurements, and the trend of the yield with the H 2 flow rate was observed to be relatively constant. Therefore, the NP yield was less sensitive to a change in H 2 flow rate than a change in plasma power for this experimental setup.
XPS was performed for the samples deposited at H 2 flow rates of 0-150 sccm to evaluate the surface composition with varying H 2 flow rates. The survey scans for each sample confirmed the presence of Al, C, O, and Cl species. The XPS survey scan for the Al NP sample deposited at 60 W, 75 sccm Ar, and 150 sccm H 2 is presented in figure S2. A highresolution Al 2p spectrum exhibits two peaks, one at 74.6 eV, corresponding to oxidized Al (Al 3+ ), and the other at 72 eV, corresponding to metallic Al (Al 0 ), further confirming the  synthesis of Al NPs. The high-resolution Al 2p spectrum for the Al NP samples deposited at 60 W, 75 sccm Ar, and 50, 100, and 150 sccm H 2 is presented in figure S3.
The surface composition of the samples by atomic percent at 0, 50, 100, and 150 sccm H 2 flow rates based on the O 1 s, Cl 2p, and Al 2p are plotted in figure 9. Atomic percentages were calculated from high-resolution scans using PHI Multipak software. The Al atomic content varied from 13.9% at 0 sccm, 18.6% at 50 sccm, 29.5% at 100 sccm, and 28.8% at 150 sccm. It should be mentioned that XPS only probes the Al NP surfaces, and that the chemical composition shown in figure 9 is not necessarily characteristic of the bulk of the NPs.
These data suggested that the addition of H 2 results in an increase in the surface Al content. In the absence of H 2, the Cl content of the NP surface was the highest and it decreased with increasing H 2 flow, suggesting that H 2 scavenges a minor amount of Cl from the surface of the NP after formation. However, it should be noted that the decrease in Cl content with the addition of H 2 is not significant (<5% change).
In the plasma, the formation of Al particles from AlCl 3 in the presence of H 2 results from precursor decomposition to form chlorinated species. The H 2 in the plasma reacts to form HCl, which can be corrosive to the Al NPs, resulting in etching. However, Cl 2 has a more pronounced etching effect on Al due to a higher rate coefficient when compared to that between Al and HCl [42]. Therefore, the addition of H 2 tends to reduce etching behavior from the Cl 2 and drives the reaction to favor Al metal.
The O content did not change significantly between samples where H 2 was added. Furthermore, the presence of Al-O bonds confirmed the presence of a thin oxide layer on the stable Al NP, which is consistent with the TEM images (figure 7) [43]. To summarize, the addition of H 2 does not significantly influence the particle size, distribution, or morphology but has a strong impact on the crystallinity of Al NPs.

Double probe measurements and particle heating
We hypothesized that the NP temperature has a major impact on Al NP particle nucleation and growth, and that NP  Figure 11. Variation of the temperature of a 10 nm diameter Al NP with the ion density, ranging from 0 to 5 × 10 12 cm −3 linearly and 5 × 10 12 cm −3 to 10 14 cm −3 logarithmically, respectively. temperature in the plasma differs between the CCP described in this study and the previously published ICP system [27].
To evaluate the temperature of Al NPs exposed in a plasma, double probe measurements [32,44] were performed to obtain the ion densities in both the CCP and ICP systems, as shown in figures 10(a) and (c). In the CCP using pure Ar, 60 W input power yields ion densities between the two electrodes of around 0.6 to 3 × 10 12 cm −3 ( figure 10(b)), while the AlCl 3 precursor injection reduces the ion densities to between 2 and 7 × 10 11 cm −3 for the same 60 W input power.
To replicate the plasma conditions used in the previous ICP study [27], the electrode pair was replaced with a 7-turn 3.2 cm long induction coil. The discharge was initiated in the CCP mode (also called E-mode) and the power increased to 50 W, at which the transition to the ICP mode (also called H-mode) occurred. Because the electron density increases by more than an order of magnitude during the E-H mode transition due to the improved inductive power coupling efficiency with increasing plasma density [45], the power could be lowered to 20 W while maintaining ICP mode operation. Higher powers in the ICP mode exceeded the range of our source meter. With pure Ar, 20 W input power already yielded approximately 1 × 10 13 cm −3 ion density in the center of the induction coils, while the ion density decays exponentially by more than one order of magnitude per centimeter towards the plasma periphery ( figure 10(d)). ICP operation with precursor injection required a minimum plasma power of 135 W to avoid quenching of the inductive mode. Due to the high-power input and plasma volume expansion at this power, the ion density is around 6 × 10 12 cm −3 . The center ion density is expected to be of the order of 10 13 to 10 14 cm −3 but could not be measured directly because the probe currents exceeded the measurement limit of our source meter. Figure 11 shows the temperature of a hypothetical 10 nm diameter Al NP derived from the heating model in section 2.2. To show conditions for both the CCP and ICP in one plot, the ion density varies linearly from 0 to 5 × 10 12 cm −3 and logarithmically from 5 × 10 12 cm −3 to 10 14 cm −3 . For ion densities less than 3 × 10 12 cm −3 , the NP temperature increases linearly with ion density since conduction cooling is the dominant cooling term for temperatures less than ∼1200 K. This ion density range correlates with the ion densities observed in the present CCP system. However, as ion density increases above 10 13 cm −3 , which we found for the ICP system, the Al NP temperature increases much slower with ion density. This is caused by the rapidly increasing heat loss due to NP evaporation when the particle temperature exceeds around 1350 K. At these high temperatures, the vapor pressure increases exponentially with temperature and the evaporation cooling term starts to dominate over conduction heat loss.
In the present CCP system with precursor flowing, the maximum plasma density measured was 7 × 10 12 cm −3 . Due to its vapor pressure increasing exponentially with temperature, NP evaporation does not play a significant role in insufficient plasma heating. This suggests that plasma densities in the CCP can be controlled in a range where NP nucleation and growth can occur in the entire plasma volume without evaporation being a factor. This enables good control over particle size and agglomeration by having plasma power as an effective control parameter.
By contrast, in the ICP system [27], the large ion densities >10 13 cm −3 lead to much higher NP temperatures and likely cause rapid evaporation of NPs. Hence, NP nucleation and growth in the dense center of the ICP plasma may not be feasible, and NP nucleation is likely limited to the plasma periphery downstream of the center, where the plasma density is maximum in the ICP system. This allows for more loss of Al atoms to the reactor walls and likely caused the lower NP yield for the ICP system compared to the CCP. Because the plasma density also decreases rapidly in the downstream ICP periphery, particles presumably become less negatively charged, which would increase NP agglomeration and lead to the broader size distributions observed in the ICP [27].

Conclusion
Low-pressure nonthermal plasma synthesis of Al NPs with controlled sizes was achieved in a CCP reactor using anhydrous AlCl 3 as a precursor. The results successfully demonstrated the production of Al NPs with high crystallinity and increased yields up to nearly 100 mg h −1 , which was ten times greater than that previously achieved (10.2 mg h −1 ) by an ICP reactor. The addition of H 2 at flow rates up to 150 sccm has been shown to significantly improve the crystallinity of the Al NPs. Furthermore, plasma power provided control over the Al NP size distribution in the power range from 40 to 80 W. The mean geometric particle diameter increased from 8.2 to 20.8 nm with a geometric standard deviation ranging from 1.25 to 1.40, which tended to increase as the input power was increased. The resultant Al NPs from this reactor showed apparent pyrophoricity at room temperature and atmospheric pressure, as pictured in the supporting information, affirming the high reactivity predicted for Al.
The control over particle size and particle monodispersity was found to be superior in this work when compared to previous work using an ICP reactor. This can be interpreted by the differences between the CCP and ICP systems based on the plasma densities in these two systems and the associated differences in NP heating behavior within each plasma. In the CCP system, plasma densities are in a range where the particle temperature is controlled by conductive cooling to the neutral gas and evaporative cooling does not play a role. Hence, particle nucleation and growth are possible in the entire plasma volume. In the ICP system, due to the much higher density in the plasma center, evaporation is rapid and NP nucleation is likely only possible in the plasma periphery. With ICP, the plasma density drops rapidly, which presumably leads to a lower NP yield, less NP charging, and less control over the growth process. Hence, the CCP scheme presented here offers superior control over NP size and material yield compared to that produced by an ICP scheme. Future work with AlCl 3 should consider the use of optical emission spectroscopy to more completely understand the nature of the Cl species within the plasma. This technique should be used to more comprehensively assess the etching of Al NPs in the plasma.