Pool boiling experimental investigation on in-situ hierarchical Cu(OH) 2 nanograss

This study fabricated the in-situ Cu(OH) 2 hierarchical nanograss surface (HNS) via immersion method for simultaneous enhancement of the critical heat flux (CHF) and heat transfer coefficient (HTC) in diverse experimental conditions. Nanograss stripes were patterned, and then in-situ grown on the copper substrate through the chemical modification. The effect of the Cu(OH) 2 HNS on the pool boiling heat transfer performance was systematically examined. Furthermore, the optimized HNS which can exhibit the highest CHF and HTC was identified based on comparative experiments. It is found that the numerous nanoscale cavities existing in the HNS can act as the active nucleation sites for facilitating the boiling process. Experimental results reveal that the HNS can reduce the independent bubble departure diameter, increase the bubble departure frequency and significantly delay the bubble mergence due to much improved capillary pumping and replenishment of cooling liquid. According to the experiments, the CHF on the optimized HNS sample, were improved by 97.8% compared with the smooth surface and the HTC is enhanced to 2.4 W/cm 2 K, which is equivalent to an augmentation of 25.7% compared to that on smooth surface. Meanwhile, the HNS can improve the heat transfer uniformity and stability with temporal temperature variations less than 1 K at CHF, which is pivotal to the efficient thermal


Introduction
With the increase in energy density for advanced energy and electronic systems such as advanced radars, lasers, space technology, concentrated photovoltaics, electronic cooling, thermal water desalination and nuclear clear power generation, thermal management technologies with ultra-high heat flux is vitally needed (Sun et al., 2022;Inbaoli et al., 2022;Chu et al., 2022;Egbo et al., 2022;Marie et al., 2022). The large latent heat of vaporization on a two-phase system as a promising approach can meet the heat dissipation requirements, and therefore pool boiling heat transfer has been widely investigated. The onset of boiling (ONB), heat transfer coefficient (HTC), and critical heat flux (CHF) are three important parameters in evaluating the pool boiling process. ONB represents the boiling inception and the beginning of the transition from single-phase heat transfer to two-phase nucleate boiling heat transfer. Through reducing the ONB at lower wall superheat the heat transfer efficiency can be enhanced. HTC depicts the heat transfer efficiency of the boiling system and a higher HTC can significantly inhibit the energy consumption. CHF represents the heat transfer capacity which is the heat transfer limit of the boiling system, and thus a higher CHF can guarantee the safe operation at a high heat flux.
As continuous two-phase convection in heterogeneous phase change phenomenon are strongly dependent on solid-liquid interfacial characteristics such as morphology and wettability (Hu et al., 2021;Eok Kim and Seok Oh, 2022;Kangude and Srivastava, 2022;Gao et al., 2021;Godinez et al., 2021;Jiang et al., 2022;Liu et al., 2022), researchers have concentrated on the manipulation of the balance between the bubble nucleation and the hydrodynamics of the dry-out space to improve pool boiling heat transfer performance. Engineering surfaces to create pits and cavities for increasing boiling nucleation sites, inducing structures such as fins for increasing the available heat surface area, and modifying morphology and wettability for improving capillary wicking, are the most common approaches to enhance pool boiling heat transfer. Using the functional surfaces at microand nanoscales, the surface morphology and wettability can be controlled to manipulate the bubble nucleation and dynamics including the liquid and vapor transport for pool boiling enhancement. A number of research groups have reported the use of micro-and nanostructures on boiling surfaces such as micro fins (Liu et al., 2022;Kong et al., 2018;Ćoso et al., 2012;Zhou et al., 2019;Zhang et al., 2018), micro pillars   Zhang et al., 2021;Moon et al., 2016;Duan et al., 2020;Chu et al., 2012), micro meshes Vaartstra et al., 2020;Kim et al., 2017), micro channels (Cooke and Kandlikar, 2011;Gouda et al., 2018;Pi et al., 2020;Kwak et al., 2018;, nanowires (Lee et al., 2019;Jo et al., 2018b;Yao et al., 2011), nanotubes (Dharmendra et al., 2016;, and nanopores (Rishi et al., 2019;Lu et al., 2016;Sezer et al., 2019;Arya et al., 2015;Nasersharifi et al., 2018) to increase the CHF and HTC. Among these microand nanoscale structures, nanoscale structures can increase the capillary pressure for driving the surrounding liquid to the nucleation sites, which in turn rapidly releases bubbles prior to their coalescence into the vapor film. However, the excessive nanoscale structures may hinder the flow channels and reduce the liquid permeability, in which case the bubbles accumulate and merge into film vapor. On the contrary, the microscale can provide the flow channels with low resistance and facilitate the liquid permeability. Therefore, the micro/nano hybrid structures with strengthened capillary pressure and enhanced liquid permeability would be the ideal structure for improving pool boiling heat transfer performance. To date, the micro-/nanoscale hybrid structures can be divided into three categories involving microchannel with nanowires (Chen and Li, 2019;Lee et al., 2018;Chen et al., 2020), microchannel with pores (Patil and Kandlikar, 2014;Jaikumar and Kandlikar, 2015;Ha and Graham, 2019), and micro/nano porous structures (Wang et al., 2018;Khan et al., 2019). For instance, Lee et al. (2018) proposed a micro-nano hybrid surface as a promising method to enhance the boiling heat transfer, and found that the combining efforts of nanowires for liquid replenishment and microcavities for bubble generation could delay the CHF phenomenon. Jaikumar and Kandlikar (2015) fabricated the open microchannel with sintered porous coating for the pool boiling heat transfer enhancement. As reported, the sintered-throughout porous microchannel can provide the highest CHF and HTC through strengthened liquid-vapor micro convections. By using the micro/nano porous, Wang et al. (2018) found that the micro/nano porous structure can accelerate the bubble growth rate and the bubble departure for pool boiling heat transfer enhancement. However, most fabricating methods such as chemical bath, chemical vapor deposition, chemical etching and electrochemical deposition are complex, which limits their applications. It is noted that the method of solution immersion as a promising approach has been used to fabricate the nanostructures. For example, Li et al. (2021b) fabricated the hierarchical nanoengineered wicking surfaces through immersion method to investigate the wicking capability and durability, as well as the pool boiling heat transfer performance. As reported, year-long continuous surface wicking capability on the hierarchical nanoengineered surface is much higher when compared to single-tier surface structures, leading to the superior pool boiling heat transfer performance. Cheng et al. (2022) fabricated a stable and dense boehmite layer on the aluminum surface which can increase the effective nucleation sites, wettability and capillary rewetting capability for enhancing pool boiling heat transfer. In these previous researches, the surface morphology grows at random on the substrate, resulting in the bubble nucleation and growth on the entire surface range, which may influence the continuous liquid replenishment and vapor escape, leading to the deterioration of the CHF enhancement. Moreover, there has been relatively few researches on enhancing pool boiling performance using this type structure through inhibiting the bubble coalescence by manipulating the bubble dynamics.
In this work, we fabricate a hierarchical nanograss surface (HNS) using immersion method to enhance the pool boiling performance and thermal uniformity and stability, by delaying the bubble mergence through controlling of the nucleation sites and bubble size. Pool boiling experiments were conducted under the saturated temperature at atmosphere pressure conditions. The results indicate that the HNS surface is beneficial to enhance the CHF and HTC simultaneously, resulting from the effect of the unique surface morphology on delaying the bubble mergence in various scenarios.

Materials
The Cu(OH) 2 nanograss was fabricated by the immersion method in the solution containing sodium hydroxide (NaOH, AR, ≥ 96.0%) and ammonium persulfate ((NH 4 ) 2 S 2 O 8 ), which were purchased from the Sinopharm Chemical Reagent Co., Ltd. To grow the Cu(OH) 2 nanograss, a solution comprising 1.25 M NaOH and 0.1 M (NH 4 ) 2 S 2 O 8 in 100 mL deionized (DI) water was prepared and then stirred for 30 s at 25 • C.

Preparation of hierarchical nanograss surface (HNS)
Before the fabrication of the HNS, the 20 mm × 20 mm × 3 mm copper (Cu) substrates were polished with 2000 grid SiC sandpaper for removing the impurities on the substrates. Then, the Cu substrates were cleaned in the ultrasonic acetone solution for 15 min to remove the residues. Subsequently, the polished substrates were rinsed with the ultrasonic ethanol and DI water for 15 min, respectively. At last, these Cu substrates were placed in an oven and heated at 65 • C for 120 min.
The fabrication of the HNS can be divided into five steps: patterning, growing, removing, cleaning and drying. Fig. 1a shows the fabrication process of the HNS on the Cu substrate. Firstly, the rectangular insulating tapes were patterned on the Cu substrate with a spacing w, on which no seeding solution can touch the Cu substrate to grow the Cu(OH) 2 nanograss. Then, this taped Cu substrate was immersed into the 100 mL solution containing 1.25 M NaOH and 0.1 M (NH 4 ) 2 S 2 O 8 in a 250 mL glass beaker for approximately 5 min. Due to the barrier layer of the insulating tapes, the Cu(OH) 2 nanograss can merely grow on the spacing between the adjacent insulating tapes with the width of w. In the growing step, the initial color of the solution in the glass beaker became light blue immediately. After the chemical modification in the glass beaker, the insulating tapes on the Cu substrate were removed and the black stripes consisting of Cu(OH) 2 can be observed. This was followed by that the modified Cu substrates were immersed into the DI water to remove the residues. At last, the fabricated samples were placed into the vacuum oven for drying at 60 • C, and then the Cu(OH) 2 nanograss could be prepared. The fabricated samples are shown in Fig. 1b. The width w of the Cu(OH) 2 nanograss stripes in the samples varied from w = 0, 2, 3, 3.5, 4.5 and 20 mm corresponding to the S1, S2, S3, S4, S5 and S6 sample, respectively. The width of the gap d between the nearest Cu(OH) 2 nanograss stripes in the different samples was 20, 1.6, 1.25, 2, 3.25, and 0 mm corresponding to the S1, S2, S3, S4, S5 and S6 sample, respectively. Table 1 illustrates the geometric dimensions of the different HNS samples. As the w increased, the area of Cu(OH) 2 stripes (A Cu(OH) 2 ) increased to cover the whole Cu substrate (A) as much as 100%. The more Cu(OH) 2 nanograss area could offer higher nucleation site density for bubble generation. With the different spacing between the nearest Cu(OH) 2 nanograss stripes, different capillary force can be generated from the gap between nanograss, thereby improving the ability of the liquid to refill into the nucleation sites. The combination of the faster bubble nucleation and immediate liquid replenishment could provide the superior pool boiling performance by enabling rapid bubble dynamics.

Pool boiling experimental setup
Fig. 2 illustrates the pool boiling experimental setup, which consists of a boiling chamber, a heater system, an auxiliary heating system, a condensing system, a visualization system and a data acquisition system. The boiling chamber was made of quartz glass with a thickness of 5 mm and the DI water was used as the working fluid. An auxiliary heater submerged in the DI water was mounted around the PTFE block to maintain the saturation of the DI water in the boiling chamber at 1 atm. The temperature of DI water was measured by the thermocouple T 0 , and the heating power of the auxiliary heater could be controlled by the AC autotransformer. With the spiral tube installing on the top of the boiling chamber, the vapor steam generated from the water boiling can be condensed for maintaining the liquid level of DI water in the boiling chamber. On top of the heating copper block was the 20 mm square boiling surface, and the bottom of the heating copper block was cylinder which held 3 cartridge heaters. The whole heating copper block was assembled into the PTFE to maintain the one-dimensional heat transfer. The heating power of the cartridge heaters was manipulated by the DC power supply which could be regulated by the AC autotransformer. Three thermocouples were placed every 5 mm below the boiling surface, and three K-type thermocouples were embedded into the copper block at the same horizontal plane location to measure the local temperature of the boiling surface. Through the data acquisition system (Keysight, DAQ 970 A) connected with the computer, all the temperature signals of thermocouples were collected. The bubble behaviors including bubble nucleation, bubble growth, bubble merge and bubble departure were captured by the high-speed camera.

Test procedures and data processing
Before the pool boiling test, the DI water was heated to the saturation temperature (T sat = 100 • C) by the auxiliary heater for around 60 min to expel the excess gas. Afterwards, the voltage of the cartridge heaters embedded in the copper block was increased step by step for improving the heat flux to the boiling surfaces. After the stepwise increase in the voltage, the temperature reached a steady state after approximately 15 min. The temperature signal was recorded every 1 min and the temperature variation of each thermocouple was less than 0.5 • C. Temperature signals at three horizontal planes (T 1 , T 2 and T 3 ) were processed by averaging the three groups of thermocouples (i.e., (T 11 + T 21 + T 31 )/3). When the temperature significantly increased compared with the earlier value, the CHF phenomenon could be observed. In this condition, the pool boiling experimental setup should be shut down to avoid the damage to the PTFE block.
By the one-dimensional Fourier's law, the input heat flux q can be calculated as follows: where λ Cu is the thermal conductivity of the copper, dT/dz is the axial temperature gradient of the Cu block calculated by the three thermocouples readings, ∆z is the distance between the adjacent thermocouple locations. The temperature gradient in the copper block was calculated using the three points backward Taylor's series approximation. (2) where T 1 , T 2 and T 3 were the average temperature at three horizontal planes. Therefore, the HTC (h) could be determined as follows: where ∆z 0 is the distance between the bottom of the boiling surface and the thermocouple of T 1 ; T w is the wall superheat of the bottom temperature of the boiling surface; T sat represents the saturation temperature of the DI water.

Uncertainty analysis
The K-type thermocouples were calibrated and absolute uncertainties were ± 0.2 K. The thermal conductivity of copper was estimated as 395 ± 5 W/m K and the uncertainties in thermocouple locations were around 0.2 mm. According to the error-propagation law (Li et al., 2021a), the uncertainties of heat flux, heat transfer coefficient, and superheat were determined using standard procedures as follows: Thus, when the heat flux was about 42.1 W/cm 2 , the corresponding relative uncertainties of the heat flux, heat transfer coefficient, and wall superheat were 7.6%, 11.1% and 8.2%, respectively. And the uncertainty details of the measured and calculated parameters are listed Table 2 below. . It was clear that the dense nanograss have uniformly covered the Cu substrate, approximate to a piece of flourish meadow. From the high magnification SEM images (Fig. 3d), we can find the nanograss has the diameter range from 1 to about 10 µm with a nanoscale or microscale interspace distance, developed randomly from the copper substrate. Fig. 3e and f show the SEM side view images of the nanograss (FE-SEM, Gemini, ZEISS). The height of nanograss grown on the Cu substrate could reach approximately 1 µm. It is noted that the dense nanograss could provide significantly sufficient nucleation sites for pool boiling. Fig. 4a, b, c and d compare the SEM images (× 30 magnifications) of Sample 2 to 5, which provided the width of the nanograss strip is w = 2, 3, 3.5 and 4.5 mm, corresponding to the gaps of 1.6, 1.25, 2 and 3.25 mm between the adjacent nanograss strips, respectively. Fig. 4e, f, g and h compare the surface roughness of Sample 2, 3, 4 and 5, which is featured as 4, 4.594, 4.356 and 4.81 µm using confocal laser scanning microscope (CLSM, VK-250, KEYENCE), respectively. From Fig. 4, the gap between the nanograss strips was sufficiently large to permit flow permeability for efficient rewetting of the nucleation sites to improve the boiling heat transfer.

Surface compositions
The copper substrate can be oxidized to Cu 2+ ions by oxidant of (NH 4 ) 2 S 2 O 8 in alkaline NaOH environment, releasing NH 3 during the process which contributes to the formation of the nanograss morphology (Chen et al., 2015), and the Cu 2+ ions captured the OH − to form Cu(OH) 2 nucleus on the copper substrate. The whole surface modification process can be explained by following equation:   and those marked with the circle is the Cu substrate. X-ray photoelectron spectroscopy (XPS) was conducted to further identify the composition and purity of the Cu(OH) 2 nanograss. The XPS spectra are presented in Fig. 5b, c and d, which is corrected for specimen charging by referencing the C 1s to 284.6 eV (Fig. 5b). As seen in Fig. 5c, two distinct peaks at 935.1 eV and 954.5 eV can be assigned to binding energy of Cu 2p 3/2 and Cu 2p 1/2 , respectively, indicating the presence of the Cu 2+ on the sample. These two peaks together with their shake-up satellites 961.5 eV, 942.8 eV and 941.1 eV are characteristic of d 9 Cu (II) compounds. Fig. 5d demonstrates the spectrum of O 1s, which can be fitted to two peaks with binding energies at 530.5 eV and 531.5 eV that contribute to the oxygen in Cu(OH) 2 . In terms of the above, the nanograss stripes are examined as chemicals composed by Cu(OH) 2 .

Verification of test facility
Experiments were firstly conducted based on the smooth copper surface to verify the experimental system by being compared with the correlation and literatures (Jo et al., 2018b;Yao et al., 2011;Han Kim et al., 2015;Rohsenow, 2022;Sarangi et al., 2015;Amiri et al., 2014;Zuber, 2022). Meanwhile, the fundamental results obtained on the smooth surface can be used as the benchmark to those on structural surfaces, showing the significant improvement of the heat transfer performance due to the in-situ growth of the Cu(OH) 2 nanograss. Two classic correlations in regards to water pool boiling heat transfer on a smooth surface are applied to verify the reliability of the experimental facility. The experimental data in Fig. 6 presents a good agreement with the predicted curve from the classic Rohsenow's correlation (Liang et al., 2020), which is given below where C sf is a coefficient depending on the boiling surface/liquid combination, which is 0.0152 for DI water on copper surface, and n = 1 for water. From the figure it is seen that the overall trend of Fig. 6. Boiling curves of the current experiment and references (smooth copper surface) (Jo et al., 2018b;Yao et al., 2011;Han Kim et al., 2015;Rohsenow, 2022;Sarangi et al., 2015;Amiri et al., 2014;Zuber, 2022).
the measured heat flux is in consistence to the results predicted by the Rohsenow's correlation. In addition to the Rohsenow's correlation, the Zuber's correlation (Zuber, 2022) applies and is given as follows: Fig. 6 compares the CHF values on the smooth surfaces from the literatures, ranging from 10 to 100 W/cm 2 . The large deviation between the experimental data have also been reported elsewhere on smooth surfaces because of the difference in the surface area, the edge effect, the surface materials, the surface shapes and the surface roughness. Smaller substrates exhibit the higher CHF than large ones. The Zuber correlation predicted the CHF on a smooth surface as 100 W/cm 2 , which has a large deviation from the actual experimental values summarized in Fig. 6. The CHF by the current experiment is measured as approximately 35.8 W/cm 2 , slightly lower than some comparable values due to the larger copper substrate used in this study (20 mm × 20 mm). Therefore, the data measured by this experimental device is reliable.

Pool boiling heat transfer performance
Fig. 7a presents the boiling curves of different samples, in which the nanograss stripes' width w is 0, 2, 3, 3.5, 4.5 and 20 mm, respectively, the depth is around 1 µm and spacing of the adjacent nanograss stripes d is 20, 1.6, 1.25, 2, 3.25 and 0 mm, respectively. The S1 with w = 0 mm included in this figure is the boiling curve on the smooth surface for comparison. It is found that boiling curves of the nanograss surfaces have the higher CHF and are shift leftward compared with that of the smooth surface, indicating that the nucleating pool boiling heat transfer coefficient can be enhanced. According to the experiments, the CHF is improved to 70.80 W/cm 2 , indicating a significant improvement of 97.8% compared to the smooth surface. Fig. 7b plots heat transfer coefficient versus heat flux on different samples, in the same conditions to Fig. 7a. It is found that the pool boiling heat transfer coefficient of the nanograss surfaces are much higher than that of the smooth surface. The highest HTC was measured as 2.4 W/cm 2 K, which was improved by 26.3% compared to the smooth surface. In all nanograss surfaces, the incipient wall superheat (as indicated in Table 3) significantly reduced from around 9.7 K to less than 7.7 K. With the different width of nanograss stripes, it is observed that the pool boiling heat transfer performance can be enhanced in different degrees.
Although the CHF depends on the w in Fig. 1, it is not proportional to w, indicating that the width of nanograss stripes (w) and spacing of the adjacent nanograss stripes (d) act synergistically to produce the highest CHF on the S4 sample. When the entire surface of the sample is grown with the Cu(OH) 2 nanograss, as shown in S6, the CHF is higher than the smooth surface, but the HTC is insignificantly improved compared to the smooth surface. This indicates that the growth of the Cu(OH) 2 nanograss can enhance the CHF, but has inhibited the further improvement of the HTC when the nanograss completely covered the copper substrate. Fig. 7c plots the HTC values as a function of wall superheat of the whole samples, in which the S4 exhibits the highest HTC. As can be seen, due to an early ONB, an enhanced HTC can be achieved at a lower wall superheat. Nevertheless, the HTC is slightly reduced as the wall superheat increases after the HTC reaches the highest value compared with the smooth surface. Fig. 7d compares the CHF and the corresponding effective HTC h eff values. In general, a high h eff corresponds to a high CHF because h eff represents the slope of the curve q ∼h eff · ∆T CHF . In this study, the variations in the ∆T values are not significant (20 K ≤ ∆T ≤ 33 K). Therefore, the CHF is positively correlated to the h eff . It should be noted that the whole nanograss stripe coated surfaces can enhance the CHF and h eff , and the S4 exhibits the highest h eff corresponding to the highest CHF. These indicate that the S4 which provides the 3.5 mm width of Cu(OH) 2 nanograss stripes and 2 mm spacing between adjacent nanograss stripes is the optimal dimensions for simultaneous enhancement of the CHF and h eff . Table 4 compares ∆T CHF , CHF, HTC max , CHF smooth and the other pool boiling conditions that were reported previously. The boiling heat transfer of various structured surfaces is related to various parameters and the characteristics of pool boiling setups. For instance, the saturation of pool boiling working fluid affects the pool boiling performance. Moreover, the geometric dimensions and surface materials, as well as the surface morphology also play an important role in the pool boiling heat transfer performance. According to Table 4, it is worth mention that the CHF is improved by 97.8% compared to the smooth surface and in the meanwhile the HTC is enhanced to 2.4 W/cm 2 K, which exhibits an excellent pool boiling performance compared with previous studies.

Bubble nucleation
According to Table 2, the nanograss stripes have increased boiling heat transfer performance, in terms of CHF and ONB. For ONB, the nanograss stripes on the Cu substrate (S2-S5) can significantly reduce the ONB to 1.8, 1.4, 1.2 and 2.5 K, respectively. Nevertheless, the Cu substrate which is entirely covered nanograss (S6) exhibits the slighter lower ONB compared with  the smooth surface (S1). Fig. 8 shows the photographs of the ONB phenomenon on the various samples. For the smooth surface (S1) and the fully covered with the nanograss, the bubbles nucleate randomly on the surfaces as illustrated in Fig. 8a and 8f. It is because that these surfaces are overall flat and the contact line between the liquid and the surface is horizontally uniform. The bubbles thereby generate arbitrarily due to the random active nucleation sites on the Cu substrate. In comparisons, it is observed that the isolated bubble is located on the nanograss stripe of Cu substrate at the early boiling stage. Moreover, the bubbles prefer to nucleate and grow at the boundaries of the nanograss stripes, as depicted in Fig. 8b, c, d and e. The boundaries of the nanograss stripes are more suitable for the bubble nucleation. In addition, the active nucleation sites can significantly increase on the S3 and S4 surface because of the larger area of nanograss stripes, resulting in the increase in the number of the bubbles. The more bubbles depart from the surface, the more heat can be carried out, leading to the enhanced HTC at the low wall superheat. The nanograss stripes on the Cu substrate can facilitate the enhanced CHF and HTC. Nevertheless, with further changing of the width and spacing of nanograss stripes, the nanograss stripes on the Cu substrate become closer and closer, resulting in the grown bubbles that are too close to depart independently due to further coalescence before departure. Therefore, it is found that there exists as optimal stripe design with optimized width and spacing for the boiling heat transfer performance.
As reported, the microcavity become active if its feature size is larger than the critical nucleation radius (r * ) given as below (Wen et al., 2017): For water at atmospheric pressure, the critical bubble radius varies from 32 to 1.6 µm for the wall superheat ranging from 1 to 20 K. It is noted that the microcavities (1-10 µm) of the nanograss stripes (Fig. 3) can be act as the effective active nucleation sites. Fig. 9 shows the nucleation sites density of bubble versus wall superheat using different samples. The active nucleation site density (N a ) is defined as the average count of boiling sites on unit surface area, that is N a = n/A, where A is the surface area of the total field (20 × 20 mm), and n is the count of nucleation bubble within the field. As illustrated, in the condition of the same heat flux, the bubble nucleation site density of the S4 is the largest among the whole samples. Fig. 10 show the bubble characteristics on each sample, represented by the relation of the bubble diameter and the frequency with the heat flux. It is clear that the bubble departure diameter decreases, and the bubble departure frequency increases, which is beneficial to increase the CHF because of delay of the  bubble coalescence resulting from the decrease in bubble departure diameter and the increase in bubble departure frequency. As sown in Fig. 10a, S3 and S4 has a role in reducing the bubble departure diameter and increasing bubble departure frequency significantly. On the S2, S5 and S6 surfaces, compared with the smooth surface S1, the CHF on these surfaces were also enhanced. This is because that the hemi-spreading phenomenon reported in previous works facilitates the water replenishment to reach the dry spots and improves the CHF compared with the smooth surface. Furthermore, nanograss has a positive effect on single bubble characteristics versus the smooth surface. Nanograss has the native features of 1-10 µm spacing which can act as the active nucleation sites for easier bubble generation compared with the smooth surface. Moreover, the optimized width and spacing of nanograss stripes can further decrease 2/3 of the bubble departure diameter and increase 1.5-fold higher bubble departure frequency compared with the smooth surface. As aforementioned, the bubble departure diameter and frequency are related with the delay of the formation of the vapor film and the water supply to the hot spots. Therefore, it is concluded that the enhanced boiling heat transfer performance on the S2, S3, S4, S5 and S6 surfaces is due to the positive effect of the hemi-spreading and single bubble features. In addition, the nanograss stripes have the smaller bubble diameter and higher bubble departure frequency than the entire nanograss covered surface S6, which indicates the nanograss stripes can enhance the pool boiling performance through changing the width and spacing of the nanograss stripes.

Bubble departure characteristics
Nevertheless, because the single bubble features on the S2, S3, S4 and S5 are almost the same, and the effect of width and spacing of nanograss stripes on the boiling enhancement is also unclear. Therefore, the bubble merging that causes vapor film on the surfaces is considered in the analysis of the thermal enhancement.

Bubble merging characteristics
As known, CHF phenomenon occurs when the bubble merge into the vapor blanket to prevent the water replenishment for cooling the hot spots. Thus, an evaluation of the bubble merging is important to the CHF enhancement. As aforementioned, the single bubble departure diameter and frequency without bubble merging were investigated. It is noted that a new concept of bubble merging diameter on the nanograss stripes surfaces (S2,  S3, S4 and S5) is defined for explaining the boiling enhancement. Fig. 11 shows the bubble coalescence phenomenon on the S2 surface with time. When the bubbles nucleate on the S2 nanograss stripes' boundaries, because the bubble merging occurs in the lateral direction, small bubbles depart from the surface with a larger merged bubble. Based on these results, the size of the merged bubbles was further measured for evaluating the merged bubble departure diameter distributions.
As shown in Fig. 12, merged bubble sizes were measured at three different heat flux (23, 30 and 42 W/cm 2 ). The whole nanograss stripes surfaces illustrated a Gaussian distribution in the below graphs, and the Gaussian function can be listed as follows (Lee et al., 2018): where y 0,cv , A cv , x c,cv and w cv represents the off-set, the area under the curve, the center of the curve and the width of the curve, respectively. As illustrated, the wide shape of Gauss fit indicates the variation of bubble size is large and the sharp shape of Gauss fit means the variation of bubble size is small. As comparisons, the S3 and S4 have the narrowest shape of the Gauss fit curves at three different heat fluxes, indicating that the S3 and S4 have the smaller variation of merged bubble diameter which exhibits bubble merging phenomenon occurrence on the S3 and S4 less than the other surfaces (S2 and S5). Thereby, it is predicted that the bubble merging can be inhibited by the artificial bubble nucleation sites -nanograss stripes. By comparing the S3 and S4 surface, the peak value of the merged bubble diameters on S4 at three different heat fluxes are smaller than on the S3. This means that well-arranged bubbles are generated on the S4 through controlling the width and spacing of the nanograss stripes to limit the bubble generation on the stripes. As a consequence, wellarranged smaller bubbles are generated on the nanograss stripes with optimized width and spacing, and bubble merging can be delayed. Fig. 13 shows the average merged bubble departure diameter and the standard deviation of merged bubble departure diameter maintained the minimized at the three different heat fluxes (23, 30 and 42 W/cm 2 ), compared with the other surfaces. On the country, the standard deviation of the S2, S3 and S5 increases as the heat flux increases, while the S4 has the smaller and more stable standard deviation at the high heat flux. This means that bubble merging phenomenon can be minimized on the S4 surface, due to its better arrangement of bubble nucleation sites through controlling the width and spacing of the nanograss stripes. Thus, the boiling heat transfer performance can be maximized by delaying the bubble merging and supplying the liquid to rewet the hot nucleation sites for inhibiting the formation of vapor blanket.

Thermal stability and uniformity
The thermal stability and uniformity on the boiling surfaces play a critical role in the persistent boiling applications, because the dry-out space resulting from the local ultra-high temperature can be avoided through maintaining the overall uniformity of the surface temperature. As above-mentioned, the nanograss stripes can better arrange the bubble nucleation sites for bubble  generation to avoid the bubble merging phenomenon. Here, the standard deviation of the wall temperature was evaluated on the S1, S2, S3, S4, S5 and S6 with time at the location 2. As shown in Fig. 14a, it is evident that the temperature variation of the smooth surface S1 is larger than the other surfaces and the standard deviation on S1 surface is 1.83 K at the CHF. This means that the large overall temperature variation cannot be avoided because of the sharp increase of the local temperature when the CHF phenomenon occurs. With covering the nanograss, the temperature variation can be decreased. As illustrated in Fig. 14f, the standard deviation of the temperature at the CHF decreases to 0.82 K on the S6, which is more than 50% reduction compared with the S1. This indicates that the nanograss grown on the copper substrate can increase the overall temperature uniformity.
Furthermore, by comparing the S2, S3, S4 and S5 (Fig. 14b, c,  d and e), different width and spacing on the nanograss stripes can lead to the different standard deviations. Here noted that the S3 has the lowest standard deviation of 0.45 K at CHF and the S4 has the lowest standard deviations of 0.04 K and 0.09 K at the low heat flux before the CHF. The reason for this is that smaller and better-arranged bubbles can be generated on the nanograss stripes with optimized width, and the spacing between the adjacent nanograss stripes can act as the liquid supply region, leading to the bubble merging delay for the boiling heat transfer enhancement. Thus, the CHF phenomenon can be delayed to avoid the local temperature increasing sharply and maintain the overall temperature uniformity.

Boiling enhancement mechanism on the HNS
The nucleated bubbles become superheated because of the phase change of the DI water when maintaining the saturate temperature (100 • C) at the atmospheric pressure. As shown in Fig. 15a, the superheated bubbles with a low density will rise up from the nucleation sites with a large bubble size and a slow detachment. In comparisons, the entirely covered nanograss can provide the microdefects which can act as the random nucleation sites for bubble generation (Fig. 15b). This means that the nanograss covered surface can obtain the earlier boiling inception than on the smooth surface. Moreover, the water outside the nucleation sites which maintains the lower temperature induces more effective cooling for the hot nucleated spots via the strengthened capillary pumping effect. The random nucleation with smaller size and higher departure frequency can be obtained on the entirely covered nanograss, which can provide the more effective cooling capability. Nevertheless, the superheated bubbles generating from the random nucleation sites will merge into the large bubbles, ultimately forming the vapor blanket which can obstruct the rewetting of water.
On the basis of the entirely-covered nanograss, the hierarchical nanograss stripes surface (HNS) with well-arranged nucleation sites region and liquid replenishment region was fabricated for cooling the fluid. As shown in Fig. 15c, the superheated bubbles nucleate on the boundaries or centers of the well-arranged nanograss stripes. With the existence of the spacing between the adjacent nanograss stripes, bubble merging can be delayed significantly. The rewetting water driven by the nanograss can flow through the narrow spacing between the adjacent nanograss stripes, forming an artificial division for preventing bubble merging. Due to the separation of bubble nucleation region and liquid replenishment region, smaller and faster nucleated bubbles can generate for carrying out more heat, and the bubble merging can be delayed for inhibiting the formation of vapor blanket. Furthermore, the small, fast and well-arranged bubbles generated on the HNS can also provide the temporal contact between the bubbles and the boiling surface for maintaining the long-time high-efficiency heat transfer performance as evidenced by the low standard deviations of temperature variation.
However, a HNS samples (S2, S3, S4 and S5) exhibit the different boiling performance due to the different spacing and width of the nanograss stripes. As reported, the vapor column spacing has been found to be determined using the capillary length λ c , which is calculated as follows (Liang et al., 2020): which is 2.5 mm for saturated water at atmospheric pressure. By comparing the S2 and S3, the independent bubble departure size is significantly smaller than that of the S2, and the independent bubble departure frequency is higher than that of the S2. These mean the S3 provide the smaller and faster bubbles for carrying out heat from the boiling surface compared with the S2 surface.
On the S4 surface, which can provide an optimal width of d = 2 mm (closest to the capillary length 2.5 mm), the frequency of bubble coalescence can be further reduced, and smaller bubbles can be successfully detached from the nanograss stripes, with no time for bubble merging. However, the S3 has more nucleation sites due to the larger area of nanograss stripes than that of the  S4. As a result, numerous independent bubbles can be generated on the nanograss stripes comparing with the S4 at the low heat flux. Nevertheless, the narrower spacing facilitates the numerous bubbles to merge into the large bubbles at the high heat flux, thereby the CHF of the S3 is lower than the S4. A decrease in CHF can be observed for the increase in the spacing beyond S4, the number of active nucleation sites is reduced, resulting in the less bubbles generated on the boiling surface. This result indicates that the boiling enhancement provided by the nanograss stripes diminishes when the spacing is larger than the capillary length. Thus, with the spacing of 2 mm, the alignment of the vapor bubble nucleation region and liquid supply region, enables the escape of vapor bubbles at the designed nucleation sites, which delays the bubble merging and the formation of larger vapor column (see Fig. 16).

Conclusions
In this study, well-arranged nanograss stripes surface were insitu grown for improving the CHF and HTC simultaneously. The main conclusions based on experiments are as follows: (1) In-situ hierarchical nanograss stripes (HNS) with different width and spacing for pool boiling enhancement is fabricated by the immersion method. The nanograss slantly from the substrate possess a height of around 1 µm, a diameter ranging from 1 to 10 µm and an interspace distance in micro/nanoscale., The composition of nanograss stripes is identified as Cu(OH) 2 by the XRD and XRS analysis.
(2) These dense HNS can act as additional nucleation sites and provide much improved capillary pumping, which is beneficial to the enhancement of the pool boiling heat transfer performance. The boundaries of the nanograss stripes are more suitable as the active nucleation sites. Moreover, significant increase in active nucleation sites can be observed on the S3 and S4 surfaces due to the larger area of the nanograss stripes. (3) Experiments show that the optimized width of 3.5 mm and spacing of 2 mm on S4 results in the largest improvements of CHF and HTC. This is because the bubble behaviors including the observed independent bubble departure size can be minimized, the bubble departure frequency can be maximized, and the bubble mergence is significantly delayed on the S4 sample. (4) Temporal uniformity of the boiling surfaces can be prominently improved by the HNS. This indicates that sharp increase in local temperature can be avoided for the longtime boiling applications. It is explained that the separation of the bubble nucleation region and liquid supply region efficiently promote the bubble nucleation and prevent the formation of vapor blanket.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request