常壓平行板型氮與氨混合氣體介電質電漿的數值模擬研究

Numerical Investigation of a Parallel-Plate Atmospheric-Pressure Nitrogen/Ammonia Dielectric Barrier Discharge

李富利(Fu-Li Li)

指導教授 ： 吳宗信

國立交通大學/工學院/機械工程系所/博士(2013年)

https://doi.org/10.6842/NCTU.2013.00733

摘要

在本論文中，利用有限體積法解一維電漿流體模型程式，並進行模擬氮氣混合0-2% 氨氣以及其含微量氧氣的常壓平板型介電質電漿。模擬中使用30 kHz頻率的脈衝式或正弦波電源驅動常壓介電質電漿。本論文研究主要可分成四部份，以下將依序簡介每一部份：
在第一部份，模擬常壓平板型氮與氨混合氣體介電質電漿。建構「氮/氨電漿化學模組」包含23種粒子和141條氣相化學反應式模擬電漿。模擬的電流密度於相位和大小與實驗量測的數據有相當的一致性。模擬的結果顯示氮氣混合0-2%氨氣電漿均為典型的類湯森電漿（Townsend-like discharge）。因為離子的數量超過電子的數量太多，致使在遍及整個週期中電漿空間沒有電中性區域。我們發現於電漿崩潰（breakdown）發生過程和之後N2+和N4+等離子含量最多，類似純氮氣介電質電漿，其中NH4+離子隨添加氨氣濃度增高而快速增加。增加氨氣濃度至氮氣電漿中，胺基（NH2）和H原子隨之增加。另外，除背景氣體氮和氨等以外，中性氣體以N原子濃度最高。胺基在很多需要官能基接植（functional group incorporation）應用裡扮演著相當重要的角色。
在第二部份，針對常壓平板型氮與氨混合氣體介電質電漿進行廣泛的參數研究，包含交流正弦波電壓源不同的電壓峰值（6-10 kV）、不同頻率（10-100 kHz）、不同介電質材料（石英和陶瓷）、不同介電質厚度（0.5-2 mm）以及不同電漿空間距離（0.1-1.2 mm）等。研究結果顯示電漿的電流密度和電漿密度隨下列情形而增加：電壓增加、頻率增加、在相同介電質厚度條件下，介電質常數增加、在相同介電質常數條件下，介電質厚度變薄以及電漿空間距離增加。在所有模擬結果中，發現電漿均為典型的類湯森電漿。N2+、N4+以及NH4+都是最重要的帶電粒子，N、N2*、NH2以及H都是最重要的中性粒子。
在第三部份，以數值與實驗方式對常壓氮氨介電質電漿在考慮微量氧氣雜質進行研究它的光源產生機制：包含NO-γ、NO-β以及N2-SPS（second postive system）等。建構完成的「氮/氧/氨電漿化學模組」包含48種粒子和235條氣相化學反應式模擬電漿。模擬結果趨勢與使用放射光譜（optical emission spectroscopy）量測電漿產生的光強度相當符合。研究結果發現NO-γ隨著氨氣含量增加而滅少，主要因為N2(A3Σu+)及NO(A)被氨氣消耗。NO-β隨著氨氣含量增加而滅少，主要因為氨氣與N原子和O原子進行化學反應而耗損。N2-SPS隨著氨氣含量增加而滅少，主要因為氨氣具高度電子親和力以及介穩態氮氣聯合離子化(associate ionization)效應變弱使得電子生成速率減弱。同時因為在電漿崩潰區間內離子濃度遠超過電子濃度及電場空間分佈僅受帶電粒子存在輕微的影響，添加0-1%氨氣的氮氣（含微量氧氣雜質）的電漿仍視為典型的類湯森電漿。
在第四部份，發展簡化電漿化學反應機制方法，並對前述完整「氮/氧/氨電漿化學模組」包含48種粒子和235條化學反應式進行簡化。簡化電漿化學模組結果包含33種粒子和87條化學反應式。在氮氣含微量氧氣條件下添加氨氣濃度從0%至1%，使用完整和簡化的電漿化學模擬氮/氧/氨介電電漿結果比較，結果顯示33種粒子密度和電漿的電流密度均具良好一致性。使用簡化的電漿化學模組模擬電漿的計算時間有效縮短2.1倍，且具有相同電性以及33種粒子密度之均方根誤差都小於1.8%。
最後，將本論文研究的主要發現做為總結，並條列出對未來研究的建議方向。

關鍵字

電漿 ； 常壓介電質電漿 ； 類湯森電漿 ； 流體模型 ； 有限體積法 ； 氮/氧/氨電漿

並列摘要

In this thesis, the planar atmospheric-pressure dielectric barrier discharges (AP-DBD) of nitrogen mixed with ammonia (0-2%) considering with and without oxygen impurity were simulated using one-dimensional self-consistent fluid modeling by applying the cell-centered finite-volume method. These AP-DBDs were driven by a 30 kHz power source with distorted sinusoidal or purely sinusoidal voltages. Researches in this thesis are divided into four major parts and are described in the following in turn.
In the first part, the N2/NH3 AP-DBDs were numerically investigated using a 1-D fluid modeling with 23 species and 141 reaction channels. The simulated discharge current densities are found to be in good agreement with the experimental data in both phase and magnitude. The simulated results show that the discharges of N2 mixed with NH3 (0-2%) are all typical Townsend-like discharges because the ions always outnumber the electrons greatly which leads to no quasi-neutral region in the gap throughout the cycle. N2+ and N4+ are found to be the most abundant charged species during and after the breakdown process, respectively, like a pure nitrogen DBD. NH4+ increases rapidly with increasing addition of NH3 initially and levels off with further increase of NH3. In addition, N is the most dominant neutral species, except the background species, N2 and NH3. NH2 and H are the second dominant species, which increase with the increasing addition of NH3. The existence of abundant NH2 plays an important role in those applications which require functional group incorporation.
In the second part, an extensive parametric study of fluid modeling of N2/NH3 AP-DBD by varying the voltage amplitude of AC power source with sinusoidal voltages (6-10 kV), voltage frequency of AC power source with sinusoidal voltages (10-100 kHz), dielectric materials (quartz and ceramic), dielectric thickness (0.5-2 mm), and gap distance (0.1-1.2 mm), has been investigated. The results show that the discharge current density and species densities increase with 1) increasing amplitude of applied voltage, 2) increasing frequency of applied voltage, 3) increasing dielectric constant at the same thickness, 4) decreasing dielectric thickness at the same dielectric constant, and 5) increasing gap distance. For all these cases, we have found that they are typical Townsend-like discharges in an average sense in which N2+, N4+ and NH4+ are found to be the most dominant ions species, and N, N2*, NH2, and H are the most dominant neutral species.
In the third part, the mechanisms of the light emissions, including NO-γ, NO-β and N2-SPS (second postive system), produced in a N2/NH3 AP-DBD considering realistic oxygen impurity was investigated numerically and experimentally. Self-consistent, one-dimensional fluid modeling was used to the numerically simulate the discharge process with 48 species and 235 reaction channels. An optical emission spectroscopy (OES) was used to measure the relative intensities of the light emission. The simulations of the light emission intensities for the above-mentioned OES lines generally reproduced the trends observed in the experiments caused by changes in the NH3 concentration. All of the predicted intensities of NO-γ, NO-β and N2-SPS decreased with increasing amount of NH3 caused by various reaction mechanisms. The former is due to the loss of N2(A) and NO(A) by the reaction of NH3 with N2(A) and NO(A) respectively. The decrease of NO-β is due to the depletion of N and O because of NH3, and the decrease of N2-SPS is due to electron attachment to NH3 and a weaker metastable-metastable associative ionization of N2. All of the simulated results demonstrate that the discharges are typically Townsend-like because ions outnumber electrons and the electric field across the gap is distorted only slightly by the charged particles during the breakdown.
In the final part, a reduced chemical kinetics model for a planar atmospheric-pressure N2/O2/NH3 dielectric barrier discharge is proposed and validated by benchmarking against a more complete version as shown in the third part. The set of complete chemistry, including 48 species and 235 reactions, has been successfully reduced to that consisting of 33 species and 87 reactions with a very limited loss of accuracy. The results show that the computational time for 1-D fluid modeling using the reduced chemistry is 2.1 times faster with essentially the same electrical properties and produces less than 1.8% of the root mean squared errors of major species compared to using the complete chemistry, when the oxygen (impurity) is taken to be 30 ppm and the ammonia varies in the range of 0-1%. The current density produced by the reduced chemistry is also found to be in excellent agreement with the current density produced by the complete chemistry.
Finally, major findings and recommendations for future study are summarized and outlined at the end of the thesis.

並列關鍵字

plasma ； atmospheric-pressure dielectric barrier discharge ； Townsend-like discharge ； fluid modeling ； finite-volume method ； nitrogen/oxygen/ammonia discharge

參考文獻

[1] F. Barbir and T. Gomez, Int. J. Hydrogen Energy 22, 1027 (1997)

[2] J. Klemberg-Sapieha, O. Kuttel, L. Martinu, and M. Wertheimer, J. Vac. Sci. Technol. A 9, 2975 (1991)

[5] P.R. McCurdy, C.I. Butoi, K.L. Williams, and E.R. Fisher, J. Phys. Chem. B 103, 6919 (1999)

[7] Y. Choi, J. Kim, and Y. Hwang, Thin solid films 506, 389 (2006)

[10] F. Tochikubo, T. Chiba, and T. Watanabe, Jpn. J. Appl. Phys. 38, 5244 (1999)