主序後星在宇宙塵埃的生命週期中扮演關鍵角色。由於主序後星數量眾多， 因此能有 效率的透過強恆星風和超新星噴發其核心中合成的原料，並以此提供星際介質中的重 元素。在這篇論文中，我研究了主序後星的質量流失，深入研究並探索它們過去一生 中的塵埃質量流失事件。主序後星的典型研究方法是僅考慮當前的質量流失，或假設 它們在其生命週期中經歷近似穩定的質量流失。塵埃從恆星中心離開時會冷卻，可以 利用塵埃的熱輻射有效地追其質量流失歷程。本論文使用Herschel / PACS遠紅外望遠 鏡（波長70微米、100微米和160微米）和JCMT / SCUBA-2次毫米波望遠鏡（450微米 和850微米）來觀測。這兩組觀測結果非常強大，可追主序後星周圍包層的徑向變化。 本論文透過觀察研究表面亮度、溫度、塵埃質量- 柱密度和徑向的塵埃輻射率光譜指 數中的特徵，確定了銀河系主序後星取樣的質量流失變化。將這些結果與恆定流出模 型的預測進行比較，本論文發現此兩種情景決定的總塵埃質量間有顯著偏差。因此進 一步顯示了質量流失變化的影響不容忽視。本論文利用輻射轉移模型確定了唧筒座U星 的分離殼層複雜性，並指出進一步研究的需要。兩個著名主序後星IRC + 10216和鯨魚 座o 星的次毫米波週期的分析顯示，可能可使用主序後星的次毫米波光變曲線來有效研 究它們包層內部的塵埃形成和分解性質。IRC + 10216顯示可見光和次毫米觀測的偏移 相位約為3/4。使用模型預測和過去的報告，顯示恆星脈衝和塵埃形成/分解之間的關 係。這篇論文包含了位於銀河系及麥哲倫雲星系，在中紅外波段觀測到但遠紅外波段 沒有觀測到的主序後星。由Herschel HERITAGE普查計劃，波長為100微米的調查，可 以對這些主序後星中冷塵埃的遠紅外波段成份存在與否進行統計研究。本論文成功提 出由最高質量流失率來源組成的三疊檢測，展示了堆疊法的優點。本論文展示結構的 變化率及質量流失歷程，並強調這些在銀河系及之外的天體，分析研究其塵埃產生的 重要性。;Evolved stars play a key role in the life cycle of dust in the universe. Through strong winds and supernovae they inject the material reprocessed in their cores to the interstellar medium, replenishing the interstellar medium with heavy elements. As evolved stars are so numerous they are extremely e ective in this role. In this thesis we study this mass loss delving deep into their past, exploring dust mass-loss events which occurred throughout their lifetime as evolved stars. The typical treatment of evolved stars has been to only account for present day mass loss or assume they are quasi-stable undergoing constant mass loss throughout their lifetime. Historic mass loss can be e ectively traced using thermal emission from the dust which cools down as it moves away from the central star. We exploit Herschel/PACS far-infrared (70 m; 100 m and 160 m) and JCMT/SCUBA- 2 sub-millimetre (450 m and 850 m) observations to study this historic mass loss. These two sets of observations are especially powerful, tracing the radial variation in the circumstellar envelopes of evolved stars. We establish the presence of variations in mass loss for a sample of Milky Way evolved stars by studying features observed in surface-brightness, temperature, dust mass-column density and spectral index of dust emissivity radial pro les. By comparing these results to predictions for a constant-out ow model we nd signi cant deviations between the total dust masses determined for the two scenarios, demonstrating that the e ect of mass-loss variations cannot be ignored. We determine the complexities of the detached shell of U Ant with the aid of radiative transfer modelling, prompting the need for further study at higher angular resolution. The analysis of the sub-mm periodicity of IRC+10216 and o Ceti, two well-known evolved stars, shows that it may be possible to use sub-mm light curves of evolved stars to study the nature of dust formation/destruction in their inner envelope e ectively. The optical and sub-mm light curves of IRC+10216 shows an o set in phase of 3=4. Using radiative transfer modelling as well as ndings in literature we attribute the sub-mm variability and phase lag to both the relationship between stellar pulse and dust formation/destruction cycle as well as a second unknown mechanism which needs to be narrowed down in the future. Moving on from the Milky Way to the Magellanic clouds we combine cutouts of evolved stars identi ed in the mid-infrared but not detected in the far-IR. The cutouts are generated from the Herschel HERITAGE survey at 100 m allowing a statistical study of the existence or absence of a far-IR component as a result of cold dust in these evolved stars. We successfully produce detections for three stacks comprised of the highest massloss rate sources, showcasing the merits of the stacking method. With this thesis we emphasise the variations in structure and hence historic mass loss of evolved stars. By doing so we highlight the importance of including these variations in studies which include dust production by evolved stars in the future.
