This thesis aims to discuss the origin of extremely metal-poor (EMP) stars, especially whether it is normal supernovae (SNe) with Eexp ~ 1 × 10(51) ergs or hypernovae (HNe) with E(exp) ≳ 10 × 10(51) ergs from comparison of their observations and theoretical supernova (SN) yields.

EMP stars with [Fe/H] ≲ − 3 may have been formed just a few generations after the first-generation Population (Pop) III stars, or they may even represent the second generation. Their abundance patterns may be the result of nucleosynthesis in even one single core-collapse SN. In other words, SNe of Pop III or almost metal-free stars should reproduce the abundance patterns of EMP stars.

The observed abundances of metal-poor halo stars show quite interesting patterns. There are significant differences between the abundance patterns in the iron-peak elements below and above [Fe/H] ~ − 2.5. For [Fe/H] ≲ − 2.5, the mean values of [Cr/Fe] and [Mn/Fe] decrease toward lower metallicity, while [Co/Fe] and [Zn/Fe] increase (McWilliam et al. 1995; Cayrel et al. 2004).

Umeda & Nomoto (2002) shows that the trends of [Zn,Co/Fe] and [Cr,Mn/Fe] can be related to the variations of explosion energy of core-collapse SNe, i.e., high [Zn,Co/Fe] and low [Cr,Mn/Fe] can be explained by HNe while low [Zn,Co/Fe] and high [Cr,Mn/Fe] by normal SNe. This is because of the larger mass ratio between the complete Si-burning region and the incomplete Si-burning region in the HN model than in the normal SN model.

Although HNe are considered to be the possible origin of EMP stars from their result, there remain some problems to be solved in order to conclude that the origin of EMP stars is HNe.

First, we should modify Ye and entropy to some extent in order to reproduce Co, Mn and Sc. As for K, we cannot reproduce EMP stars even with these modifications. Second, we do not consider the contribution from the innermost matter to SN yields. Third, SN yields in our group do not include neutrino processes. Fourth, there is a possibility of uncertainty in SN yields from a different set of nuclear reaction rates.

In this thesis, I approach the remaining problems with a different approach in each chapter. From all the results in this thesis, I make a conclusion about which is the more possible origin of EMP stars, normal SNe or HNe.

(Chapter 3)

As for Pop III and almost metal-free SNe, Umeda & Nomoto (2002) and Heger & Woosley (2010) reach the contrary conclusion even though their method and approach are similar. Umeda & Nomoto (2002) suggests HNe, while Heger & Woosley (2010) suggests normal SNe as Pop III and almost metal-free SNe.

These previous works do not include any contribution from hot-bubbles in the innermost region of SNe. However, recent multi-dimensional simulations have shown that both the neutron-rich and proton-rich matters are ejected from the hot-bubble regions (e.g., Janka et al. 2003). This innermost matter with various Ye and entropy is considered to be the origin of the heavier elements than Zn, but also can be the important site for the lighter elements (e.g., Hoffman et al. 1996).

In Chapter 3, I perform hot-bubble calculations using one-dimensional trajectory mimicking multi-dimensional simulations, and discuss whether normal SNe with hot-bubbles can reproduce EMP stars especially paying attention to Fe-peak elements.

As a result, I find that neutron-rich (Ye = 0.45-0.49) and proton-rich (Ye = 0.51-0.55) matter can increase Zn/Fe and Co/Fe ratios as observed, but tend to overproduce the other Fe-peak elements such as Ni. Among hot-bubble matter with various parameters, neutron-rich and low-entropy of s/kb ~ 5 matter is the best to reproduce Co, but at the same time, overproduces Ni.

There has been no work that cuts to the heart of the matter, and it has remained to be mystery whether Zn and Co in EMP stars reflect HNe or normal SNe with hot-bubbles. From my work, it becomes clear for the first time that HNe are the only origin of Zn and Co in EMP stars, and that there is no problem to use these elements as a barometer of explosion energy. This is an important result for unclosing the SN mechanism.

(Chapter 4)

I also discuss weak r-process elements (Sr, Y and Zr) in Chapter 4. I investigate normal SN and HN models with neutron-rich matter ejection from the inner region of the conventional mass-cut. I find that explosive nucleosynthesis in a HN can reproduce the high abundances of Sr, Y, and Zr, but that the enhancements of Sr, Y, and Zr are not archived by nucleosynthesis in a normal SN. I also find that it is dependent on entropy whether elements from Ga to Rb are produced with those from Sr to Zr . It will be possible to determine of what entropy matter contributes to weak r-process stars from the observational abundances from Ga to Rb.

(Chapter 5)

In Chapter 5, I calculate normal SN and HN yields applying the latest progenitor models and neutrino reaction rates, and compare these yields with EMP stars. I also investigate production process of each nuclide in detail.

I find that !-processes can enhance the Mn abundance with Ev = 3 × 10(53) ergs to fit the EMP stars in the normal SN model, while cannot enhance in the HN model. Even Ev = 9 × 10(53) ergs is not enough to reproduce [Mn/Fe] in the HN models. In the HN models, neutrino flux becomes smaller because the radius becomes larger by the shockwave than in the normal SN models. Therefore, the effects of neutrino reactions becomes smaller in the HN models than in the normal SN models. The !-processes also enhance [Sc/Fe] in the normal SN model to some extent. However, the enhancement is not enough to fit EMP stars. As for Co, there is no enhancement with v-processes both in the SN and HN models.

(Chapter 6)

The quantity of K is generally underestimated by the models of supernovae. These days, various SN explosion models are being suggested. It is no wonder that new hydrodynamics never before seen in explosion simulations actually occurs. In this situation, it is not enough to calculate post-process nucleosynthesis using the hydrodynamic trajectories of SN simulations. In Chapter 6, I investigate favorable conditions for producing K by performing a wide parametric search using analytical hydrodynamic models.

I find that the case with proton-rich and high initial density of ! 109 [g/cm3] is the only one where [K/Fe] becomes high enough to reproduce EMP stars. My result means that proton-rich and high density of 109 [g/cm3] actually exists in SNe, or that the origin of K is other than SNe, i.e., stellar evolution.

(Chapter 7)

In my calculation, I adopt Thielemann's REACLIB, BASEL (Thielemann et al. 1987), and complement reaction rates which is not included in BASEL set with a new set of REACLIB (JINA REACLIB). The reason for my using BASEL is that our group's previous works has been using BASEL, and that I would like to maintain consistency with the previous results. However, it is true that we should make sure to what extent those results are changed with a different set of reaction rates.

In Chapter 7, I recalculate SN and HN yields applying a different set of nuclear reaction rates, and investigate whether uncertainty from reaction rates can change the abundances largely or not. I find that the reaction rates of (63)Ga(p,γ)(64)Ge by NON-SMOKER code (Rauscher and Thielemann 2000) is at least not good for reproducing the observations because it increases [Cu/Fe] and decreases [Zn/Fe]. I also find that [Co/Fe] increases with the rate of 59Cu(p,")60Zn by NON-SMOKER, but the enhancement is not enough to reproduce the observation. Although some uncertainty from different sets of reaction rates does exist, HN models are still better to reproduce [Zn,Co/Fe] in EMP stars.

(Chapter 8)

In Chapter 8, discussion and conclusion are given. From all the results in this thesis, I conclude that HNe are still the most possible origin of EMP stars.

本論文は9章からなる。第1章はイントロダクションであり、超金属欠乏星と呼ばれる銀河系ハローに存在する鉄の含有量が太陽の1000分の1以下しかない星の重元素組成が銀河系形成初期に爆発した大質量星で起こった元素合成の情報を保持していることが述べられている。また、理論的モデルをもとに計算した元素合成過程の結果と超金属欠乏星を分光観測した結果得られる元素組成を比較することで初期銀河で形成された大質量星の性質とその元素合成過程に関する情報を引き出すことが本論文の目的であることが示されている。

第2章では本研究で用いられている理論的モデルと数値計算法について説明されている。第3章では、超金属欠乏星に検出された亜鉛の起源を調べている。先行研究では亜鉛と鉄の組成比が爆発エネルギーが超新星より一桁以上大きい極超新星で再現できるとする研究と、超新星でもニュートリノによって高温になったホットバブルと呼ばれる部分の寄与を計算に組み入れることができれば再現できると主張する研究があり、決着を付ける計算を行っている。その計算では、超新星爆発が非球対称であることと爆発機構が未だ不明であることから、 球対称超新星爆発モデルに基づいて得られた 流体素片の熱力学的な量の時間変化や核子当りの電子数を流体力学計算の結果から人工的に変更し、その不定性を補っている。 広範囲にパラメータを振って、観測された亜鉛や鉄族元素と鉄の組成比が実現される領域を探索したことが本研究の特徴である。 その結果、エントロピーが高いホットバブル領域で核子当りの電子数が少ない(中性子過剰)なら観測された亜鉛と鉄の比が再現できるとこを見出したが、コバルトの観測量を再現できないことが分かった。さらに爆発エネルギーの大きな極超新星と呼ばれる爆発での元素合成の計算と観測との比較も行った。爆発エネルギーが大きいためより高温な領域が出現し、一番深い領域で亜鉛が大量にできることが示され、観測された亜鉛は極超新星で合成されたと結論づけた。

第4章では、極超新星の中でも中性子過剰な領域でストロンチウム(Sr)やイットリウム(Y)やジリコニウム(Zr)などの弱r-過程元素と呼ばれる元素が合成されるかを調べている。金属欠乏星で検出される中性子捕獲過程元素のうち原子数が56<Z<76の範囲の元素の組成はどの星でもほぼ同じであるのに対し、弱r-過程元素の組成は星ごとに違っていることが述べられている。さらに、超新星爆発機構が不明なために、色々なシナリオに基づいた先行研究では弱r-過程元素の組成についての結果が一致しないことが述べられている。ここでは観測された弱r-過程元素と亜鉛の組成を再現する爆発の仕方を、前章で用いられた単純化したモデルを用いて調べた。その結果、超新星モデルでも観測された弱r-過程元素を再現することは可能であるが、より良く観測を再現するのは極超新星モデルであると結論づけている。

第5章では、ニュートリノによって元素を壊して合成される過程によって組成にどれくらいの影響が出るかを調べ、極超新星よりも爆発エネルギーの小さな超新星の方がこの影響が顕著であることを示した。

第6章は金属欠乏星に検出されたカリウム元素について超新星爆発での合成の可能性を調べ、爆発前の星の進化段階で合成することが必要であることを示した。第7章では広く用いられている2つの公開されている核反応率データベースについて、その原子核モデルの違いについて説明し、それらを用いて計算した元素合成の結果を比較し、その影響が極めて限定的であることを確認している。第8章はまとめと結論であり、第9章は付録で、本研究で計算した各モデルの各元素量が表として掲載されている。

以上、本論文は、球対称爆発モデルの不定性をパラメータ化するという独自の手法を用いて超金属欠乏星に観測された重元素組成を再現することで、銀河初期に存在した大質量星が極超新星爆発を起こした形跡が超金属欠乏星の元素組成に顕著に残っていることを初めて示した重要な研究であり、高く評価できる。

なお、本論文第3章は梅田秀之との共同研究、第4章は梅田秀之、冨永望との共同研究、第5章は梅田秀之、吉田敬との共同研究であるが、論文提出者が主体となって計算及び結果の解析を行ったもので、論文提出者の寄与が十分であると判断する。したがって、博士(理学)の学位を授与できると認める。