创新系统“混沌”边缘具有较强的创新潜力，易于新兴技术创新网络的形成与发展。结合生命周期理论，分析了新兴技术形成过程及其与原有技术发展轨迹的关系，深入研究了“液态化”创新网络的形成机理。通过搜集自动驾驶汽车领域专利数据开展了实证分析，运用专利引用分析与社会网络分析相结合的方法，绘制IPC共现网络和专利权人引文网络，通过计算网络各项指标，验证了新兴技术创新网络“液态化”网络的技术特征与知识流动特征。实证结果表明：自动驾驶汽车领域的技术构成具有高异质性、技术领域间存在广泛融合性；知识流动网络存在小世界效应，呈现易流动、流向广、效率高，网络边界模糊，核心创新主体众多且多样化的特点。

    With the explosive growth of emerging technologies, the emerging industries have proliferated, which in turn accelerate technology iterations. An important characteristic of innovation in emerging technological fields is that it does not occur in isolation, innovations are generated and implemented by networks of interacting organizations and individuals. Thus, understanding what is the most efficient kind of innovative network becomes crucial for pursuing innovation performance and economic growth. 
    Within the innovative network, knowledge spillovers across technologies, frequent technology diffusion and convergence across different sectors occur. Thus, the innovative network gradually presents the characteristics of fuzzy network boundaries, flexible network connection, heterogeneity of innovators and easy flow of knowledge. Meanwhile, the network structure evolves to be between the ruled and the irregular status, showing a trend of de-core. The extant literature has not mentioned this kind of network structure. To address this gap, this paper proposes a "Liquification" innovative network, whose structure is between multi-core network and complete graphs. There is a high-density connection inside the "Liquification" network, making it easier to develop adjacent possibilities and spread new ideas or innovations. At the same time, it saves the newly generated useful linkages for a long time. Once the "Liquification" network is formed, innovation and invention will continue to emerge. This paper aims to study the formation mechanism of "Liquification" network，and use empirical test to further examine its characteristics.
    In mechanism analysis of "Liquification" network, we found that the technology diffusion and absorption can speed up technological development and improve its limit, delaying the coming of technology recession, thus extending the life cycle of the original technology. The frequent knowledge diffusion and integration among various innovative subjects shortens the distance between different technology networks. Boundaries between technological networks become blurred, and the diffusion of technology no longer depends on specific paths. The orderly diffusion process tends to be disorderly. The collision of information between different innovators changes the way of linking. Therefore, the innovative network presents feature such as knowledge heterogeneity, flexible links, strong network connectivity, strong knowledge mobility, widespread technology integration, and fuzzy network boundaries. The emerging technology innovative network present to be "liquified".
    In order to further verify the characteristics of the "Liquification" network in the mechanism analysis, this paper collects the patent data of the autonomous vehicle field in the Derwent Innovation Index and conducts an empirical analysis. Autonomous vehicle is a smart car that is driven by computer system, considered to be a disruptive innovation according to the emerging technology maturity curve released by Gartner. Observing networks in different time slices, we found that the technical domain increased sharply, providing potential opportunities for technological convergence. And the whole network connectivity was enhanced with a decreasing betweenness centrality and an increasing closeness centrality, indicating that integration between different technical fields is becoming more and more common and is no longer limited to some certain specific key technologies. 
   The results show that the formation of "Liquification" network can improve technology limits, accelerate technology development and extend the technology life cycle. The characteristics of "Liquification" network include knowledge heterogeneity, flexible links, strong network connectivity, strong knowledge mobility, widespread technology integration, and blurred network boundaries. It is also found that the network shows a de-core trend as the "Liquification" network formed.