With the rapid development of scientific technology, the study on the neural mechanisms of scientific metaphoric language is methodologically and epistemologically significant. However, as an important part of conceptual metaphors, scientific metaphors have been overlooked by the previous electrophysiological studies of metaphors. This study is the first to use ERP to compare the cognitive neural mechanisms of literal and metaphoric expressions from scientific and daily contexts. The aim of the present study is to investigate the neural mechanism of scientific language by observing the differences between scientific language (metaphoric and literal) and daily expressions as well as between scientific metaphors and literal expressions (scientific and daily). A list of 120 Chinese sentences was formed consisting of three categories: scientific metaphoric scientific literal, and, daily literal with 40 sentences in each sentence type. The scientific sentences were created based on the basic scientific terms and concepts at the level of middle school covering mathematics, physics and chemistry. All the 17 participants were from the natural scientific majors in order to avoid the possible difficulties in understanding those scientific sentences during the experiment. In the experiment, ERPs were time-locked to the onset of the last word of the sentence and were obtained by stimulus-locked averaging of the EEG recorded in each condition. The resulting amplitudes of N400 or LPC were entered into 3 condition × 3 region (frontal F3, Fz, F4, central C3, Cz, C4, parietal P3, Pz, P4) × 3 hemisphere (left F3, C3, P3, midline Fz, Cz, Pz, right F4, C4, P4) ANOVAs for repeated measures. In the time window of N400 (300-500ms), there were significant main effects of condition F(2, 32) = 9.90, p < 0.001，ηp2 = 0.38. Scientific metaphors elicited the most negative N400. Significant condition × region interactions were found, F(4, 64) = 3.84，p = 0.057，ηp2 = 0.19. Post-hoc tests showed that scientific language (metaphoric and literal) elicited more negative N400 than daily literal sentences in the parietal regions (ps < 0.05). Scientific metaphors elicited more negative N400 than literal expressions (scientific and daily) in the central and parietal regions (ps < 0.05). In the time window of LPC (600-900ms), there were significant main effects of condition F(2, 32) = 3.10，p = 0.065，ηp2 = 0.16. Scientific metaphors elicited the least positive LPC. Significant condition × region interactions were found, F(4, 64) = 4.93，p = 0.01，ηp2 = 0.24. Post-hoc tests showed that in the parietal region relative to daily literals, scientific metaphors elicited less positive LPC (p = 0.012) and scientific literals elicited marginally less positive LPC (p = 0.077). Besides, the subtraction waveforms were obtained by subtracting the amplitude of daily literals from that of scientific metaphors and scientific literals respectively. The subtraction waveforms clearly show that scientific metaphors elicited another late negativity peaking around 800ms following the N400. ANOVAs for repeated measures produced significant main effect of condition, F(1, 16) = 5.74，p = 0.029，ηp2 = 0.26. Scientific metaphors elicited more negative late negativity than scientific literals. There are two main findings in the present study. Firstly from the perspective of scientific language, the distance between the two semantic domains of scientific language is larger than that of daily language, thus requiring more cognitive resources to process the meaning. Both the right and left hemispheres are important in processing scientific language, but the parietal region, especially the right part, is vital. Secondly, from the perspective of scientific metaphors, our findings replicate previous studies, namely, the increased N400 amplitude to scientific metaphors could be reflecting the initial anomaly in literal meaning, while the late negativity could be the manifestation of a further attempt to gain meaning, this time using a non-literal route as predicted by the sequential view. Both the right and left hemispheres show their involvement in understanding metaphors. Besides, our findings support the graded salience hypothesis. The complexity and abstractness of scientific expressions reduced dramatically their familiarity compared to literal expressions. So, the meaning of a scientific expression is inferred by contextual mechanisms after the more salient daily meaning is processed.