A digital Polymerase Chain Reaction (PCR) system sample segmentation chip was developed for the “division” method for samples of a digital PCR system, and a capillary microarray was used for trace biological sample detection. First, arrayed silicon substrates were fabricated using Micro-electron-mechanical Systems(MEMS) technology, subsequently silicon wafers were thinned by the process of high-efficiency, low-damage, and ultra-precision grinding and then combined through chemical modification methods. A capillary microarray with a hydrophobic surface and hydrophilic inner wall was successfully prepared. The structure of the capillary microarray was characterized by scanning electron microscopy (SEM). SEM results showed that the structure of the capillary microarray is a through-hole microarray. The hydrophobicity of the surface of the capillary microarray was characterized by the contact angle, and the contact angles of the surface before and after chemical treatment were compared. The results show that the surface of the capillary microarray is hydrophobic after chemical treatment, and the contact angle is 118°. The hydrophilicity of the inner walls of the capillary microarray was characterized by an energy dispersive spectrometer (EDS). The results show that only Si and O elements are present in the inner walls of the capillary microarray, forming a hydrophilic group (—Si—OH). Thus, the inner walls of the capillary microarray were hydrophilic. The sample segmentation performance of the capillary microarray was characterized by measurement microscopy and fluorescence microscopy. The results show that the capillary microarray divided the sample into uniform independent units. Characterized by the laser confocal scanner, the overall sample segmentation effect of the capillary microarray is intuitively reflected, and the sample addition rate of the chip is 93.8% by counting and calculation. Finally, a capillary microarray chip with a hydrophobic surface and hydrophilic inner wall was successfully prepared and exhibited excellent sample segmentation performance, which offers broad application prospects for the field of biomedics.