Sample preparation.
To prepare a reference model of tooth preparation, the right maxillary first molar was milled under the following conditions (occlusal reduction, 1.5 mm; axial reduction, 1.2 mm; finish line design, chamfer) using a milling unit (Ezis HM; DDS, Seoul, Republic of Korea). To reproduce the oral environment, a lithium disilicate ceramic (IPS e.max CAD; Ivoclar Vivadent AG, Schaan, Liechtenstein) having a transparency similar to that of natural teeth was used. After the crystallization process according to the manufacturer’s recommendations, to reduce the gloss of the surface, the surface was polished using diamond rotary instruments (852.FG.010; Jota AG, Rüthi, Switzerland). The adjacent teeth were manufactured using a 3D printer (Megprinter; Megagen, Daegu, Republic of Korea), transparent silicone (Elite Transparent; Zhermack, Badia Polesine, Italy) was used to reproduce the oral environment, and red pigment (406 red; Shinhan, Seoul, Republic of Korea) was used and replace with semitransparent silicone.
Fabrication of interim crowns and evaluation of intaglio surface trueness.
To determine the number of interim crowns (sample size) to be fabricated per finishing line locations, three pilot experiments were performed prior to this study. Based on the results of the pilot experiment, the sample size was determined using power analysis software (G*Power v3.1.9.2; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) (N = 16; effect size [f] = 0.63; actual power = 99.11%; power = 99%; α = 0.05).
To obtain a reference virtual model of tooth preparation, a precise surface scanning was performed using a contact scanner (DS10; Renishaw plc, Gloucestershire, UK) (Fig. 1). To obtain a high-resolution virtual model, five standard tessellation language (STL) files were acquired through contact scanning and five STL files were merged after optimization alignment by using a 3D mesh software program (Geomagic Design X; 3D Systems, Rock Hill, USA).
The reference model of tooth preparation was adapted to the conditions of each group and fixed to the reference model without movement. The supragingival finishing line was located approximately 0.5 mm above from level of the gingiva, whereas the subgingival finishing line was located approximately 0.5 mm below from the level of the gingiva. The equigingival finishing line was located at the level of the gingiva. Additionally, at the subgingival finishing line, a gingival displacement cord (# 2 Ultrapak; Ultradent, South Jordan, UT, USA) was packed into the gingival sulcus below the finishing line. The depth of the subgingival finishing line was confirmed using a periodontal probe (CP 15 UNC; HU-Friedy, CHI, USA).
To obtain a test virtual model of tooth preparation, an intraoral scanner (i500; MEDIT, Seoul, Republic of Korea) was used to scan a reference model at the supragingival, equigingival, subgingival, and subgingival finish line locations with gingival displacement cords (N = 16 per locations; Fig. 1). All scanning and analysis procedures were performed by an experienced investigator (K.S.).
Sixteen test virtual models acquired per finishing line locations and a reference virtual model were extracted as STL files for interim crown fabrication. In a dental CAD software program (3Shape, Copenhagen, Denmark), the design of interim crowns was performed under the same conditions of a 60-µm cement space. The STL file of the interim crown designed based on the reference virtual model was designated as a CAD reference model (CRM) for the evaluation of intaglio surface trueness (Fig. 1). Interim crowns designed based on the test virtual model were fabricated using a stereolithography 3D printer (ZENITH; Dentis, Daegu, Republic of Korea) with 0° parallel to the vat bottom. In consideration of the printing and repetition accuracy according to the position of the printed object in the vat, the interim crowns produced in four groups were divided into quarters and adjusted to the same position and number when printing once. For the photopolymerization resin for the interim crown, 3D printing resin (For interim crown; Dentis, Daegu, Republic of Korea) was used. For interim crowns after printing, all residual resin was removed according to the manufacturer’s recommendations, and postphotopolymerization was performed using a light-curing unit (CUREDEN; Kwang Myung DAICOM, Seoul, Republic of Korea). All evaluations were completed within 3 hours after printing in consideration of the dimensional change according to the time change after printing. The intaglio surface of interim crowns after all posttreatments were scanned using an intraoral scanner (i500; MEDIT, Seoul, Republic of Korea), and the STL file was designated as the CAD test model (CTM) for the evaluation of the intaglio surface trueness (Fig. 1).
Through the evaluation of the intaglio surface trueness, the accuracy of the intaglio surface of interim crowns manufactured according to the finishing line locations was compared (Fig. 1). CRM and CTM alignment and 3D comparison were performed using a 3D inspection software program (Geomagic Control X; 3D Systems, Rock Hill, SC, USA) (Fig. 1). The area of the intaglio surface was segmented based on the margin of CRM. To evaluate the intaglio surface area in detail, it was divided into the marginal, axial, and occlusal regions. CRM and CTM were aligned based on the segmented intaglio surface, and the root mean square was calculated as follows based on all cloud points of the CRM intaglio surface (1):

where Di represents the gap distance of point i of CRM and CTM and n is the number of all points evaluated.
Evaluation of the marginal and internal fit.
The silicone replica technique was performed to evaluate the marginal and internal fit of interim crowns. After filling the silicone indicator (Aquasil Ultra XLV; Dentsply Detrey GmbH, Konstanz, Germany) in the intaglio surface of the interim crown and accurately positioning it on the tooth preparation, use a jig capable of continuously applying the force (300 gf) on the occlusal surface. Continuous force was applied until the polymerization of the silicone was completed (Fig. 2a). The silicone indicator attached to the intaglio surface of the interim crown was filled with silicone of a different color, and the silicone replica was separated from the interim crown after polymerization. For identical cutting of the silicone replica, an industrial CAD software program (SolidWorks 2014 software; Dassault Systems SolidWorks Corp., Waltham, MA, USA) was used to design a jig based on CRM, and the jig was fabricated using a 3D printer (Megprinter; Megagen, Daegu, Republic of Korea) (Fig. 2b). The jig for cutting the silicone replica is designed to cut the buccolingual and mesiodistal planes based on the center of the interim crown (Fig. 2c). The distance in the silicone replica (Fig. 2d) was measured using an optical microscope (IMS 1080P; SOMETECH, Seoul, Republic of Korea). As for the measurement point of the marginal fit, the marginal gap (MG), which measures the marginal opening, and absolute marginal discrepancy (AMD), which measures the distance between the finishing line and the margin of the prosthesis, were evaluated (Fig. 3). The measurement points of the internal fit are the chamfer gap, which measures the distance between the center of the chamfer curvature of tooth preparation, angle gap, which measures the distance between the center of the angle curvature, and axial gap, which measures the distance between the center of the chamfer and the angle (Fig. 3). The occlusal gap was evaluated by measuring the distance between the center of the occlusal and middle point of the angle (Fig. 3).
Statistical analysis.
Statistical analysis was performed using a statistical software program (SPSS Ver 25.0; IBM, Chicago, USA) (α = 0.05). Since all the acquired data had a normal distribution, a parametric statistical analysis was used. Statistical comparison of the marginal and internal fit and intaglio surface trueness per groups was performed using one-way analysis of variance and the Tukey HSD test. The correlation between marginal region trueness and marginal fit (AMD and MG) was evaluated using Pearson correlation analysis.