Loading web-font TeX/Main/Regular
Influence of Substrate on Microvia Structures in Printed Circuit Boards During Reflow | IEEE Journals & Magazine | IEEE Xplore

Influence of Substrate on Microvia Structures in Printed Circuit Boards During Reflow


Finite element model of FR4 substrate with Microvia and characterization of failure risk zones based on Microvia stress outcomes.

Abstract:

The automotive industry’s increasing dependence on compact electronic packages requires printed circuit boards with robust microvia interconnections. With a focus on mini...Show More

Abstract:

The automotive industry’s increasing dependence on compact electronic packages requires printed circuit boards with robust microvia interconnections. With a focus on minimizing failure rates and reducing costs, automotive manufacturers require that microvia reliability have zero failures. To achieve this, the automotive industry requires identification and understanding of the uncontrollable factors related to substrates that influence microvias during the reflow process. However, conventional studies not accounting for the non-uniform nature of substrates fail to capture the real-time impact of composite substrates on microvias during reflow, highlighting the need for comprehensive substrate analysis. In this paper, we present a detailed finite element model substrate layer to analyze the influence of the substrate material on microvia structures using the design of experiment technique using the Taguchi method. The results showed that the uneven distribution of fiber and resin ratios within the substrate layer affects the microvia structures. Microvias in resin-rich areas are more susceptible to failure with 33% higher stresses than those present in fiber-rich areas. Furthermore, printed circuit board substrate is characterized into different zones based on microvia failure risk levels.
Finite element model of FR4 substrate with Microvia and characterization of failure risk zones based on Microvia stress outcomes.
Published in: IEEE Access ( Volume: 11)
Page(s): 142487 - 142494
Date of Publication: 14 December 2023
Electronic ISSN: 2169-3536

SECTION I.

Introduction

The printed circuit board (PCB) technology has recently evolved from single-level buried structures to copper-filled stacked microvias. The electronics industry creates highly efficient products with reduced component sizes on printed circuit boards (PCBs) [1], [2]. Institute of Printed Circuits (IPC) standards define microvias as buried or blind vias with diameters equal to or less than 150 ~\mu \text{m} [3], [4]. A PCB contains several traditional layers placed between multiple high-density interconnect (HDI) layers [5]. Microvias are used as electrical interconnects among different conducting layers in HDI technology [6], [7]. A significant number ranging from a few thousand to millions of microvias are used on a single PCB.

With the sudden increase of features in the automotive industry, the demand has grown for compact PCBs with higher densities of components and their reliability concerns. For example, flame retardant (FR4) is the most used PCB laminate material that presents a major reliability concern because of its large coefficient of thermal expansion (CTE) in the thickness direction of PCBs [8], [9]. As miniaturization advances, it is thought that several factors significantly influence PCB reliability. These factors include the effect of via diameter, aspect ratio [10], [11], [12], [13], [14], [15], and the plating thickness of PCBs [16]. Therefore, to optimize the design margin for miniaturization, FR4 was treated as a homogenous body, averaging out the collective impact on microvias. FR4 is typically a composite in which plain fiberglass woven cloth and resin are laminated [17]. The microvia is made using the drilling process and most of the surrounding area is glass fiber and resin [18]. No studies have been conducted on finding the interaction between microvia and its immediate surroundings. Figure 1 shows a schematic diagram of a single-layer PCB substrate with different fiber and resin concentrations across the board.

FIGURE 1. - Schematic diagram of the substrate and top view of fiberglass.
FIGURE 1.

Schematic diagram of the substrate and top view of fiberglass.

Microvia reliability research has primarily centered around the core focuses of design optimization and understanding the failure mechanisms of microvias. Some researchers have focused on experimental investigations of the reliability of single level microvias [19], [20]. Some publications have addressed the microvia reliability of single level unfilled or epoxy-filled microvias. However, few researchers have addressed the reliability of copper-filled microvias [21], [22]. Recent research found that concentration of small elements in electroless copper of microvia affects the reliability of HDI substrates [23], [24]. Birch [25] tested stacked electroplated microvias and found that two-level stacked microvias are relatively more reliable than four-level stacked microvias. Previous research has found different types of failure in microvias including cracks around corners, knees, pads, and interfacial separation [19], [20], [21]. Ning et al. [26] investigated the voiding effect on the reliability of copper-filled stacked microvias in detail with different characteristics of voiding such as void shapes, location, and sizes. However, previous studies did not investigate the substrate layer’s influence on microvia structures.

In reliability investigations, it is easy to control individual variables in numerical simulations, but evaluating the combined effects of factors poses challenges. The Taguchi method invented by Japanese scientist Dr. Genichi Taguchi addresses this issue by providing the assessment of combined factors and their impact on product reliability through a reduced number of simulations. The Taguchi method focuses on the effective application of engineering strategies instead of advanced-level statistical techniques and is widely used in the design of experiments (DOE) [27], [28], [29], [30].

The reliability of microvias in automobile PCBs is crucial. Automotive PCBs more often need to survive in severe environments than other applications and demand robust microvia interconnections to reduce failure-related costs. Thus, the automotive industry requires ensuring a zero-failure rate from microvia reliability. Additionally, the large number of microvias makes it impractical to test each microvia of PCB lots due to time and cost constraints. Previous studies that regard substrates as uniform fail to capture the real-time impact of substrates on microvias. Therefore, there is a need to figure out the real impact of the composite substrate on microvias during reflow.

In this paper, a comprehensive composite finite element model is developed to research the intricate dynamics of substrate layers and their influence on microvias. By utilizing the DOE technique using the Taguchi method, this paper explores critical factors affecting microvia failure above glass transition temperatures. The authors found that uneven distribution of resin and fiber in the substrate resulted in different stress concentrations on microvias. This research presents a major contribution to understanding the behavior of composite substrates and provides valuable insights for optimizing microvia reliability in automotive PCBs.

The rest of the paper is organized as follows. The novel approach of this research is described in Section II. Finite element modeling details are presented in Section III. After the Taguchi methodology with the proposed factors has been described in Section IV, Section V presents the simulation results. Section VI presents the different characterized failure risk areas for microvias. Section VII is the discussion and Section VIII concludes the paper.

SECTION II.

Novel Approach

Glass fiber cloth is a key component in the FR4 laminate and substrate layers of PCB. Glass fiber consists of long-bundled glass spindles that are woven and impregnated with resin. It is then cured to form a solid laminate providing electrical insulation among conductive layers of the board. Glass fiber along with the restraining force it exerts on the microvia has been identified as a combined factor affecting reflow reliability of microvias [31], [32]. Figure 1 illustrates that areas on the glass weave laminate exist where the resin concentration is higher than the fiber and vice versa. Therefore, it is necessary to determine the areas across the substrate where a high possibility of microvias failure exists. Fiberglass cloth has low CTE while epoxy resin has high CTE. The substrate throughout the layer consists of resin-rich and fiber-rich sites. So, the entire PCB does not carry an equal risk of microvia failure. Furthermore, the stress concentration of microvias in different positions on PCB yields varying results due to irregular distributions of fiber and resin as shown in Figure 2.

FIGURE 2. - Illustration of microvias at various positions in different fiber resin concentrations.
FIGURE 2.

Illustration of microvias at various positions in different fiber resin concentrations.

The main challenge for this problem includes addressing the imprecise microvia positioning (which affects reflow reliability) and understanding the role of fiberglass in FR4 substrates. To address these issues, several factors related to PCB behavior including geometry and material properties were analyzed. Among these factors, the key variables (including the microvia location and resin properties) were selected for the finite element analysis (FEA). Instead of using a single material value of FR4, the material value of each component with specific properties within the FR4 substrate was used. It provided a unique and clear understanding of the impact of each element of FR4 on microvias. Different failure risk zones were statistically identified by modeling single-level microvias at different positions on the fiber-resin substrate. For validation purposes, the material properties of the proposed finite element model were measured and were found to fall within the range of reported properties in the literature [26], [33].

SECTION III.

Finite Element Modeling

To determine the behavior of fiberglass and resin towards microvia, 3-D finite element models were created to study the stress evolution during thermal loading. Given that the purpose of this work is to research the impact of the substrate layer on microvias, this study is limited to modeling the single level microvia along with a composite substrate layer having fiberglass and epoxy resin. Analyzing single level microvia is preferred due to its simpler structure and placement of single glass fiber cloth layer. For stacked and staggered microvia types, FR4 sheets contain complicated structures including multi-layers of glass fibers with the phase difference [32]. The thickness of the substrate layer model falls within the range of commercially produced substrate sheets [34]. There can be several thousand to millions of microvias in a single PCB layer. However, in this study, a fiber-reinforced composite substrate is modeled having only one microvia structure. Several models were made by changing the microvia locations and positions in the fiberglass. Since fiber cloth is woven in a periodic shape, when modeling the glass fiber cloth, the shape of the fiber becomes too complicated if each fiber thread is individually modeled. Therefore, each fiber bundle is modeled as a homogenous body. The fiberglass weave has no bond among the fiber threads, but it is impregnated with resin.

Due to impregnation, the resin layer in between the fiber bodies is negligibly small and is not considered in modeling. However, the effect of resin is accounted for through the material properties, which adhere to the composite rule of mixtures, ensuring balanced characteristics. For microvia modeling, both capture and target pad diameters are 200 ~\mu \text{m} , microvia diameter from the capture pad end is 110 ~\mu \text{m} and from target pad end is 100 ~\mu \text{m} . The height of microvia is 100 ~\mu \text{m} . Figure 3 shows the finite element analysis model of woven fiberglass with microvia interconnect. The fiberglass and microvias are encased within a resin layer.

FIGURE 3. - Finite element model of FR4 substrate including the model of microvia.
FIGURE 3.

Finite element model of FR4 substrate including the model of microvia.

The glass fiber cloth has a single layer in which threads are woven in X- and Z- direction. Figure 4 shows the illustration of fiber cloth with design parameters that were considered for the finite element modeling of fiberglass 1 and 2. The detailed dimensional measurements of glass cloth are shown in Table 1. The fiberglass is an orthotropic material with two directions (i.e., X- and Z- directions) orthogonal to each other in the board plane and a third direction (Y-direction) out of the plane. In this study, the orthotropic glass fiber material properties were used and are listed in Table 2 [32].

TABLE 1 Dimensional Details of Finite Element Model of Glass Fiber 1 & 2 in \mu\text{m}
Table 1- 
Dimensional Details of Finite Element Model of Glass Fiber 1 & 2 in 
$\mu\text{m}$
TABLE 2 Glass Fiber Properties (Orthotropic)
Table 2- 
Glass Fiber Properties (Orthotropic)
FIGURE 4. - Schematic diagram of glass fiber cloth cross section with design nomenclature.
FIGURE 4.

Schematic diagram of glass fiber cloth cross section with design nomenclature.

The equivalent CTE of fiber in X- and Y- (thickness) has the value of 15 ppm/° since it has a fiber-resin matrix and was used following the composite rules of mixtures. The isotropic resin is supposed to change remarkably above the glass transition temperature T_{g} of 137 °C, listed in Table 3 [33]. The CTE of copper is 17 ppm/°, E = 120 GPa is the elastic modulus, and the Poisson ratio is 0.35 [33].

TABLE 3 Resin Properties (Isotropic-Glass Transition Temperature-137 °C)
Table 3- 
Resin Properties (Isotropic-Glass Transition Temperature-137 °C)

FEA was conducted for the determination of thermo-mechanical stress evolution in microvias under thermal loading. As for the temperature load, the initial temperature was set at room temperature, 25 °C. The temperature load was given in two steps elevated from room temperature i-e., from 25 °C to 240 °C. The temperature holding time was not considered because the analysis in this study was not an absolute evaluation of strain occurring in microvia and base material. In this study, von Mises stress was observed to determine the location of the potential failure site and maximum damage. The authors measured the equivalent CTE and equivalent Young’s modulus of the composite model. The results were compared with the typical range of FR4 material properties reported in the literature [14], [26], [33] to verify that the results are logical and fair to use. The results are shown in Table 4.

TABLE 4 Equivalent Material Properties of FR4 Substrate Model
Table 4- 
Equivalent Material Properties of FR4 Substrate Model

SECTION IV.

Taguchi Methodology

The Taguchi method is generally used to enhance the quality and efficiency of products. It is one of the best experimental methodologies to be performed with a different number of factors and their levels. In this study, four different factors with three levels of each were chosen to conduct analysis. Hence, an orthogonal array L_{9} with nine rows and four columns was chosen. The four parameters along with three levels are shown in Table 5. Taguchi factors from A to C, are schematically shown in Figure 5. Factor D is the resin material property i.e., CTE is shown in Table 5. Since the microvia location is the key factor in finding the high-potential failure risk areas, the design factor ‘Via Position’ was also considered. Three via positions were modeled from the center of the resin-rich area as shown in Figure 5(c). Epoxy resin influences PCB with respect to glass transition temperature (T_{g} ) limit (smaller CTE values under T_{g} and much larger CTE values above T_{g} ) [33]. Three levels of CTE for the Taguchi trial run were established by increment and decrement of original CTE by 10% and are shown in Table 5. The trial run order with corresponding factor and their levels is shown in Table 6.

TABLE 5 Taguchi Factor and Their Levels for Finite Element Modeling
Table 5- 
Taguchi Factor and Their Levels for Finite Element Modeling
TABLE 6 Taguchi L9 (34) Orthogonal Array With Respective Values of Factors
Table 6- 
Taguchi L9 (34) Orthogonal Array With Respective Values of Factors
FIGURE 5. - Schematic of Taguchi factors. (a) illustration of Taguchi factor A i.e., microvia locations at PCB substrate (b) Schematic of Taguchi factor B i.e., resin thickness (c) Schematic of Taguchi factor C i.e., via positions on each via location along z(i), along x(ii), and at 45° between z and x(iii).
FIGURE 5.

Schematic of Taguchi factors. (a) illustration of Taguchi factor A i.e., microvia locations at PCB substrate (b) Schematic of Taguchi factor B i.e., resin thickness (c) Schematic of Taguchi factor C i.e., via positions on each via location along z(i), along x(ii), and at 45° between z and x(iii).

SECTION V.

Results

The von Mises stress across microvias in nine simulations was observed and is shown in Figure 6. The response table and mean effect of each factor were determined using statistical analysis, shown in Table 7 and Figure 7, respectively. From Table 7, CTE resulted as the most significant parameter while substrate layer thickness was the least significant factor. Figure 7 reveals the gradual decrease of mean stress across microvia from the microvia position 80 ~\mu \text{m} to 280 ~\mu \text{m} . However, it is crucial to investigate the situation of stress specifically in open resin window sites (center of the resin-rich area). Therefore, three additional finite element simulations were conducted by modeling microvias at the center of the resin-rich area. The remaining Taguchi factors were kept the same for base comparison. The microvias observed high stresses at the center of the resin-rich area as shown in Figure 8 with a mean stress of 1084.3 MPa.

TABLE 7 Response Table of Stress Means (MPa) of Simulations Including Contributions of Each Factor Toward Stress Evolution Around Microvias
Table 7- 
Response Table of Stress Means (MPa) of Simulations Including Contributions of Each Factor Toward Stress Evolution Around Microvias
FIGURE 6. - Microvia stress distribution simulation results across substrate models of Taguchi L9 orthogonal array (OA): Blue color indicates minimum stresses whereas red color indicates maximum stresses on microvias. Each microvia model displayed maximum stress concentration at the corners.
FIGURE 6.

Microvia stress distribution simulation results across substrate models of Taguchi L9 orthogonal array (OA): Blue color indicates minimum stresses whereas red color indicates maximum stresses on microvias. Each microvia model displayed maximum stress concentration at the corners.

FIGURE 7. - Main effects plot for stress means of each Taguchi factor. Y-axis is the same for all result trends i.e., Mean stress. X-axis (A) microvia locations across the substrate (B) resin thickness in 
$\mu \text{m}$
 (C) microvia positions on fixed via locations across substrate in 
$\mu \text{m}$
 (D) coefficient of thermal expansion of the resin in ppm/°.
FIGURE 7.

Main effects plot for stress means of each Taguchi factor. Y-axis is the same for all result trends i.e., Mean stress. X-axis (A) microvia locations across the substrate (B) resin thickness in \mu \text{m} (C) microvia positions on fixed via locations across substrate in \mu \text{m} (D) coefficient of thermal expansion of the resin in ppm/°.

FIGURE 8. - Simulation results of microvias at open resin window location on the substrate.
FIGURE 8.

Simulation results of microvias at open resin window location on the substrate.

SECTION VI.

Risk Area Calculations

With the help of the main effect plot shown in Figure 7, several microvia possibilities were plotted on the PCB with respective risk levels as shown in Figure. 9. Identifying the precise position of a microvia in PCBs just by observation can be challenging. Thus, to find a high failure risk area, multiple microvia positions were illustrated based on the possibility of microvia presence at those specific locations on the PCB. Microvias located in resin-rich areas exhibited a higher risk of failure, while microvias located in fiber-rich regions demonstrated lower risk of failure. Since many microvias are present in a single PCB, it is impractical to determine the exact number of microvias present in a higher failure risk zone on the board. However, due to the uniform pattern of the glass fiber cloth across the board, a unit cell was considered in this study to calculate the high-risk zone area for microvias failing in percentage. The illustrated microvias inside the unit cell are shown in Figure 10. Risk calculations were done statistically based on the risk levels of microvias in the unit cell. Only 6.3% of the area turned out to be a high failure risk area for microvia failure, whereas 38.2% area of the unit cell had a minimal risk for microvia failure. Table 7 shows the details of the calculated percentage of PCB area in a unit cell that possesses high failure risk for microvias. From Table 8, it is evident that the substrate has a very small proportion of the total board area in which microvias demonstrate a high failure rate. Figure 10 shows that the unit cell contains a high failure area on each corner, demonstrating a uniform pattern throughout the PCB. The highest mean stress outcome of 1084MPa was observed at the center of the resin-rich area and moderately decreased when a microvia was positioned at some distance from the center of a resin-rich area as shown in Figure 11.

TABLE 8 Risk Zone Percentage Area in the Unit Cell (Colored Notation from High [Red] to Low [Green] Risk Area)
Table 8- 
Risk Zone Percentage Area in the Unit Cell (Colored Notation from High [Red] to Low [Green] Risk Area)
FIGURE 9. - Illustration of microvia failure risk zones from high(red) to low (yellow) risk on the substrate.
FIGURE 9.

Illustration of microvia failure risk zones from high(red) to low (yellow) risk on the substrate.

FIGURE 10. - Illustration of the unit cell of the board to calculate risk zones around the cell in percentage.
FIGURE 10.

Illustration of the unit cell of the board to calculate risk zones around the cell in percentage.

FIGURE 11. - Mean stress values of microvias in various positions on substrate.
FIGURE 11.

Mean stress values of microvias in various positions on substrate.

SECTION VII.

Discussion

The results of this study indicate that resin-rich sites of the substrate are susceptible to microvia failure. This study demonstrated that the CTE in the thickness direction of the substrate layer had a significant impact on microvia structure. CTE effect on the microvia reliability investigations is well-known [8], [26]. However, this study showed that the CTE doesn’t uniformly impact the microvia structure throughout the substrate layer. Resin-rich sites had a larger CTE impact on microvia structures than fiber-rich sites. Microvias located in open resin window sites experienced the mean stress concentration of 1084 MPa at CTEs of 180–220 ppm/° (above T_{g} ). Conversely, within the same CTE range, mean stresses of 718 MPa were exhibited by microvia structure around fiber rich sites. Yet the higher CTE 220 ppm/° (above T_{g} ) experienced the highest stresses regardless of microvia locations (see Table 7, CTE at level 3), which provides agreement with the results of the study by Ning [8], where the highest CTE: 70 ppm/° are associated with the longest strain values, ultimately highest induced stresses.

Additionally, the fiber-rich and resin-rich areas of the substrate had a significant impact on the microvia structure. The fiberglass weave bundle strengthened the microvia structure lowering the risk of failures above the glass transition temperature. The microvias at the center of the resin-rich area observed a maximum mean stress concentration of 1084 MPa whereas the microvia at the location of the fiberglass weave observed 727MPa mean stress around the corners (as shown in Figure 11). A 33% increase in the mean stress concentration on microvias was observed that was present in a high failure risk zone, compared to the stress concentration present in a low failure risk zone. Throughout the fiberglass cloth, open resin windows are the locations that have 0% fiber content and 100% resin content. Based on the outcomes shown in Figure 8, open resin window locations of fiberglass have the highest potential failure risk for microvias. However, the area of the high microvia failure-risk zone is too small i.e., only 6.3% of the area in the unit cell contains the high failure site for microvia that might be replaced in design optimization. Additionally, to mitigate failure risks, careful consideration must be given to designing the pitch in a manner that avoids high-risk areas.

SECTION VIII.

Conclusion

This paper shows that composite substrates have an impact on microvias during reflow. The design of experiments for finite element analysis using the Taguchi method was presented to study the effect of fiberglass weave and resin on microvias. Findings reveal that microvias in open resin windows with pure resin content experience 33% higher stress concentration than those in low resin content areas. Consequently, this study classified the fiber-reinforced resin layer into different zones based on the failure risk of microvias.

Among all the parameters studied in this work, the thermal expansion coefficient and microvia location have the most significant influence on microvia structures. Specifically, at the center of the open resin window, microvia exhibited the highest stress of 1219.43 MPa among all simulation results. This indicates that the possibility of microvia failure could be increased if microvia is located at the open resin window location of the substrate with high CTE. Therefore, more attention should be given to these two parameters during the designing and testing phase of the HDI boards. The fiberglass weave in the substrate provides strength to the microvia at elevated temperatures and prevents it from early failures. Moreover, recognition of high-risk failure sites can guide design improvements, considering the design parameters such as microvia diameters and pitch to improve reliability during the design phase.

References

References is not available for this document.