Next Generation Molding Compound Materials for Flip Chip Matrix Array Molded Packages
Next Generation Molding Compound Materials for Flip Chip Matrix Array
Molded Packages
Lim Chong Sim a, Mun Leong Loke a, Kang Eu Ong a, Enrico Garcia a, Mohamed Syazwan Osman a, Leonel Arana b, Yoshihiro Tomita c, and Jiro Kubota c
a Assembly & Test Technology Development – Malaysia, Kulim, Malaysia
b Chandler Assembly Technology Development, Assembly & Test Technology Development, Chandler, Arizona
c Pathfinding, Assembly & Test Technology Development, Tsukuba, Japan
lim.chong.sim@https://www.sodocs.net/doc/0117221314.html,
Abstract—This manuscript describes the recent pathfinding and development activities on the molding compound for the flip chip matrix molded array (FCMX) platforms. The key focus of this manuscript is on the materials and packaging challenges associated with the high K molding compound and exposed second level interconnect molding. In addition the advantages and disadvantages of transfer and compression molding are discussed.
1. Introduction
FCMX packages are the next generation of packaging technologies designed for the mobility and ultra mobility product segments. Over the last few years, the mobile computing technologies have evolved into increasingly complex systems, resulting in significant packaging challenges. The key challenges often associated with the FCMX platforms are strip warpage, which could potentially result in high yield loss for the downstream modules. High strip warpage may also lead
to excessive package warpage, potentially violating the outgoing coplanarity specifications by the Joint Electron Device Engineering Council (JEDEC), which stipulates that a package with a ball grid array (BGA) pitch of < 0.5mm needs to meet an outgoing ball-coplanarity specification of under 100um.
Various technology options have been explored previously to address the warpage concerns. This includes having the right media design to restrain the strips from warping during the assembly process flow. Although media acts as an effective restrainer, it is an expensive solution, which would typically require some level of compromise within the assembly process flow
to enable integration of the media solution. Therefore from the manufacturability perspective, molding compound still remains as the most viable option for warpage controls.
In this manuscript, discussions are focused on the molding compound issues and requirements for the different packaging technologies. Sections 3 and 4 are focused on flip chip matrix molded ball grid array generation 3 (FCMB3) high K molding compound development, while Section 5 focuses on the flip chip matrix molded ball grid array generation 4 (FCMB4) ultra long gellation compression mold formulation. In addition the different molding compound technologies that drive the formulation developments are described. 2. Background: Molding Compound Drivers 2.1. Generic Molding Compound Requirements
The main function of molding compound is to protect the package from moisture, thermal and environmental exposure to ensure that it’s able to withstand its stipulated reliability requirements. In addition molding compound provides the package with the robustness for mechanical handling, packaging and assembly. Overall the molding compound requirements can be further summarized to the following key items:
?Low moisture absorption – “Pop-corning”
phenomenon occurs because of pressure building
up from the out gassing of moisture within the packages during the level 3 preconditioning (L3P)
stress resulting in interfacial or bulk fracture. In general, moisture absorption is lower for lower Glass Transition (T g ) resins such as the biphenyl
due to the lower free volume within the cross-
linked resins.
?Adhesion to multiple interface – Good adhesion to multiple interfaces such as underfill, solder resist
and silicon die is an important criteria to ensure
that there is no delamination between the interfaces
during assembly, which could result in crack propagation during temperature cycle stress.
?Low cure shrinkage – One of the key factors affecting strip warpage is the shrinkage of the molding compound during curing. Cure shrinkage
is defined by the differences between the total shrinkage with the thermal shrinkage.
Given the contrasting differences in packaging architecture between FCMB3 and FCMB4 (to be discussed further in sections 3 and 5), a generic molding compound requirements is not sufficient to
meet the overall package performance. Therefore a significant amount of work was done to formulate different molding compound formulations to meet the competing needs of each packaging technology options.
2.2. FCMX Packaging Roadmaps and Trends
Figure 1 shows the projected packaging trends for the
FCMX platforms, which is based on the dimensional scaling of the transistor’s density and customer’s requirements. The key driver is the scaling of power delivery requirements due to an increase in transistors density in order to meet the customer’s request for an increase in functionality. This consequently results in an increase in junction temperature, which reduces the silicon performance. The other driver which is based on projected customer requirements is the dimensional scaling of the package Z height requirements, which results in a lower silicon thickness and a thinner substrate technology.
Figure 1. Projected packaging trends for FCMX platforms.
2.3. New or Complex Packaging Technology
The emergence of new and complex packaging technologies due to design constrains have resulted in a new set of molding compound requirements. Figure 2 illustrates the revisions in packaging architectures that the FCMB4 pathfinding platform has undertaken due to changes in the product specific roadmaps (PSR). In general, both the overmold and exposed second level interconnect packages are assembled using the 1269 silicon architecture on a high density interconnect (HDI) direct laser lamination (DLL3) strip technology, which is attained through the laminations of multiple build up materials with laser via [1]. The key
advantages of the coreless DLL3 strip is the increase in signaling rates for high speed inputs outputs (HSIO) requirements and a reduction in the overall strip Z height. However this technology suffers from several drawbacks, including high warpage due to reduced thickness and absence of a core, which consequently results in yield loss at chip attach. The details on both the packaging technologies (1) Discrete over mold package and (2) Exposed second level interconnect molding will be discussed in details in the following sections.
Figure 2. Revisions in FCMB4 Packaging Architecture
2.4. Transfer vs. Compression Molding
Transfer molding has been Intel’s? legacy molding compound process for the Flip chip molded matrix array package 1 (FCMMAP 1), wire bonded and Flash platforms. However moving forward, the scalability for the transfer mold is at stake due to product requirements for a thinner mold clearance and lower temperature processing. Internal data collection on a package with 100um mold clearance indicated that the present transfer mold technology would be broken for molding compound with spiral flow <110cm as shown in Figure 3.
Figure 3. Projected packaging trends for FCMX platforms.
For the transfer molding process, molding compound pellets are placed into the plunger pots of the heated mold chase as shown in Figure 4. Once the mold compound is molten, the plunger transfers the molding compound through the runners into the cavity of the mold chase, and air in the chase cavities is vented out through strategically located venting holes. To ensure complete mold fill throughout the cavity of the mold chase, it’s important to define the right flow and melt viscosity properties. Nevertheless this is still a challenge for the high K molding compound
formulation because of the increase in melt viscosity with higher filler loading.
Figure 4. Transfer molding bottom chase with plunger pots.
The other known molding technique which was recently explored during the FCMB4 pathfinding is the compression mold. Compression mold shown in Figure 5 can be thought of as a runnerless mold system, because the molding compound is placed directly into
the cavity of the mold chase prior to clamping [2]. The
compression molding technique is unique because of its
ability to support both powder and liquid molding compound formulation. In addition compression mold has been demonstrated to be able to achieve thinner mold cap clearance. Compared to the transfer mold process, compression mold drives a reduction in through put time (TPT) with the omission of the molding compound transfer process shown in Figure 6.
Figure 5. Compression molding bottom chase filled with mold.
Figure 6. Potential TPT reduction with compression mold.
From the materials perspective, compression molding drives a separate set of requirements compared to the transfer molding formulations. For instance lubrication
wax is not required in the formulation because of the presence of mold film, which prevents the molding compound from sticking to the mold chase. In addition compression molding compound formulation typically requires a longer gelation time to enable it to flow and fill up the cavities within the mold chase.
2.5. Critical Molding Parameters
Among the key critical factors for warpage reduction is design geometry, materials properties and processing conditions. From the materials perspective, key
properties that influence the outcome of strip warpage
are shrinkage factor, glass transition temperatures (T g ), coefficient of thermal expansion (CTE) and the modulus. Shrinkage factor is a dominant force for warpage control at room temperature, while the T g and CTE are crucial to controlling the dynamic warpage of the strips at reflow temperatures because of CTE mismatch.
2.6. Test Vehicle Comparisons
Various test vehicles (TV) were designed and taped out during the FCMB3 development phase and FCMB4 pathfinding phase as shown in Table 1. The key differences between both the platforms lie with the substrate technology used. FCMB3 Pinang TV is an LDI substrate with a core, while the FCMB4 Joslin 2 and Tonkin TV are DLL3 strips designed to meet the high
density design rules. Due to the absence of cores, DLL3 strips are very flimsy, with high incoming warpage variations.
Table 1. Test Vehicle comparison between FCMB3 and FCMB4.
FCMB3 FCMB4 Attributes Pinang Joslin 2 Tonkin Die size 11 x 11mm 8 x 6mm 11 x 11mm Package size 13.8 x 13.8mm 13.8 x 13.8mm 13.8 x 13.8mm Substrate LDI DLL3 HDI DLL3 HDI
Silicon 1266 1266 1268 Ball pitch
0.5mm 0.413mm 0.5mm
3. High K Molding Compound for
Overmolded Package
80% Others MHS
Mold cure
60%
Clamp Injection
3.1. Motivation for High K Mold Compound Technology Traditionally, for an over molded package (e.g FCMB3), the key challenge is to enable an effective heat dissipation path from the silicon to ensure that the overall package is meeting the TDP requirements. This is because any increment in temperature by a magnitude of 10-15o C above the T j limit of 110o C may result in a 2X difference in the lifespan of the device [3]. Technology roadmaps have projected that the future FCMX platform may be required to envelop products with power dissipations up to 35 Watts, thus emphasizing the need for an effective package and
40%
T i m e
20% 0%
Transfer mold
Compression mold WB Compression mold FC Process material optimization
system level thermal solutions. Figure 7 shows that 90% of the heat generated by the silicon is dissipated through the board, and the remaining 10% dissipated to the surroundings via convection. This shows that molding compound could play an effective role in heat dissipation by virtue of providing a larger surface area for heat flow.
Figure 7. Heat transfer path for a typical FCMX package.
Thermal modeling in Table 2 has indicated that the conventional silica (SiO 2) filled molding compound formulation with a thermal conductivity value of
around 0.8W/mK would not be able to meet the Thermal Design Power (TDP) targets for a 5 Watts power dissipation product. Instead a 3.2 W/m.K molding compound would be required.
Table 2. T j max estimation based on 150um mold clearance.
To increase its thermal conductivity, various thermally conductive ceramic fillers such as Alumina (Al 2O 3) and Silicon Carbide (SiC) were investigated. The addition
of thermally conductive fillers adversely impacts the
molding compound properties (higher CTE, higher modulus and shorter spiral flow). Therefore, the challenge lies in developing a thermally conductive molding compound technology, which is processable.
3.2. High K Molding Compound Formulations
Market survey on existing suppliers showed that the high K molding compound is a generally new formulation with low levels of market penetration within the industries. Therefore collaborations with several of the molding compound suppliers were formed to specially develop a formulation that will meet Intel’s thermal conductivity requirements of 5W/m.K. The key factors considered in the formulation were the choice of fillers and resin chemistries. For this a variety of ceramic fillers such as Al 2O 3, Silicon Carbide (SiC) and Aluminum Nitride (AlN) were all considered to enhance the thermal conductivity of the molding compound as shown in Table 3. Each of these fillers has their own sets of advantages and disadvantages. For instance although AlN fillers have the highest bulk thermal conductivity of ~270 W/m.K, it is still an unsuitable choice because of its hygroscopic nature which results in it being easily reversible into Al(OH)3 and ammonia (NH 3) under the presence of moisture [4].
Table 3. Physical properties of thermal conductive fillers.
Al 2O 3
SiC AlN Density (g/cm 3) 4.00 3.22 3.26
CTE (10-6o C -1
) 7.60 4.51 4.40 Thermal conductivity (W/m.K) 30 220 270 Dielectric constant – 1MHz (k) 9.3 8.9 8.9 Young Modulus (GPa) 386 400 308
As discussed in the previous sections, it’s important to strike a balance between the thermal conductivity targets with the flowability of the material. Figure 8 shows the correlation between melt viscosities and thermal conductivity. From the graphs it was established that the increment in filler ratio not only increases the thermal conductivity but also drives an increase in melt
viscosities due to an increase in filler-matrix interactions. Some of the typical approaches used by molding compound suppliers to control the melt viscosities include selecting a lower viscosity resin to begin with.
For this the biphenyl resin is somewhat a preferred option among the molding compound suppliers because of its low moisture uptake and good adhesion to multiple interfaces. In addition melt viscosities can be controlled by optimizing the filler distribution ratio and the hardener system used.
1000
Thermal conductivity Melt viscosity 6100
Figure 8. Correlation between thermal conductivity and melt viscosity.
Table 4 shows the list of all the molding compound materials which were evaluated during the FCMB4 pathfinding time frame. They include the X-43-3244-2 formulation, which serves as the baseline for data collection. The other formulations are the X-43-3248-1A and X-43-3248-1C, which were reiterations from the X-43-3248-1 formulation. The key differences between the materials lie in the filler systems used. To achieve a thermal conductivity of ~5W/m.K, the filler loading (wt%) of Al 2O 3 + SiO 2 fillers for the X-43-3248-1A formulation were increased from 75wt% to 77wt%,
1030102030 40 Vol.% T h e r m a l c o n d u c t i v i t y (W /m .K )
M e l t v i s c o s i t y a .s )
(P
while for the X-43-3248-1C formulation, the SiO2 fillers were replaced with higher conductive SiC fillers. In addition silicone particles were added into the formulations to act to prevent excessive mold shrinkage during the curing process.
Table 4. High K molding compound formulation comparison. Supplier
(Materials)
Material description Key properties
Shin-Etsu (X-43-3244-2) FCMB3 baseline
Multi functional resin, 75
Vol% Al2O3 + SiO2
hybrid
T g – 183o C
CTE 1/2 – 14/43ppm
K – 3.2W/m.K
Spiral flow – 88cm
Shin-Etsu (X-43-3248-1) Biphenyl resin, 75 Vol%
Al2O3 + SiO2 hybrid
T g – 124o C
CTE 1/2 - 13/53 ppm
K - 3.2W/m.K
Spiral flow – 127cm
Shin-Etsu
(X-43-3248-1A) Biphenyl resin, 77 Vol%
Al2O3 + SiO2 hybrid
T g – 120o C
CTE 1/2 - 12/47ppm
K - 4.3W/m.K
Spiral flow – 78cm
Shin-Etsu
(X-43-3248-1C) Biphenyl resin, 75 Vol%
Al2O3 + SiC hybrid
T g – 122o C
CTE 1/2 – 12/47ppm
K = 4.7W/m.K
Spiral flow – 70cm
Kyocera (G1250HT4) Low molecular weight,
79% Al2O3 + SiO2 hybrid
T g – 130o C
CTE 1/2 - 14/54ppm
K = 4.6W/m.K
Spiral flow – 105cm
Sumitomo (X80744) Biphenyl resin, 77 Vol%
Al2O3 fillers
T g – 130o C
CTE 1/2 = 10/35ppm
K = 5.0W/m.K
Spiral flow – 90cm
* All the data reported in Table 4 are suppliers’ data
3.3. High K Molding Compound Assembly Issues
Figure 9 shows the overall process flow for the FCMB3 and FCMB4 over-molded package. The first step is to pre-bake the strips from underfill (CUF) module to drive off moisture, after which the out going strips is subjected to a plasma treatment to activate the surface in order to enhance the adhesion with the substrates. The following step is molding followed by a post mold cure to complete the cross-linking of the polymers.
Figure 9. Assembly process flow for FCMB4/FCMB3 discrete overmold packages.
One of the key challenges associated with the high K molding compound formulation is the significant increase in melt viscosity. As shown in Table 4, the molding compound viscosity increases significantly with an increase in filler loading. This may result in a range of issues due to the tight mold clearance between the die. In Table 5 mechanical models predicted that the mold cap requirements for a 5 Watts package needs to be < 100um thick if a 3.2W/m.K molding compound material were used.
Table 5.T j max estimation for 10W power dissipation.
Thermal conductivity (W/m.K)
Mold
clearance 0.8 3.2 5.0 10 20 250 um 179.4 116.1 108.5 101.8 98.4 150 um 145.6 107.7
103.1 99.1 97.0 100 um 128.8 103.4 100.4 97.7 96.4 50 um 111.9 99.2 97.7 96.4 95.7 Besides incomplete fill, other key issues associated with the mold flowability are mold flow marks, shown in Figure 10. This phenomenon occurs only on molding compound formulations with very low spiral flow, (e.g X-43-3248-1A). Although flow mark may not have any known reliability implications, it may impact the 2 dimension unit identification (2DID) readability and may also fail the customers’ quality acceptance criteria. From the process perspective, lowering the molding temperature may help to improve mold flow marks. However the only complication is a potential reduction in adhesion.
Figure 10. Mold flow marks on the strip after cure.
3.4. Key Challenges: Excessive Strip Warpage Managing incoming strip warpage and subsequently keeping the strips flat throughout the assembly pipeline is the biggest assembly challenge for the FCMB4 platform. This is important; because high strip warpage has been known to interact with the downstream modules such as ball attach and saw singulation resulting in yield loss. Embedding glass cloths into the substrate build up layers helps to improve the overall strip warpage; but it’s still insufficient to drive the warpage below the limits acceptable by the downstream modules as shown in Figure 11. Mechanical modeling predicted that the glass cloth locations would have a certain degree of influence on the overall strip warpage. It was predicted that the Joslin TV with the bottom layer glass cloth would have the lowest strip warpage, followed by the center layer glass cloths. On the other hand, glass cloth layers also have its own set of
disadvantages, including limitations on the package design rules and a higher cost target. In this section our discussions will center on the warpage response for the Joslin DLL3 1 layer glass cloth TV with different molding compound materials.
Figure 11. Dynamic warpage for Joslin and Wilmot.
Figure 12 compares the cluster level warpage of
ShinEtsu X-43-3248-1C, Kyocera G1250GT-4 and the Sumitomo X80744 high K molding compound formulations. Warpage measurements on the incoming strips showed that the strips are warped in the ‘smiley’ directions. Once the post mold cured, the strip warpage increased in the ‘smiley’ direction for both the X-43-3248-1C and G1250GT4. This trend was nevertheless different for the Sumitomo X80744 molded strip, with
the warpage direction changing from ‘smiley’ to ‘crying’ after post mold cure process.
Figure 12. Strip level warpage comparison at different process steps.
As discussed, the strip warpage mechanism is typically influenced by the T g , CTE and modulus of the molding compound. One example is the increase in strip warpage along the smiley directions for X-43-3248-1C and G1250GT4 material. This is due to its low T g , which results in higher expansion of the mold due to the transformation from crystalline to a rubbery phase at temperatures above T g . In addition strip warpage is influenced by the mold compound shrinkage during the curing process. For example it is believed that the polymeric shrinkage of the X80744 mold during the curing process triggers a compressive force towards the substrate resulting in changes in warpage magnitude from “smiley’ to ‘crying’ as shown in Figure 13. To reduce the cure and thermal shrinkage, various reiterations have been previously investigated. This includes the addition of silicone micro phases which reacts with the resin system to expand the domains against the polymerization shrinkage.
(4) Post (3) Post mold cure (2) Post mold
(1) Post underfill
Figure 13.
Warpage mechanism at different process steps: (1)
Substrate warp because of CTE mismatch with die (2) Infliction in
warpage direction after post mold cure because of mold shrinkage (3)
Slight reduction in warpage because of relaxation in polymer chains with extended cure (4) Post singulation warpage.
P o s t m o l d I n c o m i n g I n c o m i n g I n c o m i n g P o s t c u r e P o s t c u r e P o s t c u r e P o s t c u r e
P o s t m o l d P o s t c u r e P o s t m o l d P o s t m o l d P o s t m o l d P o s t m o l d I n c o m i n g P o s t c u r e I n c o m i n g I n c o m i n g
4. High K Mold Compound Reliability
Issues
4.1. Introduction
Passing the stipulated reliability stress test is an integral requirement for package certification. This requires the units to pass electrically (E-Test) at Level 3 preconditioning (L3P), and the subsequent readouts of unbiased highly accelerated stress test (HAST) 130o C/85%RH for 100 hours or temperature cycle profile B for up to 1000 cycles respectively. At every intermediate read out, the packages are screened with scanning acoustic microscope C-mode (CSAM) to check for any abnormalities at the interface.
4.2. Reliability Fall Outs
Table 6 lists the reliability data for the X-43-3244-2 molding compound at different reliability stress read outs. Two defects, namely mold-die delamination and mold cracking, were observed from the CSAM screening and will be discussed in further detail in the following subsections.
Table 6. Reliability fall out for mold-die delam and mold crack . EOL L3P uHAST (50hr) TCB (500X) BHAST
(2V)
Mold-die 0/430 19/430 1/99 23/99 23/23
Mold
crack
0/160 0/160 0/80 34/80 NA
4.3. Reliability Issues: Mold-Die Delamination
Figure 14 shows the CSAM and cross-section images on a unit with mold-die delamination after the L3P stress test. From the CSAM images, it was clear that mold-die delamination occurred only at the corners of the die. Cross sections revealed that mold-die delamination had a high potential risk of incurring electrical failures because
of propagation along the mold to die path towards the mold-underfill interface.
Figure 14. CSAM images of mold-die delamination (R) and cross section images of the delamination (L).
The root cause for mold-die delamination is suspected to be a combination of poor adhesion and high peel stress at the corners of the die. This hypothesis was supported by the die thickness data in Figure 15, which concurred that the thicker dies (270um) have a significantly lower fall out than the thinner die (225um). The reason is packages with a thicker mold cap warps more due to the dominance of the molding compound, which exerts more peel stress at the corners of the silicon.
Figure 15. X-43-3244-2 mold-die delamination fall out for two
different die thickness (225um and 270um).
One of the solution paths considered to resolve mold-die delamination, was to increase the adhesion of the molding compound to the silicon interface. Therefore a new molding compound, X-43-3248-1, formulated with the biphenyl resin as its matrix, was brought in for further evaluation. Besides having good adhesion, the
biphenyl resin is also known to have low moisture uptake compared to the higher T g multifunctional resin. Figure 16 shows the results for the button shear test which was conducted after the post mold cure and L3P stress. The results showed that the X-43-3248-1 had adhesion strength in the magnitude of 30% higher compared to the X-43-3244-2 and the silica filled KMC2500VAT1. However the strength degrades marginally after the L3P stress test due to thermal degradation on the polymer matrix.
Figure 16. Mold-die adhesion strength comparison between the X-43-3248-1 with the X-43-3244-2 and KMC2500VAT1.
Table 7 shows the results for the X-43-3248-1 molding compound formulation at different reliability stress read
outs. The result after L3P + uHAST 130oC/85%RH 100 hours is clean from any mold-die delamination. However when subjected to L3P + TCB500X stress, all of the 40 units loaded into the stress chamber failed for mold-die delamination based on CSAM inspections. This observation showed that the mold-die delamination defect observed on X-43-3248-1 molding compound is not moisture driven defect but is due to stress, aggravated by the low T g (124o C), which falls within the
range of the TCB profile.
Table 7. X-43-3248-1 fall out for mold-die delam.
Mold compoun
d EOL L3P
uHAST
(50hr)
TCB (500X)
X-43-3248-1
0/80 0/80 0/40 40/40
Besides evaluating different molding compound formulations, equal focus was given to assessing the risk for mold-die delamination from the thermal perspective.
Thermal modeling using the finite element method was
carried out to comprehend the gain in maximum junction temperature (T jmax ) if a delamination is present at the corners of the die. A sensitivity model for different
cumulative areas of delamination - 0.5%, 2%, 10%, 20%, 30% and 40% - was constructed to predict the minimum cumulative area of delamination that would be 270um/225um/490um 490um
Post mold cure
Post L3P stress X-43-3244-2 X-43-3248-1
KMC2500VAT1
required before a significant rise in temperature will be
observed.
The result for the sensitivity model shown in Figure 17, suggests that the rise in die corner temperature would be significant only if the total area of delamination for the four corners exceeds 10%, while the rise in die center temperature would be significant only if the total delamination area exceeds 40%. Therefore based on the modeling results, it was concluded that the thermal risk induced by the mold-die delamination defect is low, because none of the packages post reliability stress have a total delamination area exceeding 10%.
Figure 17. Temperature rise at die center, die corner and mold top for different corner defects assumptions.
4.4. Reliability Issues: Post TCB Mold-crack
Table 8 quantifies the number of units with mold cracks at different reliability stress read out. The results showed that mold crack occurred only during the temperature cycle profile B (TCB) reliability stress test, which indicates that mold cracking is a thermo-mechanical stress induced defect. Visual observations in Figure 18 on all the CSAM images showed that the cracks initiate from the high stress edges of the 2DID and the Instruction set for execution (ISET) mark, indicating that flaw is pre-requisite for crack propagation to occur.
Table 8. Reliability fall out for mold crack at different stress read out .
EOL L3P
uHAST (50hr)
TCB
(500X) TCB (1000X)
Mold
crack 0/160 0/160 0/80 34/80 40/53
Figure 18. Mold crack CSAM images at different stress read out.
The key concern with mold cracking is the potential propagation of the crack length towards the silicon, which would result in units failing electrically due to die cracking shown in Figure 19. It was also observed that the units with mold cracks have resulted in mold-die delamination, which is a risk for thermal dissipation if the delamination exceeds 10% of the total area. To comprehend the root cause of mold cracking, a lab scale simulation GRC focusing on two possible root causes, molding compound formulation or laser mark depth was carried out as shown in Table 9. The motivation for including a non high K mold compound leg was to comprehend if the higher CTE of the high K molding compound could have induced more stress compared to a non high K mold formulation. This GRC was carried out by pre-heating the units on a hot plate at 300o C for 10 minutes before being soaked in liquid nitrogen for another 10 minutes to simulate the TCB stress test. This was repeated for a total of 4 cycles, with visual inspections being carried out for mold cracking after every cycle.
Figure 19. Die crack from a post decap mold crack unit.
Table 9. Lab scale GRC leg to decouple mold crack root cause.
Molding compound Laser power (Watts) High K mold 1.43W 2.04W Non high K mold
1.43W
2.04W
The results in Table 9 showed that there was no mold crack observed for either the high and non high K molding compound after 4 cycles on the units marked with a 1.43 Watts laser power. On the other hand, units which were marked with a 2.04 Watts laser power showed mold cracking after 1 cycle of temperature simulation. This clearly showed that mold cracking is caused by the difference in mark depth induced by the different laser power. Hence it’s important to control the laser power and laser pulse of the 2DID to reduce the overall mark depth. The setback of doing so is the possible degradation in contrast for the 2DID readability.
Therefore for risk mitigation, the other option, which was pursued without having to reduce the mark depth, was to move the 2DID mark away from the die shadow region to reduce the corner stress. This was based on the mechanical modeling quasi-static analysis, which
showed that the higher stress region was within the die shadow because of the higher CTE mismatch between the mold-die compared to the mold-substrate interface shown in Figure 20.
Figure 20. Die crack from a post decap mold crack unit.
5. Long Gellation Compression Mold for
ESM packages
5.1. Motivation for (ESM) Packaging Technology
Exposed second level interconnect molding (ESM), was an innovative packaging solution evaluated by the FCMB4 pathfinding team to resolve die paddle indentation shown in Figure 21. Our observations
showed that die paddle indentation occurs after the chip attach process, due to CTE differences between the die and substrate. The magnitude of die paddle indentation is influenced by three factors, which are the die thickness, package form factor and copper density. Mechanical modeling has shown that a reduction in package form factor, increase in die thickness and copper density, helps to reduce the magnitude of die paddle indentation.
Figure 21. Die paddle indentation on the SLI side after CAM reflow.
The key concern with die paddle indentation is controlled collapsed chip connection (C4) open joints resulting in package electrical opens. To resolve the die paddle indentation, the DLL3 strips have to be physically stiff prior to the chip attach process. Therefore, the ESM process option, which involves the molding the second level interconnect (SLI) solder spheres prior to chip attach was evaluated as shown in Figure 22 as a potential solution path.
Figure 22. Exposed second level interconnect molding process flow.
The molding process for the ESM packages can be done with either the transfer or compression molding technique, using a mold film to obtain the desired the mold cap thickness as shown in Figure 23. The mold film thickness is determined from the targeted mold cap thickness, based on the rule of thumb which specifies that a molding film is compressible up to 30% of its total thickness. Conventional molding techniques for package on package (PoP), such as top gate molding, are deemed unsuitable for FCMB4 due to the tight package Z height requirements. Besides warpage, the key challenges for the ESM packages are: (1) Incomplete mold fill due to
the tight ball pitch; (2) Inconsistent mold cap thickness; (3) Mold interactions with SLI solder spheres/paste.
Figure 23. DLL3 strips before and after molding on the SLI.
Mold film
Figure 24. Mold film for control of mold cap thickness.
5.2. Materials Response for ESM Packages
The lack of available legacy information, for the team to leverage from such as mold ability, warpage controls and heat dissipation, proved to be a huge challenge for the FCMB4 pathfinding team. Therefore for a start, flow, mechanics and thermal modeling were heavily relied on to provide the 1st cut assessments. Thermal modeling in Figure 25 indicated that there is no additional margin gained in T jmax if either high K (5W/m.K) or non high K (0.8w/m.K) molding compound was used. This is because the heat dissipation path for the ESM packages is through the top side of the package via natural convection rather than through the bottom side towards the board.
Conventional process flow
Figure 25. Comparison of PoP options in thin enclosure.
From the mechanics perspective, it was proposed that a low CTE, high T g molding compound material would be most suitable to drive a reduction in unit level warpage. Since molding is carried out prior to the chip-attach and underfill processes, the strips would have to undergo two additional heat cycles. Therefore by choosing a lower CTE and higher T g material the strips would be subjected to less thermo-mechanical mismatching with the substrates, thus reducing the amount of residual stress formed within the package upon cool down.
Besides heat dissipation and warpage, the other key assembly issue which was observed during the assembly of the ESM packages was incomplete mold fill. This was attributed to the restriction in mold flow because of the tight ball pitch (~0.413mm) and mold clearance requirements. From the assembly perspective, the mold clearance was targeted to be less than 50% of the total ball height to ensure that there was sufficient exposed solder for the surface mount process. As discussed in section 2.4, one of the key differences between transfer in the compression mold process is the mechanism of the mold flow. The flow for the compression mold process is dictated by gravitational/ capillary effect. This requires the compression mold formulation to have low gellation and low viscosity characteristics to address the concern of incomplete fill.
In summary it was agreed upon that the molding compound properties that are most suitable for the ESM packages need to be a long gellation, low CTE, high T g non high K molding compound, which is a contrasting difference from the requirements for an over molded package. Examples of some of the formulations which were brought in during the FCMB4 pathfinding are shown in Table 10.
Table 10. ESM molding compound candidates’ comparison.
Supplier
(Materials)
Material description Key properties
Hitachi
CEL9240HF10Z
Baseline material
Silica filled biphenyl
resin + multi aromatic
T g – 125o C
CTE 1/2 – 10/39ppm
K – 0.8W/m.K
Spiral flow – 120cm
Hitachi
CEL9470HF10Z
High Tg, high CTE
SiO2 filled LMW
biphenyl
T g – 145o C
CTE 1/2 – 9/38 ppm
K – 0.8W/m.K
Spiral flow – 146cm
Hitachi
CEL9740HF10Z
High Tg, low CTE
SiO2 filled LMW
biphenyl
T g – 149o C
CTE 1/2 - 8/32ppm
K – 0.8W/m.K
Spiral flow – 101cm
Shin-Etsu
(X-43-3248-1A)
Low T g. High CTE
Biphenyl resin, 77 Vol%
Al2O3 + SiO2 hybrid
T g – 127o C
CTE 1/2 – 12/49ppm
Spiral flow – 97.5m
Modulus -38.9GPa
Shin-Etsu
(X-43-3248-2F)
Middle T g. High CTE
Biphenyl resin, 75 Vol%
Al2O3 + SiO2 hybrid
T g – 134o C
CTE 1/2 – 12/49ppm
Spiral flow – 95cm
Modulus – 29.8 GPa
* All the data reported in Table 7 are suppliers’ data
5.3. ESM Assembly Challenges: Package Warpage Figure 26 compares the dynamic warpage response for Joslin CUF only 0 glass cloth (no molding) with ESM 0 glass cloth as a function of temperature. The dynamic warpage response is literally the same for both options, with package warpage changing from convex to concave at higher temperatures due to changes in the effective CTE and modulus after glass transition temperature. In comparison the overall warpage magnitude for the ESM packages is lower than the CUF only packages. This is attributed to the molding compound restraining the overall strip warpage at room and high temperature.
Figure 26. Dynamic warpage response comparing CUF.
Figure 27 shows the unit level coplan results for all the four different molding compound materials described in Table 7. Based on the results shown, it’s clear that the unit with the lowest coplan is the X-43-3248-1A material, followed by the CEL9470HF10Z, SEC-X-43-3248-2F and CEL9740HF10Z, respectively. This observation shows that the effect of warpage is primarily dominated by the modulus of the materials, followed by T g and CTE as secondary modulators. Modulus plays an effective role in warpage controls by virtue of locking the strips down after the molding process, while T g and
CTE is consistent with the mechanical model predictions in section 5.2 that described the impact of residual stress on strip warpage during cool down.
Figure 27. Unit level coplan for different molding compound materials.
Besides molding compound formulation, warpage is also influenced by other dimensional properties like the mold
cap thickness. Figure 28 is a sensitivity study comparing the warpage magnitude for 100, 150 and 200um at both end of line (EOL) and surface mount (SMT) temperatures. Results showed that EOL warpage reduction was more pronounced with a thicker mold cap compared to a thinner mold cap. However the drive for an increase in the mold cap thickness is very much dependant on the ball size used. Hence it’s important to strike a balance between SMT and mold cap thickness.
Figure 28. Package flatness as a function of mold cap thickness.
5.3. Key challenges: Molding Residue on Ball
One of the key technology challenges in enabling
compression molding for ESM packages is mold residue on ball, shown in figure 29. Surface analysis on the mold residue with Energy Dispersive X-Ray Spectroscopy
(EDX) showed that the residue is a combination of mold resin and fillers. During the compression process, the solder spheres come into direct contact with the molding compound granules resulting in the impregnation of mold on the solder spheres. The main concerns with the mold residue on ball are test rejects due to the high resistance on ball and SMT non wet.
Figure 29. Mold residue on ball after compression molding process.
To address the concerns highlighted above, various methods were evaluated to remove the mold residue from the ball. One of the techniques is laser de-flash, which uses optical imaging to target only specific mold residue spots with a 1064nm, 3-20W, 5kHZ pulsed laser ablation. Results showed that laser ablation is effective in removing resin from the molding compound but has limitations in removing portions which are filled with alumina fillers (e.g. X-43-3248-1A formulation). This is because alumina filler is opaque to ultra-violet rays. At very high fluences, it is possible to ablate alumina, but when it’s distributed in the mold as 25um spheres, the laser cannot go directly after the alumina. Instead the ablation occurs around the surrounding resin, thus holding the particle in place.
Besides laser ablation, jet deflux for residue removal was
also evaluated. The jet deflux system shown in Figure 30 is a mechanical process involving high pressure water system heated at 80o C. The key challenge for this technology option is ball oxidation due to presence of moisture and temperature, which could further induce non wet concerns at SMT.
Figure 30. Mold residue removal with the jet deflux system.
5.4. Key Challenges: Mold Cap Thickness
Mold cap thickness consistency is a challenge for the ESM package due to high ball count and tight ball pitch (~0.413mm). Mold cap thickness variation is important as it can significantly impact the warpage performance of the package. Figure 31 compares the mold cap thickness for compression mold and transfer mold with results showing a significantly higher variation for the transfer mold compared to the compression mold with the same molding compound material (CEL9240HF10Z). This data clearly shows that
compression mold has better scalability compared to the transfer mold for thinner mold cap.
Figure 31. Mold cap thickness for transfer and compression mold.
Besides warpage mold cap thickness is also known to play a role in solder extrusion after second reflow. The mechanism shown in Figure 32, describes how solder extrusion occurs during the 2nd reflow. During the chip
attach reflow, the Tin Silver Copper (SAC) 405 solder alloys at the SLI will melt, undergoing a transformation from solidus to liquidus state at peak temperature of 230o C. At the same time, huge thermo-mechanical stresses develop around the ball because of package warpage, resulting in force acting towards the molten alloy. Once the threshold of force developed exceeds the surface tension of the molten alloy, solder extrusion would occur.
Figure 32. Solder extrusion mechanism during 2nd
reflow.
The key to addressing solder extrusion is to reduce the effective thermo-mechanical stresses acting towards the solder spheres. One way is to reduce the mold cap thickness used for the SLI molding. Our results in Figure 33 comparing two different mold cap thickness (135um and 200um) showed no solder extrusion with the thinner mold cap for a 280um solder ball. From this finding it’s important to define the right mold cap thickness to balance between package warpage and solder extrusion.
Figure 33. Comparison of the impact of different mold cap thickness towards solder extrusion.
4. Conclusions
The roadmap for the next generation molding compound is being shaped by the need to have good warpage controls and excellent heat dissipation qualities. Various technology challenges with respect to flow, stress and thermal will continue to push the boundaries of manufacturability. Hence materials formulations, new equipment technologies and innovative packaging architectures have to be continuously developed to ensure scalability for future FCMX platforms.
Acknowledgments
The authors would like to acknowledge the following individuals for their contributions in this manuscript: Gilbert De Guia, Sean Chan, Lam Pooi Kit, Eric Yong, Wang Kang Ji for mold-die delamination and mold crack data, Rudge Vethayanagam and Wong Au Seong for data on the ESM, Sim Huay Huay for cross-sections on the solder extrusion units, David Xu for initiating the trans high K molding compound development.
References
[1] C. Gurumurthy & H. Azimi., “Challenges in HDI substrates,’
Intel Assembly and Test Technology Journal, available on-line, Intel intranet, https://www.sodocs.net/doc/0117221314.html, . Volume 10, 2007.
[2] J. Brand et al., “Compression mold: A manufacturable low
temperature molding solution,’ Intel Assembly and Test Technology Journal, available on-line, Intel intranet, https://www.sodocs.net/doc/0117221314.html, . Volume 9, 2006.
[3] R. Viswanath et al., “Thermal performance challenges for silicon
to systems,’ Intel Technology Journal, available on-line, Intel intranet, https://www.sodocs.net/doc/0117221314.html, . Q3, 2000.
[4] D. Tracy, L. Nguyen, R. Giberti, A. Gallo, C. Bischof, J.N.
Sweet, and A.W. Hsia, “Reliability of aluminum nitride filed mold compound,” in Proc. IEEE Electronic Components and Technol. Conf. (ECTC), San Jose, CA, May. 18-21, 1997, pp.72-77.
Intel ?
is a registered trademark of Intel Corporation or its subsidiaries
in the United States and other countries.
*Other brands and names are the property of their respective owner.
vlookup函数的使用方法及实例.doc
vlookup函数的使用方法及实例vlookup函数的使用方法及实例 excel中vlookup函数的应用，重要在于实践。 下面我们先了就下函数的构成;接着举个例子说下;最后总结下急提下遇到的相关问题： (本作者采用的是excel2003版，不过这函数在任何版本都适应) 2首先我们介绍下使用的函数vlookup 的几个参数，vlookup是判断引用数据的函数，它总共有四个参数，依次是： 1、判断的条件 2、跟踪数据的区域 3、返回第几列的数据 4、是否精确匹配 该函数的语法规则如下： =VLOOKUP(lookup_value,table_array,col_index_num,range_looku p) 该函数的语法规则可以查看到，如下图： (excel07版) 如下图，已知表sheet1中的数据如下，如何在数据表二sheet2 中如下引用：当学号随机出现的时候，如何在B列显示其对应的物理成绩? 根据问题的需求，这个公式应该是：
vmdk文件损坏打不开怎么修复vmware vmdk文件损坏打不开修复方法一 EasyRecovery数据恢复软件支持恢复VMDK文件并存储在本地文件系统中。由于数据和有关虚拟服务器的配置信息都存储在VMDK文件中，而每个虚拟系统下通常又有多个VMDK镜像，此时选择正确的VMDK镜像对成功的完成文件恢复扫描而言就显得至关重要了。载入VMDK镜像并选择对应的卷，以开始扫描VMDK文件。 根据EasyRecovery软件给出的提示操作，完成VMDK文件恢复。 当然要想保证VMDK文件恢复的顺利进行，还需注意以下几点： 1、当发现数据丢失之后，不要进行任何操作，因操作系统运行时产生的虚拟内存和临时文件会破坏数据或覆盖数据; 2、不要轻易尝试Windows的系统还原功能，这并不会找回丢失的文件，只会为后期的恢复添置不必要的障碍; 3、不要反复使用杀毒软件，这些操作是无法找回丢失文件的。 vmware vmdk文件损坏打不开修复方法二
交互式多模型算法仿真与分析
硕037班 刘文3110038020 2011/4/20交互式多模型仿真与分析IMM算法与GBP算法的比较，算法实现和运动目标跟踪仿真，IMM算法的特性分析 多源信息融合实验报告
交互式多模型仿真与分析 一、 算法综述 由于混合系统的结构是未知的或者随机突变的，在估计系统状态参数的同时还需要对系统的运动模式进行辨识，所以不可能通过建立起一个固定的模型对系统状态进行效果较好的估计。针对这一问题，多模型的估计方法提出通过一个模型集{}(),1,2,,j M m j r == 中不同模型的切换来匹配不同目标的运动或者同一目标不同阶段的运动，达到运动模式的实时辨识的效果。 目前主要的多模型方法包括一阶广义贝叶斯方法（BGP1），二阶广义贝叶斯方法（GPB2）以及交互式多模型方法等（IMM ）。这些多模型方法的共同点是基于马尔科夫链对各自的模型集进行切换或者融合，他们的主要设计流程如下图： M={m1,m2,...mk} K 时刻输入 值的形式 图一 多模型设计方法 其中，滤波器的重初始化方式是区分不同多模型算法的主要标准。由于多模型方法都是基于一个马尔科夫链来切换与模型的，对于元素为r 的模型集{}(),1,2,,j M m j r == ，从0时刻到k 时刻，其可能的模型切换轨迹为 120,12{,,}k i i i k trace k M m m m = ，其中k i k m 表示K-1到K 时刻，模型切换到第k i 个， k i 可取1,2,,r ，即0,k trace M 总共有k r 种可能。再令1 2 1 ,,,,k k i i i i μ+ 为K+1时刻经由轨迹0,k trace M 输入到第1k i +个模型滤波器的加权系数，则输入可以表示为 0,11 2 1 12|,,,,|,,,???k k trace k k k i M k k i i i i k k i i i x x μ++=?∑ 可见轨迹0,k trace M 的复杂度直接影响到算法计算量和存储量。虽然全轨迹的
五种大数据压缩算法
?哈弗曼编码 A method for the construction of minimum-re-dundancy codes, 耿国华1数据结构1北京:高等教育出版社,2005:182—190 严蔚敏,吴伟民.数据结构(C语言版)[M].北京:清华大学出版社,1997. 冯桂,林其伟,陈东华.信息论与编码技术[M].北京:清华大学出版社,2007. 刘大有,唐海鹰,孙舒杨,等.数据结构[M].北京:高等教育出版社,2001 ?压缩实现 速度要求 为了让它(huffman.cpp)快速运行，同时不使用任何动态库，比如STL或者MFC。它压缩1M数据少于100ms（P3处理器，主频1G）。 压缩过程 压缩代码非常简单，首先用ASCII值初始化511个哈夫曼节点： CHuffmanNode nodes[511]; for(int nCount = 0; nCount < 256; nCount++) nodes[nCount].byAscii = nCount; 其次，计算在输入缓冲区数据中，每个ASCII码出现的频率： for(nCount = 0; nCount < nSrcLen; nCount++) nodes[pSrc[nCount]].nFrequency++; 然后，根据频率进行排序： qsort(nodes, 256, sizeof(CHuffmanNode), frequencyCompare); 哈夫曼树，获取每个ASCII码对应的位序列： int nNodeCount = GetHuffmanTree(nodes); 构造哈夫曼树 构造哈夫曼树非常简单，将所有的节点放到一个队列中，用一个节点替换两个频率最低的节点，新节点的频率就是这两个节点的频率之和。这样，新节点就是两个被替换节点的父
LZ77压缩算法实验报告
LZ77压缩算法实验报告 一、实验内容 使用C++编程实现LZ77压缩算法的实现。 二、实验目的 用LZ77实现文件的压缩。 三、实验环境 1、软件环境：Visual C++ 6.0 2、编程语言：C++ 四、实验原理 LZ77 算法在某种意义上又可以称为“滑动窗口压缩”，这是由于该算法将一个虚拟的，可以跟随压缩进程滑动的窗口作为术语字典，要压缩的字符串如果在该窗口中出现，则输出其出现位置和长度。使用固定大小窗口进行术语匹配，而不是在所有已经编码的信息中匹配，是因为匹配算法的时间消耗往往很多，必须限制字典的大小才能保证算法的效率；随着压缩的进程滑动字典窗口，使其中总包含最近编码过的信息，是因为对大多数信息而言，要编码的字符串往往在最近的上下文中更容易找到匹配串。 五、LZ77算法的基本流程 1、从当前压缩位置开始，考察未编码的数据，并试图在滑动窗口中找出最长的匹 配字符串，如果找到，则进行步骤2，否则进行步骤3。 2、输出三元符号组( off, len, c )。其中off 为窗口中匹
配字符串相对窗口边 界的偏移，len 为可匹配的长度，c 为下一个字符。然后将窗口向后滑动len + 1 个字符，继续步骤1。 3、输出三元符号组( 0, 0, c )。其中c 为下一个字符。然后将窗口向后滑动 len + 1 个字符，继续步骤1。 六、源程序 /********************************************************************* * * Project description: * Lz77 compression/decompression algorithm. * *********************************************************************/ #include VLOOKUP函数调用方法如下：（本次以提取RRU挂高数据为例） 一、本次涉及的相关文档。 1．《某地区TD宏站现场勘测数据汇总表》如表1-1，共1000多站，本次共列出104个站点的信息： 查看原文档请双击图标：某地区TD宏站现场 查勘数据汇总表，表1-1抓图如下： 2．某工程报价单，共30个宏站，如表1-2（本报价单其他信息均删除，只保留了站点名） 查看原文档请双击图标：某工程报价单.xlsx ，表1-2抓图如下： 二、本次我们以从表1-1中提取表1-2中30个站点的RRU挂高为例，具体步骤如下： 1．先在表1-2中增加“RRU挂高”这一列，然后先提取“某城关水泵厂南”的RRU挂高。操作方法为双击下图所示灰色表格，然后鼠标左键单击列表上面的fx插入函 数。 2．点fx后弹出如下图标，在下拉列表中选择“VLOOKUP”，点确定。 3．点确定后，弹出VLOOKUP函数调用表，包含4个部分（lookup_value、Table_array、C ol_index_num、Range_lookup）。 lookup_value:需要在数据表首列进行搜索的值，本次值为表1-1中的位置B2，用 鼠标单击表1-1中的“某城关水泵厂南”，即可自动输入。。 Table_array：需要在其中搜索数据的信息表，即在表1-2中选择一个搜索区域， 注意所选区域第一列必须是与Lookup_value中查找数值相匹配的 列（本次表1-1中的B列），最后一列必须大于等于RRU挂高那一列 （大于等于C列）,至于下拉行数肯定要大于等于106行。如下图： 选择相关区域后，VLOOKUP表中的Table_array会自动输入表1-1中所选区域，如 下图： 实验名称：LZSS压缩算法实验报告 一、实验内容 使用Visual 6..0 C++编程实现LZ77压缩算法。 二、实验目的 用LZSS实现文件的压缩。 三、实验原理 LZSS压缩算法是词典编码无损压缩技术的一种。LZSS压缩算法的字典模型使用了自适应的方式，也就是说，将已经编码过的信息作为字典， 四、实验环境 1、软件环境：Visual C++ 6.0 2、编程语言：C++ 五、实验代码 #include -- need byte-swap function for big endian CPU */ #define READ_LE32(X) *(uint32_t *)(X) #define WRITE_LE32(X,Y) *(uint32_t *)(X) = (Y) /* this assumes sizeof(long)==4 */ typedef unsigned long uint32_t; /* text (input) size counter, code (output) size counter, and counter for reporting progress every 1K bytes */ static unsigned long g_text_size, g_code_size, g_print_count; /* ring buffer of size N, with extra F-1 bytes to facilitate string comparison */ static unsigned char g_ring_buffer[N + F - 1]; /* position and length of longest match; set by insert_node() */ static unsigned g_match_pos, g_match_len; /* left & right children & parent -- these constitute binary search tree */ static unsigned g_left_child[N + 1], g_right_child[N + 257], g_parent[N + 1]; /* input & output files */ static FILE *g_infile, *g_outfile; /***************************************************************************** initialize trees *****************************************************************************/ static void init_tree(void) { unsigned i; /* For i = 0 to N - 1, g_right_child[i] and g_left_child[i] will be the right and left children of node i. These nodes need not be initialized. Also, g_parent[i] is the parent of node i. These are initialized to NIL (= N), which stands for 'not used.' For i = 0 to 255, g_right_child[N + i + 1] is the root of the tree for strings that begin with character i. These are initialized to NIL. Note there are 256 trees. */ for(i = N + 1; i <= N + 256; i++) g_right_child[i] = NIL; for(i = 0; i < N; i++) g_parent[i] = NIL; } /***************************************************************************** Inserts string of length F, g_ring_buffer[r..r+F-1], into one of the trees (g_ring_buffer[r]'th tree) and returns the longest-match position and length via the global variables g_match_pos and g_match_len. If g_match_len = F, then removes the old node in favor of the new one, because the old one will be deleted sooner. excel博大精深，其使用中有许多细节的地方需要注意。 vlookup函数的使用，其语法我就不解释了，百度很多，其实我自己也没看懂语法的解释，下面就按照我自己的理解来说说怎么用的。首先，这个函数是将一个表中的数据导入另一个表中，其中这两个表有一列数据是相同项，但是排列顺序不同。举例说明； 表1 表2 将表1中的face量一列导入表2中，但两表中的名称一列的排列顺序是不同的。此时需要使用vlookup函数。 下面介绍vlookup的使用方法。 将鼠标放到表2中的D2单元格上，点击fx,会出现一个对话框，里面有vlookup函数。若在常用函数里面没有，下拉找“查找与引用”，里面有此函数。点确定。表示此函数是在表2中的D2单元格中应用。 此时出现对话框： 在第个格里输入B2，直接用鼠标在表2中点击B2单元格即可。表示需要在查找的对象是表2中的B2单元格中的内容。 然后是第二个格，点表1，用鼠标选择整个表的所有数据。表示要在表1中的B1—C14区域查找表2中的B2单元格中的内容。 第三个格里输入在表2中要导入的列数在表1中的列数的数字。在此例中为C列，其列数数字为2.表示将表1中（B1—C14）区域中查找到的单元格里的内容相对应的列（第2列）中的单元格中的内容（face量列中的数据）导入表2中相应的单元格（D2）。 最后一个格中输入“0”。表示查找不到就出现#N/A。点确定，即出现相应数据，然后下拉复制格式。 当下拉出现这种情况的时候： 其实是其查找区域在下拉过程中随着行的改变而改变了。需要对查找区域做一下固定。其方法为，在选择区域后，在区域前面加“$”号（$B$1：$C$14）。 多媒体数据压缩实验报告 篇一：多媒体实验报告_文件压缩 课程设计报告 实验题目：文件压缩程序 姓名：指导教师：学院：计算机学院专业：计算机科学与技术学号： 提交报告时间：20年月日 四川大学 一，需求分析： 有两种形式的重复存在于计算机数据中，文件压缩程序就是对这两种重复进行了压 缩。 一种是短语形式的重复，即三个字节以上的重复，对于这种重复，压缩程序用两个数字：1.重复位置距当前压缩位置的距离；2.重复的长度，来表示这个重复，假设这两个数字各占一个字节，于是数据便得到了压缩。 第二种重复为单字节的重复，一个字节只有256种可能的取值，所以这种重复是必然的。给 256 种字节取值重新编码，使出现较多的字节使用较短的编码，出现较少的字节使用较长的编码，这样一来，变短的字节相对于变长的字节更多，文件的总长度就会减少，并且，字节使用比例越不均 匀，压缩比例就越大。 编码式压缩必须在短语式压缩之后进行，因为编码式压缩后，原先八位二进制值的字节就被破坏了，这样文件中短语式重复的倾向也会被破坏（除非先进行解码）。另外，短语式压缩后的结果：那些剩下的未被匹配的单、双字节和得到匹配的距离、长度值仍然具有取值分布不均匀性，因此，两种压缩方式的顺序不能变。 本程序设计只做了编码式压缩，采用Huffman编码进行压缩和解压缩。Huffman编码是一种可变长编码方式，是二叉树的一种特殊转化形式。编码的原理是：将使用次数多的代码转换成长度较短的代码，而使用次数少的可以使用较长的编码，并且保持编码的唯一可解性。根据 ascii 码文件中各 ascii 字符出现的频率情况创建 Huffman 树，再将各字符对应的哈夫曼编码写入文件中。同时，亦可根据对应的哈夫曼树，将哈夫曼编码文件解压成字符文件. 一、概要设计： 压缩过程的实现: 压缩过程的流程是清晰而简单的: 1. 创建 Huffman 树 2. 打开需压缩文件 3. 将需压缩文件中的每个 ascii 码对应的 huffman 编码按 bit 单位输出生成压缩文件压缩结束。 VLOOKUP函数的使用方法(入门级) VLOOKUP函数是Excel中几个最重函数之一，为了方便大家学习，兰色幻想特针对VLOOKUP 函数的使用和扩展应用，进行一次全面综合的说明。本文为入门部分 一、入门级 VLOOKUP是一个查找函数，给定一个查找的目标，它就能从指定的查找区域中查找返回想要查找到的值。它的基本语法为： VLOOKUP（查找目标，查找范围，返回值的列数，精确OR模糊查找) 下面以一个实例来介绍一下这四个参数的使用 例1：如下图所示，要求根据表二中的姓名，查找姓名所对应的年龄。 公式：B13 =VLOOKUP(A13,$B$2:$D$8,3,0) 参数说明： 1 查找目标：就是你指定的查找的内容或单元格引用。本例中表二A列的姓名就是查找目标。我们要根据表二的“姓名”在表一中A列进行查找。 公式：B13 =VLOOKUP(A13,$B$2:$D$8,3,0) 2 查找范围（VLOOKUP(A13,$B$2:$D$8,3,0) ）：指定了查找目标，如果没有说从哪里查找，EXCEL肯定会很为难。所以下一步我们就要指定从哪个范围中进行查找。VLOOKUP的这第二个参数可以从一个单元格区域中查找，也可以从一个常量数组或内存数组中查找。本例中要从表一中进行查找，那么范围我们要怎么指定呢？这里也是极易出错的地方。大家一定要注意，给定的第二个参数查找范围要符合以下条件才不会出错： A 查找目标一定要在该区域的第一列。本例中查找表二的姓名，那么姓名所对应的表一的姓名列，那么表一的姓名列（列）一定要是查找区域的第一列。象本例中，给定的区域要从第二列开始，即$B$2:$D$8，而不能是$A$2:$D$8。因为查找的“姓名”不在$A$2:$D$8区域的第一列。 B 该区域中一定要包含要返回值所在的列，本例中要返回的值是年龄。年龄列（表一的D列）一定要包括在这个范围内，即：$B$2:$D$8，如果写成$B$2:$C$8就是错的。 3 返回值的列数（B13 =VLOOKUP(A13,$B$2:$D$8,3,0)）。这是VLOOKUP第3个参数。它是一个整数值。它怎么得来的呢。它是“返回值”在第二个参数给定的区域中的列数。本例中我们 价值工程 置，是一项十分有意义的工作。另外恶意代码的检测和分析是一个长期的过程，应对其新的特征和发展趋势作进一步研究，建立完善的分析库。 参考文献： [1]CNCERT/CC.https://www.sodocs.net/doc/0117221314.html,/publish/main/46/index.html. [2]LO R,LEVITTK,OL SSONN R.MFC:a malicious code filter [J].Computer and Security,1995,14(6):541-566. [3]KA SP ER SKY L.The evolution of technologies used to detect malicious code [M].Moscow:Kaspersky Lap,2007. [4]LC Briand,J Feng,Y Labiche.Experimenting with Genetic Algorithms and Coupling Measures to devise optimal integration test orders.Software Engineering with Computational Intelligence,Kluwer,2003. [5]Steven A.Hofmeyr,Stephanie Forrest,Anil Somayaji.Intrusion Detection using Sequences of System calls.Journal of Computer Security Vol,Jun.1998. [6]李华,刘智,覃征,张小松.基于行为分析和特征码的恶意代码检测技术[J].计算机应用研究，2011，28（3）：1127-1129. [7]刘威，刘鑫，杜振华.2010年我国恶意代码新特点的研究.第26次全国计算机安全学术交流会论文集，2011，（09）. [8]IDIKA N,MATHUR A P.A Survey of Malware Detection Techniques [R].Tehnical Report,Department of Computer Science,Purdue University,2007. 0引言 现有的压缩算法有很多种，但是都存在一定的局限性，比如：LZw [1]。主要是针对数据量较大的图像之类的进行压缩，不适合对简单报文的压缩。比如说，传输中有长度限制的数据，而实际传输的数据大于限制传输的数据长度，总体数据长度在100字节左右，此时使用一些流行算法反而达不到压缩的目的，甚至增大数据的长度。本文假设该批数据为纯数字数据，实现压缩并解压缩算法。 1数据压缩概念 数据压缩是指在不丢失信息的前提下，缩减数据量以减少存储空间，提高其传输、存储和处理效率的一种技术方法。或按照一定的算法对数据进行重新组织，减少数据的冗余和存储的空间。常用的压缩方式[2,3]有统计编码、预测编码、变换编码和混合编码等。统计编码包含哈夫曼编码、算术编码、游程编码、字典编码等。 2常见几种压缩算法的比较2.1霍夫曼编码压缩[4]：也是一种常用的压缩方法。其基本原理是频繁使用的数据用较短的代码代替，很少使用 的数据用较长的代码代替，每个数据的代码各不相同。这些代码都是二进制码，且码的长度是可变的。 2.2LZW 压缩方法[5,6]：LZW 压缩技术比其它大多数压缩技术都复杂，压缩效率也较高。其基本原理是把每一个第一次出现的字符串用一个数值来编码，在还原程序中再将这个数值还成原来的字符串，如用数值0x100代替字符串ccddeee"这样每当出现该字符串时，都用0x100代替，起到了压缩的作用。 3简单报文数据压缩算法及实现 3.1算法的基本思想数字0-9在内存中占用的位最 大为4bit ， 而一个字节有8个bit ，显然一个字节至少可以保存两个数字，而一个字符型的数字在内存中是占用一个字节的，那么就可以实现2:1的压缩，压缩算法有几种，比如，一个自己的高四位保存一个数字，低四位保存另外一个数字，或者，一组数字字符可以转换为一个n 字节的数值。N 为C 语言某种数值类型的所占的字节长度，本文讨论后一种算法的实现。 3.2算法步骤 ①确定一种C 语言的数值类型。 —————————————————————— —作者简介：安建梅（1981-），女，山西忻州人，助理实验室，研究方 向为软件开发与软交换技术；季松华（1978-），男，江苏 南通人，高级软件工程师，研究方向为软件开发。 数据快速压缩算法的研究以及C 语言实现 The Study of Data Compression and Encryption Algorithm and Realization with C Language 安建梅①AN Jian-mei ；季松华②JI Song-hua （①重庆文理学院软件工程学院，永川402160；②中信网络科技股份有限公司，重庆400000）（①The Software Engineering Institute of Chongqing University of Arts and Sciences ，Chongqing 402160，China ； ②CITIC Application Service Provider Co.，Ltd.，Chongqing 400000，China ） 摘要：压缩算法有很多种，但是对需要压缩到一定长度的简单的报文进行处理时，现有的算法不仅达不到目的，并且变得复杂， 本文针对目前一些企业的需要，实现了对简单报文的压缩加密，此算法不仅可以快速对几十上百位的数据进行压缩，而且通过不断 的优化，解决了由于各种情况引发的解密错误，在解密的过程中不会出现任何差错。 Abstract:Although,there are many kinds of compression algorithm,the need for encryption and compression of a length of a simple message processing,the existing algorithm is not only counterproductive,but also complicated.To some enterprises need,this paper realizes the simple message of compression and encryption.This algorithm can not only fast for tens of hundreds of data compression,but also,solve the various conditions triggered by decryption errors through continuous optimization;therefore,the decryption process does not appear in any error. 关键词：压缩；解压缩；数字字符；简单报文Key words:compression ；decompression ；encryption ；message 中图分类号：TP39文献标识码：A 文章编号：1006-4311（2012）35-0192-02 ·192· VLOOKUP函数是Excel中的一个纵向查找函数，它与LOOKUP函数和HLOOKUP函数属于一类函数，在工作中都有广泛应用。VLOOKUP是按列查找，最终返回该列所需查询列序所对应的值；与之对应的HLOOKUP是按行查找的。 VLOOKUP函数的语法结构 整个计算机就相当于一门语言，首先我们就是要获取该函数的语法结构。以下是官网的语法结构 VLOOKUP（lookup_value, table_array, col_index_num, [range_looku p]）。 书上表述就是VLOOKUP（查找值，查找范围，查找列数，精确匹配或者近似匹配） 在我们的工作中，几乎都使用精确匹配，该项的参数一定要选择为false。否则返回值会出乎你的意料。 VLOOKUP函数使用示范 vlookup就是竖直查找，即列查找。通俗的讲，根据查找值参数，在查找范围的第一列搜索查找值，找到该值后，则返回值为：以第一列为准，往后推数查找列数值的这一列所对应的值。这也是为什么该函数叫做vlookup（v为vertic al-竖直之意，lookup即时英文的查找之意）。 现有如下手机的每日销售毛数据（图左），A分销商需要提供四个型号的销售数据（图右） 这个时候，你大概可能回去一个一个人工查找，因为我所提供的数据数量很少，但是其实工作中这种数据很庞大的，人工查找无疑即浪费时间，而且不能让A分销商相信你所提供数据的准确性。接下来，我们就需要本次的主角登场了。使用vlookup函数。 第一步：选中要输入数据的单元格，=VLOOKUP（H3,$A$3:$F$19,5,FALSE）如图 压缩编码算法设计与实现实验报告 一、实验目的：用C语言或C++编写一个实现Huffman编码的程序，理解对数据进行无损压缩的原理。 二、实验内容：设计一个有n个叶节点的huffman树，从终端读入n个叶节 点的权值。设计出huffman编码的压缩算法，完成对n个节点的编码，并写出程序予以实现。 三、实验步骤及原理： 1、原理：Huffman算法的描述 (1)初始化，根据符号权重的大小按由大到小顺序对符号进行排序。 (2)把权重最小的两个符号组成一个节点， (3)重复步骤2，得到节点P2，P3，P4……，形成一棵树。 (4)从根节点开始顺着树枝到每个叶子分别写出每个符号的代码 2、实验步骤： 根据算法原理，设计程序流程，完成代码设计。 首先，用户先输入一个数n，以实现n个叶节点的Huffman 树； 之后，输入n个权值w[1]~w[n]，注意是unsigned int型数值； 然后，建立Huffman 树； 最后，完成Huffman编码。 四、实验代码：#include int min(HuffmanTree t,int i) { int j,flag; unsigned int k = MAX; for(j=1;j<=i;j++) if(t[j].parent==0&&t[j].weight VLOOKUP函数的使用方法（入门级）一、入门级 VLOOKUP是一个查找函数，给定一个查找的目标，它就能从指定的查找区域中查找返回想要查找到的值。它的基本语法为： VLOOKUP（查找目标，查找范围，返回值的列数，精确OR模糊查找) 下面以一个实例来介绍一下这四个参数的使用 例1：如下图所示，要求根据表二中的姓名，查找姓名所对应的年龄。 公式：B13 =VLOOKUP(A13,$B$2:$D$8,3,0) 参数说明： 1 查找目标：就是你指定的查找的内容或单元格引用。本例中表二A列的姓名就是查找目标。我们要根据表二的“姓名”在表一中A列进行查找。 公式：B13 =VLOOKUP(A13,$B$2:$D$8,3,0) 2 查找范围（VLOOKUP(A13,$B$2:$D$8,3,0) ）：指定了查找目标，如果没有说从哪里查找，EXCEL肯定会很为难。所以下一步我们就要指定从哪 个范围中进行查找。VLOOKUP的这第二个参数可以从一个单元格区域中查找， 也可以从一个常量数组或内存数组中查找。本例中要从表一中进行查找，那么范 围我们要怎么指定呢？这里也是极易出错的地方。大家一定要注意，给定的第二 个参数查找范围要符合以下条件才不会出错： A 查找目标一定要在该区域的第一列。本例中查找表二的姓名，那么姓名 所对应的表一的姓名列，那么表一的姓名列（列）一定要是查找区域的第一列。 象本例中，给定的区域要从第二列开始，即$B$2:$D$8，而不能是$A$2:$D$8。 因为查找的“姓名”不在$A$2:$D$8区域的第一列。 B 该区域中一定要包含要返回值所在的列，本例中要返回的值是年龄。年 龄列（表一的D列）一定要包括在这个范围内，即：$B$2:$D$8，如果写成$B $2:$C$8就是错的。 3 返回值的列数（B13 =VLOOKUP(A13,$B$2:$D$8,3,0)）。这是VLO OKUP第3个参数。它是一个整数值。它怎么得来的呢。它是“返回值”在第 二个参数给定的区域中的列数。本例中我们要返回的是“年龄”，它是第二个参 数查找范围$B$2:$D$8的第3列。这里一定要注意，列数不是在工作表中的列 数（不是第4列），而是在查找范围区域的第几列。如果本例中要是查找姓名所 交互式多模型算法卡尔曼滤波仿真 1 模型建立 分别以加速度u=0、1、2代表三个不同的运动模型 状态方程x(k+1)=a*x(k)+b*w(k)+d*u 观察方程z(k)=c*x(k)+v(k) 其中，a=[1 dt;0 1],b=[dt^2/2;dt],d=[dt^2/2;dt],c=[1 0] 2 程序代码 由两个功能函数组成，imm_main用来实现imm 算法，move_model1用来生成仿真数据，初始化运动参数 function imm_main %交互式多模型算法主程序 %初始化有关参数 move_model %调用运动模型初始化及仿真运动状态生成函数 load movedata %调入有关参数初始值（a b d c u position velocity pmeas dt tg q r x_hat p_var） p_tran=[0.8 0.1 0.1;0.2 0.7 0.1;0.1 0.2 0.7];%转移概率 p_pri=[0.1;0.6;0.3];%模型先验概率 n=1:2:5; %提取各模型方差矩阵 k=0; %记录仿真步数 like=[0;0;0];%视然函数 x_hat_hat=zeros(2,3);%三模型运动状态重初始化矩阵 u_=zeros(3,3);%混合概率矩阵 c_=[0;0;0];%三模型概率更新系数 %数据保存有关参数初始化 phat=[];%保存位置估值 vhat=[];%保存速度估值 xhat=[0;0];%融合和运动状态 z=0;%量测偏差（一维位置） pvar=zeros(2,2);%融合后估计方差 for t=0:dt:tg; %dt为为仿真步长；tg为仿真时间长度 k=k+1;%记录仿真步数 ct=0; %三模型概率更新系数 c_max=[0 0 0];%混合概率规范系数 p_var_hat=zeros(2,6);%方差估计重初始化矩阵, %[x_hat_hat p_var_hat]=model_reinitialization(p_tran,p_pri,x_hat,p_var);%调用重初始化函数，进行混合估计，生成三个模型重初始化后的运动状态、方差 %混合概率计算 for i=1:3 u_(i,:)=p_tran(i,:)*p_pri(i); end for i=1:3 c_max=c_max+u_(i,:); end 目 录 实验一用C/C++语言实现游程编码 实验二用C/C++语言实现算术编码 实验三用C/C++语言实现LZW编码 实验四用C/C++语言实现2D-DCT变换13 实验一用C/C++语言实现游程编码 1. 实验目的 1) 通过实验进一步掌握游程编码的原理； 2) 用C/C++语言实现游程编码。 2. 实验要求 给出数字字符，能正确输出编码。 3. 实验内容 现实中有许多这样的图像，在一幅图像中具有许多颜色相同的图块。在这些图块中，许多行上都具有相同的颜色，或者在一行上有许多连续的象素都具有相同的颜色值。在这种情况下就不需要存储每一个象素的颜色值，而仅仅存储一个象素的颜色值，以及具有相同颜色的象素数目就可以，或者存储一个象素的颜色值，以及具有相同颜色值的行数。这种压缩编码称为游程编码，常用(run length encoding，RLE)表示，具有相同颜色并且是连续的象素数目称为游程长度。 为了叙述方便，假定一幅灰度图像，第n行的象素值为： 用RLE编码方法得到的代码为：0@81@38@501@40@8。代码中用黑体表示的数字是游程长度，黑体字后面的数字代表象素的颜色值。例如黑体字50代表有连续50个象素具有相同的颜色值，它的颜色值是8。 对比RLE编码前后的代码数可以发现，在编码前要用73个代码表示这一行的数据，而编码后只要用11个代码表示代表原来的73个代码，压缩前后的数据量之比约为7:1，即压缩比为7:1。这说明RLE确实是一种压缩技术，而且这种编码技术相当直观，也非常经济。RLE所能获得的压缩比有多大，这主要是取决于图像本身的特点。如果图像中具有相同颜色的图像块越大，图像块数目越少，获得的压缩比就越高。反之，压缩比就越小。 译码时按照与编码时采用的相同规则进行，还原后得到的数据与压缩前的数据完全相同。因此，RLE是无损压缩技术。 VLOOKUP函数 使用举例 如图 vlookup函数示例 所示，我们要在A2:F12区域中提取100003、100004、100005、100007、100010五人的全年总计销量，并对应的输入到I4:I8中。一个一个的手动查找在数据量大的时候十分繁琐，因此这里使用VLOOKUP函数演示： 首先在I4单元格输入“=Vlookup(”，此时Excel就会提示4个参数。 Vlookup结果演示 第一个参数，很显然，我们要让100003对应的是I4，这里就输入“H4,” ； 第二个参数，这里输入我们要查找的区域(绝对引用)，即“$A$2:$F$12,”； 第三个参数，“全年总计”是区域的第六列，所以这里输入“6”，输入“5”就会输入第四季度的项目了； 第四个参数，因为我们要精确的查找工号，所以留空即可。 最后补全最后的右括号“)”，得到公式“=VLOOKUP(H4,$A$2:$F$12,6)”，使用填充柄填充其他单元格即可完成查找操作。 VLOOKUP函数使用注意事项 说到VLOOKUP函数，相信大家都会使用，而且都使用得很熟练了。不过，有几个细节问题，大家在使用时还是留心一下的好。 一．VLOOKUP的语法 VLOOKUP函数的完整语法是这样的： VLOOKUP(lookup_value,table_array,col_index_num,range_lookup) 1．括号里有四个参数，是必需的。最后一个参数range_lookup是个逻辑值，我们常常输入一个0字，或者False;其实也可以输入一个1字，或者true。两者有什么区别呢？前者表示的是完整寻找，找不到就传回错误值#N/A；后者先是找一模一样的，找不到再去找很接近的值，还找不到也只好传回错误值#N/A。这对我们其实也没有什么实际意义，只是满足好奇而已，有兴趣的朋友可以去体验体验。 2．Lookup_value是一个很重要的参数，它可以是数值、文字字符串、或参照地址。我们常常用的是参照地址。用这个参数时，有三点要特别提醒：A）参照地址的单元格格式类别与去搜寻的单元格格式的类别要一致，否则的话有时明明看到有资料，就是抓不过来。特别是参照地址的值是数字时，最为明显，若搜寻的单元格格式类别为文字，虽然看起来都是123，但是就是抓不出东西来的。VLOOKUP函数的使用方法(图解说明_很详细)
LZSS压缩算法实验报告
excel中的vlookup函数的使用方法及注意事项
多媒体数据压缩实验报告
VLOOKUP函数的使用方法(从入门到精通)
数据快速压缩算法的C语言实现
vlookup函数的使用方法实例
压缩编码算法设计与实现实验报告
VLOOKUP函数地使用方法
交互式多模型算法卡尔曼滤波仿真
数据压缩实验指导书
vlookup函数使用说明