Научная статья

Коммуникация, связь, радиоэлектроника и цифровые приборы

A relatively hard polishing padis used first to planarize the wafer surface using a chemically reactive and abrasive slurry. A second polishing step is then carried out on a relatively soft polishing pad using a slurry to remove or reduce scratches introduced by polishing with the hard pad.



47.26 KB

0 чел.



A process for polishing a layer on a semiconductor wafer in which the incidence of undesirable scratches on the polished surface is reduced by using a multiple step polishing pro­cedure. A relatively   hard  polishing  padis  used  first  to

planarize the wafer surface, using a chemically reactive and abrasive slurry. A second polishing step is then carried out on a relatively soft polishing pad, using a slurry to remove or reduce scratches introduced by polishing with the hard pad. A final polishing step is performed on the soft polishing pad using de-ionized water to remove particles from the surface of the wafer.


This is a continuation of application Ser. No. 08/512,771, filed Aug. 9, 1995, now abandoned.


when forming the metal layer into the desired circuit inter­ connections. The resulting filaments of metal remaining  in the CMP scratches after forming the interconnections can cause a short circuit fault to occur if the scratch is proximate

5   to contacts or interconnection lines in the metal layer.

Thus, what is required is a chemical-mechanical polishing process in which scratches on the wafer surface are avoided or removed prior to deposition of a subsequent conductive layer.

This invention relates to the production of semiconductor  10 integrated circuits, and more specifically to a process for manufacturing integrated circuits in which scratches due to chemical-mechanical polishing are reduced.


In accordance with the present invention, there is pro­ vided a process for planarizing a layer formed on a wafer during fabrication  of  a semiconductor  integrated  circuit. A


Modern ultra-large scale integrated (ULSI) circuits are constructed with up to several millions of active devices, such as transistors and capacitors, formed in a semiconduc­ tor substrate. Interconnections between the active devices are created by providing a plurality of conductive intercon­ nection layers, such as polysilicon and metal, which are etched to form conductors for carrying signals between the various active devices. The individual interconnection layers are nominally electrically isolated from one another, and from the silicon substrate, by an insulative interlayer dielec­ tric (ILD), such as silicon dioxide (Si02) produced by chemical vapor deposition (CVD). The conductive layers and interlayer dielectric are deposited on the silicon sub­ strate wafer in succession, with each layer being, for example, of the order of 1 micron in thickness. The ILD conformably covers the underlying layer (e.g. a metal layer etched to form conductive interconnects) such that the upper surface of the ILD is characterized by a series of non-planar

steps which correspond in height  and width to the underly­ ing interconnect  lines.

These height variations in the upper surface of the ILD can have deleterious effects on the subsequent steps and layers applied in forming the integrated circuit. For example,

 15 first polishing step of the process involves polishing  the wafer  on  a  first  polishing  pad  using  a  slurry  having  a

chemical reagent and a suspension of abrasive particles. The first polishing pad is a relatively hard pad (relatively low compressibility)  and is used  to abrade a surface portion of

20 the layer so as to planarize the surface of the layer. A second polishing step of the process is then performed on a second polishing pad, also using a slurry. The second polishing pad

is a relatively soft pad (relatively high compressibility), and is used to remove scratches from the planarized layer surface

25  which  may  have  resulted  from  the  first  polishing  step.

Finally,  a  third  polishing  step  is performed  on  the  wafer

using the second polishing pad but de-ionized water instead of the abrasive slurry. The third polishing step removes the slurry solution and particles from the surface of the wafer.

30  In order to abrade the surface of the  layer  being planarized, the pressure used to apply the wafer to the first polishing pad is relatively high, and is preferably in  the range of about 3.5 to about 9 pounds-per-square-inch (PSI). The scratch removal phase comprising the second polishing

35 step involves less pressure between the wafer and the second polishing pad, in the range of about 2 to about 5 PSI. Finally, the rinsing stage comprising the third polishing step involves the least pressure, and is preferably of the order of 0.5 to 1.5 PSI.

a non-planar dielectric surface can interfere with the optical   40

resolution of subsequent photolithographic processing  steps.

This can make the high resolution lines required for compact ULSI circuits difficult to produce. Additionally, if the height variations in the ILD surface  are severe, there  is a danger that insufficient metal coverage can occur at the step height   45

variations  in  the  subsequent  conductor  layer,  which  can

result in open circuit flaws.

In order to combat these difficulties, various techniques have been developed in an attempt to better planarize the upper  surface  of  the ILD.  One  approach, referred  to  as  50 chemical-mechanical  planarization  or  polishing  (CMP), employs abrasive polishing to remove the surface height variations of the dielectric layer. According to this method

the semiconductor wafer is pressed against a moving pol­ ishing surface that is wetted  with  a chemically  reactive,  55 abrasive slurry. Slurries are usually either basic or acidic and generally contain a suspension of alumina or silica particles. The polishing surface and wafer are moved relative to one another in an abrasive fashion to remove protruding portions

of the dielectric layer. The abrasive polishing process con-   60

tinues until the surface of the ILD is largely flattened.

One problem which has been encountered with subjecting semiconductor  wafers  to  chemical-mechanical  polishing  is that scratches can be produced on the polished surface of the wafer (e.g. on the surface of the ILD). Metal deposited on   65 the ILD as the next layer of the integrated circuit fills these scratches,  but  can  then  be  difficult  to  remove  therefrom

Rotation of both the polishing pad and the wafer is employed for each  of the polishing steps, with rotation speeds typically in the range of about 10 to 20 revolutions­ per-minute (RPM) for the first and second polishing steps. The third polishing step is preferably carried out with higher rotational speeds, such as greater than 30 RPM for each of the polishing pad and the wafer. In the preferred embodi- ments the third polishing step involves rotational speeds of the order of 60 RPM which facilitates cleansing of the wafer of slurry and abraded particles with the aid of the supplied water.

The wafer is polished on the first polishing pad to remove a desired amount of material from the layer being polished, and the time required for polishing during the first polishing step is adjusted accordingly. For  example, for many appli­ cations the first polishing step time will be in the range of about one minute to about four minutes. The second pol- ishing step, used for removing or reducing scratches on the wafer, generally achieves better results with longer polishing times, although it has been found that  a second polishing step time in the range of about 30 to 45 seconds can yield adequate results. It is preferred that the second polishing step be performed for at least 30 seconds. The third polishing step, which is primarily for removing particles from the wafer surface with water as a rinsing agent, can be per­ formed for any suitable length of time, and one minute of this processing step has been found to generally be suffi- cient.

Other features and advantages of the present  invention will be apparent from the appended claims, and from  the detailed description  of the invention which follows below.


The invention is described in greater detail hereinafter, by way of example only, with reference to the accompanying drawings, wherein:

ing surface is typically a planar pad made of a porous material, such  as blown poyurethane mounted on a planar rotatable  platen.

FIG. 2 illustrates a semiconductor wafer 100 following

5 the CMP processing, wherein the surface features of the ILD 160 have been generally removed. A prior art polishing process may involve an initial polishing step on a relatively hard polishing pad (e.g. a polishing pad known by the name IC-60 manufactured by Radel, Inc.) using an abrasive slurry

FIG. 1 is a cross-section of a portion of a conventional  10

semiconductor substrate.

FIG. 2 is a cross-section of a conventional semiconductor substrate  after  CMP processing.

FIG. 3 is a cross-section of the semiconductor substrate

in order to abrade the surface of the ILD to a generally planar state (FIG. 2). The initial polishing step may in some instances be followed by a rinsing or buffing step in order to remove particles from the surface of the wafer which may have adhered thereto during the planarizing step. The rinsing

illustrated  in  FIG.  2  after  a  photoresist   layer  has  been deposited.

FIG. 4 is a cross-section of the semiconductor substrate of FIG. 3 after the formation of etched contact openings.

FIG. 5 is a cross-section of the semiconductor substrate of FIG. 4 after a conductive material deposition.

FIG. 6 is a cross-section of the semiconductor substrate illustrated in FIG. 5 after the conductive material has been etched back.

FIG. 7 is a cut-away side view of a conventional chemical-mechanical polishing apparatus; and

FIG. 8 is a flowchart showing an example of the method of the present invention.


A novel chemical-mechanical polishing process for semi­ conductor integrated circuit formation is described. In the following description, numerous specific details are set forth, such as specific materials and process parameters, etc. in order to provide a thorough understanding of the present invention. It will be obvious, however, to one skilled in the art that the present invention may be practiced without these specific details. In other instances, well known semiconduc­ tor processes and machinery have not been described in particular detail in order to avoid unnecessarily obscuring the present invention.

In FIGS. 1to 4, there is shown a cross-sectional repre­ sentation of a potion of a semiconductor wafer at various stages of processing according to the prior art. These figures illustrate a difficulty which can arise when scratches are formed on the surface of an interlayer dielectric (ILD) by a chemical-mechanical polishing step.

FIG. 1 shows a cross-section view of a portion of a semiconductor wafer 100 comprising a silicon substrate 120 having a polysilicon line 140 formed thereon. An insulating interlayer dielectric 160 is deposited over the silicon sub­ strate 120 and polysilicon line 140. The conformal nature of the ILD 160 causes the surface 180 thereof to generally reflect the uneven topography of the underlying layers. The uneven surface 180 of the ILD 160 can have disadvanta­ geous effects on the deposition, patterning and etching of subsequent layers. Consequently, a technique known as chemical-mechanical polishing (CMP) has been developed in order to planarize the surface of ILD 160 prior to subsequent processing. In general, chemical-mechanical polishing processes involve pressing the semiconductor wafer against a moving polishing surface that is wetted with a chemically reactive, abrasive slurry. Slurries are usually either basic or acidic and generally contain a suspension of alumina or silica particles as an abrasive agent. The polish-

 15 or buffing step may be performed using a CMP apparatus and  a  relatively  soft  polishing  pad  (such  as  a  SUBA IV

polishing pad from Radel, Inc.) which is supplied with water while the wafer is applied thereto. However, in some instances the chemical-mechanical polishing process of the

20 prior art can leave a scratch 200 in the surface of the polished layer, which can eventually result in a flawed semiconductor circuit  as described herein below.

The next step in the process involves depositing a photo­ resist layer 220, and patterning the layer with openings 222

25 for the formation of interlayer contacts to the silicon sub­ strate 120 and polysilicon line 140 (FIG. 3). Etching of the ILD   160  is  then  performed   according  to  the  patterned

photo-resist 220 using, for example, a conventional dry plasma etch process. This results in the formation of etched

30  contact openings 224A, 224B and 22C, as illustrated in FIG.

4. In this example, the scratch 200 in the surface of the lLD

160 is in the region of adjacent contact openings 224B and 224C. A layer 230 of a conductive material such as tungsten is then deposited on the wafer, to fill the contact openings so

35  as to create interlayer contacts. The conductive material 230 is then removed from the surface of the ILD 160 to leave plugs of conductive material 232A, 232B and 232C forming the interlayer contacts. The removal of the conductive material  230  is performed  using  an  etching  process  or  a

40   polish-back  process  of  chemical-mechanical  polishing.

However, as shown in FIG. 6, the depression in the surface

of ILD 160 resulting from the scratch 200 can disadvanta­ geously retain a portion of the conductive material on the surface of ILD 160, causing a conductive bridge 240 short-

45 circuiting adjacent interlayer contacts 232B and 232C. This short circuit results in a flawed integrated circuit. Accordingly, it can be seen that scratches to a semiconductor wafer introduced during a chemical-mechanical polishing processing step can, in some circumstances, produce flawed

50 integrated circuits, and thereby reduce the yield of a fabri­ cation  process.

*   * *  *  *


А также другие работы, которые могут Вас заинтересовать

38430. Многокритериальный анализ решений по обеспечению безопасности техногенного объекта с расширенным понятием безопасности 735 KB
  Экспертные подходы многокритериальных принятий решений на основе сравнений многокритериальных альтернатив обеспечения социотехнической безопасности техногенного объекта ТО Определение наилучшей альтернативы. Методы ELECTRE ранжирования многокритериальных альтернатив. Применения МАИ для многокритериальных сравнений альтернатив оценки безопасности техногенного объекта
38431. Метод расчета мехатронной системы привода телескопа на основе равновесно-оптимальной балансировки 3.15 MB
  Cтабильноэффективный компромисс в ММС СТЭК ММС это объединение стабильности и эффективности в рамках множества решений от полного совпадения данных свойств до обеспечения определенной степени сближения в условиях информационнотактических расширений соглашений. СТЭК в иерархических системах дополняет СТЭК ММС СТЭК ИС.3 П Парето граница АВ; Н Нэшравновесие; УКУ область угрозконтругроз; ИТ идеальная точка; УК – оптимальная часть Пграницы на основе узкого конуса ; Ш – точка Шепли; СНД – ПаретоНэш область компромиссов ПНОК...
38432. Моделирование процесса нанесения краски устройством с применением робота Kawasaki 3.31 MB
  Определить параметры траекторного движения захвата – декартовы координаты углы Эйлера скорости обеспечивающие непрерывное точное и безошибочное выполнение технологических операций. Пульт выполняет серию важных задач: Ручное управление роботом Обучение данных позиции координат Обучение вспомогательных данных блочное программирование Рис. В языке используется термин позиция так как этот термин выбран в стандарте ISO фактически же позицией является совокупность трёх координат конца центра схвата TCP а также трёх эйлеровых...
38433. Разработка и исследования метода сетевого оператора для адаптивного управления динамическим объектом 3.77 MB
  Решение задачи синтеза системы управления — есть поиск управления, как функции от пространственных координат. При этом сложнее всего получить структуру функции многомерного управления. До недавнего времени данная задача решалась следующим образом: исследователь определял структуру математического выражения, оставляя параметры неопределенными, затем их значения находились с помощью численных методов в соответствии с заданным критерием управления.
38434. Разработка и исследование искусственной нейронной сети для управления динамическим объектом с переменными параметрами 2.08 MB
  Искусственные нейронные сети используются в качестве регулятора многомерных и многосвязных динамических объектов. Применение искусственных нейронных сетей для целей управления является одной из многочисленных областей относительного нового раздела современной науки – нейроинформатики..
38435. Разработка системы конкурентно-оптимального прогноза управления предприятием на основе динамической модели олигополии 3.31 MB
  Cтабильноэффективный компромисс в ММС СТЭК ММС – это объединение стабильности и эффективности в рамках множества решений – от полного совпадения данных свойств в одной точке пространства J или U до обеспечения возможной степени сближения в условиях информационнотактических расширений соглашений. СТЭК ММС дополняют СТЭК в иерархических системах СТЭК ИС где реализуется право первого хода на основе субъективной информации что составляет тему отдельного исследования. Компромиссы на основе комбинации ПаретоНэшУКУШеплиподходовП –...
38436. Разработка и исследование метода аналитического программирования для структурно-параметрического синтеза системы управления динамическим объектом 14.23 MB
  Сложность задачи состоит в том, что в общей постановке для нелинейного объекта с произвольными критериями качества практически невозможно получить аналитическое решение. Поэтому известные методы для решения, как правило, неэффективны, поскольку используют специальные свойства объектов и функционалов.
38437. Многокритериальный синтез позиционного управления на основе многопрограммной стабилизации 2.76 MB
  Комбинированный метод многокритериального синтеза позиционного управления формирует аналитический вид управления, как набор параметров и известных функций состояния из состава «сетевого оператора» конечной сети этих функций и операций над ними
38438. Разработка моделирование процесса поддержки заданных климатических условий в помещении в системе InTouch 2.09 MB
  Трехдиапазонный регулятор температуры 60 3. Ведь отапливать рабочие помещения в выходные и праздничные дни не следует так интенсивно как по будням или скажем интенсивность отопления должна зависеть от температуры за окном а не от календарного времени года: вспомним хотя бы минувшую зиму когда в январе была плюсовая температура а отопление по интенсивности было “зимним†приходилось открывать окна в зданиях а можно было всего лишь снизить мощность обогрева тем самым сэкономить значительные средства. Возможные колебания...